Biochemical Differences in the αβ T Cell Receptor·CD3 Surface Complex between CD8+ and CD4+ Human Mature T Lymphocytes

We have reported the existence of biochemical and conformational differences in the αβ T cell receptor (TCR) complex between CD4+ and CD8+ CD3γ-deficient (γ-) mature T cells. In the present study, we have furthered our understanding and extended the observations to primary T lymphocytes from normal (γ+) individuals. Surface TCR·CD3 components from CD4+ γ- T cells, other than CD3γ, were detectable and similar in size to CD4+ γ+ controls. Their native TCR·CD3 complex was also similar to CD4+ γ+ controls, except for an αβ(δϵ)2ζ2 instead of an αβγϵδϵζ2 stoichiometry. In contrast, the surface TCRα, TCRβ, and CD3δ chains of CD8+ γ- T cells did not possess their usual sizes. Using confocal immunofluorescence, TCRα was hardly detectable in CD8+ γ- T cells. Blue native gels (BN-PAGE) demonstrated the existence of a heterogeneous population of TCR·CD3 in these cells. Using primary peripheral blood T lymphocytes from normal (γ+) donors, we performed a broad epitopic scan. In contrast to all other TCR·CD3-specific monoclonal antibodies, RW2-8C8 stained CD8+ better than it did CD4+ T cells, and the difference was dependent on glycosylation of the TCR·CD3 complex but independent of T cell activation or differentiation. RW2-8C8 staining of CD8+ T cells was shown to be more dependent on lipid raft integrity than that of CD4+ T cells. Finally, immunoprecipitation studies on purified primary CD4+ and CD8+ T cells revealed the existence of TCR glycosylation differences between the two. Collectively, these results are consistent with the existence of conformational or topological lineage-specific differences in the TCR·CD3 from CD4+ and CD8+ wild type T cells. The differences may be relevant for cis interactions during antigen recognition and signal transduction.

We have reported the existence of biochemical and conformational differences in the ␣␤ T cell receptor (TCR) complex between CD4 ؉ and CD8 ؉ CD3␥-deficient (␥ ؊ ) mature T cells. In the present study, we have furthered our understanding and extended the observations to primary T lymphocytes from normal (␥ ؉ ) individuals. Surface TCR⅐CD3 components from CD4 ؉ ␥ ؊ T cells, other than CD3␥, were detectable and similar in size to CD4 ؉ ␥ ؉ controls. Their native TCR⅐CD3 complex was also similar to CD4 ؉ ␥ ؉ controls, except for an ␣␤(␦⑀) 2 2 instead of an ␣␤␥⑀␦⑀ 2 stoichiometry. In contrast, the surface TCR␣, TCR␤, and CD3␦ chains of CD8 ؉ ␥ ؊ T cells did not possess their usual sizes. Using confocal immunofluorescence, TCR␣ was hardly detectable in CD8 ؉ ␥ ؊ T cells. Blue native gels (BN-PAGE) demonstrated the existence of a heterogeneous population of TCR⅐CD3 in these cells. Using primary peripheral blood T lymphocytes from normal (␥ ؉ ) donors, we performed a broad epitopic scan. In contrast to all other TCR⅐CD3-specific monoclonal antibodies, RW2-8C8 stained CD8 ؉ better than it did CD4 ؉ T cells, and the difference was dependent on glycosylation of the TCR⅐CD3 complex but independent of T cell activation or differentiation. RW2-8C8 staining of CD8 ؉ T cells was shown to be more dependent on lipid raft integrity than that of CD4 ؉ T cells. Finally, immunoprecipitation studies on purified primary CD4 ؉ and CD8 ؉ T cells revealed the existence of TCR glycosylation differences between the two. Collectively, these results are consistent with the existence of conformational or topological lineage-specific differences in the TCR⅐CD3 from CD4 ؉ and CD8 ؉ wild type T cells. The differences may be relevant for cis interactions during antigen recognition and signal transduction.
␣␤ T lymphocytes recognize peptide-major histocompatibility complex ligands by means of a multimeric protein complex termed the ␣␤ T cell receptor (TCR) 1 CD3 complex (TCR⅐CD3). This structure is composed of a variable ␣␤ TCR dimer that binds antigens and three invariant dimers (CD3␥⑀, ␦⑀, and ) that are in charge of TCR⅐CD3 complex transport, stabilization, and signal transduction (1). The minimum stoichiometry, therefore, is believed to be ␣␤␥⑀␦⑀ 2 .
