Unselected CD4+8+ rat thymocytes, generated in vitro from their direct precursors, are readily converted to functional TCRhigh T cells by stimulation with immobilized TCR-specific mAb plus IL-2. Lineage decision invariably occurs toward CD4−8+, regardless of the timing of TCR stimulation after entry into the CD4+8+ compartment or the concentration of TCR-specific mAb used for stimulation. CD4-specific mAb synergizes with suboptimal TCR-specific mAb in inducing T cell maturation, but lineage decision remains exclusively CD4−8+. These results contrast with those obtained in mice, in which Abs to the TCR complex were shown to promote CD4+8− T cell maturation from CD4+8+ thymocytes. Surprisingly, when rat and mouse CD4+8+ thymocytes were stimulated with PMA/ionomycin under identical conditions, the opposite lineage commitment was observed, i.e., mouse thymocytes responded with the generation of CD4+8− and rat thymocytes with the generation of CD4−8+ cells. It thus seems that CD4+8+ thymocytes of the two species respond with opposite lineage decisions to strong activating signals such as given by TCR-specific mAb or PMA/ionomycin. A possible key to this difference lies in the availability of p56lck for coreceptor-supported signaling. We show that in contrast to mouse CD4+8+ thymocytes, which express both a complete and a truncated CD8α-chain (CD8α′) unable to bind p56lck, rat thymocytes only express full-length CD8α molecules. Mice, but not rats, therefore may use CD8α′ as a “dominant negative” coreceptor chain to attenuate the CD8 signal, thereby facilitating MHC class II recognition through the higher amount of p56lck delivered, and rats may use a different mechanism for MHC class distinction during positive selection.
Tcell repertoire selection is controlled by the interaction of clonally distributed TCR molecules expressed on immature CD4+8+ thymocytes with MHC/peptide complexes at the surface of thymic cortical epithelial cells. The rescue of cells with self-MHC-restricted TCR is accompanied by lineage decision: thymocytes selected by recognition of MHC class I molecules down-regulate CD4 and become CTL precursors maintaining the coreceptor CD8 (1, 2), whereas those selected by MHC class II down-regulate CD8 and become Th cell precursors maintaining the coreceptor CD4 (3, 4).
Based on experiments in mice expressing class I or class II-restricted transgenic TCR and defective or transgenic MHC or coreceptor molecules, several models have been proposed to explain the coordination of TCR class specificity with CD4/8 lineage commitment during positive selection (for a recent summary, see 5 . These models contain elements that have been named “instructive” if MHC class recognition informs the CD4+8+ cell of the appropriate lineage decision (6); “stochastic” if lineage commitment is independent of MHC class recognition and coreceptor engagement, requiring a “selective” element in which coligation of the TCR with the appropriate coreceptor rescues only those cells that made the right choice (7); and “default” if failure to engage the TCR by one MHC class will result in commitment to the opposite lineage (8).
The concept of an instructive element in CD4/8 lineage decision has recently found support from coreceptor-domain shuffling experiments conducted in TCR-transgenic mice. In animals containing a chimeric coreceptor with an extracellular CD8α and an intracellular CD4 domain, CD4+8+ thymocytes expressing an MHC class I-restricted transgenic TCR were diverted into the CD4 subset during positive selection (9). Like CD4 itself, this chimeric coreceptor engages more of the nonreceptor kinase p56lck (Lck)3 than CD8 (10, 11, 12). Since Lck plays a central role in the initiation of TCR signaling (13, 14), it was proposed that the elevated amounts of Lck delivered to the TCR complex by the cytoplasmic CD4 domain of the chimeric coreceptor compared with the wild-type CD8 molecule during MHC I recognition resulted in an increased signaling strength that instructed the immature thymocyte to initiate CD4 lineage commitment (9). A mechanism in which quantitative differences in Lck delivery are used for MHC class identification during positive selection may also provide an explanation for the puzzling finding that in addition to the full-length CD8α molecule, mouse CD4+8+ thymocytes co-express a tail-less isoform of CD8α generated by alternative splicing that is unable to bind Lck (15, 16). These CD8α′/β heterodimers are lost from the surface after positive selection and could act as “dominant negative” molecules to attenuate the Lck signal during class I-mediated positive selection.
