T Cell Receptor Signaling Inhibits Glucocorticoid-induced Apoptosis by Repressing the SRG3 Expression via Ras Activation*

Activation of T cell antigen receptor (TCR) signaling inhibits glucocorticoid (GC)-induced apoptosis of T cells. However, the detailed mechanism regarding how activated T cells are protected from GC-induced apoptosis is unclear. Previously, we have shown that the expression level of SRG3, a murine homolog of BAF155 in humans, correlated well with the GC sensitivity of T cells either in vitro or in vivo. Intriguingly, the expression of SRG3 decreased upon positive selection in the thymus. Here we have shown that TCR signaling inhibits the SRG3 expression via Ras activation and thereby renders primary thymocytes and some thymoma cells resistant to GC-mediated apoptosis. By using pharmacological inhibitors, we have shown that Ras-mediated down-regulation of the SRG3 gene expression is mediated by MEK/ERK and phosphatidylinositol 3-kinase pathways. Moreover, TCR signals repressed the SRG3 transcription through the putative binding sites for E proteins and Ets family transcription factors in the proximal region of the SRG3 promoter. Introduction of mutations in these elements rendered the SRG3 promoter immune to the Ras or TCR signals. Taken together, these observations suggest that TCR signals result in GC desensitization in immature T cells by repressing SRG3 gene expression via Ras activation.

Signals triggered by the activation of glucocorticoid (GC) 1 receptor (GR) or T cell antigen receptor (TCR) alone potently induce death of T cells (1)(2)(3)(4). However, simultaneous activation of both receptors fails to induce apoptosis by antagonizing the other signaling pathway (5)(6)(7)(8)(9)(10)(11)(12). There has been some evidence that GCs inhibit TCR-induced apoptosis by repressing FasL expression (13). In addition, by observing thymocyte development in mice with altered levels of GCs or GR, it has been shown that in interplay with TCR, GCs affect thymic development (14 -17). Reciprocally, proper stimulation through TCR/ CD3 antagonized GR activity and might contribute to GC resistance in T cells largely through the Ras activation of MEK/ ERK cascade (18). Activation of Ras alone, instead of TCR/CD3 engagement, was sufficient to inhibit GC-mediated apoptosis by activating Raf and Ral.GDS as mediators of the survival signals.
Clonal deletion of immature T cells, which are nonfunctional or self-reactive, is crucial for maintenance of the normal immune system. In particular, glucocorticoids (GCs) have been suggested to trigger programmed cell death of T cells and may help to eliminate developing thymocytes that fail to differentiate properly (17,19,20). They are either produced by thymic epithelium and possibly thymocytes themselves or are transferred from the adrenal gland in an endocrine manner (9). Due to their lipophilic properties, GCs can diffuse into the cytoplasm of target cells including thymocytes, bind to and activate their receptor, GR, which upon ligand binding translocates into the nucleus and associates with specific GC-response elements (GREs) to evoke various effects by modulating the expression of specific genes either positively or negatively depending on cell and gene context. GC sensitivity of thymocytes is developmentally regulated during maturation in the thymus. Immature CD4 ϩ CD8 ϩ double-positive (DP) thymocytes are known to be exquisitely susceptible to GC-induced apoptosis, whereas mature CD4 ϩ or CD8 ϩ single-positive (SP) T cells are relatively resistant (4,21). Thus, it is conceivable that the immature thymocytes to be positively selected should be protected from apoptotic actions of GCs. Thus far, many studies have addressed the cross-talk pathway for the inhibition of GC-induced apoptosis by TCR-or Notch-mediated signaling (6,22,23). Only a small population of DP thymocytes that acquire GC resistance by signaling through TCR/CD3 or NotchI appear to survive GC-triggered cell death and differentiate into mature SP thymocytes. Relatively little is known, however, regarding the specific nuclear target that couples these surface signals with GC resistance and thereby contributes to distinguishing thymocytes fated to die by GCs from those destined to survive.
SRG3 (for SWI3-related gene), a murine homolog of yeast SWI3 or human BAF155, was originally isolated as a gene expressed at higher levels in the thymus than in the periphery (24). It is presumed to play crucial roles as a core component of the SWI/SNF complex by remodeling chromatin structure (25,26). Recently, we found that the level of SRG3 expression is critical in determining GC sensitivity of developing thymocytes (24,27,28). The SRG3 protein was required for GC-induced * This work was supported in part by a grant from the Korea Science and Engineering Foundation through Protein Network Research Center (to R. H. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by BK21 Program from the Ministry of Education and Human Resources Development.
In the present study, we found that TCR/CD3 signaling inhibits GC-induced apoptosis of primary thymocytes and immature DP cells through SRG3 down-modulation. Signals from TCR/CD3 repressed the SRG3 transcription via Ras activation. As a result, the physical interaction between SRG3 and GR was inhibited, resulting in GC desensitization. However, thymocytes in transgenic mice overexpressing SRG3 were more sensitive to GC-triggered apoptosis compared with wild-type controls, and TCR signals did not suppress the cell death as potently as in wild-type mice. TCR repression of SRG3 expression was largely mediated through activation of the Ras/MEK/ ERK or PI3K pathway, which was previously suggested to be critical for TCR inhibition of GC-induced apoptosis. In addition, we found that TCR signals suppress the SRG3 transcription through the putative binding sites for E proteins and Ets family transcription factors in the SRG3 promoter. Collectively, these observations suggest a model for the role of SRG3 in the cross-talk pathway for inhibition of GC-mediated apoptosis by TCR signals.

EXPERIMENTAL PROCEDURES
Mice and Cells-Transgenic mice overexpressing SRG3 in FVB background were previously described (28). The S49.1 murine thymoma cell was purchased from the American Type Culture Collection (ATCC) and maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (FBS). The murine T cell hybridoma, KCIT1-8.5, was obtained from Dr. Y. Choi (University of Pennsylvania School of Medicine, Philadelphia) and maintained in RPMI 1640 containing 10% FBS supplemented with 50 M 2-mercaptoethanol (2-ME). The murine double-positive (DP) thymoma, 16610D9, was provided by Dr. C. Murre (University of California, San Diego, La Jolla, CA) and cultured in Opti-MEM (Invitrogen) containing 10% FBS supplemented with 50 M 2-ME. All media were supplemented with penicillin and streptomycin.
