Enhanced GITR/GITRL interactions augment IL-27 expression and induce IL-10-producing Tr-1 like cells


Correspondence Dr. Yijun Carrier, CTI-Boston, Pfizer, Center for Life Sciences Building 3 Blackfan Circle, 18th Floor Boston, MA 02118

Fax: +1-617-274-3248

e-mail: yijun.carrier@pfizer.com


The glucocorticoid-induced TNFR-related (GITR) protein is a coactivating receptor that is constitutively expressed on Treg cells and induced on activated T cells. To better under-stand the role of long-term GITR signaling, we generated a mouse that constitutively expresses GITR ligand (GITRL) on APCs that mimics the physiological distribution of GITRL in vivo. Despite a five-fold expansion of the Treg-cell pool, there is increased activation and depletion of naive T cells in the transgenic (Tg) mice, suggesting that the increased number of Treg cells cannot fully suppress T-cell activation. Interestingly, GITRL Tg mice have multiorgan lymphocytic infiltrates yet display no overt autoimmunity, indicating the existence of a compensatory immunoregulatory mechanism(s). In the spleens and tissue infiltrates ofGITRL Tg mice, we found increased numbers of Foxp3 IL-10-producing type 1 regulatory T (Tr-1)-like cells that suppress naïve T-cell proliferation in an IL-10-dependent fashion. Increased IL-27 production from Tg APCs and activation of c-Maf in the Tr1-like cells suggest a possible mechanism for their induction. Our results demonstrate that enhanced GITR/GITRL interactions have a pleiotropic role on the regulation of T-cell responses, which includes promoting the differentiation of Tr-1-like cells, which contribute to the maintenance of peripheral T-cell tolerance.


T cells mediate a variety of immune functions that are regulated by a complex network of receptor/ligand pairs and are primarily dependent on costimulatory signaling through the TCR [[1]]. One such receptor/ligand pair that controls T-cell homeostasis is the glucocorticoid-induced TNFR-related (GITR) receptor and its ligand, GITRL [[2]]. In T cells, GITR is constitutively expressed at a high basal level on CD25+Foxp3+ Treg cells and is induced rapidly, yet transiently, on activated CD4+ and CD8+ conventional effector T (Teff) cells. Its ligation by GITRL exerts a costimulatory signal that triggers TCR-dependent proliferation and the expansion of Treg and Teff cells [[3]].

GITRL is constitutively present on resting APCs, and its expression is downregulated by triggering the BCR, CD40, or several TLRs [[4, 5]]. Previous studies have shown that immune signals induced by GITR/GITRL interactions modulate both innate and adaptive immune responses [[5, 6]], however the functional outcome of such interaction is largely dependent on each specific model tested and is often controversial. Reduced inflammatory responses to acute and chronic lung inflammation were observed in GITR−/− mice compared with GITR+/+ mice, indicating a proinflammatory role of GITR in tissue inflammation [[7]]. Studies using agonist anti-GITR Ab or GITRL-Fc as an adjuvant further demonstrated GITR enhancement of both humoral and cellular immunity [[8-10]], and a GITR-mediated override of the suppressive capacity of Foxp3+ Treg cells [[5]], emphasizing a role of GITR in promoting Teff cell function in vivo. Interestingly, other studies have shown that GITR/GITRL interactions expand Treg cells nonspecifically in the presence of IL-2 [[4, 5]] and that Treg cells regain their suppressive activity upon removal of GITR signaling [[10]], suggesting the final outcome of GITR/GITRL interaction would be stronger immune suppression, especially at the late stage of an immune response when GITR signaling wanes due to downregulation of GITRL on APCs. This notion was supported by the absence of auto-immunity in animals treated with agonistic GITR-Ab due to the existence of functional Treg cells [[9, 10]].

To better understand the in vivo role of T-cell signaling through GITR, especially under long-term GITR stimulation, we generated a mouse that constitutively expresses GITRL on MHC class II+ cells. In contrast to a previously reported B cell-specific GITRL transgenic (Tg) mouse that has a balanced proliferation of both Teff and Treg compartments with an intact naive (Tn) population [[11]], we found excessive T-cell activation and multiorgan lymphocytic infiltrations in our Tg model. However, we also discovered that long-term GITR signaling could generate additional/alternative immunosuppressive cells, including the differentiation of Tr-1like cells by increased levels of IL-27, which may control peripheral tissue inflammation through the production of the anti-inflammatory cytokine IL-10.


