Anti-inflammatory Roles of Glucocorticoids Are Mediated by Foxp3⁺ Regulatory T Cells via a miR-342-Dependent Mechanism

SUMMARY

Glucocorticoids (GC) are the mainstay treatment option for inflammatory conditions. Despite the broad usage of GC, the mechanisms by which GC exerts its effects remain elusive. Here, utilizing murine autoimmune and allergic inflammation models, we report that Foxp3⁺ regulatory T (Treg) cells are irreplaceable GC target cells in vivo. Dexamethasone (Dex) administered in the absence of Treg cells s completely lost its ability to control inflammation, and the lack of glucocorticoid receptor in Treg cells alone resulted in the loss of therapeutic ability of Dex. Mechanistically, Dex induced miR-342–3p specifically in Treg cells and miR-342–3p directly targeted the mTORC2 component, Rictor. Altering miRNA-342–3p or Rictor expression in Treg cells dysregulated metabolic programming in Treg cells, controlling their regulatory functions in vivo. Our results uncover a previously unknown contribution of Treg cells during glucocorticoid-mediated treatment of inflammation and the underlying mechanisms operated via the Dex-miR-342-Rictor axis.

In Brief

Glucocorticoids are the first line treatment option for acute and chronic inflammatory conditions, yet the underlying mechanisms remain largely elusive. Kim et al demonstrate that Foxp3⁺ regulatory T cells are key mediators of glucocorticoid-induced treatment by miR-342-dependent metabolic control in Treg cells.

Graphical Abstract

INTRODUCTION

Glucocorticoids (GC) are a class of cholesterol-derived corticosteroid hormones playing highly diverse roles in homeostasis, development, metabolism, and immunity (Barnes, 2011; Cain and Cidlowski, 2017). Due to potent immunosuppressive and anti-inflammatory properties, synthetic GC such as dexamethasone (Dex) serves as the frontline treatment option for various inflammatory conditions including multiple sclerosis and asthma (Alangari, 2014; Barnes et al., 1997). GC exerts its functions by binding the cytosolic glucocorticoid receptor (GR) expressed in almost every nucleated cell (Necela and Cidlowski, 2004; Oakley and Cidlowski, 2013). The cytosolic GR, which remains inactive, undergoes conformational change upon ligand binding, translocates into the nucleus, where it binds the glucocorticoid-responsive DNA elements, and acts as transcriptional activators or repressors (Barnes, 2006; Necela and Cidlowski, 2004). Despite the long history of GC discovery and broad applications to attenuate inflammatory responses, the precise cellular mechanisms underlying GC-induced immune modulation are not entirely understood.

The GR is ubiquitously expressed. Yet, there is emerging evidence that the target cells mediating the GC effects differ depending on the types or tissues of inflammation (Cain and Cidlowski, 2017; Franco et al., 2019). Utilizing cell-type-specific GR-deficient animal models, it has been shown that conventional T cells are the primary GC target cells during Dex-mediated treatment of autoimmune inflammation in the central nervous system (CNS) (Wüst et al., 2008). On the other hand, myeloid and airway epithelial cells have been shown to be the GC-responsive cells during contact hypersensitivity and airway inflammation, respectively (Klaßen et al., 2017; Tuckermann et al., 2007). In addition to the complexity behind the cell-type-specific GC actions, the ‘effector’ mechanisms underlying GC-induced immune modulation are thought to be diverse. GC induces apoptotic cell death in inflammatory cells such as airway infiltrating eosinophils (Meagher et al., 1996; Ohta and Yamashita, 1999; Schleimer and Bochner, 1994), although some cell types express resistance against GC-induced apoptosis (Gruver-Yates and Cidlowski, 2013; Liles et al., 1995).

Foxp3-expressing regulatory T (Treg) cells are the central regulator of immunity and tolerance, and defects in Treg cell generation or function result in systemic autoimmune inflammation (Bennett et al., 2001; Bluestone and Tang, 2018; Josefowicz et al., 2012). GC is known to modulate Treg cell homeostasis (Cari et al., 2019; Ugor et al., 2018). More specifically, GC can amplify interleukin-2 (IL-2)-dependent Treg cell expansion and enhance their suppressive activity to control autoimmune inflammation and graft-versus-host diseases (Chen et al., 2006; Xie et al., 2009). GC may upregulate Foxp3 mRNA expression in asthmatic patients (Karagiannidis et al., 2004) and promote the generation of antigen-specific induced Treg cells via GC-induced leucine zipper (GILZ) protein (Bereshchenko et al., 2014; Hamdi et al., 2007). However, discrepant findings have also been reported (Olsen et al., 2015; Sbiera et al., 2011). Therefore, the precise roles of GC in Treg cell biology and the underlying mechanisms remain elusive.

In the present study, we report that Treg cells are necessary for GC to control inflammatory diseases, autoimmune neuroinflammation, and allergic airway inflammation. Dex administered completely lost its treatment effects when Treg cells were absent. Likewise, inflammation in Treg cell-specific GR-deficient mice was not alleviated by Dex administration, suggesting a direct involvement of GC signaling in Treg cells during the treatment. Mechanistically, from transcriptome analysis we uncovered miRNA-342–3p induced by Dex specifically in Treg cells. miR-342–3p targeted mTORC2 component, Rictor. Dex-induced miR-342–3p played a key role in regulating metabolic profiles of Treg cells in response to Dex stimulation. Modulating miR-342 and Rictor expression directly controlled Treg cell suppressive function in vivo. Here, we report that Treg cells are indispensable direct target cells of Dex to control both autoimmune and allergic inflammation and that miR-342 and Rictor-dependent metabolic control in Treg cells plays a key role during glucocorticoid-mediated treatment.

RESULTS

Dexamethasone Fails to Attenuate Inflammation without Treg Cells

Systemic GC is the mainstay treatment option for inflammatory disorders. Experimental autoimmune encephalomyelitis (EAE) is a mouse model of autoimmune neuroinflammation induced by myelin oligodendrocyte glycoprotein (MOG) peptide immunization. Dexamethasone (Dex) administered at the onset of the disease effectively attenuated the development of clinical signs (Figure S1A). The accumulation of CD4⁺ T cells expressing inflammatory cytokines in the CNS tissues was significantly reduced by the treatment (Figures S1B and S1C). Anti-inflammatory effects of Dex were also examined in murine model of allergic airway inflammation induced by cockroach antigens (CA), in which lung-infiltrating CD4⁺ T cells secreting T helper-2 (Th2) cell type cytokines drive the inflammation (Figure S1D) (Jang et al., 2017). Dex systemically delivered during CA challenge diminished inflammatory cell infiltration in the airway and lung tissues (Figures S1E and S1G). CD4⁺ T cells expressing inflammatory cytokines in the lung tissues were also reduced following the treatment (Figure S1F). Therefore, Dex attenuates Th17- and Th2 cell-associated inflammatory responses.

Treg cells play an essential role in regulating immune responses during autoimmune and allergic inflammation (Dominguez-Villar and Hafler, 2018; Noval Rivas and Chatila, 2016). As the effect of GCon Treg cells in inflammation remains unclear, we sought to directly test the role of Treg cells during Dex treatment of inflammation. EAE was induced in Foxp3^DTR mice that express the diphtheria toxin receptor (DTR) under the Foxp3 promoter, and the mice were treated with Dex at the disease onset in the presence or absence of Treg cells (DTx or vehicle injected). Treg cell depletion following DTx treatment was validated in every experiment (data not shown). Mice rapidly succumbed to death following endogenous Treg cell depletion at the EAE onset (Figure 1A; Kim et al., 2019). However, Dex administration was unable to rescue mice from lethal disease when Treg cells were absent (Figure 1A). The accumulation of total CD4⁺ T cells and of T cells expressing proinflammatory cytokines in the CNS was greatly reduced by Dex treatment, and the reduction was lost when Treg cells were absent (Figures 1B and 1C). Histopathologic examination further validated Treg-cell-dependent Dex treatment effects (Figure 1D). CNS expression of inflammatory cytokine mRNA, Ifng, and Il17, and the key transcription factors, Tbx21 and Rorc, followed similar expression patterns (Figure 1E). Treg-cell-dependent Dex effects were similarly observed in allergic airway inflammation. Dex treatment during CA challenge failed to diminish eosinophil and CD4⁺ T cell infiltration in the bronchoalveolar lavage (BAL) and lung tissues following Treg cell depletion (Figures 1F and 1G; Figure S2A). Likewise, Dex administration failed to curtail inflammatory CD4⁺ T cell accumulation in the lung when Treg cells were depleted (Figure 1G; Figure S2B). Secretion of IL-4 and IL-13 in the BAL fluid and the expression of the Muc5a and Muc5b genes was reduced by Dex, but the reduction was only seen when Treg cells were present (Figure 1H; Figure S2C). Lung inflammation and airway resistance showed similar Treg-cell-dependent treatment effects (Figures 1I and 1J). Treg-cell-dependent treatment by Dex was also found following intranasal Dex administration, a route commonly used in asthmatic patients, suggesting that both systemically and locally administered Dex requires Treg cells for its anti-inflammatory effects (Figure 1K; Figure S2D).