Cell Stimulation, Immunoprecipitation, Immunoblotting, and Flow Cytometry-Single cell suspensions of thymocytes (1-2 ϫ 10 6 ) from 4-to 5-week-old mice were suspended in 1 ml of RPMI 1640 medium containing 10% FBS and 50 M 2-ME supplemented with glutamine, penicillin, and streptomycin and cultured in medium alone or treated with PMA ϩ ionomycin for 3 h. After further incubation for 12 h in the absence or presence of Dex treatment, the cells were stained with FITC-conjugated annexin V and propidium iodide (PI) according to the manufacturer's protocol. PI-negative (live) cells were electronically gated and analyzed by flow cytometry (20,000 -30,000 events) with the CellQuest TM software using FACStar (BD Biosciences). Whole-cell ex-tracts from the cells unstimulated or stimulated with PMA ϩ ionomycin for 12 h were immunoprecipitated with anti-GR antibodies (M-20x; Santa Cruz Biotechnology) as described previously (29) and subjected to immunoblotting (28). For stimulating thymoma cells, 6-well plates were coated with appropriate antibodies at 10 g/ml at 4°C overnight followed by washing with phosphate-buffered saline. Five million cells at 10 6 cells/ml were added to each well and incubated at 37°C for 24 h. For stimulation of T cell lines by PMA ϩ ionomycin, cells were treated with 7.4 ng/ml PMA and 0.26 g/ml ionomycin or 0.2 ng/ml PMA and 0.25 g/ml ionomycin. The cells (10 6 cells) were then stained with FITCconjugated anti-CD69 antibody and propidium iodide (PI). After washing with phosphate-buffered saline, the surface expression of CD69 was analyzed by flow cytometry. The remaining cells were used for immunoblotting.
Northern Blot Analysis-Total RNA was prepared from cells by resuspension in TRIzol® reagent (Invitrogen) according to the manufacturer's instructions. Total RNA (7.5-10 g) was resolved on a 1.2% formaldehyde gel and then blotted to Hybond TM -N (Amersham Biosciences). Northern blots were probed with Id3 and SRG3 probes. The 405-bp Id3 probe was generated by random priming using the EcoRI fragment of Id3/S-003 vector, and the specific probe for SRG3 was generated as described previously (27).

RESULTS
Repression of SRG3 Expression upon TCR/CD3 Engagement Independently of TCR-induced Cell Death-We found previously that SRG3 acts as a coactivator for GR to up-regulate the transcriptional activity of this receptor and thereby render thymocytes sensitive to GC-mediated apoptosis (24,28). The relative sensitivity of developing thymocytes to the GCs was modulated depending on the expression level of SRG3 proteins. Intriguingly, the level of SRG3 mRNA in the positively selected thymocytes (CD3 high CD69 ϩ ) in the thymus was about three times lower than in the pre-selected thymocytes (CD3 low CD69 Ϫ ) (27). Because selection events in the thymus are active processes that involve various signals emanating from TCR/CD3 complex and other accessory molecules, and TCR/CD3 signaling has been shown to antagonize GC-induced apoptosis of T cells (19), it is possible that TCR-mediated signals render thymocytes resistant to GCs by modulating the SRG3 expression level. To assess whether SRG3 expression is directly modulated by antibody-mediated TCR/CD3 engagement, the murine DP thymoma cell line, 16610D9, was cultured in plates coated with anti-CD3⑀ or anti-TCR␤ antibody, alone or in combination with anti-CD4 antibody. The 16610D9 cell line is derived from a spontaneous thymoma in a p53-deficient mouse and exhibits characteristics typical of primary DP thymocytes (HSA high TCR med CD5 low CD44 low CD69 low ) (31). The expression of CD69, an early T cell activation marker (32), was specifically induced by antibody-mediated TCR/CD3 engagement, whereas it remained unchanged when treated with anti-hamster IgG control antibody (Fig. 1A). Subsequently, the level of SRG3 proteins in the whole-cell extracts from control and TCR-stimulated cells was examined by immunoblotting. As shown in Fig. 1C, the expression level of SRG3 protein decreased maximally by 55.3% upon T cell activation, whereas that of GR did not. The degree of SRG3 down-regulation correlated well with that of CD69 induction, suggesting that SRG3 repression may depend on the intensity of TCR/CD3 signals. When another thymoma cell line, S49.1, was activated with anti-CD3⑀, the expression of SRG3 also decreased maximally by 64% (Fig. 1D). These findings suggest that the expression of SRG3 in immature thymoma cells is down-regulated in response to signals through TCR/CD3 ligation. In addition to antibody-mediated TCR/CD3 cross-linking, cotreatment of a defined combination of PMA, a protein kinase C activator and ionomycin, a calcium ionophore also resulted in a decrease in SRG3 protein expression (Fig. 1D). Intracellular signaling triggered by the combinations of these drugs was shown to mimic the anti-apoptotic effect of proper cross-linking of the TCR/CD3 complex and induce thymocyte survival and differentiation (6,(33)(34)(35)(36)(37)(38). Because combination of PMA ϩ ionomycin may bypass the early consequences of TCR cross-linking, including membrane proximal signals in thymocytes, these results suggest that TCR-proximal signals are not required for TCR-mediated repression of SRG3 gene expression.
