Although the transcription factor Foxp3 is implicated in regulating glucocorticoid-induced TNF receptor (GITR) expression in the T regulatory cell lineage, little is known about how GITR is transcriptionally regulated in conventional T cells. In this study, we provide evidence that TCR-mediated GITR expression depends on the ligand affinity and the maturity of conventional T cells. A genetic dissection of GITR transcriptional control revealed that of the three transcription factors downstream of the classical NF-κB pathway (RelA, cRel, and NF-κB1), RelA is a critical positive regulator of GITR expression, although cRel and NF-κB1 also play a positive regulatory role. Consistent with this finding, inhibiting NF-κB using Bay11-7082 reduces GITR up-regulation. In contrast, NFAT acts as a negative regulator of GITR expression. This was evidenced by our findings that agents suppressing NFAT activity (e.g., cyclosporin A and FK506) enhanced TCR-mediated GITR expression, whereas agents enhancing NFAT activity (e.g., lithium chloride) suppressed TCR-mediated GITR up-regulation. Critically, the induction of GITR was found to confer protection to conventional T cells from TCR-mediated apoptosis. We propose therefore that two major transcriptional factors activated downstream of the TCR, namely, NF-κB and NFAT, act reciprocally to balance TCR-mediated GITR expression in conventional T cells, an outcome that appears to influence cell survival.
GITR has received attention over the past decade because of its involvement in regulating immune responses and its potential for altering the course of various immunopathological diseases (1). Mouse glucocorticoid-induced TNF receptor (GITR),5 a member of the TNFR superfamily, is induced on T cells following TCR stimulation or by glucocorticoid dexamethasone (2). Expression of GITR varies on T cells. Whereas low levels of GITR are constitutively expressed on freshly isolated T cells, these levels increase upon TCR ligation both in humans and mice; indeed, this increase is more pronounced than that induced by glucocorticoids (3, 4). Furthermore, a subset of CD4 T cells, CD4+CD25+Foxp3+ (regulatory T (TReg) cells) constitutively express high levels of GITR on the cell surface (5, 6). GITR is also expressed by B cells, macrophages, and dendritic cells (DCs) in vitro and in vivo following LPS treatment (1, 6), although the functional relevance of GITR expression by non-T cells is unresolved. Mouse GITR ligand (GITRL) is expressed on bone marrow-derived and freshly isolated splenic DCs, B cells, and macrophages (7, 8). GITRL is also up-regulated soon after LPS treatment of B cells, bone marrow-derived DCs, and macrophages in vitro (7). These findings suggest that GITRL on APC engage GITR on T cells to provide costimulatory signals. The high level of constitutive GITR expression by TReg cells implies a critical role for GITR in these cells, a notion supported by the finding that the addition of GITR Abs diminishes the suppressive activity of TReg cells (5, 6). However, recently, it was established that expression of GITR by responder cells, rather than TReg cells, is critical for the reversal of TReg cell-mediated suppression of proliferation (8). For non-TReg cells, soluble mouse GITRL and agonistic GITR-specific Abs promote Ag-specific proliferation (7, 9, 10). GITR signaling also activates NF-κB, which enhanced T cell survival (11, 12).
Currently, the molecular basis of GITR regulation on T cells is poorly understood. In a recent study, the transcription factor Foxp3 in TReg cells was shown to repress NFAT:AP-1-dependent transcription and to form a cooperative complex with NFAT (13). The net outcome of this interaction was repression of IL-2 expression but up-regulation of the expression of the TReg cell markers CTLA4, CD25, and GITR. Naive conventional T cells express very little Foxp3, but markedly up-regulate GITR upon TCR ligation. Therefore, alternate non-Foxp3 pathways must be utilized to control GITR transcription in these non-TReg cells. In this study, we provide several lines of evidence to demonstrate that for both conventional CD4+ and CD8+ T cells, NF-κB transcriptional factors act as positive regulators of GITR expression while NFAT functions as a negative regulator of GITR expression in these T cells.
