Distinct roles for Syk and ZAP-70 during early thymocyte development

We used competitive repopulation of the thymus in irradiated hosts as a sensitive and discriminating assay for in vivo thymocyte development. Such a competitive repopulation strategy might, for example, reveal subtle differences in proliferative capacities or usage of limiting growth factors. We used stem cell progenitors from zap70^−/− and syk^−/− embryos or animals and distinguished the two donor populations based on the congenic markers CD45.1 and CD45.2 (20). This strategy is outlined in Fig. S1 (available). Lethal irradiation of WT hosts and reconstitution with donor stem cells resulted in ∼98–99% replacement of all thymocytes within 4 wk by all genotypes tested (unpublished data). The 1–2% residual host cells detected were always within the mature SP subsets. All other subsets were routinely replaced >99% by donor-derived cells.

Because thymocyte development occurs as a well-characterized series of stages, the percentage represented by each genotype at a given stage is indicative of the developmental fitness of the thymocyte up until that stage. Any impairment or enhancement in fitness would result in a decrease or increase, respectively, at the next stage of development. The competitive repopulation of the thymus using syk^−/− versus zap70^−/− progenitors is shown in Fig. 1. Injection of equal amounts of BM or fetal liver (FL) resulted in roughly equal representation at the earliest DN2 and DN3 stages identified. Equal representation of syk^−/− and zap70^−/− thymocytes within these earliest subsets suggests that neither kinase has an important unique function in development until at least the DN3 stage. In striking contrast, the two populations are differentially represented at multiple stages immediately after DN3. At least two changes seemed to occur: Syk^−/− thymocytes became underrepresented at the DN4 stage when compared with DN3, and after that, zap70^−/− thymocytes became underrepresented while progressing from the DN4 to the DP stage. These could be explained in multiple ways. One possibility is that Syk provides a positive regulatory role at the DN3 stage of development (thus, syk^−/− accumulates at DN3), whereas ZAP-70 provides a different positive regulatory role at the DN4 and ISP stages. Another possibility is that ZAP-70 provides a negative regulatory role at the DN3 stage, allowing zap70^−/− thymocytes a greater fitness advantage in transitioning from DN3 to DN4. Similarly, syk^−/− thymocytes may have an advantage in progression from DN4 to DP. Our results differ from previous work that had reported that ZAP-70's only unique function in thymocyte development was during thymic selection, at the DP stage (21). We also saw this effect in our assays, as ZAP-70 deficiency resulted in a complete block in positive selection as shown by the decrease in representation of zap70^−/− thymocytes between the DP and SP4 stages. These data reveal that Syk and ZAP-70 uniquely contribute to the developmental fitness of early thymocytes at distinct stages well before positive selection occurs.

To understand the basis of differential fitness between syk^−/− and zap70^−/− developing thymocytes, we sought to determine the expression patterns of both kinases during all stages of thymocyte development. To investigate the expression patterns of Syk and ZAP-70 kinases, we quantitated the relative levels of mRNA in all thymocyte subsets. We sorted cells and used real-time fluorigenic RT-PCR to quantitate mRNA expression. The DN subsets were isolated based on standard phenotyping: CD25⁺CD44⁺ (DN2), CD25⁺CD44⁻ (DN3), and CD25⁻CD44⁻ (DN4). We found that Syk mRNA was robustly expressed at DN2, whereas Zap70 was hardly detectable, almost 100-fold less than in peripheral T cells (Fig. 2 A). Zap70 is first up-regulated within the DN3 population and increases thereafter. Interestingly, Syk mRNA expression decreases from DN3 to DN4 and spikes at the DP stage, before decreasing again at the SP stage. This pattern of Syk family kinase mRNA expression is consistent with another recent study (22).

After characterizing mRNA expression, we investigated ZAP-70 and Syk protein expression. We previously characterized Syk (23), but not ZAP-70, protein expression during thymocyte development. To do so, we conjugated an mAb previously made in this laboratory against ZAP-70 (24) to Alexa Fluor dyes. This gave a 25-fold increased signal in T versus B cells. This was specific, as NK cells and the 1–2% remaining T cells from zap70^−/− mice provided a negative control (unpublished data).

Expression of ZAP-70 and Syk in all stages of thymocyte development is shown in Fig. 2 B. The overall pattern of expression of Syk agrees with our previous findings (23). We found ZAP-70 expressed at low levels during the earliest stages of development. In fact, robust ZAP-70 expression is not seen until DN4, which exhibits a four- to fivefold increase in ZAP-70 levels compared with DN3. ZAP-70 protein continues to increase thereafter (Fig. 2, B and C). CD8⁺TCRβ^hi SP thymocytes had slightly more ZAP-70 than CD4⁺ SP thymocytes, owing to their slightly larger size (unpublished data).

