Loss of N-terminal Charged Residues of Mouse CD3ϵ Chains Generates Isoforms Modulating Antigen T Cell Receptor-mediated Signals and T Cell Receptor-CD3 Interactions*

The antigen T cell receptor (TCR)-CD3 complexes present on the cell surface of CD4+ T lymphocytes and T cell lines express CD3ϵ chain isoforms with different isoelectric points (pI), with important structural and functional consequences. The pI values of the isoforms fit the predicted pI values of CD3ϵ chains lacking one, two, and three negatively charged amino acid residues present in the N-terminal region. Different T cells have different ratios of CD3ϵ chain isoforms. At a high pI, degraded CD3ϵ isoforms can be better recognized by certain anti-CD3 monoclonal antibodies such as YCD3-1, the ability of which to bind to the TCR-CD3 complex is directly correlated with the pI of CD3ϵ. The abundance of CD3ϵ isoforms can be modified by treatment of T cells with the proteinase inhibitor phenanthroline. In addition, these CD3ϵ isoforms have functional importance. This is shown, first, by the different structure of TCR-CD3 complexes in cells possessing different amounts of isoforms (as observed in surface biotinylation experiments), by their different antigen responses, and by the stronger interaction between low pI CD3ϵ isoforms and the TCR. Second, incubation of cells with phenanthroline diminished the proportion of degraded high pI CD3ϵ isoforms, but also the ability of the cells to deliver early TCR activation signals. Third, cells expressing mutant CD3ϵ chains lacking N-terminal acid residues showed facilitated recognition by antibody YCD3-1 and enhanced TCR-mediated activation. Furthermore, the binding avidity of antibody YCD3-1 was different in distinct thymus populations. These results suggest that changes in CD3ϵ N-terminal chains might help to fine-tune the response of the TCR to its ligands in distinct activation situations or in thymus selection.

The antigen T cell receptor (TCR) 4 complex is responsible for antigen recognition through TCR␣␤ (or TCR␥␦) variable heterodimers. These are noncovalently associated with CD3␥, CD3␦, CD3⑀, and (CD247) polypeptides involved in the initiation of signal transduction and the control of complex expression (reviewed in Refs. 1 and 2). Current structural models of the TCR-CD3 complex support the idea that each minimal subunit contains one TCR␣␤ (or TCR␥␦) heterodimer and two CD3⑀ polypeptides per TCR heterodimer (3)(4)(5)(6)(7)(8). CD3⑀ ectodomains pair noncovalently with one CD3␥ or CD3␦ chain through unique excluding sites in their G strands and membrane-proximal stalk sequences, and it is thus assumed that there is one CD3⑀␦ and one CD3⑀␥ dimer per complex (9 -14). Finally, it has been shown that covalently linked chain homodimers are needed for efficient transport of complete TCR-CD3 complexes from the endoplasmic reticulum to the cell surface (reviewed in Ref. 2), and data from in vitro assembly of TCR-CD3 chains suggest the association of one dimer per complex (3,4). From these data, it follows that the minimal TCR-CD3 complex unit contains eight polypeptides (␣␤⅐⑀␦⅐⑀␥⅐ 2 ).
The data summarized above have been generated using many different cells and cell lines in diverse experimental approaches, and it would be reasonable to conclude that the TCR-CD3 complex is a constant structure, the components of which (beyond those differences arising from V region variability) are equal in all T cells. However, the existing data also indicate that there are profound quantitative and qualitative changes during the development of not only T lymphocytes (15), but also among different mature T cell subsets or T cell lines. Concerning CD3, different ratios of CD3␥ and CD3␦ polypeptides in TCR-CD3 complexes from different T cell lines have been reported (9). Furthermore, CD3␦ chains are absent, and there are wide differences in CD3␥ chain glycosylation in ␥␦ T cells (16). Even so, TCR␥␦-CD3 complexes contain two CD3⑀␥ dimers per complex (8). TCR-CD3 complexes naturally form aggregates, the degree of aggregation of which within one cell or among T cells cannot be easily ascribed to the nature of the TCR antigen recognition unit (7,17). Differences in TCR-CD3 structure among human or mouse T cell lines have been also detected biochemically and by differences in their relative recognition by anti-TCR or anti-CD3 antibodies (18 -20).
All these differences might have important functional consequences if they can alter the factors determining the efficiency of TCR-mediated signals. These include the efficiency of spontaneous (7,17) or ligand-induced oligomerization/polymerization of TCR-CD3 complexes, the sensitivity to induction of conformational changes upon ligand binding (21), and the efficiency of coreceptor association with the TCR-CD3 complex (22,23). In turn, these differences could correlate with the amount and distribution of molecules involved in the activation of distinct pathways and/or the threshold and kinetic patterns of T cell activation.