Mature CD4 ϩ and CD8 ϩ ␣␤ T cells differ sharply in their major histocompatibility complex ligands, but their TCR⅐CD3 complex is believed to be qualitatively identical. The reduced ␣␤ TCR⅐CD3 staining levels observed in CD8 ϩ T cells, relative to CD4 ϩ T cells, were therefore reported as quantitative under this assumption (2). Unexpectedly, peripheral blood ␣␤ TCR⅐CD3 expression was shown to be more impaired in CD8 ϩ than in CD4 ϩ cells when CD3␥ (3,4) or CD3␦ (5) was absent. These observations were followed by the description of conformational and biochemical differences in the TCR⅐CD3 complex between CD8 ϩ and CD4 ϩ CD3␥ deficient (␥ Ϫ ) T lymphocytes (6). Biosynthetic studies showed that CD8 ϩ but not CD4 ϩ ␥ Ϫ T cells lacked normal TCR␣. Instead, the CD4 ϩ ␥ Ϫ T cells contained a small ␣␤ heterodimer composed of abnormally glycosylated TCR␤ and an abnormally small CD3-associated chain that was not recognized by TCR␣-specific antibodies.
In the present study, we have extended these biochemical studies to the cell surface and provide further evidence for the existence of qualitative differences in the ␣␤ TCR⅐CD3 complex between CD8 ϩ and CD4 ϩ T lymphocytes, particularly when CD3␥ is lacking, but also in normal T cells.
Human PBLs were obtained with informed consent from normal donors. Purified CD4 ϩ and CD8 ϩ T cells were isolated immunomagnetically using standard procedures according to manufacturer's instructions (Miltenyi Biotec, Bergisch Gladbach, Germany). Purities Ͼ90% were the rule. Human postnatal thymocytes were isolated from thymus fragments removed during corrective cardiac surgery of patients aged 1 month to 4 years with informed parental consent.
TCR Labeling and Immunoprecipitation-Surface labeling ( 125 I and * This work was supported by grants from the Ministerio de Ciencia y Tecnología (BMC2002-3247) and the Comunidad Autónoma de Madrid (21/01) (to J. R. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Endo-␤-N-acetylglucosaminidase H and N-glycosidase F (Roche Applied Science) treatments were done as described (6,7). Finally, samples were resuspended in Laemmli sample buffer and boiled for 5 min before a short spin at 12,000 ϫ g. SDS-PAGE was performed on 10% polyacrylamide gels and analyzed by autoradiography.
Confocal Microscopy-5 ϫ 10 6 cells were washed twice in phosphatebuffered saline/1% bovine serum albumin and adjusted to a final concentration of 2.5 ϫ 10 5 /ml in the same solution. 2.5 ϫ 10 4 cells were dropped on non-treated glass slides by cytocentrifugation at 400 rpm for 3-4 min in a Cytospin 3 cytocentrifuge (Shandon, Pittsburgh, PA). Samples were air-dried and then fixed in acetone at room temperature for 5 min. Slides were again air-dried and subsequently stained with the corresponding mAb. Briefly, 100 l of either TCR␣-or TCR␤-specific mAb (1:5 dilution) were dispensed onto the fixed cells, and the slides were incubated at room temperature in a wet box for 40 min. Samples were washed twice in phosphate-buffered saline, dried carefully, and stained with a mouse IgG-specific Cy5-conjugated secondary mAb (100 l, 1:200 dilution) for another 40 min in a wet box. After two more washes with phosphate-buffered saline, samples were dried and stained (40 min, wet box) with either CD4-or CD8␣-specific fluorescein isothiocyanate-conjugated mAb from Pharmingen (100 l, 1:2 dilution). After two final washes, a drop of 1,4-diazabicyclo[2.2.2]octane (DABCO) mounting medium was added to the slides, and cells were analyzed under a Bio-Rad MRC-1024 confocal microscope.
NANAse (Sigma) treatments were done as described (12). NANAse effects were controlled externally by the increased WT31 binding of a ␥␦ ϩ T cell line termed Peer (data not shown) and internally by the decreased binding of a CD43-specific antibody, as described (13).
