Increased GITRL Impairs the Function of Myeloid-Derived Suppressor Cells and Exacerbates Primary Sjögren Syndrome

Primary Sjögren syndrome (pSS), a systemic autoimmune disease, is characterized by lymphocytic infiltration of the exocrine glands, primarily salivary and lachrymal glands, leading to the loss of secretary function (1, 2). Although the cause of pSS remains largely unclear, many studies have demonstrated the involvement of dendritic cells, NK cells, B cells, and T cells in the pathogenesis of SS (2–5). Recent studies have found that Th1 and Th17 immune responses play a critical role in the development of pSS (3, 6–9).

Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of immature myeloid cells with immunosuppressive functions (10). Murine MDSCs are characterized by coexpression of CD11b and Gr-1, which can be further subdivided into CD11b⁺Ly-6G⁻Ly-6C^high monocytic MDSCs (M-MDSCs) and CD11b⁺Ly-6G⁺Ly-6C^low polymorphonuclear MDSCs (PMN-MDSCs). Human MDSCs are CD11b⁺CD33⁺HLA-DR⁻ cells. MDSCs have been shown to promote tumor progression by suppressing T cell–mediated antitumor immunity in cancer (11, 12). Recently, MDSCs are shown to be involved in the pathogenesis of various autoimmune disorders, including type 1 diabetes (13), multiple sclerosis (MS) (14–16), rheumatoid arthritis (RA) (17, 18), and systemic lupus erythematosus (19–21), but the role of MDSCs in the development of pSS has remained poorly understood. Glucocorticoid-induced TNFR family–related protein (GITR) is a type I transmembrane protein of TNF superfamily that is highly expressed on regulatory T cells, has low levels on conventional effector T cells, and is rapidly upregulated after activation (22). The ligand for GITR (GITRL) is a type II transmembrane protein predominantly expressed on endothelial cells, dendritic cells, macrophages, and B cells (23). The interaction of GITR and GITRL has been demonstrated to modulate both innate and adaptive immune responses (24). The engagement of GITR on effector T cells exhibits a positive costimulatory signal leading to T cell proliferation and cytokine production (25). However, GITR stimulation on regulatory T cells has been suggested to abolish the suppressive capacity of the cells (26, 27). Previous studies, including our recent findings, have revealed that GITR/GITRL pathway is involved in the pathogenesis of autoimmune diseases, including RA (28, 29), MS (30), and autoimmune diabetes (31). Recently, GITRL is reported to be closely associated with the disease severity in pSS patients and MRL-Fas^lpr mice (32, 33). However, the underlying mechanism for GITRL in regulating the progression of pSS remains largely unclear.

In this study, we characterized the functional changes of MDSCs in experimental SS (ESS) mice and pSS patients and identified a critical role of GITRL in regulating the suppressive function of MDSCs in the pathogenesis of pSS.

Female C57BL/6 mice at 8 wk old were purchased from Experimental Animal Center of Yangzhou University. Mice were housed in a specific pathogen-free animal facility, and all the experiments were approved by the Committee on the Use of Live Animals in Research and Teaching of Jiangsu University.

The ESS mouse model was induced as we previously described (8). Briefly, bilateral salivary glands (SG) were isolated from female C57BL/6 mice for homogenization in PBS to prepare SG proteins. Naive mice were immunized with SG proteins emulsified in an equal volume of CFA (Sigma-Aldrich) to a concentration of 2 mg/ml (100 μl each mouse) s.c. on the neck on days 0 and 7. On day 14, the booster injection was performed with a dosage of 1 mg/ml SG proteins emulsified in Freund Incomplete Adjuvant (Sigma-Aldrich). Naive mice immunized with adjuvant alone served as adjuvant controls.

Saliva flow rates were measured as previously described (8). In brief, mice were anesthetized and injected i.p. with pilocarpine (Sigma-Aldrich) at a dosage of 5 mg/kg body weight. Saliva was then collected using a 20-μl pipet tip from the oral cavity for 15 min.

CD11b⁺Gr-1⁺ MDSCs were isolated from the spleens of ESS mice using a FACSAria II SORP (Becton Dickinson) cell sorter. M-MDSCs and PMN-MDSCs were isolated using a Mouse MDSC Isolation Kit (Miltenyi Biotec) following the manufacturer’s protocol. MDSC suppression assay was performed as we previously described (34).

