Abstract
17β-Estradiol (E2) suppresses the development of experimental autoimmune encephalomyelitis (EAE) through estrogen receptor (ER) α, yet the cellular targets remain elusive. We have used an adoptive transfer model of myelin oligodendrocyte glycoprotein–specific CD4+ T cells from 2D2 TCR transgenic mice. We show that in the recipient mice, ERα expression in bystander CD4+ T cells, rather than in cognate 2D2 T cells, is required for the inhibition of Th17 cell differentiation by E2. Coadministration of estrogen-primed WT, but not ERα-deficient CD4+ T cells, with naive 2D2 T cells lacking ERα inhibited the development of Th17 cell–mediated EAE. Suppression of Th17 cells and protection from EAE were maintained when ERα was deleted in Foxp3+ regulatory T cells. We showed that in vivo PD-L1 blockade alleviated the anti-inflammatory action of E2 and that PD-1 expression on cognate but not bystander T cells was required for the E2-dependent inhibition of Th17 differentiation. In cotransfer experiments, we found that only WT but not PD-1KO 2D2 T cells were amenable to E2-dependent inhibition of Th17 differentiation. These results support the conclusion that the restriction of Th17 cell development by E2-primed bystander CD4+ T cells requires cell-intrinsic PD-1 signaling within cognate T cells rather than induction of regulatory 2D2 T cells through PD-1 engagement. Altogether, our results indicate that pregnancy-level concentrations of estrogen signal in conventional Foxp3neg CD4+ T cells to limit the differentiation of cognate Th17 cells through a trans-acting mechanism of suppression that requires a functional PD-1/PD-L1 regulatory axis.
Introduction
Multiple sclerosis is a T cell–mediated autoimmune disease with a female to male ratio of 3:1. Multiple sclerosis and its mouse model, experimental autoimmune encephalomyelitis (EAE), are characterized by the infiltration of inflammatory leukocytes, including autoreactive T cells, into the CNS, resulting in myelin damage (1). Reduced disease activity in women with multiple sclerosis is commonly observed during pregnancy, suggesting that sex hormones, such as estrogens, could downmodulate the autoimmune response and inflammation (2–4). Indeed, studies in EAE have clearly established protective effects of estrogen administration on disease activity (5–10). Likewise, protective effects of estrogens have been reported in clinical trials using estriol (11–13) or 17β-estradiol (E2) (14) in relapsing-remitting multiple sclerosis patients. These disease-modulating effects of exogenous estrogens in CNS autoimmunity are thought to account for the beneficial effects of pregnancy on clinical symptoms in women with multiple sclerosis (15, 16).
Estrogens bind to nuclear receptors, thereby directly or indirectly regulating the expression of numerous genes. Estrogens can also activate intracellular signaling pathways. There are two estrogen receptors (ERs): ERα and ERβ. These ERs, most notably ERα, are functionally expressed by many immune cells, including T cells (10), B cells (17), macrophages (18, 19), and dendritic cells (DCs) (20–24), suggesting that estrogens could regulate immunity through a direct effect on immunocompetent cells. Alternatively, ERs are expressed by CNS-resident cells such as neurons (25), astrocytes (25–27), and microglia (28). Recent works have shown that estrogens or ER-selective ligands can exert potent neuroprotective effects by acting in CNS-resident cells through ERα (25–27) or ERβ (28, 29). Thus, emerging evidence supports the notion that E2 could inhibit CNS autoimmunity and inflammation through distinct nonoverlapping mechanisms, namely anti-inflammatory and neuroprotective actions (15, 16).
In the active EAE model, priming of autoantigen-specific CD4 T cells in the secondary lymphoid tissues by DCs, which produce polarizing cytokines such as IL-12, IL-23, and IL-6, promotes the development of pathogenic CD4 T cells into Th1 and Th17 phenotype (30, 31). Both Th1 and Th17 cells can induce EAE, albeit with different symptoms, depending on the preferential recruitment of either cell type to the spinal cord or the brain parenchyma, respectively (32, 33). IL-23–driven GM-CSF secretion by pathogenic Th17 cells is essential for the induction of neuroinflammation in EAE (34, 35). A consistent observation in E2-mediated EAE protection is the strong anti-inflammatory effect characterized by the inhibition of autoantigen-specific CD4+ T cell responses in lymphoid organs, including both Th1 (5, 7, 8, 10, 36) and Th17 cells (9, 10). This immunosuppressive mechanism triggered by E2 treatment occurs early in the course of the myelin oligodendrocyte glycoprotein (MOG)–specific immune response in draining lymph nodes (10) and is likely to account for the sustained suppression of EAE generally observed in E2-treated mice (5, 7, 36).
