Autonomous Maturation of α/β T Lineage Cells in the Absence of Cooh-Terminal Src Kinase (Csk)

The normal development of α/β T cells is controlled by signals originating from the pre-TCR, TCR, and coreceptors CD4 and CD8. Pre-TCR–derived signals are necessary for the progression from CD4⁻CD8⁻ double-negative (DN) to CD4⁺CD8⁺ double-positive (DP) thymocytes (for a review, see reference 1). Signaling from the TCR and coreceptors are required for the complex maturation of DP thymocytes into CD4⁺ or CD8⁺ single-positive (SP) cells (for reviews, see references 2 and 3). TCRs and CD4 or CD8 coreceptors on DP thymocytes interact with MHC/peptide complexes present on thymic stromal cells. If this interaction is of sufficient avidity, the DP cells receive a survival signal and are positively selected to proceed in development to the SP stage. In the case of too strong MHC/peptide–TCR interaction, the cells are negatively selected and die by apoptosis, similar to those DP cells which receive no survival signal for lack of MHC/peptide–TCR interaction (for a review, see reference 3). Positive selection is accompanied by the commitment to either the CD4 or CD8 lineage of T cells. The outcome of lineage commitment and positive selection is coordinated, as usually CD4 SP thymocytes are self-MHC class II restricted, whereas CD8 SP thymocytes are restricted to self-MHC class I molecules. Consequently, mice, which cannot express MHC class I, MHC class II, or both MHC class I and class II molecules, lack CD8 T cells, CD4 T cells, or both CD4 plus CD8 T cells, respectively 4,5,6,7,8,9.

The first biochemical changes detectable upon TCR engagement are the activation of the Src family protein tyrosine kinases (PTKs) Lck and Fyn, which physically associate with CD4/CD8 coreceptors and the TCR, respectively 10,11,12. The combined deletion of lck and fyn genes, as well as the overexpression of a dominant negative lck transgene in mice result in the arrest of thymocyte maturation at the DN stage of development 13,14,15. Conversely, expression of constitutively active Lck or Fyn drives recombination activating gene (rag)-deficient DN thymocytes to mature to the DP stage of development, and activated Lck is implicated in mediating allelic exclusion at the TCR β locus 14,16,17. Hence, Lck and Fyn are involved at least in the early steps of α/β T cell developmental leading to the generation of DP thymocytes. It is unknown whether the activation of Lck and Fyn in the presence of TCR and coreceptors is sufficient to generate all signals required for the complete developmental program of thymocyte maturation.

To investigate if Src family PTKs are involved in all steps of α/β T cell development, we took the approach of conditional inactivation of the negative regulator of Src family PTKs (COOH-terminal Src kinase [Csk]) during α/β T cell development. We previously showed that α/β T cells can develop from csk-deficient DN thymocytes in the absence of pre-TCR and α/β TCR 18. This TCR-independent T cell maturation passes through all stages of thymocyte development and is accompanied by an increase in the specific activity of Lck and Fyn in csk-deficient thymocytes. From our previous analysis, it remained unclear whether γ/δ TCR–derived signals contribute to the α/β TCR–independent generation of T cells. Furthermore, Csk substrates other than Lck and Fyn could be responsible for the autonomous T cells maturation. Finally, it is unknown whether the deletion of csk in thymocytes that express a TCR causes an uncoupling of TCR engagement from thymocyte maturation, or if TCR-derived signals dominate over the activation of Lck and Fyn caused by the absence of Csk.

Here we provide evidence that α/β T cell development in the absence of Csk is also independent of γ/δ TCR expression and that no other substrate of Csk is sufficient to drive T cell development in the absence of Lck and Fyn. Furthermore, we show that the positive selection of csk-deficient thymocytes is largely uncoupled from TCR-derived input on the developmental fate, whereas TCR-mediated negative selection appears normal in the absence of Csk.