We sorted CD25⁺ DN cells (pooled DN2/3) and performed immunoblot analysis on lysed cells (Fig. 2 D) to confirm low ZAP-70 protein expression in DN3. We also sorted CD25⁻ DN cells and compared these cells with various lymphoid cell lines. This analysis confirmed that Syk is easily detectable within DN2/3 cells, whereas ZAP-70 is expressed at low levels. Overall, Syk and ZAP-70 are reciprocally expressed across thymocyte development, where Syk is highly expressed in early development and ZAP-70 is highly expressed in later development. However, there is considerable overlap where both kinases are detectable, including the DN3, DN4, and ISP stages.

Intracellular flow cytometry, combined with our initial competitive repopulation assays, suggested that Syk may be serving a positive regulatory role at DN3 and serves a reduced role from the DN4 through the DP stages as its protein expression disappears. To test this possibility, we competed Syk-deficient progenitors against the WT in a repopulating thymus. This ruled out confounding effects brought on by competing against zap70^−/− thymocytes. Because Syk is highly expressed in DN1 and possibly earlier, we assayed representation of DN1 after varying the ratios of FL injected into the host as a measure of the fitness of colonizing the thymus. Analysis of DN1 representation revealed no substantial changes in donor ratios (Fig. 3 A) compared with what was injected. Therefore, colonization is not substantially affected in syk^−/− thymocyte progenitors after lethal irradiation of the host, consistent with our results in Fig. 1.

We then assessed donor representation within subsequent stages of thymocyte development. Fig. 3 B shows that syk^−/− thymocytes were consistently underrepresented only after DN3. This was true at all starting ratios. Interestingly, syk^−/− thymocytes were overrepresented within the CD4⁺ SP and CD8⁺ SP TCR^hi populations (not depicted). This was true whether the irradiated host was CD45.1⁺ or CD45.2⁺, suggesting that syk^−/− thymocytes may be more fit than the WT at progressing from the DP to the SP stage.

We focused our experiments on whether Syk deficiency impairs TCRβ rearrangement and protein expression. DN3 cells from WT and syk^−/− thymocytes had equal percentages of intracellular TCRβ (TCRβic⁺) cells (Fig. 3 C). However, cell-cycle progression was impaired 40–50% in syk^−/− DN3 (Fig. 3 D), as shown by decreased BrdU incorporation. In contrast, DN4 cell cycling was comparable in syk^−/− and WT cells, indicating less of a role for Syk at this stage. Because apoptosis is difficult to detect within a steady-state thymus, we could not detect reproducible differences in apoptosis between genotypes within the mixed chimeras. Therefore, our competitive repopulation experiments indicated that Syk-deficient thymocytes have a fitness defect compared with WT thymocytes at the DN3 stage, and loss of Syk is only partially compensated for by ZAP-70. Furthermore, although TCRβ rearrangement seems to occur normally, Syk deficiency impairs cell-cycle progression at the DN3 stage.

DN3 cells are a mixture of at least two populations: small cells actively rearranging TCRβ loci (DN3-E or DN3a) and large cells that have successfully expressed a TCRβ chain, initiated pre-TCR signaling, and entered a cell cycle (DN3-L or DN3b) (25, 26). Our repopulation assays suggested that Syk is required for normal progression through one or both of these stages. To investigate which stage is affected, we analyzed TCRβic⁺ expression and compared it to cell size to discriminate DN3-E and DN3-L cells within WT and syk^−/− thymocytes. For a given mouse, TCRβic⁺ correlated well with increased cell size in WT thymocytes. However, in syk^−/− thymocytes within the same mouse, TCRβic⁺ did not correlate with increased cell size, where the DN3-L population was greatly reduced despite normal levels of TCRβic⁺ expression (Fig. 3 C; and Fig. 4, A and B). We performed a similar analysis of DN3 from WT and zap70^−/− mixed chimeric mice (further discussed in the following section). Zap70^−/− thymocytes had normal numbers of DN3-L cell formation after TCRβic⁺ expression. Similar results were seen in mice reconstituted with varying ratios, or that were noncompetitively reconstituted, and in nonreconstituted WT and zap70^−/− mice. Collectively, these results indicate that Syk has a nonredundant function in early pre-TCR signaling and is needed for normal blast formation and cell-cycle progression after TCRβ selection and protein expression. Additionally, our mixed chimeric studies indicate that this defect is thymocyte intrinsic.