In a previous study (18), we detected differences in the recognition of mouse CD3 by monoclonal antibodies that were linked to differences in the N-terminal sequence of CD3⑀ as determined by recognition with an N-terminal peptide-specific antibody. These differences were not due to alternative splicing of the mini-exons coding for the N-terminal sequence of CD3⑀, but were due to degradation by proteinases (including metalloproteinases) sensitive to phenanthroline. Interestingly, the association of CD3 with the TCR is weaker when CD3⑀ N-terminal sequences are degraded (18). Here, we have taken advantage of the presence of negatively charged amino acid residues in the N-terminal sequence of CD3⑀ to further analyze this phenomenon. Removal of these charged residues should affect the isoelectric point of CD3⑀, rendering isoforms with distinct pI values that could be distinguished by two-dimensional PAGE. Furthermore, we have generated mutant CD3⑀ chains lacking charged amino acids and showed that their loss can enhance the response to TCR-mediated activation.
Cells and Cell Lines-The mouse Th2 T cell lines SR.D10 (D10) (33), D10.TCR2.3 (34), and AK-8 (35,36) are specific for a peptide comprising residues 134 -146 of chicken conalbumin and major histocompatibility complex (MHC) class II of the haplotype. AE103 is an I-A k -specific Th1 clone (37). They were cultured and grown as described previously in detail (18). Spleen and thymus cell suspensions were obtained from BALB/c mice.
Plasmids and Transfectants-A mouse CD3⑀ cDNA (nucleotides 69 -650 of GenBank TM /EBI Data Bank accession number NM007648, including the mouse CD3⑀ open reading frame between nucleotides 80 and 649) was amplified by reverse transcription-PCR from total RNA extracted from SR.D10 cells using oligonucleotides 5Ј-GAGAGAGAATTCTGAGAG-GATGCGG-3Ј (sense) and 5Ј-GTCAGACTGCTCTCTGAT-TCAGGCC-3Ј (antisense). The amplified cDNA was ligated into the pTA-TOPO plasmid vector (Invitrogen). A C-terminal sequence tag for five residues of the vesicular stomatitis virus glycoprotein was introduced by PCR amplification using the CD3⑀ cDNA cloned in pTA-TOPO as a template and primers 5Ј-GAGAGAGAATTCTGAGAGGATGCGG-3Ј (sense) and 5Ј-CGGAATTCTCATTTGCCAAGCCGGTTGACTGCTC-TCTGATTCAGGCC-3Ј (antisense). The PCR product was digested with the restriction enzyme EcoRI and ligated into the pSR␣ vector (pSR␣/CD3⑀-VSV). The insert was extracted using EcoRI and ligated into the bicistronic pIRES2-EGFP expression vector (Clontech). This plasmid (pIRES2-EGFP/ CD3⑀) was grown in Escherichia coli DH5␣, purified, and sequenced. A cDNA coding for CD3⑀ with three acidic N-terminal residues (Asp 2 , Asp 3 , Glu 5 ) mutated to Gly by A-to-G substitutions in the relevant codons was obtained using the QuikChange site-directed mutagenesis kit (Stratagene) using pSR␣/CD3⑀-VSV as template and primers 5Ј-GGCACTT-GCCAGGGCGGTGCCGGGAACATTGGATACAAAGTC-TCC-3Ј (sense) and 5Ј-GGAGACTTTGTATCCAATGTT-CCCGGCACCGCCCTGGCAAGTGCC-3Ј (antisense). The plasmid containing the mutant CD3⑀ DNA was grown and sequenced as described above to confirm the mutations. After digestion with EcoRI, the insert was subcloned into the pIRES2-EGFP vector to generate the pIRES2-EGFP/CD3⑀MUT expression vector.
Cell Activation-For antigen activation, 10 4 T cells were cultured in Click's medium supplemented with 10% inactivated fetal calf serum (culture medium) in 96-well plates (Costar) with mitomycin C-treated 10 5 B10.BR spleen cells as antigen-presenting cells plus the indicated concentrations of conalbumin-(134 -146) peptide. Antigen-presenting cells were previously depleted of T cells by incubation with anti-CD90 antibody and rabbit complement. After 72 h of incubation at 37°C and 5% CO 2 , proliferation was determined by a colorimetric assay using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide as described (38). For activation with pervanadate, a pervanadate solution was prepared by adding 5 l of 1 M H 2 O 2 to 1 ml of 5 mM NaVO 4 (pH 7.0) in culture medium that was left for 20 min at room temperature. Pervanadate was then added to cells in culture medium (10 l of pervanadate solution/100 l of cells at 10 8 cells/ml). After 5 min of incubation at 37°C in a water bath, the reaction was stopped with ice-cold phosphate-buffered saline (PBS), 0.5 mM EDTA, and 1 mM NaVO 4 .