Materials and Methods
Reagents and Abs
OVA, lithium chloride, cyclosporin A (CsA), ionomycin and PMA were obtained from Sigma-Aldrich. Compounds PD98059, SP600125, BAY11-7082, and LY294002 were purchased from Calbiochem. FK506 was obtained from Janssen-Cilag. MHC class II-binding OVA peptide OVA323–339 and class I Kb-binding OVA peptide SIINFEKL (OVA257–264) and its altered peptide ligands (14) were obtained from Mimotopes. Anti-CD3 Ab, 145-2C11, was produced by the mAb facility at our Institute.
C57BL/6 were used as wild-type control mice. TCR-transgenic (Tg) OT-I and DO11 mice were bred and maintained at The Walter and Eliza Hall Institute.
cRel−/− and NF-κB1−/− mice on a C57BL/6 background have been described previously (15, 16). Ly5.1 Rag−/− mice were engrafted with rela+/+ or rela−/− E14 liver hemopoietic progenitors derived from intercrossing rela+/− mice. Foxo3a−/− mice (from Dr. N. Motoyama, National Institute for Longevity Studies, Obv, Japan) were backcrossed to C57BL/6 for five generations and Foxo3a+/+ littermates were used as controls (17).
Generation of GITR Tg mice
To derive a GITR minigene, thymic cDNA from DO11 mice injected with OVA peptide was used to amplify exons 1–3 using specific primers (gitr forward, GGATCCACCATGGGGGCATGGG and exon 3 reverse, TGGTCCAAAGTCTGCAGTGACC). The genomic NOD/Lt DNA from exons 3 to 5 was also amplified using specific primers (exon 3 forward, GCCATGGGCACCTTCTCCG and gitr reverse, ACTAGTCTCGAGGCCTCATGGCCACCGACC). The GITR minigene was amplified with the two PCR products as templates using primers gitr forward and gitr reverse The minigene was cloned behind the human CD2 enhancer/promoter (18) and the polyadenylation signal was derived from the rabbit globin gene. Tg mice were generated by microinjection of fertilized eggs from C57BL/6 and NOD/Lt. Two Tg lines (CD2p-GITR) were obtained for each strain. Offspring were screened for overexpression of murine GITR on T lymphocytes using FITC-conjugated DTA-1.
Flow cytometric analysis of lymphocytes and cell sorting
Ab used in this study to stain cell surface markers included PE-Cy7- or allophycocyanin-conjugated anti-mouse CD4 (clone RM4-5; BD Pharmingen), allophycocyanin-Cy7-conjugated anti-mouse CD8 (clone 53-6.7; BD Pharmingen), FITC- or PE-conjugated rat anti-mouse CD69 Ab (BD Pharmingen), and biotinylated (R&D Systems) or FITC-conjugated (in house) anti-mouse GITR (DTA-1). For surface staining, cells were incubated with optimally diluted Ab for 30 min on ice. Viable cells (determined by propidium iodide exclusion) were analyzed using a FACSAria (BD Biosciences) in conjunction with CellQuest software (BD Biosciences). In some experiments, cell subsets were also sorted on FACSAria.
In vivo peptide injection
For detecting GITR mRNA, OVA Tg DO11 mice were either untreated or injected with 0.5 mg of peptide i.p. After 3 h, thymocytes were prepared and sorted for CD4+CD8+ double-positive (DP) and CD4+ single-positive (SP) cells.
Lymph node cells or purified CD8+ T cells in 10% FCS-RPMI 1640 were cultured in a 1-ml volume in 48-well tissue culture plates. Cells were cultured in the absence or presence of anti-CD3 Ab, ionomycin, PMA, or ionomycin plus PMA. For OVA-specific TCR Tg CD8 T cells, cells were stimulated with peptide SIINFEKL and its altered peptide ligands (14, 19). Cells were cultured for 24–48 h for the detection of GITR proteins on the cell surface and for 3 h for the detection of GITR mRNA.