Recombinant mouse GITRL protein was produced and purified as previously described (29, 35). The control protein was prepared with the same protocol except with no mouse GITRL insert in the vector. Endotoxin was removed by ToxinEraser endotoxin removal resin (GenScript) and measured by E-TOXATE Kits (Sigma-Aldrich). The endotoxin levels of both GITRL and control are <0.1 endotoxin units/ml. The eluted proteins were kept in sterile PBS. The recombinant human GITRL (hGITRL) protein was obtained from R&D Systems.

Single-cell suspensions were initially blocked with Fc Block (BD Biosciences) at 4°C for 10 min, and surface/intracellular markers were stained with relevant fluorochrome-conjugated mAbs: anti-mouse CD40, CD80, CD86, MHC class II (MHC-II), GITR, GITRL, F4/80, and CD19 from eBioscience; anti-mouse CD11b, Gr-1, Ly6G, and Ly6C from BioLegend; anti-mouse arginase-1 from R&D Systems; and anti-human CD11b, HLA-DR, CD33, GITR, and arginase-1 from BioLegend. For intracellular staining, cells were stimulated with PMA (50 ng/ml; Sigma-Aldrich), ionomycin (1 μg/ml; Enzo Life Sciences), monensin (2 μg/ml; Enzo Life Sciences). After 5 h, cells were stained with anti-CD4 mAb (eBioscience), fixed, permeabilized, and stained with anti–IFN-γ mAb or anti–IL-17 mAb (eBioscience) according to the Intracellular Staining Kit (Invitrogen) instructions. Flow cytometry was performed using FACSCalibur flow cytometer (Becton Dickinson), and data were analyzed using FlowJo software.

The quantitative real-time PCR were performed as previously described (36). The sequences for the primers used are as follows: mouse GITR, forward -5′-CTCAGGAGAAGCACTATGGG’-3′, reverse-5′-AGCTGGGCAAGTCTTGTAG’-3′; and β-actin, forward -5′-TGGAATCCTGTGGCATCCATGAAAC-3′, reverse-5′-TAAAACGCAGCTCAGTAACAGTCCG-3′. β-Actin was used as an internal control.

Autoantibodies against SG proteins, ANA (Elabscience), anti–M3 muscarinic receptor (M3R) Abs, IFN-γ and IL-17 (eBioscience) in the serum of ESS mice, and hGITRL in serum (R&D Systems) were assessed by ELISA.

The activity of arginase and NO concentration was measured as previously described (37).

GITR short interfering RNA (siGITR), and negative controls were synthesized by RiboBio. Oligonucleotide transfection was performed with Entranster-R (Engreen Biosystem) according to the manufacturer’s instructions.

SG tissues were paraffin embedded, sectioned, and stained with H&E. For immunofluorescence microscopy, frozen sections were incubated with Alexa Fluor 488–conjugated anti-CD11b mAb (BioLegend), biotin anti–Gr-1 mAb (BioLegend) and streptavidin Alexa Fluor 594 conjugate (Invitrogen). Hoechst 33342 (Beyotime) was used to label nuclei. Matched isotype controls (eBioscience) were used for control staining.

Whole-blood samples were collected from patients with pSS (n = 25) and healthy controls (n = 25). All patients were diagnosed according to the American–European Consensus Group criteria (3). The demographic and clinical characteristics of the patients and controls are shown in Supplemental Table I. The study protocols and consent forms were approved by the Institutional Medical Ethics Review Board of Jiangsu University.

PBMCs were isolated from healthy volunteer donors by density gradient centrifugation using Ficoll-Hypaque solution (TBD Science). PBMCs were cultured with GM-CSF (10 ng/ml; PeproTech) and IL-6 (10 ng/ml; PeproTech) for 7 d, and the cytokines were refreshed every 2–3 d. After 1 wk, all cells were collected, and adherent cells were removed using nonprotease cell detachment solution Detachin (Genlantis). CD33⁺ cells were isolated using anti-CD33 magnetic MicroBeads (Miltenyi Biotec). The purity of isolated cells was >90%.

