EphA Receptors Inhibit Anti-CD3-Induced Apoptosis in Thymocytes¹

The differentiation of T cells occurs primarily in the thymus both during embryo development and in adult organisms. Recent observations based on RT-PCR, Northern blot, and immunohistochemical analysis suggest that receptors of both the erythropoietin-producing hepatocyte (Eph)⁴ A (EphA1, EphA2, EphA3, and EphA4) and EphB (EphB6 and EphB2) subfamilies of tyrosine kinase receptors are expressed in the thymus (1, 2, 3, 4, 5). Moreover, the EphA1 and EphA3 receptors were shown to be specifically expressed by thymocytes, whereas expression of the complementary ephrin-A1 and ephrin-A2 ligands was observed in thymic epithelial cells (3). Although these experimental approaches do not specifically detect cell surface expression of Eph receptors or examine their functional status, this expression pattern suggests that thymocyte-expressed EphA receptors may undergo ligand stimulation and modulate thymocyte behavior and differentiation.

Thymocytes that interact strongly with self-Ags have the potential to differentiate into self-reactive T lymphocytes and subsequently to initiate autoimmune responses. Therefore, these cells are negatively selected by induction of apoptosis during the double-positive (DP), CD4⁺CD8⁺, stage of their differentiation (6, 7). Oppositely, the survival and differentiation of DP thymocytes into mature T cells (positive selection) is mediated by low avidity TCR-self Ag interaction. Cosignaling through numerous receptors contributes to fine-tuning this process of selection (8). The possibility that Eph receptors may act as coreceptors for the TCR in thymocytes is greatly strengthened by the recent observation of Luo et al. (9) that Eph receptors colocalize with the TCR complex in T lymphocytes.

The Eph receptors form the largest family of receptor tyrosine kinases (RTK), with at least 14 members. Membrane-bound molecules termed ephrins operate as ligands for the Eph receptors, initiating their signaling upon interaction (10, 11, 12). Ephrins are attached to the cell membrane either via a glycosylphosphatidylinositol anchor (ephrin-A type) or by a transmembrane domain (ephrin-B type) (13, 14). Both subfamilies of ephrins can also operate as receptors and transmit signals inside the cell in a process termed reverse signaling upon interaction with Ephs (15, 16, 17, 18, 19, 20).

The Eph receptors are divided into subfamilies A and B, according to their sequence similarity and ligand binding preferences (interaction with ephrin-A or ephrin-B ligands, respectively) (11). Within both subfamilies the specificity of ligand-receptor interaction is degenerate, and each given receptor can bind multiple ligands, although with distinct affinities. We have also recently demonstrated that different Eph receptors of the same subfamily can form heterocomplexes and cooperate in their signal transduction (21). It is likely that under physiological conditions ephrins do not, in fact, activate a single specific Eph receptor, but, rather, a complex of subfamily receptors. These properties make the study of Eph function and signaling extremely complex; consequently, many experiments often describe the function not of a particular receptor, but of the combination of receptors from a particular subfamily expressed by the cell of interest (22).

Eph receptors play a critical role in developmental and differentiation processes (23, 24) and are known to be involved in formation of the CNS (14), angiogenesis (25), and hemopoietic stem cell differentiation (26). The majority of their biological functions are executed by control over cell adhesion/repulsion and cell morphology. These cellular responses are based on the ability of Ephs to regulate the activity of integrin receptors and cytoskeletal rearrangements (27, 28, 29, 30, 31, 32). Eph receptors have been shown to interact with a variety of signaling molecules, including Cbl (21); Fyn, Src, and Abl tyrosine kinases (33, 34, 35, 36, 37); r-Ras (27); and Nck, Crk, and SLAP adaptor proteins (34, 38, 39). Unlike many other RTKs, the Eph receptors are not commonly capable of activating the Ras-Raf-MAPK pathway. Conversely, members of both the EphA and EphB subfamilies have been reported to inhibit activation of the MAPKs ERK1 and ERK2, although this inhibition appears to be cell type restricted (40, 41).

Because both Ephs and ephrins are cell surface molecules, Eph receptor activation requires cell-cell attachment and is polarized toward areas of cell-cell contact (42). Thus, Eph receptors would appear to be ideally positioned to regulate TCR engagement upon interaction with APCs.