In our earlier studies on the development of rat thymocytes, we employed an in vitro system in which “virgin” CD4+8+ thymocytes are generated by overnight incubation of their direct precursors (17). These cells can be followed as a single cohort of synchronously differentiating cells, which express cell surface TCR and other markers of CD4+8+ thymocytes at the levels characteristic of their ex vivo isolated counterparts (18, 19). We have previously shown that stimulation of such “virgin” CD4+8+ thymocytes with immobilized TCR-specific mAb induces down-regulation of CD4 and CD8 (18) and expression of IL-2Rβ (20, 21). Inclusion of IL-2 rescues approximately 50% of input cells and, without proliferation, converts them within 2 days to phenotypically and functionally mature TCRhigh cells that are exclusively of the CD4−8+ phenotype (18). In these studies, the CD8 isoform detected was predominantly CD8αα rather than CD8αβ, which characterizes thymus-derived T cells (20). However, as will be shown below, CD8αβ cells are efficiently and exclusively produced in this system if conditions for stimulation are appropriately modified.
While in vitro generation of rat CD8 T cells from CD4+8+ precursors has proven very robust and efficient, we have been unsuccessful at reproducing these findings in mice using either freshly isolated or in vitro-generated CD4+8+ thymocytes (our unpublished observations). On the other hand, the differentiation of CD4+8−, but not CD4−8+, lineage cells from mouse CD4+8+ precursors by stimulation with TCR-specific (hybrid) mAb has been reported by a number of groups (22-25). To reconcile these divergent results obtained with rats and mice, we have tried to identify conditions for the in vitro induction of CD4 lineage commitment in rat CD4+8+ thymocytes and have, by replacing species-specific mAbs and cytokines with the nonspecific stimulants PMA and ionomycin, directly compared the response of rat and mouse CD4+8+ cells to a strong activating signal with regard to lineage commitment. Finally, we have investigated whether rat CD4+8+ thymocytes express CD8α′, a possible tool for MHC class discrimination during repertoire selection in mice.
Materials and Methods
Young adult Lewis rats and BALB/c mice of both sexes, bred at the Institute’s facilities, were used.
Rat-specific mAbs W3/25 and OX-35 (both anti-CD4), OX-8 (anti-CD8α), 341 (anti-CD8β), OX-44 (anti-CD53), and R73 (anti-TCRα/β), were obtained from PharMingen, San Diego, CA, and from Serotec, Oxford, U.K., as purified Ab or Ab conjugates. Mouse CD4-specific mAb L3T4 was from Boehringer, Mannheim, Germany, and 53-6.7 (anti-CD8) and H57-597 (anti-TCRα/β) were from PharMingen. Phycoerythrin-conjugated F(ab′)2 donkey anti-mouse Ig was obtained from Dianova GmbH (Hamburg, Germany), and streptavidin-Red670 was obtained from Life Technologies, Eggenstein, Germany.
Immunofluorescence and flow cytometry
For 2- or 3-color FACS analysis, 5 × 104 to 2 × 105 cells in 100 μl PBS/0.2% BSA/0.02% sodium azide were incubated for 15 min on ice with an unlabeled mAb, followed by a 15-min treatment with F(ab′)2 donkey anti-mouse phycoerythrin, 10 μg/ml normal mouse IgG (Sigma Chemicals Co., St. Louis, MO), and FITC- or biotin-conjugated mAb to the second marker. Finally, biotinylated mAb were developed with streptavidin-RED670 (Life Technologies). Flow cytometry was performed with a FACScan flow cytometer, using LYSYS II software for acquisition and Cellquest software for analysis (all from Becton Dickinson, Mountain View, CA). Routinely, 10,000 events were analyzed. Results are shown as log10 fluorescence intensities on a four-decade scale displayed as dot plots or histograms.
