Surface Expression and Functional Competence of CD3-Independent TCR ζ-Chains in Immature Thymocytes¹

Development of αβ thymocytes is characterized by the transition of cells through an ordered sequence of distinct phenotypes, which can be defined by the expression of the coreceptor molecules CD4 and CD8. Early in development, the transition of the most immature double-negative (DN)³ to the double-positive (DP) stage is largely dependent upon the productive rearrangement of TCRβ genes and expression of TCR β-chains in association with monomorphic pTα chains 1 . Cell-autonomous signaling by the pre-TCR results in β selection, i.e., selection of TCRβ-expressing cells for survival; other consequences include expansion, down-regulation of the IL-2R (CD25), expression of both coreceptor molecules, and induction of a second wave of RAG expression, allowing TCRα rearrangements to occur 2, 3 .

Ligation of the αβ TCR in mature T cells induces activation of Src family kinases, such as Lck and Fyn, as well as phosphorylation of the immunoreceptor tyrosine-based activation motifs (ITAM) of CD3 and ζ subunits 4 . The phosphorylation of the two tyrosines present in one ITAM results in recruitment of the tyrosine kinase ZAP-70 into the TCR/CD3/ζ complex through the binding of tandem SH2 domains 5, 6 . This association permits the enzymatic activity of ZAP-70, resulting in phosphorylation of downstream targets 7 . A similar sequence of events appears to take place also in signal transduction by the pre-TCR, since mice lacking both Lck and Fyn 8, 9 and mice lacking both ZAP-70 and Syk 10 exhibit a phenotype that is strikingly similar to the one of RAG^−/− mice 11 , implying a critical role for these molecules in the proximal signal transduction by the pre-TCR. The role played by CD3 and ζ-chains in pre-TCR assembly and signaling has been only partially elucidated. Studies with mutant mice have revealed an absolute requirement for CD3ε-chains 12 , no essential requirement for CD3δ-chains 13, 14 , and a major role for ζ-chains 15, 16, 17, 18 . ζ^−/− mice have a scarcely populated thymus and display an anomalous progression from the DN to the DP stage, in that the CD4⁻8⁻25⁻ stage that precedes the DP stage is absent 19 . Therefore, a specific role for ζ-chain in controlling proliferation and differentiation of DN cells has been hypothesized.

In RAG-2-deficient mice, the absence of TCRβ rearrangements leads to a developmental block at the DN CD25⁺ stage 11 . CD25⁺ cells express low levels of CD3γ, δ, and ε subunits at the surface 20, 21 . Efficient cellular proliferation, down-regulation of CD25, and production of small cortical cells that are characteristic of the physiologic transition driven by the pre-TCR can be induced in fetal thymus organ cultures by addition of anti-CD3ε mAb 22 and, in vivo, by injection of mice with anti-CD3ε Abs 20, 21 . Hence, CD3 appears functionally competent to promote these developmental steps in the absence of a fully assembled pre-TCR and in fact in the absence of the pre-TCR α-chain 23 .

We found that ζ-chains, despite the lack of detectable physical association with the surface CD3 complex, are functionally coupled to the signaling cascade initiated by anti-CD3 treatment of RAG-2^−/− thymocytes since they become associated with phosphorylated ZAP-70. This complex can be immunoprecipitated from the membrane fraction of anti-CD3-treated RAG-2^−/− thymocytes, suggesting its recruitment to the cell membrane. Moreover, the ζ-chain dimers that are not linked to CD3 can be detected in pre-TCR-expressing cells. The comparison of ZAP-70, p95^vav, and p120^cbl phosphorylation upon anti-CD3 treatment shows the same pattern in RAG-2^−/− thymocytes and pre-TCR-expressing cells, and differs from activated RAG-2^−/−ζ^−/− thymocytes in which ZAP-70 and p120^cbl phosphorylation is barely detectable. Therefore, an important role of the ζ-chain/ZAP-70 complex in mediating the transition from the DN to the DP thymocyte stage is postulated.