Expanded Treg cells in GITRL Tg mice are unable to control constant naïve T cell activation in vivo

In GITRL Tg mouse, 80∼90% of cells that constitutively express MHC class II molecules, including the majority of B cells, DCs, NK cells, and a fraction of macrophages, constitutively express Tg GITRL on the surface (Supporting Information Fig. 1 and Supporting Information Table 1). Surface expression of the GITR receptor on APCs and T cells is decreased, especially on Foxp3+ Treg cells (MFI: 15.9 ± 0.6), to the level expressed on non-Treg CD4+ cells (MFI: 15.28 ± 0.5) of WT mice (Fig. 1A and B), a phenomenon likely due to ligand stimulation. These data indicate that we have generated a mouse with physiological pattern of expression of GITRL on all types of APCs and that GITR is functionally engaged by its ligand in vivo.

Figure 1.

Analysis of lymphocyte populations and T-cell subpopulations in the spleens of GITRL Tg mice. (A) The frequency of GITR+ cells on gated CD11c+, CD19+, CD3+, and CD11b+ splenic populations is shown as mean ± SDof n = 8 mice. (B) Expression of GITR and Foxp3 in the CD3+CD4+ spleen cells of WT and Tg mice. One individual representative of n = 8/group is shown. (C) The absolute numbers of total splenocytes, B cells, total T cells, and CD4+ and CD8+T-cell subsets from WT and Tg mice at the indicated ages are shown as mean ± SD of n = 8–13/age group. Frequencies and absolute numbers of (D) Treg cells, (E) Teff/mem cells, and (F) Tn cells, and (G) the Tn/Teff ratio in WT and Tg mice at the indicated ages are shown as mean ± SD of n = 8–13/age group. (A, B) Data shown are representative of three independent experiments using 12–14 week old mice. *p < 0.01, unpaired Student's t-test.

The outcome of this enhanced GITR/GITRL interaction is reflected in the numbers and ratios of lymphocyte subsets in the spleens of GITRL Tg mice, in line with the important involvement of GITR signaling in T-cell homeostasis. GITRL Tg mice have enlarged spleens and peripheral lymph nodes (data not shown) and increased cellularity in these organs (Fig. 1C and Supporting Information Table 1). The frequency and absolute numbers of B cells, T cells, DCs, and macrophages are all increased in the Tg mice compared with the WT littermates of the same ages. Despite a suggested role of GITR signaling in CD8+ T-cell activation, the cellularity of CD8+ T cells in Tg mice is comparable with that in WT littermates (Fig. 1C), indicating that GITRL overexpression does not appear to modulate CD8+ T cells. The expansion of CD4+ T-cell compartment likely accounts for the total expansion of T cells in Tg mice.

Treg cells are expanded to 30∼50% of the total CD4+ T-cell population, approximately five times more Treg cells per Tg mouse as compared with that of WT littermates (Fig. 1D). In vitro, Treg cells from Tg and WT mice displayed similar capacities to inhibit anti-CD3-induced naïve T-cell proliferation (data not shown). As expected, CD44+CD62L T effector/memory (Teff/Tmem) cells also expanded rapidly and peaked at ∼10 times more cells (Fig. 1E). However, CD44CD62Lhi naïve (Tn) cells had a steady decline in Tg mice (Fig. 1F). The absolute number of Tn cells in Tg mice dropped significantly starting at ∼9 weeks of age (when the rapid expansion phase stopped in the spleen). At the age of 20 weeks, the Tn/Tact ratio in the CD4+ pool of WT littermates was ∼2.9 whereas in Tg mice it was only 0.2 (Fig. 1G). These results demonstrate that enhanced GITR signaling triggered by the overexpression of GITRL leads to increased T-cell activation and reduction of naïve T cells in vivo, despite the presence of abundant numbers of Treg cells.

Figure 2.

GITRL Tg mice have multi-organ lymphocytic infiltration without overt autoimmunity. (A) H&E staining of liver, pancreas, and kidney from 14-week-old WT or Tg mice showing perivascular infiltrations (arrows). (B) The amounts of anti-ds DNA IgG in the serum of WT and Tg mice at age of 14 weeks were normalized to that of a MRL/lpr mouse with systemic lupus erythematosus and shown as individual values and mean ± SD of n = 10/group. (C, D) H&E staining of colon from (C) 14-week-old WT or Tg mouse, or (D) Tg mice at age of 20 weeks (healthy colonic mucosal epithelium with goblet cells), 48 weeks (diffuse, minimal expansion of the mucosa by lymphocytes), and 58 weeks (diffuse, mild expansion of the mucosa by edema, and lymphocytes). All micrographs shown are at 10× magnification, one mouse representative of three. Data shown are representative of two experiments.