All mice used in these experiments were maintained in a conventional animal facility and analyzed at 6–8 wk of age. Maintenance and analysis of mice were done according to the German animal protection law. The generation of csk^fl/fl mice has been described 18. Csk^fl/fllck-cre, csk^fl/flrag-1^−/−MX-cre csk^fl/fl β2m^−/−Aα^−/−lck-cre, and csk^fl/fllck-creHY-TCRtg mice are on mixed genetic background of 129, CB.20, and C57Bl/6. csk^fl/fllck-creHY-TCRtg mice were bred onto H-2^b or H-2^d MHC background. MX-cre mice were provided by R. Kühn and K. Rajewsky (University of Cologne, Cologne, Germany). Lck-cre mice were provided by J. Marth (University of California San Diego, San Diego, CA). Aα^−/− and β2m^−/− mice on C57Bl/6 genetic background were provided by H. Bluethmann and H.J. Fehling (Roche, Basel, Switzerland), respectively. rag-1^−/− mice on C57Bl/6 genetic background were obtained from The Jackson Laboratory. Lck^−/− mice on C57Bl/6 genetic background and fyn^−/− mice were provided by T. Mak (Amgen, Toronto, Canada) and R. Perlmutter (Merck, Rahway, NJ), respectively.

Mice were killed with CO₂ and lymphoid organs removed. Single cell suspensions were prepared by gentle tearing of spleens, lymph nodes, or thymi in RPMI 1640 (GIBCO BRL) containing 10% FCS (Roche). When necessary, erythrocytes were lysed by incubating cells in red blood cell lysis buffer (0.75% NH₄Cl, 100 mM Tris/HCl, pH 7.65) for 2–4 min at room temperature. Lysis was stopped by adding 10 ml of compete RPMI.

Fluorescence staining was performed as described 19. In brief, 1–2 × 10⁶ cells were washed in PBS/1% BSA/0.01% sodium azide (PBA) and incubated with biotinylated antibodies at saturating concentrations in PBA for 20 min at 4°C. After washing with PBA, cells were incubated in PBA containing fluorochrome-conjugated antibodies and streptavidin for 20 min at 4°C. After a final washing step, cells were resuspended in 100–300 μl PBA and analyzed on a FACSCalibur™ flow cytometer (Becton Dickinson). For the exclusion of dead cells during analysis, propidium iodide (PI; 0.2 μg/ml; Molecular Probes) or TO-PRO-3 (1 nM; Molecular Probes) was added to the cells before acquisition. The following reagents were purchased from BD PharMingen: CD3ε (145-2C11), CD4 (RM4-5), CD5 (53-7.3), CD8α (53-6.7), CD24/HSA (M1/69), CD25 (7D4), CD69 (H1.2F3), CD90/Thy1.2 (53-2.1), TCR β (H57-597), and streptavidin-cychrome. Streptavidin-Cy7–PE was purchased from Medac. Biotinylated T3.70 and FITC-coupled F23.1 antibodies were gifts from H. von Boehmer (Dana-Farber, Boston, MA). Antibodies against B220 and Mac-1 were produced at the Institute for Genetics (Cologne, Germany).

CD4 T cells were purified by MACS depletion with anti-CD8, anti-B220, and anti–Mac-1 antibodies according to the manufacturer's instructions (Miltenyi Biotec), followed by FACS^®-assisted enrichment of CD4^hi and CD4^lo cells on a FACStar™ (Becton Dickinson) with anti-CD4 and anti-CD8 antibodies. The purity of sorted cells was >95%.

Genomic DNA was isolated from 10⁶–10⁷ cells in suspension following the protocol by Laird et al. 20, followed by phenol/chloroform extraction. Southern blot analysis was done by standard procedures 21. Genomic DNA was digested with EcoRI, size fractionated on 0.7% agarose gels, and blotted onto nitrocellulose membranes (Geenscreen; NEN Life Science Products) by capillary transfer. Csk-specific fragments were detected with the radiolabeled probe C 18.

