Mislocalization of Lck impairs thymocyte differentiation and can promote development of thymomas

Thymocyte differentiation is regulated at 2 key checkpoints. The first acknowledges successful rearrangement and expression of a T-cell receptor β (TCRβ) chain, which, in combination with the pre-Tα chain, permits ligand-independent signaling. As a result, CD4⁻CD8⁻ double-negative 3 (DN3) thymocytes enforce allelic exclusion of the TCRβ locus, proliferate, and express CD4 and CD8 coreceptors. The second checkpoint occurs upon expression of a rearranged TCRα chain by CD4⁺CD8⁺ double-positive (DP) thymocytes, which facilitates ligand-dependent signals transduced by the αβTCR interacting with the major histocompatibility complex (MHC) and results in 3 distinct cell fates: (1) “death by neglect” in the absence of signal; (2) negative selection, in the case of high-avidity interactions; and (3) positive selection from low-avidity interactions. Src-family protein tyrosine kinases (SFKs), particularly Lck, are critical for transition through these checkpoints.

Studies have shown that thymocyte development is substantially blocked at the DN3 stage in the absence of Lck, and is completely blocked in the absence of both Lck and Fyn.¹  Furthermore, positive selection of DP thymocytes is greatly impaired and few mature T cells are present in the periphery of Lck^−/− mice. Loss of Fyn alone does not impair thymocyte development, and peripheral T cells are present in normal numbers in Fyn^−/− mice.²  Therefore, although SFKs have partially redundant functions in thymopoiesis, Lck is critical for normal thymocyte development and Fyn plays a minor compensatory role.

Several studies have focused on the role of functional domains within Lck and Fyn. Both proteins have N-terminal sites of myristoylation and palmitoylation critical for membrane association, which share little homology and are called “unique” domains. In addition, they have highly homologous Src homology 3 (SH3) and SH2 domains, a catalytic domain, and a C-terminal-negative regulatory tyrosine residue (Tyr505 for Lck, Tyr528 for Fyn).^3,4  This unique domain is critical for the association of Lck with CD4 and CD8,^5-7  and is required for recruitment of Fyn to the CD3 complex.⁵  Furthermore, the unique domain might be important in the association of Lck with phosphatases and dephosphorylation of Tyr505 by CD45,^8,9  whereas studies using truncated versions of Lck have shown that the unique domain may also affect substrate specificity.¹⁰ 

Given that the unique domains are key in determining the cellular localization of Lck and Fyn, it seems likely that these regions are fundamental determinants of the specific roles of these kinases in vivo. We sought to test this hypothesis by generating transgenic mice on an Lck^−/− background that express a “domain-swap” version of Lck in which the unique domain has been replaced with that of Fyn; these are called Fyn-unique Lck (FU-Lck). T-cell development in these mice was compared with control transgenic mice expressing similar levels of wild-type (WT) Lck. These experiments showed that the Lck unique domain was important for both DN-DP transition and selection of single-positive (SP) cells during thymocyte development. Interestingly, when domain-swap FU-Lck transgenic mice, but not WT Lck transgenic mice, were crossed to an Lck-inducible transgenic background, thymomas developed in 100% of individuals. This process differed from previous reports of thymomas caused by overexpression of dysregulated Lck in that tumor development was dependent on the expression of TCR components. These data indicate a critical role for the unique domain of Lck in the normal development of T cells, and provide a mechanistic basis for the differing roles of Lck and Fyn in T cells.

The FU-Lck transgene was constructed by swapping the Lck unique domain (a.ac. 1-61) for the Fyn unique domain (a.ac 1-85) and adding a C-terminal V5 epitope tag. FU-Lck cDNA was sequenced and cloned into the VA expression vector containing the human CD2 control regions.¹¹  Control Lck cDNA containing a V5 tag was subcloned into the VA vector. Transgenic mice were generated by pro-nuclear injection into (CBAxC57Bl/10)F₂ fertilized oocytes. Mice positive for transgene expression were identified by slot-blot and polymerase chain reaction (PCR) analysis and bred to an Lck^−/− background.¹²  Where indicated, mice were crossed with Lck-inducible (Lck^ind) Lck^−/−13  and Lck^indLck^−/−Rag1^−/− F5 TCR-transgenic mice (Rag1 stands for recombination activating gene 1).¹⁴  Lck^ind mice were fed doxycycline in their food (1 mg/g) from birth. C57BL/6 mice served as WT controls. Mice were maintained under local and United Kingdom Home Office guidelines.