To determine early activation events in SR.D10 cells, anti-TCR antibody 3D3 or a control antibody was adsorbed (5 g/ml) to polystyrene microbeads. These beads were mixed with cells (10 7 in 100 l of culture medium/sample) at a 1:1 ratio as described (32). Cells were incubated for 5 min at 37°C in a water bath, and the reaction was stopped with ice-cold PBS, 0.5 mM EDTA, and 1 mM NaVO 4 . For intracellular detection of IFN-␥ in AE103 transfectants, cells were incubated at 37°C in 24-well culture plates precoated with anti-TCR antibody F23.1 (10 g/ml, 10 6 cells/ml of culture medium, 1 ml/well, in the presence of 10 g/ml anti-CD28 antibody). After 90 min of incubation at 37°C, brefeldin (10 g/ml) was added. For TCR-independent activation, the cells were stimulated with phorbol 12-myristate 13-acetate (20 ng/ml) plus 1 M ionomycin.
Immunoprecipitation, Isoelectric Focusing (IEF), and Immunoblotting-Immunoprecipitations of cell-surface TCR or CD3 were performed with 2 ϫ 10 7 cells/determination as described (18). Briefly, cells were incubated with antibodies in cold PBS containing 2% inactivated fetal calf serum and 0.1% sodium azide (staining buffer). Unbound antibodies were washed, and the cells were lysed on ice for 30 min with lysis buffer (1 ml/10 7 cells; 10 mM CHAPS in 50 mM Tris-HCl and 150 mM NaCl (pH 7.6) containing 1 mM MgCl 2 , 1 mM EGTA, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 1 mM NaVO 4 ). After centrifugation, postnuclear lysates were precleared by rotation with an irrelevant antibody coupled to Sepharose and centrifuged, and immunoprecipitation was carried out for 2 h at 4°C by rotation of the precleared supernatants with affinity-purified rabbit antimouse or anti-rat antibodies coupled to Sepharose as appropriate. The immunoprecipitates were eventually washed five times with cold wash buffer (2 mM CHAPS in 50 mM Tris-HCl and 150 mM NaCl (pH 7.6) containing 1 mM MgCl 2 , 1 mM EGTA, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 1 mM NaVO 4 ) and extracted with SDS-PAGE sample buffer. Immunoprecipitations from cell lysates were performed as above, except that precipitation was per-formed with anti-rat antibodies coupled to Sepharose beads previously incubated with rat anti-CD3 antibody YCD3-1 (3 g/sample) and washed.
Immunoblotting of immunoprecipitates and lysates was performed as described (32) using horseradish peroxidaseconjugated protein A for immunoprecipitates, horseradish peroxidase-conjugated ExtrAvidin for biotinylated samples, and horseradish peroxidase-conjugated anti-rabbit or antimouse IgG for cell lysates.
Surface and Intracellular Staining and Flow Cytometry-For surface staining, cells (0.5 ϫ 10 6 /sample in 0.05 ml of cold staining buffer) were incubated for 30 min on ice with biotinylated antibodies. After washing with staining buffer, the cells were incubated for an additional 30 min on ice with phycoerythrinconjugated streptavidin in staining buffer plus fluorescein isothiocyanate-or DyLight 649-coupled antibodies as indicated. The cells were washed and analyzed with a FACScan flow cytometer (Coulter Electronics). Three-color analysis of thymus populations was performed with a FACSCalibur flow cytometer (BD Biosciences) with CellQuest acquisition and analysis software.
For intracellular staining, cells (2 ϫ 10 6 ) were washed with cold PBS and fixed for 5 min with 0.25 ml of 4% paraformaldehyde in PBS at room temperature. After adding 1 ml of PBS, 0.1% bovine serum albumin, 0.05% sodium azide, 1 mM MgSO 4 , 1 mM CaCl 2 , and 10 mM HEPES (pH 7.2) (PBS/bovine serum albumin), the cells were spun and frozen at Ϫ70°C in PBS and 10% Me 2 SO until used. The cells were washed with 0.1% saponin (Sigma) in PBS/bovine serum albumin (PBS/saponin) and blocked at 4°C for 30 min with 5% nonfat milk in PBS/saponin. The cells were then stained with phycoerythrin-conjugated anti-mouse IFN-␥ antibody XMG1.2 in 5% nonfat milk in PBS/saponin for 30 in the cold. After washing three times with PBS/saponin, the cells were analyzed as described above.

CD3⑀ Isoforms with Different pI Values Are Detected in TCR-CD3
Complexes on the Surface of CD4 ϩ T cells-Our previous data suggested that one anti-CD3 antibody (YCD3-1) has different avidity for CD3 expressed by different CD4 ϩ T cells (18). Particularly, it binds with high avidity to CD3 expressed by the SR.D10 cell line, whereas its avidity is low for CD3 in AE103 cells. In contrast, other anti-CD3 antibodies (i.e. 500A2) or anti-TCR antibodies show no major differences in avidity for the same cells (18). High avidity of antibody YCD3-1 correlates with higher N-terminal degradation by metalloproteinases as determined using anti-CD3⑀ N-terminal peptide antibodies and proteinase inhibitors such as phenanthroline. CD3⑀ chains from different species, including mouse, possess negatively charged amino acid residues in their N-terminal sequence that affect their predicted pI values (Fig. 1A). We reasoned that CD3⑀ immunoprecipitates from different mouse T cell lines and/or using different anti-CD3 or anti-TCR antibodies should show distinct patterns of CD3⑀ chain isoforms, which could be distinguished by IEF.