Blue Native (BN)-PAGE and Western Blotting-Published methods were used for the two-step affinity purification of the TCR (14). Briefly, cells were stimulated with pervanadate and lysed with 1% digitonin lysis buffer containing sodium orthovanadate. Phosphorylated proteins were purified with 1 g of 4G10 and 3 l of protein G-Sepharose, washed three times, and subsequently eluted in BN buffer (500 mM ⑀-aminocaproic acid, 20 mM NaCl, 10% glycerol, 2 mM EDTA, and 20 mM bis-Tris, pH 7) including 50 mM phenylphosphate, 1 unit of alkaline phosphatase, and 1% digitonin. Dephosphorylated TCRs were separated by BN-PAGE as described (15). Western blots were developed with the antiantiserum 448. Ferritin monomers (f1; 400 kDa), dimers (f2; 880 kDa), and trimers (f3; 1320 kDa) were included as molecular mass standards. Under these conditions, the normal dephosphorylated ␣␤␥⑀␦⑀ 2 TCR⅐CD3 runs at ϳ400 kDa (f1) despite the fact that the relative molecular masses of the isolated dimers add up to only 230 kDa (90 ϩ 50 ϩ 50 ϩ 40). Therefore, shape as well as molecular mass is important in BN-PAGE gels. For antibody-based gel shift experiments, samples were pre-incubated with ␤F1 (TCR␤) mAb or Fab fragments of SP34 (CD3⑀) before electrophoresis. A detailed description and analysis of the wild-type TCR by BN-PAGE has been submitted. 2

RESULTS
CD4 ϩ and CD8 ϩ CD3␥-deficient T Lymphocytes Express Biochemically Different Membrane ␣␤ TCR⅐CD3 Complexes-Natural CD3␥-deficient (␥ Ϫ ) T cells were biotinylated or radioiodinated, lysed, and immunoprecipitated with CD3-specific as well as with TCR-specific mAb and compared with normal controls (␥ ϩ ). The results showed that, in contrast to biosynthetic experiments, the amount of surface CD3-associated TCR (and ) proteins was very low in ␥ Ϫ cells (Fig. 1A). This precluded a meaningful analysis of the TCR composition of the complex by this method. We have previously reported by immunoprecipitation that surface CD3␦ from CD8 ϩ ␥ Ϫ cells showed an impaired electrophoretic mobility in comparison with controls because of its abnormal glycosylation (Fig. 1A,  right). This finding correlated with complete endo-␤-N-acetylglucosaminidase H resistance (6). Therefore, we next analyzed the CD3␦ from CD4 ϩ ␥ Ϫ cells. The results demonstrated that, in contrast to CD8 ϩ ␥ Ϫ cells, the CD3␦ from CD4 ϩ ␥ Ϫ cells was similar to that from controls, both in terms of relative mobility and endo-␤-N-acetylglucosaminidase H sensitivity (Fig. 1B). The direct immunoprecipitation of TCR chains showed that CD4 ϩ ␥ Ϫ cells were comparable with CD4 ϩ ␥ ϩ cells in terms of TCR chain composition, whereas CD8 ϩ ␥ Ϫ cells clearly lacked normal surface TCR␣ and TCR␤ chains (Fig. 2). These differences were TCR-specific, because control immunoprecipitates of major histocompatibility complex class I molecules were equivalent in all cells.
To further characterize the surface TCR chains of the mutant cells, immunofluorescence confocal analyses were performed using the same mAb. Expression of the CD4 or CD8 coreceptors was used as a positive or negative control in each cell type. The results indicated that both TCR␣ and TCR␤ chains showed normal (mostly surface) distribution and fluorescence intensities in CD4 ϩ ␥ Ϫ lymphocytes (Fig. 3). CD8 ϩ ␥ Ϫ cells also showed normal expression of TCR␤ by this assay, but TCR␣ was hardly detectable and, when present (10 -20% of cells), mostly intracellular. The discrepancy in TCR␤ expression by CD4 ϩ ␥ Ϫ cells between these data and the immunoprecipitation results (above) may be due to technical differences, such as the use of acetone for confocal analysis.
