RORγt expression in T_regs promotes systemic lupus erythematosus via IL‐17 secretion, alteration of T_reg phenotype and suppression of Th2 responses

Summary

Systemic lupus erythematosus (SLE) is a common autoimmune disorder with a complex and poorly understood immunopathogenesis. However, a pathogenic role for the T helper type 17 (Th17) axis was demonstrated by many studies, while regulatory T cells (T_regs) were shown to mediate protection. Recently, we and others characterized a novel and independent T cell population expressing both the T_reg characteristic transcription factor forkhead box protein 3 (FoxP3) and the Th17‐defining retinoic acid receptor‐related orphan nuclear receptor γt (RORγt). Studies in a model of acute glomerulonephritis unveiled potent regulatory, but also proinflammatory, functions of RORγt⁺FoxP3⁺ T_regs. This bi‐functional nature prompted us to suggest the name ‘biT_regs’. Importantly, the pathogenic biT_reg effects were dependent upon expression of RORγt. We thus aimed to evaluate the contribution of RORγt⁺FoxP3⁺ biT_regs to pristane‐induced SLE and explored the therapeutic potential of interference with RORγt activation. Our analyses revealed expansion of IL‐17 producing biT_regs in a distinctive time–course and organ‐specific pattern, coincident with the development of autoimmunity and tissue injury. Importantly, specific ablation of RORγt activation in endogenous biT_regs resulted in significant amelioration of pristane‐induced pulmonary vasculitis and lupus nephritis. As potential mechanisms underlying the observed protection, we found that secretion of IL‐17 by biT_regs was abrogated completely in FoxP3^Cre × RORC^fl/fl mice. Furthermore, T_regs showed a more activated phenotype after cell‐specific inactivation of RORγt signalling. Finally, and remarkably, biT_regs were found to potently suppress anti‐inflammatory Th2 immunity in a RORγt‐dependent manner. Our study thus identifies biT_regs as novel players in SLE and advocates RORγt‐directed interventions as promising therapeutic strategies.

Introduction

Systemic lupus erythematosus (SLE) is a complex and relatively common autoimmune disorder, which can affect multiple organs including skin, joints, lungs, kidneys and the central nervous system 1, 2. As SLE causes high morbidity and mortality in a rather young collective of patients, the search for new therapeutic strategies is a priority for scientists worldwide 3, 4. Despite intensive research, the events which lead to development of SLE and eventually cause the associated organ pathologies still remain widely elusive 5, 6. However, a central role for CD4⁺ T helper cells in disease pathogenesis has been demonstrated by multiple studies in mice and humans, including our own 7, 8, 9, 10, 11. Recently, we and others particularly highlighted the importance of the T helper type 17/interleukin‐17 (Th17/IL‐17) axis for systemic autoimmunity and lupus nephritis 10, 12, 13. Importantly, mice deficient in IL‐17A or F were shown to be protected from disease in three independent studies, using the murine model of pristane‐induced SLE 14, 15, 16. Furthermore, recent studies could identify a specialized regulatory T cell subset, which is tailor‐made for down‐regulation of pathogenic Th17 responses 17, 18. Lack of these T regulatory 17 (T_reg17) cells resulted in much increased SLE associated mortality and organ pathologies 19. A crucial role for both, Th17 cells and T_regs in SLE is therefore evident. Interestingly, we and others could recently describe an intriguing novel T cell subset, expressing the unusual combination of the T_reg master transcription factor forkhead box protein 3 (FoxP3), together with the Th17 master regulator retinoic acid receptor‐related orphan nuclear receptor γt (RORγt) 20, 21, 22. Functional characterization revealed that these RORγt FoxP3 double‐positive T cells displayed both regulatory as well as proinflammatory functions in a model of acute crescentic glomerulonephritis 22. RORγt⁺FoxP3⁺ T cells were shown to produce the proinflammatory cytokine IL‐17, but at the same time they secreted high amounts of immunoregulatory IL‐10, transforming growth factor (TGF)‐β and IL‐35. Furthermore, they potently suppressed T effector cell (T_eff) responses in vitro, identifying them as T_regs 20, 22. Given their bi‐functional properties, we proposed to name these RORγt⁺FoxP3⁺ T cells ‘biT_regs’. In accordance with their unique cytokine profile and transcription factor expression, biT_regs showed both pro‐ and anti‐inflammatory in‐vivo functions in crescentic glomerulonephritis 22. Exogenous biT_reg transfer ameliorated renal inflammation to a similar extent as conventional RORγt‐negative T_regs, although the suppressive mechanisms seem to be different. Conversely, endogenous biT_regs displayed additional proinflammatory functions. Interestingly, these proinflammatory biT_reg functions depended upon activation of RORγt 22. This is clinically highly relevant, as multiple pharmaceutical companies are currently developing RORγt blocking agents and some Phases I and II studies are already ongoing 23, 24, 25. Furthermore, a landmark study which was published during preparation of this paper confirmed the biological relevance of biT_regs and surprisingly identified them as potent down‐regulators of anti‐inflammatory Th2 immunity 26. This effect, which might explain some of the proinflammatory biT_reg functions, also depended upon activation of the transcription factor RORγt. Given their multiple immune modulatory functions, biT_regs are highly likely to contribute to development of SLE. Because, however, nothing is known to date about the clinical relevance of biT_regs, we decided to study their role in SLE. A major hallmark of human SLE is activation of the Type I interferon pathway, as has just recently been highlighted again 27. We thus chose to use the pristane model of SLE, which is currently the only available murine model characterized by a strong interferon signature 28. In particular, our study addressed the following aspects: (1) characterization of biT_reg dynamics in the different organ systems affected during the course of SLE, (2) analysis of the role of RORγt in biT_regs with special focus on IL‐17 secretion and regulation of Th2 immunity and (3) assessment of the contribution of biT_regs to organ pathologies in pristane‐induced SLE.

Materials and methods

Animals

LoxP‐site flanked RORC^fl/fl mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). FoxP3^YFP‐Cre mice were a kind gift from Alexander Y. Rudensky (Memorial Sloan‐Kettering Cancer Center, New York, NY, USA). All animals used in this study were on a C57BL/6 background and were raised under specific pathogen‐free conditions at our animal facility.