Because antibody 500A2 showed little variation among T cells in terms of avidity, immunoprecipitation of CD3 with antibody 500A2 was initially used as a mean to assess the CD3⑀ isoform distribution in AE103 and SR.D10 cells (Fig. 1, B and C, upper panels). Immunoprecipitates were separated by IEF (first dimension) and PAGE (second dimension) and immunoblotted using rabbit anti-mouse CD3⑀ extracellular domain antibodies. As predicted, different CD3⑀ isoforms were present within the pI range for complete chains or chains lacking one or two aspartic acid residues of the N-terminal sequence. The most abundant species in both cell lines had a pI close to 7, fitting the predicted pI for CD3⑀ chains lacking one aspartic acid. However, at a low pI, complete CD3⑀ chains were abundant in AE103 cells, but almost absent in SR.D10 precipitates (Fig. 1, B and C, upper panels).
Immunoprecipitation of surface CD3⑀ using antibody YCD3-1 showed that this antibody bound better to isoforms with higher pI values, i.e. those lacking one or two N-terminal aspartic acids. Thus, these isoforms are over-represented in YCD3-1 immunoprecipitates from either AE103 or SR.D10 cells (Fig. 1, B and C; see also Fig. 2A), in agreement with the predicted behavior of this antibody based on our previous data (18).
Our previous results also suggested that the less degraded CD3⑀ chains maintain strong ties with the TCR, whereas TCR-CD3 bonds are looser for degraded CD3⑀ chains (18). This was confirmed in immunoprecipitates of surface TCR from SR.D10 cells. Indeed, as depicted in Fig. 1C (lower panel), despite the low proportion of complete low pI CD3⑀ chains among CD3 chains from SR.D10 cells observed in anti-CD3 precipitates (upper and middle panels), these low pI chains coprecipitated very efficiently with the TCR. Fig. 1D shows that the different isoforms detected in the cell lines could also be found in precipitates of surface CD3 from freshly isolated T cells from normal mouse spleen, indicating that the presence of CD3⑀ chains with varying pI values is not a phenomenon peculiar to selected cells cultured in vitro. Immunoblotting with anti-CD3⑀ antibody revealed CD3⑀ with distinct pI values. In addition, anti-CD3 antibody YCD3-1 preferentially precipitated CD3⑀ chains with higher pI values than did antibody 500A2. C, shown is a comparison of CD3⑀ chains from surface TCR-CD3 complexes immunoprecipitated from SR.D10 cells with anti-CD3 or anti-TCR antibody. Immunoprecipitates using anti-CD3 antibody 500A2 or YCD3-1 or anti-TCR F23.1 antibody, as indicated, were separated and immunoblotted as described for B. D, immunoprecipitation of CD3 from the surface of TCR-CD3 complexes of spleen T lymphocytes performed using antibody YCD3-1 showed that CD3⑀ chains with different pI values are present in normal spleen T lymphocytes. pI CD3⑀ chain isoforms in immunoprecipitates were separated and detected as described for B. All results shown are representative of at least two different experiments.

Structural and Functional Consequences of CD3⑀ Isoform
Variation-The differences in CD3⑀ isoform content among cells might likely influence the overall structure of the TCR-CD3 complex. It has been shown that ligand-induced changes in the TCR can be reflected as differences in the accessibility of small ligands such as biotin to TCR-CD3 polypeptides (40). Consequently, we checked for possible differences between SR.D10 and AE103 cells in the exposure of CD3⑀ by determining the efficiency of CD3⑀ biotinylation in each cell line ( Fig.  2A). Biotinylation of CD3⑀ chains from AE103 cells was comparatively better than that of CD3⑀ chains from SR.D10 cells. This seemed a specific phenomenon, as biotinylation of an unrelated surface polypeptide (CD4) was comparatively weaker in AE103 cells (Fig. 2, A and B).
As shown in Fig. 1 and Ref. 18, monoclonal antibody YCD3-1 binds surface CD3⑀ isoforms lacking one or two N-terminal amino acid residues better than other anti-CD3 antibodies. Thus, the avidity of antibody YCD3-1 for a cell line can be taken as an indication of the relative content of high pI CD3⑀ isoforms in that line. To analyze whether this might influence TCR signaling, we also compared the antigen response of three different cell lines with the same antigen specificity but different avidity for YCD3-1 binding (18). As shown in Fig. 2C, there was a strong correlation between the ability of antibody YCD3-1 to bind to the cells and the efficiency of antigen peptide to activate the cells.
