The CD3ε Subunit of the TCR Contains Endocytosis Signals¹

The TCR expressed in the majority of peripheral T lymphocytes is a complex composed of the hypervariable TCR αβ heterodimer noncovalently linked to the monomorphic CD3γ-, δ-, ε-, and ζ-chains (TCR-CD3). The α- and β-chains contain Ig-like variable domains responsible for Ag and superantigen recognition (1), whereas the CD3γ-, δ-, ε-, and ζ-chains are involved in signal transduction (2). Cell surface expression of the TCR-CD3 complex requires the assembly of at least three chain pairs, αβ, ζζ, and γε or δε. Unassembled subunits or partial complexes are not expressed on the plasma membrane, because they are retained along the secretion pathway and eventually degraded (3, 4).

Once on the plasma membrane, TCR-CD3 complexes are continuously internalized and recycled back to the cell surface (5, 6). Moreover, soon after TCR ligation by Ag, superantigen, or mAbs, the cell surface expression of TCR-CD3 is down-modulated. Down-modulation results from increased receptor internalization and degradation (5, 7, 8, 9, 10).

Internalization of cell surface receptors requires the presence of particular sequences, mainly located in their cytosolic regions, which mediate the interaction of receptors with molecular components of the cellular endocytic machinery. These sequences belong mainly to two families, which are the tyrosine-based motif and the di-leucine motif. The tyrosine-based motif involves a critical aromatic residue (usually tyrosine) placed in a context of one or more amino acids with large hydrophobic side chains (YxxΦ), where x represents any amino acid and Φ a hydrophobic residue. Likewise, sequences composed of two leucines, or leucine and another hydrophobic amino acid, are also involved in receptor endocytosis. Both the tyrosine-based motifs and the di-leucine motifs have been reported to mediate the localization of receptors into clathrin-coated pits (11, 12).

Because the TCR-CD3 complex is composed of multiple subunits, its internalization may depend on sequences present in one or several of its subunits. Interestingly, internalization of TCR-CD3 complexes is determined by one or several CD3 chains, and this depends on the kind of stimulus-inducing internalization. For instance, TCR-CD3 internalization induced by phorbol esters requires the presence of a di-leucine-based motif present in the cytosolic region of CD3γ (13, 14). However, this motif is not indispensable for TCR-CD3 internalization induced by enterotoxin superantigens or anti-TCR-CD3 mAbs (9, 15, 16). Internalization of TCR-CD3 induced by an anti-CD3 mAb occurs in the absence of the cytosolic region of either CD3γ or CD3δ, but it is blocked when both cytosolic regions are missing (15). However, the absence of both CD3γ and CD3δ cytosolic regions still allows TCR-CD3 internalization induced by peptide Ag.³ Therefore, under physiologic conditions, TCR-CD3 internalization likely involves determinants present in TCR-CD3 subunits different from CD3γ and CD3δ. Because TCRα and -β display short cytosolic regions, it appears likely that CD3ε could bear determinants involved in TCR-CD3 internalization.

To investigate whether CD3ε itself displays endocytosis sequences, we expressed CD3ε independently of other TCR-CD3 subunits, and we analyzed its capacity to be internalized. We show here that CD3ε contains endocytosis determinants. Therefore, CD3ε could potentially contribute to the internalization of TCR-CD3 complexes. Moreover, the presence of endocytosis sequences could allow the retrieval of unassembled CD3ε chains that could have escaped retention in the endoplasmic reticulum (ER)⁴ and reached the cell surface independently of a fully formed TCR-CD3 complex.

Chemicals were from Sigma Chemical (St. Louis, MO). ¹²⁵I was from Amersham-Pharmacia Biotech (Piscataway, NJ). Human transferrin (Sigma Chemical) was loaded with iron and coupled to lissamine rhodamine (Eastman Kodak Co., Rochester, NY) or to fluorescein isothiocyanate (Molecular Probes, Eugene, OR) as previously described (17, 18). The mouse anti-CD3ε mAb (SP34, IgG3) has been previously described (19). F(ab′)₂ fluorescein-labeled rabbit anti-mouse Ig was from Dako (Glostrup, Denmark), and F(ab′)₂ PE-labeled goat anti-mouse Ig Fc was from Immunotech (Marseille, France). The antiserum to rat growth hormone (GH) for immunocytochemistry (lot AFP4115) was obtained from the National Institute of Diabetes and Digestive and Kidney Diseases (National Institutes of Health, Bethesda, MD). The mouse mAb against human CD8α (OKT8, IgG2a) was from Ortho Diagnostic Systems (Raritan, NJ).