For detection of mRNA, RNA was extracted from cell pellets with Tri Reagent (Ambion) and was reverse-transcribed with Moloney murine leukemia virus-reverse transcriptase in a 50-μl reaction at 37°C. PCR was performed using 1–2 μl of reverse transcriptase reaction mixture in a standard 50-μl reaction. PCR conditions for all primer sets were 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s. A series of cycle numbers (15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35) were tested for each molecule so that the product being amplified was in the early exponential phase. Primers used included GITR forward (fwd), 5′-TGATCACCATGGGGGCATGGGCCATGCTG and reverse (rev) 5′-ACTAGTCTCGAGGCCTCATGGCCACCGACC; Actin fwd, 5′-GTGGGCCGCCCTAGGCACCA and rev, 5′-CTCTTTGATGTCACGCACGATTC; OX40 fwd, CGAATTCCACCATGTATGTGTGGGTTCAG and rev CGGGATCCTCAGGAGCCACCAAGGTGGG.
Detection of endogenous GITR mRNA
Lymph node cells from OT-1 and GITR Tg OT-1 mice were cultured with or without cognate peptide in the absence or presence of DTA-1. In some experiments, CD25−CD4+ lymph node cells from C57BL/6 and C57BL/6 GITR Tg mice were stimulated with anti-CD3 (0.1–10 μg/ml). Cells were cultured at 0.5–2 × 106/ml in 48-well plates between 4 and 16 h. For detection of endogenous GITR mRNA, total RNA was extracted from cultured cells. RT-PCR was performed with primers detecting the endogenous 3′ untranslated region of GITR which differs from the Tg GITR: forward, CCAAGCCAGACGCTACAAGAC and reverse, GATCTTGTCTAAACGTGGTGC. In these experiments, we used CD3 as control sequences (forward, ACACTTTCTGGGGCATCCTG and reverse, TGATGATTATGGCTACTGCTG) given the recent report that CD3 mRNA is not up-regulated T cells after activation, whereas many household genes like actin are up-regulated (20). The amplified products were subjected to density analysis using Adobe Photoshop (Bio-Rad) and standardized to CD3. Three independent experiments were performed. For CD69 expression, parallel cultures were harvested. Cells were then stained for CD4, CD8, and CD69. Gated CD8 T cells were shown for their CD69.
Nuclear extracts were prepared as previously described (21). Protein concentrations were determined using a Bio-Rad protein assay kit. The protocol used for gel shift assays was based on that described previously (22). Briefly, 1.5 μg of nuclear extract was incubated with 50 nM 32P-labeled double-stranded oligonucleotide probe in a buffer containing 10 mM HEPES (pH 7.8), 50 mM sodium glutamate, 5 mM MgCl2, 1 mM EDTA, 1 mM DTT, 5% glycerol, and 1 μg of poly(dI:dC) on ice for 20 min. Two micrograms of Ab was then added and the reactions were left for another 20 min on ice (total volume, 20 μl). The reactions were then loaded onto a 5% nondenaturing polyacrylamide gel in 1× Tris-borate-EDTA and subjected to electrophoresis at 250 V for 2 h. The gels were dried onto a double layer of 3MM paper and exposed to film. The κB3 probe (5′-CGTAAGCAGCGGGAAATCCCCC-3′), previously described (21), was used in all reactions. The Abs used were specific to NF-κB1 p50 (sc-1192X), RelA (sc-372X), and cRel (sc-71X) and were all obtained from Santa Cruz Biotechnology.
Culture supernatants were harvested 24 h after culture and assayed for IL-2 by ELISA. Recombinant cytokines as standard, coating Ab, and biotinylated Ab were obtained from BD Biosciences.