The Proteinase Inhibitor Phenanthroline Alters the Ratio of CD3⑀ Isoforms-Our previous data suggested that metalloproteinases sensitive to inhibition by 1,10-phenanthroline are involved in N-terminal degradation of CD3⑀ chains (18). Consequently, we predicted that treatment of cells with 1,10-phenanthroline should enrich the proportion of low pI CD3⑀ chains. These experiments were carried out with SR.D10 cells, a cell line with a high proportion of high pI CD3⑀ isoforms. Analysis of CD3 immunoprecipitates from the surface of SR.D10 cells revealed that phenanthroline-treated cells indeed had clearly lower amounts of high pI CD3⑀ chains (i.e. pI 7.8) than did untreated cells compared with the amount of pI 6.91 chains in each cell (Fig. 3A). This effect was also observed in immunoprecipitates of CD3 from total cell lysates (data not shown). Intriguingly, 1,10-phenanthroline did not prevent the degradation of low pI CD3⑀ chains (i.e. chains of pI Ͻ7.0), suggesting that there are proteinase(s) mediating the first steps of N-terminal degradation that are different and insensitive to inhibition by phenanthroline.
The role of CD3⑀ N-terminal charged amino acids in TCRmediated activation was then studied by analyzing the effect of phenanthroline. Interestingly, incubation with phenanthroline was sufficient to alter the early steps of activation induced by clonotypic anti-TCR antibodies (Fig. 3, B-D). Thus, phenanthroline treatment inhibited early Tyr phosphorylation of ZAP-70 in cells activated by anti-TCR antibody, whereas the effect was clearly lower if a TCR-independent stimulus such as pervanadate was used (Fig. 3B). A marked inhibition by phenanthroline of TCR activation of downstream pathways such as the ERK mitogen-activated protein was also observed (Fig. 3C,  lanes 1 and 2). This is in contrast with its modest effect on ERK FIGURE 2. Differences in biotinylation of CD3⑀ and antigen activation among different T cell lines. A, cells from lines SR.D10 and AE103 were surface-biotinylated and immunoprecipitated with anti-CD3 antibody 500A2 (left panels) or anti-CD4 antibody GK1.5 (right panels) as a control. Biotinylation of the precipitated surface CD3⑀ and CD4 polypeptides was assessed by blotting with streptavidin (upper panels), and the relative protein load was determined by immunoblotting using rabbit anti-CD3⑀ and anti-CD4 antibodies (lower panels). B, the relative intensity of biotinylation of CD3⑀ in AE103 cells was clearly stronger than that in SR.D10 cells. The intensity of the signal was corrected for protein load by determining the ratio of the absorbance of biotinylated polypeptides to the absorbance of specific immunoblots. The same determination in CD4 showed that the enhanced biotinylation of CD3⑀ was specific to this particular surface polypeptide. The absorbance ratio for SR.D10 cells was considered as 1 in each case. C, a higher content of high pI CD3⑀ isoforms correlated with a lower antigen activation threshold in three Th2 cell lines (SR.D10, AK-8, and D10.TCR2.3) of the same pMHC specificity. The concentration of antibody YCD3-1 needed to achieve 20% of maximal binding was taken as an indication of the CD3⑀ isoform content (i.e. a lower concentration of antibody YCD3-1 indicates a higher proportion of high pI isoforms; see Fig. 1 and Ref. 18). The concentration (micromolar) of antigen (Ag) peptide needed for 50% maximal proliferation (see "Experimental Procedures") was considered for antigen sensitivity. Representative results of two independent experiments are shown. activation induced by pervanadate (Fig. 3C, lanes 5 and 6) or phorbol esters plus ionomycin (data not shown).

The N-terminal Negatively Charged Residues of CD3⑀ Control Recognition by Antibody YCD3-1 and TCR-mediated
Activation-To further analyze the role of CD3⑀ N-terminal aspartic and glutamic acid residues in TCR-CD3 structure and TCR-mediated activation, we analyzed the behavior of cells expressing CD3⑀ constructs in which Glu 2 , Glu 3 , and Asp 5 were mutated to Gly. For this analysis, we chose AE103 cells, which, according to our previous data (18), show the lowest fraction of high pI CD3⑀ isoforms. The cells were transfected with bicistronic vectors coding for enhanced green fluorescent protein (EGFP) alone, for EGFP and wild-type CD3⑀ (pIRES2-EGFP/ CD3⑀), or EGFP and mutant CD3⑀ chains (pIRES2-EGFP/ CD3⑀MUT). We observed that the level of expression of the EGFP fusion protein was similar in all transfected cells, although transfected wild-type cells (pIRES2-EGFP/CD3⑀) expressed slightly higher amounts of EGFP as determined by flow cytometry (Fig. 4A).