Animal experiments and functional studies

Pristane disease was induced in 8–10‐week‐old male or female FoxP3^Cre × RORC^fl/fl mice and matched FoxP3^Cre × RORC^wt/wt (referred to as FoxP3^Cre) littermate controls by single intraperitoneal (i.p.) injection of 500 µl of pristane oil (2,6,10,14‐tetramethylpentadecane; Sigma‐Aldrich, St Louis, MO, USA) 14, 19. Organs were harvested between 1 week and 9 months after injection, as indicated. Animal experiments were performed according to national and institutional animal care and ethical guidelines and were approved by local committees (approval codes 37/11, 45/12, 73/14 and 07/15).

Morphological studies

Glomerular abnormalities were determined in 50 glomeruli per mouse in a blinded manner, as published previously 14, 29. These included glomerular hypercellularity, crescent formation, fibrinoid necrosis, segmental proliferation, hyalinosis and capillary wall thickening. Lung tissue was perfused with 500 µl formalin and fixed overnight, washed with ethanol, paraffin‐embedded and periodic acid‐Schiff (PAS)‐stained 30. Leucocyte infiltration, granuloma formation, alveolar‐wall thickening and alveolar haemorrhage were assessed in a semi‐quantitative score according to the percentage of tissue affected (0–25% = 1, 25–50% = 2, 50–75% = 3, >75% = 4). Mean scores from 15 high‐power fields were calculated. Numbers of pulmonary granulomas were determined in 10 low‐power fields (lpf, magnification × 100) per tissue section; their size was quantified using the Zeiss Axio Vision software (Carl Zeiss, Jena, Germany). For immunohistochemistry, tissue was stained with antibodies against CD3 (A0452; Dako, Hamburg, Germany), F4/80 (BM8; BMA Biomedicals, Hiddenhausen, Germany), MAC2 (M3/38; Cedarlane‐Laboratories, Burlington, ON, Canada), granulocyte‐differentiation antigen‐1 (GR‐1) (NIMP‐R14; Hycult Biotech, Uden, the Netherlands), FoxP3 (FJK‐16s; eBioscience, San Diego, CA, USA) or KI67 (D3B5; Cell Signaling, Danvers, MA, USA) and developed with a polymer‐based secondary antibody–alkaline phosphatase kit (POLAP; Zytomed, Berlin, Germany), as published previously. Fifty glomerular cross‐sections (gcs) per kidney and 30 high‐power fields (hpf, magnification × 400) per kidney and lung section were counted in a blinded fashion.

Isolation of leucocytes from various tissues

Spleens were harvested in Hanks’ balanced salt solution (HBSS) and passed through 70‐µm nylon meshes. After lysis of erythrocytes, cells were washed and passed over 40‐µm meshes, counted and resuspended in phosphate‐buffered saline (PBS) for either culture or fluorescence activated cell sorter (FACS) analysis. Kidneys were minced and incubated in digestion medium [RPMI‐1640 medium containing 10% fetal calf serum (FCS), 1% HEPES, 1% penicillin/streptomycin, 8 µg/ml collagenase D and 0·4 µg/ml DNase] at 37°C for 40 min. Kidney tissue was then homogenized using a gentle magnetic affinity cell sorter (MACS) dissociator (Miltenyi Biotech, Bergisch Gladbach, Germany) and purified using Percoll‐gradient centrifugation 31. The peritoneal cavity was washed with 5 ml ice‐cold PBS to harvest cells at the indicated time‐points after pristane administration 19. For analysis of leucocytes from lungs, organs were removed, minced and incubated in digestion medium at 37°C for 45 min. Tissues were then homogenized using a gentle MACS dissociator (Miltenyi Biotech), filtered over a 70‐µm mesh and purified using Percoll‐gradient centrifugation 30. Peripheral blood was drawn into ethylenediamine tetraacetic acid (EDTA)‐coated tubes and red blood cell lysis was performed.

Systemic humoral immune responses

Circulating anti‐ds‐DNA and anti‐U1‐ribonucleoprotein (RNP) antibodies from serum were analysed by ELISA at the indicated dilutions after coating microtitre plates with either poly‐L‐lysine (Sigma‐Aldrich) and calf thymus DNA (Worthington) or U1‐RNP (Arotec). For analysis of total non‐antigen‐specific immunoglobulins, ELISA plates were precoated with anti‐mouse immunoglobulin (Ig)G antibodies (Jackson Immuno Research). For all analyses, serum samples were applied at the dilutions indicated. The following secondary antibodies were used for detection: total IgG, IgG1 (both Southern Biotech), IgG2b (Invitrogen), IgG2c (Bethyl) and IgG3 (Jackson Immuno Research). For detection of IgE, sera were diluted 1 : 200 and a commercially available ELISA was utilized (Biolegend, San Diego, CA, USA), following the manufacturer's instructions.

Analyses of cytokines from serum and spleen cell cultures

Splenocytes (4 × 10⁶ cells/ml) were cultured under standard conditions in the presence of 1 µg/ml anti‐CD3 (eBioscience, San Diego, CA, USA) and supernatants were harvested after 72 h. Commercially available ELISAs were used for detection of IFN‐γ and IL‐17A (Biolegend). Serum was obtained from mice at 5 and 9 months after pristane induction and was analysed for levels of IL‐17 and IFN‐γ by ELISA (Biolegend) 19.