In this paper, we show that functional EphA receptors are expressed on the surface of the murine thymocytes, including CD4⁺CD8⁺ DP and CD4⁺CD8⁻, CD4⁻CD8⁺ single-positive (SP) cells. We demonstrate that stimulation of murine thymocytes with ephrin-A1 attenuates TCR-mediated induction of apoptosis in DP thymocytes. Ephrin-A1 also inhibits CD25 (IL-2Rα) and CD69 expression in both DP and SP thymocyte populations and IL-2 secretion by the SP thymocyte population. Finally, we demonstrate that EphA receptor activation in thymocytes inhibits TCR-mediated initiation of the RAS-MAPK pathway. These findings all strongly suggest that EphA receptor engagement is antithetical to TCR responses and thus, given high ephrin-A1 and ephrin-A2 expression by thymic epithelial cells (3), will play an important role in thymocyte differentiation in vivo.

Monoclonal anti-phosphotyrosine was obtained from Upstate Biotechnology. Abs to MAPK and phospho-MAPK were purchased from Santa Cruz Biotechnology. Soluble dimerized ephrin-A1-Fc, ephrin-B3-Fc, and fibroblast growth factor receptor (FGFR)-Fc fusion proteins were purchased from R&D Systems. Human IgG was obtained from Serotec. Anti-mouse CD25, CD4, CD8, CD5, CD69, and CD3 and matching isotype controls for flow cytometric analysis were purchased from BD Biosciences. Anti-CD3 (145-2C11) and anti-CD28 (37.51) for cell stimulation were also obtained from BD Biosciences. CD4 and CD8 Dynabeads and CD4 Detachabead were purchased from Dynal Biotech.

Thymi were obtained from 5- to 7-wk-old female BALB/c mice. Mononuclear cells were isolated by Percoll density gradient centrifugation, and adherent cells were removed by incubation on plastic at 37°C for 60 min. The resulting thymocytes are typically 95% CD3⁺. Purified DP and SP thymocyte subsets were obtained by cell sorting anti-CD4-Cy-PE- and anti-CD8-PE-labeled thymocytes. DP thymocytes were also isolated by Percoll density fractionation (43, 44). Purified thymocyte populations were analyzed by flow cytometry for purity >93%. All animal experiments were conducted according to Canadian federal regulations and Hospital for Sick Children guidelines.

Ephrin-A1-Fc and anti-CD3 were immobilized at 20 μg/ml, unless otherwise indicated, on 24-well plastic tissue culture dishes for 1.5 h at 37°C. Soluble protein was removed by three washes with PBS. Human IgG or irrelevant FGFR-Fc fusion protein were immobilized at 20 μg/ml in all ephrin-A1-negative points as a control for the nonspecific effects of the Fc portion of the ephrin-A1-Fc fusion protein. Concentrations of human IgG were adjusted where required to keep the protein concentration constant. Anti-CD28 was used as a soluble Ab at 1 μg/ml in the culture medium. Cells (2 × 10⁶) were added in 1 ml of culture medium to each well.

Thymocytes (2 × 10⁶) were incubated in 1% serum with or without immobilized ephrin-A1-Fc fusion protein and in anti-CD3 Ab with or without soluble anti-CD28. Immobilized irrelevant Ab or FGFR-Fc fusion protein were used as a control for ephrin-A1-Fc where necessary. CD25/CD69 expression was then analyzed by staining with FITC-labeled anti-CD25 or CD69-PE and relevant isotype controls. For analysis of apoptosis, 1 × 10⁶ treated thymocytes were stained with FITC- or PE-labeled annexin V. The annexin V-FITC and annexin V-PE apoptosis detection kits (R&D Systems) were used according to the manufacturer’s instructions. In all cases thymocytes were gated to exclude debris, and a minimum of 10,000 cells were analyzed for each treatment. All experiments were repeated at least three times.

Thymocytes were stimulated with anti-CD3 and ephrin-A1 as indicated for 20 h at 37°C, and the conditioned medium was collected. The concentration of IL-2 in each sample was determined by ELISA according to the manufacturer’s instructions (R&D Systems).