Antibodies were from eBioscience, BD Pharmingen, or Cell Signaling Technology. For intracellular staining, cells were permeabilized with 0.5% saponin or fixed in 2% formaldehyde and permeabilized in 90% ice-cold methanol. Samples were collected on a flow cytometer (FACSCalibur, BD BioSciences) and analyzed using FlowJo Version 8.8.6 software (TreeStar).

For analysis of Ca²⁺ mobilization, thymocytes were loaded with indo-1-AM calcium sensor dye (Sigma-Aldrich) and stained with biotinylated TCR (H57-597) and CD4 (RM4.5) monoclonal antibodies (mAbs). CD8α-PE (53-6.7) and CD4-FITC (YTA3.1) Abs were used to identify thymus cell populations. Biotinylated Abs were cross-linked using streptavidin-allophycocyanin conjugates, and samples were collected on a flow cytometer (LSR II; BD BioSciences).

DNA was prepared from thymocytes after lysis in 100mM Tris (pH 8), 5mM ethylenediaminetetraacetic acid (EDTA), 0.2% sodium dodecyl sulfate (SDS), 200mM NaCl, and 100 μg/mL proteinase K. The primers used were (Dβ2.1) GTAGGCACCTGTGGGGAAGAAACT and (Jβ2.7) TGAGAGCTGTCTCCTA C TATCCATT.¹⁵ 

Cells were lysed in 1% Triton X-100, 0.5% n-dodecyl-b-D-maltoside, 50mM Tris-HCl (pH 7.5), 150mM NaCl, 20mM EDTA, 1mM NaF, and 1mM sodium orthovanadate containing protease inhibitors, and samples were separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE). For immunoprecipitations, lysates were precleared with 10 μL of protein A-Sepharose® (Sigma-Aldrich) slurry before immunoprecipitation with 2 μg of anti-CD3ϵ (2C11), anti-CD4 (YTA3.1), or anti-CD8β (YTS-156) coupled to Sepharose beads. Immunoprecipitates were washed with lysis buffer, and proteins were transferred to polyvinylidene difluoride membranes (Millipore). Membranes were incubated in blocking reagent (LI-COR Biosciences) before immunoblotting with anti-V5, anti-extracellular signal–regulated kinase 2 (anti-ERK2; Santa Cruz Biotechnology), anti-phosphotyrosine (anti-pTyr; 4G10), anti–ζ-chain-associated protein kinase 70 (anti-ZAP70), anti-ZAP70 pY319, anti-Lck, anti-Lck pY505, anti-Src pY416, and anti-caspase 3 (Cell Signaling Technology). Proteins were detected with secondary Abs and visualized using an infrared imaging system (Odyssey; LI-COR Biosciences).

To study the role of the unique domain of Lck, we generated transgenic mice expressing WT Lck (Lck^tg) or a domain-swapped version of Lck in which the unique domain was replaced with that of Fyn (FU-Lck^tg) (Figure 1A). Western blot analysis using a V5 mAb specific for an introduced epitope tag showed that line 1 of the Lck^tg mice and line 3 of the FU-Lck^tg mice expressed similar levels of protein (Figure 1B). In both cases, the level of expression of transgenic Lck was low (< 5% of WT levels; supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). These mice allowed us to determine the impact of expression of low levels of WT Lck on T-cell development, as well as the function of the Lck unique domain in this process.