Preparation of cells
Immature rat CD4−8+ thymocytes were isolated by treating thymocyte suspensions with saturating amounts of R73 and W3/25 mAbs and removing the labeled cells by rosetting with rabbit anti-mouse Ig (Dakopatts, Hamburg, Germany)-coated sheep erythrocytes. The remaining cells were treated with OX35 and OX44 mAb, followed by sheep anti-mouse Ig ferritin particles (Miltenyi Biotec GmbH, Bergisch-Gladbach, Germany), before being passed through a magnetic-activated cell sorter (Miltenyi GmbH). The resulting population consisted of more than 99% immature CD4−8+ cells with a viability of >90% as determined by trypan blue exclusion. Peanut agglutinin (PNA) positive rat and mouse thymocytes were purified by panning on PNA (Sigma)-coated plates (20 μg/ml, carbonate buffer pH 9.5 for 24 h) in PBS, 5% FCS at 1 × 107 cells in 10 ml per 100 mm plate (Greiner, Sulzfeld, Germany) for 90 min at 4°C. After 3 washes, bound cells were eluted with 10 ml 0.2 M β-galactose in PBS/FCS for 10 min at room temperature, and washed twice in PBS/FCS.
A total of 5 × 105 thymocytes/ml of supplemented RPMI 1640 (18) were cultured in 24-well plates (Costar, Cambridge, MA). For TCR stimulation, culture wells were precoated overnight with rabbit anti-mouse Ig (40 μg/ml in carbonate buffer, pH 9.5) followed by a 2-h incubation with mAb R73 and/or OX35 in balanced salt solution at the concentrations given, and extensive washing. In all, 500 U/ml of human rIL-2 (a kind gift of Hoechst AG, Frankfurt, Germany) was added where given. Identical results were obtained using recombinant rat IL-2 (the kind gift of Dr. Mason, Oxford, U.K.). Where indicated, PMA and ionomycin (both from Sigma) were added at 0.4 ng and 0.2 μg/ml, respectively. In experiments not shown, the following cytokine preparations were included: 0.1% of rat IL-4 containing supernatant prepared from transfected Chinese hamster ovary cells (26) with a titer of 10−5 in an MHC class II induction assay using rat B cells; 50% culture supernatant from mouse IL-7-transfected 3T3 cells (the kind gift of Drs. Rolink and Melchers) that promoted recovery of immature rat thymocytes when used at 30% final concentration; 20% supernatant of Con A-stimulated peripheral T cells (CASUP) prepared from rat spleen stimulated for 24 h with 5 μg of Con A/ml at 107 cells/ml; and supernatant of rat thymic stromal cells prepared by dispase digestion of lymphocyte-depleted thymus fragments and cultured for 24 h at 106/ml.
Coreceptor re-expression assay
Conditions for pronase-stripping and coreceptor re-expression followed the protocol developed by Suzuki and colleagues (8).
Immunoprecipitation and Western blotting
Thymocytes (108/ml) or nylon wool-passed lymph node cells (6 × 107/ml) were lysed with 1% Nonidet P-40 in the presence of 1 mM Na3VO4, 20 mM NaF, 1 μg/ml leupeptin, and 1 μg/ml aprotinin, precipitated with OX-8 mAb or normal Ig using protein G-coupled Sepharose beads (Pharmacia, Uppsala, Sweden), and washed extensively with lysis buffer. Endo F-treatment was performed as described in reference 15, using 1 U Endo F (Boehringer, Mannheim, Germany) per sample. Reducing SDS-PAGE and transfer to nitrocellulose membranes followed standard procedures. Blots were probed with biotinylated OX-8 mAb and developed with horseradish peroxidase-conjugated streptavidin and the enhanced chemiluminescence system (Amersham, Braunschweig, Germany).
Rat CD8 T cells induced from virgin CD4+8+ thymocytes are de novo differentiated cells
Stimulation of rat virgin CD4+8+ thymocytes, generated in vitro from their direct precursors, with immobilized TCR-specific mAb and IL-2 converts about 50% of input cells within 2 days to phenotypically and functionally mature TCRhigh cells, which are exclusively CD4−8+ (18). Although the purity of input cells and absence of measurable proliferation during this phenotypic conversion made an expansion of contaminating mature CD8 cells highly unlikely, this is now formally excluded by the use of LEW.1F rats as thymocyte donors: in that strain, 3% of immature CD4+8+ cells express TCR utilizing the Vα8.2 segment, whereas due to RT1f-driven thymic overselection, 14% of mature CD8 T cells and thymocytes are Vα8.2+ (27). As shown in Figure 1, CD8 T cells derived by in vitro stimulation of LEW.1F virgin CD4+8+ thymocytes contain the same frequency of Vα8.2+ cells as their immature precursors, indicating that they were not derived from mature contaminants but generated by de novo differentiation.