Cells were quickly resuspended in ice-cold lysis buffer consisting of 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Nonidet P-40, 5 mM EDTA, 1 mM PMSF, and 1 mM sodium orthovanadate. After solubilization on ice, debris was removed by centrifugation, followed by addition of SDS reducing sample buffer. Samples were separated on SDS-polyacrylamide gels and transferred to nitrocellulose membranes (Amersham Biosciences). Membranes were blocked for 1 h at room temperature with 5% blotting grade nonfat milk in PBS/TBS. Immunoblotting Abs were placed in PBS-0.1% Tween 20/TBS-0.1% Tween 20 and incubated overnight at 4°C. After washing, bound Abs were detected with HRP-conjugated secondary Ab (Amersham Biosciences) and LumiGlo chemiluminescent reagents (Kirkegaard & Perry Laboratories).

Previous studies have demonstrated that receptors of the EphA subfamily are expressed in the thymus and thymocytes (1, 2, 3). However, the cell surface expression and functional status of these receptors have not been characterized. To examine the cell surface expression and ligand-binding ability of EphA receptors on murine thymocytes, we incubated purified murine thymocytes on ice with soluble recombinant ephrin-A1 fused to the Fc portion of human IgG, stained with fluorescent-labeled anti-human Fc, and detected bound ligand by flow cytometry. Incubation with anti-human Fc alone was used as a specificity control. This assay analyzes both the expression of EphA receptors on the outer membrane of the thymocytes and their ability to bind the ephrin-A1 ligand. Ephrin-A1 is known to be expressed on thymic epithelial cells (3) and thus has the potential to interact with thymocyte-expressed EphA receptors in vivo. Ephrin-A1 interacts with all receptors of the EphA subfamily (10), so cell surface expression of any EphA receptor should be detected. Flow cytometric analysis of revealed that most murine thymocytes expressed EphA receptors capable of binding ephrin-A1 (or ephrin-A3). Approximately 80–85% of the DP thymocyte population expressed EphA receptors on their surface, and even greater expression (> 90%) was detected on CD4⁺CD8⁻ and CD4⁻CD8⁺ SP cells (Fig. 1 A).

Ligand binding to the extracellular domain of RTKs induces receptor dimerization or oligomerization and stimulates their intrinsic tyrosine kinase activity. As a consequence, RTKs undergo autophosphorylation, providing specific docking sites for cytoplasmic signaling proteins and thus initiating specific signaling cascades (45). To confirm the functional activity of the EphA receptors expressed by murine thymocytes, we stimulated the cells with dimeric ephrin-A1-Fc fusion protein at 37°C, which is sufficient to induce receptor dimerization and subsequent phosphorylation (46, 47). In the absence of precipitating Abs for most ephrin-A receptors, we used the previously reported ability of the ephrin-A1 ligand to pull-down EphA receptors (41). Ligand-receptor complexes were pulled down from the lysates of ephrin-A1-stimulated cells by addition of protein G-Sepharose beads (which bind the Fc portion of the fusion protein). Ephrin-A1, but not an unrelated FGFR-Fc fusion protein, precipitated a singular tyrosine-phosphorylated protein of ∼130 kDa, the expected size for a member of the EphA subfamily (Fig. 1 B). These observations suggest that the majority of murine thymocytes express functional receptor tyrosine kinases of the EphA subfamily on their surface that are capable of interacting with EphA ligands, undergoing tyrosine phosphorylation and thus initiating intracellular signals.

Having demonstrated the expression of functional EphA receptors on thymocytes, we examined the ability of EphA activation to influence TCR-mediated responses, because Eph receptors have been shown to colocalize with the TCR complex in mature T lymphocytes (9). DP (CD4⁺CD8⁺) thymocytes, strongly stimulated through their TCR by self-MHC/Ag complex, are negatively selected by induction of apoptosis. This process that can be mimicked in vitro by stimulation with high concentrations of immobilized anti-CD3 Ab (48). To investigate a potential role for EphA receptors in regulating the induction of apoptosis, we costimulated purified murine thymocytes with immobilized anti-CD3 and recombinant ephrin-A1-Fc and analyzed the induction of apoptosis by annexin V-FITC binding. Human IgG was used as a specificity control in all points without ephrin-A1.