According to our previous results ( Figs. 1 and 3) (18), we expected that AE103 cells expressing mutant CD3⑀ would mimic the behavior of cell lines expressing high amounts of degraded high pI CD3⑀ isoforms, such as SR.D10. Accordingly, expression of mutant CD3⑀ should enhance the avidity of antibody YCD3-1 for cells transfected with the pIRES2-EGFP/ CD3⑀MUT vector. This was indeed the case (Fig. 4B), so the concentration of antibody YCD3-1 needed to achieve 50% of maximal binding was ϳ3-fold lower in the mutant CD3⑀ transfectants (Fig. 4C). These changes were not observed when the avidity of anti-TCR antibody F23.1 was determined in the same cells (Fig. 4, B and C). These data are in agreement with the results from IEF analysis of CD3⑀ chains from the transfected cells. Thus, cells transfected with pIRES2-EGFP/CD3⑀ had a high amount of non-degraded low pI CD3⑀ chains (Fig. 4D,  upper panel, box a). On the other hand, cells transfected with pIRES2-EGFP/CD3⑀MUT lacking negatively charged residues had enhanced amounts of high pI CD3⑀ chains (Fig. 4D, lower  panel, box c). These differences were also reflected in the absorbance ratios for these isoforms in the transfectants, such that box a/b ratios were 1.02 and 0.56 for wild-type and mutant CD3⑀ cells, respectively, and box a/c ratios were 0.88 and 0.31 for these same cells.
The role of CD3⑀ charged residues in modulating TCR responses was analyzed in cells expressing exogenous wild-type or mutant CD3⑀ chains. Fig. 5 shows that, in agreement with our prediction, expression of mutant CD3⑀ clearly enhanced FIGURE 3. Effect of the metalloproteinase inhibitor 1,10-phenanthroline on CD3⑀ isoforms and TCR activation. A, SR.D10 cells were treated or not with 0.1 mM 1,10-phenanthroline for 16 h, as indicated, and surface CD3 (antibody YCD3-1) was immunoprecipitated (IP), separated by IEF and PAGE, and immunoblotted with anti-CD3⑀ antibodies. CD3⑀ isoforms with pI 6.91 and 7.8 are framed by boxes a and b, respectively, showing that isoform b was reduced in immunoprecipitates from phenanthroline-treated cells. Results are representative of three different experiments. The absorbance ratio between the CD3⑀ isoforms with pI 6.91 (box a) and 7.8 (box b) is indicated and shows that the relative abundance of isoform a was increased in phenanthroline-treated cells compared with untreated control cells. B and C, 1,10-phenanthroline inhibited early TCR-mediated ZAP-70 and ERK activation, respectively. SR.D10 cells pretreated or not with 1,10-phenanthroline were activated for 5 min with clonotypic anti-TCR antibody 3D3 (TCR; 5 g/ml) adsorbed to polystyrene microspheres or with a control antibody (Control). As a positive TCR-independent activation control, the cells were activated with pervanadate (PV). Lanes 1 and 2 (marked as *), 30-min exposure; lanes 3-6, 1-min exposure. P-Tyr, phosphotyrosine; P-ERK, phospho-ERK. D, absorbance was normalized for protein load of phospho-ZAP-70 and phospho-ERK. The absorbance of phosphorylated proteins in untreated control cells was considered 1 in each case. Data are from one experiment of two performed with similar results. the number of IFN-␥-producing cells and the level of cytokines produced by each cell when the cells were activated by platebound anti-TCR antibody in the presence of anti-CD28 antibody (Fig. 5). In contrast, these parameters were not enhanced, but were lower in the mutant CD3⑀ transfectants activated by a TCR-independent stimulus such as phorbol 12-myristate 13-acetate plus ionomycin (Fig. 5).
Differences in CD3⑀ Isoform Expression among Thymocyte Subpopulations-As shown above, the avidity of antibody YCD3-1 for CD3 varies depending on the relative abundance of distinct CD3⑀ isoforms, viz. it is higher as the fraction of high pI CD3⑀ increases. To determine whether different populations of normal T cells have different CD3⑀ isoform profiles, different T cell subsets from the thymus (Fig. 6A) or the spleen (data not shown) were selected on the basis of their expression of the CD4 and CD8 coreceptors and stained with anti-TCR or anti-CD3 antibodies. Abundant immature CD4 ϩ CD8 ϩ thymocytes expressed low levels of TCR-CD3 on their surface (Fig. 6B), and antibody YCD3-1 bound with a relatively low avidity to these cells (Fig. 6C). A small subpopulation of CD4 ϩ CD8 ϩ thymocytes expressing high TCR-CD3 levels (double-positive (DP) CD4 ϩ CD8 ϩ TCR hi ) is a more mature population of DP cells. Interestingly, when DP CD4 ϩ CD8 ϩ TCR hi thymocytes (ϳ5% of all DP CD4 ϩ CD8 ϩ cells) were analyzed, we observed high avidity binding of antibody YCD3-1 to these cells, a feature linked to lower TCR activation thresholds (Fig. 6, B and C). The avidity of YCD3-1 binding was lower in the mature single-positive (SP) CD4 ϩ cells, which are presumed to be the prime target for negative selection in vivo (41,42), and was even lower in the SP CD8 ϩ thymocytes (Fig. 6, B and C). In contrast, the avidity of anti-TCR antibodies was not different among the different subsets, even though there were large differences in the level of TCR expression (Fig. 6, B and C). The avidity of anti-TCR or anti-CD3 antibodies for spleen CD4 ϩ and CD8 ϩ T lymphocytes was similar to that observed in the SP thymus populations (data not shown).

