Interleukin-21 Promotes Germinal Center Reaction by Skewing the Follicular Regulatory T Cell to Follicular Helper T Cell Balance in Autoimmune BXD2 Mice

Abstract

Objective

Follicular regulatory T (Tfr) cells act as the regulatory counterpart of follicular T helper (Tfh) cells to suppress germinal center (GC) B cell differentiation. We recently identified that interleukin 21 (IL-21) promoted Tfh differentiation in autoimmune BXD2 mice that develop spontaneous GCs. The objective of this study was to determine the modulatory effects of IL-21 on Tfr and the Tfr/Tfh balance in BXD2 mice.

Methods

The percentage and phenotype of Tfr were determined in BXD2 and BXD2-Il21^−/− mice. The effects of IL-21 on Tfr and the ratio of Tfr/Tfh were evaluated. Sorted Tfr cells from BXD2-Il21^−/− mice were co-cultured with Tfh and B cells, or transferred into BXD2 mice to determine their function.

Results

GC B cells and Tfh cells were significantly reduced, but the percentage of Tfr cells was 2-fold higher in BXD2-Il21^−/− mice than in WT-BXD2. Adenovirus-IL-21 administration to BXD2-Il21^−/− mice decreased Tfr and the ratio of Tfr/Tfh but increased GC B cells in the spleen. rmIL-21 suppressed Foxp3 and significantly reduced Tgfb1, Il2 and Gitr but enhanced Il21, Il6, Pd1, Cxcr5 and Icos in Tfr cells. IL-21 also counteracted Tfr-mediated inhibition of antibody secretion in the Tfh-B cell co-culture system. Transfer of Tfr cells into young BXD2 mice reduced GC size and decreased autoantibody-producing B cells.

Conclusion

High levels of IL-21 selectively enhanced Tfh differentiation but inhibited Tfr commitment and their suppressive function on Tfh and B cells, suggesting that IL-21 skews the balance from Tfr to Tfh to promote autoreactive GC reactions in BXD2 mice.

Introduction

Abnormal selection and development of high affinity autoantibody-producing-B cells in germinal center (GC) is a central feature of autoimmune diseases including systemic lupus erythematosus (SLE) and rheumatoid arthritis. Both pro-inflammatory T helper cells and regulatory T (Treg) cells can regulate the formation of GCs. Importantly, the development of antibody-producing plasma cells within the GC requires help from CXCR5⁺ICOS⁺PD-1⁺ follicular T helper (Tfh) cells, the differentiation of which is Bcl6-dependent and IL-21-mediated (1–3). An increase in the numbers or activity of Tfh cells has been correlated with the pathogenesis and severity of disease in GC-dependent autoimmune conditions (4–8). “Regulatory” cells within the GC control the number and the function of Tfh and GC B cells. In mice, Qa-1⁺ CD8⁺ T cells regulate Tfh cells in vivo, whereas in humans, CD69⁻CD25⁺CD4 T cells are capable of migrating into GCs to suppress GC B cell response (9–11). Notably, a recently described subset of follicular regulatory T (Tfr) cells, which share features of Tfh and Treg cells, has been found to regulate the differentiation of GC B cells in vivo (12, 13). However, little is known about how Tfr cells are regulated, although the PD-1-PD-L1 interaction has been reported to inhibit these cells in the lymph nodes and blood (14).

Aberrant T cell homeostasis also contributes to the development of autoimmune diseases. An imbalance between Treg and Th17 is associated with disease activity in lupus prone mice and SLE patients (15). However, the imbalance between Tfh and Tfr cells in the pathogenesis of autoimmunity has not been explored.

The cytokine milieu is critical to control the development of pathogenic and non-pathogenic immune responses. Increased level of IL-21 has been detected in the sera of SLE patients (16) and lupus prone mice (17). IL-21 acts in an autocrine manner to promote the generation of Tfh cells (3, 18) and is considered the signature cytokine of Tfh cells (2, 19, 20). Conversely, IL-21 also has been shown to negatively regulate the number of conventional Treg cells in IL-21 deficient mice (21). In this study, we report that, in autoimmune BXD2 mice that develop spontaneous autoreactive GCs in the spleen, high level of IL-21 plays a critical role in promoting autoimmunity by selectively enhancing Tfh development, inhibiting Tfr programming, as well as counteracting the suppressive function of Tfr cell in vitro and in vivo.

Materials and Methods

Mice

C57BL/6 (B6) and BXD2 mice were obtained from the Jackson Laboratory. B6-Il21^−/− and B6-Il21r^−/− mice obtained from the Mutant Mice Regional Resource Center (Davis, CA) were backcrossed with BXD2 mice for eight generations. All mice were housed under specific pathogen-free conditions in the University of Alabama at Birmingham (UAB) Mouse Facility. All mouse procedures were approved by The UAB Institutional Animal Care and Use Committee. Female mice were used in each experiment.