Figure 1. Dex-Induced Treatment Requires Treg Cells.

(A–E) Foxp3^DTR mice were induced for EAE and injected with Dex daily on days 9–11 and/or with DTx on day 7. (A) The mice were daily scored for clinical diseases and survival.

(B) CD4⁺ T cells infiltrating the CNS tissues were enumerated on day 14.

(C) Intracellular IFN-γ⁺ and IL-17⁺ CD4⁺ T cells in the CNS were measured on day 14.

(D) H&E staining of the spinal cords (day 14 post immunization, original magnification x100). Scale bar, 10 μm. (E) qPCR analysis of the Ifng, Il17, Rorc, and Tbx21 expression in the CNS. Veh (n = 4–9), Dex (n = 4–13), DTx (n = 4–9), and DTx + Dex (n = 4–9) from two to three independent experiments.

(F–K) Foxp3^DTR mice were injected i.p. with CA in alum adjuvant on days 0 and 7. Starting at day 14, mice were intranasally challenged for 4 consecutive days with CA. DTx was injected 1 day before and on the day of first CA challenge.

(F–J) Vehicle or Dex was intraperitoneally injected on the first and third days of CA challenge (days 14 and 16). Mice were sacrificed 24 h after the last challenge. Total numbers of eosinophils and CD4⁺ T cells in the BAL (F) and of cytokine expressing CD4⁺ T cells in the lung (G) were calculated by flow cyometry analysis. (H) IL-4 and IL-13 secretion in the BALF was determined using a CBA assay. (I) H&E staining and histology score of the lung tissues are shown (original magnification × 4). Scale bar, 250 μm. (J) Airway resistance was measured by flexivent experiments.

(K) Dex (0.25 mg/kg) was administered intranasally on the first and third days of CA challenge. Eosinophils and CD4⁺ T cells in the BAL were enumerated. Veh (n = 4–13), Dex (n = 4–14), DTx (n = 4), and DTx + Dex (n = 4–18) from two to three independent experiments. Each symbol represents an individually tested animal. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, as determined by Mann-Whitney or Kruskal-Wallis nonparametric test. Please also see Figure S2.

Dex therapeutically given to mice with ongoing EAE rapidly lowered the clinical signs (Figure 2A). Dex-induced attenuation also required the presence of Treg cells, as Treg cell depletion abrogated the treatment effects (Figure 2A). The inability of Dex to reverse EAE without Treg cells was not due to severity of the disease, because similar Treg cell dependency was also seen during mild EAE induced by suboptimal MOG peptide immunization (Figure 2C). Immunizing mice with low dose of MOG peptide induced delayed and mild disease (peaked at the clinical score of ~2), and Dex treatment further lowered the disease severity (Figure 2C). Mild EAE turned into a lethal, albeit slightly delayed, EAE when Treg cells were depleted, and Dex treatment had no impact on rescuing mice from the disease (Figures 2C and 2E). Lethality was rescued by adoptive transfer of thymus derived regulatory (tTreg) or in vitro generated regulatory (iTreg) cells (Figures 2B, 2C, and 2D), and Dex treatment further diminished the disease severity (Figures 2B and 2D). Likewise, iTreg cells transferred reduced inflammatory CD4⁺ T cell infiltration and lung inflammation in endogenous Treg-cell-depleted mice treated with Dex (Figure S3). Collectively, these results demonstrate that the presence of Treg cells is necessary for Dex to control autoimmune and allergic inflammation in vivo.

Figure 2. Treg Cells Transferred Restore Dex-Induced Treatment in Treg-Depleted Mice.

(A) Foxp3^DTR mice were induced for EAE. Dex treatment started at the average score of 2 (i.e., days 13–15, arrows). n = 5–7 for each group.

(B) Foxp3^DTR mice (n = 5–6) induced for EAE and treated with DTx or with DTx + Dex as above were transferred with flow cytometry sorted tTreg cells (1×10⁶ cells per recipient).

(C–E) Foxp3^DTR mice were immunized with MOG 35–55 (150 μg for C or 300 μg for D), treated with Dex for 3 days (arrows). Mice received 2×10⁶ Foxp3⁺ flow cytometry sorted iTreg cells together with DTx on day 7. The mice were daily monitored and scored for clinical diseases and survival. n = 3–13 for each group. **p < 0.01; ***p < 0.001, as determined by Mann-Whitney or Kruskal-Wallis nonparametric test. Please also see Figure S3.

Dex Directly Acts on GR Expressed in Treg Cells for its Functions

GC mediates its immunomodulatory effects through cytosolic receptors, the glucocorticoid receptors (GR, encoded by the Nr3c1 gene) that are ubiquitously expressed (Kadmiel and Cidlowski, 2013; Oakley and Cidlowski, 2013). The loss of treatment effects of Dex without Treg cells suggested that Treg cells may have been the primary target cells of GC in vivo, although conventional T cells are also known as potential Dex targets capable of mediating its anti-inflammatory effects (Engler et al., 2017; Wüst et al., 2008). To directly test whether GC signaling (i.e., GR expression) in Treg cells is necessary for Dex treatment effects, we utilized Treg-cell-specific Nr3c1^−/− (Foxp3-cre x Nr3c1^fl/fl, Nr3c1^ΔTreg) mice. Treg-cell-specific GR deletion was validated by genomic DNA PCR (data not shown). Dex-induced dual specificity phosphatase 1 (Dusp1) gene expression was lost in Treg cells from Nr3c1^ΔTreg but not from control Nr3c1^WT mice, whereas conventional T cells from both types of mice upregulated Dusp1 expression following Dex stimulation (data not shown). T cell development in the thymus was not affected by the GR deficiency in Treg cells (Figure S4A), and peripheral CD4⁺ and Treg cells were comparable between the groups (Figure S4B). Both conventional CD4⁺ and Treg cells displayed similar surface phenotypes including Nrp1, Tim3, ICOS, GITR, PD1, and CD25, as well as in vivo turnover (Figure S4C; data not shown). Cellular expression of inflammatory cytokines and lineage transcription factors was also comparable between the groups (Figure S4D). Therefore, the lack of endogenous GC signaling in Treg cells appears to have little impact on immune homeostasis at steady state. Both Nr3c1^ΔTreg and Nr3c1^WT mice developed comparable EAE (Figure 3A, left). Following Dex treatment, EAE was significantly attenuated in Nr3c1^WT mice, whereas the disease in Nr3c1^ΔTreg mice remained unchanged (Figure 3A, right). Consistent with the disease severity, CD4⁺ T cell infiltration in the CNS was significantly greater in Dex treated Nr3c1^ΔTreg mice (Figure 3B). However, Treg cell accumulation was comparable (Figure 3B), suggesting that the inability of Dex to control inflammation cannot be attributed to Treg cell recruitment. Treg cell expression of surface phenotypes was similar between the groups (Figure 3C). CNS infiltrating CD4⁺ T cells from Dex-treated Nr3c1^WT mice secreted less IFN-γ and more IL-10 following recall stimulation compared to those from Dex-treated Nr3c1^ΔTreg mice (Figure 3D). The amount of serum cytokines (IL-17, IFN-γ, and TNF-α) was found greater in Dex-treated Nr3c1^ΔTreg mice (Figure 3E). The expression of proinflammatory cytokines in the CNS tissues measured by qPCR analysis also supported the results (Figure 3F). Similar results were observed in allergic inflammation model. Dex treatment significantly diminished inflammatory cell recruitment in the BAL cells in Nr3c1^WT but not in Nr3c1^ΔTreg mice (Figure 3G), and lung-infiltrating CD4⁺ T cells expressing inflammatory cytokines followed the same pattern (Figure 3H). Histopathologic examination of the lung tissues showed key involvement of GC signaling in Treg cells during Dex-mediated treatment of lung inflammation (Figure 3I). Lastly, lung infiltration of Treg cells was reduced by Dex treatment in Nr3c1^WT but not in Nr3c1^ΔTreg mice, although expression of Treg-cell-associated markers and lineage transcription factors remained comparable between the groups (Figure 3J; Figure S5). Therefore, direct Dex signaling via GR in Treg cells is essential for Dex to control inflammation.