Several hypotheses have been proposed to explain the anti-inflammatory effect of E2 on EAE, such as enhanced expansion of Foxp3+ regulatory T (Treg) cells (9, 37), induction of tolerogenic DCs (38) or regulatory B cells (39), and direct inhibition of Th17 cells (40). However, the precise dominant cellular targets in which estrogen signals in vivo to inhibit Th17 cells priming in draining lymph nodes had remained elusive until we established, using a conditional knockout strategy, that ERα expression in T cells, but not DCs nor macrophages, was necessary and sufficient to mediate the protective effect of E2 on EAE (10).
In the current study, we seek to precisely identify the T cell populations targeted by E2 through the generation of new mouse models in which ERα was selectively invalidated in cognate MOG-specific CD4+ T cells or bystander CD4+ T cells, including Foxp3+ Treg cells. Altogether, our findings identified bystander Foxp3neg CD4+ T cells as the main cellular targets of E2 that restrict Th17 cell differentiation in trans and protect from CNS autoimmunity through a mechanism that requires PD-1 on cognate CD4+ T cells but not Treg cells.
Materials and Methods
Mice
Female C57BL/6JRJ (B6) mice were purchased from the Centre d’Elevage R. Janvier (Le Genest St. Isle, France). Mice with a disrupted Esr1 gene (hereafter called ERαKO), previously described (41), were backcrossed onto the B6 background for at least 10 generations. To generate mice lacking T and B cells and deficient for ERα, B6, and ERα+/−, mice were crossed to B6 Rag2KO mice. To generate TCR transgenic mice lacking ERα, B6 ERα+/− mice were crossed to OVA-specific OT-2 CD45.1 mice or 2D2 CD45.1 mice (42) expressing a MOG35–55/I-Ab–specific TCR. Mice selectively lacking ERα in T lymphocytes or in DCs have been described elsewhere (10). They were obtained by crossing ERαfl/fl mice (41) with mice expressing the Cre recombinase under the control of CD4 (43) or CD11c (44) promoter, respectively. The targeted exon 2 of the Esr1 gene encodes the first zinc finger of the DNA-binding domain of ERα. Cre-mediated recombination of the conditional floxed allele leads to unambiguous inactivation of ERα. In the current study, we generated mice selectively lacking ERα in Treg cells (Foxp3+ cells) (Foxp3YFPCre ERαfl/fl) by crossing ERαfl/fl mice with mice expressing the Cre recombinase and the YFP protein fluorescent genes under the control of Foxp3 promoter (45). PD-1KO (46) and 2D2 PD-1KO (47) mice have been previously described. For the generation of irradiation bone marrow chimeras, mice were γ-irradiated (8.5 Gy [850 rad], [137Cs] source) the day before reconstitution with 5 × 106 PD-1KO or WT bone marrow cells per mouse. Mice were used 8 wk after reconstitution. All mice were housed in specific pathogen-free conditions. All animals were handled according to the Animal Care and Use of Laboratory Animal guidelines of the French Ministry of Research (study approval number 05187.01).
Induction of EAE
Purified CD4+ T cells (4–5 × 106 cells) from CD4Cre ERαfl/fl, Foxp3YFPCre ERαfl/fl, and their littermate control mice (ERαfl/fl or Foxp3YFPCre ERα+/+) pretreated or not with E2, were transferred i.v. together with isolated naive 2D2 T cells (2 × 104 cells) into castrated RAGKO ERαKO mice (male or female). One day after adoptive transfer, recipient mice were immunized s.c. in the flanks with 50 μg of MOG35–55 peptide (Neosystem, Strasbourg, France) emulsified in IFA supplemented with Mycobacterium tuberculosis (H37RA; BD, St. Louis, MO) (CFA at 1 mg/ml) and were injected i.p. with 200 ng of pertussis toxin (Calbiochem, Darmstadt, Germany) on day 0. Mice receiving CD4+ T cells pretreated with E2 were also treated with E2 to maintain E2 signals all along the experiment. Mice were daily scored for clinical signs of disease as follows: 0, no detectable signs of EAE; 1, complete limp tail; 2, limp tail and hindlimb weakness; 3, severe hindlimb weakness; 4, complete bilateral hindlimb paralysis; 5, complete hindlimb paralysis and forelimb weakness; 5.5, total paralysis of both forelimbs and hindlimbs; and 6, death. At the peak of disease, mice were sacrificed, and mononuclear cells were isolated from the spinal cord or brain, counted, and analyzed for cytokine expression. Cumulative disease indexes were calculated as the sum of daily clinical score of each individual mouse.