The viability of the activated cells was not significantly affected by antibody-mediated TCR/CD3 cross-linking, although cotreatment of PMA ϩ ionomycin led to a slight increase in cell death (data not shown). However, to exclude the possibility that the TCR/CD3-mediated SRG3 down-regulation resulted from nonspecific protein degradation in cells undergoing TCRinduced cell death, a T cell hybridoma, KCIT1-8.5, was used. The KCIT1-8.5 cell line is resistant to TCR-induced apoptosis due to the defect in the T cell death-associated gene 51 (TDAG51), which couples TCR stimulation with CD95 (Fas) expression but is still competent to be activated by TCR triggering (39,40). Stimulation of KCIT1-8.5 with anti-CD3⑀ or anti-TCR␤ antibody alone was sufficient for CD69 induction with no significant effect on the cell viability (Fig. 1B). As shown in Fig. 1E, activation of KCIT1-8.5 also resulted in a dramatic reduction in SRG3 proteins by 82%. These results indicate that SRG3 expression is reduced upon T cell activation independently of TCR-induced cell death.
TCR-mediated Repression of SRG3 Transcription via Ras Activation of MEK/ERK and PI3K Pathways-To determine whether TCR signals modulate the SRG3 expression at the transcriptional level, 16610D9 DP cells were stimulated with PMA ϩ ionomycin for varying amounts of time, and the mRNA levels of SRG3 and Id3 were quantified by Northern blot analysis. As reported previously (41), activation of these T cells rapidly induced Id3 transcripts ( Fig. 2A). Id3 induction became much stronger as the period of stimulation lasted longer than 6 h. However, the level of SRG3 transcripts began to be downregulated, albeit slowly, in the early phase but led to a dramatic decrease 6 h after T cell activation ( Fig. 2A). The suppression of SRG3 transcription upon T cell stimulation by either antibody-mediated TCR/CD3 cross-linking or cotreatment of PMA ϩ ionomycin was well reflected by a change in the SRG3 protein expression ( Fig. 2C and data not shown). Because protein expression of the SRG3 gene seemed to depend on the strength of TCR signals (Fig. 1, A and C), we further investigated how SRG3 transcription is regulated when much weaker activation signals are transduced. At lower concentrations of PMA ϩ ionomycin, Id3 transcripts were not significantly induced until 18 h (Fig. 2B). Concomitantly, SRG3 transcripts were not reduced until 18 h. Collectively, these results indicate that SRG3 transcription is repressed upon T cell activation.
Engagement of TCR/CD3 complex was shown to activate intricate intracellular signaling networks, including Ras/ MAPK pathway (42)(43)(44). Recently, it has been reported that TCR activation of Ras/MEK/ERK cascade plays a crucial role in inhibiting GC-induced apoptosis of T cells (18). Because the level of SRG3 was reduced upon TCR stimulation (Figs. 1 and 2) and correlated well with GC sensitivity of T cells (24,27,28), it is likely that the signaling molecules implicated in the cross-talk pathways for the TCR inhibition of GC-mediated apoptosis may also participate in down-regulating SRG3 expression. To address this possibility, we next examined if activated Ras affects the SRG3 expression. To this goal, we generated pSRG3-Luc, a reporter plasmid driven by 1,145 nucleotides upstream of the SRG3 translation start site (designated Ϫ1,145). Subsequently, we measured the pSRG3-Luc reporter activity in whole-cell extracts from thymoma cells cotransfected with varying amounts of K-RasV12. As shown in Fig. 3A, the SRG3 promoter activity in both 16610D9 and S49.1 cells was reduced by the K-RasV12 expression in a dose-dependent fashion. Similar results were obtained with H-RasV12, suggesting that Ras-mediated SRG3 down-regulation is not isoform-specific to K-Ras (Fig. 3B). Taken together, these results suggest that TCR/CD3 triggering activates Ras, leading to a decrease in both the promoter activity and the protein expression of the SRG3 gene.
To delineate the downstream effector(s) of Ras, we subsequently measured the effect of specific inhibitors of the Rasactivated pathways on the SRG3 promoter activity in 16610D9 DP cells that had been cotransfected with K-RasV12. Specific inhibition of MEK1/MEK2 or PI3K pathway by PD98059 or wortmannin, respectively, blocked the Ras activity to repress the SRG3 promoter activity, albeit less efficiently in the latter (Fig. 3C). In contrast, SB203580, a specific inhibitor of p38 kinase, did not exhibit significant suppressive activity against the Ras inhibition of the SRG3 promoter activity. To confirm these results, we subsequently measured the SRG3 promoter activity in the 16610D9 DP cells cotransfected with partial loss-of-function mutants of K-RasV12 which specifically activate only one of the downstream effectors of Ras (45). RasV12S35 or RasV12C40, which selectively activates Raf or PI3K, respectively, negatively regulated the SRG3 promoter activity (Fig. 3D). RasV12G37, which activates only Ral.GDS, however, could suppress the SRG3 promoter activity only slightly. Taken together, these observations indicate that activation of Raf/MEK/ERK and/or PI3K pathway may be crucial for TCR/Ras-mediated SRG3 down-regulation.
TCR/Ras Suppression of the SRG3 Transcription Conferred in the SRG3 Proximal Promoter Region-To delimit the specific promoter element(s) responsible for the TCR/Ras-mediated SRG3 repression, we generated various reporter constructs driven by serially truncated forms of the SRG3 promoter, and we assessed their responsiveness to TCR signals or activated Ras. In 16610D9 cells, the basal promoter activity was considerably diminished by ϳ70% after reducing the fragment from Ϫ1145 to Ϫ891 (D2 to D3), but further 5Ј truncation up to Ϫ210 did not significantly affect the promoter activity (Fig. 4A). The promoter activities of all 5Ј truncation constructs (D2 to D7) were suppressed upon PMA ϩ ionomycin treatment. Notably, the keen responsiveness to the PMA ϩ ionomycin was still apparent in the shortest 210-bp construct (D7), suggesting that major element(s) responsive to TCR signals may exist in this region. On the other hand, 3Ј truncation of the 289-bp segment in the context of D2 construct (D8) completely abolished the base-line activity, although that of 210 bp (D9) yielded a mild increase in luciferase activity compared with the 289-bp deletion, arguing that the proximal 289-bp region contains part of the SRG3 core promoter. Consistent with the data using 5Ј serial deletion mutants, 3Ј truncation of proximal 210 bp rendered the SRG3 promoter immune to TCR-triggered SRG3 suppression, corroborating the fact that major TCR-response element(s) may exist in this region (Fig. 4A). The promoter activities of all constructs examined were also repressed by K-RasV12 in a dose-dependent manner (Fig. 4B). In this case, the shortest D7 construct driven by the 210-bp proximal SRG3 promoter region was highly repressible upon Ras activation. Taken together, these results suggest that suppression of the SRG3 transcription by TCR or Ras signaling is largely conferred in the proximal 210-bp region of the SRG3 basal promoter.