GITRL Tg mice have multiorgan lymphocytic infiltration, yet display no overt autoimmunity

In light of the extensive T-cell activation and accumulation of Teff cells in GITRL Tg mice, we examined Tg mice for signs of systemic autoimmunity and inflammation. All Tg mice appeared healthy, bred well, and no spontaneous death was observed in mice up to 58 weeks of age. In addition to the enlarged spleens and lymph nodes, we observed massive perivascular infiltrations in multiple organs, including the lung, liver, intestinal tract, pancreas, sal-ivary gland, and kidney (Fig. 2A). These infiltrates were found to consist mainly of lymphocytes. Interestingly, there were no overt signs of autoimmunity such as proteinuria, high levels of glucose in the blood, or high titers of anti-dsDNA Abs (Fig. 2B and data not shown). In an attempt to delineate the mechanism(s) underlying the lack of overt pathology, we carefully examined the Tg mice at 6, 14, 20, 33, 48, and 58 weeks of ages for signs of inflammatory bowel disease (IBD), a disease often present in animals with systemic autoimmunity. Tg mice younger than 20 weeks of age had occasional small, focal aggregates of lymphocytes in duodenum, jejunum, ileum, and colon, as was also observed in WT mice (Fig. 2C and Supporting Information Fig. 2). In 48- and 58-week-old GITRL Tg mice, there was a diffuse, minimal-to-mild expansion of the small and large intestinal mucosae due to edema and lymphocytic infiltrates (Fig. 2D and Supporting Information Fig. 2). Although these diffuse infiltrations of mucosa by lymphocytes appeared to represent progression of inflammation, severe histological changes were not observed, including villous atrophy/fusion, mucosal ulceration, crypt abscessation, hemorrhage, mixed cellular infiltration, bacterial overgrowth, and fibrosis, which can be observed in standard murine colitis models [[12]]. In agreement with these microscopic findings, we did not observe any diarrhea or bleeding (by occult blood test) in Tg mice of any age. We also did not observe weight loss, although compared with WT mice, Tg mice gained less weight after 20 weeks of age. Together, these data suggest that the lymphocytic infiltration did not elicit IBD in these mice.

Increased IL-10 production by T cells in GITRL Tg mice

After an initial rapid cell activation and expansion (<age 6 weeks), cell numbers plateaued at the age of 15∼20 weeks in the Tg mice (Fig. 1C), suggesting compensatory mechanisms developed and helped the immune system return to homeostasis. Further studies were thus designed to investigate the mechanisms that prevent the progression of activation, especially the differentiation of T cells into autoreactive pathogenic Teff cells, in the presence of strong GITR costimulation.

Despite excessive T-cell activation, we found significantly lower sera IL-2 levels in Tg mice compared with WT littermates (Fig. 3A). Cytokines that are important for Th1 cell differentiation (TNF-α and IL-12), but not for Th2 differentiation (IL-4, IL-5, or IL-13) were elevated. Interestingly, levels of IL-10, a potent anti-inflammatory cytokine, were significantly elevated in the Tg sera samples (Fig. 3A). To find the source of IL-10 secretion, we measured IL-10 mRNA levels in sorted T, B, myeloid, and DCs from spleens of WT and Tg mice (Fig. 3B). The IL-10 transcripts appeared more abundant in T cells than any other cell types; the T cells in Tg mice contained 10-fold more copies than WT T cells, suggesting that T cells could be the major source of IL-10 in the Tg mice. Consistent with this hypothesis, we found a significant fraction of memory T cells from Tg spleen produced IL-10 (Fig. 3C). T cells that produced IFN-γ also were increased in the spleens of Tg mice. Similar frequency of IL-2+ and IL-17A+ cells were detected in WT and Tg mice, along with a slight increase in IL-4+ cells. Together, these data suggest that T cells activated by strong GITR costimulation were skewed to Th1 cells that underwent significant expansion beginning at 4 weeks of age (Supporting Information Fig. 3A). This process occurred despite the presence of abundant Treg cells, indicating that T-cell differentiation also may not be subject to suppression by Treg cells in these Tg mice.

Figure 3.

Increased IL-10 production by T cells in GITRL Tg mice. (A) Cytokines in the sera are shown as individual values and mean ± SD of n = 10/group, 14 weeks of age. IL-5 and IL-13 were under detection limits. Statistical significance was determined by unpaired Student's t-test. (B) Expression of Il10 mRNA in FACS-sorted CD3+, CD19+, CD11b+, and CD11c+ splenocytes of 12-week-old mice is shown as mean ± SD of triplicate samples. Data are relative to mRNA copies in DCs of WT mice. *p < 0.01, unpaired Student's t-test. (C) Intracellular expression of the indicated cytokines in gated CD3+CD4+ spleen cells of one representative individual of each group is shown. Mice were used at age of 12 weeks. Data shown are representative of three (A, C) or four (B) independent experiments (n = 5/group).