Figure 3. Dex Directly Acts on GR Expressed in Treg Cells for its Functions.

(A–F) Treg cell-specific Nr3c1^−/− and control mice (n = 7–11) were induced for EAE and treated with vehicle or Dex on days 9–11 (arrows).

(A) Clinical score was daily monitored.

(B) Total number of CNS-infiltrating CD4⁺ T cells and Foxp3⁺ Treg cells was enumerated on day 14 post induction. (C) The mean fluorescence intensity (MFI) of ICOS, CD25, CD44, and CTLA4 on CNS infiltrating Foxp3⁺ Treg cells was determined by flow cytometry.

(D) CNS cells were isolated and restimulated with MOG peptide for 48 h. IL-10 and IFN-γ secretion in the culture supernatant was determined by ELISA.

(E) Serum IL17, TNF-α, and IFN-γ was determined by CBA assay (day 14 post EAE induction).

(F) Ifng and Il17 mRNA expression in the CNS was determined by qPCR on day 14 post immunization. Data shown are from the mean ± SD or individually tested mouse of three independent experiments.

(G–J) Treg-cell-specific Nr3c1^−/− and control mice were sensitized with CA in alum adjuvant and intranasally challenged as above. Dex (0.25 mg/kg) was administered intranasally on the first and third days of allergen challenge. Total numbers of eosinophils and CD4⁺ T cells in the BAL (G) and of cytokine expressing CD4 T cells in the lung (H) were determined by flow cytometry analysis. (I) H&E staining and histology score of the lung tissues (original magnification x10). Scale bar, 100 μm. (J) The total number and frequency of lung infiltrating Foxp3⁺ Treg cells, and the mean fluorescence intensity of Foxp3 and CD25 of Treg cells were determined by flow cytometry. Each symbol represents an individually tested animal. Veh (n = 6–8) and Dex (n = 7–9) from three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, as determined by Mann-Whitney or Kruskal-Wallis nonparametric test. Please also see Figure S4.

Differential Gene Signatures Are Induced by Dex in Treg Cells

To gain closer insights into the molecular mechanisms underlying Treg-cell-dependent Dex effects, we performed an RNA-seq experiment comparing gene expression between Treg and CD4⁺ T effector cells isolated from the inflamed lung tissues of mice induced for allergic inflammation and treated with Dex or vehicle. In particular, we compared genes differentially expressed by Dex over vehicle treatment in each population (Treg versus Teff cells, after normalization by vehicle group, p < 0.001) and the total 1752 genes were identified (Figure S6). These genes were reanalyzed using repetitive normalization by vehicle treatment in each cell type, and 150 genes were identified based on the expression pattern by Dex treatment (|FC|>2, p < 0.001, Figure S6). Differentially expressed genes were equally distributed based on the functions and location (Figure 4A). We particularly focused on the genes whose expression is elevated by Dex treatment only in Treg cells. They included Ccr9, Lrrc32, Esam, Ly6c1, Napb, Rragd, and Evl (Figure 4B; Figure S7). Treg-cell-specific upregulation by Dex was validated by qPCR analysis. Since Treg-cell-dependent Dex treatment effects were found in both allergic and autoimmune inflammation models, the validation was performed using Treg and effector cells isolated from inflamed tissues from EAE (Figure 4C) and from allergic inflammation (Figure 4D). In support of the RNA-seq data, Dex treatment significantly elevated the expression of the selected genes in Treg cells, while the same treatment lowered the expression in effector T cells (Figures 4C and 4D). Of note, Dex-induced downregulation in effector T cells was dependent on Treg cells or on Treg cell GR expression, as Treg cell depletion or GR deficiency in Treg cells abrogated the downregulation (Figures 4C and 4D). Glucocorticoid-induced leucine-zipper (Gilz), GR (Gra), dual specificity phosphatase 1 (Dusp1), and FK506 binding protein 5 (Fkbp5) are canonical Dex-inducible genes (Ayroldi and Riccardi, 2009; Kelly et al., 2012; Tchen et al., 2010), and they were equally induced by Dex treatment in both Treg and effector cells (Figure 4E). Moreover, in Nr3c1^ΔTreg mice treated with Dex, the canonical gene expression was lost only in Treg but not in effector cells, validating Treg cell-specific loss of GR signaling (Figure 4E). Unlike Treg-cell-dependent genes shown above, Treg cell deletion had no impact on the expression of the canonical Dex-induced genes in effector T cells (Figure 4E). Therefore, these results demonstrate groups of genes controlled by Dex treatment via different mechanisms between Treg and effector cells.

Figure 4. Differentially Expressed Genes between Treg Cells and T Effector following Dex Treatment.

(A, B, and D) Foxp3^GFP or Foxp3^DTR mice were induced for allergic inflammation, treated with vehicle or Dex as described above. At sacrifice, Foxp3⁺ Treg and effector CD4⁺ T cells were flow cytometry sorted from the lung tissues.

(A and B) RNA-seq analysis was performed (n = 3 each group). Based on the function and location, gene sets were marked for each category.

(B) Heatmap shows expression of genes (|fold change| > 2, p < 0.001 in Treg cells versus effector cells).

(C–E) RNA was isolated from the sorted cells and evaluated for gene expression by qPCR. Expression of the indicated genes was normalized to the housekeeping Gapdh gene, and relative quantification (RQ) value was calculated.

(C and E) Foxp3^DTR or Treg-cell-specific Nr3c1^−/− mice induced for EAE and treated with Dex or DTx + Dex were sacrificed as above. Treg cells and effector CD4⁺ T cells were flow cytometry sorted from the CNS tissues, and relative expression of the indicated genes was determined by qPCR. Data are the mean ± SD of three independent experiments. *p < 0.05; **p < 0.01, as determined by Kruskal-Wallis nonparametric test. Please also see Figure S5.

Dex Induces microRNA-342 Specifically in Treg Cells

Ena-vasodilator stimulated phosphoprotein (EVL) is an actin-associated protein involved in various processes dependent on cytoskeleton remodeling and cell polarity (Trichet et al., 2008) and is one of the Dex-induced genes in Treg cells (Figures 4C and 4D). Notably, the third intron of the Evl gene encodes a highly conserved microRNA-342 (miR-342) (Figure 5A) (Grady et al., 2008; Li et al., 2018), and the expression of intronic miR-342 mirrors that of the host gene, Evl (De Marchis et al., 2009; Grady et al., 2008). Indeed, miR-342–3p expression was significantly enhanced by Dex treatment in Treg cells in vivo, whereas the expression was reduced in effector T cells by the treatment (Figure 5B). GR expression of Treg cells was necessary for miR-342–3p induction by Dex (Figure 5B). Searching for its target molecules, we found that the Rictor, a mTORC2 complex molecule, was a potential target of miR-342–3p based on the sequence homology (Figure S8). Indeed, Rictor expression was inversely regulated by Dex treatment in Treg and effector cells via a GR-dependent manner (Figure 5C). Increased expression of the miR-342–3p and downregulation of the Rictor were also observed in Treg cells isolated from inflamed lung tissues, and effector T cells displayed a reversed expression pattern (Figure 5D). Therefore, Rictor expression appears to be controlled by miR-342–3p induced by Dex stimulation.

Figure 5. Dex Induces miRNA-342–3p Specifically in Induced Treg Cells.

(A) Schematic representation of the Evl gene and the location of the conserved miRNA-342.

(B and C) Foxp3^DTR or Treg-cell-specific Nr3c1^−/− mice induced for EAE and treated with Dex were sacrificed day 14 post induction as in Figure 4. qPCR analysis of miR-342–3p and Rictor expression in the Treg and effector cells isolated from the CNS.

(D) Foxp3^DTR mice induced for allergic inflammation were treated with vehicle or Dex as described in Figure 1. At sacrifice, Treg and effector CD4 T cells flow cytometry sorted from the lung were evaluated for gene expression by qPCR.

(E and F) Relative expression of the indicated miRNAs in naive or in vitro differentiated CD4⁺ T cells with or without Dex.

(G) miR-342–3p expression in the thymus-derived tTreg cells. tTreg cells isolated from Foxp3^GFP mice were activated with or without Dex. miR-342–3p and Rictor expression was determined by qPCR analysis.

(H) Immuno blot analysis for Rictor, Raptor, and β-actin in lysates from iTreg cells treated with or without Dex.

(I) Eosinophils and B cells isolated from the BAL and mediastinal lymph node of mice with allergic inflammation, respectively, were evaluated for miR-342–3p expression by qPCR.