We have previously shown the TCR-independent development of csk-deficient α/β T lineage cells in csk^fl/flTCRβ^−/−MX-cre mice. As these mice can express γ/δ TCR, we wanted to exclude a possible involvement of γ/δ TCR–derived signaling in the observed phenotype. Mice were generated to carry homozygous null mutation for rag-1 in combination with two loxP-flanked csk alleles and the MX-cre transgene (csk^fl/flrag-1^−/−MX-cre; references 18, 22, and 23). The MX-cre transgene is interferon inducible, but shows basal activity which results in the expression of Cre in a small fraction of all thymocyte populations including DN cells, as determined by Southern blot analysis of Cre-mediated deletion of a single loxP-flanked csk allele in these cells 18. Thymi of Csk^fl/flrag-1^−/−MX-cre mice without injection of interferon are ∼10-fold larger than thymi of rag-1^−/− mice and close to the size of control thymi of csk^fl/fl rag-1^+/− mice (Fig. 1 A). Southern blot analysis of total thymocyte genomic DNA provides evidence that the thymus of csk^fl/flrag-1^−/−MX-cre mice consists of exclusively csk-deficient cells (Fig. 1 B). The analysis of csk-deficient thymocytes reveals the presence of DP and CD4-SP cells. The fraction of DP thymocytes is slightly decreased and the percentage of CD4-SP cells is increased compared with csk^fl/flrag-1^+/− control mice. The analysis of thymocyte maturation markers on DP and CD4-SP cells of csk^fl/flrag-1^−/−MX-cre mice shows increases in CD69 and CD5 surface expression on CD4-SP cells as signs of positive selection and also decreased CD24/HSA surface expression on CD4-SP cells, which indicates further thymocyte maturation (Fig. 1 C). The fraction of CD4-SP cells as well as CD69⁺HSA^lo cells is increased in csk^fl/flrag-1^−/−MX-cre mice compared with csk^fl/flrag-1^+/− control mice (Fig. 1A and Fig. D). CD8-SP cells are largely absent in csk^fl/flrag-1^−/−MX-cre mice. Surface expression levels of CD4 on DP and CD4-SP thymocytes are reduced in csk^fl/flrag-1^−/−MX-cre mice, compared with similar cells of control mice (Fig. 1 A). These csk-deficient cells leave the thymus to form a sizeable peripheral compartment of CD4^lo cells in the spleen and lymph nodes (Fig. 1 D, and data not shown). Surface expression of CD90/Thy1 and CD5 suggests the T lineage identity of these CD4^lo cells (Fig. 1 D, and data not shown).

Considering that T lineage development from csk-deficient thymocytes is possible without any TCR, the question arises whether the TCR can at all influence the fate of csk-deficient thymocytes. To address this issue, T lineage development in csk^fl/fllck-cre mice was analyzed. In these mice, the cre-transgene is under the control of the proximal lck promoter. As expected from earlier studies 24,25, this promoter limits Cre expression and csk deletion to DN and DP thymocytes (data not shown). The numbers of thymocytes and splenic T cells in csk^fl/fllck-cre mice are comparable to csk^fl/fl control mice (Table). Analysis of thymocyte subpopulations in csk^fl/fllck-cre mice reveals that the percentages of DP and CD8-SP cells are slightly reduced, whereas CD4-SP and DN fractions are about twofold increased compared with thymi of csk^fl/fl control mice (Table, and Fig. 2 A, top). Changes in the expression of thymocyte maturation markers CD69, CD5, and CD24/HSA are similar to those described for csk^fl/flrag-1^−/−MX-cre mice (data not shown). A skewing of the CD4/CD8 cell ratio is also apparent in peripheral T cells of csk^fl/fllck-cre mice, which have strongly reduced numbers of CD8 T cells (Fig. 2 A, bottom, and data not shown).