Flow cytometry

Cells were stained with fluorochrome‐labelled antibodies against CD45, CD3, CD4, CD8, CD19, CD44, CD69, CD62L, CCR6, γδTCR, inducible T cell co‐stimulator (ICOS), glucocorticoid‐induced TNFR family‐related gene (GITR), CD103, cytotoxic T lymphocyte‐associated protein 4 (CTLA‐4), CXCR3 (eBioscience), C‐X‐C chemokine receptor type 5 (CXCR5), programmed death 1 (PD‐1), IgD, IgM, CD138 and PD‐L1 (Biolegend). For intracellular and intranuclear staining, samples were processed using a commercial intranuclear staining kit (FoxP3 kit; eBioscience). Fluorochrome‐labelled antibodies against IL‐4, IL‐5, IL‐13, IL‐17, IFN‐γ, FoxP3 (eBioscience), RORγt, KI67 (both BD Biosciences, Heidelberg, Germany) and Gata‐3 (Biolegend) were employed, as published recently 19, 22. For cytokine staining, cells were activated with phorbol myristate acetate (PMA) (50 ng/ml; Sigma‐Aldrich) and ionomycin (1 µg/ml; Calbiochem‐Merck, Temecula, CA, USA) for 3.5 h. After 30 min of incubation, brefeldin A (10 µg/ml; Sigma‐Aldrich) was added. LIVE/DEAD staining (Invitrogen Molecular Probes, Eugene, OR, USA) was used to exclude dead cells during flow cytometry. Experiments were performed on a BD LSRII Cytometer (Becton Dickinson, Heidelberg, Germany).

T_reg suppression assay

CD4⁺ spleen cells were enriched using magnetic‐activated cell sorting according to the manufacturer's protocol (MACS CD4⁺ T cell Kit II; Miltenyi Biotec). T_regs and T_eff cells were isolated by FACS sorting (performed on a BD ARIAIII Cytometer; Becton Dickinson). A total of 1 × 10⁵ CD45⁺CD4⁺yellow fluorescent protein (YFP)⁻ effector T cells from FoxP3^Cre mice were then cultured for 72 h in anti‐CD3 monoclonal antibody (mAb) (5 µg/ml; BD Biosciences) precoated 96‐well plates either alone or in co‐culture with CD45⁺CD4⁺YFP⁺ T_regs from FoxP3^Cre or FoxP3^Cre × RORC^fl/fl mice at the ratios indicated. Suppressive capacity was determined by IL‐2 ELISA performed from the supernatants, as published recently 17, 22, 32. For analyses of IFN‐γ, IL‐10, IL‐4, IL‐13 and TNF‐α in the supernatants, cytometric bead array assays were performed using a commercial kit (LEGENDplex^TM mouse Th cytokine mix and match subpanel; BioLegend). To assess T_eff cell proliferation, cells from the above culture experiments were harvested and analysed for their KI67 expression by FACS.

Quantitative real‐time polymerase chain reaction (PCR) analysis

Total RNA of renal cortex and spleen tissue was isolated according to a standard Trizol protocol and purified utilizing a Nucleospin kit (Macherey & Nagel, Düren, Germany). Real‐time PCR was performed after cDNA transcription as described previously (all primer sequences are available upon request) and results were normalized to expression of 18S rRNA 17.

Statistical analyses

Results are expressed as mean ± standard error of the mean (s.e.m.). Groups were compared by Student's t‐test. A P‐value < 0·05 was considered statistically significant.

Results

IL‐17⁺ biT_regs expand during systemic pristane‐induced lupus

In order to characterize more clearly the pristane model of SLE, we initially performed time–course analyses of the multiple organ manifestations. After pristane injection, mice develop acute sterile peritonitis and non‐immune complex pulmonary capillaritis. Approximately 3 weeks later, immunologically active peritoneal and pulmonary granulomas become apparent. Subsequently, autoimmunity is established with increasing serum levels of various autoantibodies, which we found to be present already in low amounts at week 8. Deposition of immune complexes then leads to progressive development of lupus nephritis which, according to our analyses, became evident by light microscopy at approximately 5 months (data not shown). In Fig. 1a a schematic overview is depicted. In order to assess the potential role of biT_regs, we analysed their frequencies in multiple organs before and at different time–points after injection of pristane. biT_regs were found to be present in both peritoneal lavage cells and lungs of healthy animals. In acute antigen‐independent sterile peritonitis and pulmonary vasculitis, their population expanded only slightly (Fig. 1b,c). However, at later time‐points, coincident with development of peritoneal and pulmonary granulomas, biT_reg percentages increased significantly (2·5 ± 0·27% of peritoneal FoxP3⁺ T_regs at baseline versus 5·56 ± 0·98% at 5 weeks and 7·57 ± 0·99% at 8 weeks; 5·5 ± 0·46% of pulmonary FoxP3⁺ T_regs at baseline, 18·96 ± 3·17% at 5 weeks and 21·4 ± 0·86% of FoxP3⁺ T_regs at week 8) (Fig. 1b,c). In contrast, spleens which are only mildly affected during pristane‐induced SLE showed merely a slight expansion of biT_regs (Fig. 1d). Finally, we studied biT_regs during the course of immune complex lupus nephritis. Interestingly, our analyses revealed a distinctive pattern of renal infiltration. While only few biT_regs were detected at 2 months after pristane injection (1·3 ± 0·38% of FoxP3⁺ T_regs), their frequencies increased to reach a maximum at 5 months (10·62 ± 1·7% of FoxP3⁺ T_regs). Afterwards, biT_reg proportions declined steadily and were almost back to baseline levels (0·3 ± 0·04% of FoxP3⁺ T_regs) at 9 months (1·3 ± 0·38% of FoxP3⁺ T_regs) (Fig. 1e). A significant proportion of biT_regs produced IL‐17 in all examined organs at all time‐points (Fig. 1b–e). A representative FACS plot of pulmonary biT_regs, including their IL‐17 secretion, is shown in Fig. 1f.

Figure 1.