In a last set of experiments, the native surface TCR⅐CD3 complexes from CD4 ϩ and CD8 ϩ mutant cells were compared (14,15). The results indicated that the TCR⅐CD3 complex from CD4 ϩ ␥ Ϫ T lymphocytes was similar to normal complexes from ␥ ϩ T cells (CD4 ϩ , CD8 ϩ , or Jurkat), although with delayed electrophoretic mobility (Fig. 4A). This observation was compatible with either a different stoichiometry or a different structure (such as glycosylation or conformation changes) of the mutant complex (see "Experimental Procedures"). To distinguish between these two possibilities, antigenic shift experiments were performed. One shift indicates that there is a single chain per complex, and two shifts denote two binding sites per complex. The results indicated that the mutant TCR⅐CD3 complex had a normal stoichiometry when compared with that of ␥ ϩ T cells (Jurkat, Fig. 4B, or PBL, not shown). It contained one TCR␤ chain (hence a single ␣␤ heterodimer) and two CD3⑀ chains (probably representing two CD3␦⑀ dimers, because ␥ was absent). In summary, these experiments support an ␣␤(␦⑀) 2 2 model for the mutant TCR⅐CD3 complex in CD4 ϩ ␥ Ϫ cells, where the ␥⑀ dimer is replaced by the analogous ␦⑀ dimer. Therefore, other types of modifications, such as glyco-sylation or conformation, may explain the delayed migration of the mutant complex. In sharp contrast, CD8 ϩ ␥ Ϫ cells expressed a very different TCR⅐CD3 complex as compared with their CD4 ϩ ␥ Ϫ counterparts. Native TCR⅐CD3 complexes isolated from CD8 ϩ ␥ Ϫ cells migrated along a broad smear, suggesting the existence of a heterogeneous population of complexes in these cells (Fig. 4A). The stoichiometry of these heterogeneous TCR⅐CD3 complexes could not be determined by antigenic shift due to the basal broad smear.

FIG. 3. Confocal analysis of TCR␣ and TCR␤ expression by
CD3␥-deficient CD4 ؉ and CD8 ؉ T cells. 2.5 ϫ 10 4 CD4 ϩ or CD8 ϩ cells, both ␥ ϩ and ␥ Ϫ , were cytocentrifuged, acetone-fixed, and stained with purified TCR␣-and TCR␤-specific mAbs and Cy3-conjugated secondary antibodies. CD4-and CD8-fluorescein isothiocyanate stainings were included as a positive control. Cy3 and fluorescein isothiocyanate emissions were acquired simultaneously. The images show a medial optical cut of representative cells. CD4 staining was intracellular in this particular preparation.
From these studies we conclude that CD4 ϩ ␥ Ϫ and CD8 ϩ ␥ Ϫ cells shared a poor surface association of CD3 to TCR chains. We also conclude, by using several experimental approaches, that their surface ␣␤ TCR⅐CD3 complexes differed biochemically in the CD3 and the TCR components, as well as in the complete native complex.
CD4 ϩ and CD8 ϩ Primary T Lymphocytes Express Biochemically Different Membrane ␣␤ TCR⅐CD3 Complexes-Normal primary CD4 ϩ T cells show higher TCR⅐CD3 staining levels than do CD8 ϩ cells (ϳ1.5-fold with OKT3; Ref. 2). This difference was interpreted as quantitative under the assumption that the ␣␤ TCR⅐CD3 complex is identical in both cell types. However, as shown above and in Ref. 6, the analysis of ␥ Ϫ TCR⅐CD3 complexes suggested the existence of hitherto unrecognized structural differences between both cell lineages. These results prompted us to search for qualitative ␣␤ TCR⅐CD3 complex differences in normal CD4 ϩ and CD8 ϩ primary T lymphocytes. To test this hypothesis, a broad epitopic scan of the TCR⅐CD3 complex was performed by three-color flow cytometry using several CD3-or TCR-specific mAbs within gated CD4 ϩ and CD8 ϩ subsets. To exclude CD8 ϩ NK cells (mostly CD8␣␣ ϩ , which stain dull for CD8 mAb), only CD8 bright cells (CD8␣␤ ϩ T cells) were analyzed. As a control, CD18-specific mAbs were used, which stained primary CD4 ϩ and CD8 ϩ T cells similarly. The data are depicted in Fig. 5A as MFI ratios of each mAb in CD4 ϩ relative to CD8 ϩ T cells, with representative histograms in Fig. 5B. The results showed that, for most of the mAb assayed, CD4 ϩ lymphocytes stained better (MFI ratio Ͼ1.5, Fig. 5A) than did CD8 ϩ lymphocytes, as described (2). However this was not the case with two mAbs that stained CD8 ϩ T cells similarly (MFI ratio ϭ 1.2, WT31; Ref. 16) or even better (MFI ratio Ͻ1, RW2-8C8; Ref. 17) than did CD4 ϩ T cells. These observations are not consistent with the quanti-tative interpretation but, rather, with the existence of conformational (i.e. qualitative) differences in the ␣␤ TCR⅐CD3 complex between CD4 ϩ and CD8 ϩ T cells.