The analysis of TCR surface expression on α/β T lineage cells in csk^fl/fllck-cre mice reveals a small population of CD4^loTCR⁻ cells present in csk^fl/fllck-cre mice (Table and Fig. 2 B). These cells are likely to be descendents of DN thymocytes that acquired the csk deletion before a successful rearrangement of their TCR β genes. Overexpression of an Lck transgene expressed in DN thymocytes has been reported to mediate allelic exclusion at the TCR β locus 17. The activation of Lck as a consequence of csk deletion in DN thymocytes is expected to act similarly and therefore suppress TCR β gene rearrangements. Hence, similar to csk^fl/flTCRβ^−/−MX-cre and csk^fl/flrag-1^−/−MX-cre mice, TCR-deficient cells develop in csk^fl/fllck-cre mice.

CD4 expression is reduced on thymocytes and peripheral CD4 T cells of csk^fl/fllck-cre mice, similar to csk^fl/flrag-1^−/− MX-cre mice (Fig. 2 A). Also, CD8 expression levels are reduced on the remaining peripheral CD8 T cells (Fig. 2 A). Surface expression levels of TCR on the majority of CD4-SP thymocytes and CD4 T cells is three- to fourfold reduced compared with csk^fl/fl control cells (Fig. 2 B). The Southern blot analysis of genomic DNA isolated from total thymocytes or FACS^®-sorted CD4^hi and CD4^lo peripheral T cells shows a clear correlation between low CD4 and TCR expression and deleted csk loci (Fig. 2 C).

Are TCR^loCD4^lo or TCR^loCD8^lo csk-deficient cells dependent on MHC molecules for their positive selection? To address this question, csk^fl/fllck-cre mice were bred onto β2m^−/−Aα^−/− (MHC⁻) background 5,8. The development of CD4-SP thymocytes and CD4 T cells in csk^fl/flMHC⁻lck-cre mice largely resembles that seen in csk^fl/flMHC⁺ controls (Fig. 3 A). The development of CD8 lineage cells is unaltered compared with csk^fl/flMHC⁻ control mice (Fig. 3 A). Despite the absence of MHC molecules, thymocytes in csk^fl/fl MHC⁻lck-cre mice show typical signs of positive selection. Upregulation of CD5 and CD69 on wild-type DP thymocytes are the first signs of positive selection 26,27. Expression of both markers is similarly increased in a fraction of DP cells in csk^fl/flMHC⁻lck-cre mice and csk^fl/flMHC⁺ control mice, compared with the csk^fl/flMHC⁻ control mice (Fig. 3 B). Also, the thymocyte maturation marker CD24/HSA is downregulated similarly in CD4-SP cells of control and csk^fl/flMHC⁻lck-cre mice (Fig. 3 B).