Increased Foxp3 IL-10-producing T cells with suppressive function in adult GITRL Tg mice

Tr-1 cells are CD4+ T cells that do not use the Foxp3 transcription factor, yet predominantly produce IL-10 and are involved in T-cell suppression and immune resolution [[13]]. In WT spleens, less than 2% of CD4+ T cells were IL-10 producers, and half were Foxp3IL-10+ cells (Fig. 4A); this ratio was maintained as mice aged (Fig. 4B). However, in the spleens of GITRL Tg mice, the Foxp3IL-10+ subset, dramatically increased from ∼1% at 2 weeks of age to 6∼7% of total CD4+ T cells at 12 weeks (Fig. 4A and B). A significant fraction of these cells also coproduced IFN-γ (Fig. 4C and D), suggesting that chronic GITR/GITRL signaling induced the generation of Foxp3IL-10-producing Tr-1-like cells.

Figure 4.

Increased Foxp3IL-10-producing T cells with suppressive function in adult GITRL Tg mice. Intracellular expression of (A) IL-10 and Foxp3 in the CD3+CD4+ spleen cells or (C) IL-10 and IFN-γ in the CD3+CD4+Foxp3 spleen cells of WT and Tg mice at 12 weeks of age are shown as one individual representative of n = 5/group. (B) The frequencies of Foxp3IL-10+ cells in the splenic CD4+ T-cell population at the indicated ages are shown as mean ± SD of n = 8–13/group. (D) The percentages of splenic IL-10+ and IFN-γ+IL-10+ cells in the CD3+CD4+Foxp3 fraction of WT and Tg mice at 12 weeks of age are shown as individual values and mean of n = 5/group. (E, F) The expression of (E) Il10 and (F) Ifng mRNA in sorted CD4+T-cell subsets (as described in Supporting Information Fig. 4A) of 12-week-old mice are shown as mean ± SD of triplicate samples. (G) Tn cells from WT mice as Tresponder and Tr-1-like cells from Tg mice were cocultured at the indicated ratios without or with anti-IL-10 Ab. Proliferation was measured at 48 h and shown as mean ± SD% proliferation of controls that contained only Tresponder cells (n = 3). All data are representative of three experiments. *p < 0.01, unpaired Student's t-test, E and F: Tr-1 versus all other T-cell subsets.

We next isolated IL-10-producing CD4+ T cells from the Tg spleen to further investigate their function. Based on the phenotype defined in a recent report using IL-10/GFP+-Tr-1 cells [[14]], we postulated that these IL-10-producing Tr-1 cells were derived from activated Teff cells, and that they were in the CD4+CD25CD44+CD127 fraction. Thus, we sorted CD4+CD25CD44+CD127 Teff cells from Tg spleen cells, and compared their IL-10 expression to sorted memory T cells (CD4+CD25CD44+CD127+), Tn (CD4+CD25CD44CD127+) cells, and Treg (CD4+CD25+) cells (Supporting Information Fig. 4). Consistent with the previous report, only the CD44+CD127 fraction of Teff cells had high levels of Il10 mRNA expression (Fig. 4E), suggesting that this fraction of Teff cells contained Tr-1-like cells. Interestingly, this was the population that also expressed highest level of Ifng transcripts (Fig. 4F). Furthermore, the CD44+CD127 fraction of Teff (Tr-1-like) cells inhibited the proliferation of naïve WT T cells in response to anti-CD3 stimulation in vitro in a dose-dependent manner (Fig. 4G). Addition of neutralizing anti-IL-10 mAb relieved suppression of T-cell proliferation, indicating that suppression by the Tg Tr-1-like cells was mediated by IL-10 to a large extent. Therefore, in addition to the expansion of Treg cells, we discovered that enhanced GITR signaling induced another type of suppressive T cells in the Tg mice.

Increased IL-27 production in GITRL Tg mice promotes Tr-1 differentiation

We then explored possible mechanisms that would cause the induction of IL-10 from activated T cells in the GITRL Tg mice. IL-10 itself has long been suggested as an inducer of Tr-1 cells, particularly when it comes from immature DCs. Although Tg DCs had higher levels of IL-10 expression compared with WT DCs (Fig. 3B), they displayed a “mature DC” phenotype with elevated surface expression of signature activation markers CD86, CCR7, FcγR III/II (CD16/32), and IFN-γR (CD119) (Fig. 5A).

Figure 5.