All chimeras and mutations were constructed by PCR using Taq DNA polymerase (New England Biolabs, Beverly, MA) essentially as previously described (20, 21). Human CD8 and CD3ε cDNAs were used as templates, with appropriate oligonucleotides designed to introduce the amino acid changes. All of the PCR products were digested with XhoI and BamHI and cloned in the pSRα expression vector (22). The constructs were all checked by complete DNA sequencing.

COS cells were grown in DMEM containing 4 g/L glucose and 10% FCS. COS cells were transfected by electroporation. Routinely, one 100-mm plate of cells grown to subconfluence was used for each transfection. Cells were resuspended in 200 μl of DMEM supplemented with 10% FCS and 10 mM HEPES buffer (pH 7.2) and gently mixed with the DNA transfection mixture (6 μg of pSRα-CD3ε vector, 17 μg of pSK plasmid used as carrier DNA, and 5 μl of 1.5 M NaCl) and then electroporated in 0.4-cm cuvettes at 200 V, 900 μF, using an EasyjecT apparatus (Eurogentec, Seraing, Belgium). The cells were carefully resuspended by pipetting and plated in a 100-mm petri dish. After 24 h of culture, the cells were transferred to 24-well plates. Rat basophilic leukemia (RBL) cells were grown in RPMI 1640 supplemented with 10% FCS and 10 mM HEPES buffer (pH 7.2). RBL cells were transfected with 20 μg of pSRα-CD3ε (R183S) by electroporation at 260 V, 900 μF. Stably transfected cells were selected in 1 mg/ml G418 (Life Technologies, Paisley, U.K.). Positive clones were selected by immunofluorescence and flow cytometry using the anti-CD3ε mAb, SP34. Jurkat cells lacking TCR-CD3 surface expression (clone 31-13) (23) were grown in RPMI 1640 supplemented with 10% FCS and 10 mM HEPES buffer (pH 7.2). Cells were transfected with 20 μg of pSRα-CD8/ε by electroporation at 260 V, 900 μF. Stably transfected cells were selected in 2 mg/ml G418 (Life Technologies). Positive clones were selected by immunofluorescence and flow cytometry using an anti-CD8 mAb, OKT8.

Experiments were conducted essentially as previously described (8). Fluorochrome-labeled transferrin was used at 150 nM.

Experiments were conducted 48 h after transfection on COS cells growing in 24-well plates. Cells were cooled on ice for 3 min and washed once at 4°C with DMEM supplemented with 10% FCS and 10 mM HEPES buffer (pH 7.2). The medium was thoroughly removed, and 75 μl of the same cold medium containing ¹²⁵I-labeled anti-CD3ε mAb (SP34) at a final concentration of 40 nM was added per well. Cells were incubated for 30 min at 4°C to allow binding of the Ab and then at 37°C for the appropriate times to allow internalization. At the end of each time point, cells were washed three times at 4°C with DMEM. The remaining surface-bound Ab was then removed by two successive acid washes (8 min with 800 μl of DMEM containing 25 mM sodium acetate, pH 1.8). At the end, the cells were lysed in 400 μl of 0.1 M NaOH solution and the wells washed with 800 μl of distilled water. Acid treatment removed the surface-bound Ab with an efficiency of 80–95%. Nonspecific internalization was determined using cells incubated in the presence of a 10-fold excess of unlabeled Ab or using the same amount of mock-transfected COS cells. Both techniques gave similar results. Data were corrected taking into account nonspecific internalization values as well as the efficiency of the acid wash, and the percentage of internalized receptors was calculated as previously described (24).

Cell surface expression of CD3ε was measured by immunofluorescence and flow cytometry at various times after binding of soluble or plastic-bound Abs as previously described (25).