Anti-TCR plus IL-2 induces CD8 lineage decision regardless of timing after entry into the CD4+8+compartment
Some models of positive selection and lineage decision of CD4+8+ thymocytes contain a “default” element in which a cell that fails to be engaged in one differentiation pathway spontaneously opts for the second one (Ref. 8, reviewed in 5 . We therefore asked whether TCR engagement at different timepoints after entry into the CD4+8+ compartment might reveal a loss of responsiveness to CD8 lineage instruction, and possibly a gain of CD4 lineage commitment. Immature CD4−8+ cells or their 1-, 2-, or 3-day-old CD4+8+ progeny were stimulated with immobilized TCR-specific mAb plus IL-2 and analyzed daily for coreceptor expression. As can be seen in Figure 2, this stimulation protocol invariably resulted in the induction of the CD4−8+ phenotype, even if the cells had spent their maximum lifetime of 3 days as CD4+8+ cells. Mature levels of TCR expression after release from TCR engagement and stability of CD4/8 phenotype after surface stripping and re-expression confirmed that lineage decision and maturation had indeed occurred (not shown, but see below). Thus, the capacity to enter the mature CD4−8+ subset is maintained in CD4+8+ rat thymocytes for the entire observation period in vitro, which coincides with the lifespan determined for the corresponding mouse subset in vivo (28, 29). Note also that CD4+8+ cells maintained in medium remained homogeneously CD4high8high, indicating that at least outside the thymic microenvironment, spontaneous down-regulation of one or the other coreceptor does not occur with time.
Thymic stromal cells, their supernatants, IL-4, IL-7, or CASUP do not induce lineage decision toward CD4
As one factor determining the unidirectional lineage decision observed in the present system, IL-2 was considered. Accordingly, we tested to determine whether other cytokines might allow the generation of CD4 cells. As potential candidates, rat IL-4 and IL-7 were tested in recombinant form, and in a broader approach, CASUP or supernatant obtained from freshly isolated thymic stromal cells were added to the cultures. With the exception of IL-7, which slightly improved cell recovery, and CASUP, which contained IL-2 and had the same effect, none of these additions affected the kinetics or outcome of in vitro differentiation, i.e., they did not lead to the appearance of CD4/8 single positive or TCRhigh cells when added by themselves or together with soluble or immobilized TCR-specific mAb, nor did they affect the differentiation of CD8 cells when added together with IL-2 (data not shown). Thus, using a limited panel of available rat cytokines and crude sources of unseparated supernatants, we have to date not been able to identify a cofactor that will direct the response to TCR stimulation by immobilized mAb toward the generation of CD4+8− cells.
Maintenance of CD8 lineage decision after reconstitution of suboptimal TCR stimulation by CD4 engagement
Due to their MHC class specificity, the CD4 and CD8 coreceptors may participate in lineage instruction of thymocytes undergoing positive selection. To investigate whether mAb-mediated co-ligation of TCR and CD4 would induce lineage decision toward CD4 rather than CD8, virgin CD4+8+ thymocytes were stimulated with immobilized TCR-specific mAb and IL-2 over a wide range of concentrations in the presence or absence of coimmobilized CD4-specific mAb. After 2 days of stimulation, cells were rested for 1 day to allow up-regulation of the TCR and analyzed for the expression of CD4, CD8α, CD8β, and TCR-αβ. As shown in Figure 3, the fraction of CD4+8+ immature thymocytes induced to differentiate increased with the concentration of TCR-specific mAb employed, resulting in the generation of CD4−8+ TCRhigh cells. Moreover, while CD4−8+ T cells generated by low concentrations of TCR-specific mAb were of the CD4− CD8α+β+ phenotype, those obtained with high concentrations of TCR-specific mAb contained many CD4−8α+β− cells. This is in line with our previously published experiments in which even higher concentrations of TCR-specific mAb were routinely used and led to the predominant induction of the CD4−8α+β− phenotype (18, 20). When suboptimal amounts of TCR-specific mAb were employed, the reduced number of cells addressed was strongly increased by the coimmobilization of CD4-specific mAb. At the same time, however, CD4 costimulation did not induce a CD4+8− phenotype but rather resulted in an increased yield of CD4−8α+β+ TCRhigh cells. In additional experiments (not shown), coligation of TCR and CD4 by mAb was again combined with the cytokine supplements listed above. With the exception of IL-2 and of CASUP, which allowed the differentiation of TCRhigh CD4−8+ cells, none of these additions allowed T cell maturation of anti-TCR- plus anti-CD4-stimulated cells in this in vitro system. Also, as expected, the same synergism described in Figure 3 for TCR- and CD4-specific Abs was also observed when instead of anti-CD4, anti-CD8 mAb were employed (not shown).