The DNA sequence analyses of the 210-bp basal promoter revealed that this region includes potential binding sites for Sp, Ets, and E protein family transcription factors. To examine directly whether TCR signals affect the ability of these transcription factors to interact with the putative binding sites in the SRG3 promoter, nuclear extracts were prepared from 16610D9 cells that had been stimulated with PMA alone or PMA ϩ ionomycin for varying amounts of time and were examined by EMSA for any change in binding activity upon activation. The labeled oligonucleotides encompassing the potential binding sites for each transcription factor were used as probes, such as Sp-binding sites at Ϫ152 with consensus 5Ј-GGGGCGTGGC-3Ј (Sp152), Ϫ114 with consensus 5Ј-GGGCGGGGC-3Ј (Sp114), Ets-binding site at Ϫ130 with consensus 5Ј-GGAA-3Ј (EBS130), and E box element at Ϫ84 with consensus 5Ј-CATCTG-3Ј (E84). Oligonucleotides containing E box element in the E5 site (as a positive control) or Oct-1binding site (as a negative control) were also used. As reported previously (41), E2A/HEB binding to E5 probes was rapidly blocked upon T cell stimulation, whereas the amount of Oct-1 binding complexes remained unchanged (Fig. 5, E and F). Notably, PMA ϩ ionomycin also rapidly inhibited the assembly of DNA-binding complexes at the putative E box element (E84) in the SRG3 promoter whose migration rate was similar to that of the E2A/HEB complexes binding to the E5 site (Fig. 5D). In contrast, the complex formation at the potential Ets-binding site (EBS130) was significantly induced upon PMA ϩ ionomycin treatment, whereas the Sp binding was not significantly affected (Fig. 5, A-C). Treatment of PMA alone did not affect the complex formation at EBS130 element (Fig. 5B), suggesting that Ras/MAPK signaling alone does not affect DNA binding to the EBS130 element, but calcium-dependent signaling seems to play important role in modulating EBS130 occupancy. Collectively, these observations suggest that the putative E box element and Ets-binding site in the SRG3 proximal promoter may play a major role in TCR repression of the SRG3 transcription.
To confirm the regulatory role of the putative E box element (E84) and Ets-binding site (EBS130) in the SRG3 promoter, mutations were introduced in the E84 or EBS130 element in the context of the D7 construct driven by 210-bp proximal SRG3 promoter (Fig. 6A), and their effects on the basal pro-moter activity and responsiveness to the TCR or Ras signaling were assessed. As shown in Fig. 6B, mutation of either E84 or EBS130 element diminished the basal promoter activity by nearly half, and simultaneous mutation of both elements substantially repressed the promoter activity to one-fifth, suggesting that the intact E84 and EBS130 contribute the basal activity of the SRG3 promoter via distinct mechanisms. In addition, the promoter activities of all constructs examined were highly repressible upon PMA ϩ ionomycin treatment. Irrespective of the types of mutations introduced, in response to PMA ϩ ionomycin treatment, the SRG3 promoter activities were constitutively suppressed up to the level that is similar to the basal activity of the SRG3 promoter with mutations in both the E84 and EBS 130 elements (Fig. 6B). Consistent with the data obtained by gel-shift assay (Fig. 5), these results suggest that TCR-triggered signals down-regulate the SRG3 transcription by inactivating the putative binding site for E proteins and Ets-binding sites.
Activated Ras also down-regulated the SRG3 promoter activity, but less efficiently compared with the PMA ϩ ionomycin treatment, suggesting that TCR stimulation exhibits more potent effects than Ras activation in repressing the SRG3 transcription probably by activating other additional signaling FIG. 4. TCR/Ras responsiveness of the SRG3 promoter is mainly conferred in the proximal 210-bp promoter region. A, reporter constructs driven by serially truncated forms of the SRG3 promoter were transfected into 16610D9 cells. The transfected cells were treated with PMA ϩ ionomycin (PϩI) or ethanol (EtOH) for 6 h prior to measuring luciferase activity. Bars ϭ S.E. B, the proximal 210-bp SRG3 promoter region is highly repressible upon Ras activation. The Ras responsiveness of each SRG3 reporter construct was measured by measuring luciferase activities in whole-cell extracts from 16610D9 cells cotransfected with K-RasV12 as indicated. Bars ϭ S.E. pathways besides Ras activation. Still, mutation of either E84 or EBS130 element rendered the SRG3 promoter less responsive to Ras activation (Fig. 6B). Furthermore, mutation of both elements rendered the promoter completely immune to the Ras or TCR signals, corroborating that TCR/Ras signals repress the SRG3 expression through the putative E box element (E84) and Ets-binding site (EBS130) in the SRG3 promoter.