Anti-CD3 stimulation induced thymocyte apoptosis as expected, whereas stimulation with ephrin-A1 alone had no effect. However, costimulation with ephrin-A1 was found to strongly decrease the anti-CD3-induced response (Fig. 2). On the average, 15–25% of thymocytes (2 × 10⁶ cells/assay) underwent apoptosis in response to anti-CD3 in 20 h. The 20-h incubation period was chosen because although greater anti-CD3-induced apoptosis could be achieved by extension of the time, a significant increase in spontaneous cell death was also observed. At least 10,000 cells/assay were analyzed by flow cytometry.

The inhibitory effect of ephrin-A1 on anti-CD3-induced apoptosis was concentration dependent and reached its maximum at an ephrin-A1 concentration of 10 μg/ml (Fig. 3,A), whereas costimulation with an unrelated FGFR-Fc fusion protein was ineffective even at 20 μg/ml (Fig. 3,B). Similarly, little effect on induction of cell death was observed upon costimulation with the ephrin-B subfamily ligand, ephrin-B3 (Fig. 3 C), demonstrating the specific nature of the ephrin-A1-induced responses. Ephrin-A1 alone did not decrease the level of spontaneous cell death of cultured thymocytes, suggesting that it acted to specifically inhibit anti-CD3-induced cell apoptosis, rather than as a general survival factor. This also suggested that stimulation with ephrin-A1 could not rescue cells already committed to the apoptotic pathway.

Although anti-CD3 was quite effective at inducing thymocyte apoptosis in our hands (15–25% of cells), much higher levels of programmed cell death could be achieved by costimulation through the CD28 coreceptor. Anti-CD3/CD28 stimulation induced apoptosis in 40–50% of thymocytes in 20 h, but this could also be strongly inhibited by ephrin-A1 (Fig. 3 D).

Because the process of negative selection primarily involves CD4⁺CD8⁺ DP thymocytes, we specifically analyzed apoptosis in this population. To exclude possible secondary effects on DP cells due to simultaneous inhibition of SP cell responses, such as cytokine secretion, we examined the response of purified DP cells. Most DP cells express EphA receptors, and in agreement, costimulation with ephrin-A1 significantly inhibited TCR-induced cell death (Fig. 3,E). Induction of cell death was repeatedly inhibited by 80–90%, suggesting that EphA receptors may help regulate the selection process. Anti-CD3 and anti-CD3/CD28 stimuli also induced apoptosis in CD4 SP thymocytes (Fig. 3 F), but not as efficiently in CD8 SP cells (not shown). CD4 SP responses to both stimuli were strongly inhibited by ephrin-A1 costimulation, as observed for DP thymocytes.

In response to TCR stimulation, thymocytes up-regulate expression of the IL-2R α-chain (CD25). We observed CD25 up-regulation upon stimulation of the total thymocyte population with immobilized anti-CD3 (Fig. 4,A), typically in 20–25% of thymocytes after 20 h (2 × 10⁶ cells/assay). Costimulation with ephrin-A1 was found to strongly inhibit this response. Inhibition was ephrin-A1 concentration dependent, and immobilization at 10 μg/ml was sufficient to significantly suppress TCR-mediated induction of CD25 expression, typically demonstrating inhibition of 50–70%. In contrast, control FGFR-Fc protein was ineffective even at 20 μg/ml. A more detailed analysis of CD25 expression in CD4⁺CD8⁺ DP and CD4⁺ SP cells revealed significant blockages in CD25 up-regulation in both populations by ephrin-A1 (Fig. 4, B and C). Although CD25 induction in 20 h was much greater on SP cells (average, 50% of cells) than on DP cells (average, 8–12% of cells), both were strongly inhibited by ephrin-A1. Analysis of CD25 up-regulation after only a short period of stimulation (8 h), to control for the increased viability seen when ephrin-A1 was included with the proapoptotic stimulus, revealed a similar inhibition (Fig. 4,D). Both CD4 and CD8 SP thymocyte populations demonstrated increased CD25 expression that was strongly inhibited by ephrin-A1 (Fig. 4,E). Because within the 8-h time period the induction of CD25 on DP cells could not be detected, no analysis of these cells was possible. However, analysis of purified DP cells stimulated over a 20-h period revealed the same pattern as that observed for DP cells within the total thymocyte population. Ephrin-A1 strongly inhibited the induction of CD25 by both anti-CD3 and anti-CD3/CD28 (Fig. 4 F).