To determine whether these conformational differences were developmentally regulated or regulated by activation (as described for activated ␥␦ T cells, 18), mature CD4 ϩ and CD8 ϩ T lymphocytes were analyzed and compared for CD3 (and CD18) expression in four consecutive differentiation stages as follows: 1) intrathymic single positive (CD4 ϩ CD8 Ϫ or CD4 Ϫ CD8 ϩ ) early mature T cells (gated as described under "Experimental Procedures"); 2) peripheral blood naïve T cells (defined by CD45RA ϩ expression, which identifies recent thymic emigrants, Ref. 19); 3) peripheral blood memory T cells (CD45RO ϩ ); and 4) recently activated peripheral blood T cells (gated as CD69 ϩ after phytohemagglutinin incubation) or constitutively activated H. saimiri (HVS)-transformed T cells (20). The results indicated that the increased RW2-8C8 binding of CD8 ϩ as compared with that of CD4 ϩ T cells was essentially independent of their activation or differentiation status, although it was more apparent in peripheral T cells (naïve, memory, or activated) than in intrathymic single positive T cells (Fig. 5C). In contrast, increased UCHT1 binding of CD4 ϩ T cells as compared with that of CD8 ϩ T cells was clearly regulated by activation and differentiation, as it occurred only in intrathymic and naïve T cells but not in memory or activated T cells. These changes were specific for the TCR⅐CD3 complex, as CD18 was regulated in an opposite fashion. Interestingly, equivalent UCHT1 binding of CD4 ϩ and CD8 ϩ memory T cells was acquired through both a decreased binding of CD4 ϩ T cells and an increased binding of CD8 ϩ T cells relative to their naïve counterparts, whereas RW2-8C8 binding differences were unaltered by the naïve/memory transition (not shown). Therefore, CD4 ϩ and CD8 ϩ T cells indeed regulated their ␣␤ TCR⅐CD3 FIG. 4. Native surface TCR⅐CD3 complex analyses of CD3␥-deficient CD4 ؉ and CD8 ؉ T cells. A, equal numbers of ␥ ϩ and ␥ Ϫ cells, either CD4 ϩ or CD8 ϩ , were stimulated with pervanadate and lysed in 1% digitonin before the purification of phosphorylated proteins using anti-phosphotyrosine-immunoprecipitation. Bound proteins were eluted, dephosphorylated, and separated by BN-PAGE. Western blots were developed with anti-. Ferritin monomers (f1), dimers (f2), and trimers (f3) were included as molecular weight standards. B, antibody-based gel shift assay of CD4 ϩ ␥ Ϫ or Jurkat T cells using the indicated mAbs.
conformational differences but did so differently for each epitope, and the RW2-8C8 epitope was well conserved along mature T cell differentiation.
It has been proposed that lipid raft integrity affects the topological arrangement of the TCR⅐CD3 complex on the cell surface (10). Also, it is believed that a fraction of TCR⅐CD3 is raft-associated because of its association with the cytoplasmic portion of the CD8 (or CD4) coreceptor (21). We therefore reasoned that the conformational TCR⅐CD3 differences observed between CD4 ϩ and CD8 ϩ T cells could be due, at least in part, to the differential arrangement of lipid raft-associated surface TCR⅐CD3 clusters or arrays. To test this hypothesis, primary T lymphocytes were treated with M␤CD and analyzed for CD3 expression within the CD4 ϩ and CD8 bright T cell subsets using several TCR⅐CD3-specific mAbs (Fig. 5, D and E). M␤CD disrupts lipid microdomains by extracting cholesterol from plasma membranes (22). WT31 and OKT3 were used as negative and positive controls for the TCR⅐CD3 complex, respectively, as described (11). Two different CD18-specific mAbs were used as additional negative controls (10). The result showed that the TCR⅐CD3 complex was more dependent on lipid raft integrity in CD8 ϩ as compared with CD4 ϩ T cells when probed with several antibodies (RW2-8C8, OKT3, Leu4, Cris7, UCHT1, and BMA031) but not with WT31 (as expected; Ref. 11). These results are consistent with the existence of more lipid raft-associated TCR⅐CD3 domains (or isoforms) in CD8 ϩ than in CD4 ϩ T cells. Interestingly, RW2-8C8 seemed to bind a relatively lipid raft-independent TCR⅐CD3 epitope (or isoform) in CD4 ϩ , which was lipid raft-dependent in CD8 ϩ T cells. It was surprising to find that the lipid raft-dependent TCR⅐CD3specific mAbs were those that bound CD4 ϩ better than CD8 ϩ T cells (compare Fig. 5, A and D).