E2A/HEB Protein Complex Specifically Binds to the Putative E Box Element in the SRG3
Promoter-In thymocytes, E2A protein mostly binds to the E box element with the consensus CANNTG sequence as a complex with HeLa E box-binding (HEB) protein, another basic helix-loop-helix family transcription factor highly expressed in developing T cells (46). The protein complexes assembled on the E84 element in the SRG3 Site-directed mutagenesis was performed to alter the sequences of the EBS130 element from CCGGAAGA into CCGAGAGA or the E84 element from CATCTG into CTGCAG in the context of D7 reporter shown in Fig. 4A. B, each reporter construct designated was cotransfected with or without K-RasV12 into 16610D9 cells, and luciferase activity was determined 48 h after transfection. The cells transfected with each construct only were treated with PMA ϩ ionomycin (PϩI) for 6 h prior to measuring luciferase activity. S, Et, or E denotes the putative binding site for Sp, Ets, or E protein family transcription factors, respectively. Bars ϭ S.E. promoter exhibited similar mobility to the E2A/HEB complexes on the E5 site, and their binding activities were also rapidly inhibited with similar kinetics upon T cell activation (Fig. 5, D  and E). Thus it is likely that E2A/HEB binds to the E84 element and activates SRG3 transcription. In an attempt to clarify these ideas, we next planned to examine whether E2A and/or HEB protein specifically binds to the E84 element. To this end, nuclear extracts prepared from primary thymocytes in C57BL/6 mice or 16610D9 immature DP thymocytes were analyzed by EMSA using the labeled E84 oligonucleotides as a probe. Preincubation of the nuclear extracts with antibodies against E protein members such as E47, E12, or HEB inhibited the DNA binding of the complex containing each protein to the E84 probe, whereas antibodies against unrelated proteins such as K-Ras or actin did not (Fig. 7, A and B). Because E47 and E12 proteins are encoded by the E2A gene and produced by alternative splicing of the exon encoding the helix-loop-helix domain, this result indicates that E2A/HEB complex directly interacts with the E84 element in the SRG3 promoter.
To confirm further the specificity of the E2A/HEB binding to the E84 element, we conducted competition assay by using the oligonucleotides listed in Fig. 7C as competitor DNAs. Addition of excessive and unlabeled E84 or E5 oligonucleotides that contain E box elements kept the E2A and HEB proteins from binding to the labeled E84 probes (Fig. 7C). However, unla-beled Oct-1 oligonucleotides that do not contain an E box element could not compete with the E84 probes to bind to E47/ HEB complexes. Taken together, these results suggest that E47/HEB heterodimeric complex may directly interact with the putative E box element in the SRG3 promoter.
If E2A/HEB activates the expression of SRG3 by directly binding to the E84 element, there may exist significant correlation between occupancy of E2A/HEB at the E84 element and SRG3 expression level. In addition, because SRG3 was proved to exacerbate the susceptibility of T cells to the GC-triggered cell death (24,27,28), the expression level of SRG3 may correlate with GC sensitivity of various cell types. To address this possibility, we subsequently investigated cell type specificity of E2A/HEB binding to the E84 element. To this end, EMSA was performed with nuclear extracts from various murine cells such as primary thymocytes in C57BL/6 mice, 16610D9 thymoma cell line, mature lymph node lymphocytes in C57BL/6 mice, EL4 thymoma cell line, and NIH3T3 fibroblast cell line. The first two are representatives of GC-sensitive, immature T cells (SRG3 high ), and the next two are those of GC-resistant, mature T cells (SRG3 low ), whereas the last represents non-lymphoid cells (SRG3 low ). Intriguingly, the binding of the E2A/HEB complex to the E84 probes was mainly detected in only GC-sensitive, immature T cells but not in GC-resistant cells including mature T cells (Fig. 7D). These results suggest that the occu- FIG. 7. Specific binding of E2A/HEB protein complex to the putative E box element in the SRG3 promoter. Supershift assay of E2A/HEB complex was performed using nuclear extracts prepared from 16610D9 cells (A) or freshly isolated thymocytes derived from C57BL/6 mice (B). The nuclear extracts were pre-incubated with antibodies (Ab) as indicated on the top of each lane at room temperature for 30 min, and radiolabeled oligonucleotide probes encompassing the putative E box element in the SRG3 promoter (E84) were added. Anti-K-Ras and anti-actin antibodies served as nonspecific controls. E2A/HEB indicates the E2A/HEB-containing complexes, and NS means nonspecific binding complexes. C, competition assay. Molar excess (ϫ50 and 100) of the indicated cold probes were pre-incubated with nuclear extracts prepared from thymocytes in C57BL/6 mice prior to the addition of the labeled E84 probe. D, cell type-specific binding of the E2A/HEB complex to the E84 element. Equal amounts of nuclear extracts were prepared from freshly isolated thymocytes from C57BL/6 mice, 16610D9, lymph node T cells from C57BL/6 mice, EL4, and NIH3T3 cells, and complex formation at E84 element was analyzed by EMSA. pancy of E2A/HEB at the E84 element strongly correlates with the level of SRG3 expression and GC sensitivity of various cell types.

A Variety of Ets Family Transcription Factors Can Interact with the Putative Ets-binding Site in the SRG3 Promoter-To
determine whether members of the Ets domain transcription factor family can associate with the EBS130 element, we carried out EMSAs with nuclear extracts from thymocytes in C57BL/6 mice using the labeled EBS130 probe in the presence or absence of various unlabeled competitor oligonucleotides containing functional Ets-binding sites. The wild-type EBS130 oligonucleotide efficiently abolished binding of protein complexes to the EBS130 probe, whereas mutant EBS130 oligonucleotide was unable to inhibit binding of the complexes even at high concentrations (Fig. 8A, lanes 1-5). Commercial oligonucleotides that have been shown to contain Ets-related binding sites for Ets-1/PEA3 or PU.1/GABP␣, respectively, competed effectively with EBS130 probes (Fig. 8A, lanes 6 -9). However, commercial mutant Ets-1, Elk-1/SRF, or Oct-1 oligonucleotides are unable to compete with the EBS130 probe (Fig. 8A, lanes  10 -15). In addition, as shown in Fig. 8B, both unlabeled wildtype Ets-1 and EBS130 oligonucleotides could effectively block the Ets-1 binding to the commercial Ets-1 probes, whereas mutant Ets-1 and EBS130 competitors could not. To analyze in more detail the interaction of the EBS130 element with Ets family transcription factors, we examined whether functionally important Ets-binding sites in the regulatory regions of a variety of genes including Ets sites of several B or T lymphoidspecific genes can compete with EBS130 probe. Oligonucleotides encompassing the mb-1 promoter Ets site, which interacts with Ets-1, Ets-2, and PU.1 (Fig. 8C, lanes 6 and 7), the Fas promoter site, which interacts with GABP␣ and AP1 (Fig. 8C,  lanes 12 and 13), the IgH enhancer site, which interacts with ELF-1 (Fig. 8D, lanes 6 and 7), the polyoma virus enhancer, which interacts with ERG (Fig. 8D, lanes 10 and 11), and the E74 Ets site, which interacts with TEL-2 (Fig. 8D, lanes 12 and  13), still competed efficiently with labeled EBS130 probes. However, virtually no competition was observed with the other sites including the PEG-3 site, which binds PEA3 (Fig. 8C,  lanes 8 and 9), major histocompatibility complex class II promoter, which binds PU.1 (Fig. 8C, lanes 10 and 11), and GPIX Ets site, which interacts with Fli-1 (Fig. 8D, lanes 8 and 9). Results from these competition studies suggest that the EBS130 element can interact with various, but not all functionally relevant, Ets family transcription factor(s) whose binding preference is similar to Ets-1, GABP␣, ELF-1, ERG, and FIG. 8. Ets family transcription factors interact with the EBS130 element in the SRG3 promoter. EMSA was performed with nuclear extracts from thymocytes in C57BL/6 mice using labeled EBS130 (A, C, and D) or commercial Ets-1 (B) oligonucleotide as a probe. Nuclear extracts were pre-incubated in the molar excess (ϫ50 and 100) of the indicated cold probes prior to the addition of each labeled probe. Sequences present in the regulatory regions of various genes that have been shown to contain functionally relevant Ets-binding sites and commercial oligonucleotides containing wild-type or mutant Ets-binding sites as designated on top of each lane are detailed under "Experimental Procedures." TEL-2 but divergent from the PEA3, PU.1, and Fli-1.