T cell functional avidity determines GITR up-regulation on T cells
It is well accepted that the TCR engagement induces GITR up-regulation by both conventional CD4 and CD8 T cells (2, 4). In this study, we revealed using TCR Tg mice and antigenic peptide ligands, further complexities of TCR-dependent GITR induction. First, the canonical peptide SIINFEKL and the strong agonist SEINFEKL promptly stimulate OVA-specific CD8 T cells from OVA TCR Tg mice (OT-I), leading to GITR up-regulation within 24 h. However, stimulation with high-dose SIINFEKL (>2000 ng/ml) failed to up-regulate GITR on CD8 T cells, but still up-regulated CD69 (Fig. 1,a). Second, at the other end of the spectrum, the partial agonist/antagonist peptides SIINFEDL or SIINFEGL at doses between 100 and 50,000 ng/ml were also very weak at up-regulating GITR, despite CD69 being up-regulated (Fig. 1,b). Third, more mature SP thymocytes up-regulated GITR upon in vivo Ag stimulation while immature DP did not, even though the DP cells responded to the same Ag as determined by CD69 up-regulation (data not shown). To confirm this and to exclude the possibility that the failure to detect GITR induction was the result of immature DP cells dying, we examined GITR transcription. Three hours after peptide injection (a time frame that preceded any potential cell death), SP thymocytes but not immature DP thymocytes displayed elevated expression of GITR mRNA (Fig. 1 c).
These results show that TCR-mediated GITR up-regulation was affected by the Ag dose, peptide efficacy/avidity, and the developmental state of the T cell. Moreover, there were several circumstances where T cells were activated but there was little or no GITR up-regulation, indicating that there appear to be negative regulatory mechanisms for GITR expression during T cell activation.
The NF-κB transcription factor RelA is selectively activated in T cells by agonist but not the partial agonist/antagonist peptides
To better understand the molecular basis of the TCR-mediated regulation of GITR on conventional T cells, we chose to examine the requirements of certain signaling pathways engaged downstream of the TCR in control of GITR expression. We focused on the NF-κB transcription factors that critically control the molecular programs elicited by TCR activation. To this end, we evaluate NF-κB activation by T cells following TCR stimulation. OT-1 T cells were stimulated with SIINFEKL or partial agonist SIINFEGL. EMSA was performed with nuclear extracts from T cells that were stimulated under various conditions. Both a low dose (0.25 μg/ml) and high dose (25 μg/ml) of SIINFEKL induced the nuclear translocation of RelA while a high dose of SIINFEGL (25 μg/ml) did not induce RelA nuclear translocation (Fig. 2). Stimulation with SIINFEKL also induced high amounts of NF-κB-1 although unstimulated and SIINFEGL-stimulated T cells had also some NF-κB-1 in nuclear extracts. cRel was not detected in all preparations (data not shown).
The NF-κB transcription factor RelA is nondispensable for GITR up-regulation on TCR-activated T cells
Next, we examined the role of NF-κB in GITR regulation using T cells from mice that lacked either RelA, cRel, or NF-κB1. Naive T cells displayed increased surface GITR when they were stimulated with 10–1000 ng/ml anti-CD3 Ab alone. With a suboptimal dose of anti-CD3, costimulation with anti-CD28 Ab further augmented GITR expression by T cells (Fig. 3). Among the different NF-κB-deficient T cells examined, RelA−/− T lymphocytes displayed the most marked deficit in GITR up-regulation. When a low concentration of anti-CD3 (10 ng/ml) was used, T cells virtually had no GITR up-regulation with or without anti-CD28 Ab (Fig. 3,a). When a high concentration of anti-CD3 Ab (1000 ng/ml) was used, GITR was induced on RelA−/− T cells but the level was always lower compared with that of RelA+/+ T cells (Fig. 3 a). The deficit in GITR induction was similar for both CD4 and CD8 T cells.