Expression of either transgene had no effect on T-cell development in Lck^+/+ backgrounds (data not shown), so analysis was done on an Lck^−/− background. T-cell development was assessed by fluorescence-activated cell sorting (FACS) analysis. As previously reported for Lck^−/− mice,¹²  the proportions, but not the numbers, of DN cells were increased, whereas both the proportions and numbers of DP and CD4 SP thymocytes were decreased compared with WT controls (Figure 1C, Table 1). Despite the low expression levels, Lck^tg substantially rescued DP and SP cell numbers and proportions in Lck^−/− mice (Figure 1C, Table 1), although these did not reach WT levels. Expression of FU-Lck^tg resulted in an intermediate phenotype between that of the Lck^tg mice and that of the Lck^−/− mice. Thus, the numbers and proportions of DP and SP thymocytes in FU-Lck^tg mice were elevated compared with Lck^−/− mice, but were reduced by approximately 50% and approximately 85%, respectively, compared with Lck^tg mice (Figure 1C, Table 1). The majority of the CD8 SP thymocytes in both Lck^−/− and FU-Lck^tg mice were immature, TCR-negative cells (Table 1), indicative of a late block in progression to the DP stage.

Further analyses were performed to compare DN populations. An accumulation of CD44⁻CD25⁺ DN3 cells was apparent in Lck^−/− thymi (Figure 1D). Expression of the Lck^tg partially rescued this phenotype, although more DN3 cells were present compared with WT thymi. Expression of FU-Lck^tg resulted in a DN phenotype intermediate between that of Lck^tg mice and that of the Lck^−/− controls (Figure 1D). By contrast, proportions of γδTCR⁺ cells were similar in all groups of mice (Figure 1E).

Levels of TCRβ and CD5 expression on DP thymocytes were assessed. TCRβ expression on Lck^−/− DP thymocytes was elevated above that of WT cells (Figure 1F), reflecting the requirement for Lck in the down-regulation of surface TCR expression.¹⁶  TCR expression on FU-Lck^tg DP thymocytes was similar to Lck^−/− levels (Figure 1F), whereas expression on Lck^tg cells was intermediate between WT and Lck^−/−. Thymocyte CD5 expression is developmentally regulated and controlled by TCR signals and avidity.¹⁷  As expected, our results showed that few Lck^−/− DP thymocytes expressed high levels of CD5 (Figure 1F). More DP cells from FU-Lck^tg thymi expressed high levels of CD5 compared with Lck^−/− cells (34% and 14%, respectively). These data indicate that the FU-Lck protein can transduce TCR signals, resulting in the up-regulation of CD5 in DP thymocytes. Lck^tg DP thymocytes expressed levels of CD5 that were consistently above WT levels (Figure 1F), which may reflect increased expression of TCR on the former relative to WT DP cells.¹⁷ 

The ability of FU-Lck^tg and Lck^tg to facilitate positive selection of F5 TCR-transgenic thymocytes was addressed. In Rag1^−/−F5 TCR-transgenic mice, Lck expression is essential for the development of DP and CD8SP cells. Expression of either FU-Lck^tg or Lck^tg rescued DN-DP progression to a variable degree in Rag1^−/−Lck^−/− F5 mice (data not shown). However, conversion of DP to mature CD8 SP thymocytes was profoundly defective in mice expressing FU-Lck^tg compared with WT Lck or Lck^tg, which was also reflected in the reduced numbers of CD8SP cells (supplemental Figure 2A-B), confirming the importance of the Lck unique domain in positive selection.

These data show that the unique domain is critical for Lck function in T-cell development. Expression of low levels of WT Lck relieved the DN3 block in Lck^−/− mice to a greater extent than expression of similar levels of domain-swap FU-Lck, leading to differentiation of more DP cells. Similarly, DP-SP progression was impaired in FU-Lck^tg thymi because fewer cells up-regulated CD5 and progressed to mature SP cells than in Lck^tg animals. However, the fact that FU-Lck^tg expression partially rescues DP and SP cell numbers in Lck^−/− mice suggests that possession of the specific unique domain is not an absolute prerequisite for Lck function in αβ T-cell development. In contrast to development of αβ lineage T cells, numbers of thymic γδTCR⁺ cells were previously shown to be unaffected by Lck deficiency,¹⁸  and similar numbers of γδTCR⁺ cells were present in the thymi of all groups of mice examined in the present study. In contrast, thymic natural killer (NK) T-cell development was profoundly blocked in Lck^−/−, FU-Lck^tg, and Lck^tg mice, suggesting that the development of this T-cell subset was dependent on higher Lck expression levels than either αβ or γδ T cells (supplemental Figure 2B). This precludes analysis of the role of the Lck unique domain in invariant NKT (iNKT)-cell development in the current work.