(J) Expression of miR-342–3p and miR-16 in the spinal cord and brain tissues was determined (day 14 post EAE induction). Data are the mean ± SD of three independent experiments. *p < 0.05; **p < 0.01, as determined by Mann-Whitney or Kruskal-Wallis nonparametric test. Please also see Figure S6.

We next determined whether Dex-induced miR-342–3p expression is Treg cell-specific. miR-342–3p expression was pronounced in in vitro differentiated Treg (iTreg) cells compared to other conventional effector T cell subsets differentiated in vitro (Figure 5E). Other miRs known to be expressed in Treg cells, including miR-146, miR-155, miR-15b, and miR-16, were highly expressed in iTreg cells (Figure 5E; Figure S9A). However, miR-342–3p was the only miR elevated by Dex treatment in Treg cells among those tested (Figure 5F; Figure S9B). Similar expression pattern was also observed in thymus-derived tTreg cells (Figure 5G). Dex stimulation further enhanced miR-342–3p expression, while the Rictor expression was downregulated (Figure 5G), suggesting that both subsets of Treg cells (thymus derived and induced) similarly respond to Dex. The expression of Rptor, a mTORC1 complex molecule, remained unchanged by Dex treatment, suggesting a mTORC2-specific effect (Figure S9C). Of note, Gilz expression was comparably induced by Dex treatment in all tested effector subsets (Figure S9C). Immunoblot analysis further validated diminished Rictor but not Raptor expression in iTreg cells following Dex stimulation (Figure 5H). Dex stimulation did not induce miR-342–3p expression in other immune cells such as BAL eosinophils or lymph node B cells from the mice induced for allergic inflammation, suggesting Dex-induced miR-342–3p expression in Treg lineage cells (Figure 5I). Elevated miR-342–3p expression by Dex in Treg cells also occurred in vivo. Dex treatment increased miR-342–3p expression in the CNS tissues of EAE mice, and Treg cell depletion abrogated the expression, while miR-16 expression remained comparable by Dex treatment (Figure 5J). Collectively, these results demonstrate that Dex induces miR-342–3p expression specifically in iTreg cells.

miRNA-342 Controls Dex-Mediated Treg Cell Functions

We next examined whether miR-342–3p plays a role in Treg cell functions during Dex-mediated control of inflammation. iTreg cells were generated in the presence or absence of Dex. miR-342–3p expression was elevated in iTreg cells compared to Th0 cells, and further enhanced by Dex treatment (Figure 6A). Dex treatment increased Foxp3 mRNA expression (Figure 6A), although Foxp3 expression based on GFP expression remained comparable between the groups (Figure 6B). Inhibiting miR-342 expression using miR-342–3p specific antisense oligonucleotide inhibitors abolished miR-342–3p expression in iTreg cells as well as in tTreg cells even with Dex stimulation (Figure 6A; data not shown). miR-342 inhibition slightly but significantly diminished the Foxp3 mRNA expression and Dex treatment was unable to restore the expression (Figure 6A). Yet, Foxp3 protein expression remained unchanged by miR-342 inhibition (Figure 6B). miR-342 inhibition drastically reversed Rictor expression, supporting that the Rictor is a direct target of miR-342–3p, whereas the Rptor expression remained unchanged in all tested conditions (Figure 6A). Evl expression was further enhanced by Dex treatment regardless of miR-342 inhibition, validating that miR-342–3p is a downstream molecule of the Evl (Figure 6A). To test the role of miR-342–3p in Treg cells in vivo, iTreg cells treated with miR-342-specific or control inhibitors in the presence or absence of Dex were adoptively transferred into Foxp3^DTR mice induced for EAE and treated with DTx at the onset. As shown above, endogenous Treg cell depletion at EAE onset resulted in lethal disease and exogenous Treg cell transfer rescued mice from the lethality (Figure 6C). Control inhibitor treated iTreg cell transfer rescued mice from death, and Dex stimulation of iTreg cells further lowered the disease severity (Figure 6C). By contrast, miR-342 inhibitor treated iTreg recipients developed severe (but not lethal) EAE compared to control iTreg cell recipients, and more importantly, Dex stimulation of these iTreg cells had no impact on attenuating the disease severity, suggesting that miR-342–3p expression in Treg cells appears to be critical for Dex-mediated Treg suppressive function in vivo (Figure 6C).

Figure 6. miR-342–3p Controls Treg Functions and Metabolism.

(A and B) Naive CD4⁺ T cells isolated from Foxp3^GFP mice were nucleofected with control (NC) or miR-342–3p specific antisense oligonucleotide inhibitor and then activated in vitro under Th0 or iTreg cell conditions with or without Dex for 72 h.

(A) Foxp3, miR-342–3p, Rictor, Rptor, and Evl expression were determined by qPCR analysis.

(B) Foxp3 (GFP) expression was determined by flow cytometry analysis.

(C) Foxp3^DTR mice were induced for EAE and treated with DTx day 7 post induction. 2×10⁶ control iTreg cells or miR-342–3p inhibitor treated iTreg cells as above were transferred on day 7. Dex treatment was performed on days 9 through 11. The mice were daily monitored and scored for clinical diseases. N.C (n = 5), Dex (n = 7), miR-342–3p inh (n = 8), and miR-342–3p inh + Dex (n = 8) from two to three independent experiments.

(D) Expression and phosphorylation of the indicated proteins in control or miR-342–3p inhibitor treated iTreg cells with or without Dex were determined by immuno blot analysis.

(E–G) Naive CD4⁺ T cells nucleofected with the indicated inhibitors or si-Rictor were stimulated under iTreg condition with or without Dex for 72 h. (E) OCR and EACR measurement. Quantitation of basal respiration and glycolysis are shown. n = 5 each group. (F) qPCR analysis of Rictor expression. (G) qPCR analysis of Hk2 and Hif1a expression. Data are the mean ± SD of two to three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, as determined by Kruskal-Wallis nonparametric test. Please also see Figure S7.

miR-342 Controls Treg Cell Metabolism

Rictor is a key adaptor protein of the mTORC2 complexes (Guertin et al., 2006; Oh and Jacinto, 2011). Unlike Raptor, Treg-cell-specific Rictor deletion does not result in gross abnormalities, although it has been implicated that Rictor may play a role in Treg maintenance (Chapman and Chi, 2014; Zeng et al., 2013). It has been recently reported that mTORC2 inactivation or Rictor deficiency may improve Treg cell function (Charbonnier et al., 2019). Since Dex-induced miR-342–3p in iTreg cells targets Rictor and miR-342 directly controls Treg functions in vivo, we asked whether miR-342-dependent inhibition of Rictor has impact on the mTOR pathway and on the metabolic profile of Treg cells during Dex stimulation. Dex treatment significantly downregulated Rictor protein expression, and miR-342 inhibition completely reversed the downregulation even with Dex treatment (Figure 6D; Figure S10A). Rictor is involved in phosphorylating Akt at Ser 473 site (Dibble et al., 2009; Sarbassov et al., 2005). Indeed, Akt Ser473 phosphorylation was diminished by Dex treatment, while it was enhanced by miR-342 inhibition (Figure 6D; Figure S10A). Akt Thr308 phosphorylation induced by PDK1 was not affected by Dex and/or miR-342 inhibition (Figure 6D; Figure S10A). Fully activated Akt phosphorylates FoxO1 (Pan et al., 2017), and indeed enhanced FoxO1 transcription factor phosphorylation was observed following miR-342 inhibition (Figure 6D; Figure S10A). Dex treatment inhibited S6K activity, whereas miR-342 inhibition enhanced S6 phosphorylation regardless of Dex (Figure 6D; Figure S10A).