Increased IL-27 expression by APCs and activated IL-27 signaling in GITRLTg T cells. (A) Mean fluorescent intensity (MFI) of indicated surface protein expression on gated CD11c+ and CD11b+ subsets is shown as mean + SD of five samples. (B, E, F) The expression of (B) Il27 mRNA in CD11c+ and CD11b+ cells, (E) Ahr and Maf mRNA, and (F) Il21, Il21r, and Gzmb mRNA in sorted CD4+T-cell subsets from spleens of 14-week-old mice is shown as mean ± SD of triplicate samples. Data were expressed as relative to copies in WT mice (B) or in WT Tnaive subset (E, F). (C) The surface expression of IL-27R (bold line histogram) on CD3+CD4+ spleen cells is overlayed with the isotype control Ab (shaded histogram). The frequency (top) and MFI (bottom) of IL-27R+ cells are depicted as mean ± SD of n = 5 samples. (D) The surface expression of ICOS and CD62L on CD3+CD4+ spleen cells is shown from one individual representative of n = 5 in each group. Experiments were performed two (E–F) or three (A–D) times using 12-week-old mice. *p < 0.05, unpaired Student's t-test, WT versus Tg. **p < 0.01, unpaired Student's t-test, Ahr: WT versus Tg; Maf and Gzmb: Tr-1 versus all other subsets.

IL-27 is produced by activated APCs and has been implicated in driving the differentiation of Tr-1 cells via activation of the transcription factors aryl hydrocarbon receptor (Ahr) and c-Maf in mice and human [[15-17]]. Significant increases in the transcriptional levels of IL-27 in the Tg APCs were observed when compared with those of WT littermates (Fig. 5B). In addition, a significant number (∼30%) of CD4+ cells in Tg spleens were IL-27R+ and a similar fraction also expressed ICOS on the cell surface (Fig. 5C and D), indicating possible IL-27R signaling in T cells. Expression of the Ahr transcript in all CD4+ T-cell subsets of Tg mice was higher than those in T-cell subsets of WT mice; levels of Ahr mRNA in Tg Tr-1-like cells were comparable with WT Treg cells (Fig. 5E). As expected, under the influence of both Ahr and ICOS [[16, 18]], the expression of Maf (encodes transcription factor c-Maf) mRNA was highest in the Tr-1-like subset of Tg mice. Consistent with the increased transcriptional activity of c-Maf, the expression of the c-Maf target genes [[16, 17]], Il10 (Fig. 4E), Gzmb (encodes granzyme B), Il21, and Il21r (Fig. 4F), was all increased significantly. Together, these data suggest that the enhanced signaling through GITR/GITRL in the Tg mice increases the in vivo production of IL-27 by APCs, activates the IL-27 signaling pathway in T cells, and induces IL-10-producing Tr-1-like cells in Tg mice. These cells have a phenotype resembling the de novo generated Tr-1 cells via IL-27 and Ahr/c-Maf signaling [[16, 17]].

Tr-1 cells, in addition to Treg cells, infiltrate the lung tissues in GITRL Tg mice

To further understand the function of Tr-1 and Treg cells in controlling peripheral tissue inflammation, we characterized the infiltrating lymphocytes in the lung, one of the most heavily infiltrated organs. As shown in Fig. 6A, significant perivascular lymphocytic infiltrates were readily observed at the age of 7 weeks. At 58 weeks of age, the infiltrates became very severe, yet no overt changes in pathology were observed (Fig. 6A, enlarged), such as the thickening of vessel walls, which often is one of the first signals of inflammation in the lung. FACS analysis of lung infiltrates revealed increased numbers of T and B cells, normal levels of macrophages or neutrophils, and undetectable levels of eosinophils or DCs in Tg mice (data not shown). This observation was further confirmed when cell-type-specific gene expression was quantified in the lung tissues. As shown in Fig. 6B, only the expression of the T-cell-specific gene, Thy1, was significantly increased in the lungs of Tg mice. The expression of the macrophage-specific gene, Emr1 (encodes F4/80) and neutrophil-specific gene, Il8rb (encodes IL-8 Rβ) was similar in Tg and WT lungs, while transcripts of the eosinophil-specific gene, Epx (encodes eosinophil peroxidase), were not detected in either Tg or WT lungs. Together, these data indicate that despite the massive infiltration of T cells, an influx of destructive inflammatory cells was absent. Additionally, the lack of tissue damage suggests that the effector function of the infiltrating T cells (mainly Th1 cells, data not shown) was impaired in the lung tissues of GITRL Tg mice.

Figure 6.

IL-10-producing Tcells infiltrate in the lung. (A) H& E staining of lung tissues at 7 and 58 weeks of age are shown at 10× magnification; enlarged views (boxed areas) from 58-week-old mice are shown at 40× magnification. Arrows indicate the intact structures of bronchia and blood vessels with walls comprised by a single layer of cells, one mouse representative of three. Experiments were performed twice. Expression of (B) Thy1, Emr1, and Il8rb mRNA and (C) Il10 mRNA in the lung tissues of WT and Tg mice is shown. Gapdh or Thy1 was used for normalization of total or T-cell numbers as indicated. Data are expressed as relative to copies in WT mice and depicted as individual values with mean ± SD of n = 5/group. Expression of (D) IL-10 and CD11b on CD45+ or (E) IL-10 and Foxp3 on CD45+CD4+ lymphocytes purified from lung tissues is shown from one individual representative of n = 5/group. (B–E) Experiments were performed three times using 12–16 week old mice. *p < 0.05, unpaired Student's t-test.