Examination of peripheral T cells in Lck^tg- and FU-Lck^tg-transgenic mice showed reduced proportions of αβ T cells in lymph nodes from Lck^−/− and FU-Lck^tg mice compared with either WT or Lck^tg mice (Figure 2A), which is consistent with the observed thymus defects. Furthermore, the ratio of CD4:CD8 T cells was approximately 1.5:1 in WT and Lck^tg mice, whereas in Lck^−/− and FU-Lck^tg mice, this ratio was reversed (Figure 2B).

Quantification of CD4 and CD8 T-cell numbers showed that these were reduced in lymph nodes from Lck^−/−, Lck^tg, and FU-Lck^tg mice compared with WT mice, which is in agreement with the analysis of thymocyte development (Table 2). Lck^tg mice had 3.5-fold more CD4 T cells and 2-fold more CD8 T cells than FU-Lck^tg mice (Table 2). Nonetheless, numbers of CD4 T cells, but not CD8 T cells, were significantly higher in FU-Lck^tg mice compared with Lck^−/− mice (Table 2).

Interestingly, the proportions of γδ T cells were elevated in Lck^−/−, Lck^tg, and FU-Lck^tg compared with WT lymph nodes (Figure 2A). Furthermore, although absolute numbers of γδ T cells were similar in WT and Lck^−/− mice, they were approximately 2- and 4-fold higher in Lck^tg and FU-Lck^tg mice, respectively (Table 2). γδ T cells can develop thymically and extrathymically.¹⁹  It is unclear whether the greater numbers of γδ T cells found in Lck^tg and FU-Lck^tg mice resulted from expansion of thymic-derived γδ T cells in lymph nodes or through increased differentiation of extrathymic γδ T cells.

Levels of surface CD44 expression on αβ T cells are used as an indicator of prior antigen experience; high expression of CD44 is maintained on activated or memory cells. However, T cells generated in lymphopenic conditions also up-regulate CD44 as a result of homeostatic expansion.²⁰  Although the majority of WT CD4 and CD8 T cells expressed low levels of CD44 (Figure 2C-D), Lck^−/− and FU-Lck^tg CD4 and CD8 T cells expressed high levels of CD44 (Figure 2C-D). Most CD4 T cells from Lck^tg mice expressed low levels of CD44, although some expansion of CD44^hi CD8 T cells was apparent. In summary, these data indicate that defects in T-cell development in Lck^−/− and FU-Lck^tg mice result in the presence of reduced numbers of peripheral αβ CD4 and CD8 T cells that expand in the periphery, becoming CD44^hi. Interestingly, γδ T-cell numbers were not affected in Lck^−/− mice and were elevated in Lck^tg and FU-Lck^tg mice.

To circumvent the block in T-cell development in FU-Lck^tg mice, we introduced an inducible Lck transgene. In Lck^ind mice, Lck expression is controlled by the induction of a tetracycline-responsive transgene by adding doxycycline to the diet.¹³  When crossed to the Lck^−/− background, expression of the Lck^ind transgene bypasses defects in T-cell development and allows the accumulation of peripheral T cells. After doxycycline is removed from the diet, thymus involution occurs and peripheral T cells lose Lck protein expression.

Analysis of Lck^tg/Lck^ind mice maintained on a doxycycline diet showed that coexpression of these transgenes resulted in normal proportions of thymocyte populations (Figure 3A). In contrast, thymi from FU-Lck^tg/Lck^ind mice maintained on a doxycycline diet were grossly enlarged and necessitated euthanasia of mice > 12 wks old (supplemental Figure 3A). FACS analysis demonstrated the presence of increased proportions of CD8 SP cells (Figure 3A). Analysis of forward light scatter showed that DP and CD8 SP cells from FU-Lck^tg/Lck^ind mice were substantially larger than their counterparts in Lck^tg/Lck^ind mice (Figure 3B). Strikingly, FU-Lck^tg/Lck^ind DP and CD8 SP cells did not express detectable levels of surface αβTCR (Figure 3C). Furthermore, both DP and CD8 SP cells from FU-Lck^tg/Lck^ind mice expressed low levels of CD5 and high levels of CD24 compared with control cells (Figure 3D-E), indicative of an immature phenotype. The presence of enlarged thymi was apparent in 100% of FU-Lck^tg/Lck^ind mice analyzed at > 10 weeks of age (n = 17), and although the proportions of DN, DP, and SP cells varied between individual mice, in all cases CD4⁻CD8⁺TCR⁻ cells were present in increased numbers (supplemental Figures 3B-D). This phenotype was not simply a consequence of increased Lck levels, because the WT Lck^tg together with the Lck^ind transgene led to normal maturation and surface expression of TCR.