E2 treatment
For E2 treatment, we used the standard protocol described elsewhere (10). Briefly, female mice were implanted s.c. in the scapular region with an E2 pellet (2.5 mg, 60-d release; Innovative Research of America) 7 d before CD4+ T cells purification or immunization with MOG35–55 peptide according to the experiments. Control untreated mice were sham operated and received no pellet.
T cell assays
Naive 2D2 T cells (1–2 × 106 or 0.15 × 106 cells) either purified from WT or ERαKO mice were injected into WT, CD4Cre ERαfl/fl, CD11cCre ERαfl/fl, or Foxp3YFPCre ERαfl/fl recipient mice pretreated or not with E2. The following day, recipient mice were immunized in the flank with MOG35–55/CFA as for EAE induction. Five to six days postimmunization, draining lymph nodes were harvested. For the recall assay with lymph node cells (LNC), total LNC (0.5 × 106 cells) were in vitro stimulated with MOG35–55 peptide in HL-1 synthetic medium. Cultures were incubated for 72 h in a humidified atmosphere of 5% CO2 in air. For T cell proliferation assays, cells were pulsed during the last 8 h of culture with 1 μCi of [3H]TdR (GE Healthcare, Little Chalfont, U.K.). Cells were harvested onto glass fiber filter membranes, and [3H]TdR incorporation was measured by a MicroBeta TriLux luminescence counter (PerkinElmer, Waltham, MA). For cytokine analysis, LNC were in vitro stimulated with 10 μM MOG35–55 peptide in HL-1 synthetic medium, and cytokine concentrations were measured in 72-h culture supernatant by ELISA using specific mAb pairs from BD Pharmingen. LNC (3 × 106 cells) were also ex vivo stimulated for 5 h with PMA/ionomycin (50 and 500 ng/ml, respectively) in presence of brefeldin A for the last 2 h. IL-17, IFN-γ, and GM-CSF intracellular production of 2D2 CD45.1+ CD4 T cells was then assessed by flow cytometry.
Isolation of CNS-infiltrating cells
On the indicated day postimmunization, mice were anesthetized and perfused intracardially with ice-cold PBS. Brain and spinal cord were harvested, grinded, and digested for 40 min with collagenase D (1 mg/ml) (Roche, Indianapolis, IN) and DNase I (0.2 mg/ml) (Sigma) in HBSS containing 20 mM hepes under continuous agitation at 37°C. Cells were then separated by centrifugation (1000 rpm for 40 min, 18°C) on a discontinuous isotonic Percoll gradient containing 70 and 37% layers. Mononuclear cells at the 70–37% interface were collected and washed extensively in FACS buffer containing 1% FCS, 2 mM EDTA, and 0.1% NaN3 in PBS.
Statistical analysis
For comparison of EAE clinical scores between groups, repeated measures two-way ANOVA was performed followed by a Bonferroni post hoc test by using GraphPad Prism 4 (GraphPad Software, La Jolla, CA). Otherwise, pairwise comparisons between groups were conducted either using the Mann–Whitney U test or one-way ANOVA test. All graphs show mean and SEM. A p value <0.05 was considered significant. *p < 0.05, **p < 0.01, **p < 0.01, and ****p < 0.0001.
Results
E2 treatment inhibits the development of IFN-γ–, IL-17–, and GM-CSF–producing MOG-specific 2D2 CD4+ T cells in draining lymph nodes
To directly assess the inhibitory effect of E2 on CD4+ T cell priming and differentiation in vivo, we set up an adoptive transfer model in which naive MOG-specific 2D2 Ly5.1+ CD4+ T cells were i.v. injected into Ly5.2 B6 mice treated or not with E2 6 d before. Mice were then immunized with MOG peptide in CFA, and draining lymph nodes were collected 5 d later (Fig. 1A). As shown in Fig. 1B, 2D2 CD4+ T cells were readily detected in LN cells from both control and E2-treated mice with similar frequency and absolute numbers. These data suggested that T cell homing to inflamed LN as well as expansion of MOG-specific T cells were not impaired by E2 treatment. Accordingly, the recall proliferative response of 2D2 CD4+ T cells induced by MOG35–55 in vitro was similar between both groups (Fig. 1C). Lack of effect of E2 treatment on recruitment and proliferation of Ag-specific T cells in LN strikingly contrasted with the strong inhibitory effect of E2 treatment on the development of MOG-specific 2D2 effector Th cells, producing either IL-17 or IFN-γ (Fig. 1D). Unlike inflammatory cytokines, IL-4 production by MOG-specific 2D2 CD4+ T cells was not altered by E2 treatment (Fig. 1E). In agreement with these results, the frequency of 2D2 CD4+ T cells producing either IL-17 or IFN-γ was strongly reduced in E2-treated mice (Fig. 1F). Likewise, E2 treatment also inhibited the development of GM-CSF–producing 2D2 CD4+ T cells (Fig. 1G). Thus, raising serum E2 concentrations in recipient mice can restrict the differentiation of adoptively transferred naive T cells into effector Th1 and Th17, including GM-CSF–producing T cells, while sparing Th2 development.