Opposite lineage decisions induced in mouse and rat CD4+8+ thymocytes by PMA/ionomycin
The ease with which mature CD4−8+ T cells are induced from immature rat CD4+8+ precursors by stimulation with TCR-specific mAb plus IL-2 prompted us to repeat these experiments in mice. Despite extensive variations in cytokine supplements, all attempts to generate mature T cells from TCR-stimulated mouse CD4+8+ thymocytes in single-cell culture remained unsuccessful (data not shown). However, other groups have shown that mouse TCR-specific mAb or hybrid Abs are able to induce maturation of CD4 T cells in vivo (22, 23) or fetal thymic organ culture (FTOC) (24), and that TCR-specific mAb coimmobilized with mAb to various thymocyte cell surface molecules induces CD4 lineage commitment in single-cell cultures of mouse CD4+8+ thymocytes (25). Furthermore, transient treatment of CD4+8+ mouse thymocytes with the protein kinase C activator PMA plus the Ca-ionophor ionomycin was shown to result in the generation of CD4+8− thymocytes from CD4+8+ precursors in vitro (30). Together, these results suggested that a “strong” signal such as those provided by the high-affinity interactions of Abs with TCR or by forced protein kinase C activation and elevation of Ca2+ levels is perceived by mouse CD4+8+ thymocytes as a signal to enter the CD4 lineage, and that rat CD4+8+ thymocytes might respond with the opposite decision. To directly test this possibility, CD4+8+ mouse and rat thymocytes were enriched by panning on PNA-coated plates and subjected to a 20-h PMA/ionomycin pulse followed by 20 h of rest in medium. After PMA/ionomycin treatment, an aliquot of the cells was subjected to coreceptor stripping by pronase treatment (8), and the re-expressed CD4 and CD8 molecules were analyzed after the resting period. As shown in Figure 4, A and B, mouse and rat CD4+8+ thymocytes responded to this artificial stimulus with the expression of opposite CD4/8 phenotypes: in agreement with published results, CD4+8− cells were obtained from mouse CD4+8+ thymocytes. The parallel rat cultures did not generate this subset but yielded CD4−8+ cells. Again, these CD8 cells expressed both CD8α- and β-chains (data not shown). Interestingly, CD4−8+ and CD4+8− thymocytes generated in this fashion from mouse and rat CD4+8+ cells, respectively, contained both TCRhigh and TCRlow/negative cells (Fig. 4 C). This suggests that circumvention of TCR signaling by PMA/ionomycin induces lineage commitment even in those CD4+8+ cells that have failed to express appropriately rearranged and pairing TCRα-chains, and argues against selective survival or expansion of mature cells contained in the PNA+ thymocyte preparations, which always contain some single positive cells. In the case of PMA/ionomycin-driven differentiation of rat cells, the contribution of contaminating mature CD4−8+ cells to the observed enrichment of the CD4−8+ phenotype was further excluded by the use of the LEW.1F strain as the thymocyte donor and verification of preselection Vα8.2 usage as described above (not shown). These data indicate that immature CD4+8+ thymocytes from rat and mice are programmed for opposite lineage decisions in response to the strong stimulus provided by the pharmacologic agents PMA and ionomycin, and possibly also to physiologic, TCR/coreceptor-mediated signals.