Young adult (4–6-wk-old) C57BL/6, RAG-2^−/−, RAG-1^−/−, and CB17 scid/scid (SCID) mice were obtained from the animal colony of the Insitut Pasteur (Paris, France). ζ^−/+ mice were kindly provided by Dr. B. Malissen (Centre d’Immunologie INSERM-CNRS, Marseille, France). SCB.29 24 and M14T 25 thymocyte cell lines were used. The mouse mAbs employed were anti-CD3ε 145-2C11 26 , anti-ζ G3 27 , and anti-phosphotyrosine 4G10 (Upstate Biotechnology, Lake Placid, NY). The following rabbit antisera were used: anti-calnexin C-terminal peptide (StressGen, Victoria, B.C.), anti-TCRζ (kindly provided by Dr. L. Samelson, National Institute of Health, Bethesda, MD), anti-ZAP-70 and anti-p95^vav (Santa Cruz Biotechnology, Santa Cruz, CA), and anti-p120^cbl (kindly provided by Dr. A. Veillette, University of Toronto, Canada).

For cell surface iodination, 50 × 10⁶ thymocytes obtained from C57BL/6 and RAG-2^−/− mice were treated with 0.1 mg/ml water-soluble Bolton-Hunter reagent (Pierce, Rockford, IL) in PBS at 4°C for 30 min, and the reaction was quenched by addition of 5% Ig-free FCS and 0.1 mg/ml lysine in HBSS 28 . Cell surface proteins were labeled with ¹²⁵I by the lactoperoxidase method 29 and extracted at 4°C in 1% digitonin or 0.5% Triton X-100 lysis buffer (0.15 M NaCl, 10 mM Tris, pH 8, 1 mM EGTA, 1 mM MgCl₂, and protease inhibitors); anti-CD3ε mAb was added to the lysates at the final concentration of 10 μg/ml and immune complexes were precipitated by protein A-Sepharose (Pharmacia Biotech, Uppsala, Sweden). Cell surface biotinylation was performed as described 30 . In brief, 50 × 10⁶ viable cells were washed in PBS and incubated in 1 ml of PBS, 5 mM NHS-LC-biotin (Pierce, Rockford, IL) for 1 h at 4°C. After washing in PBS, 1 mM glycine, the cells were lysed in 1% digitonin lysis buffer. Immunodepletion of CD3ε- or ζ-chain was achieved by three sequential immunoprecipitations of 1% digitonin cell lysates employing G3 and 145-2C11 mAbs, respectively, at 20 μg/ml, followed by protein A-Sepharose.

For tyrosine phosphorylation experiments, thymocytes at 10⁸ cells/ml were either left untreated or incubated at 4°C for 30 min with anti-CD3ε mAb. After three washes, cells were incubated at 37°C for 2 min, then anti-hamster Ig serum was added and cells were incubated at 37°C for 2 min. For in vivo stimulation, mice were i.v. injected with 50 μg of anti-CD3ε mAb and thymocytes were recovered at day 5 after mAb administration. Washed cells (>85% CD4⁺8⁺) were lysed in 0.5 or 1% Triton X-100 lysis buffer. The indicated rabbit antisera (2 μl) were added to the lysates, and immune complexes were precipitated by protein A-Sepharose, run under reducing conditions in SDS-PAGE, and immunoblotted as described below.

For two-dimensional nonreducing versus reducing SDS/PAGE, immunoprecipitates were run in SDS sample buffer under nonreducing conditions in a discontinuous Laemmli SDS-polyacrylamide (5–15% gradient) gel. The first dimension strips were then equilibrated in reduced SDS sample buffer for 30 min at room temperature and then run into a second 5–15% gradient SDS-polyacrylamide gel. The gel was then dried and subjected to autoradiography at −70°C. For two-dimensional IEF/PAGE, immune complexes were eluted in IEF sample buffer for 3 h at room temperature and then resolved by IEF in a horizontal apparatus (Pharmacia Biotech), followed by SDS-PAGE in a 5–15% gradient gel 31 . The gels were blotted in transfer buffer (100 mM glycine, 0.1% SDS, 10 mM Tris, 25% ethanol, pH 8.3) onto nitrocellulose membrane (Hybond-ECL; Amersham, Little Chalfont, U.K.), and the membranes were subjected to autoradiography at −70°C. To probe the transferred proteins with anti-calnexin rabbit antiserum, membranes were blocked for 1 h at room temperature in PBS containing 5% nonfat dry milk and 0.1% Tween-20 (PBS-milk), washed three times in PBS, 0.1% Tween-20 (PBS-Tween), followed by overnight incubation with rabbit antiserum at 4°C. After three washes (30 min each) in PBS-Tween, membranes were incubated for 1 h with horseradish peroxidase-conjugated anti-rabbit Igs in PBS-milk and washed three times in PBS-Tween. After incubation in ECL detection fluid, the blots were exposed to Hyperfilm ECL (Amersham). For immunoblot of total cell lysates, 10⁷ thymocytes were lysed in 100 μl lysis buffer containing 1% Triton X-100. The lysates were run on two-dimensional nonreducing versus reducing SDS-PAGE and immunoblotted with anti-ζ rabbit serum, as described above.