E2 inhibits early MOG-specific Th1/Th17 cell development in draining lymph nodes through ERα signaling in bystander but not cognate CD4+ T cells
We first investigated whether the E2 inhibitory effect was mediated through ERα signaling in cognate CD4+ T cells. To address this issue, we generated ERα-deficient Ly5.1 2D2 TCR transgenic mice from which we purified naive CD4+ T cells that were used in the adoptive transfer model depicted in Fig. 1A. After priming with MOG in CFA, similar frequency and numbers of ERα+/+ and ERαKO 2D2 CD4+ T cells were recovered from the draining lymph nodes of B6 recipient mice whether they were treated or not with E2 (Fig. 2A). In E2-treated mice, the inhibition of IL-17+ 2D2 T cell development was not dependent on ERα expression in cognate 2D2 T cells (Fig. 2B, 2C), whether 2D2 cells were restimulated ex vivo with PMA/ionomycin (Fig. 2B) or their cognate peptide Ag (Fig. 2C). Similar adoptive transfer experiments were performed using naive OVA-specific OT-2 CD4+ T cells expressing or not expressing ERα (Fig. 2D). As for 2D2 CD4+ T cells, WT or ERαKO IL-17+ OT-2 CD4 T cells developed at a similar frequency in control untreated mice, indicating that at steady-state, ERα signaling had no major impact on Th17 development in this experimental setting (Fig. 2D). Again, treatment of recipient mice with E2 strongly inhibited the development of IL-17+ OT-2 cells irrespective of ERα expression in these cells (Fig. 2D). Altogether, these data show that E2/ERα signaling in cognate CD4+ T cells is dispensable for the inhibition of Th17 cell differentiation.
We next examined the respective contribution of ERα signaling in recipient T lymphocytes or DCs in the E2-mediated inhibition of 2D2 Th1/Th17 cell priming using conditional ERα KO mice lacking ERα expression in T lymphocytes (CD4Cre ERαfl/fl) or in the DC lineage (CD11cCre ERαfl/fl). In agreement with our previous work showing that E2 could efficiently prevent EAE in mice lacking ERα in the DC lineage (10), we found that the priming of adoptively transferred 2D2 CD4+ T cells into IFN-γ–, IL-17–, and GM-CSF–producing CD4+ T cells was still inhibited by E2 in the CD11cCre ERαfl/fl mice as efficiently as in WT mice (Supplemental Fig. 1A, 1B). In striking contrast to the CD11cCre ERαfl/fl mice, the inhibitory effect of E2 on 2D2 Th1/Th17 cell priming was lost in CD4Cre ERαfl/fl mice lacking ERα expression in T lymphocytes (Fig. 2G, 2H). In agreement with previous data, the frequency of 2D2 CD4+ T cells that expanded in draining lymph nodes were similar between ERαfl/fl or CD4Cre ERαfl/fl recipient mice whether they were treated or not with E2 (Fig. 2E). Whereas the recall proliferative response of 2D2 CD4+ T cells stimulated with MOG peptide in vitro was also similar between all groups (Fig. 2F), IL-17 production was strongly reduced in E2-treated ERαfl/fl control mice (Fig. 2G). This inhibitory effect of E2 was lost in CD4Cre ERαfl/fl recipient mice (Fig. 2G). Direct analysis of the frequency of IFN-γ– or IL-17–producing 2D2 CD4+ T cells confirmed the absence of E2-mediated inhibition in recipient mice lacking ERα in T cells (Fig. 2H, Supplemental Fig. 2).