To analyze the capacity of CD3ε to be internalized, this chain was expressed independently of other TCR-CD3 subunits. Isolated CD3ε cannot reach the cell surface due to the presence of an ER retention signal in its cytosolic region (20, 26). We therefore expressed a CD3ε form containing a mutation in its ER retention signal (R183S). As shown in Fig. 1 A, this mutant was stably expressed on the cell surface of RBL cells. Immunoprecipitation experiments showed that CD3ε is mainly expressed in these cells as disulfide-linked dimers, with monomeric and trimeric forms also present (data not shown). Disulfide-linked CD3ε homodimers have been reported to exist in T lymphocytes, where they coexist with CD3ε-γ and CD3ε-δ heterodimers (27). Likewise, homodimeric and trimeric forms of CD3ε have been observed in COS cells coexpressing several CD3 subunits (28), as well as in in vitro translated and assembled TCR-CD3 complexes (29). This suggests that the folding and self-assembly of CD3ε chains in the presence or absence of other TCR-CD3 subunits are similar. Therefore, these polypeptides are suitable for investigating the presence of internalization determinants in CD3ε.

To test whether CD3ε bears internalization signals, cells were incubated in the presence of anti-CD3ε mAb and rhodamine-labeled transferrin and then fixed, permeabilized, and stained with fluorescein-labeled secondary Abs. As shown in Fig. 1,B, cells incubated at 37°C, but not those incubated at 4°C, internalized the anti-CD3ε mAb, which colocalized with transferrin (Fig. 1,B). Transferrin accompanies its receptor through the endocytic and recycling pathway and is therefore widely used as a marker for early endocytic organelles (30, 31). Therefore, our data indicate that CD3ε is internalized and reaches endocytic intracellular compartments. Furthermore, cross-linking with a surface-bound mAb induced the down-regulation of CD3ε from the cell surface, as assessed by flow cytometry (Fig. 1 C). Altogether, these data show that CD3ε can be internalized in the absence of other components of the TCR-CD3 complex, thus indicating that this subunit displays internalization sequences.

To localize putative endocytosis sequences in the cytosolic tail of CD3ε, a panel of mutants encompassing the whole cytosolic region was generated (Fig. 2,A). All of these mutants were expressed on the cell surface of transfected COS cells at comparable levels (Fig. 2,B). The capacity of these mutants to be internalized was then measured using a ¹²⁵I-labeled anti-CD3ε mAb. Consistent with the results shown in Fig. 1, three CD3ε mutants containing single-residue changes in the ER retention sequence (L180S, R183K, and R183S) were internalized equally well (Fig. 3,A). A panel of truncated CD3ε molecules was then tested. As shown in Fig. 3,B, the deletion of residues 181–185 (Δ1) did not inhibit CD3ε internalization, as compared with single-residue mutants described in Fig. 3 A. However, a larger deletion involving residues 166–185 (Δ2) significantly reduced the capacity of CD3ε to be internalized. This suggests that some sequences contained in the 166–180 region could be involved in CD3ε internalization. Interestingly, this region contains two tyrosine-based motifs (Y166EPI and Y177SGL), which could support CD3ε internalization (12). These motifs are part of the ITAM of this subunit that is involved in signal transduction.

Although the deletion of residues 166–185 (Δ2) inhibited CD3ε internalization, the inhibition was only partial (Fig. 3,B), indicating that CD3ε bears additional internalization sequences. To determine whether the cytosolic tail of CD3ε contains additional internalization sequences, the effect of additional truncations in CD3ε internalization was analyzed. Thus, CD3ε molecules lacking residues 166–185 and other residues of the cytosolic region were tested (see Fig. 2,B, Δ3–Δ6). As shown in Fig. 3 C, none of the additional deletions of the CD3ε cytosolic tail had a further inhibitory effect on the internalization capacity of the Δ2 mutant.

Altogether, these data indicate that no other endocytosis sequences were present in the cytosolic tail of CD3ε, besides the endocytosis signals present between amino acids 166 and 180. These results suggest that CD3ε could also contain endocytic signals outside the cytosolic region.