(a) The temporal course of pristane‐induced inflammation and organ‐specific pathologies is shown. Intraperitoneal injection leads to development of acute innate mediated peritonitis and non‐immune complex pulmonary capillaritis. Subsequently peritoneal and pulmonary granulomas develop and autoantibody formation is initiated. During the following months, immune complex lupus nephritis develops. (b) Retinoic acid receptor‐related orphan nuclear receptor γt (RORγt)⁺forkhead box protein 3 (FoxP3)⁺ bi‐functional regulatory T cell (biT_reg) percentages among FoxP3⁺ peritoneal lavage cells at the indicated time‐points after pristane injection (left) and interleukin (IL)‐17 production by biT_regs (right). (c) Pulmonary (d) splenic and (e) renal biT_reg percentages among FoxP3⁺ T_regs and percentages of IL‐17⁺ biT_regs at the indicated time‐points after pristane injection. (f) A representative fluorescence activated cell sorter (FACS) plot from inflamed lungs at 5 weeks after pristane injection shows RORγt⁺FoxP3⁺ biT_regs, RORγt⁺FoxP3^– T helper type 17 (Th17) cells and FoxP3⁺RORγt^– conventional T_regs (cT_regs). IL‐17 production by biT_regs is shown on the right. Numbers indicate percentages of FoxP3⁺ cells. Dotted lines indicate basal biT_reg percentages in healthy mice. Bars show mean ± standard error of the mean (s.e.m.). *P < 0·05; **P < 0·01; ***P < 0·001 versus basal percentages.

RORγt activation in biT_regs aggravates pulmonary capillaritis

In order to study the role of RORγt in biT_regs, we generated FoxP3^Cre × RORC^fl/fl mice, which specifically lack RORγt activation in T_regs. As pristane‐injected mice develop severe, often lethal, haemorrhagic pulmonary capillaritis, we first studied lung pathology in FoxP3^Cre × RORC^fl/fl and FoxP3^Cre control mice. At 3 weeks after pristane injection, no macroscopic or histological differences were noted (Supporting information, Fig. S1a,b). However, at later stages, when biT_regs expand and pulmonary inflammation becomes granulomatous, we found striking protection of mice lacking RORγt in T_regs. Macroscopic pulmonary haemorrhage was reduced significantly in FoxP3^Cre × RORC^fl/fl mice (haemorrhage score 2·9 ± 0·3 versus 1·67 ± 0·3) (Fig. 2a). In line with this, histological signs of vasculitis were ameliorated (vasculitis score 1·63 ± 0·3 versus 0·77 ± 0·2) (Fig. 2a). Interestingly, granulomas were also reduced in numbers (6 ± 1·0 versus 2·5 ± 0·6) and size (41·23 ± 4·3 versus 26·88 ± 5·4 × 10³ µm²) in mice with T_reg‐specific deficiency of RORγt (Fig. 2b). Furthermore, our analyses revealed a significant reduction of infiltrating T cells, both by immunohistochemistry (17·16 ± 1·6 versus 11·52 ± 1·4 CD3⁺ cells/hpf) as well as FACS analysis (55·66 ± 1·5 versus 49·48 ± 2·4% CD4⁺ of total CD3⁺ cells) of lungs (Fig. 2c). Similarly, numbers of GR‐1‐positive neutrophils were also reduced in FoxP3^Cre × RORC^fl/fl mice (9·72 ± 0·7 versus 6·8 ± 0·4 GR‐1⁺ cells/hpf) (Fig. 2d).

Figure 2.

(a) Representative photographs of periodic acid‐Schiff (PAS)‐stained lungs from indicated mouse strains at 8 weeks after pristane injection (left). Quantification of macroscopic haemorrhage as well as vasculitis score is shown (right) (original magnification ×200). (b) Representative photographs of pulmonary granulomas and quantification of granuloma numbers and size (original magnification ×400). (c) Immunohistochemical staining for pan T cell marker CD3 (left). Quantification of pulmonary CD3⁺ T cell numbers by immunohistochemistry and CD4⁺ T helper cell percentages by fluorescence activated cell sorter (FACS) as indicated (original magnification ×200). (d) Immunohistochemical staining and quantification of infiltrating granulocyte‐differentiation antigen‐1 (GR‐1)⁺ neutrophils (original magnification ×200). A = alveolus; B = bronchus; V = vessel; arrows indicate pristane droplets. Squares represent individual animals, horizontal lines indicate means. Bars show mean ± standard error of the mean (s.e.m.). *P < 0·05; **P < 0·01.

Lupus nephritis is ameliorated in FoxP3^Cre × RORC^fl/fl mice

At later stages, pristane‐injected mice develop proliferative immune complex glomerulonephritis. We thus studied renal disease and found significant amelioration of lupus nephritis at 9 months in FoxP3^Cre × RORC^fl/fl mice, as evidenced by protection from glomerular injury (58·9 ± 6·3 versus 26·18 ± 4·4% abnormal glomeruli) (Fig. 3a). In line with this, glomerular cell proliferation, a hallmark of lupus nephritis, was also reduced (1·0 ± 0·07 versus 0·67 ± 0·07 KI67⁺ cells/gcs) (Fig. 3b). Importantly, protection from histological damage also resulted in less functional injury, as indicated by lower blood urea nitrogen (BUN) levels (24·4 ± 1·2 versus 19·6 ± 1·6 mg/dl) (Fig. 3c). Finally, we found that loss of RORγt activation in T_regs also protected from development of albuminuria (239·5 ± 20·4 versus 151·7 ± 18·7 mg/g albumin per creatinine) (Fig. 3d).

Figure 3.

(a) Representative photographs of periodic acid‐Schiff (PAS)‐stained kidney sections at 9 months after pristane injection, as well as quantification of glomerular injury (original magnification ×400). (b) Representative photographs of immunohistochemical staining of KI67 and quantification of glomerular KI67⁺ cells (original magnification ×400). (c) Quantification of serum blood urea nitrogen (BUN) levels. (d) Quantification of urinary albumin/creatinine ratios at the indicated time‐points before and after injection of pristane. Bars show mean ± standard error of the mean (s.e.m.). Squares represent individual animals, horizontal lines indicate means. *P < 0·05; **P < 0·01.