Human CD4⁺ T cells were isolated from healthy volunteer donors using anti-CD4 MicroBeads (Miltenyi Biotec). CD4⁺ T cells were labeled with CFSE (5 μM; Invitrogen) and cocultured with human CD33⁺ cells at a ratio of 2:1 in 96-well plates (Costar) in the presence of anti-CD3 (eBioscience) and anti-CD28 (eBioscience) mAbs for 3 d. CFSE fluorescence intensity was analyzed to determine the proliferation of CD4⁺ T cells by flow cytometry.

The statistical significance was determined by the Student t test or one-way ANOVA. Correlations were determined by a Spearman correlation coefficient. All analyses were performed using SPSS 16.0 software. Any p values <0.05 were considered statistically significant.

In ESS mice, the kinetic changes of CD11b⁺Gr-1⁺ MDSCs were analyzed in the spleen, cervical lymph nodes (CLN), and peripheral blood during the disease progression. Compared with the naive mice (day 0) or control mice immunized with adjuvant alone, the frequencies of MDSCs remarkably increased and reached a peak in ESS mice around day 14 or 35 postimmunization (Fig. 1A–C). Flow cytometric analysis identified a population of MDSCs infiltrated in SG 10 wk after immunization (Fig. 1D), which was further confirmed by immunofluorescence microscopy (Fig. 1E). Similarly, both the number and the proportion of two subsets, CD11b⁺Ly-6G⁻Ly-6C^high M-MDSCs and CD11b⁺Ly-6G⁺Ly-6C^low PMN-MDSCs were also increased during the course of ESS development (Fig. 1F, 1G). In addition, mice treated with adjuvant alone showed only a slight increase of MDSCs at the early stage (Fig. 1A; Supplemental Fig. 1).

We next examined both phenotypic and functional features of the increased MDSCs; the expression of CD40, CD80, CD86, and MHC-II on MDSCs and their suppressive capacity on T cell proliferation were analyzed. As shown in Fig. 2A, MDSCs (M-MDSCs/PMN-MDSCs) isolated from the early stage of ESS (day 18) expressed low levels of CD40, CD80, CD86, and MHC-II, which displayed an immature phenotype. However, the expression of these molecules was gradually upregulated along the disease progression. MDSCs from the late stage of ESS (10 wk) expressed high levels of these surface markers, which exhibited a more differentiated mature phenotype. Coculture of MDSCs (M-MDSCs/PMN-MDSCs) and T cells showed that early-stage MDSCs displayed potent activity in suppressing T cell proliferation, whereas the late-stage cells almost lost their suppressive capacity (Fig. 2B). Similarly, the suppressive factors secreted by MDSCs, including arginase and NO, were also downregulated during the development of ESS (Fig. 2C). Furthermore, MDSCs infiltrated in SG also produced little arginase, even slightly less than the late-stage MDSCs in spleens (Fig. 2D). Together, these results suggest that the suppressive capacity of MDSCs was downregulated during the progression of ESS.

We further examined the suppressive function of MDSCs in vivo. Purified early-stage or late-stage MDSCs were adoptively transferred to ESS mice on days 18 and 25 (Fig. 3A). Remarkably, early-stage MDSCs, which possessed strong suppressive capacity, significantly reversed the saliva flow rate to normal level, whereas the late-stage cells did not show any obvious effect (Fig. 3B). Furthermore, levels of autoantibodies against total SG Ags, ANA, and anti-M3R Abs were also decreased after early-stage MDSC treatment (Fig. 3C). Notably, the early-stage MDSC treated group displayed smaller CLN and SG when compared with the control group (Fig. 3D). Histological examination further identified the pathological changes in local submandibular glands. Only little lymphocytic infiltration was observed in SG after early-stage MDSC treatment. In contrast, the late-stage MDSC treatment did not show any therapeutic effect with pronounced tissue destruction and multiple lymphocytic foci detected in SG (Fig. 3E). In ESS mice, both Th1 and Th17 responses were also reduced after the treatment of early-stage MDSCs, whereas the late-stage MDSC treatment did not result in any change in T cell responses (Fig. 3F, 3G). Similar changes were also observed in the serum levels of IFN-γ and IL-17 (Fig. 3H). Taken together, these findings reveal that early-stage MDSCs could alleviate the progression of ESS and suppress Th1/Th17 responses, but late-stage MDSCs possess no immunosuppressive function.