Among T cells, Esr1 mRNA transcripts were expressed at higher level in CD4+ T cell as compared with CD8+ T cells. In CD4+ T cells, Esr-1 gene expression was similar between CD45RBhi CD25− cells and CD45RBlow CD25+ cells (Supplemental Fig. 1C). As expected, ERα transcripts were undetectable in T lymphocytes purified from CD4Cre ERαfl/flmice (Supplemental Fig. 1C). Inhibition of 2D2 T cell priming into IL-17– and IFN-γ–producing cells still persisted in CD8+ T cell–depleted mice, indicating that CD8+ T cells are dispensable for the suppressive effect of E2 on MOG-specific Th1/Th17 differentiation (Supplemental Fig. 1D). Altogether, our results indicated that E2 inhibits early Th1/Th17 cell development in draining lymph nodes through a trans-acting mechanism of suppression, which required ERα signaling in the host CD4+ T cell compartment but not in the cognate MOG-specific 2D2 CD4+ T cell population.
ERα signaling in bystander CD4+ T cells inhibits MOG-specific Th17 cell development and protects from EAE
To directly demonstrate that ERα expression within endogenous host CD4+ T cells was necessary and sufficient for the E2-mediated protection of EAE, we set up an experimental model in which ERαKO Rag2KO mice were replenished with CD4+ T cells purified from either WT or CD4Cre ERαfl/fl C57BL/6 mice together with low numbers of naive ERα−/− 2D2 Ly5.1 CD4+ T cells (Fig. 3A). The following day, mice were immunized with MOG-CFA to induce EAE. In this adoptive transfer model, 2D2 CD4+ T cells, but not the bystander CD4+ T cells, cause EAE (48, 49). Moreover, only the bystander CD4+ T cells expressed ERα and were therefore the only cells able to respond to E2 treatment (Fig. 3A). Upon immunization, strong and rapid EAE developed in mice transferred with ERαKO 2D2 T cells in the absence of E2 treatment, whether the cotransferred bystander CD4+ T cell population expressed ERα or not (Fig. 3B). By contrast, in the presence of E2, disease was significantly less severe in mice injected with WT ERαfl/fl CD4+ T cells as compared with mice that received ERα-deficient CD4+ T cells (Fig. 3C). Disease protection in this group was associated with reduced inflammatory infiltrates in the spinal cord as shown by anti-CD45 staining on spinal cord section by confocal microscopy (Fig. 3D) and FACS analysis of inflammatory leukocytes extracted from the spinal cords (Fig. 3E). Absolute numbers of CNS-infiltrating CD4+ T cells, including pathogenic RORγt+ 2D2 CD45.1+ T cells, were strongly reduced in mice protected from EAE because of the cotransfer of ERα-responsive bystander CD4+ T cells (Fig. 3F).
Similar results were obtained in this adoptive transfer model using low numbers of ERαKO 2D2 T cells (Supplemental Fig. 2). As shown in Supplemental Fig. 2, the absolute numbers of 2D2+ T cells were selectively reduced in WT B6 ERαfl/fl mice but not in mice lacking ERα in T lymphocytes. The MOG35–55–specific production of IL-17, GM-CSF, and IFN-γ was significantly inhibited in E2-treated WT mice but not in E2-treated CD4Cre ERαfl/fl mice (Supplemental Fig. 2). In agreement with the data in Figs. 1, 2, E2 treatment significantly reduced the frequency and absolute numbers of IL-17–producing 2D2 CD4+ T cells in WT B6 ERαfl/fl mice but not in mice lacking ERα in T lymphocytes (Supplemental Fig. 2). This model was then used throughout the rest of the study.
ERα signaling in bystander CD4+ Foxp3neg T cells is necessary and sufficient to inhibit pathogenicity of Th17 cell in trans
As bystander CD4+ T cells contain Foxp3+ Treg cells, we sought to determine whether E2/ERα signaling in this subset was important or not to mediate EAE protection. Therefore, we generated mice selectively lacking ERα in Treg cells using a reported/deleter mouse model (45) as described (Fig. 4A). As expected, genomic Cre-mediated deletion of the targeted Esr1 exon was selectively observed in Treg cells (CD25+CD45RBlow CD4+) from Foxp3YFP-Cre ERαfl/fl mice but not in naive CD4 T cells (CD45RBhigh), CD8 T cells, or B cells (Fig. 4B). Male Foxp3YFPCre ERαfl/fl and female Foxp3YFPCre/YFPCre ERαfl/fl mice were born at a Mendelian frequency and lacked splenomegaly, lymphadenopathy, or other clinical signs of autoimmune pathology up to 14 wk of age. Flow cytometric analyses revealed unaltered frequencies and absolute numbers of thymic and peripheral CD8+ and CD4+ T cells, including Foxp3+ CD4+ T cell subsets in the mesenteric LN or spleen in female Foxp3YFP-Cre/YFP-Cre ERαfl/fl mice (Foxp3YFPCre ERαfl/fl) as compared with Foxp3YFPCre/YFP-Cre ERα+/+ littermate controls (Foxp3YFPCre ERα+/+) (Fig. 4C).