Renal inflammatory cell infiltration is reduced in mice with RORγt‐deficient T_regs

Because infiltrating renal leucocytes are key players of renal tissue injury, we next analysed renal inflammatory cell infiltration. In line with amelioration of renal damage, immunohistochemistry revealed a significant reduction of both glomerular (2·9 ± 0·3 versus 1·84 ± 0·2 CD3⁺ cells/gcs) and interstitial (29·58 ± 1·1 versus 24·08 ± 1·5 CD3⁺ cells/hpf) T cells in FoxP3^Cre × RORC^fl/fl mice (Fig. 4a). Percentages of regulatory T cells, however, were not significantly different (0·022 ± 0·002 versus 0·017 ± 0·002 FoxP3⁺/CD3⁺ cells per hpf) (Fig. 4b). Similarly, we found unchanged proportions of renal Th1 and Th17 cells (Supporting information, Fig. S2a,b). Interestingly, however, percentages of total IL‐17⁺ renal leucocytes were significantly lower in kidneys of FoxP3^Cre × RORC^fl/fl mice (2·11 ± 0·16 versus 1·6 ± 0·07% IL‐17⁺ of CD45⁺ cells), indicating reduced influx of IL‐17⁺ populations different from Th17 cells (Fig. 4c). Furthermore, we also detected lower numbers of glomerular (3·74 ± 0·55 versus 2·2 ± 0·27 MAC‐2⁺ cells/gcs) (Fig. 4d) and interstitial (21·53 ± 2·1 versus 13·86 ± 1·2 F4/80⁺ cells/hpf) (Fig. 4e) macrophages in mice with T_reg‐specific deletion of RORγt.

Figure 4.

Representative photographs and quantification of glomerular and interstitial renal (a) CD3⁺ T cell and (b) forkhead box protein 3 (FoxP3)⁺ regulatory T cell (T_reg) infiltration at 9 months after pristane injection (original magnification ×400). (c) Fluorescence activated cell sorter (FACS) analysis of total renal infiltrating interleukin (IL)−17⁺ leucocytes (one of two sets shown). (d) Representative photographs and quantification of glomerular MAC2⁺ monocyte/macrophage infiltration (original magnification ×400). (e) Representative photographs and quantification of renal interstitial F4/80⁺ monocyte/macrophage infiltration (original magnification ×200). Squares represent individual animals, horizontal lines indicate means. *P < 0·05; **P < 0·01.

Humoral autoimmunity remains unaltered in FoxP3^Cre × RORC^fl/fl mice

Next, we wanted to explore the mechanisms leading to tissue protection from pristane‐induced SLE in FoxP3^Cre × RORC^fl/fl mice. We therefore assessed the effects of RORγt deactivation in biT_regs on humoral immune responses. In this respect, we found unaltered levels of serum total IgG as well as identical amounts of all analysed subclasses at 9 months after pristane injection (Supporting information, Fig. S3a). Similarly, formation of IgG autoantibodies of all subclasses against dsDNA (Fig. 5a) as well as U1‐RNP was unchanged in FoxP3^Cre × RORC^fl/fl mice (Fig. 5b). In line with unchanged serum antibody levels, we documented a similar immune complex deposition in kidneys of pristane‐injected mice, as evidenced by complement C3 (0·84 ± 0·1 versus 0·59 ± 0·05 C3 deposition score) (Fig 5c) and mouse IgG (mIgG) staining (0·53 ± 0·06 versus 0·46 ± 0·05 mIgG deposition score) (Fig. 5d). Similarly, analyses of splenic B cell populations showed similar percentages of total B cells (16·7 ± 1·0 versus 16·38 ± 2·2% CD19⁺CD138^– of CD45⁺ cells), switched memory B cells (9·24 ± 3·6 versus 14·55 ± 1·3% IgM^–IgD^– of CD19⁺CD138^– cells) as well as plasma cells (1·34 ± 0·39 versus 0·52 ± 0·07% CD138^high of CD45⁺ cells) in both strains of mice (Supporting information, Fig. S3b–d). Finally, we found that percentages of splenic T follicular helper cells (TfH) were also unaffected by abrogation of RORγt activation in T_regs (1·3 ± 0·29 versus 2·48 ± 0·85% CXCR5⁺PD‐1⁺ of CD4⁺FoxP3^– cells) (Supporting information, Fig. S3e).

Figure 5.

(a) Quantification of immunoglobulin (Ig)G anti‐ds‐DNA antibodies, as well as the indicated IgG subclasses from serum of forkhead box protein 3 (FoxP3)^Cre wild‐type and FoxP3^Cre × retinoic acid receptor‐related orphan nuclear receptor (ROR)C^fl/fl mice in serial dilutions by enzyme‐linked immunosorbent assay (ELISA). (b) Quantification of serum anti‐U1‐ribonucleoprotein (RNP) autoantibodies of the indicated IgG subclasses in serial dilutions by ELISA. (c) Immunohistochemical staining and quantification of glomerular complement C3 deposition (original magnification ×400). (d) Immunohistochemical staining and quantification of glomerular mouse IgG (mIgG) deposition (original magnification ×400). Squares represent individual animals, horizontal lines indicate means.

IL‐17 secretion by biT_regs is dependent upon RORγt

In a next step, we aimed to explore alterations of cellular immune responses. As IL‐17 is a known downstream target of RORγt, we analysed IL‐17 expression in FoxP3^Cre × RORC^fl/fl mice. Analysis of peritoneal lavage cells at 3 weeks after pristane injection confirmed specific and complete absence of RORγt in FoxP3⁺ T_regs (4·58 ± 1·5 versus 0·28 ± 0·3% biT_regs of FoxP3⁺ cells) (Fig. 6a, left). In line with our hypothesis, IL‐17 production by T_regs was abrogated totally in FoxP3^Cre × RORC^fl/fl mice, indicating absolute dependency on RORγt (2·33 ± 0·48 versus 0·15 ± 0·06% IL‐17⁺ of FoxP3⁺ cells) (Fig. 6a, right). Similarly, analysis of vasculitic lungs showed absence of RORγt⁺ biT_regs (4·06 ± 1·0 versus 0·55 ± 0·33% biT_regs of FoxP3⁺ cells) as well as abolished IL‐17 production by T_regs (2·9 ± 0·74 versus 0·35 ± 0·19% IL‐17⁺ of FoxP3⁺ cells) (Fig. 6b). Finally, we analysed nephritic kidneys at 5 months after pristane injection and again found absence of biT_regs in FoxP3^Cre × RORC^fl/fl mice (10·62 ± 1·7 versus 0·8 ± 0·17% biT_regs of FoxP3⁺ cells) as well as abrogation of IL‐17 secretion by renal FoxP3⁺ T_reg cells (0·92 ± 0·27 versus 0·24 ± 0·09% IL‐17⁺ of FoxP3⁺ cells) (Fig. 6c).