To examine the potential factors responsible for the altered function of MDSCs, we found that GITR was expressed on total MDSCs and two subpopulations (Fig. 4A). Moreover, the level of GITR expression was upregulated during the disease development in ESS mice (Fig. 4B). Notably, GITRL expression was also gradually increased in the spleen, CLN, and SG along the disease progression (Fig. 4C). To further ascertain the GITRL-expressing cell populations, we detected increased GITRL expression on macrophages and B cells in ESS mice (Fig. 4D, 4E). To investigate whether the function of MDSCs could be regulated by the increased GITRL via GITR/GITRL pathway, we used recombinant GITRL, a GITR agonist, to trigger GITR signaling on early-stage MDSCs isolated from ESS mice. After GITRL treatment, levels of CD40, CD80, CD86, and MHC-II on total MDSCs, M-MDSCs, and PMN-MDSCs were enhanced (Fig. 4F), and the suppressive function of MDSCs (M-MDSCs and PMN-MDSCs) on CD4⁺ T cells was decreased (Fig. 4G). Similarly, reduced arginase and NO was observed after GITRL stimulation (Fig. 4H). All these results suggest GITRL could promote the maturation of MDSCs and downregulate the suppressive capability of the cells.

We next investigated the effect of GITRL-treated MDSCs in inhibiting the ESS development. Early-stage MDSCs treated with or without GITRL were adoptively transferred to ESS mice on days 18 and 25, respectively (Supplemental Fig. 2A). When compared with the control MDSC (Ctrl-MDSC) group, saliva flow rate was significantly decreased, and the levels of autoantibodies were enhanced after transferring with GITRL-treated MDSCs (Supplemental Fig. 2B–E). In addition, GITRL-MDSC treatment showed bigger CLN and SG, and more lymphocytes were detected in local submandibular glands (Supplemental Fig. 2F, 2G). Furthermore, GITRL-MDSC treatment did not reduce Th1 and Th17 responses, whereas the Ctrl-MDSCs significantly inhibit the immune responses (Supplemental Fig. 2H–J). Together, these findings suggest that GITRL may contribute to the reduced immunosuppressive function of MDSCs during ESS development.

Exogenous GITRL was administrated into the ESS mice starting from day 18 for three times (Fig. 5A), and the GITRL treatment significantly exacerbated the disease development. The reduced saliva flow rate, enhanced autoantibodies, and pronounced lymphocytic infiltration in submandibular glands were observed (Fig. 5B–E). As shown in Fig. 5F, the expression of costimulatory molecules and MHC-II on MDSCs (M-MDSCs/PMN-MDSCs) was increased at different levels, and the suppressive capacity of MDSCs in ESS mice was markedly reduced after GITRL administration (Fig. 5G, 5H).

To further determine whether the altered function of MDSCs in vivo was mediated via the engagement of GITR, we adoptively transferred MDSCs with silenced GITR gene expression into ESS mice on day 32 (Fig. 6A). As expected, the therapeutic effect of MDSCs in inhibiting ESS progression was minimal. However, MDSCs with silenced GITR gene showed potent therapeutic effect in ESS mice. Notably, the saliva flow rate was recovered to normal level and the serum autoantibodies were significantly reduced in siGITR-MDSCs–treated ESS mice (Fig. 6B–E). When compared with the Ctrl-MDSCs and ESS group, the sizes of CLN and SG were smaller, with significantly fewer lymphocytes infiltrated in submandibular glands in siGITR-MDSC group (Fig. 6F, 6G). Furthermore, flow cytometric analysis detected markedly reduced Th1 and Th17 responses in ESS mice treated with siGITR-MDSCs (Fig. 6H–J). Together, these data suggest that the GITR/GITRL pathway is involved in modulating the suppressive capacity of MDSCs.