We also examined the ability of ephrin-A1 stimulation to block TCR-initiated secretion of the IL-2 cytokine. The source of IL-2 in the thymus is unclear, but appears to include secretion resulting from TCR-mediated stimulation of SP thymocytes. To determine whether EphA receptor activation might also inhibit the production of IL-2, we analyzed conditioned media from anti-CD3- and anti-CD3/ephrin-A1-stimulated total murine thymocytes for IL-2 by ELISA. In agreement with previous observations (48), the stimulation of thymocytes in vitro with immobilized anti-CD3 was sufficient to induce IL-2 production. This was, however, dramatically attenuated upon costimulation with ephrin-A1 (Fig. 5,A). A separate analysis of anti-CD3-induced IL-2 production by FACS-purified DP and SP thymocyte populations revealed that the SP cells were indeed responsible for all the IL-2 production in culture, and this could be inhibited by ephrin-A1 (Fig. 5 B). Thus, EphA receptor activation would appear to inhibit TCR signaling, resulting in both IL-2Rα expression and IL-2 secretion.

Ephrin-A1 costimulation was also found to inhibit anti-CD3 induced up-regulation of CD69, a much earlier marker of activation (Fig. 6,A). Significant CD69 expression could be observed 7–8 h after stimulation with either anti-CD3 alone or combined with CD28 (Fig. 6,B). However, in both cases, ephrin-A1 inhibited induction of CD69 expression typically by 50%. CD4/CD8 subpopulation analysis revealed CD69 induction in all subsets (at an equivalent time, CD25 induction was only observed on SP cells) and inhibition of all subsets by ephrin-A1 (Fig. 6,C). Examination of the DP thymocyte population based on relative CD3 expression (low and medium (immature preselection DP thymocytes) and high (mature, positively selected DP thymocytes)) revealed that the induction of CD69 expression in the CD3^med and CD3^high groups could be strongly inhibited by ephrin-A1 (Fig. 6,D), although no significant inhibition was observed in the most immature CD3^low cells. We previously demonstrated that ephrin-A1 costimulation inhibited apoptosis of purified DP cells (Fig. 3 E), and as expected, inhibition of apoptosis (and CD25 expression) could be observed in the CD3^med-high subpopulations (not shown), which constitute most of these thymocytes. These findings suggest EphA receptor activation inhibits multiple aspects of thymocyte responses to TCR signaling, and that this inhibitory response is found in all but the earliest stages of thymocyte differentiation.

Ras-MAPK pathway activation is one of the major events initiated by TCR engagement and is crucial for the development of a variety of TCR-mediated responses (49). Recent observations have demonstrated that EphA subfamily receptors can inhibit the Ras-Raf-MAPK pathway in some types of cell (41). To follow the influence of EphA receptors on anti-CD3-induced activation of the Ras-MAPK pathway in thymocytes, we measured ERK1/ERK2 MAPK activation in anti-CD3- and anti-CD3/ephrin-A1-stimulated cells with anti-phospho-ERK Ab, which specifically recognizes phosphorylated, and consequently activated, ERKs. Anti-CD3 stimulation of thymocytes caused a significant increase in ERK phosphorylation. However, this was significantly reduced upon costimulation with ephrin-A1 (Fig. 7). This suggests that in thymocytes, as in some other types of cells, EphA receptors can negatively regulate Ras-Raf-MAPK pathway activation and that EphA receptor-mediated blockage of TCR-induced responses may be at least in part due to inhibition of the MAPK pathway.

The differentiation of thymocytes is strictly guided by a complex combination of stimuli provided by cells of nonlymphoid origin in the thymus. Thymic epithelial cells have been demonstrated to express ephrin-A1 and ephrin-A2 ligands (3), and we have shown in this study that functional receptors of the EphA subfamily are expressed on the surface of the majority of murine thymocytes, including both DP and SP populations. Both negative selection of self-reactive DP thymocytes and positive selection of weakly activated DP thymocytes are critical in thymocyte maturation and the development of a healthy functional T cell repertoire. Our experiments demonstrate that EphA receptors can modulate TCR-induced apoptosis of purified thymocytes in culture and are therefore likely to regulate the selection process in vivo.