Figure 6.

(a) A representative fluorescence activated cell sorter (FACS) plot and quantification of peritoneal lavage bi‐functional regulatory T cells (biT_regs) (left) as well as interleukin (IL)‐17 production by biT_regs (right) at 3 weeks after pristane injection in forkhead box protein 3 (FoxP3)^Cre wild‐type and FoxP3^Cre × retinoic acid receptor‐related orphan nuclear receptor (ROR)C^fl/fl mice (right). (b) Quantification of pulmonary biT_regs (left) and IL‐17 production (right) at 3 weeks after pristane injection in the indicated mouse strains. (c) A representative FACS plot and quantification of renal biT_regs (left) and their IL‐17 production (right) at 5 months after pristane injection. Numbers in FACS plots indicate percentages of FoxP3⁺ cells. Squares represent individual animals, horizontal lines indicate means. *P < 0·05; **P < 0·01; ***P < 0·001.

RORγt deficiency alters the phenotype of T_regs

In order to evaluate whether RORγt also affects intrinsic T_reg functions, we studied FoxP3 protein expression on the single‐cell level. Interestingly, we found significantly higher FoxP3 mean fluorescence intensities in T_regs from the blood of naive FoxP3^Cre × RORC^fl/fl mice [1376 ± 24 versus 1567 ± 36 FoxP3 mean fluorescence intensity (MFI)]. The same was true for T_regs from both peritoneal lavage cells (702·6 ± 79·8 versus 1212 ± 200·3 FoxP3 MFI) and lungs (500·7 ± 55·7 versus 778·5 ± 120 FoxP3 MFI) 3 weeks after pristane injection (Fig. 7a). Detailed analysis of T_reg activation marker molecules revealed enhanced surface expression of CD103 (7·9 ± 1·0 versus 15·33 ± 1·7% of FoxP3⁺ cells), GITR (7·67 ± 0·68 versus 22·9 ± 1·1% of FoxP3⁺ cells) and ICOS (9·06 ± 0·91 versus 21·95 ± 2·14% of FoxP3⁺ cells) on RORγt‐deficient T_regs. Levels of CTLA‐4 and PD‐L1 remained unaffected (Fig. 7b). In congruence with elevated expression of activation markers, we found higher T_reg activation (37·07 ± 2·6 versus 45·53 ± 0·86% CD69⁺CD62L^– and 21·68 ±1·2 versus 36·57 ± 2·5% CD44⁺CD62L^– of FoxP3⁺ cells) (Fig. 7c) as well as proliferative activity (77·09 ± 2·1 versus 83·43 ± 1·5% KI67⁺ of FoxP3⁺ cells) (Fig. 7d). In‐vitro suppression of cytokine production by T_regs, including IL‐2, Th1 characteristic IFN‐γ, Th2 prototype IL‐4 and IL‐13, as well as Th17‐associated TNF‐α, were unchanged in co‐culture experiments with T_eff cells (Supporting information, Fig. S4a–d). Induction of IL‐10 production was also similar between the groups (Supporting information, Fig. S4e). Finally, suppression of T_eff cell proliferation was unaffected in FoxP3^Cre × RORC^fl/fl mice (Supporting information, Fig. S4f).

Figure 7.

(a) Quantification of forkhead box protein 3 (FoxP3) mean fluorescence intensity (MFI) in regulatory T cells (T_regs) from blood of naive mice as well as peritoneal lavage and pulmonary T_regs at 3 weeks after pristane injection from indicated mouse strains (left) and a representative fluorescence activated cell sorter (FACS) plot of FoxP3 MFI in peritoneal lavage T_regs (right). (b) Analysis of the indicated surface markers on T_regs from blood of naive mice. (c) Quantification of activation markers on T_regs from blood of naive mice of the indicated genotypes. (d) Assessment of T_reg proliferative activity by FACS analysis of KI67 expression in naive blood T_regs. Squares represent individual animals, horizontal lines indicate means. *P < 0·05; **P < 0·01; ***P < 0·001.

biT_regs control Th2 immunity in a RORγt‐dependent manner

Next, we wanted to assess whether the observed absence of IL‐17 secretion and enhanced T_reg activation in FoxP3^Cre × RORC^fl/fl mice might affect systemic immunity. We therefore analysed immune responses in the blood of naive mice. Remarkably, we found a spontaneous hyper‐Th2 phenotype with an increase in Gata3 (0·47 ± 0·04 versus 0·69 ± 0·05% Gata3⁺ among CD4⁺FoxP3^– cells), IL‐5 (0·27 ± 0·09 versus 0·64 ± 0·08% IL‐5⁺ among CD4⁺FoxP3^– cells) and IL‐13 (1·9 ± 0·41 versus 4·3 ± 0·29% IL‐13⁺ among CD4⁺FoxP3^– cells)‐positive T helper cells in FoxP3^Cre × RORC^fl/fl animals (Fig. 8a). In line with this, we also found enhanced Th2 responses in peritoneal lavage cells (Fig. 8b, Supporting information, Fig. S5a) and lungs (Fig. 8c) during pristane‐induced inflammation. Analyses of the renal cellular infiltrate supported these findings and also showed broad skewing towards Th2 immunity (0·41 ± 0·08 versus 0·73 ± 0·1% IL‐4⁺, 0·95 ± 0·14 versus 2·0 ± 0·5% IL‐5⁺ and 4·14 ± 0·2 versus 6·62 ± 0·7% IL‐13⁺ among CD4⁺FoxP3^– cells) (Fig. 8d). Conversely, Th1 and Th17 responses remained unaffected by knock‐out of RORγt in T_regs in all examined organs (Supporting information, Fig. S5b–f). Similarly, serum levels of IFN‐γ and IL‐17 were identical (Supporting information, Fig. S5g), as was splenic and renal IFN‐γ mRNA expression (Supporting information, Fig. S5h). Finally, we analysed humoral Th2 immunity and found strikingly elevated levels of IgE antibodies in serum of naive FoxP3^Cre × RORC^fl/fl mice (358 ± 26·8 versus 801·2 ± 110 ng/ml) as well as after disease induction (670·9 ± 116 versus 1633 ± 194 ng/ml) (Fig. 8e).