Treg cells utilize fatty acid oxidation and oxidative phosphorylation (OXPHOS) to meet their bioenergetic needs, and glycolysis is under tight control (Gerriets et al., 2015; Howie et al., 2017; Michalek et al., 2011). Foxp3 supports OXPHOS and suppresses glycolysis (Angelin et al., 2017). Enforcing glycolysis in Treg cells by upregulating glucose transporter Glut1 activates mTORC1 pathway and reduces Foxp3 expression as well as the suppressive function (Gerriets et al., 2016). Based on miR-342-dependent regulation of Rictor and other metabolic regulators, we determined metabolic profiles of Treg cells stimulated with Dex in the presence or absence of miR-342 inhibition. Basal oxygen consumption rate (OCR), an indication of oxidative phosphorylation, was elevated by Dex treatment, whereas miR-342 inhibition lowered the OCR regardless of Dex stimulation (Figure 6E; Figure S10B). Conversely, Dex treatment reduced extracellular acidification rate (ECAR), a measurement of overall glycolytic flux (Figure 6E; Figure S10B). miR-342 inhibition reprogrammed Treg cells to glycolytic pathways and Dex treatment had little impact on glycolytic pathways induced by miR-342 inhibition (Figure 6E; Figure S10B). In support, the expression of hexokinase 2 (Hk2), a key mediator of aerobic glycolysis (Wolf et al., 2011), and hypoxia inducible factor 1a (Hif1a), a key factor upregulating many enzymes in glycolytic pathway (Mathupala et al., 2001), was decreased by Dex, and miR-342 inhibition reversed the Dex-induced reduction (Figure 6G; Figure S10B). To examine if miR-342 inhibitor induced alteration in Treg cell metabolism is attributed to Rictor, the cells were cotreated with si-Rictor. si-Rictor treatment abolished Rictor expression even with miR-342 inhibitor treatment (Figure 6F). These Treg cells restored OXPHOS and suppressed glycolysis (Figure 6E). Hk2 and Hif1a expression further corroborated the findings (Figure 6G). When tTreg cells were treated with Dex, miR-342 inhibitor, and si-Rictor, we observed similar metabolic profiles (Figure S11). Thus, Dex controls the metabolic profiles in both tTreg and iTreg cells via the miR-342-Rictor pathways. Dex had no impact on OXPHOS and glycolysis in tTreg cells isolated from Nr3c1^ΔTreg mice, suggesting that Dex induces metabolic programming via GC signaling (Figure S12). miR-342 inhibition abolished Dex-induced OXPHOS and glycolysis regardless of GR expression (Figure S12).

To further validate that miR-342–3p controls Treg cell function, we examined if miR-342–3p overexpression in Treg cells enhances their suppressive function. miR-342–3p overexpression significantly reduced the Rictor expression in Treg cells even without Dex treatment (Figure 7A). In vivo, miR-342 mimic transfected iTreg cells effectively attenuated EAE comparable to Dex treated WT iTreg cell recipients, even without Dex treatment (Figure 7B), suggesting that miR-342 is capable of enhancing Treg cell suppressive function independently of Dex. If Dex-mediated Rictor downregulation via miR-342–3p in Treg cells strengthens suppressive functions of Treg cells by supporting OXPHOS and by suppressing glycolytic pathways, it is then possible that enforced Rictor expression may abrogate such effects. To test this possibility, we overexpressed Rictor in Treg cells. miR-342–3p expression remained unchanged by Rictor overexpression, although Hk2 and Hif1a mRNA expression was substantially upregulated, suggesting altered metabolic program by Rictor overexpression (Figure 7C). In vivo, Rictor overexpressing iTreg cell recipients with EAE did not respond to Dex treatment, and the disease severity remained unchanged, while control iTreg cell recipients displayed attenuated disease following Dex treatment (Figure 7D). Therefore, these results demonstrate that enforced Rictor expression in Treg cells negates Dex treatment effects in vivo, further supporting the key contribution of the miR-342–3p-Rictor axis to Treg cell functions during Dex treatment.

Figure 7. Dex Controls Treg Functions via the miR-342–3p-Rictor Axis.

(A and B) Naive CD4⁺ T cells isolated from Foxp3^GFP mice were nucleofected with control (NC) or miR-342–3p mimic, and then activated in vitro under iTreg conditions with or without Dex for 72 h.

(A) miR-342–3p and Rictor expression were determined by qPCR analysis.

(B) Foxp3^DTR mice induced for EAE and treated with DTx day 7 post induction were transferred with 2×10⁶ control or miR-342–3p mimic treated iTreg cells and daily treated with Dex on days 9–11 (arrows). The mice were daily monitored and scored for clinical diseases. n = 5–9 each group.

(C and D) Naive CD4⁺ T cells isolated from Foxp3^GFP mice were nucleofected with control or Rictor expression vectors and subsequently polarized under iTreg condition with or without Dex for 72 h.

(C) qPCR analysis of miR-342–3p, Rictor, Hk2, and Hif1a expression.

(D) Foxp3^DTR mice induced for EAE and treated with DTx day 7 post induction were transferred with 2×10⁶ control or Rictor overexpressing iTreg cells. Dex treatment was performed on days 9 through 11 as above (arrows). The mice were daily monitored and scored for clinical diseases. n = 6–10 each group. Data are the mean ± SD of two to three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001, as determined by Mann-Whitney or Kruskal-Wallis nonparametric test.

DISCUSSION

In this study, we report that Treg cells were necessary for Dex to control autoimmune and allergic inflammation. Treg cell depletion itself induces severe inflammatory responses from conventional T cells, raising the possibility that the ineffective Dex effect could result from heightened Th cell responses (Arvey et al., 2014). However, we found that the lack of GR only in Treg cells impaired Dex effects in vivo, supporting a direct role for Treg cells in response to Dex stimulation. Mechanistically, Dex induced miR-342–3p in Treg cells, and the mTORC2 complex adaptor molecule, Rictor, was the direct target of miR-342–3p. Inhibiting miR-342 expression in Treg cells reversed Dex-mediated Treg-cell-dependent inhibition of the inflammation in vivo. miR-342–3p-mediated Rictor downregulation enforced Dex-induced OXPHOS metabolic profiles in Treg cells. Inhibiting miR-342 expression reversed Rictor expression, caused metabolic reprogramming to glycolytic pathways, and impaired Treg cell suppressive functions, during which Dex administration was unable to reverse impaired Treg cell functions. Likewise, miR-342 overexpression enhanced Treg cell functions even without Dex treatment, while Rictor overexpression in Treg cells abrogated Dex-mediated inhibition of inflammation in vivo. Collectively, these results uncover a previously unknown mechanism by which Treg cells mediate Dex-induced inhibition of inflammatory responses via a miR-342- and Rictor-dependent mechanism.

GC including Dex activates cytosolic GR expressed in almost every cell type (Oakley and Cidlowski, 2013). Previous studies demonstrate that the in vivo GC target cells differ depending on the tissues or the types of inflammatory responses. Lck-cre-induced GR deletion, targeting the exon 3 encoding the first zinc finger DNA binding domain (Tronche et al., 1999), uncovered that T cells are the primary GC-responding cells in Th1 and Th17 cell-mediated autoimmune inflammation (Engler et al., 2017; Wüst et al., 2008), whereas T cell expression of the GR seems dispensable during GC-mediated treatment of contact hypersensitivity (Tuckermann et al., 2007). In the present study, we utilized GR-floxed mice targeting the exon 3, independently developed by Ashwell and colleagues (Mittelstadt et al., 2012), to delete GR expression in Treg cells and found that Treg cells are necessary for GC to control both Th1 and Th17 cell-associated autoimmune and Th2 cell-associated allergic airway inflammation. Of note, Wüst et al. (2008) ruled out that Treg cells play an anti-inflammatory role of Dex in EAE based on the finding that Dex treatment diminishes peripheral Treg cell frequency in control but not in Lck-cre Nr3c1^fl/fl mice (Wüst et al., 2008). We found comparable Treg cell accumulation in the CNS in Dex-treated Nr3c1^ΔTreg mice. Since Lck-cre-induced deletion of the GR in all T cells including Treg cells could have masked the contribution of Treg cells in EAE, we would argue that Treg-cell-specific Nr3c1 deletion is the better approach to precisely test the contribution of Treg cells during Dex-induced treatment. Alternatively, different Dex doses used may account for the discrepancy in these studies (100mg/kg versus 5mg/kg). A study from Klaßen et al. reported that airway epithelial cells are the major GC target cells and that Nr3c1 deletion in total T cells had no impact on GC-mediated inhibition of allergic inflammation (Klaßen et al., 2017). Again, the reason behind the discrepancy is not clear. We used 2.5mg/kg of Dex in cockroach antigen-induced inflammation, while Klaben et al. used 10mg/kg of Dex in OVA-induced allergic inflammation. Although the dose of GC used may play a role in GC action on different target cells, we found that increasing Dex dose to 10mg/kg in our model still required the presence of Treg cells for Dex to control cockroach antigen-induced airway inflammation (Nguyen and Min, unpublished results). Alternatively, allergens with protease activity, such as cockroach antigens, are known to enhance inflammatory responses via epithelial activation (Kale et al., 2017), potentially altering GC responsiveness. The nature of allergens may dictate the target cells through which Dex mediates anti-inflammatory roles.