Flow cytometric analysis was performed to examine the frequencies of CD11b⁺CD33⁺HLA-DR⁻ MDSCs in peripheral blood from pSS patients and healthy controls. Compared with healthy controls, the percentage of MDSCs was significantly increased in patients with pSS (Fig. 7A). A further analysis showed that these expanded MDSCs in pSS patients secreted low levels of arginase (Fig. 7B), and the level of arginase was negatively correlated with the disease activity (Fig. 7C). In pSS patients, both the serum level of GITRL and GITR expression on MDSCs were significantly increased and displayed positive correlations with the disease activity of pSS (Fig. 7D–G). Moreover, the arginase level in MDSCs from pSS patients was negatively correlated with GITRL and GITR (Fig. 7H, 7I). To further determine a role of GITRL in regulating the function of human MDSCs, MDSCs induced from healthy donors were treated with recombinant hGITRL. Similar to the findings from mice, the suppressive capacity of MDSCs on CD4⁺ T cells was significantly decreased after GITRL treatment (Fig. 7J), whereas the arginase level in MDSCs was also reduced (Fig. 7K). Together, these data suggest that the suppressive function of MDSCs in pSS patients is also regulated by GITR/GITRL pathway.

Recent studies on the immune suppressive function of MDSCs have generated significant interest in exploring their potential in the treatment of autoimmune diseases (14, 15, 17, 38, 39). In this study, we found that frequencies of MDSCs and their subsets were increased, but MDSCs gradually lost their suppressive function during the development of ESS, thus leading to markedly enhanced Th1 and Th17 responses. Moreover, we demonstrated that increased GITRL attenuated the suppressive function of MDSCs via activating GITR/GITRL pathway. Taken together, our findings have revealed that the increased GITRL could promote the maturation of MDSCs and downregulate the suppressive capacity of these cells, contributing to the exacerbated progression of pSS.

Extensive studies have characterized the function of MDSCs in the development of autoimmune diseases. The expansion of MDSCs has been reported in several autoimmune diseases such as MS, RA, type I diabetes, autoimmune uveoretinitis, and systemic lupus erythematosus (13–18, 39). Interestingly, the contrasting results on the functions of MDSCs in different autoimmune diseases have been obtained. In experimental autoimmune encephalomyelitis (EAE) mice, Ioannou et al. (15) found that only granulocytic MDSCs (G-MDSCs) subpopulation was increased, whereas adoptive transfer of G-MDSCs isolated from EAE mice significantly suppressed the development of EAE and inhibited Th1 and Th17 immune responses. However, Yi et al. (16) demonstrated that M-MDSCs in EAE could promote the differentiation of Th17, whereas depletion of MDSCs markedly reduced Th17 response and alleviated the severity of EAE. Similarly, the conflicting results were also observed in the autoimmune arthritis. Fujii et al. (17) identified a protective role of G-MDSCs in collagen-induced arthritis mice, whereas M-MDSCs were found to promote Th17 cell polarization and exacerbated the disease progression in collagen-induced arthritis mice (18). Because of their extraordinary heterogeneity and plasticity in phenotypes and functions, MDSCs at different stages of disease may play a differential role in either promoting or alleviating disease progression. In this study, we found that early-stage MDSCs displayed strong suppressive capacity, whereas the late-stage MDSCs almost lost their inhibitory effects on T cells. Moreover, the late-stage MDSCs expressed increased levels of CD40, CD80, CD86, and MHC-II, which exhibited a more mature phenotype of myeloid cells. Consistent with recent findings by Qi et al. (40), we also observed increased MDSCs in both pSS patients and ESS mice. Interestingly, Qi et al. (40) found that transfer of splenic MDSCs from NOD mice aggravated SS-like symptoms in NOD mice, whereas depletion of MDSCs alleviated SS progression in NOD mice, suggesting a proinflammatory role of MDSCs in SS progression. However, we showed that adoptive transfer of MDSCs purified from ESS mice at early stage significantly alleviated disease severity in ESS mice (Fig. 3). These seemingly contrasting findings might be possibly due to different mouse models used for assessing SS development and MDSCs in different disease stages. Because NOD mice develop both diabetes and SS-like symptoms, our current studies have used a well-established ESS model induced in normal mice with no diabetic involvement (8). Intriguingly, CD11b⁺Gr-1⁺ cells infiltrated in SG expressed low level of arginase, suggesting that MDSCs in local tissues may not possess immunosuppressive function, but probably promote local inflammation. In this study, we observed the kinetic changes of MDSCs in different tissues or organs during ESS development. Although our data suggested the possible involvement of MDSCs from CLN and SG in disease pathology, it is currently unclear about the relative importance of these MDSCs compared with splenic MDSCs in the pathogenesis of ESS. Because the total number of MDSCs in the spleen is much larger than those in draining lymph mode and SG, it is likely that splenic MDSCs may exert more potent function in suppressing systemic immune responses possibly by inhibiting the function of effector T cells and B cells. However, it remains to be investigated how MDSCs in draining lymph mode and SG are modulated by local microenvironment and lose their immunosuppressive function during SS pathogenesis, a notion that is supported by recent evidence on the functional differences of MDSCs in peripheral lymphoid organs and MDSCs isolated from the tumor tissue (10). Additionally, we show that adoptive transfer of early-stage MDSCs to ESS mice has significantly alleviated the severity of ESS, indicating the therapeutic effect of MDSCs on the disease. However, transfer of late-stage MDSCs showed no therapeutic effect to alleviate the disease in ESS mice, in which more pronounced lymphocytic infiltration and enhanced Th1/Th17 immune responses were observed. Together, these data suggest that the local microenvironment at different disease stages may modulate the functional status of MDSCs, which will render these cells with either proinflammatory or immunosuppressive properties. Therefore, to explore the potential effective MDSC-based therapies, it is important for the maintenance of their immunosuppressive function in vivo.