¹²⁵I-labeled anti-CD3ε immunoprecipitates were boiled twice: 2 min in 1% SDS and 2 min in deglycosylation buffer (20 mM sodium phosphate, pH 7.2, 10 mM NaN₃, 50 mM EDTA, 0.5% Nonidet P-40). One sample was stored at −20°C, another one was incubated at 37°C for 16 h, and a third one was incubated at 37°C for 16 h in the presence of 2 mU neuraminidase, 2.5 mU O-glycosidase, and 0.4 U N-glycosidase F (Boehringer Mannheim, Mannheim, Germany). The deglycosylation products were analyzed by SDS-PAGE.

After two washes in ice-cold PBS, C57BL/6 and RAG-2^−/− thymocytes were resuspended at 50 × 10⁶/300 μl in hypotonic buffer (20 mM Tris-HCl, pH 7.5, 1 mM EGTA, 1 mM MgCl₂, 0.5 mM DTT, and protease inhibitors) and incubated 10 min on ice. Cells were then disrupted by homogenization on ice with a Dounce homogenizer (30 strokes at low speed and 10 strokes at maximum speed). Salt concentration was adjusted to 150 mM NaCl, and intact cells, nuclei, and cytoskeleton were pelleted by centrifugation at 5000 rpm for 5 min in microcentrifuge (Eppendorf, Hamburg, Germany) at 4°C. After two washes in hypotonic buffer, the pellet (P1) was resuspended in Laemmli sample buffer. The low-speed supernatant was centrifuged at 100,000 × g for 30 min; the resulting pellet (P100) was considered the membrane fraction, and the supernantant (S100) was considered the soluble proteins fraction. The P100 fraction was dissolved in 0.5% Triton X-100 lysis buffer and immunoprecipitated either with anti-CD3ε or anti-ZAP-70 Abs, whereas the S100 fraction was directly immunoprecipitated. All samples were resolved by SDS-PAGE in a 5–15% gradient gel and immunoblotted with the indicated Abs.

Recently, incomplete endoplasmic reticulum retention has been demonstrated to take place in immature thymocytes and CD3 has been shown to be expressed at the cell surface, devoid of chains of the TCR for Ag and in association with calnexin, a molecular chaperone previously considered to reside exclusively in the endoplasmic reticulum 32, 33 . To characterize the subunit composition of CD3 on the surface of thymocytes from RAG-2^−/− mice, thymocytes from RAG-2^−/− and C57BL/6 mice were surface labeled by the ¹²⁵I-lactoperoxidase method and lysed in digitonin, and cell lysates were immunoprecipitated with anti-CD3ε mAbs. As evident from two-dimensional nonreducing versus reducing SDS/PAGE gel analysis (Fig. 1 A), two spots were missing in RAG-2^−/− mice: one (40–45 kDa) represents the reduced TCR α- and β-chains in C57BL/6 thymocytes, while the second (16 kDa) represents reduced ζ-chains. The latter were undetectable in RAG-2^−/− mice even after long exposures. Common to both precipitates are four spots that migrate either along (90, 26, and 21 kDa) or slightly above (25 kDa) the diagonal. While the 26, 21, and 25 kDa proteins were identified as δ, γ, and ε CD3 chains, the 90-kDa molecule is likely to be calnexin.