GR-deficient Treg cells are unable to control experimental T cell-induced colitic inflammation, although GR expression (i.e., GC signal) plays little role in Treg cell homeostasis and survival at steady state (Rocamora-Reverte et al., 2019). IFN-γ production in GR deficient Treg cells was noticed, suggesting that endogenous GC signal may prevent Treg cells from acquiring Th1 cell-like phenotypes (Rocamora-Reverte et al., 2019). In agreement with this report, we found that the lack of GR in Treg cells does not alter immune homeostasis in the primary and secondary lymphoid tissues at steady state. Likewise, EAE severity remained comparable between Treg-cell-specific Nr3c1 deficient and control mice without Dex administration, suggesting that endogenous GC signaling in Treg cells does not appear to affect Treg cell’s ability to control inflammation in the CNS tissues. This is in contrast with a previous study that Lck-cre induced Nr3c1 deletion results in early onset of EAE (Wüst et al., 2008). Therefore, endogenous GC signaling may have distinct impact on conventional T versus Treg cells. Further investigation is needed to better understand Dex effects in different cell types.

The facts that miR-342–3p expression is induced by Dex stimulation only in Treg but not in other effector lineage cells and that miR-342–3p plays an instrumental role during Dex-mediated control of inflammation raise several questions. First, how is miR-342 expression specifically regulated in Treg cells? miR-342 expression is controlled by its host gene, Evl, encoding cytoskeletal effector protein implicated in the motility and adhesion (Estin et al., 2017). In vitro generated Foxp3⁺ Treg cells express higher Evl compared to other lineage effector subsets including Th17 lineage cells, suggesting that TGF-β may not be involved. However, we cannot exclude the possibility that TGF-β signaling in the absence of inflammatory factors such as IL-6 may be important for the Evl expression. Alternatively, Foxp3 expression itself may be a factor driving the expression of miR-342–3p. In fact, Dex enhances miR-342–3p and Foxp3 gene expression in developing iTreg cells (Kim and Min, unpublished results). The mechanisms behind miR-342 and Treg cell development are currently being examined. Second, how does miR-342 control Treg cell function? Based on the sequence homology, we found that Rictor expression was directly controlled by miR-342–3p. Metabolic profiles, particularly the mTOR pathways, of Treg cells have extensively been studied (Kishore et al., 2017; Zeng and Chi, 2015). Unlike Rptor, deletion of the Rictor in Treg cells does not alter Treg cell homeostasis, and Rictor-deficient Treg cells express normal Treg cell gene signatures and mitochondrial activity (Zeng et al., 2013). However, our study demonstrates that Rictor expression in Treg cells was critical especially in maintaining OXPHOS metabolic programming and in limiting glycolytic pathway activation during Dex treatment, for neutralizing miR-342–3p expression with miR-342-specific antisense inhibitors and overexpressing miR-342–3p using miR mimics abolishes and bolsters Treg cell suppressive functions in vivo, respectively. It is worth noting that the cells in which miR-342 expression was inhibited or Rictor was overexpressed become GC refractory. The condition, referred to as steroid resistance, is often associated with severe inflammatory conditions. It will thus be important to examine the causal relationship between GC resistance and miR-342–3p expression (or metabolic programming) of Treg cells. Third, is Rictor expression controlled by miR-342–3p the sole mechanism underlying GC treatment? Since a single miRNA could target multiple genes (Creighton et al., 2012), there may be additional factors that are controlled by miR-342 and are involved in Treg-cell-dependent control of inflammatory responses. We are currently investigating this question.

The fact that Treg cells are necessary in responding to therapeutically administered GC is a compelling yet puzzling finding, especially given the ubiquitous expression of the GC receptor in every cell type. The molecular basis underlying the Treg-cell-specific GC responses will require future investigation. It is estimated that there are more than 1000 GR binding sites, i.e., glucocorticoid responsive element (GRE), in the genome (Vockley et al., 2016). It is particularly worth noting that GR itself is a Foxp3 binding protein (Rudra et al., 2012). Therefore, depending on the cell types, different GR signaling complexes including DNA binding proteins and other transcription modulators may be formed, driving cell-type-specific regulation (Franco et al., 2019). In support of this notion, we found a cohort of genes that is commonly induced or that is specifically altered in Treg cells in response to Dex stimulation. It will therefore be important to compare the DNA binding landscape of GR complexes between different cell types, especially lymphocyte subsets.

In summary, the current study identifies a previously unappreciated mechanism by which Treg cells play a key role during GC-induced treatment of inflammation. The study uncovers miR-342–3p as a potential regulator of metabolic programming in Treg cells in response to GC stimulation. Further investigation may enable us to understand how GC modulates inflammation and possibly to identify therapeutic targets to treat severe inflammatory conditions including steroid-resistant inflammation by targeting the miR-342-Rictor axis in Treg cells.

STAR★METHODS

Detailed methods are provided in the online version of this paper and include the following

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Booki Min (minb@ccf.org or booki.min@northwestern.edu).

Materials Availability

This study did not generate new unique reagents.

Data and Code Availability

The accession number for the RNA-seq reported in this paper is GEO: GSE136969.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals

C57BL/6 (B6), C57BL/6 Foxp3^DTR, Foxp3^Cre/YFP mice, and Nr3c1^fl/fl mice were purchased from the Jackson Laboratory (Bar Harbor, ME). C57BL/6 Foxp3^GFP mice were obtained from Dr. Yasmine Belkaid (NIH) (Bettelli et al., 2006). Six to twelve-week old male and female (sex- and age-matched) mice were used for the experiments. All of the mice were maintained in a specific pathogen-free facility located at the Lerner Research Institute. All experiments were performed in accordance with the guidelines of the Lerner Research Institute Animal Care and Use Committee.

METHODS DETAILS

EAE induction

EAE was induced by injecting mice subcutaneously into the flanks with 200 μL of emulsion containing 300 μg or 150 μg MOG_35–55 peptide (BioSynthesis, Lewisville, TX) and 5mg/mL Mycobacterium tuberculosis strain H37Ra (Difco, Detroit, MI). In addition, the mice received 200 ng pertussis toxin (Sigma, St. Louis, MO) intraperitoneally injected on days 0 and 2 after immunization. Vehicle or Dex (5mg/kg, Sigma) were intraperitoneally injected on days 9–11 after immunization. Clinical signs of EAE were assessed daily according to the standard 5-point scale. For depletion of Treg cells in Foxp3^DTR mice, diphtheria toxin (DTx, Sigma) was injected intraperitoneally (1 μg) on day 7 after immunization. Spinal cords were fixed in 4% paraformaldehyde, and paraffin embedded, and sections were stained with hematoxylin and eosin (H&E).

Allergic Airway inflammation

Mice were injected intraperitoneally on days 0 and 7 with 5 μg of Cockroach antigen (CA, Greer Laboratory, Lenoir, NC) mixed in 100 μL of aluminum hydroxide adjuvant (Alum, Sigma). Starting on day 14, the mice were intranasally challenged for 4 consecutive days with 5 μg CA in PBS. Vehicle or Dex (2.5 mg/kg) were intraperitoneally injected on the first and third days of allergen challenge (day 14 and day 16). Mice were sacrificed 24 h after the last Ag challenge. In some experiments, Treg cells were depleted by intraperitoneal injection of 1 μg of DTx. Lung tissues were prepared from paraffin-embedded blocks and stained with H&E. Lung mechanics was measured using the FlexiVent ventilator (FlexiVent, Scireq). Lung resistance was measured in response to increasing doses of inhaled methacholine, as previously described (Nguyen et al., 2019).

Antibodies and Flow cytometry

At sacrifice, BALF cells, lung, and CNS tissue were collected and prepared as previously described (Kim et al., 2019; Nguyen et al., 2019). Briefly, the cells were stained with anti-CD4 (RM4–5, BioLegend), anti-Ly6G (1A8, BD Biosciences), and anti-Siglec-F (1RNM44N, eBioscience) mAbs. For intracellular staining, cells were stained with anti-IL-4 (11B11, eBioscience), anti-IL-13 (eBio13A, eBioscience), anti-IL-17 (TC11–18H10, BD Bioscience), anti-IFN-γ (XMG1.2, BD Bioscience), anti-IL-5 (TRFK5, BD Biosciences), anti-IL-10 (JES5–16E3, BD Biosciences) and anti-IL-2 (JES6–5H4, BD Biosciences) mAbs. Cells were acquired using a FACSLSRII or FACSFortessa X-20 (BD Biosciences) and analyzed using a FlowJo software (Tree Star Inc.). For intracellular staining, cells were harvested separately and stimulated ex vivo with PMA (10ng/mL, Millipore-Sigma) and ionomycin (1 μM, Millipore-Sigma) for 4 h in the presence of 2 μM monensin (Calbiochem) during the last 2 h of stimulation. Cells were immediately fixed with 4% paraformaldehyde, permeabilized, and stained with fluorescence-conjugated mAbs.