Figure 8.

(a) Quantification of T helper (Th) cells expressing the Th2 characteristic transcription factor Gata3 or the indicated Th2 cytokines in peripheral blood from naive mice by flow cytometry. (b) Flow cytometric analysis of peritoneal Th cells expressing Gata3 or the indicated Th2 prototype cytokines from peritoneal lavage cells 3 weeks after pristane injection. (c) Gata3 expression by T helper cells in lungs at 3 weeks after pristane injection. (d) Quantification of renal T helper cells expressing the indicated Th2 cytokines at 2 months after pristane injection. (e) Quantification of immunoglobulin (Ig)E levels by enzyme‐linked immunosorbent assay (ELISA) from serum of naive mice and at 5 months after pristane injection. Squares represent individual animals, horizontal lines indicate means. *P < 0·05; **P < 0·01; ***P < 0·001.

Discussion

Our study aimed to evaluate the role of the newly identified RORγt⁺FoxP3⁺ biT_regs 20, 22, 26, 33 in systemic lupus erythematosus. As nothing is known about the occurrence and biology of these multi‐potent cells in SLE, we started with a thorough analysis of the temporal and organ‐specific pattern of biT_reg expansion. In the first 2 weeks after pristane injection mice develop sterile peritonitis, which is mediated by innate immunity 34. During this acute, non‐autoantigen‐dependent inflammation, biT_reg levels remained at baseline. Another early and antigen‐independent manifestation of pristane is acute pulmonary vasculitis with diffuse alveolar haemorrhage 35. Similarly to peritonitis, we did not find relevant expansion of biT_regs during this innate phase of pulmonary inflammation. However, the persistence of inflammatory responses after pristane injection leads to chronically progressive development of autoimmunity and formation of peritoneal and pulmonary lymphogranulomas, which became apparent from approximately 3 weeks post‐injection. Interestingly, during this stage of disease, biT_reg percentages increased significantly in both organs, indicating a role for establishment of granulomas and autoimmunity. In addition, these findings suggest that biT_reg expansion occurs in an antigen‐dependent manner, which is in full agreement with a recent report 26. With increasing time after pristane injection, various types of autoantibodies develop 36, 37. These are deposited in the kidneys and result in progressive immune complex‐dependent lupus nephritis. During this process, renal biT_regs showed a very distinct and concerted time–course. Their population started to expand early, in parallel with increasing renal inflammation and reached a maximum at approximately 5 months. Subsequently, biT_reg percentages decreased slowly and were almost back to baseline levels at 9 months after pristane injection. Interestingly, this distinguished time–course of renal biT_regs in chronically developing lupus nephritis parallels their dynamics in the nephrotoxic nephritis model of acute glomerulonephritis 22. In contrast to the massive peritoneal, pulmonary and renal biT_reg expansion, they increased only slightly in spleens. This concurs with the fact that pristane injection results only in a mild splenic inflammatory response. In summary, biT_regs thus seem to be early mediators of adaptive inflammation and tissue injury during pristane‐induced SLE. Importantly, and in accordance with previous findings by us and others 21, 22, 38, 39, a robust fraction of biT_regs produces IL‐17 in all organs and at all investigated time‐points. Given the observed expansion of biT_regs in all affected organs, we hypothesized that biT_regs might play key functional roles during SLE development. As our earlier data suggest that activation of RORγt mediates the pathogenic functions of biT_regs 22, we generated mice with selective deficiency of RORγt in FoxP3⁺ T_regs. In a first step, we wanted to explore whether abrogation of RORγt activation would result in protection from pulmonary vasculitis, which is a rare but life‐threatening complication of SLE in humans 40. Analysis of lungs at an early time‐point, before development of autoimmunity and expansion of biT_regs, showed similar degrees of injury. However, at later stages, FoxP3^Cre × RORC^fl/fl mice were protected significantly from pulmonary vasculitis in terms of haemorrhage, leucocyte infiltration and histological damage. Furthermore, pulmonary granulomas were reduced both in size and numbers. Next, we aimed to study the development of lupus nephritis, which is one of the most severe organ manifestations of SLE in humans and associates with a bad prognosis. In this respect, we found that FoxP3^Cre × RORC^fl/fl mice were protected significantly, as evidenced by less histological injury and glomerular cell proliferation. In line with this, functional parameters of renal injury as serum BUN levels and albuminuria were also reduced significantly. Furthermore, we found much decreased renal proinflammatory leucocyte infiltration in mice lacking RORγt activation in T_regs. Interestingly, frequencies of IL‐17‐producing total renal leucocytes were also reduced, despite unaltered renal Th17 responses. This indicates effects of T_reg‐specific RORγt deletion on IL‐17⁺ leucocyte populations different from Th17 cells as, for example, γδTcells and T_regs themselves. In order to explore the mechanisms by which abrogation of RORγt activation in biT_regs protects from SLE organ manifestations, we next assessed development of humoral autoimmunity. However, our analyses revealed unchanged levels of total serum IgG as well as all measured subclasses in FoxP3^Cre × RORC^fl/fl mice. Similarly, serum levels of SLE characteristic anti‐ds‐DNA, and anti‐U1‐RNP autoantibodies of all subclasses remained unaltered. Consistent with unaffected serum antibodies, we found the same extent of renal complement C3 and IgG deposition in both groups of mice. In congruence, analyses of splenic B cell subpopulations, plasma cells and T follicular helper cells showed no difference.