Relative to RelA−/− T cells, the deficit in GITR up-regulation by T cells from cRel−/− mice was smaller (Fig. 3 b). Again, the deficit was clearer when cells were stimulated with a low concentration of anti-CD3; the deficit exists for both CD4 and CD8 T cells. Unlike RelA−/− T cells, cRel−/− T cells had no defect in GITR up-regulation when anti-CD28 was added along with anti-CD3 Ab (data not shown).
Although NF-κB1 was readily detected from nuclear extracts of T cells (Fig. 2), T cells from nf-κB1−/− mice has minimal defects in GITR up-regulation (data not shown). It indicates that NF-κB1 has a minor role in GITR up-regulation.
A partial block in GITR up-regulation on T cells from mice deficient in single subunits of NF-κB (Fig. 3, a and b) raised the possibility that functional overlap among these transcription factors may be masking a great overall role for the NF-κB pathway in TCR-induced GITR expression. Therefore, we used an inhibitor of NF-κB activation, aiming to achieve global inhibition of this pathway. The compound Bay11-7082 inhibits IκB kinase phosphorylation selectively and irreversibly (23). In the presence of 2–5 mmol of Bay11-7082, GITR expression on T cells activated with 100-1000 ng/ml anti-CD3 Ab was reduced to similar levels seen on unstimulated T cells (Fig. 3 c). Suppression of anti-CD3-mediated GITR up-regulation by Bay11-7082 was also observed at the level of mRNA (data not shown).
Given that GITR ligation could activate NF-κB (11), we tested here whether ligation of GITR could further up-regulate GITR. We found that agonist anti-GITR Ab with or without TCR stimulation on T cells from GITR Tg mice did not enhance significantly endogenous GITR over the levels stimulated with TCR stimulation alone (Fig. 3,d). Nevertheless, the ligation of GITR by agonist anti-GITR Ab did increase CD69 up-regulation of naive T cells from GITR Tg mice but not wild-type mice when TCR was stimulated with either cognate Ag (Fig. 3 d) or polyclonal anti-CD3 (data not shown), indicating that GITR does behave as a costimulatory molecule for T cells in our system.
The transcription factor NFAT negatively regulates GITR expression
Although NF-κB acts predominantly as a positive regulator for GITR regulation, our finding that strong TCR signals inhibit GITR induction prompted an examination of what transcription factors are required to suppress the GITR gene expression. We reasoned that transcription factor NFAT might be a candidate for a negative regulator for GITR based on the following observations:
When we used PMA and ionomycin for T cell stimulation, we found that ionomycin, which is a strong activator of NFAT (24), antagonized the ability of PMA alone to up-regulate GITR. Purified CD8 T cells, when stimulated with PMA alone, effectively up-regulated GITR as well as CD69. Compared with PMA, ionomycin alone weakly stimulated GITR expression, even though at doses of 500-1500 ng/ml ionomycin, CD69 was up-regulated. Furthermore, at these higher doses, ionomycin clearly suppressed PMA-stimulated GITR up-regulation, while at lower doses it synergized with PMA to up-regulate CD69 (Fig. 4,a). Notably, high doses of ionomycin alone induced apoptosis of T cells. To rule out that low GITR expression by ionomycin-stimulated T cells was purely attributed to cell death, purified CD8 T cells were stimulated with PMA, ionomycin, or both agents for 3 h (a time point preceding the onset of substantial T cell death). GITR up-regulation, reflected by mRNA expression, was observed in cells stimulated by PMA, but not in cells stimulated with ionomycin or ionomycin plus PMA (Fig. 4 b).
Next, we used the calcineurin inhibitors CsA and FK 506 in our culture system to test their effect on GITR regulation. CsA and FK506 inhibit the calcium-dependent serine/threonine phosphatase calcineurin and its substrate, the transcription factor NFAT, by forming complexes with immunophilins (25). To test the impact of NFAT suppression on GITR regulation, lymph node T cells were stimulated with ionomycin plus PMA for 24 h and then GITR expression on T cells (both CD4 and CD8) was analyzed. Although ionomycin inhibited PMA-induced GITR up-regulation, the addition of CsA (200–2000 ng/ml) or FK506 (20–200 ng/ml) counteracted this inhibition (Fig. 5,a). As previously reported, CsA and FK506 suppressed the induction of CD69 (Fig. 5 a), albeit modestly, and abolished IL-2 production (data not shown) by T cells stimulated with both ionomycin and PMA. Thus, these calcineurin inhibitors have differential effects on activation markers (up-regulate GITR, down-regulate CD69).