We next examined the effect of E2 treatment on the in vivo priming of 2D2 T cells in recipient Foxp3YFPCre ERαfl/fl mice. Again, although the relative frequency of CD4+ T cells or 2D2 T cells in draining lymph nodes was not affected by E2 treatment (Fig. 5A), a dramatic downregulation in the production of IFN-γ and IL-17 by MOG-stimulated 2D2 CD4+ T cells was observed in E2-treated mice (Fig. 5B). Likewise, E2 treatment significantly reduced the frequency and absolute numbers of IL-17–producing 2D2 CD4+ T cells (Fig. 5C) as well as the development of CCR6+ RORγt+ Th17 cells (Fig. 5D). Thus, ERα signaling in Foxp3+ Treg cells is dispensable for E2-mediated inhibition of Th17 cell priming in lymph nodes.
Using the adoptive transfer model described in Fig. 3A, we then compared the capacity of bystander CD4+ T cells in which ERα was deleted in Foxp3+ Treg cells (Foxp3YFPCre ERαfl/fl) to inhibit EAE induced by 2D2 Ly5.1+ ERαKO CD4+ T cells in the presence of E2. In accordance with our previous experiments, strong EAE developed in mice coinjected with ERα-deficient CD4+ T cells starting at day 9 postimmunization, whereas the other groups that received WT or Foxp3YFPCre ERαfl/fl CD4 T cells were significantly protected (Fig. 5E). Unexpectedly, the protective effect of bystander CD4+ T cells was further enhanced when ERα was deleted in Foxp3+ Treg cells as compared with WT CD4+ T cells (Fig. 5E). In agreement with clinical EAE, the absolute numbers of CD45+ inflammatory infiltrates in the spinal cord cells were strongly reduced in mice injected with Foxp3YFPCre ERαfl/fl CD4+ T cells as compared with E2-unresponsive CD4Cre ERαfl/fl T cells, whereas mice injected with WT CD4 T cells showed an intermediate phenotype (Fig. 5F). By contrast, no disease protection was observed in mice cotransferred with CD4+ T cells from Foxp3YFPCre ERαfl/fl or control Foxp3YFPCre ERα+/+ mice in the absence of E2 supplementation (Fig. 6A). In this experimental setting, inflammatory infiltrates were qualitatively similar in both groups as well as the frequencies of CNS-infiltrating 2D2 CD4 T cells producing IFN-γ, IL-17, and/or GM-CSF (Fig. 6B, 6C). Thus, in absence of sustained E2 supplementation, bystander CD4 T cells do not suppress cognate 2D2 T cell activation and EAE development whether they express ERα in Foxp3+ Treg cells or not (Fig. 6). Altogether, our results demonstrate that estrogen signaling through ERα within bystander Foxp3neg CD4+ T cells is necessary and sufficient to restrict differentiation of encephalitogenic Th17 cells in trans and protect from CNS autoimmunity.
The PD-1/PD-L1 axis is required for E2-mediated Th17 cell inhibition
Previous works have shown that protective action of E2 on EAE was lost in mice lacking either PD-1 (9) or PD-L1 (39); we therefore investigated whether a functional PD-1/PD-L1 axis was required for the E2-mediated inhibition of Th17 priming. E2 treatment strongly decreased total cell numbers in draining lymph nodes, including 2D2 CD4+ T cells (Fig. 7A). This anti-inflammatory effect of E2 was lost in recipient mice treated with PD-L1 blocking mAb (Fig. 7A). In these mice, the frequency and absolute numbers of IL-17–producing 2D2 CD4+ T cells are similar to the values obtained in control untreated mice (Fig. 7B). Although, we still noticed a trend toward an inhibitory action of E2 in anti–PD-L1 treated mice; this was NS (Fig. 7B). As E2-dependent upregulation of PD-1 on CD4+ T cells has been previously reported (50), we next determined whether expression of PD-1 on bystander CD4+ T cells was involved in the trans inhibition of Th17 cell priming. Chimeric mice reconstituted with PD-1–deficient bone marrow cells were used as recipients. The frequency of Foxp3+ Treg cells was similar in draining lymph nodes from WT and PD-1–deficient bone marrow chimeras. In both groups, E2 treatment did not upregulate the frequency of Foxp3+ Treg cells, whether they express or not PD-1 molecules (Supplemental Fig. 3A, 3B). As expected, we noticed a strong upregulation of PD-1 on 2D2 CD4+ T cells by day 6 postimmunization as a consequence of MOG-driven T cell activation in situ (Supplemental Fig. 3C). No differences were observed between control and E2-treated mice, indicating that MOG-driven 2D2 CD4+ T cell activation was not affected by estrogen signaling. By contrast, the frequency (Fig. 7C, 7D) and absolute numbers (Fig. 7E) of IL-17–producing 2D2 CD4+ T cells were significantly inhibited in both E2-treated WT and PD-1KO chimeras.