To analyze in a more direct way whether the TCR can influence the positive selection and lineage fate of csk-deficient DP thymocytes, we introduced TCR transgenes into csk^fl/fllck-cre mice. For an MHC class II–restricted TCR transgene, we used the DO11.10 TCR transgenic mouse strain, which expresses the DO11.10 TCR that recognizes a chicken ovalbumin–derived peptide in the context of I-A^d MHC molecules 28. As expected for MHC class II–restricted transgenic TCR, mostly CD4 T cells develop in both control and csk-deficient mice. No obvious differences could be detected in the numbers of control and csk-deficient thymocyte populations and peripheral T cells (data not shown). Notably, in both thymocytes and peripheral CD4 T cells the expression levels of CD4 as well as the transgenic TCR are reduced when csk deletion is achieved by the lck-cre transgene (Fig. 4A and Fig. B). For an MHC class I–restricted TCR transgene, we used the HY-TCR transgenic mouse strain. Csk^fl/fllck-creHY-TCRtg mice were bred on H-2D^b or H-2D^d background. The HY-TCR is specific for the male HY antigen presented by H-2D^b molecules 29. In female mice, HY-TCR transgenic cells are positively selected by an unknown peptide and only CD8 lineage T cells express both transgenic TCR α and TCR β chains (30; Fig. 5a and Fig. b). The analysis of positive selection of DP thymocytes in female mice expressing transgenic TCR and H-2D^b MHC molecules reveals a strong reduction in the percentage of CD8-SP thymocytes in the absence of Csk compared with control HY-TCR transgenic mice (Fig. 5 A, top panels). At the same time, the percentage of DP cells is reduced whereas CD4-SP and DN cell fractions are increased about twofold. Despite reduced numbers of CD8-SP thymocytes in csk^fl/fllck-creHY-TCRtg mice, a sizable number of peripheral CD8 T cells exists in these mice (Fig. 4 A). As evident from the expression of the transgenic TCR α chain on these CD8 T cells, positive selection of CD8 T cells is not completely abrogated (Fig. 5 B). In normal female HY-TCR transgenic mice, all CD4 cells express an endogenous TCR α chain which forms an MHC class II–restricted TCR in combination with the transgenic TCR β chain 30. Therefore, the transgenic TCR α chain is not expressed on CD4 T cells in these mice. The analysis of surface expression of transgenic TCR on csk-deficient CD4 T lineage cells reveals that in contrast to control mice, a large fraction of CD4-SP thymocytes and peripheral CD4 T cells express the transgenic TCR α chain (Fig. 5 B). Similar to the phenotype of csk^fl/fl lck-cre mice, surface expression levels of CD4 and CD8 are reduced on thymocytes and peripheral T cells of csk^fl/fllck-creHY-TCRtg mice (Fig. 5 A). Similar to the DO11.10 TCR transgene, expression from a multicopy transgenic TCR locus does not compensate for the reduction of expression of transgenic TCR α and β chains both on thymocytes and peripheral T cells of csk^fl/fllck-creHY-TCRtg mice (Fig. 5 B, and data not shown). Southern blot analysis of the csk locus on genomic DNA of total thymocytes shows efficient csk deletion in csk^fl/fllck-creHY-TCRtg mice.

Finally, the analysis of csk^fl/fllck-creHY-TCRtg mice on the nonselecting H-2D^d background reveals that even in the absence of the positively selecting H-2D^b MHC molecules, CD4^loTCR^lo cells expressing the transgenic TCR α chain can develop (Fig. 5 D). Unlike on H-2D^b background, there is no efficient development of T3.70⁺ CD8 T cells in csk^fl/fllck-creHY-TCRtg mice on H-2D^d background (data not shown).

The analysis of positive selection of α/β TCR–expressing csk-deficient cells suggests that selection signals originating from the expressed TCR have little influence on the fate of csk-deficient DP thymocytes. To test whether TCR signaling has any impact at all on selection steps during thymocyte development, we analyzed negative selection of the autoreactive HY-TCR–expressing thymocytes in male csk^fl/fllck-creHY-TCRtg mice on H-2D^b background. Similar to male control mice, thymocyte numbers are drastically reduced in csk^fl/fllck-creHY-TCRtg mice compared with female control mice (Fig. 5 A and 6 A). Our unpublished data show that the lck-cre transgene leads to deletion of the loxP-flanked csk gene in up to 25% of DN thymocytes. We also found that cre-mediated deletion of csk in 15% of DN is sufficient to generate a full-sized thymus in TCR β or rag-1–deficient mice (18; Fig. 1). If negative selection would be defective in csk-deficient thymocytes, the small fraction of csk-deficient cells detected by Southern blot analysis in total thymocytes of csk^fl/fllck-creHY-TCRtg mice (Fig. 5 B) should give rise to thymocyte populations similar to those observed in csk^fl/flrag-1^−/−MX-cre mice (Fig. 1). As no increase in thymocyte numbers or percentages of DP and SP cell populations occurs in male csk^fl/fllck-creHY-TCRtg mice, negative selection in the HY-TCR system functions normally in the absence of Csk.