Rat thymocytes do not express a truncated form of CD8α
Based on the above results, we considered the possibility that mouse CD4+8+ thymocytes are programmed to respond to a strong signal with lineage decision toward CD4 because under physiologic conditions, more Lck will be delivered by CD4 than by CD8, and consequently, this may not hold true for their rat counterparts. Since in mice, a tail-less CD8α′ isoform is exclusively expressed in immature mouse thymocytes (16), where it may attenuate the CD8-mediated Lck signal at that stage, the existence of a truncated CD8α molecule was investigated in rats by Western blotting using a mAb (OX-8) that recognizes the CD8α membrane-proximal hinge region. Immunoprecipitates from lysates of rat thymocytes and of peripheral T cells were either directly examined or treated with endoglycosidase F after immunoprecipitation with OX-8 to improve resolution. As can be seen in Figure 5, a single band migrating at the apparent m.w. expected for the intact CD8α-chain was observed in both thymocytes and peripheral T cells. Thus, in contrast to results obtained with mouse CD4+8+ thymocytes, where the CD8α′ molecule is readily detected biochemically (16), this molecular species is absent from the rat.
The experiments presented in this communication indicate that CD4+8+ thymocytes from mice and rats differ in their interpretation of external stimuli that induce CD4/8 lineage commitment. This was directly demonstrated by employing identical conditions for isolation and stimulation of CD4+8+ cells from the two species, i.e., enrichment by PNA binding and activation with PMA and ionomycin. This approach eliminates potential differences in signal delivery that may be encountered with species-specific mAbs or cytokines. Confirming the results of others, we found that a 20-h pulse treatment with PMA/ionomycin induced down-regulation of CD8 in CD4+8+ mouse thymocytes (30); in contrast, CD4+8+ thymocytes from rats down-regulated CD4 under the same conditions.
These results are in agreement with those obtained with mAb-mediated TCR stimulation. In the rat, maturation of both CD8αα (18, 20) and CD8αβ T cells (this paper) is easily induced by stimulating CD4+8+ thymocytes with TCR-specific mAb and IL-2 (18), but this protocol is unsuccessful in mice (our unpublished observations). In that species, however, a number of studies have demonstrated lineage decision toward CD4 if TCR-specific mAb were employed to drive positive selection. In the experiments by Müller and Kyewski, hybrid mAb cross-linking the TCR to the surface of thymic epithelial cells efficiently induced CD4, but not CD8 T cell maturation in vivo (31), even if MHC II or CD4 were absent (23). In a similar approach employing FTOC from class II-deficient mice, Takahama and coworkers found that TCR- and CD3-specific mAb rescued CD4 T cell development (22). Finally, Bommhardt et al. recently reported that cross-linking of CD3 to either CD4 or CD8 using recombinant F(ab′)2 Abs drives CD4 T cell development in FTOC in an MHC-independent fashion (24). In addition to these experiments in which TCR-specific mAb induce CD4 T cell differentiation in an intact thymic microenvironment, an in vitro differentiation system for mouse CD4+8+ thymocytes was recently described in which coimmobilization of mAb with one of several thymocyte cell surface molecules supported anti-TCR-driven T cell maturation in single-cell culture. Again, the phenotype of differentiated cells was exclusively CD4+8− (25).
At first sight, these and our present results, in which CD4- and TCR-specific Ab synergized in the generation of CD8 T cells from rat CD4+8+ thymocytes, contradict instructive models of lineage commitment because they show that neither the “correct” MHC class nor the relevant coreceptor need to be engaged for efficient conversion of mouse or rat CD4+8+ precursors to CD4 or CD8 T cells, respectively.
We do, however, interpret these results in support of the “signal strength” hypothesis of lineage instruction, and postulate that while mouse CD4+8+ thymocytes interpret a strong signal as MHC II recognition and thus commit toward CD4, rat CD4+8+ thymocytes are differently programmed and respond to a strong signal with CD8 commitment. This hypothesis is based on the notion that due to the manyfold higher stability of TCR-mAb complexes compared with those formed between TCR and MHC/peptide in positive selection (32), stimulation with TCR-specific mAb will lead to the formation of stable TCR aggregates on which the intracellular signaling machinery, including Lck, will efficiently assemble even in the absence of coreceptor engagement, thereby providing a strong signal. Indeed, peripheral T cells, which require coreceptor participation in Ag-driven responses, are readily activated with TCR-specific mAb without the need for recruitment of coreceptors to the TCR complex (33), and synergism between TCR and coreceptor-specific mAb only becomes apparent under conditions of suboptimal stimulation by TCR-specific mAb (34).