Inhibition of Dexamethasone-induced Apoptosis by TCR Signaling through SRG3 Down-regulation-It was reported previously that SRG3 enhances GC sensitivity of developing thymocytes by physically interacting with GR and thus potentiating its transcriptional activity (24,28). The level of SRG3-GR complex formation is determined depending on the expression level of SRG3 and is critical in regulating GC sensitivity in T cells (27). Therefore, TCR-mediated repression of SRG3 expression should result in a decrease in SRG3-GR complex formation, leading to GC desensitization in T cells. To clarify these ideas, primary thymocytes from wild-type C57BL/6 mice were cultured in the presence or absence of PMA ϩ ionomycin. Twelve hours later, whole-cell extracts were prepared, and complexes containing GR were precipitated with anti-GR antibodies followed by immunoblotting with antibodies against SRG3, GR, and actin. As expected, PMA ϩ ionomycin treatment led to a decrease in the level of SRG3 proteins by half, whereas that of GR and actin remained unchanged (Fig. 9, input). As a result, complex formation between SRG3 and GR was inhibited (Fig.  9, IP:GR). However, there remained some SRG3 proteins in the supernatant upon precipitation of SRG3-GR immunocomplexes (data not shown), suggesting that all of the SRG3 proteins may not associate with GR.
To examine if repression of SRG3 expression in response to TCR signals can render thymocytes resistant to GC-induced apoptosis, primary thymocytes from either wild-type or transgenic mice overexpressing SRG3 under the control of human CD2 promoter (hCD2-SRG3 ϩ Tg mice) in FVB background were cultured in medium alone or pre-treated with PMA ϩ ionomycin. Twelve hours after further incubation in the presence or absence of Dex, a synthetic GC, we measured the portion of apoptotic cells by flow cytometry after staining with annexin V. As shown in Fig. 10A, in wild-type mice, Dex alone at 10 Ϫ7 or 10 Ϫ8 M rapidly induced apoptosis in about 30% of thymocytes, whereas the addition of PMA ϩ ionomycin efficiently blocked Dex-induced apoptosis. However, in hCD2-SRG3 ϩ Tg mice overexpressing SRG3, a higher proportion of thymocytes underwent Dex-induced apoptosis compared with wild-type thymocytes, and the addition of PMA ϩ ionomycin did not inhibit the cell death as potently as in wild-type mice (Fig. 10B). Compared with wild-type controls, overexpression of SRG3 rendered thymocytes more sensitive to Dex-mediated apoptosis even in the presence of PMA ϩ ionomycin (Fig. 10C). However, the protective effect of PMA ϩ ionomycin against Dex-induced apoptosis was still apparent for both wild-type and SRG3 transgenic mice. This may be partly because the expression of SRG3 proteins driven by endogenous SRG3 alleles may be modulated in the same way in both mice, and thus a decrease in endogenous SRG3 levels in thymocytes from SRG3 transgenic mice upon PMA ϩ ionomycin treatment can also contribute to preventing cell death induced by Dex treatment as in wild-type mice. This effect appears to be more severe because the level of SRG3 proteins overexpressed in the transgenic mice was only twice the level of normal thymocytes (28). Otherwise, it is possible that the level of SRG3 proteins can be regulated at the translational level, although less efficiently in down-regulating SRG3 protein expression than transcriptional repression. However, as shown in Fig. 10D, it was apparent that overexpression of SRG3 decreases the rate of protection upon PMA ϩ ionomycin treatment against Dex-mediated thymocyte apoptosis, especially when Dex was treated at low concentrations. When we assessed the portion of cells that survive Dex-mediated apoptosis by PMA ϩ ionomycin treatment, ϳ70 -80 or 80 -90% for 10 Ϫ7 or 10 Ϫ8 M Dex, respectively, was rescued in wild-type mice, whereas about 50 -60 or 40 -50% for 10 Ϫ7 or 10 Ϫ8 M Dex, respectively, was alive in the SRG3 transgenic mice (Fig. 10D). Taken together, these results suggest that signals emanating from TCR result in SRG3 down-regulation and the concomitant decrease in the SRG3-GR complex, rendering thymocytes resistant to GCs. DISCUSSION Immature DP thymocytes seem to acquire GC resistance as they mature into SP thymocytes in the thymus. Indeed, DP thymocytes are known to undergo apoptosis upon exposure to GCs, whereas mature SP T cells are much less susceptible to GCs (19). Although it has yet to be clearly demonstrated what determines GC sensitivity of developing thymocytes, surface signals through TCR/CD3, Notch1, and CD28 have been considered important in rendering GC resistance (6,22,23). In particular, it has been suggested that TCR/CD3 inhibition of GC-induced apoptosis is largely mediated by the Ras/MEK/ ERK pathway (18). However, the specific nuclear target that exerts inhibitory effects on GC-mediated cell death in response to TCR signals remains unknown.