Indeed, the analysis of an anti-CD3ε immunoprecipitate from RAG-2^{−/− 125}I surface-labeled thymocytes by two-dimensional IEF/SDS-PAGE revealed a 90-kDa protein with an acidic isoelectric point together with the CD3δ, ε, and γ subunits at positions expected from their size and charge (Fig. 1,B). Immunoblotting with a rabbit anti-calnexin antiserum detected a protein in the same position representing the 90-kDa surface-labeled acidic molecule (Fig. 1,D). Treatment of the anti-CD3ε immunoprecipitates with peptide-N-glycosidase F and O-glycosidase did not provoke any shift in the electrophoretic mobility of the 90-kDa protein band, whereas the CD3γ and δ subunits were sensitive to the same enzymatic digestion (Fig. 1 C), as described 26 , being resolved as a single band with an apparent molecular mass of 20 kDa.

Although ζ-chains were detectable in total cell lysates from thymocytes of rearrangement-deficient mice (Fig. 2, C–F), we failed to detect ζ-chains that were associated with CD3ε molecules on the cell membrane of RAG-2^−/− thymocytes in many attempts utilizing distinct cell surface-labeling procedures, i.e., ¹²⁵I or biotin labeling, and using different conditions of lysis. Furthermore, immunoprecipitation of digitonin lysates from unlabeled thymocytes with anti-CD3ε mAb, followed by immunoblotting of immune complexes with rabbit anti-ζ antiserum, did not reveal any association of CD3ε molecules with ζ-chains even after injection of mice with anti-CD3ε mAb (data not shown). The same protocol has proven to be extremely sensitive in detecting CD3-associated ζ-chain, when performed with C57BL/6 thymocytes. These results are consistent with previous reports demonstrating that ζ-chains do not associate with calnexin during TCR-CD3 folding 34, 35, 36, 37 .

We then analyzed a possible functional recruitment of ζ-chains generated by cross-linking CD3ε on the surface of RAG-2^−/− thymocytes, in a way analogous to that employed in pre-TCR-expressing cells 38 : initially, immunoprecipitation with anti-ZAP-70 Abs of lysates of RAG-2^−/− thymocytes stimulated in vitro with anti-CD3ε mAbs did not reveal an enhancement of the weak constitutive phosphorylation of ZAP-70 (Fig. 3 A, lanes 1 and 2). However, phosphorylation of ZAP-70 as well as the p21 and p23 isoforms of phosphorylated ζ-chains were detected once the experiment was performed with thymocytes from RAG-2^−/− mice that were injected with anti-CD3ε mAb, with no significant variations upon restimulation in vitro (lanes 3 and 4). In control experiments, thymocytes from C57BL/6 mice were analyzed: phosphorylation of ZAP-70 as well as two ZAP-70-associated phosphoproteins were found. The two proteins with molecular mass of 24 and 28 kDa correspond most likely to CD3 chains. Furthermore, the expected increase in the constitutive phosphorylation of ZAP-70-associated ζ-chains was observed (lanes 5 and 6) 39 .

Since we failed to unambiguously detect phosphorylated CD3 chains associated with ZAP-70 in activated RAG-2^−/− thymocytes, we verified phosphorylation of CD3 chains following Ab injection, by anti-CD3ε mAb immunoprecipitation from the membrane fraction of RAG-2^−/− thymocytes lysed in 0.5% Triton X-100. We immunoblotted the immunoprecipitate resolved in two-dimensional nonreducing versus reducing SDS/PAGE gel, with anti-phosphotyrosine mAb. The resulting autoradiograph showed the presence of an intensely phosphorylated spot corresponding to CD3ε and two spots corresponding to CD3γ- and δ-chains (Fig. 3 B), which are absent in membranes from unstimulated RAG-2^−/− thymocytes. We conclude therefore that CD3γ-, δ-, and ε-chains are phosphorylated upon in vivo stimulation of RAG-2^−/− thymocytes, while phosphorylated ζ-chains appear to be the dominant recruitment site of ZAP-70 under these experimental conditions.