Because the TCR⅐CD3 protein components are equivalent in CD8 ϩ and CD4 ϩ T cells, it was also possible that the conformational differences were due, in part, to glycosylation variability of the TCR⅐CD3 complex, as shown for ␥␦ T cells (12, 18). To test this hypothesis, primary T cells were digested with NANAse and analyzed for CD3 expression within the CD4 ϩ and CD8 bright T cell subsets using RW2-8C8 and Leu4 (Fig. 6A). Decreased binding of CD43 was used as an internal control of NANAse treatment, as described (13). The results showed that NANAse-treated CD4 ϩ T cells became indistinguishable from CD8 ϩ T cells using RW2-8C8 but not Leu4, suggesting that NANA-associated glycosylation of the TCR⅐CD3 components in CD4 ϩ T cells partially hides (or builds) the RW2-8C8 epitope.
To further characterize the RW2-8C8 epitope, competition experiments were performed on several mature and immature T lymphocyte types using phycoerythrin-labeled Leu4 (CD3specific) or BMA031 (framework TCR␣␤-specific). The results CD8 ϩ T cells with the indicated mAb in n independent donors. MFI ratios above or below 1 (indicated by the horizontal dotted line) reflect an increased or decreased mAb binding to CD4 ϩ cells, respectively. As an invariant control, CD18/CD11b expression was evaluated in parallel with two different mAbs recognizing different epitopes of the same molecule. B, representative reactivity patterns of selected mAbs in CD4 ϩ (open histograms) and CD8 bright (filled histograms) PBLs. Profiles are shown as the logarithm of relative fluorescence versus the number of cells. The vertical line in each panel indicates the upper limit of background fluorescence using isotype-matched irrelevant mAb. C, MFI ratios as in panel A for the indicated mAb and T cell subsets from at least three independent donors (HVS, H. saimiri-transformed; Rest, resting; Act, activated). D, purified PBLs from n independent donors were treated with M␤CD to disrupt rafts and analyzed for surface binding of the indicated mAbs within CD4 ϩ and CD8 ϩ T cell subsets. The results are given relative to untreated cells. E, representative reactivity patterns of selected mAbs in CD4 ϩ and CD8 bright PBLs before (thin histograms) or after (thick histograms) M␤CD treatment. The numbers denote the percentage of binding as in panel D.
FIG. 5. Comparative cell surface TCR⅐CD3 expression by normal CD4 ؉ and CD8 ؉ peripheral blood lymphocytes and dependence on lipid raft integrity. A, MFI ratios Ϯ S.D. of CD4 ϩ relative to (Fig. 6B) showed that preincubation with unlabeled RW2-8C8 blocked BMA031 but not Leu4 (or UCHT1, not shown) binding. In contrast, another CD3-specific mAb (UCHT1) or an irrelevant mAb (CD18) did not preclude BMA031 binding. However, RW2-8C8 is CD3-rather than TCR␣␤-specific, as it was shown to bind TCR␥␦ ϩ BMA031 Ϫ PBLs (Fig. 6B, bottom right). Therefore, RW2-8C8 binds to a CD3 determinant that is closer to (or more influenced by mAb binding to) the BMA031 epitope than to the Leu4 or UCHT1 epitopes. Alternatively, these results are consistent with the existence of two isoforms of surface TCR⅐CD3 complexes, one that cannot be engaged by Leu4 or BMA031 if RW2-8C8 is present and another in which RW2-8C8 binding blocks BMA031 but not Leu4 or UCHT1 binding. Primary CD4 ϩ and CD8 ϩ ␥ Ϫ T lymphocytes were also analyzed for the expression of the epitope recognized by RW2-8C8. The results (Fig. 6C) indicated