We reported previously that the expression level of SRG3 correlated well with GC sensitivity of thymocytes both in vitro and in vivo (24,28). With a reduction in the SRG3 expression across the DP to SP transition, developing thymocytes became resistant to GC-induced apoptosis. The GC sensitivity of thymocytes can be altered by overexpression or repression of SRG3 proteins in transgenic mice. For example, thymocytes derived from mice overexpressing antisense SRG3 transcripts (lck-␣SRG3 ϩ Tg), which showed about 50% decline in SRG3 expression with intact GR expression, became GC-resistant (27). In contrast, SRG3 expression in the lymph node T cells from mice overexpressing SRG3 (hCD2-SRG3 ϩ Tg) increased by about 2-fold, and this conferred more sensitivity to GCs to these cells that would have been resistant to the hormone (28). These results suggest that SRG3 acts as a positive regulator of GC sensitivity in developing thymocytes during maturation in the thymus. Most interesting, we observed that SRG3 expression decreased immediately after positive selection (27), which suggests that some signals delivered during positive selection may rescue DP thymocytes from GC-induced apoptosis by downregulating SRG3 expression. In the present study, we found that signals from TCR/CD3 protect primary thymocytes and FIG. 9. TCR signaling represses the SRG3 expression and thereby results in a decrease in the SRG3-GR complex. Thymocytes derived from wild-type FVB mice were cultured in the presence or absence of 0.1 ng/ml (for 0.1PϩI) or 0.2 ng/ml (for 0.2PϩI) PMA together with 0.25 g/ml ionomycin as indicated on top of each panel. 12 h later, equal amounts of whole-cell extracts (Input) were prepared and immunoblotted (IB) with antibodies against SRG3, GR, or actin. The amounts of actin serve as a control. Whole-cell extracts prepared from thymocytes in wild-type FVB mice incubated with or without PMA ϩ ionomycin for 12 h were immunoprecipitated (IP) with anti-GR antibody (M-20x), and the immunoprecipitates (IP:GR) were immunoblotted with the indicated antibodies. Control lane represents whole-cell extracts prepared from freshly isolated thymocytes. some thymomas from GC-induced apoptosis by repressing SRG3 expression and concomitantly lowering the level of the SRG3-GR complex. Transient stimulation of thymocytes with a combination of PMA ϩ ionomycin, which triggers intracellular signaling similar to signals triggered by TCR cross-linking, significantly antagonized the GC-induced apoptosis of thymocytes from wild-type mice, although it did not potently inhibit the cell death in thymocytes from transgenic mice overexpressing SRG3 (Fig. 10). Overexpression of SRG3 in transgenic mice rendered thymocytes more sensitive to GCs compared with wild-type thymocytes (Fig. 10). Because the SRG3 expression might be driven by both ectopic SRG3 transgenes and endogenous SRG3 alleles in thymocytes from the SRG3 transgenic mice, which leads to an increase in the SRG3 proteins at approximately two times higher compared with wild-type mice (28), and repression of SRG3 gene expression by PMA ϩ ionomycin mainly occurred at the transcriptional level (Fig. 2), it is likely that PMA ϩ ionomycin might suppress the endogenous SRG3 gene transcription to down-regulate the overall amounts of SRG3 proteins even in the SRG3 transgenic mice and thereby contribute to decreasing the GC sensitivity (Fig. 10). However, because the total amount of SRG3 proteins in thy-mocytes from the SRG3 transgenic mice was still higher than that in wild-type thymocytes through transgenic SRG3 expression even in the presence of PMA ϩ ionomycin, the rate of protection against Dex-triggered cell death upon PMA ϩ ionomycin treatment was higher in wild-type mice than in the SRG3 transgenic mice (Fig. 10D). In addition, SRG3 expression in both thymocytes and some thymoma cells was suppressed upon activation by either antibody-mediated TCR/CD3 engagement or PMA ϩ ionomycin treatment (Figs. 1, 2, and 9). Because cotreatment of PMA ϩ ionomycin has been shown to mimic intracellular signaling pathways triggered by TCR/CD3 cross-linking bypassing TCR-proximal signaling, these results suggest that the membrane-proximal signaling pathways including tyrosine kinase activation is not required for downregulation of SRG3 expression in activated T cells. By using KCIT1-8.5 cell line, we also found that TCR-mediated SRG3 down-regulation is a specific response to TCR signaling (Fig. 1,  B and E).
SRG3 plays a unique role as a potentiator in GC-induced apoptosis by physically associating with GR and thereby enhancing its transcriptional activity. At present, it is generally accepted that GR-mediated transcriptional regulation is criti- 10. Activation of thymocytes with combinations of PMA and ionomycin results in inhibition of Dex-induced apoptosis by modulating SRG3 expression. Primary thymocytes derived from wild-type FVB mice (A) or transgenic (Tg) mice overexpressing SRG3 (hCD2-SRG3 ϩ Tg) (B) were unstimulated or pre-stimulated in culture with 0.1 or 0.2 ng/ml PMA ϩ 0.25 g/ml ionomycin (0.1PϩI or 0.2PϩI) for 3 h, and different concentrations of Dex (at 10 Ϫ7 or 10 Ϫ8 M) were added. After an additional 12 h of incubation, the cells were stained with annexin V-FITC and analyzed by flow cytometry. The percentage of cells undergoing apoptosis is indicated. C, the average of results (mean Ϯ S.E.) from three separate experiments from A and B. D, the plot depicts the relative apoptosis (%) when the rate of thymocyte apoptosis upon exposure to Dex alone in Fig. 1C is set to 100%. cal for the induction of GC-mediated apoptosis (19,47). The GR has been shown to modify gene transcription by either directly binding to the specific GRE as a homodimer or physically associating as an activated monomer with other transcription factors such as AP1, NF-B, CREB, and STATs (48). The GR can also regulate gene expression by recruiting chromatinmodifying complexes such as SWI/SNF, histone acetyltransferase, or histone deacetyltransferase (49). GRE-mediated transcription was inhibited by overexpressing antisense SRG3 transcripts or dominant negative SRG3 fragments inhibiting the SRG3-GR interaction (28), whereas it was enhanced in a dose-dependent manner by overexpressing SRG3 (data not shown). These results suggest that SRG3 may modulate the sensitivity to GC-induced apoptosis by directly regulating the transcriptional activity of GR. In addition, it has been also shown that GR physically interacts with the SWI/SNF complex, and trans-activation of GR requires Brg1/BAF chromatin remodeling complex (29,50). Whether SRG3 affects the transcriptional activity of GR independently or as a component of SWI/SNF complex remains to be elucidated.