We next wished to directly characterize the biochemical nature of the initial pre-TCR signal in WT and syk^−/− thymocytes. However, traditional biochemical methods were prohibitive because of the paucity of DN subsets and the limitations of making sufficient chimeric mice. As an alternative to traditional biochemistry, we used flow cytometry to detect phosphorylation site–specific antibody staining in TCRβic⁻ DN3-E cells before and after CD3 cross-linking as a correlate of kinase activation. The phospho–ZAP-70 (Tyr319)/Syk (Tyr352) antibody used detects activated forms of both ZAP-70 and Syk. Small amounts of CD3 reach the surface of Rag1-deficient DN3 cells (27), and we speculated that this would be true of normal TCRβic⁻ DN3-E cells. Anti-CD3 stimulation of WT DN3-E thymocytes resulted in a twofold increase in phosphorylated Syk/ZAP proteins (Fig. 4 C, top left). Preincubating with PP2, an Src-kinase specific inhibitor, decreased the phospho-ZAP/Syk signal below background (not depicted), confirming that the signal emanates from a CD3-Src kinase–dependent pathway. Similar stimulations of syk^−/− DN3-E thymocytes revealed that any phosphorylation of ZAP-70 is undetectable (Fig. 4 C, top right). As a control, CD4⁺ SP thymocytes from WT and syk^−/− thymocytes (which express equal levels of ZAP-70 and no Syk) were equally phosphorylated on Syk family kinases after CD3 cross-linking. These data suggest that Syk, not ZAP-70, is preferentially activated downstream of the CD3 complex within WT DN3-E cells and that this may account for the blast defect seen in syk^−/− TCRβic⁺ DN3 thymocytes.

We next wished to further test whether ZAP-70 serves a positive regulatory role before positive selection. Our initial competitive repopulation experiments suggested that ZAP-70 might provide two novel functions: a negative regulatory function at DN3 as well as a positive function immediately after DN3 and before positive selection. Unlike syk^−/− thymocytes, zap70^−/− thymocytes had no discernable fitness impairment or enhancement at the DN3 stage, as measured by competitive repopulation versus the WT (Fig. 5 A). Blast size, TCRβ rearrangement, and protein expression were similar to the WT, as was cell-cycle status in DN3 and DN4 cells (Fig. 5, B and C). These data suggest that ZAP-70 has no notable role (positive or negative) at the DN3 stage that is not compensated for by Syk in the zap70^−/− thymocyte during competitive repopulation. Competitive repopulation revealed that SP thymocytes are lacking in zap70^−/− thymocyte subsets, consistent with the reported phenotype (21).

Surprisingly, zap70^−/− thymocytes were also substantially impaired before the DP stage. Zap70^−/− thymocytes were largely impaired even before the CD8⁺CD4^−/loTCRβ^lo ISP stage (Fig. 5 A). Thus, competition of zap70^−/− versus the WT and syk^−/− versus the WT is entirely complementary with our results from directly competing syk^−/− against zap70^−/− (Fig. 1, Fig. 3, and Fig. 5). Collectively, these data suggest a novel positive regulatory function for ZAP-70 during the DN4, ISP, and DP stages and before positive selection, as well as a distinct positive role for Syk at DN3-E. The individual competitions against the WT argue against any enhanced fitness upon deleting the kinases and any obvious negative regulatory functions.

It is possible that ZAP-70 is not involved in pre-TCR signaling but has some other function that is important and merely correlates with the pre-TCR⁺ cellular stage. For example, the pre-TCR may generate a single Syk-dependent signal at the DN3-E stage that leads to the expected 6–10 cell divisions and characteristic gene expression changes, whereas ZAP-70 functions in a parallel but different pathway. Alternatively, initial Syk-dependent pre-TCR signaling may give way to continuous ZAP-70–dependent pre-TCR and/or TCR signaling that is MHC independent between the DN3 and DP stages.