Nucleofection

Nucleofection was performed with Mouse T Cell Nucleofector Kit and Nucleofector device (Amaxa, Koelin, Germany). First, 1 × 10⁶ naive CD4⁺ T cells were resuspended in 100 μL nucleofector solution. 100 pM oligonucleotides (including miR-342–3p inhibitor and control-NC5; Integrated DNA Technologies, San Jose, CA) were added into the solution and mixed gently. Then the mixtures were gently transferred to electroporation cuvettes and placed in the Nucleofector device. Cells were nucleofected in the X-001 program. Finally, transfected cells were transferred to a 24-well plate with 1.5 mL prepared Mouse T Cell Nucleofector Medium in each plate and incubated in a humidified 37°C/5% CO² incubator for 2 h. Then, transfected cells were activated under the indicated polarizing conditions and treated with Dex (50nM).

In vitro T cell differentiation and adoptive transfer

NaiveCD4⁺Foxp3^–CD44^low T cells from lymph nodes of Foxp3^GFP mice were sorted by cell sorting and treated with immobilizedanti-CD3 (2C11) and soluble anti-CD28 (37.51) mAbs in the presence of recombinant cytokines and antibodies to promote effector T cell differentiation. For Th1 and Th2 cells, IL-12 (10ng/mL)/anti-IL-4 (10 μg/mL) and IL-4 (1000U/mL)/anti-IFN-γ (10 μg/mL) were used, respectively. For Th17 cells, IL-6 (10ng/mL)/TGF-β (2.5ng/mL) and anti-IL-4/anti-IFN-γ (10 μg/mL) were used. For iTreg cells, IL-2 (100U/mL) and 5 ng/mLTGF-β (2.5 ng/mL) were used. After 3 days, iTreg cells (CD4⁺ Foxp3⁺) cells were cell sorted using a FACSAria (BD, San Jose,CA) prior to transfer. Cytokines and antibodies were purchased from Peprotech (Rocky Hill, NJ) and eBioscience (San Diego, CA), respectively.

Cytokine analysis by a cytometric bead array

The serum and BALF were collected from mice induced for autoimmune and allergic inflammation, respectively. The cytokines and IL-13 secretion was quantified using a cytometric bead array (CBA) kit (BD Biosciences) and an enhanced sensitivity flex set (BD Biosciences), respectively, according to the manufacturer’s instructions. The data were analyzed using the CBA software, and the standard curve for each cytokine was generated using the mixed cytokine standard.

ELISA

CNS cells isolated from mice with ongoing EAE were in vitro stimulated with 20 μg/mL MOG_35–55 peptide for 48 h. ELISA kits were used to detect the levels of IL-10 and IFN-γ according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN). Five repeated wells were set for each sample and standard product, and the optical density was measured at 492 nm using a microplate reader (Thermo Fisher Scientific).

Quantitative RT-PCR

RNA was isolated from inflamed lung, CNS or FACS sorted cells using a GeneJet RNA isolation kit (Thermo Fisher Scientific) or Quick-RNA Microprep kit (Zymo Research, Irvine, CA). cDNA was synthesized using Moloney Murine Leukemia Virus (M-MLV) reverse transcriptase (Promega, Madison, WI). For miRNA detection, cDNA was prepared using the miRscript cDNA synthesis kit (QIAGEN, Hilden, Germany), according to manufacturer’s protocols. qPCR reactions were set up with the miScript SYBR Green PCR Kit. qPCR analysis was performed using a QuantStudio 3 Real-Time PCR System (Applied Biosystems, Waltham, MA) using TaqMan reagents (Applied Biosystems) or SYBR green master mixes (Applied Biosystems). The gene expression level of miRNAs was normalized to that of snoRNA U61 and of mRNAs to that of Gapdh.

Immuno blot analysis

Cells were lysed in lysis buffer (Cell Signaling Technologies). Protein separation was performed on4%polyacrylamide gel and electro blotted on nitrocellulose membrane (Whatman Inc., Florham Park, NJ.). The membrane was incubated for 60 min in5%skimmed milk powder in PBS with 0.1% Tween (PBS-T), and were then incubated overnight at 4°C with the appropriate primary antibody (see star method, Cell Signaling Technologies, Danvers, MA). On the next day, the membrane was washed with PBS-T, and incubated with the appropriate peroxidase secondary antibody (Cell Signaling Technologies) for 1 h at room temperature. Bands were then detected using an enhanced chemiluminescent HRP substrate (supersignal west femto maximum sensitivity substrate; Fisher Scientific) and visualized after exposure onto HyBlot CL Autoradiography Film (Denville Scientific) and processing through an Ecomax X-ray film processor.

Metabolic assay

Cellular metabolism was measured using an XF24 cellular flux analyzer instrument from Seahorse Bioscience (North Billerica, MA, USA). Transfected cells were prepared as described above and adhered to the plate with Cell-Tak (1 μg/well, Corning). OXPHOS was measured using a Mitostress test kit (Agilent Technologies, Wilmington, DE) according to the manufacturer’s instructions. Oligomycin (1 μM), FCCP (2 μM) and rotenone/antimycin A (0.5 μM) were successively injected while measuring OCR. Basal respiration was defined as baseline OCR. Maximal respiration was defined as OCR after FCCP treatment. ATP production was calculated as the difference between basal respiration and OCR after oligomycin treatment, and spare capacity was calculated as the difference between maximal and basal respiration. Extracellular acidification rate (ECAR) was measured using the XFe24 Extracellular Flux Analyzer and the glycolysis stress test kit (Seahorse Bioscience). Glucose (1mM), oligomycin (2 μM) and 2-desoxy-D-glucose (0.5 μM) were then successively added while measuring ECAR. Glycolytic capacity was defined as ECAR after glucose and oligomycin injection.