As well as the increased cell surface expression of GITR, mRNA levels for GITR increased in cells stimulated with ionomycin plus PMA in the presence of FK506 (data not shown). Neither CsA nor FK506 had an effect on GITR up-regulation by T cells stimulated with PMA alone. We also observed a similar pattern with OX40, a molecule that is structurally related to GITR (Fig. 5 c). Ionomycin suppressed PMA-induced OX40 mRNA production. Addition of FK506 reversed the suppression by ionomycin.
Similar to our findings for ionomycin/PMA-induced activation, addition of FK506 was found to reverse the suppression of GITR expression by high-dose peptide (Fig. 5 b).
Lithium chloride, an inhibitor of the NFAT export kinase GSK3, was also found to suppress TCR-mediated GITR up-regulation. Lithium chloride has been used to increase NFAT activity by inactivating the NFAT export kinase GSK3, resulting in prolonged dephosphorylation of NFATc and sustained nuclear localization (26, 27). Addition of lithium chloride (at doses between 10 and 40 mM) but not potassium chloride suppressed anti-CD3 Ab-induced GITR up-regulation on CD4 and CD8 T cells (Fig. 6). Lithium chloride in the absence of anti-CD3 stimulated T cells to express CD69 stimulation, albeit to a lesser extent than for TCR stimulation.
Non-NF-κB transcription factors are involved in GITR regulation
Although experiments with NF-κB-deficient mice established a critical role for NF-κB (particularly RelA) as a positive regulator of GITR expression, lower levels of GITR can still be induced in the absence of NF-κB. This suggested that other transcription factors may also participate in GITR regulation. We tested two candidates known to be involved in T cell activation, namely, Foxo3a and AP-1. T cells isolated from Foxo3a−/− mice consistently showed higher expression of GITR compared with Foxo3a+/+ littermates when stimulated with anti-CD3 Ab (Fig. 7,a). Even in the unstimulated state, Foxo3a−/− T cells displayed slightly higher levels of GITR expression. Given that Foxo3a is a downstream target of TCR signaling that is negatively regulated by PI3K/Akt, we tested the consequence of PI3K inhibition. Inclusion of the PI3K inhibitor LY294002 (6–12 μM) indeed suppressed anti-CD3-induced GITR up-regulation (Fig. 7,b). However, the potency of PI3K inhibition on GITR up-regulation could be largely due to its inhibition of rapamycin-sensitive mTOR, also a downstream target of PI3K/Akt, since the addition of rapamycin drastically reduced the anti-CD3-induced GITR up-regulation (data not shown). As for transcription factor AP-1, we found that the JNK inhibitor SP600125 (28) partially suppressed anti-CD3-induced GITR up-regulation (Fig. 7 c).
GITR expression endows T cells with resistance to TCR-mediated apoptosis
Although the primary aim of this study was to understand the regulation of TCR-dependent GITR expression in conventional T cells, here we also present evidence that up- and down-regulation of GITR influences T cell survival. The data outlined in Fig 1,b showed that OT-I CD8 T cells up-regulated CD69 but not GITR when stimulated with partial agonistic altered peptide ligands. We consistently observed that 24 h following the initiation of peptide-induced activation there were less viable cells in these cultures compared with unstimulated or agonistic ligand-stimulated cultures (Fig. 8). Within the duration of this study (24 h), there was no appreciable cell proliferation. To assign a specific role to GITR in protecting T cells from TCR-mediated apoptosis, we generated OT-I CD8 T cells that constitutively express GITR under the control of human CD2 promoter. When stimulated with partial agonistic altered peptide ligands such that GITR was not induced by TCR stimulation, CD8 T cells overexpressing Tg GITR survived better than wild-type OT-I cells (Fig. 8). This finding indicates that TCR-induced GITR expression is involved in activation-induced T cell survival.