Previous data showed that the expression of an “activated” Lck protein in which the regulatory Tyr505 residue had been mutated to phenylalanine (Lck^Y505F) resulted in thymomas in mice.^21,22  In Lck^Y505F transgenic mice, transformation occurred at the DN stage and was independent of recombinase activity, because tumors developed in a Rag1^−/− background.^21,22  To determine whether the development of thymomas in FU-Lck^tg/Lck^ind mice occurred in an analogous manner, mice were crossed to a Rag1^−/− background. Surprisingly, the development of thymic abnormalities was completely ablated in FU-Lck^tg/Lck^ind/Rag1^−/− mice (Figure 3F). When an F5 TCR transgene was introduced to FU-Lck^tg/Lck^ind/Rag1^−/− mice, thymi were again grossly enlarged compared with littermate controls. As described for FU-Lck^tg/Lck^ind mice, an increased proportion of FSC^highCD4⁻CD8⁺TCR⁻ cells was apparent in the thymi of FU-Lck^tg/Lck^indRag1^−/− F5 mice (Figure 3F-H).

Development of thymomas in FU-Lck^tg/Lck^ind mice required either recombinase activity or the expression of a TCR transgene, yet cells from these thymi did not express surface TCR. However, intracellular staining revealed the presence of abundant intracellular TCRβ in FU-Lck^tg/Lck^ind and FU-Lck^tg/Lck^ind/Rag1^−/− F5 thymocytes (Figure 3I and data not shown). PCR analysis of TCR Dβ2.1-Jβ2.7 gene rearrangements was used to assess clonality of individual tumors,¹⁵  and the data suggested that tumors were predominantly oligoclonal (Figure 3J).

To determine whether tumors persisted in the absence of Lck^ind transgene expression, FU-Lck^tg/Lck^ind mice were maintained on doxycycline until 12 weeks of age, and the development of a typical thymoma was confirmed in one individual (supplemental Figure 4). Littermates were removed from doxycycline for a further 4 weeks. At the time of culling, all mice were healthy and thymi were greatly reduced in size compared with the tumor from the mouse analyzed at 12 weeks (data not shown). FACS analysis demonstrated variable expression of CD4 and CD8, with 2 of 3 thymi retaining a high proportion of CD4⁻CD8⁺ cells (supplemental Figure 4). By contrast, surface TCRβ expression was detected on thymocytes from all FU-Lck/Lck^ind OFF-dox mice. These data suggest that the growth of thymomas in FU-Lck/Lck^ind mice and the internalization of TCRβ by tumor cells requires continuous expression of the Lck^ind transgene.

Western blot analyses were performed to assess the biochemical basis for the thymic phenotype of FU-Lck^tg/Lck^ind mice. Levels of pTyr in lysates of control Lck^tg/Lck^ind thymocytes were low (Figure 4A). Higher levels of pTyr were apparent in lysates from FU-Lck^tg/Lck^ind mice, with bands running at approximately 55 kDa and 70 to 75 kDa being particularly hyperphosphorylated (Figure 4A). Lck activates the downstream kinase ZAP70, and autophosphorylation on residue Tyr319 is critical for ZAP70 function.²³  Using a specific antibody, we confirmed that ZAP70 Tyr319 was hyperphosphorylated in FU-Lck^tg/Lck^ind cell lysates (Figure 4B).