DISCUSSION
Differences among TCR-CD3 complexes expressed by normal T cells are of interest, as they might quantitatively and qualitatively regulate TCR signals under physiological conditions. Our data show that TCR-CD3 complexes from mouse T cells and T cell lines contain different CD3⑀ chain isoforms that can be separated by their isoelectric point. Their pI values fit those predicted for mouse CD3⑀ chains lacking one or two N-terminal negatively charged amino acid residues. Data from this and our previous study (18) show that this phenomenon is due at least in part to degradation of the N-terminal sequence by metalloproteinases sensitive to inhibition by phenanthroline. In retrospect, the existence of different molecular species  of CD3⑀ is a likely reason for the failure of early attempts to determine the N-terminal sequence of CD3⑀ from purified protein, which were successful in establishing the N-terminal sequence of CD3␦ (43).
The N-terminal region of CD3⑀ has variable sequence and length in different species, yet a common characteristic of all known CD3 chains is the presence of negatively charged amino acids in this region (18). The N-terminal sequence of CD3⑀ might be flexible and does not have a defined position within the structures of the CD3⑀␥ and CD3⑀␦ dimers defined so far (11)(12)(13)(14). Although this region is not directly involved in the formation of CD3⑀␥ and CD3⑀␦ dimers (11)(12)(13)(14), we have shown that it modulates the interaction between CD3 dimers and the TCR heterodimer, as loss of N-terminal negative charges of CD3⑀ chains leads to weaker interaction of CD3 dimers with the TCR (Figs. 1 and 4) (18). These differences have distinct structural consequences, as exemplified by the different biotinylation of CD3⑀ chains from cell lines with different proportions of CD3⑀ isoforms ( Fig. 2A) or the ability of certain anti-CD3 antibodies such as YCD3-1 to bind with higher avidity to T cells enriched in high pI isoforms (18) or expressing mutant CD3⑀ chains lacking the N-terminal charges (Fig. 4).
The structural relationships among chains within the TCR-CD3 complex have been the subject of intense study (see Refs. 2, 44, and 45 for recent reviews). It seems that CD3⑀␦ dimers interact preferentially (3, 4, 46 -49), although not exclusively (50), with TCR␣ chains. Charged amino acids within the transmembrane domains of TCR␣ and CD3 and interactions between a sequence motif present in the stalk connecting the transmembrane and Ig domains of TCR␣ and an undetermined region of the CD3⑀␦ dimers are involved in this interaction. Furthermore, recent data indicate that the TCR C␣ D-E loop is also involved in TCR-CD3⑀␦ interactions (51). FIGURE 6. Anti-CD3 antibody YCD3-1 has a different avidity for different thymus populations. A, cells from BALB/c mouse thymus were stained with anti-CD4 (stains red) and anti-CD8 (stains green) antibodies to distinguish DP CD4 ϩ CD8 ϩ thymocytes, SP CD4 ϩ cells, and SP CD8 ϩ cells. PE, phycoerythrin; FITC, fluorescein isothiocyanate. B, TCR and CD3 staining with anti-TCR antibody F23.1 or anti-CD3 antibody YCD3-1 coupled with DyLight 649 was performed, and TCR or CD3 staining of DP CD4 ϩ CD8 ϩ cells (E), SP CD4 ϩ cells (Ⅺ), or SP CD8 ϩ cells (f) was determined. In addition, the staining of DP CD4 ϩ CD8 ϩ TCR hi cells (F) was also determined. C, the concentrations of anti-TCR antibody (Ab) F23.1 (left panel) and anti-CD3 antibody YCD3-1 (right panel) needed to achieve 50% of the maximal binding in each subpopulation were extrapolated from the data in B. Data are from one experiment of two performed with similar results.
On the other hand, available immunochemical (52), mutation (51,53), and in vitro assembly (4) data indicate that the CD3⑀␥ dimers preferentially interact with TCR␤ chains, possibly close to the "cave" beneath the TCR C␤ F-G loop, which is sided by the Lys-containing A-B loop and the glycan of Asn 185 in the TCR C␣ domain (53)(54)(55). An alternative site encompassing the TCR C␤ CCЈ loop and close to CD3⑀␦ has been also suggested based on TCR C␤ domain mutation analysis (51).
The size of CD3 dimers, their rigid nature, and the known location of anti-CD3 antibody-binding sites and N-glycosylation sites in CD3␦ and CD3␥ all point to the ABE face of CD3⑀ as a likely region of interaction with the TCR unit (11)(12)(13)(14). This would also place the N-terminal sequence of CD3⑀ analyzed here in a privileged position to regulate the interactions between the TCR and the CD3 dimers.