To evaluate the subcellular localization of the ZAP-70/ζ-chain complex in activated RAG-2^−/− thymocytes, we separated cell lysates from anti-CD3-injected mice into three fractions, containing nuclei and cytoskeleton (P1), membranes (P100), and cytoplasmic proteins (S100). As a control, we applied the same protocol to C57BL/6 thymocytes. The P1 fraction was directly processed for SDS/PAGE; the membrane fraction was dissolved in lysis buffer containing 0.5% Triton X-100, divided in two aliquots, and subjected to immunoprecipitation either with anti-ZAP-70 rabbit serum or anti-CD3ε mAb; the S100 soluble phase was directly immunoprecipitated either with anti-ZAP-70 rabbit serum or anti-CD3ε mAb. Fig. 3 C shows that we could selectively detect the p21 and p23 phosphorylated isoforms of ζ-chain in the anti-ZAP-70 immunoprecipitate of the membrane fraction derived from activated RAG-2^−/− thymocytes and not in the corresponding anti-CD3ε immunoprecipitate. Conversely, p21 phosphorylated ζ-chains were present in both the anti-CD3ε and anti-ZAP-70 immunoprecipitates of the membrane fraction from unstimulated C57BL/6 thymocytes, as expected. Therefore, phosphorylated ζ-chains were expressed at the cell membrane of activated RAG-2^−/− thymocytes in association with ZAP-70, but not with CD3.

The efficacy of cell fractionation and immunoprecipitation was checked by anti-ZAP-70 and anti-calnexin (in the case of anti-CD3ε immunoprecipitates of RAG-2^−/− thymocytes) immunoblots of the same filters. As can be seen in Fig. 3 C, at the indicated exposure times we could detect a major signal with anti-ZAP-70 serum in the immunoprecipitated S100 fraction, as shown 40 , and confinement of calnexin to the anti-CD3ε immunoprecipitate of the P100 membrane fraction of RAG-2^−/− thymocytes.

To address whether CD3-independent ζ-chains existed at the cell membrane also concomitantly with the pre-TCR complex, SCB.29 cells were surface labeled with biotin, lysed in digitonin, and immunoprecipitated with anti-CD3ε and anti-ζ-chain Abs. The results of these experiments show that ζ-chains were found in association with CD3γ-, δ-, and ε-chains (Fig. 4, lane 1), as well as isolated or very loosely associated with CD3, as evident by the poor representation of CD3 bands in anti-ζ-chain mAb immunoprecipitates (lane 3). The poor coimmunoprecipitation of CD3 proteins by anti-ζ mAb was not a peculiarity of the Ab employed, since the same mAb efficiently coprecipitated CD3 chains from C57BL/6 thymocytes and since anti-ζ rabbit Abs recognizing a different N-terminal ζ-chain epitope 41 gave identical results. The detergent was not responsible since the same results were obtained by lysing cells with Brij96 (data not shown). Finally, anti-CD3ε immunoprecipitation of lysates depleted of ζ-chain and anti-ζ immunoprecipitation after CD3 depletion showed that it was impossible to deplete either CD3 or ζ-chain by preclearing with anti-ζ and anti-CD3ε mAbs, respectively (Fig. 4, lanes 2 and 4). The efficiency of the preclearing protocol employed was checked in a thymocyte cell line (M14T) expressing mature TCR 25 , in which either anti-ζ or anti-CD3 mAb completely depleted precipitable CD3 and ζ-chains, respectively (lanes 6 and 8).

The prevalence of CD3-independent ζ-chains as recruitment site for ZAP-70 suggested that the ζ-chain/ZAP-70 complex could endow RAG-2^−/− thymocyte with efficient responsiveness to anti-CD3 stimulation and could contribute to pre-TCR-driven signaling. Therefore, we compared ZAP-70 phosphorylation of RAG-2^−/−, ζ^−/−, and pre-TCR-expressing cells stimulated by anti-CD3 Abs; in addition, we analyzed the phosphorylation of p95^vav and p120^cbl as downstream targets of Lck and ZAP-70 activation 42, 43, 44, 45, 46, 47 . In vivo stimulated RAG-2^−/− thymocyte and in vitro activated SCB.29 cells gave similar results, as all displayed markedly increased phosphorylation of all of the three molecules tested following anti-CD3 stimulation (Fig. 5, B and C). Additionally, anti-CD3 administration to RAG-2^−/− mice resulted in increased expression of p120^cbl (Fig. 5,B, lane 6). The same experiment performed with C57BL/6 thymocytes showed efficient phosphorylation of these targets under the experimental conditions employed (Fig. 5 A).