The interpretation of changes in T cell development in csk-deficient mice rests on the assumption that the negative role of Csk in the regulation of Src family PTKs activity is the main reason for the observed phenotypes. We have shown previously that the specific activities of Lck and Fyn are increased in thymocytes isolated from csk^fl/flTCRβ^−/−MX-cre mice 18. We first tested whether the activation of Fyn caused by the deletion of csk could improve thymocyte development in lck-deficient mice, which have strongly reduced thymocyte numbers and impaired positive selection into CD4-SP and CD8-SP cells. When csk^fl/fllck-cre mice were crossed onto an lck^−/− background, a moderate twofold increase of thymus cellularity was apparent compared with lck^−/− mice. However, thymi of csk^fl/fllck^−/−lck-cre mice were still five to sixfold smaller than csk-deficient and control thymi (Fig. 7 A). Notably, a small number of CD4-SP thymocytes and CD4 T cells with reduced expression levels of CD4 and TCR could develop in csk^fl/fllck^−/−lck-cre mice (Fig. 7 A). These data support the notion that the activation of Fyn is responsible for the incomplete block in thymocyte development of lck^−/− mice 14. However, with respect to the TCR-independent development of csk-deficient T lineage cells, there is no formal proof that not other substrates of Csk could be responsible for that phenotype. To address this question we bred csk^fl/fl lck-cre mice onto a fyn^−/−lck^−/− background 13,14,31,32. We then analyzed whether the autonomous development of csk-deficient T cells could still occur in csk^fl/flfyn^−/− lck^−/−lck-cre mice after deletion of csk in DN thymocytes mediated by the lck-cre transgene. The absence of both Fyn and Lck normally causes the arrest of thymocyte development early at the DN stage, and the analysis of thymocytes from csk^fl/flfyn^−/−lck^−/−lck-cre mice revealed no increase in thymus cellularity and a block at the CD25⁺CD44⁻ stage of thymocyte development similar to control csk^fl/flfyn^−/− lck^−/− mice (Fig. 7 B). Southern blot analysis of total thymocytes shows that csk deletion occurred in a small fraction of the mostly DN cells of csk^fl/flfyn^−/−lck^−/−lck-cre mice (data not shown). Given that the lck-cre transgene leads to the development of TCR-deficient cells after deletion of csk in DN thymocytes, the lack of csk-deficient T lineage cells in csk^fl/flfyn^−/−lck^−/−lck-cre mice shows that Csk substrates or interacting proteins other than Lck and Fyn are not by themselves sufficient to cause the TCR-independent development of T lineage cells in the absence of Csk.

We used the conditional inactivation of csk in T lineage cells to address the question whether the activation of Src family PTKs Lck and Fyn is sufficient to support all steps in α/β T cell development that are normally controlled by pre-TCR or TCR engagement. It was of particular interest to understand if engagement of the α/β TCR can influence the fate of csk-deficient thymocytes, which in principle can also develop in the complete absence of TCR molecules.

The phenotype caused by conditional csk deletion in rag-1–deficient mice shows that TCR⁻ T lineage cells, as evident by CD90/Thy-1 and CD5 expression, can develop in the absence of Csk (Fig. 1). This phenotype is similar to the previously reported pre-TCR, α/β TCR–independent development of α/β T lineage cells in csk^fl/flTCRβ^−/−MX-cre mice 18. Therefore, the possibility of any critical contribution from the γ/δ TCR to the α/β TCR–independent development of α/β T lineage cells in the absence of Csk can be excluded.

Based on the lack of α/β T cell development of csk^fl/fl fyn^−/−lck^−/−lck-cre mice, it can also be excluded that Csk substrates or interacting proteins other than Lck and Fyn are sufficient for the autonomous T cell development in the absence of Csk (Fig. 6). Lck and Fyn are absolutely required for the development of thymocytes past the CD25⁺CD44⁻ DN stage, even in the absence of Csk. This does not exclude that Csk may regulate other proteins or may have additional functions earlier in T cell development before the proximal lck promoter becomes active 33,34.