Analysis of CD3 mutants or hybrid proteins has been very useful in uncovering the role of specific regions in the interactions between different polypeptides within the TCR-CD3 complex and/or in the induction of specific signals. However, only those alterations in TCR-CD3 elements that are found under physiological conditions can be considered to play a functional role in cells expressing wild-type polypeptides. In addition to N-terminal degradation of CD3⑀, changes in the carbohydrate content of different chains might have an impact on the structure of the complex in ␥␦ or ␣␤ T cells (16,20,56).
T cells possess a distinctive capacity to discriminate closely related cognate peptide-MHC (pMHC) ligands. This fine discrimination capacity has been explained by two main kinds of models. Kinetic proofreading models (57)(58)(59) propose that T cells discriminate the quality of ligands by the strength and quality of TCR signals, which are dependent on the time of occupancy of the TCR. In turn, this is determined mainly by the dissociation rate of the cognate pMHC ligands (60). It has been shown that small differences in pMHC affinity can translate into large differences in activation of important intracellular activation pathways such as the ERK cascade, acting as an onoff signal at the cellular level (61).
Conformational change models (21,(62)(63)(64) are based on the assumption that the TCR can discriminate ligands on the basis of their ability (agonists) or lack thereof (non-agonists) to induce specific conformational changes in the TCR-CD3 complex. It is well established by x-ray crystallography that pMHC binding induces conformational changes in the interacting surface of the TCR␣␤ ectodomains, but with one exception (65), these changes do not translate to the constant domains of the TCR (reviewed in Ref. 44), pointing to cross-linking the TCR with enough affinity as the driving force for activation. These assumptions were challenged by the finding of a conformational change in the intracellular domain of CD3⑀ chains produced by strong stimulatory antibodies (21). This change was detected by the exposure of a cryptic polyproline sequence that could then be bound by the SH3.1 (Src homology 3.1) domain of the Nck adapter protein (21) or by one antibody (APA-1) specific for this region of CD3⑀ (66). This conformational change is a very early activation event, taking place independently and before any detectable activity of tyrosine kinases (21,66), and can discriminate between weak or strong agonist pMHC (66,67). These data has been recently put together with the finding of the presence of varying amounts of multivalent TCR-CD3 complexes on the surface of lymphocytes (7) to propose a model in which the affinity of interaction with cognate ligands increases exponentially with multivalency to reach a proofreading threshold that produces conformational changes in the complex (68). Furthermore, the conformational change in CD3⑀ required for full activation needs at least dimerization of TCRs in a given permissive geometry (69). This process would not require changes in the structure of the TCR ectodomains, but the rearrangement of the extracellular and transmembrane domains of the complex upon TCR ligation (69).
Because differences in the CD3⑀ N-terminal sequence influence the strength of CD3-TCR interactions, we proposed previously that loss of N-terminal charged residues in CD3⑀ would result in a looser TCR-CD3 interaction, facilitating ligand-induced rearrangement of the different elements within the complex (18). The fact that differences in the CD3 N-terminal chain sequence also produce alterations in the ability of TCR to deliver efficient intracellular signals might help in understanding the different sensitivity of a given TCR to single pMHC ligands in mature and immature cells (66, 70 -72). Thus, cognate pMHC ligands that are nonagonists or partial agonists for mature T cells in the periphery can deliver functionally meaningful signals in thymocytes (66, 70 -72). Consequently, altering CD3⑀ N-terminal charges might be a simple mechanism to fine-tune activation thresholds of the cell. In this regard, it should be noted that, whereas changes during ontogeny in the abundance of certain cytoplasmic proteins (i.e. phosphatases) could regulate TCR activation thresholds to a given ligand in terms of kinase activation (61,72), it is difficult to conceive such a mechanism in the case of partial agonist pMHC ligands that do not induce conformational changes in CD3⑀ of mature T cells (66), but can do so in immature DP thymocytes (70), as the conformational change is very fast and independent of metabolic activity (21,69). Our finding of marked differences in the CD3 isoform pattern of thymocyte subpopulations potentially involved in positive or negative selection might help in understanding some of these issues, as these differences might be indicative of changes in their activation threshold.
In summary, we have observed that mouse CD3⑀ chains naturally lose negatively charged residues in their N-terminal region. This process is mediated by proteinases, varies among different T lymphocytes, and has an impact on the interactions between the chains of the TCR-CD3 complex and the cells' ability to deliver TCR activation signals. We have preliminary data on similar pI heterogeneity in CD3⑀ from human T cells, 5 and the presence of negatively charged residues in the same region of all known CD3⑀ chains suggests that this might be a widespread process. The exact nature of the proteinase(s) involved and the possibility that proteinases activated in inflammatory processes could modulate the activation threshold of T cells recruited into inflammation sites by altering CD3⑀ chains should be investigated.