The region between residues 166 and 180 contains two consensus tyrosine-based motifs (Y166EPI and Y177SGL) that could support CD3ε internalization (12). To investigate whether any of these motifs is actually a tyrosine-based internalization signal for CD3ε, we replaced both tyrosine residues by serine. These amino acid substitutions were shown to abolish internalization of other receptors containing tyrosine-based endocytic signals (11). As shown in Fig. 3 D, the substitution of both tyrosine residues by serine did not inhibit CD3ε internalization. These results indicate that the two tyrosine-based sequences of CD3ε do not behave as bona fide tyrosine-based internalization signals for this molecule. Nevertheless, the deletion analysis described above strongly suggests that this region of the molecule is involved in its endocytosis.

A further proof for the presence of endocytic signals in a receptor can be obtained by analyzing whether the graft of the region containing the putative endocytic sequence confers to another protein the capacity to be endocytosed. Most of the previously described endocytosis motifs have been localized in the cytosolic region of receptors (12), although several reports have shown that transmembrane regions could also support receptor internalization (32, 33, 34, 35, 36, 37). Therefore, we generated two chimeric molecules in which the extracellular region of CD3ε was replaced by the rat GH. These chimeras contained either the transmembrane region alone, or both the transmembrane and the cytosolic region of CD3ε (Fig. 4,A). The capacity of these chimeras to be internalized was then analyzed in transiently transfected COS cells using anti-GH Abs. As shown in Fig. 4,B, both chimeras were internalized, as revealed by down-regulation experiments, although the one lacking the cytosolic region of CD3ε was less efficiently down-regulated. Moreover, both chimeras were found in intracellular vesicular compartments, as assessed by immunofluorescence and confocal microscopy (Fig. 5). Double labeling microscopy using rhodamine-coupled secondary Abs to detect the GH chimera and fluorescein-coupled transferrin showed colocalization between GH/CD3ε chimera and transferrin labeling (Fig. 5), indicating that the chimeras reached the endocytic organelles. This vesicular staining was not observed in cells incubated with irrelevant Abs, implying that the entry of the anti-GH Ab was mediated by the GH/CD3ε chimeras and not by fluid-phase endocytosis (data not shown). These data provide additional proof for the presence of endocytic signals in CD3ε and indicate that the transmembrane region of CD3ε could also contribute to its internalization.

Because all of the data described above were obtained in non-T cells, it was necessary to verify whether the internalization signals borne by isolated CD3ε molecules are also functional in T lymphocytes. To this end, the same cDNA constructs mutated in the ER retention signal described above were transfected in Jurkat T cells lacking TCR-CD3 cell surface expression. However, CD3ε constructs carrying the extracellular and transmembrane regions did not reach the cell surface. This is likely due to its association with other CD3 subunits, which are expressed intracellularly in these cells. This clone lacks TCRβ expression but expresses intracellularly all of the other subunits of the complex (23). Although the CD3ε constructs transfected lack the ER retention signal, association of these mutants with other CD3 subunits in the ER likely leads to the retention of the complexes via the ER retention signals of other subunits (4). Because the assembly into complexes of TCR-CD3 subunits involves mainly the extracellular and transmembrane regions (4, 28), we made a construct in which the extracellular and transmembrane regions of CD3ε were substituted by those of CD8α. This chimera was readily expressed in TCR-CD3-negative cells. Moreover, it was internalized as efficiently as CD3ε molecules expressed in non-T cells, as assessed by down-modulation experiments (Fig. 6), thus indicating that CD3ε carries internalization determinants that are functional in T lymphocytes.

In this report we show that CD3ε bears endocytosis signals that could contribute to the internalization and cell surface down-regulation of TCR-CD3 complexes.

Previously reported data indicated that internalization of TCR-CD3 complexes may involve determinants from one or several of its subunits, depending on the kind of stimulus inducing internalization. For instance, phorbol ester-induced TCR-CD3 internalization depends on the presence of a di-leucine-based motif located in the cytosolic region of CD3γ (13, 14). In contrast, TCR-CD3 internalization induced by superantigens or mAbs occurs even in the absence of the entire cytosolic region of CD3γ (9, 15, 16). Moreover, TCR-CD3 internalization induced by an anti-CD3 mAb can take place in the absence of the cytosolic region of either CD3γ or CD3δ, but it is blocked when both of these cytosolic regions are deleted (15). However, the absence of the cytosolic regions of CD3δ and CD3γ still allows TCR-CD3 internalization induced by peptide Ag.³ Therefore, under physiologic conditions, TCR-CD3 internalization likely involves determinants present in TCR-CD3 subunits other than CD3γ and CD3δ. The data we present here demonstrate that CD3ε itself bears endocytosis determinants and could therefore support the internalization of TCR-CD3 complexes.