The conflicting results on the role of MDSCs in preventing autoimmune diseases are mainly due to the high heterogeneity and plasticity of myeloid cells whose phenotypes and functions are largely dependent on local microenvironments (10, 41–44). In this study, we found for the first time, to our knowledge, that GITR was expressed on MDSCs in both ESS mice and pSS patients. Moreover, the expression of GITR on MDSCs was significantly increased during the progression of the disease. Consistent with previous findings, we found that increased GITRL was closely associated with the disease activity in pSS patients (32). To determine whether GITRL can modulate the function of MDSCs, we showed that the recombinant GITRL protein promoted MDSCs to differentiate into a mature phenotype and significantly attenuated the suppressive effect of MDSCs both in vitro and in vivo. Although exogenous GITRL treatment could modulate the function of MDSCs both in vitro and in vivo, the effect of GITRL on the progression of ESS might also be possibly attributed to its regulation on effector T cells because GITRL can also activate the effector T cells and induce their proliferation and cytokine production by acting as a costimulatory factor for T cells (35, 45). Concurrently, GITRL could impair the suppressive function of MDSCs in humans. Moreover, the early-stage MDSCs modified by GITRL lost their primary capacity to alleviate the ESS development. In the inflammatory environment with high level of GITRL, MDSCs interfered with GITR exhibited efficient therapeutic effect in vivo, indicating that the GITR/GITRL pathway is predominantly involved in the regulation of MDSCs. Additionally, consistent with the functional features of late-stage MDSCs, adoptive transfer of GITRL-treated MDSCs could also lead to moderately enhanced Th1 and Th17 cell responses when compared with the ESS mice, suggesting their potential proinflammatory effect. Together, these data suggest that MDSCs, as cells with high heterogeneity and plasticity, cannot only lose their suppressive capacity after encountering inflammatory environment, but also convert into pathogenic cells that might accelerate the inflammatory process in autoimmune diseases. To abrogate GITRL-mediated functional change of MDSCs in vivo, we generated MDSCs with silenced GITR expression and found that adoptive transfer of these MDSCs led to markedly inhibited Th1/Th17 responses and exhibited profound therapeutic effect on suppressing disease progression in ESS mice. In addition, we also observed significantly reduced autoantibody levels. Further studies are needed to determine whether MDSCs can directly regulate B cell response and Ab production during pSS development. Moreover, it remains to be investigated whether MDSCs with silenced GITR expression display an anti-inflammatory cytokine profile, which may facilitate their potential application for treating autoimmune diseases as novel tolerogenic cell therapies.

In conclusion, our findings demonstrate that increased GITRL modulates the phenotype and function of MDSCs, thus promoting the progression of pSS. Therefore, blocking GITR/GITRL pathway may facilitate the development of effective MDSC-based therapies for treating patients with pSS or other autoimmune diseases.

The authors have no financial conflicts of interest.