that RW2-8C8 but not Leu4 again stained CD8 ϩ T cells better than it stained CD4 ϩ T cells even when CD3␥ is absent. These results demonstrated that RW2-8C8 binds to a CD3 determinant that is more strongly expressed by CD8 ϩ than by CD4 ϩ T cells even in the absence of CD3␥. The notion that RW2-8C8 may be detecting a distinct CD3 conformational epitope was explored in several T-lineage cell types as follows: 1) a cell line expressing the pre-TCR⅐CD3, an immature BMA031 Ϫ TCR⅐CD3 complex in which the variable TCR␣ chain is replaced by the invariant pT␣ component (SupT1; Ref. 23); 2) Jurkat cells bearing the mature BMA031 ϩ TCR␣␤; and 3) CD4 ϩ CD8 ϩ (double positive) thymocytes bearing an immature BMA031 ϩ TCR␣␤ (24). The results (Fig. 6D) showed that RW2-8C8 was the best CD3 antibody in terms of SupT1 or double positive thymocyte binding, but not for the conventional TCR⅐CD3 expressed by control Jurkat T cells. Taken together, these results indicated that RW2-8C8 binds to a peculiar CD3 determinant that is close to the TCR heterodimer, more prominent than those recognized by other CD3specific mAb in immature TCR␤-containing ensembles, and more strongly expressed by CD8 ϩ than by CD4 ϩ T cells even when CD3␥ is absent.
Finally, surface radioiodination and immunoprecipitation studies using both CD3-and TCR-specific mAbs were performed to compare the TCR⅐CD3 expressed by purified primary CD4 ϩ and CD8 ϩ T cells. The results (Fig. 7) again revealed the existence of consistent differences in TCR (but not CD3) components between CD4 ϩ and CD8 ϩ T cells. The densitometry ratio of high to low molecular weight forms of ␣␤TCR (TCRtg (tg, totally glycosylated) in Fig. 7) was higher in CD4 ϩ T cells than in CD8 ϩ T cells (1.8 versus 1.4, 2.5 versus 1.6. and 1.7 versus 1.4 using CD3-, TCR␤-, or TCR␣ϩTCR␤-specific mAbs, respectively). These differences are probably due to N-linked glycosylation trimming, because they disappeared after canase treatment.
FIG. 6. Characterization of the RW2-8C8 antibody. A, purified PBLs from normal donors were NANAse-treated (ϩ) to remove sialic acid residues or left untreated (Ϫ) and analyzed for surface binding of RW2-8C8, Leu4, and CD43 (positive control) within CD4 ϩ and CD8 ϩ T cell subsets. B, mature or immature T cells were stained with Leu4phycoerythrin (or UCHT1, not shown) or BMA031-phycoerythrin 2) TCR␣ and TCR␤ chain content ( Fig. 2 and 3; and 3) native ␣␤TCR⅐CD3 migration (Fig. 4). Second, primary normal CD4 ϩ and CD8 ϩ T cells were shown to express conformationally and biochemically different surface ␣␤TCR⅐CD3 complexes (Figs. 5 and 7, respectively), which were regulated by activation and differentiation for certain epitopes, likely due to lipid raft-dependent arrangements or glycosylation. These results confirm our previous biosynthetic and phenotypic studies in CD3␥deficient T cells (6) and extend the observations to primary T lymphocytes from normal individuals. Biosynthetic studies showed that CD8 ϩ but not CD4 ϩ CD3␥-deficient T cells contained a small ␣␤ heterodimer composed of abnormally glycosylated TCR␤ and an abnormally small CD3-associated chain that was not recognized by TCR␣-specific antibodies. However, the TCR C␣ gene was normal, and the atypical 32-kDa TCR␣ observed intracellularly was indeed shown to be TCR␣ by protein sequencing using matrix-assisted laser desorption ionization time-of-flight (data not shown).