Unlike many other receptors that display only one strong internalization signal in their cytosolic regions (11, 12), our data indicate that CD3ε internalization is determined by the additive effect of determinants located in the cytosolic and transmembrane regions of the molecule. One of these is located between amino acids 166 and 180 (Fig. 3). This determinant could be formed by the tyrosine-based motifs (YxxI/L) that constitute the ITAM of CD3ε. However, these sequences do not seem to behave as typical tyrosine-based motifs described for other receptors (11), because the mutations of both tyrosine residues to serine did not alter CD3ε internalization. Rather, this sequence seems to behave as a weak internalization signal, such as those found in the IL-2 receptor β and γ subunits (35, 37). The tyrosine-based motifs within the ITAM of other receptors have been shown to be involved in internalization. However, their particular features were different depending on the receptor and on the ligand inducing internalization. For instance, tyrosine-based motifs of the ITAMs of CD3γ and CD3δ, together with a di-leucine motif, mediate the internalization of chimeric molecules containing the cytosolic tail of these CD3 chains (38). Moreover, both tyrosine residues of the ITAM of FcγRIII were shown to be required for its internalization, because the mutation of each tyrosine residue to valine or serine abolished multimeric receptor internalization (39). In contrast, although one of the tyrosine-based motifs of the ITAM of Igα is required for B cell receptor constitutive internalization, this mutation does not impair the internalization induced by multivalent ligands (40).

Sequential deletions of the CD3ε cytosolic tail between the transmembrane region and aa 166 (Fig. 2,A, Δ3–Δ6) did not inhibit the internalization capacity of the 166–185 (Δ2) deletion mutant (Fig. 3,C). This suggests that, within the cytosolic tail of CD3ε, only the region between amino acids 166 and 180 contains endocytosis signals and that the other internalization determinants could be part of the transmembrane region. In support of this, a chimera containing only the transmembrane region of CD3ε fused to the rat GH was internalized, although less efficiently than a chimera displaying both the transmembrane and cytosolic regions of CD3ε (Figs. 4 and 5). The presence of signals for sorting in the endocytic pathway in both the cytosolic and transmembrane regions has been reported for other membrane proteins (32, 33, 34, 35, 36, 37), although a transmembrane endocytic motif had not been described.

Our data demonstrate that CD3ε has the capacity to be internalized, thus suggesting that this subunit could contribute to TCR-CD3 complex internalization. Moreover, considering, on the one hand, that CD3γ and CD3δ also carry internalization signals (14, 15, 38) and, on the other hand, that CD3γ and CD3δ are not the only subunits involved in TCR-CD3 internalization,³ it is likely that TCR-CD3 complex internalization utilizes endocytic signals from all CD3 subunits. This would not be surprising, because all CD3 subunits are involved in TCR signal transduction.

Nevertheless, from our analysis, we cannot determine whether all or only some of the CD3ε endocytic determinants shown here are involved in the internalization of the TCR-CD3 complex. In fact, some of these determinants may not be accessible to the endocytic machinery when this subunit is part of a complete TCR-CD3 complex. In this case, endocytosis signals exposed only in isolated CD3ε subunits could serve to retrieve CD3ε subunits or partial CD3 complexes that would have escaped the ER retention barrier and reached the plasma membrane independently of fully assembled TCR-CD3 complexes. In this regard, it is worth noting that only low amounts of CD3ε associated with CD3γ or CD3δ were found expressed on the surface of certain thymomas as well as on immature CD4⁻CD8⁻ thymocytes but not in mature T cells (41, 42, 43). The removal of isolated CD3ε molecules from the plasma membrane may represent an additional control mechanism to prevent the expression of unassembled CD3ε subunits on the cell surface.

We thank Dr. D. Ojcius for critical reading of the manuscript.