RNA-seq

RNA was isolated using the RNeasy Micro kit (QIAGEN). All subsequent sample quality assessment was performed on a Fragment Analzyer electrophoresis system (Agilent). Total RNA was normalized prior to oligo-dT capture and cDNA synthesis with SMART-Seq v4 (Takara). The resulting cDNA was quantified using a Qubit 3.0 fluorometer (Life Technologies). Libraries were generated using the Nextera XT DNA Library Prep kit (Illumina). Medium depth sequencing (25 million reads per sample) was performed with a HiSeq 2500 (Illumina) on a Rapid Run flow cell using a 100 base pairs, Paired End run. Raw data files and processed files have been deposited in the Gene Expression Omnibus under accession number GSE136969.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical significance was determined by the Mann-Whitney (2-tailed) or Kruskal-Wallis test using the Prism 7 software (GraphPad) as indicated in the legends to the figures. A p value < 0.05 was considered statistically significant. EAE scores between groups were analyzed as disease burden per individual day with one-way-ANOVA and post hoc test as indicated. Sample sizes for each experiment are included in the figure legends.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-mouse CD4 PE-Cy7 (clone RM4-5)	Biolegend	Cat # 100528
anti-mouse IFNγ PE-CF594 (clone XMG1.2)	BD Biosciences	Cat # 562303
anti-mouse IL17 PE (clone TC11-18H10)	BD Biosciences	Cat # 559502
anti-mouse CD44 PE (clone IM7)	Thermo Fisher	Cat # 12-0441-83
anti-mouse Ly-6G, PE (clone 1A8)	BD Biosciences	Cat # 551461
anti-mouse Siglec-F, APC (clone 1RNM44N)	eBioscience	Cat # 50-1702-82
anti-mouse IL-4, APC (clone 11B11)	BD Biosciences	Cat # 557739
anti-mouse IL-13, PE (clone eBio13A)	eBioscience	Cat # 12-7133-82
anti-mouse IL-17, FITC (clone TC11-18H10)	BD Biosciences	Cat # 560220
anti-mouse IL-5, PE (clone TRFK5)	BD Biosciences	Cat # 554395
anti-mouse IL-10, APC (clone JES5-16E3)	BD Biosciences	Cat # 554468
anti-mouse IL-2, FITC (clone JES6-5H4)	BD Biosciences	Cat # 554427
anti-mouse CD62L, PE (clone MEL-14)	BD Biosciences	Cat #553151
anti-mouse CD8, APC (clone 53-6.7)	BD Biosciences	Cat #553035
anti-mouse CD4, BV421 (clone RM4-5)	Biolegend	Cat # 100563
anti-mouse CD25, PE-cy5 (clone PC61.5)	eBioscience	Cat # 15-0251-82
anti-mouse ICOS, APC (clone C398.4A)	BD Pharmingen	Cat # 565882
anti-mouse Nrp1, PE (clone 3E12)	Biolegend	Cat # 145204
anti-mouse CD44, BV510 (clone IM7)	BD Biosciences	Cat # 563114
anti-mouse GITR, PE-cy7 (clone DTA-1)	BD Biosciences	Cat # 558140
anti-mouse PD1, APC (clone J43)	Invitrogen	Cat # 17-9985-82
anti-mouse CTLA4, PE-cy7 (clone UC10-4B9)	Biolegend	Cat # 106314
anti-mouse Foxp3, FITC (clone FJK-16s)	Invitrogen	Cat # 11-5773-82
anti-mouse Tbet, APC (clone eBio4B10)	eBioscience	Cat # 50-5825-82
anti-mouse Gata3, PE (clone TWAJ)	eBioscience	Cat # 12-9966-42
anti-mouse RORγ, PE-TR (clone B20)	Invitrogen	Cat # 61-6981-82
anti-mouse Ki67, PE-cy7 (clone SolA15)	Invitrogen	Cat # 25-5698-82
anti-Rictor (clone 53A2)	CST	Cat # 2114
anti-Raptor (24C12)	CST	Cat # 2280
anti-Beta actin	CST	Cat # 4967
anti-pAKT (S473) (clone D9E)	CST	Cat # 4060
anti-pAKT (T308) (clone C31E5E)	CST	Cat # 2965
anti-AKT (clone C67E7)	CST	Cat # 4691
anti-pFoxo1a (S256)	CST	Cat # 9461
anti-Foxo1a (clone C29H4)	CST	Cat # 2880
anti-pS6 (clone D57.2.2E)	CST	Cat # 4858
Chemicals, Peptides, and Recombinant Proteins
Cockroach antigens	Greer Laboratory	Item# B26
Aluminum hydroxide (Alum)	MilliporeSigma	Cat # 239186
Anti-mouse CD3 mAb (clone 2C11)	Envigo Bioproducts	N/A
Transforming growth factor β (TGF-β)	PeproTech	Cat# 100-21
Diphtheria toxin (DTx)	Sigma-Aldrich	Cat # D0564
Pertussis toxin (PTx)	Sigma-Aldrich	Cat # P2980
Dexamethasone-Water Soluble (Dex)	Sigma-Aldrich	Cat # D2915
MOG (35–55)	Biosynthesis	Cat # 12668-01
PMA (phorbol 12-myristate 13-acetate)	Sigma-Aldrich	Cat # P1585
Ionomycin	Sigma-Aldrich	Cat # I9657
TRIzol	Invitrogen	Cat # 15596026
Power SYBR Green PCR Master Mix	Thermo Fisher	Cat # 4367659
Recombinant DNA
Myc_Rictor	Addgene	Plasmid #11367
Critical Commercial Assays
Mouse IL-10 DuoSet ELISA	R&D systems	Cat # DY417-05
Mouse IFN-gamma DuoSet ELISA	R&D systems	Cat # DY485-05
miScript II RT Kit	Qiagen	Cat # 218161
miScript SYBR Green PCR Kit	Qiagen	Cat # 218073
Cytometric Bead Array (CBA) Mouse Th1/Th2/Th17 Cytokine Kit	BD Biosciences	Cat # 560485
CBA mouse IL-13 Enhanced Sensitivity Flex Set	BD Biosciences	Cat # 562273
Quick-RNA™ Miniprep Kit	Zymo research	Cat # R1051
Mouse T cell nucleofector kit	Amaxa	Cat # VPA-1006
MojoSort™ Mouse CD4 Naïve T Cell Isolation Kit	Biolegend	Cat # 480040
SuperSignal West Femto Maximum Sensitivity Substrate	Thermo Fisher	Cat # 34095
Mouse: GAPDH FAM/MGB probe	Thermo Fisher	ID# 4352661
Mouse: Foxp3 FAM/MGB probe	Thermo Fisher	ID# Mm00475162_m1
Mouse: IFNg FAM/MGB probe	Thermo Fisher	ID# Mm01168134_m1
Mouse: IL17 FAM/MGB probe	Thermo Fisher	ID# Mm00439618_m1
Mouse: Rragd FAM/MGB probe	Thermo Fisher	ID# Mm00546741_m1
Mouse: Galnt9 FAM/MGB probe	Thermo Fisher	ID# Mm01196716_m1
Mouse: Esam FAM/MGB probe	Thermo Fisher	ID# Mm00518378_m1
Mouse: Evl FAM/MGB probe	Thermo Fisher	ID# Mm00468405_m1
Mouse: Nabp FAM/MGB probe	Thermo Fisher	ID# Mm00658947_m1
Mouse: Rictor FAM/MGB probe	Thermo Fisher	ID# Mm01307318_m1
Mouse: Raptor FAM/MGB probe	Thermo Fisher	ID# Mm01242613_m1
Mouse: Dusp1 FAM/MGB probe	Thermo Fisher	ID# Mm00457274_g1
Mouse: Fkbp5 FAM/MGB probe	Thermo Fisher	ID# Mm00487406_m1
Mouse: Muc5ac FAM/MGB probe	Thermo Fisher	ID# Mm01276718_m1
Mouse: Muc5b FAM/MGB probe	Thermo Fisher	ID# Mm00466391_m1
Mouse: Hk2 FAM/MGB probe	Thermo Fisher	ID# Mm00443385_m1
Mouse: Hif1a FAM/MGB probe	Thermo Fisher	ID# Mm00468869_m1
Mm_miR-15b_2 miScript Primer Assay	Qiagen	ID# MS00011235
Mm_miR-16_2 miScript Primer Assay	Qiagen	ID# MS00037366
Mm_miR-146_1 miScript Primer Assay	Qiagen	ID# MS00001638
Mm_miR-155_1 miScript Primer Assay	Qiagen	ID# MS00001701
Mm_miR-342_1 miScript Primer Assay	Qiagen	ID# MS00002184
5 nmol IDT miRNA Inhibitor_NC5; mG/ZEN/mCmGmAmCmUmAmUmAmCmGmCmGmCmAmAmUmAmUmGmG/3ZEN/	IDT	Ref # 210223534
5 nmol IDT miRNA Inhibitor_mmu-miR342-3p; mA/ZEN/mCmGmGmGmUmGmCmGmAmUmUmUmCmUmGmUmGmUmGmAmG/3ZEN/	IDT	Ref # 210223535
mirVan miRNA Mimic, Negative Control #1	Thermo Fisher	Cat # 4464058
mirVana miR-342 Mimic	Thermo Fisher	ID#MC12328
Accell Mouse Rictor siRNA	Horizon Discovery	Cat # A06459813-0005
Seahorse XF Glycolysis Stress Test Kit	Agilent	Cat # 103020-100
Seahorse XF Cell Mito Stress Test Kit	Agilent	Cat # 103015-100
Deposited Data
RNAseq	This paper	GEO: GSE136969
Experimental Models: Organisms/Strains
C57BL/6 mice (B6)	Jackson Laboratory	JAX: 000664
B6 Foxp3GFP (B6 background)	Generated by V. Kuchroo and Provided by Y. Belkaid (Bettelli et al., 2006)	N/A
B6.129(Cg)-Foxp3tm3(DTR/GFP)Ayr/J (B6 background)	Jackson Laboratory	JAX: 016958
B6.Cg-Nr3c1tm1.1Jda/J (B6 background)	Jackson Laboratory	JAX: 021021
B6.129(Cg)-Foxp3tm4(YFP/icre)Ayr/J (B6 background)	Jackson Laboratory	JAX: 016959
Oligonucleotides
GAPDH forward: 5’- ATGCCTGCTTCACCACCTTCT-3’	IDT	N/A
GAPDH reverse: 5’- CATGGCCTTCCGTGTTCCTA-3’	IDT	N/A
common GR forward: 5’-AAAGAGCTAGGAAAAGCCATTGTC-3’	IDT	N/A
GRα reverse: 5’-TCAGCTAACATCTCTGGGAATTCA-3’	IDT	N/A
Gilz forward: 5’- AATGCGGCCACGGATG -3’	IDT	N/A
Gilz reverse: 5’- GGACTTCACGTTTCAGTGGACA -3’	IDT	N/A
Software and Algorithms
Prism 7	GraphPad	https://www.graphpad.com/scientific-software/prism/
Flowjo	TreeSta	https://www.flowjo.com/solutions/flowjo/downloads
Seahorse Wave	Seahorse Wave

Highlights.

• Without Foxp3⁺ Treg cells, glucocorticoid loses its therapeutic effects

• Glucocorticoid induces miR-342–3p expression specifically in Treg cells

• miR-342–3p targets Rictor and regulates the metabolic programming in Treg cells

• miR-342–3p and Rictor expression controls Treg cell function in vivo