Evaluation of IL-10 expression revealed that Il10 mRNA was markedly increased (100-fold) in the lungs of Tg mice (Fig. 6C) as a result of the increased T-cell influx and a higher expression of IL-10 transcripts per cell. There was no difference detected in the percentage of IL-10-producing CD11b+ cells in lung infiltrates (Fig. 6D), suggesting that these cells were not a major source of IL-10 in this tissue. Further analysis of the CD4 compartment of lung infiltrates in the Tg mice revealed two-fold more IL-10+ Tr-1 cells than IL-10+ Treg cells, which together reached ∼6% of total CD4+ T cells (Fig. 6E). Overall, these data suggest that increased levels of IL-10 in peripheral tissues of GITRL Tg mice may provide a mechanism for the downregulation of local tissue inflammation.


Using a transgenic mouse model, we described the effects of constitutive expression of GITRL on APCs. The phenotype of the GITRL transgenic mouse indicates that a chronic increase in immunostimulation and compensating immunosuppression are both occurring in vivo, seemingly in a functionally sequential and balanced way to ensure immune and peripheral tissue homeostasis and to prevent the generation of autoimmune manifestations.

Consistent with previous observations that GITR provides a positive costimulation to Teff and Treg cells, we found increased cellularity in both T-cell populations and also an expanded APC compartment. Excessive T-cell activation led to enhanced Th1 differentiation in lymphoid tissues and the Th1-dominant lymphocytic infiltrations in multiple organs. However, there was a lack of overt autoimmunity that could be explained by the increase of compensating immunosuppression involving two types of suppressor T cells, conventional Foxp3+ Treg and Foxp3IL-10+ Tr-1 cells. Treg cell numbers are increased roughly five-fold in the spleen and three-fold in the lung tissue of Tg mice. Although the suppressive functions of Tg Treg cells appeared normal in vitro, the suppressive function of Treg cells in vivo may be reduced. Several studies have pointed out that stimulation through GITR stimulation breaks immune suppression by Treg cells [[5, 19]], and we also detected a significant increase in Teff/Tmem and a decrease in Tn cells in vivo. This defect in Treg cell-mediated immune suppression in vivo, at least in terms of T-cell activation in context of the Tg mice, could be due to an inability of the Treg cells to regulate antigen presentation in this system [[20]], since no downregulation of surface costimulatory molecules (CD80/CD86) was observed on Tg APCs (Fig. 5A). In addition, we have detected increased levels of IL-21 (Fig. 5F and Supporting Information Fig. 3B), a cytokine that renders Teff resistant to Treg-mediated suppression [[21]]. Therefore, upregulated IL-21 production could also explain the escape from Treg-mediated suppression, leading to the observation of impaired in vivo Treg cell function. However, the overall functionality of Treg cells in the Tg mice awaits further investigation.

Tr-1 cells play an important role in immune resolution by inhibiting effector functions of both innate and adaptive immune cells [[22]]. In addition to Treg cell expansion, we also noted a significant increase in the numbers of peripheral IL-10-producing Tr-1-like cells, which may have arisen after increased Th1 proliferation (Fig. 4B and Supporting Information Fig. 3) and acted as a feedback regulator to dampen the proinflammatory Th1 resp-onses [[13]]. A 100-fold increase in IL-10 mRNA in the lungs of GITRL Tg mice was detected, expressed mainly by T cells. While these tissues are heavily infiltrated with Th1 cells, their function maybe inhibited by IL-10 produced by both Tr-1 and Treg cells that are present in the infiltrates. While it is difficult to separate the relative contributions of Treg cells and Tr-1 cells in overall immune suppression, it is possible that the Tr-1 cells play a role in limiting the inflammation associated with the tissue lymphocytic infiltrates seen in the GITRL transgenic mice.

The importance of Tr-1 cells to control local inflammation has been described recently in a genetically mutated mouse model (CNS1-KO) with a significantly reduced Treg-cell presence in gut-associated lymphoid tissues. In this study, Zheng et al. reported increased numbers of Tr-1 cells [[23]] that correlated with an absence of immune-mediated lesions in the colon, small intestine, or elsewhere in CNS1-KO mice, as well as no Th17 differentiation (similar to our observations in GITRL Tg mice). These findings were unexpected, considering the current view that compromised Treg-cell function leads to enhanced Th17 differentiation and inflammation. Rather, Tr-1 cells protected these mice from dextran sulphate sodium-induced colitis.