To test this, we used an in vivo model of pre-TCR signaling. Rag1^−/− (Rag) thymocytes can be induced to differentiate into DP thymocytes by in vivo stimulation with anti-CD3 mAbs (27, 28). This process yields normal numbers of DP thymocytes and is dependent on kinases and the nonredundant T cell adapters Slp-76 and LAT (14, 15). We injected increasing amounts of anti-CD3 into rag1^−/− mice to determine whether DP development was titratable. Increasing amounts of anti-CD3 resulted in a dose-dependent increase in total thymocyte production after 5 d and was primarily representative of DP production (Fig. 6 A, bar graphs). Absolute DN4 production also increased with increasing antibody, whereas DN3 cells decreased (not depicted), probably owing to the slow rate of replacement from BM stem cells and increased differentiation to DN4. This assay allowed us to ask whether ZAP-70 is necessary for DP generation when we control the strength and specificity of a pre-TCR–like signal. Anti-CD3 stimulation resulted in 7–10-fold less DP generation at all antibody concentrations in rag1^−/−zap70^−/− (RagZAP dKO) compared with Rag controls. The difference slightly decreased with increasing antibody concentrations, suggesting that increased signaling can lead to a normalization of cell numbers. Nevertheless, within the time frame studied, RagZAP dKO DN3 required 10-fold more antibody injected to achieve the same overall cellularity (compare RagZAP dKO at 100 μg with Rag at 10 μg). We believe this result represents a ZAP-70–dependent cell intrinsic defect downstream of the CD3 pathway before the positive selection stage.

Representative phenotypic profiles of Rag and RagZAP dKO thymocytes are shown in Fig. 6 A (right). No overall increase in cell numbers was seen in RagZAP dKO stimulated with 10 μg anti-CD3, yet 68% of cells differentiated to the DP stage, reflecting decreased absolute DN3s. At 30 μg, the frequency of DP is similar with or without ZAP-70, yet the overall and DP cellularity is greater than sevenfold less without ZAP-70. The absolute counts of DN3 and DN4 are also decreased two to fivefold in RagZAP dKO compared with Rag mice (not depicted).

To determine how ZAP-70 affects generation of DP cells in this assay, we measured levels of BrdU incorporation and Annexin V binding as markers of cell-cycle status and apoptosis, respectively. For both genotypes, the fraction of cells in cell cycle in all subsets increased with increasing antibody. Examples of BrdU incorporation at 100 μg are shown in Fig. 6 B. RagZAP dKO DN3 and DN4 cells incorporated ∼80% BrdU compared with the respective Rag KO cells, and RagZAP dKO DP cells incorporated only ∼35% (Fig. 6 B). Interestingly, the relative decrease in cell cycle in RagZAP dKO compared with Rag subsets was constant at various antibody concentrations. These data suggest that the fraction of cells in cell cycle in all subsets is proportional to the CD3-mediated (i.e., pre-TCR) signal and that this signal depends on ZAP-70.

Because the initiating CD3-based signal was largely synchronized in these experiments, assessment of apoptosis was possible. For both genotypes, apoptosis was highest after minimal stimulation with 10 μg anti-CD3 and is shown in Fig. 6 C. DN3, DN4, and DP subsets in RagZAP dKO were 5–20-fold more apoptotic than Rag after 10 μg anti-CD3 (Fig. 6 C). This difference decreased to approximately twofold in all subsets at greater antibody concentrations. Thus, unlike cell cycle, the effect of ZAP-70 on survival was proportional to the pre-TCR signal strength but appeared to be threshold dependent. The cell-cycle and apoptosis results suggest that ZAP-70 affects sustained pre- TCR signaling at all stages between DN3 and DP. We attempted similar experiments using rag1^−/− and rag1^−/−syk^−/− reconstituted animals (into lethally irradiated rag1^−/− hosts) but could not achieve reproducible levels of proliferation even within rag1^−/− reconstituted mice. This may be because of radiation damage to the vascular system of the host mouse (29) and inefficient absorption of the anti-CD3 antibody.

Because ZAP-70 is quickly up-regulated within DN4 cells, we speculated that ZAP-70 might be a direct target or indicator of pre-TCR signaling. As noted before, DN3 thymocytes are a mixture of two populations: small, quiescent TCRβ-unrearranged DN3-E (DN3a) cells and large, cycling TCRβ-rearranged DN3-L (DN3b) cells that are beginning the progression toward DP. Our RNA and initial flow cytometry studies conflicted on whether ZAP-70 was first expressed within the DN3 stage. We noticed that DN3 cells had a reproducible positive shoulder of ZAP-70 protein but not Syk (Fig. 7 A), suggesting that subpopulations have differential ZAP-70 expression.