GITR has recently attracted considerable interest as a potential target of immune intervention (1, 29). This can be potentially achieved by either targeting GITR signaling or the induction of GITR. For the latter purpose, we undertook to investigate GITR regulation. It is well established that TReg cells constitutively express high levels of GITR (4, 5, 6). In contrast, naive conventional T cells for mice and humans express low levels of GITR, which can be markedly up-regulated upon TCR stimulation (2, 3, 4, 5, 6). In contrast, glucocorticoids are modest at inducing mouse GITR up-regulation (4). We now show that TCR ligation up-regulates GITR expression only modestly on immature T cells. Furthermore, we found that even for mature conventional T cells, only a certain strength of TCR stimulation (neither weak nor very strong) was capable of up-regulating GITR.
To understand how these findings coincide with transcriptional regulators controlling GITR expression, we demonstrate that members of the NF-κB family of transcription factors act as a positive regulator of TCR-mediated GITR induction. Using T cells from various NF-κB mutant mice, RelA was found to have the most profound impact on GITR expression. Under suboptimal stimulation conditions (10 ng/ml anti-CD3 Ab) that promote wild-type T cells to up-regulate GITR, rela−/− T cells failed to up-regulate GITR. However, in response to a higher concentration of anti-CD3 Ab, which delivers stronger TCR signals, rela−/− T cells did increase GITR expression, albeit never reaching the levels of expression seen on wild-type T cells under the same stimulation condition. This functional importance of RelA in TCR-induced GITR induction agrees with the promoter binding data showing that RelA but not NF-κB2p52 binds to a site within the GITR promoter (30). Currently, it remains unclear which NF-κB family member(s) RelA partners in controlling GITR transcription. In this study, we also showed that the loss of cRel has some influence on GITR up-regulation on T cells, although we did not detect significant nuclear translocation of cRel. Nevertheless, cRel has been reported to be induced by TCR stimulation (31). Despite that, NF-κB1 p50 can be readily detected from nuclear extract of stimulated T cells, the loss of NF-κB1 p50 had a minimal effect on GITR regulation. Although it is unknown whether the loss of all three NF-κB subunits downstream of the classical pathway would totally abolish GITR up-regulation, we observed that when Bay11-7082 was used to suppress NF-κB, GITR up-regulation was indeed minimal. This suggests a model in which multiple NF-κB transcription factors are essential for TCR-induced GITR expression.
Since GITR ligation can activate NF-κB (11), it raised the question whether GITR ligation can form a positive regulatory loop to further induce GITR expression. In the absence or presence of TCR stimulation, agonist anti-GITR Ab did not induce appreciable endogenous GITR mRNA on T cells from GITR Tg mice despite that ligation of GITR does provide costimulatory effect to T cells, as evidenced by CD69 up-regulation. It may be related to the subunit of NF-κB family that was activated by GITR ligation since we showed here that only RelA drastically influence GITR levels. In addition, GITR can also interact with adaptor proteins of TNF signal transduction to negatively regulate NF-κB activation (12). Thus, the net outcome of GITR signaling on NF-κB activation could be difficult to predict. Furthermore, given that transcription factor NFAT can be a negative regulator of GITR expression, it is also possible that GITR signaling can also activate other transcriptional factors that might negatively regulate GITR.