We demonstrated that these Foxp3-IL-10-producing cells inhibited naïve T-cell proliferation in vitro in an IL-10-dependent manner. Although we were not able to demonstrate directly the in vivo functions of IL-10 by using a commercially available rat anti-IL-10 mAb (data not shown), the effects of IL-10 are evident in our GITRL Tg model. For example, it is known that IL-10 regulates growth and promotes survival of resting mouse B cells, and that it inhibits Ab production against thymus-independent type I and type II Ag [[22]]. In GITRL Tg mice, the number of B cells, including B-1 cells, is almost doubled when compared to their WT littermates (Fig. 1C). NP-Ficoll-induced Ab responses were significantly decreased in Tg mice (Supporting Information Fig. 5). Another example comes from the absence of proinflammatory innate immune cells (eosinophils, neutrophils, and macrophages) in the infiltrates of the lung, where increased Tr-1 cells can be detected. The phenomenon can be explained by one of the major functions of IL-10, the inhibition of the chemotaxis response of activated macrophages and T cells, through the inhibition of chemokine production and receptor expression [[22]]. Therefore, IL-10, produced predominantly by Tr-1 like cells in this system, could effectively control both humoral and cellular immunity in vivo.

The identity of Tr-1 cells as a unique T-cell subset is still debatable due to lack of a lineage-specific transcription factor or surface marker. IL-10 is produced by various subsets of Th cells after chronic stimulation, resulting in the gradual disappearance of effector cytokines and preventing excessive inflammation and immunity [[13]]. A large percentage of IL-10-producing cells detected in GITRL Tg mice coproduced IFN-γ, indicating that these cells could be “regulatory IL-10-producing Th1 cells” that are derived from activated Th1 cells, which were increased in the memory T-cell pool of the Tg mice. However, they have a phenotype that is similar to IL-27-induced Tr-1 cells, such as an activated Ahr/c-Maf pathway and increased expression of IL-21, IL-21R, granzyme B, and T-bet (Fig. 5 and data not shown), indicating that increased IL-27 production by Tg APCs may be responsible for the generation of these cells in the Tg mice [[15-17]]. Although the origin of these Tr-1-like cells is unclear, for example, de novo generated versus differentiated from activated Th1 cells, IL-27 had been proved to be the master inducer of IL-10 by T cells activated in Th0-, Th1-, or Th17-polarized conditions [[16]].

The causal relationship of enhanced GITR signaling and prod-uction of IL-27 has not been demonstrated. There is evidence that IFN-γ signaling contributes to the transcriptional activation of the Il27 gene (encoding p28 subunit) through the recruitment of IFN-regulatory factor 1 (IRF1) to the IFN-stimulated response element (ISRE) site on the Il27 gene [[24]]; through this cytokine crosstalk, increased production of IFN-γ by T cells from GITRL Tg mice likely results in the enhancement of IL-27 secretion by APCs. In addition, considering that once macrophages and DCs are activated, they upregulate IFN-γR (Fig. 5A) and produce large amounts of IFN-γ (Supporting Information Fig. 6), this autocrine loop of IFN-γ could further enhance the production of IL-27. Of note, GITR was not detected in the FACS analysis of Tr-1-like cells, but Gitr mRNA was present in these cells at similar levels to Tn cells, yet at significantly reduced levels to Treg cells in Tg mice (Supporting Information Fig. 4B), indicating that GITR signaling could play a role in the expansion of activated T cells (Tmem and Tr-1). While we believe enhanced production of IL-27 plays a dominant role in Tr-1 differentiation, our data are inconclusive as to whether GITR directly acts as a differentiation factor or it is indirectly through a mechanism such as the induction of IL-27.

It has been demonstrated with soluble GITR-Ig fusion proteins that reverse signaling through GITRL in pDCs activates a noncanonical NF-κB pathway, leading to indoleamine 2,3-dioxygenase (IDO)-mediated protection against allergic bronchopulmonary aspergillosis [[25]]. It is possible that constitutive overexpression of GITRL may result in IDO-dependent immune regulation. When we subjected purified DCs and macrophages from WT and Tg mice to the measurement of gene expression of Ido, a trend of increased transcripts was detected in cells from Tg mice, although it did not reach statistical significance (Supporting Information Fig. 6). In addition, the expression levels of IFN-α, an important intermediate protein proposed to be critical for GITRL-mediated IDO activation [[25]], were similar in APCs from both WT and Tg mice, further indicating that the IDO pathway may not be a major contributor to the overall immune suppression that controls excessive Teff function.

Overall, the results presented here support our hypothesis that enhanced GITR/GITRL interactions in the absence of a strong TCR signal lead predominantly to the generation of Th1 cells and an increase in Treg cell numbers, with impaired in vivo functional capability to control excessive Teff cell generation. As a feedback response for the maintenance of self-tolerance, IL-10-secreting Tr-1-like cells are induced and contribute to the suppression of excess inflammation and to the prevention of associated immunopathology [[26]]. We propose that Tr-1 differentiation represents a compensatory regulatory mechanism, and that Tr-1 cells regulate, in coordination with Treg cells, the maintenance of immunological self-tolerance.