We focused on CD25⁺ DN2/3 thymocytes and assayed for simultaneous intracellular expression of TCRβ and either ZAP-70 or Syk protein to investigate the earliest detectable TCRβ-selected cell. TCRβic⁺ cleanly separated a ZAP-70^lo/− cell from a ZAP-70⁺ population. Thus, TCRβ selection correlates with ZAP-70 protein up-regulation. The increase in expression was about threefold. The antibody specificity was confirmed by staining equivalent zap70^−/− CD25⁺ DN cells (unpublished data). ZAP-70 correlated with cell size, which also correlated to TCRβic⁺ expression (Fig. 7 A; and unpublished data). We also found that ZAP-70⁺ cells corresponded to CD27⁺ DN3b cells (unpublished data) (26).

These results suggested that ZAP-70 is up-regulated after TCRβ is rearranged and expressed. Therefore, the up-regulation of ZAP-70 may be a consequence of initial pre-TCR signaling itself. To test this, we assayed ZAP-70 expression in Rag KO mice, as they are completely blocked in pre-TCR signaling and arrested at DN3. We injected increasing amounts of anti-CD3 mAb into Rag KO mice and examined ZAP-70 expression in DN3, DN4, and DP cells. ZAP-70 protein increased in a dose-dependent manner after anti-CD3 treatment, as shown for DN4 in Fig. 7 B. Treated RagZAP dKO mice confirmed the specificity of the signal (Fig. 7, B and C). ZAP-70 protein is almost completely absent in resting Rag KO DN3, consistent with a requirement of an initial pre-TCR signal for robust ZAP-70 up-regulation. ZAP-70 expression increased in DN3, DN4, and DP cells (Fig. 7 C; and unpublished data) after anti-CD3 injection in a dose-dependent manner. In addition, we assayed mice in which TCRβ rearrangement still occurs yet no signal is generated (lat^−/− and lck^−/−fyn^−/−) and found no up-regulation of ZAP-70 or CD27 and, therefore, no correlation with intracellular TCRβ (unpublished data). Collectively, our data suggest that ZAP-70 is expressed at low levels in DN3-E thymocytes and is up-regulated only after initial pre-TCR signaling in DN3-L. Thus, ZAP-70, along with CD27, is among the earliest targets or indicators of pre-TCR signaling. Developing thymocytes have the unique feature by which initial Syk-dependent pre-TCR signaling at DN3-E quickly up-regulates ZAP-70, which then replaces Syk during sustained pre-TCR/TCR signaling during the DN3-L, DN4, ISP, and DP stages.

We have examined the expression patterns and in vivo functions of the Syk family kinases Syk and ZAP-70 during early stages of thymocyte development. Syk^−/− thymocytes are impaired in transitioning from DN3 to DN4 during competitive repopulation of the thymus. Analyses of mRNA abundance and intracellular flow cytometry indicated that early T cell progenitors express high levels of Syk and little or no ZAP-70, and this remains true as cells commit to the T lineage pathway and become DN2 cells. After TCRβ selection has occurred, Syk is necessary in quiescent DN3-E (DN3a) cells for optimal initial pre-TCR signaling and entry into cell cycle and transition to DN3-L (DN3b) cells. ZAP-70 appears to have a preferred function during sustained pre-TCR signaling from the DN4 through the DP stages and aids in the normal expansion and differentiation of thymocytes despite concurrent Syk expression (through the ISP stage), after which positive selection occurs. In vivo stimulation with anti-CD3 mAbs in Rag mice confirms that ZAP-70 uniquely functions downstream of the pre-TCR/TCR after the DN3-E stage and before DP selection. Loss of ZAP-70 mostly increases apoptosis levels but also decreases cell-cycle progression, especially at the DP stage. ZAP-70 expression is induced by the pre-TCR and is quickly up-regulated, coincident with CD27 when newly formed pre–T cells transition from DN3-E to DN3-L. Our data suggest the pre-TCR is a dynamic signaling module in which the signaling machinery is fundamentally altered during development. We propose a model where Syk-dependent pre-TCR signaling is replaced with ZAP-70–dependent pre-TCR signaling.

Previous work on the role of Syk in T cell development has been conflicting. Although the original KO papers suggested that Syk had little role in development (1, 2), later reports argued for either a unique role in TCRγδ development (30) or a more quantitative role in general T cell development (31). Our work is consistent with and expands on the results of Colucci et al. (31). That study found that DN3 cells were found at higher frequencies in syk^−/− reconstituted mice, but they never competed syk^−/− thymocytes versus the WT within a given mouse. Our studies allow direct comparison of the in vivo fitness of syk^−/− versus WT and syk^−/− versus zap70^−/− thymocytes within the same mouse, allowing us to assess a quantitative impairment of the block and to conclude that the impairment is cell intrinsic. Further still, we demonstrate that ZAP-70 protein is poorly expressed within the DN3-E thymocytes, providing a logical explanation for syk^−/− impairment.