Our findings suggest that the level of ephrin-A expression on thymic APCs may help set the intensity of TCR engagement required to induce effective TCR pathway signaling and subsequently selection. This would effectively provide a defined constant against which TCR signaling in individual thymocytes must be measured to determine whether the level of TCR engagement is ineffective (resulting in deletion), productive (resulting in positive selection), or too strong (resulting in negative selection). Our findings are in accordance with observations in a number of systems that show EphA receptor signaling to primarily inhibit signaling responses and cytoskeletal rearrangements. It is likely that a similar response is induced by EphA receptor activation in thymocytes, causing cytoskeletal organization that is not conducive to productive TCR engagement and signaling.

The failure to see inhibition of responses in the most immature CD3^low DP thymocyte population suggests that TCR signaling in these cells may not be regulated by EphA receptor activation. Many of these cells differ from the more mature CD3^med/high DP thymocytes, being in the preselection stage (50, 51) where only the TCRβ-chain is rearranged. Functional rearranged TCRβ pairs with invariant pre-TCRα-chain to form a CD3-associated pre-TCR complex. Pre-TCR signaling then triggers expansion and further differentiation into TCRαβ mature T cells, a process known as β selection (52). Although EphA receptors may play a role in modulating the intensity of TCR engagement to affect selection thresholds in TCRαβ thymocytes, modulation of signals through the pre-TCR complex would not appear desirable in this preselective phase. In agreement, flow cytometric analysis of ephrin-A-Fc binding to the CD3^low thymocyte population suggests that these cells express lower levels of EphA receptors than more mature DP and SP cells (not shown).

In contrast to many RTKs, which commonly activate the Ras-MAPK pathway through engagement of the Grb2/Sos complex, EphA receptors were recently shown to inhibit activation of the MAPK pathway in certain cell types (41). Our findings expand the variety of responsive cell types, demonstrating that EphA receptors can also inhibit TCR-mediated activation of the Ras-MAPK pathway in thymocytes. Activation of the MAPK pathway is a major event in TCR signaling, critical for the induction of a variety of TCR-mediated responses in thymocytes (49) including positive regulation of IL-2 and possibly CD25 expression (53, 54, 55). The negative regulation of Ras-MAPK pathway activation may provide a mechanism by which EphA receptors can modulate TCR signaling and help to define the border between positive and negative selection of thymocytes.

Munoz et al. (3) demonstrated that application of soluble EphA receptor to fetal thymus organ cultures (FTOCs) increased apoptosis of DP thymocytes, which is in agreement with our observations. However, addition of soluble ephrin-A protein to FTOC produced the same effect. As the investigators themselves admitted (3), interpretation of their experiments is complicated by the fact that dimeric soluble EphA-Fc and ephrin-A-Fc fusion proteins were used in their FTOC system. Both proteins can certainly disrupt normal EphA-ephrin-A interaction; however, being dimeric in nature they will also activate endogenous ephrin-A or EphA signaling, potentially initiating complex responses from both the thymocytes and surrounding nonlymphoid cells. As a result, the authors were unable to conclude whether it was interference with Eph or ephrin signaling (or both) that was responsible for the observed decrease in thymocyte survival (3). Although monomeric soluble Ephs (or ephrins) do not have activating potential (47, 56) and their application to FTOC could be proposed, this would still not permit differentiation between the consequences of inhibiting EphA receptor activation and inhibition of reverse signaling through ephrin-A ligands. Not only is FTOC a complex model, but the bidirectional nature of Eph-ephrin signaling, combined with the expression of both EphA and ephrin-As on both thymocytes and surrounding nonlymphoid tissue, makes interpretation of experimental results difficult. Our experiments using costimulation of purified thymocytes through TCR and EphA receptors may therefore provide a less ambiguous approach to addressing the role of EphA receptors in the process of TCR-induced apoptosis and suggests that they are negative regulators of this process.

In summary, under our in vitro conditions with purified thymocytes, EphA receptor activation is clearly capable of inhibiting signaling through the TCR and ultimately blocking induction of apoptosis. This strongly suggests that signaling through EphA receptors may be a critical factor in determining thymocyte fate, by regulating selection thresholds in the thymus.

We thank Martina Nikolic for technical assistance.

The authors have no financial conflict of interest.