The TCR⅐CD3 complex is a very flexible structure. For instance, the ␥␦ TCR (and ␥␦ T cell development) can be quite CD3␦-independent in mice (5,25). There are descriptions of CD3-specific mAb that do not bind the ␥␦ TCR⅐CD3 complex unless it was deglycosylated previously (WT31, 12), and even T cell activation can modify CD3␥ glycosylation (18) with clear biochemical consequences. Similarly, ␣␤TCR⅐CD3 expression is possible without CD3␥ (4,8) or ␦ (5), but not without both (26). Therefore, it is not surprising that further flexibility is present in CD4 ϩ versus CD8 ϩ T cells, particularly when CD3␥ is absent, but also in normal donors. The mAbs used for surface TCR⅐CD3 detection were obtained using different antigens (bulk PBL or thymocytes, purified primary T cells or T cell lines, solubilized cell membranes, purified TCR⅐CD3 proteins, etc.) and different screening criteria (fresh, fixed, or tumor T cell binding, cytolysis blocking, PBL stimulation, etc.) (27). Therefore, a possible explanation for the observed lineage-specific binding differences could stem from the immunization or screening criteria used in each case, a situation that would favor naïve CD4 ϩ and lipid raft-dependent over naïve CD8 ϩ and lipid raft-independent TCR⅐CD3 binding with certain mAbs. These differences could be exploited for therapeutic purposes, such as lineage-specific immunosuppression during graft rejection.
What could be the mechanism responsible for the observed lineage-dependent biochemical TCR⅐CD3 differences? Because all of the subunits are present in both cases (even with the same stoichiometry, if isolated with digitonin), we believe that there must be biochemical differences in the way the complex is assembled, glycosylated, trimmed, or topologically arranged in the cell surface in each T cell subset after the lineage decision is reached and apparently also upon antigen recognition (28). The reported physical association of CD3␦ with the raft-resident coreceptor molecules CD4 and CD8 on resting T lymphocytes could contribute to the observed differences (21,29). In addition, the individual TCR⅐CD3 subunits could possess slightly distinct structures. These differences could be caused by developmentally acquired fundamental changes that distinguish the two cell types, including the glycosylation machinery (30,31) or chaperones. For instance, the inactivation of a sialyltransferase (ST3Gal-I) strongly reduced peripheral CD8 ϩ but not CD4 ϩ cell numbers (32). Further work is required to address this issue, perhaps through a comparative genomic approach that includes the chaperones and enzymes involved in glycosylation pathways. Interestingly, TCR⅐CD3 signaling capacity and dynamics are not affected by the observed biochemical differences, either with or without CD3␥ (6 -8), further suggesting that the changes are required for optimal receptorcoreceptor cis interactions during antigen recognition and signal transduction in each cell type (33). An alternative interpretation is that two different TCR populations exist on the cell membrane and that their relative proportions change in different T cell subsets. Antibodies such as RW2-8C8 may detect those changes.
␣␤TCR⅐CD3 Structure and Function-The results are consistent with a recent report on the structure and stoichiometry of the TCR⅐CD3 complex in vitro (34). The authors demonstrated that CD3␥ is important for incorporation of but not ⑀ to the complex, as shown in Fig. 1A (CD3-associated is very scarce in ␥ Ϫ cells) and in Ref. 6. They also proved that, whereas TCR␣ is strictly associated to ␦⑀ dimers, TCR␤ can interact with ␥⑀ as well as with ␦⑀ dimers. This apparent biochemical promiscuity may explain why surface TCR⅐CD3 adopts a ␣␤(␦⑀) 2 2 stoichiometry (Fig. 4) and why its expression is so notable despite the lack of CD3␥ (7). Although the FIG. 7. Immunoprecipitation of surface-iodinated TCR⅐CD3 complexes from isolated CD4 ؉ and CD8 ؉ fresh peripheral blood lymphocytes. CD4 ϩ (15 ϫ 10 6 cells) and CD8 ϩ (5 ϫ 10 6 cells) PBLs from a normal donor were immunomagnetically isolated, surface-labeled with 125 I, lysed in 1% digitonin-containing buffer, precipitated with the indicated CD3-, TCR-, or HLA-specific mAb, and digested, where indicated (ϩ), with N-glycosidase F (N-Gly) before electrophoresis under reducing conditions in 10% polyacrylamide gels. The positions of the expected TCR and CD3 proteins are indicated (dg, deglycosylated; tg, totally glycosylated). same authors showed that isolated ␣␤ and ␦⑀ dimers do not normally assemble into ␣␤(␦⑀) 2 complexes in vitro, intrathymic selection mechanisms may expand, in vivo, the otherwise rare T cell precursors that manage to assemble a viable surface TCR⅐CD3 complex with such stoichiometry. In vivo observations further support this contention, because human CD3␥-deficient individuals showed significant T cell development with only mild lymphopenia and immunodeficiency, whereas CD3␦-deficient patients had no T cells and very severe clinical abnormalities (35).