Clearly, the pre-TCR has several well-defined functions (10). However, it is not established when these functions are used. Rescue from apoptosis may be necessary at the DN3-E stage but may also be necessary at every stage from DN3-L to DP for normal population expansion. One can imagine continuous pre-TCR signaling enabling maximal cell-cycle progression at all stages. In contrast to this, decreasing Rag activity and enforcing allelic exclusion of the TCRβ locus might only be necessary at the DN3-E stage. Some of these distinct functions have been shown to diverge downstream of the pre-TCR (19). We propose that they also temporally diverge during differentiation and expansion. ZAP-70 may provide continuous pre-TCR/TCR signaling during the DN3-L, DN4, ISP, and/or DP stages that is needed for optimal development. Indeed, our data from in vivo–stimulated Rag mice demonstrate a strong correlation with anti-CD3 titration and ZAP-70 protein expression in all subsequent stages and absolute DP generation. A similar function has recently been proposed for Slp-76 (18). Continuous or constitutive pre-TCR signaling has been previously suggested. Studies of c-cbl^−/− thymocytes suggest a continuous pre-TCR signal, as these cells have pre-TCRs that fail to be continuously internalized (32). DN4 cells are shown to activate NF-κB and NFAT downstream of the pre-TCR (33). A recent study demonstrated temporal separation of Egr3 and RORγ induction after pre-TCR signaling, both of which are necessary for proper DP generation (34).

Other signaling components are also known to promote the proliferation and survival of DN and DP stages. Early stages of TCRβ selection require activation of Notch1 (11), p53 (35), FADD (36), and NF-κB (33, 37) pathways, among others, many of which are likely to be affected downstream of the pre-TCR. Because these molecules exert their effects at or immediately after the DN3 stage, they may be optimally activated by an Syk-dependent pre-TCR signal. However, Bcl-X, RORγ, and TCF/LEF all affect the survival of DP cells (38, 39), and they may be preferentially influenced downstream of a ZAP-70–dependent pre-TCR signal.

Notch and E2A activities precede/activate a host of T lineage genes such as CD3 and Lck as early as stage DN1 (40–42). Therefore, it is surprising that ZAP-70, a largely T lineage–specific protein, is not highly expressed until after the first T lineage commitment checkpoint. One teleological explanation may be that Syk is better able to transduce the initial pre-TCR signal. This could either be a quantitative or a qualitative difference. Quantitatively, Syk is a more active kinase in vitro, causes more basal activation in T cells, and is less Src kinase–dependent in T cell lines (for review see reference 43), which makes it less coreceptor dependent. In fact, the use of Syk for initial pre-TCR signaling may explain why the Lck KO is more leaky at pre-TCR signaling than at positive selection. Qualitatively, Syk may activate/inactivate specific genes. After initial pre-TCR signaling, ZAP-70 may be better able to transduce continuous pre-TCR signals that are necessary for DN to DP transitioning. A previous study addressed whether Syk can substitute for ZAP-70 during positive selection by transgenic expression of Syk in a ZAP-70–deficient mouse strain (44). These thymocytes overexpressed Syk after β selection and through the DP stage, suggesting that if Syk were overexpressed it could replace ZAP-70 during continuous pre-TCR signaling and DP expansion. However, the efficiency of DP generation was not determined, as these thymocytes were not tested by competitive repopulation, leaving open the question of whether Syk could comparably replace all ZAP-70 functions if its expression was maintained from the DN3 to the DP stage. A recent study showed that Syk-expressing B cells require the traditional mitogen-activated protein kinase kinase pathway for p38 activation, whereas ZAP-70–expressing T cells can also activate p38 in a novel mitogen-activated protein kinase kinase 3/6–independent manner (45). Thymocytes require that p38 not be activated during initial pre-TCR signaling (46), and Syk may suffice. However, p38 becomes necessary for normal transitioning from the DN4 to the DP stage (47) and, thus, may be largely induced by ZAP-70–based pre-TCR signaling. We are currently testing these possibilities.