Given that the TCR is dispensable for T cell development in the absence of Csk, we wanted to understand whether engagement of the TCR or coreceptors on csk-deficient thymocytes influences their developmental fate. Using csk^fl/fl mice in combination with the lck-cre transgene, we could analyze the development of csk-deficient thymocytes in the presence of the TCR. Similar to csk^fl/fl TCRβ^−/−MX-cre mice and csk^fl/flrag-1^−/−MX-cre mice, the number of CD8 T cells is reduced both in the thymus and in peripheral lymphoid organs of csk^fl/fllck-cre mice, whereas the number of CD4-SP thymocytes is consistently increased in the absence of Csk compared with controls (Fig. 1 A and 2 A). This bias for commitment to the CD4 T lineage could be the result of increased or extended activity of Lck and Fyn in the absence of Csk. Indeed, we have shown previously that the specific activity of Lck and Fyn are increased in csk-deficient thymocytes 18. This could be interpreted by the cell as a strong TCR-derived signal, which according to various recent reports favors development of CD4 lineage cells 35,36,37,38,39.

Development of CD4 T cells in the absence of Csk can occur without TCR/MHC interaction. Three independent results of our experiments support this conclusion. First, TCR-deficient T cells are present in csk-deficient rag-1^−/− or TCR β^−/− mice (reference 1; Fig. 1). Second, TCR-expressing T cells develop in the absence of MHC class I and MHC class II (Fig. 3). And third, transgenic TCR-expressing T cells develop in the absence of the positively selecting MHC molecule (Fig. 5 B). Despite the independence of TCR/MHC interaction, in all of these cases, the typical changes in the expression of thymocyte maturation markers (CD69, CD5, and CD24/HSA) that normally accompany positive selection are observed (Fig. 1 and Fig. 3, and data not shown). It is intriguing that even in the absence of TCR engagement, CD69 and CD5 maturation markers are expressed on a similar fraction of DP cells, as in controls. This could indicate that the deletion of csk initiates a predetermined developmental program that includes the changes in maturation marker expression at the appropriate time. Alternatively, after the deletion of csk in DP thymocytes, a gradual decline of Csk protein levels reaches thresholds at which Lck and Fyn become activated only in a small fraction of cells at a given time. This activation of Lck and Fyn may mimic a “true” TCR-derived signal and lead to the changes in maturation marker expression.

We investigated directly whether engagement of the TCR influences the development of csk-deficient thymocytes by the analysis of development of T cells expressing a transgenic TCR. Although development of TCR transgenic thymocytes in an MHC class II–restricted DO11.10 system appears largely normal in the absence of Csk, the MHC class I–restricted HY-TCR system showed very unusual features. Similar to the situation in nontransgenic csk-deficient mice, positive selection in csk^fl/fllck-creHY-TCRtg mice produces mostly CD4-SP thymocytes in contrast to the csk^fl/flHY-TCRtg controls. Also different from control mice, the csk-deficient CD4-SP as well as peripheral CD4 cells in csk^fl/fllck-creHY-TCRtg mice express both the transgenic TCR α and β chains (Fig. 4 B). Positive selection of these unusual CD4 lineage cells is unlikely mediated by their MHC class I–restricted transgenic TCR, but rather appears to be uncoupled from TCR engagement as seen for example in csk^fl/flMHC⁻lck-cre mice (Fig. 3). This interpretation is consistent with the recently published ability of activated Lck to redirect the positive selection of thymocytes expressing a different MHC class I–restricted TCR transgene from the CD8 into the CD4 lineage 38. Notably, in that study a TCR transgene was used that can function independently of the CD8 coreceptor. Thus, positive selection and commitment to the CD4 lineage by activated Lck may also be TCR and MHC independent, as in the absence of Csk.