Intriguingly, the signaling pathway implicated in the positive selection and TCR/CD3 inhibition of GC-induced apoptosis is very similar. Positive selection was impeded in mice overexpressing dominant negative component(s) of the Ras/MAPK signaling pathway such as Ras, Raf, MEK, or both Ras and MEK-1, yet negative selection remained intact (51)(52)(53)(54)(55)(56). Furthermore, three MAPK family kinases play qualitatively distinct roles in the selection of developing thymocytes (57)(58)(59). ERK played a critical role in the positive selection, whereas JNK or p38 kinase affected the negative selection (53). These results were corroborated by recent studies using transgenic mice expressing a mutant TCR ␣-chain-connecting peptide domain or mice with targeted disruption of the RasGRP, Grb2, or CD3⑀ gene (60 -64). In addition, mice deficient in Id3 or its upstream regulator, Egr-1, showed similar defects in positive selection (65,66). On the other hand, recent studies by Jamieson and Yamamoto (18) have shown that GC-induced apoptosis is also inhibited by TCR/CD3 activation of MEK/ERK via Ras. Most interesting, the same TCR/Ras/MEK/ERK pathway down-regulated SRG3 expression (Fig. 3). We have found that putative binding sites for E proteins and Ets family members located in the proximal region of the SRG3 promoter play critical roles in TCR-mediated suppression of SRG3 transcription (Figs. 4 and 5). The responsiveness of the SRG3 promoter to the TCR or Ras activation was largely conferred in the ϳ210-bp proximal SRG3 promoter region (Fig. 4). However, introduction of mutations in the E box element (E84) and Ets-binding sites (EBS130) within this SRG3 promoter region severely impaired the Ras or TCR activity to repress SRG3 promoter activity, suggesting crucial roles of these regulatory elements in TCR/Ras modulation of the SRG3 gene expression (Fig. 6B). EMSA analyses showed that E2A/HEB heterocomplex specifically associates with the E84 element and that a variety of Ets family members are able to interact with the EBS130 element (Figs. 7 and 8). Upon T cell activation, the complex formations at both elements were significantly changed (Fig. 5, B and D). E2A/HEB binding to the E84 element is rapidly blocked, which is consistent with previous reports (41) showing inhibition of E box DNA binding activity upon TCR stimulation via Id3 induction by the Ras/MAPK cascade. Indeed, E2A and HEB were shown to cooperate to activate SRG3 expression, which is inhibited by expression of dominant negative E protein inhibitor, Id3 protein (71). Furthermore, we observed that E2A/HEB occupancy at the E84 element exhibited a significant correlation with SRG3 expression level and GC susceptibility in various cell lines (Fig. 7D).
On the other hand, the binding of Ets family members to the EBS130 element was enhanced depending on the intracellular calcium levels in activated T cells. Because introduction of a mutation in the EBS130 element abolished the basal SRG3 promoter activity by half (Fig. 6B), we presume that the intact EBS130 element contributes to the basal SRG3 promoter activity in resting state by either interacting with Ets family members with high affinity, which act as an activator, or associating with Ets factors with low affinity, which exhibit repressor function. In activated T cells, however, the interaction of activator Ets factors with the EBS130 element may be inhibited, which facilitates the binding of repressor Ets factors. Although there are binding preferences for particular Ets family members to a specific site, a large overlap in binding specificities has been observed. Furthermore, previous reports (67)(68)(69) showed that both Ets-1 and Ets-2 are rapidly hyperphosphorylated upon activation of thymocytes, and phosphorylation of Ets-1 by calcium-dependent signaling negatively regulates DNA binding activity by stabilizing an inhibitory conformation that reinforces the autoinhibition of Ets-1. Indeed, we showed here by EMSAs that complex formation at the EBS130 element was efficiently blocked by functional Etsbinding sites in a variety of genes such as Ets-1/2 site in mb-1, GABP␣ site in Fas, ELF-1 site in IgH enhancer , ERG site in polyomavirus enhancer, and TEL-2 site in the E74 promoter (Fig. 8), suggesting that EBS130 has affinity for Ets family transcription factors with either activator or repressor function. Otherwise, because Ets family members are known to be regulated by a complex series of inter-or intramolecular interactions in addition to direct protein-DNA interactions (70), it is also possible that interaction between EBS130 complex and coregulatory binding partners such as Id proteins, mSin3A, NcoR, or CtBP may be enhanced in activated T cells, resulting in transcriptional repression (70). The mechanisms underlying regulation of SRG3 gene expression by Ets family members remain to be elucidated.
Involvement of the similar signaling pathway in both positive selection and protection from GC-induced apoptosis, together with the present data, provides useful insight into understanding the potential role of SRG3 during thymic positive selection. In response to the positively selecting signals that involve signals from proper cross-linking of the TCR/CD3 complex and other accessory molecules, DP thymocytes may downmodulate SRG3 and thus become resistant to GCs. In this respect, the fine regulation of SRG3 expression in response to TCR signaling may contribute to distinguishing DP thymocytes that are sensitive to GC-induced apoptosis from those resistant to GCs and further differentiate into mature SP cells.