It is certainly possible that the differential fitness of Syk and ZAP-70–deficient thymocytes may, at least in part, be reflective of the distinct expression levels of each kinase during these stages and may not be caused by a unique function of the given kinase at a certain stage. To formally test this would require a model system involving the competition of thymocytes that express comparable levels of either ZAP-70 or Syk throughout thymic development on a null background for both kinases. Additionally, as TCRγδ cells diverge from TCRαβ cells at this DN3 stage because of a purported difference in signal strength, the balance of γδ versus αβ T cells may be altered (48). A relative overexpression of ZAP-70, Syk, or both could alter this ratio by increasing the strength of the signal generated, thereby increase relative numbers of TCRγδ. Our comparative studies detail for the first time the extent to which thymocytes differentially express both Syk family kinases and show that such differential expression/and or function results in differential cellular fitness during early development, where a Syk family–driven pre-TCR signal is necessary for normal DP expansion.

Interestingly, one unusual feature that distinguishes TCRα selection from the analogous Ig light chain selection is that TCRα chains are not allelically excluded (49, 50). Although the mechanisms leading to allelic exclusion are not well understood, Syk may be uniquely suited to better transduce these signals in both lineages. Although ZAP-70 can perform some Ig heavy chain allelic exclusion (51), Syk is mostly responsible for this. It is unknown whether Syk or ZAP-70 differentially effect allelic exclusion of the TCRβ chain.

In summary, our studies identify a specific role for Syk during β selection and cell-cycle entry and uncover an unexpected relative requirement for ZAP-70 during early thymocyte development. This occurs despite overlapping expression of both kinases within DN3-L and DN4 and ISP cells. Our data are also consistent with a model in which Syk-dependent pre-TCR signaling induces ZAP-70 expression, which then replaces Syk for ZAP-70–dependent pre-TCR signaling as thymocytes progress toward the DP stage.

Zap70^−/− mice were backcrossed at least seven times onto the B6 background. Syk^+/− mice were carried as heterozygotes because of their homozygotic perinatality and were backcrossed at least 11 generations to B6. Syk^+/− was later crossed to the BoyJ (B6 mice with the CD45.1⁺ congenic marker B6.SJL-Ptprc^a Pepc^b/BoyJ) background for one generation. B6 and rag1^−/− mice used for breeding were originally obtained from the Jackson Laboratory. B6 mice used as hosts for competitive repopulation were obtained from Charles River Laboratories. BoyJ mice were purchased from Taconic and the Jackson Laboratory. For BrdU studies, 1–2 mg BrdU (Sigma-Aldrich) at 20 mg/ml was injected i.p. 1–2 h before killing the animal. All animals were housed in specific pathogen-free facility at the University of California, San Francisco according to university and National Institutes of Health guidelines.

RNA was extracted using TRIzol (Invitrogen), according to the manufacturer's instructions. cDNA was created using the Sensiscript protocol (QIAGEN). Primers and probes used for quantitative PCR span introns and were as follows. For mouse ZAP70, 5′-GCATGCGCAAGAAGCAGATT-3′ (forward primer), 5′-GGGCCTCTCGCATCATCTC-3′ (reverse primer), and [6-FAM]CTTTGTCGGCCTTCTCTGTGCCCTG[BHQ1∼Q] (probe) were used. For mouse Syk, 5′-CTGGTTCCATGGCAACATCTC-3′ (forward primer), 5′-TGGCCCTGATCAGGAATTTTC-3′ (reverse primer), and [6-FAM]TGACCCTATGAGGACCGTCTGCTCTGAT[BHQ1∼Q] (probe) were used. HPRT primers and PCR conditions were previously described (52).

Cell lysates were prepared using a 1% NP-40 lysis buffer with inhibitors and transferred to membranes after SDS-PAGE, as previously described (53). The A20 B cell line, Jurkat and P116 (ZAP-70–deficient) cells, and the pre–T cell lines Sciet27 and SCB29 were used as ZAP-70– and Syk-expressing lymphoid cell lines. Lysates were corrected for protein content before loading. Antiactin mAb was obtained from Sigma-Aldrich. Lanes were quantitated using an Image Station (model 440DF; Kodak) and ID image analysis software (version 3.5; Kodak).