Surprisingly, we found that extensive cross-linking by administration of anti-CD3 mAb to RAG-2^−/−ζ^−/− mice resulted in barely detectable phosphorylation of ZAP-70 (Fig. 5,D, lane 2); moreover, p120^cbl was only weakly phosphorylated (Fig. 5,D, lane 6), and the amount of protein was unchanged with respect to the unstimulated counterpart. However, phosphorylation of p95^vav was apparently not affected and was readily detected (Fig. 5 D, lane 4). Thus, since anti-CD3ε stimulation of ζ^−/− mice results in thymocyte development analogous to the one observed in RAG-2^−/− mice 48, 49 , subtle differences in proximal signal transduction exist that are not critical for the final outcome of extensive CD3 cross-linking in the two mouse strains.

The analysis of ζ-deficient mice has revealed the requirement of ζ-chain to achieve thymocyte expansion once productive rearrangement of TCRβ genes has occurred 15, 16, 17, 18 ; more recently, the absence of ζ-chains has been shown to result also in inefficient allelic exclusion at the TCR β-chain locus 50 . Therefore, an essential function of the ζ-chain is exerted after rearrangement of TCRβ genes and is connected to pre-TCR assembly. Transcripts encoding ζ-chains have been detected very early in thymocyte development 51 ; in this study, we show the expression of a ζ-chain dimer in the absence of TCRβ rearrangement and its functional competence in recombinase-deficient thymocyte, in which CD3 chains are sequestered by calnexin and incompletely retained in the endoplasmic reticulum 32 . In pre-TCR-expressing cells, we were unable to deplete either surface ζ-chains or CD3 proteins by preclearing with anti-CD3 and anti-ζ Abs, respectively, implying that either ζ-chains are very loosely associated with the pre-TCR/CD3 complex or exist as a physically independent entity in addition to the fully assembled complex. The existence of ζ-chain dimers, in the absence of the pre-TCR and their membrane targeting independently of CD3, favors the latter interpretation of the results. Accordingly, in mature T cells, ζ-chains have been shown to be transported to the cell surface independently of the TCR complex, leading to the hypothesis that the TCR-CD3 complex is transported to the membrane by the ζ turnover pathway 52 . Since ζ-chains represent a rate-limiting factor in TCR assembly and surface transport 53, 54 , expression of ζ-chains before the pre-TCR might allow immediate and efficient assembly/transport of pre-TCR/CD3 complex. This hypothesis would be further strengthened by the detection of surface TCR ζ-chains in CD3ε^−/− thymocytes. However, the impossibility to stimulate in vivo these cells makes this demonstration particularly difficult to obtain.

In vivo stimulation through anti-CD3ε mAb of RAG-2^−/− thymocytes leads to phosphorylation of ZAP-70, which is found in association with phosphorylated ζ-chains. The detection of CD3ε-independent ζ-chain/ZAP-70 complexes in RAG-2^−/− mice as well as in pre-TCR-expressing cells 38 could imply a physiologic relevance for the functional recruitment of ζ-chain, not stably associated to pre-TCR complex. The dual role played by ζ-chains as rate-limiting factor in TCR assembly 53, 54 and as signaling module 55, 56 makes it difficult to dissociate these two functions during normal T cell development, but it is clear that in TCRβ-deficient mice, ζ-chains can be recruited at the cell membrane as ζ/ZAP-70 complex.