As PD-1/PD-L1 pathway has been suspected to play a role in E2-mediated EAE protection (9), we next investigated whether PD-1 expressed on activated 2D2 CD4+ T cells was required for the inhibition of the Th17 response. PD-1–deficient naive 2D2 CD4+ T cells (47) were adoptively transferred in mice treated or not with E2, and the development of IL-17–producing 2D2 T cells was monitored in draining lymph nodes. Contrasting with our results with PD-1–sufficient 2D2 cells, the frequency of IL-17+ cognate T cells was not affected by E2 treatment when 2D2 cells were lacking PD-1 (Fig. 7F, 7G). To establish further the cell-intrinsic role of PD-1 on 2D2 CD4+ T cells, we performed adoptive cotransfer experiments of naive CD4+ T cells from either Ly5.1 WT and Thy1.1 PD-1KO 2D2 mice into Ly5.2 B6 mice treated or not with E2 (Fig. 8). Mice were immunized with MOG-peptide in CFA, and the draining lymph nodes were collected at day 6. LNC were stimulated with PMA-ionomycin, and the frequency of IL-17–producing cells among CD45.1+ WT and Thy1.1+ PD-1KO 2D2 T cells were assessed as shown in Fig. 8A. We observed a reduction in the relative frequency of WT 2D2 T cells in E2-treated mice relative to PD-1–deficient 2D2 T cells (Fig. 8B). This was associated with a selective inhibition in the frequency of IL-17–producing cells among the PD-1–sufficient 2D2 T cells (Fig. 8C). Indeed, the WT/PD-1KO Th17 cell ratio was significantly decreased in mice treated with E2 compared with untreated controls (Fig. 8D). These results are consistent with the conclusion that within the same individual mice, E2 treatment selectively inhibited the Th17 differentiation and expansion of PD-1–sufficient 2D2 T cells, whereas PD-1KO cells were largely unaffected.
Together, these data show that cell-intrinsic PD-1 signaling within cognate 2D2 T cells, but neither within endogenous CD4+ T cells nor Treg cells, is required for the inhibition of Th17 cell priming by E2-sensitized bystander CD4+ T cells.
Discussion
We reported previously a requirement for ERα expression in hematopoietic cells, namely T cells rather than DCs or myeloid cells, in the anti-inflammatory action of E2 administration (10). This suppressive effect, characterized by inhibition of Th1/Th17 cell priming in inflammatory lymph nodes, was elicited at steady-state serum concentrations of E2 congruent with pregnancy values. We describe in this paper an indirect path through ERα signaling in noncognate conventional CD4+ T cells as the main mechanism by which estrogens limit Th1/Th17 cell differentiation of cognate CD4+ T cells in trans and protect from EAE. We show that ERα activation in MOG-specific CD4+ T cells is totally dispensable to this action. This mechanism appears to selectively limit acquisition of Th17 effector functions in cognate CD4+ T cells but not their potential for Th2 development nor their proliferative response upon subsequent ex vivo challenge. In vivo experiments on Rag2KOERαKO mice showed that estrogen-sensitized ERα-proficient CD4+ T cells, unlike their ERα-deficient counterparts, inhibited the development of Th17 cell–mediated EAE when coadministered with naive ERα-deficient 2D2 T cells. Bystander CD4+ T cells from mice lacking Esr1 gene specifically in Treg cells were still highly proficient in their capacity to inhibit Th17 cell differentiation and EAE development. Actually, we consistently observed an enhanced capacity of these cells to suppress EAE as compared with WT CD4+ T cells. These results might suggest that in the context of inflammation, ERα signaling in Treg cells could negatively impact on Treg cell stability or regulatory potential. Indeed, a recent study has suggested that ERα may suppress Foxp3 expression and iTreg development both in vivo and in vitro (51). Altogether, our data bring to light an important suppressive property of estrogen by which ERα signaling in bystander Foxp3neg CD4+ T cells promotes their capacity to restrict the differentiation of cognate autoantigen-specific Th17 cells in trans, thereby protecting from CNS autoimmunity.