Experiments were undertaken to determine whether the hyperphosphorylated, approximately 55-kDa protein(s) represented Lck. Phosphospecific Abs showed that levels of Lck-inhibitory Tyr505 and activatory-Tyr394 phosphorylation were markedly increased in lysates from FU-Lck^tg/Lck^ind mice compared with those from Lck^tg/Lck^ind mice (Figure 4C). Additional analyses demonstrated that in some, but not all, thymomas, FU-Lck^tg protein was expressed at higher levels compared with Lck^tg protein in control mice (Figure 4D). Similarly, WT Lck protein levels were elevated approximately 2-fold in tumor cell lysates compared with control cell lysates (Figure 4D). Previous studies using intracellular staining indicated that Lck was expressed heterogeneously in thymocytes from Lck^ind-transgenic mice.¹³  The higher level of Lck in FU-Lck^tg/Lck^ind thymus lysates compared with Lck^tg/Lck^ind lysates could be a result of an outgrowth of cells expressing high levels of the Lck^ind transgene. Intracellular FACS staining demonstrated that this was indeed the case (Figure 4E and supplemental Figure 5).

Finally, it was possible that the elevated ZAP70 phosphorylation in tumor lysates was a result of the altered proportions of specific thymocyte subsets in FU-Lck^tg/Lck^ind mice compared with Lck^tg/Lck^ind mice. To address this possibility, intracellular FACS staining was performed, and the data indicated that levels of intracellular ZAP70 pTyr319 in gated DP cells were approximately 2- to 3-fold higher in cells from FU-Lck^tg/Lck^ind mice compared with controls (Figure 4F).

Cell lines are derived readily from Lck^Y505F thymomas by propagation in tissue-culture media containing serum without the addition of other growth factors.^21,22  Moreover, cell lines and thymocytes taken directly from Lck^Y505F tumors demonstrate resistance to apoptotic stimuli.^22,24  To determine the susceptibility to apoptotic stimuli, thymocytes from FU-Lck^tg/Lck^ind mice and control animals were subjected to γ-irradiation, cultured for 4 hours, and assessed for caspase activation by Western blot analysis. Blots showed that irradiation-induced activation of caspase-3, as judged by cleavage of the p32 zymogen and the appearance of the p17 active form, was comparable in all groups of mice (Figure 5). Furthermore, background levels of caspase-3 activation appeared somewhat higher in FU-Lck^tg/Lck^ind cell lysates, because levels of p17 were higher in cells that had not received irradiation compared with cells from control mice cultured under the same conditions (Figure 5). These data suggest that thymocytes from FU-Lck^tg/Lck^ind mice are sensitive to apoptotic stimuli and have elevated levels of spontaneous cell death compared with control thymocytes ex vivo.

The unique domain of Lck regulates association with CD4 and CD8,⁶  and is also important for the localization of Lck in the absence of coreceptor expression.²⁵  It was therefore useful to determine the intracellular localization of Lck^tg and FU-Lck^tg proteins in thymocytes. Immunoprecipitation of CD4 or CD8β followed by Western blot analysis showed that a large proportion of Lck^tg expressed in thymocytes of Lck^tgLck^−/− mice was coreceptor-associated (Figure 6A-B). Although a very small amount of FU-Lck^tg and Fyn (data not shown) coprecipitated with anti-CD4 and anti-CD8, this only occurred in the Lck^−/− background, presumably because WT Lck associates much better with coreceptors and displaces FU-domain–containing proteins. Nonetheless, the majority of FU-Lck^tg protein failed to coimmunoprecipitate with either CD4 or CD8. Furthermore, a small proportion of FU-Lck^tg but not Lck^tg was associated directly with the TCR complex, as shown by CD3ϵ immunoprecipitation (Figure 6). These data confirm that WT Lck^tg and FU-Lck^tg are differentially localized in thymocytes.

To ascertain the effect of altered localization of FU-Lck^tg on signal transduction directly, we analyzed antibody-induced Ca²⁺ flux in DP thymocytes. Indo-1-loaded cells were labeled with biotin-conjugated TCRβ ± CD4 Abs, and Ca²⁺ flux was monitored after cross-linking with streptavidin. Cross-linking TCR+CD4, but not TCR alone, induced a robust peak of intracellular Ca²⁺ flux in WT DP thymocytes (Figure 7A-B). In the absence of Lck, TCR + CD4 cross-linking induced only a small Ca²⁺ flux that was substantially delayed, as has been reported previously,¹⁴  and may reflect low-level Fyn association with CD4. Expression of either WT Lck^tg or FU-Lck^tg restored the ability of TCR+CD4 cross-linking to induce Ca²⁺ flux, although this was delayed in FU-Lck^tg compared with WT Lck^tg DP cells (Figure 7A). Interestingly, TCR cross-linking alone consistently induced a substantial Ca²⁺ response in FU-Lck^tg but not Lck^tg or WT DP thymocytes (Figure 7B). This response was independent of the levels of TCR, because gates were set on cells with equivalent surface TCR expression. Therefore, relocalization of FU-Lck to the TCR-CD3 complex has direct effects on signal transduction in DP thymocytes.