Restoration of TCR surface expression in ζ-deficient mice by truncated ITAM-less ζ-chains has been shown to overcome deficient signaling at this stage of development 57 . However, in an analogous experimental model, a transgenic mutant ITAM-less ζ-chain could not rescue differentiation of DN cells into DP cells of ζ^−/− mice, leading to the hypothesis that ζ-chains contribute signals that cannot be replaced by CD3 chains 58 . It has been hypothesized that CD3 subunits might become more accessible to Lck in the absence of full-length ζ-chains 59 ; indeed, we have shown that in RAG-2^−/− mice, CD3ε is not associated with ζ-chains and is phosphorylated following mAb injection. ZAP-70 is preferentially recruited by phosphorylated ζ-chains present at the cell surface physically unlinked to CD3 molecules. In the physiologic cellular environment, loosely associated and CD3-independent ζ-chain could constitute a preferential recruitment site for ZAP-70 and endow the cell with the optimal stoichiometry of available signaling complexes at this stage of thymocyte differentiation.

Efficient transition to the DP stage can be obtained in ζ^−/− mice by anti-CD3ε mAb administration 48, 49 , suggesting that ζ-chain recruitment in stimulated RAG-2^−/− thymocytes is nonessential and replaceable without functional impairment of the signaling cascade. However, it has been shown that a number of different stimuli, such as irradiation 60, 61, 62, 63 , Ras activation 64 , p53 inactivation 65, 66 , as well as abrogation of the Fas function 67 promotes the development of DP cells in rearrangement-deficient mice, thereby providing evidence for the existence of diverse signals that enable the same developmental program. The proximal signal-transduction events occurring in activated ζ^−/− thymocytes are subtly different from the ones taking place in stimulated RAG-2^−/− thymocytes and pre-TCR-bearing cells, but apparently apt in these circumstances, to determine the same developmental course. The efficient phosphorylation of p95^vav in the absence of marked ZAP-70 recruitment might be enabling and sufficient per se for such a differentiation to occur. Accordingly, Vav has been shown to be critically involved in thymocyte proliferation 68, 69, 70, 71 .

It has also been shown that coreceptor expression in immature thymocytes results in poor ZAP-70 phosphorylation after TCR stimulation due to diversion of available Lck; this effect has been postulated to be dependent on the association of Lck to cytoplasmic tails of coreceptors 72 . The absence of coreceptor expression at the DN stage would permit a free dispersion of the membrane-associated Lck pool, rendering it more available for ITAM phosphorylation and subsequent ZAP-70 activation. In addition, a basal Lck activity might also result in the observed weak phosphorylation of ZAP-70 before RAG-2^−/− thymocyte stimulation and be dependent on weak ζ phosphorylation. This hypothesis is supported by the absence of constitutive ZAP-70 phosphorylation in ζ^−/− mice. The pool of phosphorylated ZAP-70 molecules could lower the threshold for subsequent signaling by the assembled pre-TCR, rendering the cell appropriately sensitive. These potential functions of CD3-independent ζ-chains could depend on their association to developmentally regulated membrane proteins. Among potential candidates, Thy-1 could play a role since it has been shown to use ζ-chains for efficient signaling 73 . CD16 is expressed very early in thymocyte development and has been found associated to ζ homodimers 74 . Interestingly, CD16 and CD2 are coordinately expressed during T cell development with loss of CD16 and acquisition of CD2 expression characterizing a late DN stage immediately before the conversion into DP thymocytes 75 ; since CD16 could substitute for the TCR in coupling CD2 to signaling pathways by contributing ζ-chains 76 , it has been proposed that CD16 may serve a role similar to the TCR early in thymocyte ontogeny by coupling CD2 and Thy-1 to downstream signaling 77 . The efficient transition to DP stage observed in stimulated RAG-2^−/− thymocytes could benefit from the artefactual recruitment of these alternative forms of surface ζ-chain, thereby mimicking the assembly and signaling of pre-TCR.

We thank Bernard Malissen for providing ζ^−/+ mice; Larry Samelson for the generous gift of anti-ζ rabbit antiserum; André Veillette for the generous gift of anti-p120^cbl rabbit antiserum; Hung-Sia Teh for G3 hybridoma; Jean-Philippe Corre for ζ^−/+ mice breeding and screening; and Oreste Acuto, Michele Pelosi, Loretta Tuosto, and Robert Weil for helpful discussions and advice.