Apart from the demonstration that NF-κB and AP-1 act as positive regulators in GITR induction in T cells, in this study, we have provided several lines of evidence that by contrast NFAT appears to act as a transcriptional repressor of GITR expression. First, ionomycin, an agent that activates NFAT, suppresses PMA-induced GITR up-regulation, both at the protein and mRNA levels. Second, the NFAT inhibitors CsA and FK506 can reverse the ionomycin-induced suppression of GITR expression that is induced by PMA and high doses of antigenic peptide. Third, persistent NFAT activation in T cells, delivered through inhibition of NFAT export kinase GSK3 by lithium chloride, suppresses anti-CD3 Ab-induced GITR up-regulation on T cells.
How NFAT negatively regulates GITR is still unclear. In TReg cells, NFAT can form a complex with Foxp3 to repress IL-2 expression, but up-regulate CTLA4, CD25, and GITR (13). Nevertheless, how are Foxp3-NFAT complexes, a negative regulator for certain genes, able to up-regulate GITR? Does NFAT/Foxp3 act as a trans-activator for GITR transcription directly or does NFAT/Foxp3 permit other trans-activators to initiate gene transcription? Furthermore, in non-TReg cells whose Foxp3 expression is very low (32), how does NFAT negatively regulate GITR, presumably by a non-Foxp3 pathway? Several studies have shown that NFAT1 can compete with NF-κB, in particular RelA, for NF-κB binding sites and this competition leads to reduced transcription (33, 34). Alternatively, NFAT1 may also compete with NF-κB for transcriptional coactivators such as p300 (35, 36).
The TNFR superfamily encompasses many members. So, does NFAT regulate the expression of all TNFR molecules in the same fashion as GITR? We showed here that a closely related TNFR family member, OX40, seems to behave like GITR. First, negative regulation is induced by ionomycin signaling and, second, this is reversed by FK506. Correspondingly, the induction of OX40 in T cells following TCR stimulation was impaired in mice lacking cRel or NF-κB1 (data not shown). On the other hand, TNF family members that are expressed by T cells such as CD40L and FasL are clearly induced by NFAT activation (37, 38, 39, 40).
Initial studies with cell lines transfected with GITR had implicated it could protect T cells from TCR-mediated apoptosis (2). A subsequent study using primary T cells also suggested that GITR played a role in promoting cell survival, although decreased apoptosis was not demonstrated directly (11). In this study, we provided two lines of evidence to further support that GITR is a prosurvival factor for T cells. First, fine-tuning of TCR signal strength using altered peptide ligand under conditions whereby GITR was not induced led to enhanced cell death when compared with unstimulated T cells. Second, primary T cells with Tg expression of GITR survive better than nontransgenic T cells following TCR stimulation, particularly under conditions in which endogenous GITR was not induced by TCR engagement. It is well established that altered peptide ligands can lead to different outcomes for T cells, including apoptosis (41). In this study, we established that failing to induce GITR after stimulation with partial agonistic peptide may be one of the reason for T cells undergoing apoptosis. This conclusion is supported by our finding that enforced expression of GITR can rescue T cells from apoptosis. Since immature T cells do not up-regulate GITR after the engagement with agonist antigenic peptide, it also raises the intriguing possibility that the inability to induce GITR is one of the mechanisms that contributes to negative selection. Currently, the significance of abrogated GITR up-regulation by T cells stimulated with strong agonistic ligands is not clear (although one could opine that this is a potential mechanism for peripheral deletion) nor is FK506 reversal of GITR suppression under such conditions. Further investigation into GITR regulation and function may unravel such issues and may even avail ways of promoting immune intervention.
We thank Nicole Ashman for animal care and technical assistance.
The authors have no financial conflict of interest.
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.
This work was supported by National Health and Medical Research Council of Australia and Juvenile Diabetes Research Foundation (to A.M.L.). Y.Z. is supported by a RD Wright Fellowship from the National Health and Medical Research Council of Australia.
Abbreviations used in this paper: GITR, glucocorticoid-induced TNFR; Treg, regulatory T cell; DC, dendritic cell; GITRL, GITR ligand; DP, double positive; SP, single positive; fwd, forward; rev, reverse; Tg, transgenic.