The notion that bystander noncognate Foxp3neg CD4+ T cells can drastically influence the differentiation of the Ag-specific Th17 cells is not without precedent (48, 52). In an elegant study, Hirahara and colleagues (48) demonstrated that IL-27–primed nonspecific bystander T cells could limit Th17 cell differentiation of Ag-specific effector cells in trans. They provided evidence that APC-derived IL-27 acted on conventional naive CD4+ T cells to promote Stat-1–dependent PD-L1 expression. IL-27–primed noncognate PD-L1+ CD4+ T cells were able to restrict Th17 cell development and to protect from EAE through PD-1 expressed on cognate T cells (48). More recently, it was reported that type I IFN signaling in T cells could promote the development of FoxA1 Treg cells able to suppress autoreactive T cells in the CNS, again through PD-1/PD-L1 interactions (52). The PD-1/PD-L1 axis has also been suspected to play a role in E2-mediated EAE protection. Indeed, previous works have shown that protective action of E2 on EAE was lost in mice lacking either PD-1 (9) or PD-L1 (39). It was initially suggested that the anti-inflammatory effects of E2 were mediated through its capacity to drive expansion of Treg cells in vivo (37) and to enhance PD-1 expression on various cell types, including APCs (53) or Foxp3+ Treg cells (54). Our present results invalidate these assumptions and demonstrate that PD-1 expression on cognate CD4+ T cells, but not on bystander CD4+ T cells nor APCs, is required for the inhibition of Th17 differentiation by E2. Indeed, in cotransfer experiments we clearly showed that only PD-1–sufficient 2D2 T cells were amenable to E2-dependent inhibition of Th17 development, whereas PD-1KO 2D2 Th17 cells that developed in the same inflammatory lymph nodes were not affected (see Fig. 8). Thus, induction of Treg cells through PD-1 engagement of cognate 2D2 T cells is therefore unlikely to contribute to the inhibitory effect of E2. These results, rather, support the conclusion that the restriction of Th17 cell development by E2-primed bystander CD4+ T cells requires cell-intrinsic PD-1 signaling within cognate 2D2 T cells. Indeed, PD-1/PD-L1 interactions have been shown to impair Th17 differentiation in mouse models (48) and in patients with STAT3 loss- and STAT1 gain-of-function mutations (55). Whether PD-1 on MOG-specific 2D2 T cells interacts with PD-L1 expressing bystander CD4+ T cells (48, 52) or others, PD-L1 expressing cells in draining lymph nodes to inhibit encephalitogenic Th17 cell development will deserve further investigations.
In conclusion, our data now provide direct evidence that bystander Foxp3neg CD4+ T lymphocytes are the main ERα-expressing target cells orchestrating the anti-inflammatory effects of E2 in EAE. This involves the inhibition in trans of cognate Th17 cell development in inflammatory lymph nodes through a mechanism that is dependent on PD-1 expression on cognate but not bystander CD4+ T cells. Harnessing ERα signaling in Foxp3neg CD4+ T cells may therefore represent a valuable therapeutic approach to limit inflammation in multiple sclerosis. ER modulation strategy could be possibly combined with regulatory cytokines, such as IFN-β, to promote synergistic or additive anti-inflammatory effects on multiple sclerosis, as it has been reported in a recent clinical trial (14). Delineating further the molecular mechanisms by which ERα signals in Foxp3neg CD4+ T cells may help to optimize ER modulation strategies, to limit the deleterious impact of estrogen on reproductive tissues, and to improve the protective action of estrogens in CNS autoimmunity in an effort to mimic the beneficial effect of pregnancy in women with multiple sclerosis.
Acknowledgements
Foxp3-YFP-Cre and 2D2 mice were generously provided by Dr. A. Rudensky (Sloan-Kettering Institute for Cancer Research, New York, NY) and Dr. V. Kuchroo (Brigham and Women’s Hospital, Boston, MA), respectively. We thank C. Cenac and A. Gouazé for technical assistance. We also thank the staff of the flow cytometry facility of the Center of Pathophysiology Toulouse Purpan (F. L’Faqihi and V. Duplan) and the animal house staff members (INSERM UMS06).
Footnotes
This work was supported by the Association pour la Recherche sur la Sclérose en Plaques (ARSEP), the Conseil Régional Midi-Pyrénées, and the Fondation Recherche Médicale (Équipe Labellisée DEQ2000329169). S.L. and L.G. received fellowships from ARSEP.
The online version of this article contains supplemental material.
References
Disclosures
The authors have no financial conflicts of interest.