Materials and methods

Generation of GITRL Tg mice

The mouse class II MHC Eα promoter was cloned upstream of full-length mGITRL cDNA followed by a downstream bovine growth hormone polyadenylation sequence. Four founder lines on C57BL/6 background were generated. Line 23, which by flow cytometry expressed the highest GITRL levels on MHC class II cells was selected for these studies. GITRL Tg mice were maintained on a C57BL/6 background and bred under specific pathogen-free conditions in the Pfizer animal facilities. Age-matched female heterozygous mice and littermates were used at indicated ages. All protocols were approved by the Pfizer Animal Care and Use Committee.

FACS analysis

Single cell suspensions were obtained from spleens by mincing through 70 μm cell strainers followed by erythrocyte lysis. Lung infiltrates were obtained by harvesting lung tissues after whole-body PBS perfusion, digestion with 150 U/mL collagenase III in culture media at 37°C for 1 h, and Ficoll separation. For intracellular cytokine detection, cells (2 × 107 cells/mL) were first stimulated for 4 h at 37°C in the presence of 50 ng/mL PMA, 1 μg/mL ionomycin, and 1 μl/mL GolgiStop (BD Biosciences). This method only detects cytokine production from memory T cells. All cells were stained according to manufacturer's protocol and dead cells were excluded by 7AAD incorporation (BD Biosciences). Data were acquired with a LSR-II cytometer (BD Bioscience) and analyzed with FlowJo (Treestar) software.

Anti-mouse GITRL (clone 5F1, Pfizer) was generated as previously reported [[5]]. The following fluorescent- or biotin-labeled anti-mouse mAbs (and clone names) were obtained from BD Bioscience: CD3 (145-2C11), CD4 (L3T4), CD8 (Ly-2), CD11c (HL3), CD25 (PC61), CD45R (RA3-6B2), CD49b (DX5), CD62L (MEL-14), CD127 (A7R34), GITR (DTA-1), I-A/I-E (2G9), ICOS (7E.17G9), IFN-γ (XMG1.2), IL-2 (JES6-5H4), IL-4 (11B11), IL-10 (JES5-16E3), IL-17A (TC11-18H10); from eBioscience: CD19 (eBio1D3), Foxp3 (FJK-16s); from Invitrogen: CD11b (RM2828,), CD44 (RM5726); and from R&D Systems, IL-21 (149204).

Histopathology analysis

Tissue specimens were fixed in 10% buffered formalin and embedded in paraffin. Five-micrometer sections were stained according to standard protocols with H&E.

Quantitative real-time PCR

T cells, B cells, DCs, and macrophages were sorted on FACSAria III (BD Bioscience) based on surface expression of CD4, CD19, CD11c, or CD11b. More than 98% purity was routinely obtained. RNA extracted (Qiagen, RNeasy mini kits) from sorted cells was subjected to quantitative RT-PCR using TaqMan Gene Expression Assays (Applied Biosystem). Data were expressed as 2−ΔCt relative to transcripts of control samples from WT mice. Gapdh or Thy1 was used for normalization to total cell or T cell numbers, as indicated.

Regulatory T-cell suppression assay

Each T-cell subset was FACS sorted from B cell-depleted (anti-mouse CD19-microbeads) splenocytes. Splenic CD4+CD25 from naive WT mice and CD4+CD25CD44+CD127 (Tr-1) cells from Tg mice were mixed at indicated ratios in 96-well U-bottom tissue culture plates. The cells were stimulated with 1 μg/mL of soluble anti-CD3 mAb (clone 145-2C11) with RBC-lysed whole spleen cells (irradiated, 4000 rad) from WT mice as APCs, in the absence or presence of 25 μg/mL of anti-IL-10 blocking mAb (clone JES5-16E3, BD Bioscience). Cell proliferation was measured as thymidine incorporation using a microbeta liquid scintillation counter (PerkinElmer). Data were calculated as pecentage proliferation of control cultures that contained only the responder cells and presented as mean ± SD of triplicate wells.

Statistic analysis

Statistic analysis was performed using an unpaired Student's t-test.


We thank Dr. Nancy Stedman, Lawrance Mason, and Jie Cai for assistance in histological analysis and Dr. Janet Buhlmann for critical review.

Conflict of interest

All authors were full time employees of Pfizer during this study.


glucocorticoid-induced TNFR-related

Teff cell

effector T cell

Tmem cell

memory T cell

Tn cell

naïve T cell

Tr-1 cell

type 1 regulatory T cell