Although not addressed experimentally, we consider it likely that also in positive selection of thymocytes driven by TCR-specific mAb, the strong signal is delivered via Lck, even if the coreceptors are not engaged. Coreceptor-independent participation of Lck in mAb-induced TCR signaling has been demonstrated in mature T cell lines lacking these molecules (35), and Lck is not exclusively associated with CD4 and CD8. In fact, CD4 T cell maturation through a strong, Lck-mediated signal may actually be favored in experiments using TCR-specific mAb and MHC II-deficient mice (22, 23), because the amount of available free Lck is increased in the absence of constitutive CD4/MHC II interactions (36).
Even in thymic selection driven by the physiologic engagement of the TCR by MHC molecules, coreceptor-mediated Lck delivery is not an absolute prerequisite, as shown by the low-level thymic maturation of CD4 T cells, which proceeds in mice expressing CD4 without the Lck binding cytoplasmic domain (37), and by the ability of an Lck mutant lacking CD4/8 interactions to promote thymocyte maturation (38). Although the latter findings appear to contradict a role of coreceptor-delivered Lck in lineage instruction, it seems possible that in its absence, higher-affinity receptors are selected that can initiate the signaling cascade by coreceptor-independent recruitment of Lck.
In mice, a difference in the amount of coreceptor-mediated Lck delivery to the TCR may be achieved in two ways: an intrinsically lower capacity of the intracytoplasmic domain of CD8α compared with that of CD4 to bind this kinase, and the expression of a truncated CD8α isoform (CD8α′) on CD4+8+ mouse thymocytes. It is intriguing that these molecules, which make up about 40% of CD8α, are expressed at the cell surface exclusively in the selectable CD4+8+ compartment, but are intracellularly degraded in mature CD8 T cells (15). We speculate that in mouse CD4+8+ thymocytes, they may act as dominant negative molecules to attenuate CD8-mediated Lck delivery and thereby increase the sensitivity of MHC class discrimination through signal strength. Indeed, in the experiments by Itano and colleagues, diversion of thymocytes with class I-restricted transgenic TCR into the CD4 subset was not only observed in the presence of a chimeric CD8/4 molecule, but was, to a lesser degree, also facilitated by a CD8α transgene unable to provide the truncated CD8α splice variant (9). Our present finding that rat CD4+8+ thymocytes, which apparently respond to strong signaling with lineage decision toward CD8, lack the truncated CD8α isoform, is compatible with the hypothesis that sensing of MHC class through signal strength is reversed between the two species. In addition to the biochemical evidence presented, inspection of the CD8α sequence in rats and humans reveals that in neither of the two species does the potential exist for the generation of a tail-less CD8α isoform by alternative splicing. Rather, omission of exon IV, which in mice results in the formation of CD8α′ lacking both exons IV and V due to a stop codon generated by frame-shift at the beginning of exon V, would, in these species, yield very large cytoplasmic tails without recognizable homology to known protein domains (not shown). Therefore, neither rats nor humans attenuate the CD8-dependent Lck signal in immature thymocytes by a truncated CD8α isoform, and thus cannot use this hypothetical mechanism of MHC class discrimination in positive selection.
Although the finding that mAbs to the TCR and to CD4 synergize in the generation of rat CD8 T cells in vitro fits well with a species-specific modification of the signal strength model as outlined above, the additional requirement for IL-2R signaling in that system does not permit the assignment of the CD8 lineage decision solely to the intensity of the TCR/coreceptor signal perceived. Thus, we have previously shown that the in vitro generation of rat CD8 T cells by stimulation with TCR-specific mAb and IL-2 can be divided into two steps: first, TCR stimulation leads to down-regulation of CD4 and CD8, and de novo expression of the IL-2Rβ-chain (20) in all TCR+CD4+8+ thymocytes (21). IL-2 then rescues these cells and induces up-regulation of CD8, but not CD4, as well as phenotypic and functional maturation (18, 20). Accordingly, one may argue that TCR stimulation primed TCR+CD4+8+ thymocytes for positive selection but was neutral with regard to lineage decision, which was guided toward CD8 by IL-2, but might have occurred toward CD4 if some different unknown second signal had been provided. Our failure to identify such an instructional second signal for rat CD4 commitment in cytokine preparations from peripheral lymphoid cells and thymic stromal cells does, of course, not exclude its existence. On the other hand, the induction of IL-2Rβ expression on all TCR+CD4+8+ thymocytes by stimulation with TCR-specific mAb and their IL-2 dependent conversion to CD8 T cells indicate that together, these two signals instruct CD8 lineage decision in rats and argue against stochastic lineage precommitment in rat CD4+8+ thymocytes.