Despite the apparent uncoupling of TCR engagement and positive selection into the CD4 T lineage, two results of the analysis of csk^fl/fllck-creHY-TCRtg mice point to the fact that signals derived from the TCR of csk-deficient thymocytes can influence their developmental fate. First, CD8 T cells bearing the transgenic TCR do develop in csk^fl/fllck-creHY-TCRtg mice. This occurs at reduced efficiency compared with csk^fl/flHY-TCRtg control mice; however, the frequency of TCR transgenic csk-deficient CD8 T cells is significantly higher than that of CD8 T cells in csk^fl/fllck-cre mice (compare Fig. 5a and Fig. b, and Fig. 2 A). Therefore, appropriate positive selection of csk-deficient thymocytes into the correct T cell lineage can occur, albeit at reduced efficiency. Second, the analysis of male csk^fl/fllck-creHY-TCRtg mice on H-2D^b MHC background males clearly shows that TCR-mediated negative selection in the HY-TCR system in csk-deficient thymocytes is not impaired (Fig. 5). The reduced number of HY-TCR expressing CD8-SP thymocytes in female csk^fl/fllck-creHY-TCRtg mice on H-2D^b MHC background may result from a deviation of positive selection into the CD4 T lineage due to stronger TCR-derived signal in the absence of Csk. Further analysis of thymocyte development in the absence of Csk in different TCR transgenic systems is required to confirm the differential role of Csk-mediated regulation of Lck and Fyn in positive and negative selection.

The surface expression of TCR as well as CD4 and CD8 coreceptors is significantly reduced on csk-deficient compared with control thymocytes and peripheral T cells. This phenotype is observed in all mice analyzed in this study, including those expressing a transgene-encoded TCR. The correlation between low receptor expression levels and csk deficiency is established by Southern blot analysis of CD4^lo and CD4^hi cells (Fig. 2 C). Reduction of TCR expression levels in the absence of Csk could be the consequence of Lck activation in these cells. On mature T cells, the ligation of TCR by MHC/peptide ligands and presumable activation of Lck leads to the endocytosis and lysosomal degradation of TCRs 40. Also, expression of the constitutively active Tyr505Phe mutant of Lck in mature T cells causes downregulation of surface TCR and degradation in a lysosomal compartment 41. Mice expressing a transgenic TCR and constitutively active Lck transgene show reduced TCR surface expression on thymocytes compared with only TCR transgenic control mice 42. Little is known about the regulation of CD4 and CD8 surface expression. However, both positive selection of DP thymocytes, which presumably activates Src family PTKs, and stimulation of mature T cells is accompanied by the reduction of coreceptor surface expression 43,44,45,46. Downregulation of TCR and coreceptors may also in some way be coordinated; however, the low expression level of CD4 on TCR-deficient T lineage cells in csk^fl/flTCRβ^−/−MX-cre and csk^fl/flrag-1^−/−MX-cre mice argues against a tight coordination.

The reduced TCR and coreceptor expression levels seen in the absence of csk do not impair the function of mature T cells as defined by analysis of Ca²⁺ flux and proliferation in response to TCR cross-linking (data not shown, unpublished observations). It remains to be shown whether the reduced TCR and coreceptor expression levels represent a cell autonomous adaptation to the defective regulation of Lck- and Fyn-mediated signaling from the TCR and coreceptors.

In conclusion, our data show that the Csk-mediated regulation of Lck and Fyn during T cell development is essential for the establishment of developmental checkpoints that require signals from the pre-TCR and TCR to be faithfully transmitted to the developing thymocyte. It remains to be addressed whether the csk-deficient T cells populating peripheral lymphoid organs are dependent on TCR-mediated survival signals.

We thank Horst Bluethmann, Hans-Jörg Fehling, Ralf Kühn, Denis Loh, Tak Mak, Jamey Marth, Roger Perlmutter, Klaus Rajewsky, and Harald von Boehmer for providing mice and reagents. We thank members of the Rajewsky and Tarakhovsky labs for discussions and the staff at the animal facility of the University of Cologne for animal husbandry. We are indebted to Banu Meiners for invaluable help with mouse genotyping and to Marc-Werner Dobenecker for help with some experiments.

This work was supported by the Deutsche Forschungsgemeinschaft through SFB 243 Fonds der Chemie and the Irene Diamond Foundation.