Given the unchanged humoral autoimmunity, we next explored T cell immune responses. Strikingly, and in line with the importance of RORγt for Th17 cells 41, proinflammatory IL‐17 expression by biT_regs was abrogated completely in FoxP3^Cre × RORC^fl/fl mice. This observation might, at least in part, account for the observed amelioration of SLE‐induced tissue injury. In order to identify further potential mechanisms of protection, we explored whether RORγt would also affect T_reg suppression and activity. Similar to our previous data, analysing T_regs from RORC pan‐knock‐out mice 22, we found unchanged in‐vitro suppressive capacity. In vivo, however, T_reg‐selective deletion of RORγt resulted in enhanced levels of FoxP3 protein as well as increased expression of T_reg activation markers. These findings indicate that RORγt expression in T_regs impairs their in‐vivo regulatory function. Therefore, the high activation status of RORγt⁺ biT_regs, which we and others have described previously 22, 33, 42, 43, does not seem to be induced by RORγt. Rather, additional transcription factors, which are co‐activated in RORγt⁺ biT_regs as, for example, interferon regulatory factor (IRF)4 26, runt‐related transcription (Runx)1, Runx3 42 or unknown others, might be responsible. Next, we wanted to explore whether these observed T_reg alterations would result in any effects on systemic immunity. Interestingly, we found significantly elevated type 2 immune responses in naive and pristane‐challenged FoxP3^Cre × RORC^fl/fl mice, both systemically and in all inflamed target organs. Th1 and Th17 responses, in contrast, remained unaltered. Importantly, these observations are in full agreement with a very recent report, which was published during preparation of this paper 26. This emergence of biT_regs as potent regulators of type 2 immunity is particularly remarkable, as the mechanisms regulating Th2 responses remain widely unknown to date. In this regard, our data do not indicate a direct effect of biT_regs on suppression of Th2 cells, as generation of Th2 immunity was not altered in our in‐vitro co‐culture assays of T_eff with RORγt‐deficient T_regs. Rather, as suggested by Ohnmacht et al., indirect mechanisms, involving interactions between T_regs and dendritic cells, seem to be important 26.

Taken together, we found that biT_reg selective deletion of RORγt resulted in abrogation of their IL‐17 secretion, enhancement of Th2 immunity and increased T_reg activation. Importantly, these changes were associated with protection from pulmonary vasculitis and lupus nephritis. It remains unclear, however, to what extent each of these three RORγt‐dependent immune alterations contribute to disease amelioration. While it is likely that protection results from a mixed phenotype, involving all three mechanisms, future studies will need to address this aspect in more detail.

In summary, our study provides the first evidence, to our knowledge, for a crucial role of the newly defined biT_regs in SLE. Our findings thus further support research into biT_reg biology and favour RORγt‐directed interventions as novel therapeutic options for SLE.

Disclosure

None.

Supporting information

Additional Supporting information may be found in the online version of this article at the publisher's web‐site:

Fig. S1. (a) Quantification of macroscopic haemorrhage as well as vasculitis score is shown as indicated. (b) Quantification of granuloma numbers and size. Squares represent individual animals, horizontal lines indicate means. All comparisons between the groups P > 0·05 not significant (n.s.).

Fig. S2. (a) Representative fluorescence activated cell sorter (FACS) plots and (b) quantification of CD45⁺CD3⁺CD4⁺ forkhead box protein 3 (FoxP3)^– T effector cells (T_eff) secreting the indicated cytokines from kidneys of mice at 9 months after pristane injection. Numbers in fluorescence activated cell sorter (FACS) plots indicate percentages. Squares represent individual animals, horizontal lines indicate means. All comparisons between the groups P > 0·05 not significant (n.s.).

Fig. S3. (a) Levels of total immunoglobulin (Ig)G and the indicated IgG subclasses were determined by enzyme‐linked immunosorbent assay (ELISA) from serum of mice at 9 months after pristane injection at serial dilutions. (b) Percentages of splenic total B cells, (c) switched‐memory B cells (sMB) and (d) plasma cells were not different between the groups at 9 months after pristane injection. (e) Frequencies of splenic T follicular helper cells (TfH) were similar at 9 months after pristane injection. All comparisons between the groups P > 0·05 not significant (n.s.).

Fig. S4. (a–f) In‐vitro suppression assays were performed by co‐culturing wild‐type CD4⁺ T effector cells (T_eff) with regulatory T cells (T_regs) from forkhead box protein 3 (FoxP3)^Cre × retinoic acid receptor‐related orphan nuclear receptor (ROR)Cfl/fl mice or FoxP3^Cre controls at the indicated ratios (n = 4 per group). (a) Cytokine levels of interleukin (IL)−2 were analysed in co‐culture supernatants by enzyme‐linked immunosorbent assay (ELISA). Cytokine levels of (b) Th1 cytokine interferon (IFN)‐γ, (c) T helper type 2 (Th2) cytokines IL‐4 and IL‐13, (d) Th17‐associated cytokine tumour necrosis factor (TNF)‐α and (e) IL‐10 were analysed in co‐culture supernatants by cytometric bead array. (f) Suppression of T_eff proliferation was analysed by quantification of KI67 expression. Dotted lines represent T_eff alone without T_regs (n = 4). All comparisons between the groups P > 0·05 not significant (n.s.).

Fig. S5. (a) Representative fluorescence activated cell sorter (FACS) plots of expression of Gata3 and the indicated T helper type 2 (Th2) cytokines in T effector cells (T_eff) of peritoneal lavage cells at 3 weeks after pristane injection. (b–e) Fluorescence activated cell sorter (FACS) analyses of indicated Th1 and Th17 cytokine and transcription factor expression in T_eff from (b) blood of naive mice, (c) peritoneal lavage cells at 3 weeks, (d) pulmonary leucocytes at 3 weeks and (e) renal infiltrating leucocytes at 8 weeks after pristane injection. (f) Enzyme‐linked immunosorbent assay (ELISA) analysis of spleen cell cytokine production at 9 months. (g) ELISA analyses of serum cytokine levels at the indicated time‐points after pristane injection. (h) Interferon (IFN)‐γ mRNA expression levels in spleens and kidneys at 9 months after pristane injection. Numbers in FACS plots indicate percentages. Squares represent individual animals, horizontal lines indicate means. Bars represent mean ± standard error of the mean (s.e.m.). All comparisons between the groups P > 0·05 not significant (n.s.).