The mechanism by which IL-2R signaling promotes differentiation of rat CD8 T cells remains to be analyzed. In addition to the induction of genes via pathways involving janus-kinases and signal transducers and activators of transcription (39), the activation of Lck in signal transduction through the IL-2Rβ-chain (40, 41) might support or prolong the hypothetical strong Lck signal guiding CD8 development in rats. In the context of the divergent lineage decisions of rat and mouse CD4+8+ thymocytes presently described, it is of interest that IL-2R lacking the α-chain (IL-2Rβγ) are functional in rats (20) and humans (42), but not in mice (43). Therefore, rats and humans, but not mice, have the potential to use signaling through the IL-2Rβγ in CD8 commitment. From this point of view, it is not surprising that an essential contribution of the IL-2/IL-2R system to mouse CD8 T cell generation has been excluded by gene targeting (44, 45, 46, 47), but it may be premature to generalize this result to other species. Moreover, it is possible that IL-15, known to employ the IL-2Rβγ for signal transduction (48) and to be produced by thymic epithelial cells (49), is of relevance in vivo, rather than IL-2.
In the system of in vitro differentiation of rat virgin CD4+8+ thymocytes (18) presently employed, a homogeneous population of synchronously differentiating cells that has not received TCR-mediated selection signals in vivo is followed over time. Although lacking the natural microenvironment, this system closely mimics the in vivo situation with regard to kinetics of constitutive and inducible expression of cell surface receptors (19) and regulation of recombination-associated gene-1 and TCR mRNA (H.-J. Park, and T. Hünig, unpublished observations). Of note, none of the phenotypic intermediates with regard to coreceptor expression that have been observed in the mouse thymus and have given rise to complex models of positive selection and lineage decision (7, 8, 50, 51) are spontaneously formed even during prolonged culture of virgin CD4+8+ cells in vitro, suggesting that they are the result of in vivo interactions with stromal cells. With regard to the kinetics of CD4/8 commitment, we found that rat virgin CD4+8+ cells retain the potential to be converted to a CD4−8+ cell for several days, arguing against a default element in lineage decision in which a cell would spontaneously switch to the opposite lineage commitment in the absence of an instructing signal (8).
An interesting side aspect of the present work is that in vitro stimulation of rat virgin CD4+8+ cells with increasing doses of immobilized TCR-specific mAb and IL-2 results in a shift from the CD8αβ phenotype characteristic of thymus-derived T cells to the CD8αα isoform expressed by many gut-associated T cells and by activated rat CD4 T cells (52). Thus, a supraoptimal TCR-mediated signal may divert CD4+8+ in cells from their mainstream differentiation pathway into the CD8αα subset.
In summary, mouse and rat CD4+8+ cells respond to stimulation with TCR and coreceptor-specific mAb and to PMA/ionomycin treatment with opposite lineage commitment, suggesting that also under physiologic conditions of repertoire selection, the same stimulus is interpreted differently. It will be of interest to see whether, as is presently hypothesized, human and rat CD4+8+ cells generate a stronger signal in MHC I vs II recognition, and how similar initial signals may be translated into opposite lineage choices in the different species analyzed.
We thank Anneliese Schimpl and Thomas Herrmann for critically reading the manuscript and Uschi Bommhardt for reminding us of the existence of CD8α′ in mouse thymocytes.
This work was supported by the DFG through SFB 165, and by Fonds der Chemischen Industrie e.V.
Abbreviations used in this paper: Lck, p56lck; PNA, peanut agglutinin; CASUP, supernatant of Con A-stimulated peripheral T cells; FTOC, fetal thymic organ culture.