T-cell development and mature T-cell function are critically dependent on the activities of the SFKs Lck and Fyn. Lck is required for the initiation of TCR-signaling pathways, whereas Fyn contributes to these pathways but is also involved in the negative regulation of T-cell responses.^{1,12-14,26,27}  In the current study, we generated transgenic mice to address the role of intracellular localization and the function of the N-terminal unique domains of Lck. These experiments demonstrated that the unique domain is important for the role of Lck in T-cell development. Furthermore, coexpression of very low levels of mislocalized FU-Lck with an Lck-inducible transgene led to the aberrant proliferation of immature thymocytes, and ultimately in thymic tumorigenesis.

Previous studies addressed the function of the Lck unique domain through transfection of cell lines with mutant and truncated forms of the kinase. These studies showed that the unique domain regulates membrane²⁵  and coreceptor association, and might influence substrate specificity¹⁰  and interactions with regulatory phosphatases.⁸  However, the significance of these findings for the role of Lck in T-cell development had not been addressed previously. In an attempt to understand the molecular basis for the differing roles of Lck and Fyn, Abraham et al generated transgenic mice that expressed chimeric Lck-Fyn proteins in the absence of endogenous Lck.²⁸  In their study, substitution of the Lck catalytic domain for that of Fyn had little effect on DP-cell numbers but impaired positive selection. By contrast, substitution of the Lck N-terminal-unique SH3 and SH2 domains for those of Fyn severely impaired the production of DP cells in transgenic mice.²⁸  These data suggested that specific Lck catalytic activity was important for positive selection, whereas the N-terminal regulatory domains were critical for Lck function in DN cells. However, the individual requirements for the Lck SH3, SH2, and unique domains in DN cells were not clear from these experiments. A recent study demonstrated that a “knockin” mutation in the Lck SH3 domain resulted in a 50% reduction in thymocyte cell numbers in homozygous mice, due to a partial block in DN3 progression and in positive selection.²⁹  Our data indicate that replacement of the unique domain of Lck for that of Fyn also results in impaired DN3 progression. It remains possible that the low level of expression of FU-Lck^tg protein exacerbates the defect in DN thymocytes. Nonetheless, it is apparent that at a similar or even lower level of expression, WT Lck is superior in promoting DN3 progression. These data are consistent with data showing that the unique domain is important for Lck localization in coreceptor-negative cell lines.²⁵ 

Previous reports indicate that recruitment of Lck is important for the coreceptor function of CD4 and CD8,^30,31  and we found that DP maturation and positive selection were impaired in FU-Lck^tg mice. Furthermore, analysis of Ca²⁺ flux in DP thymocytes indicated that relocalization of Lck altered responses to stimulation with TCR ± CD4 cross-linking. Specifically, FU-Lck^tg DP cells were hyporesponsive to TCR+CD4 cross-linking but hyperresponsive to TCR cross-linking alone. While Ab-induced signals are likely to be markedly different from those required for selection processes in vivo, these experiments suggest that altered localization of Lck could fundamentally affect the types of signals transduced in primary thymocytes. While FU-Lck^tg expression resulted in an altered CD4:CD8 T-cell ratio, it should be noted that the defects in positive selection predominantly resulted in a quantitative reduction in SP cell numbers rather than in an overt, qualitative change in lineage selection. In this regard, both CD4 and CD8 SP cell numbers were dramatically reduced.