For competitive repopulation assays, FL versus FL or BM versus BM cells were injected i.v. into lethally irradiated B6 or BoyJ hosts, and reconstituted thymi were analyzed 4–8 wk later. FL was from timed pregnancies at E15.5–16.5, and BM was from 8–12-wk-old adults. Syk^−/− embryos were visually identified and confirmed by either PCR or flow cytometry of reconstituted blood. For some experiments, syk^−/− FL was used to generate chimeric mice, and the resulting BM was used for competing or reconstituting irradiated hosts. No differences were seen using these BM compared with FL. WT recipients were irradiated with two doses of 600 rads, 3–5 h apart. Mixed FL chimeras were injected i.v. at ratios of 1:0, 0:1, 1:3, 3:1, 1:9, or 9:1. Each FL was resuspended in 1 ml media, and 200 μl (mixed or not) was used per mouse. For BM, 5 × 10⁶ total cells were injected i.v. into each mouse. Males were used as recipients for most experiments, and females were used for some with no qualitative differences. Within an experiment, all hosts were sex and age matched and were between 5–10 wk old.

Thymi were separated to single-cell suspensions using glass slides. 1–3 × 10⁶ cells were used for each stain. mAbs against CD3, CD4, CD8, CD11b, CD19, and Gr-1 were used to exclude lineage-committed cells and delineate DN thymocytes and were conjugated to either PE or allophycocyanin (APC; BD Biosciences and eBioscience, respectively). Biotin-DX5, -TCRγδ and -CD69 (BD Biosciences) with streptavidin-APC or -PE were used to amplify the signal and/or exclude these cells. For Syk and ZAP-70 quantification in thymocyte subsets, DN1 was defined as Lin⁻ CD44⁺CD117⁺CD25⁻. T and B cells were from the LN and spleen and defined by CD3⁺CD19⁻ for T cells and CD3⁻CD19⁺ for B cells. CD25, CD44, CD117, CD45R, and TCRβ (BD Biosciences) were conjugated to PE, PE-Cy5.5, or PerCP-Cy5.5 and FITC and further characterized DN subsets. mAbs against CD45.1 and CD45.2 were conjugated to FITC, PE, or APC. mAbs against mouse/human ZAP-70 1E7 (24) and mouse Syk 5F5 (23) were conjugated to Alexa Fluor 488 or Alexa Fluor 647 at fluorophore/protein ratios of 6.5–7:1 or 5:1, respectively, using the manufacturer's instructions (Invitrogen). Anti-BrdU was purchased from Invitrogen. Annexin V–FITC was obtained from Caltag. Fc binding was blocked using anti-CD16/32 2.4G2 (Harlan). For intracellular staining, cells were surfaced stained; fixed in Fix and Perm Medium A (Caltag); stained for intracellular ZAP-70, Syk, TCRβ, BrdU, or Lck with Fc block for 30–60 min in medium B (Caltag); washed with Perm/Wash (BD Biosciences); and fixed. BrdU staining also included 4.1 mM CaCl₂ and 20 U DNaseI (Roche). For phospho-ZAP/Syk analysis, cells were stimulated with 20 μg/ml anti-CD3 (2C11), followed 20 s later by 50 μg/ml goat anti–hamster (Jackson ImmunoResearch Laboratories) secondary cross-linker and fixed to terminate stimulation. Cells were permeablized using 90% ice-cold methanol for 30 min, stained with 2 μl phospho–Zap-70 (Tyr319)/Syk (Tyr352) antibody (Cell Signaling) for 1 h, stained with surface markers and donkey anti–rabbit–PE (Jackson Immunoresearch Laboratories) for 1 h, and fixed. Data were collected on a FACSCalibur (BD Biosciences) and analyzed using FlowJo software (TreeStar, Inc.). For RNA and immunoblot analysis, cellular subsets were sorted using a sorter (MoFlo; DakoCytomation).

8–12-wk-old rag1^−/− and rag1^−/−zap70^−/− mice were used for in vivo anti-CD3 treatment. Anti-CD3 mAb 2C11 was injected i.p. at 10, 30, or 100 μg/animal diluted to 500 μl in PBS, and mice were killed and analyzed 5 d later. All experiments were repeated at least twice with similar results, and data are represented as the mean ± SD. For rag1^−/− and rag1^−/−syk^−/− experiments, chimeric mice were made from FL and were used 5–6 wk after reconstitution.

After stem cell transfer and reconstitution, competitive repopulation was assessed as shown in Fig. S1. To ensure an unbiased assessment, all subpopulations were identified first, followed by interrogation of the CD45 allele marker.

We thank A. Roque for invaluable assistance with animal husbandy and J. Kropp for assistance with real-time PCR primer design. We thank M. Hermiston for advice on animal experiments and C. MacArthur and S. Jiang for cell-sorting expertise. We thank C. Lowell and N. Killeen for critical readings of the manuscript and helpful discussions, as well as Weiss lab members for helpful discussions.

The authors have no conflicting financial interests.