Expression of both the Lck^tg and the FU-Lck^tg transgenes was much lower than that of endogenous Lck, but despite this, the Lck^tg transgene substantially rescued thymus differentiation in Lck^−/− mice and was as efficient as WT at positive selection. Similarly, transgenic mice expressing only 3% of WT CD45 activity were shown to have normal numbers of thymocyte subsets and peripheral T cells.³²  If only low levels of Lck are required for T-cell development, why are higher levels expressed in WT animals? One possibility is that higher levels of expression are required for full responsiveness of peripheral T cells. However, CD4 T cells from Lck^tg mice up-regulate activation markers and undergo proliferation in response to CD3/CD28 stimulation in a similar manner to WT cells (data not shown). Another possibility is that while low levels of Lck expression permit thymocyte selection processes, the T-cell repertoire in these mice may be skewed. Although this may be the case, we did not see any evidence of the development of autoimmune disease, which can occur with muted signaling during differentiation.³³ 

During β-selection, coreceptors are not expressed, and it is unclear how Lck associates with the pre-TCR to facilitate signaling. Despite the ability of the FU-Lck^tg to associate directly with CD3 chains, it was much poorer than Lck^tg, which showed no such association, at promoting β-selection in an Lck^−/− background. Therefore, it was surprising that when coexpressed with an Lck^ind transgene, Fu-Lck^tg mice developed thymomas. Overexpression of Lck^Y505F was previously shown to lead to thymoma development, but this occurred independently of pre-TCR signaling.²²  In the current study, some aspect of overexpression was contributory to tumor formation because FU-Lck^tg mice on a WT Lck^+/+ background were tumor-free. However, control Lck^tg/Lck^ind cells, which express as much Lck as FU-Lck^tg/Lck^ind tumor cells, never developed into tumors. Therefore, some unique aspect of coexpression of FU-Lck^tg but not Lck^tg precipitated uncontrolled growth of this population. Development of thymic lymphoproliferation was dependent on expression of either Rag1 or a TCR transgene, and was mainly represented in an immature stage, indicating that pre-TCR signals can drive the phenotype, as might αβTCR signals. Signals through pre-TCR drive proliferation rather than differentiation and are ligand independent.³⁴  It is possible that mislocalization of FU-Lck to the CD3 chains initiates a signaling cascade that, in the presence of excessive levels of Lck, results in unregulated proliferation. A further, not mutually exclusive, explanation is that FU-Lck may disrupt the function of Fyn. It is established that Fyn can negatively regulate T-cell activation^26,35  and that FU-Lck might prevent Fyn-dependent feedback inhibition pathways; however, Fyn^−/− mice do not develop tumors. Nevertheless, it is clear that coexpression of FU-Lck results in dysregulation of WT Lck, because it shows hyperphosphorylation at both the activatory Y394 and the inhibitory Y505 residues.

The ability of signaling proteins to localize to specific cellular compartments is critical for TCR signal transduction. Intracellular localization is regulated dynamically through interactions mediated via specific functional domains. Mislocalization of signaling proteins can dramatically affect the outcome of TCR signaling. Thus, constitutive targeting of CD45 or SLP76 to plasma membrane lipid microdomains results in defective TCR signaling.^36,37  Lck has long been known to associate with the CD4 and CD8 coreceptors, and deletion of the Lck unique domain critical for coreceptor association results in a failure to induce interleukin-2 promoter activity in Jurkat cells.¹⁰  In the present study, we show that the FU domain cannot substitute for that of Lck in transgenic mice, and that a mislocalized FU-Lck protein is greatly impaired in mediating signals at all stages of T-cell development. Furthermore, aberrant signaling as a consequence of coexpression of FU-Lck and WT Lck results in thymomas. These data show that the regulation of intracellular localization by unique domains is critical for SFK function in vivo.

This work was supported by grants 0251 and 05096 from Leukemia & Lymphoma Research UK (to R.J.S., A.F., and R.Z.) and from the Medical Research Council UK (to R.Z., N.P., and A.I.M.).

Contribution: R.J.S. designed and performed experiments, analyzed data, and cowrote the paper; A.F. and N.P. performed experiments; A.I.M. designed experiments; and R.Z. designed the study and cowrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

The current affiliation for A.F. is Flow Cytometry Facility, London Research Institute, Cancer Research UK, London, United Kingdom.

Correspondence: Rose Zamoyska, Institute of Immunology and Infection Research, University of Edinburgh, West Mains Rd, Edinburgh EH9 3JT,United Kingdom; e-mail: Rose.Zamoyska@ed.ac.uk.