Flow cytometry analysis

Cells were stained for surface markers with the following antibodies: Pacific-blue- or Alexa-488-anti-CD4 (RM4-5, GK1.5); Pacific-blue-anti-CD19 (6D5); PE conjugated anti-PD-1 (RMP1-30), CD44 (IM7), TGF-β1 (TW7-16B4) all from Biolegend. Alexa-647-anti-GL-7 (GL7); FITC- or PE-anti-ICOS (398.4A or 7E.17G9); PE conjugated anti-CD25 (pc61.5), GITR (DTA-1), and Fas (15A7) all from eBioscience; PE-Cy7-anti-CXCR5 (2G8, BD Biosciences); PE-anti-CTLA-4 (UC10-4F10-11, BD Pharmingen).

For nuclear transcription factor staining, cells were labeled with surface markers, then fixed and permeabilized with the Foxp3-Staining-Buffer-Set (eBioscience), according to the manufacturer's instruction. Cells were then stained with PE-anti-Bcl6 (K112-91, BD Biosciences) and PE-anti-Foxp3 (FJK-16s, eBiosciences).

For phospho-flow staining, after treatment, cells were fixed and permeabilized with the BD Phosflow™ Fix Buffer and Perm Buffer, according to the manufacturer's instruction. Surface markers staining were followed by intracellular staining with Alexa-647-rabbit-anti-phospho-Akt-Ser473 (Cell signaling) or Pacific-blue-mouse-anti-Stat3-p-Y705 (4/p-Stat3, BD Bioscience). Samples were acquired with an LSRII FACS analyzer (BD Biosciences), and data was analyzed with FlowJo software (Tree Star, Inc. Ashland, OR, USA).

Immunofluorescent staining of frozen sections and confocal imaging

Spleens frozen sections were processed as previously described (22). All reagents and antibodies were purchased from Invitrogen except specified: Biotin-PNA (Vector Laboratory) followed by SA-Alexa-350; Alexa-555-anti-IgM; Alexa-647-anti-CD4 (GK1.5, Biolegend); Application of rat anti-mouse Foxp3-biotin (FJK-16s, eBiosciences) and SA-HRP were followed by the tyramide signal amplification (TSA Kit, T20931) and SA-Alexa-488. Images were captured with a Leica DMIRBE inverted Nomarski/epifluorescence microscope outfitted with Leica TCS NT laser confocal optics.

Real-time quantitative RT-PCR

RNA isolation, cDNA synthesis, and real-time PCR reactions were carried out as described previously (22). All primers are described in Supplementary Table 1.

ELISA to detect cytokines

Serum levels of IL-21 and TGF-β1 were measured using the mouse IL-21 ELISA Ready-SET-Go!® (88–8210, eBioscience) Kit and the Human/Mouse TGF-beta-1 ELISA Ready-SET-Go! (88–7344, eBioscience) Kit, respectively, according to the manufacturer’s instructions.

ELISPOT quantification of autoantibody-producing B cells

The ELISPOT procedure was as previously described (23, 24).

Administration of Ad-LacZ and Ad-IL21

Ad-IL21 and Ad-LacZ (2×10⁹ p.f.u. per mouse) were administered intravenously (i.v.). Spleen tissue and cells were analyzed 7 days later.

Proliferation and functional assay of responder cells co-cultured with Tfr cells

B cells from WT-BXD2 mice were purified by Anti-CD19 MACS columns (Miltenyi Biotech). Tfr (CXCR5⁺GITR⁺CD4⁺) and Tfh (CXCR5⁺GITR⁻CD4⁺) cells from BXD2-Il21^−/−, BXD2-Il21r^−/− or WT-BXD2 mice were sorted by the FACSAria II sorter. Tfh cells from WT-BXD2 were labeled with CFSE and left untreated or stimulated with anti-CD3 (1 µg/ml) plus anti-CD28 (2 µg/ml); Tfh cells (1×10⁵/well) were then cultured alone or at a 1:1 ratio with Tfr or control cells and either left untreated or treated with IL-21 (50 ng/ml) for 3 days in 96-well round bottom plate. B cells (3×10⁵/well) were either left untreated or stimulated with anti-IgM (10 µg/ml) and anti-CD40 (10 µg/ml), then cultured alone or with Tfr or control cells at a 3:1 ratio, alone or with IL-21 for 5 days. B cells were also co-cultured with anti-CD3/CD28 stimulated Tfh and Tfr cells at a 3:1:1 ratio, with or without IL-21 for 5 days.

For analysis of Tfh cells proliferation, CFSE dilution was determined by flow cytometry. For B cell functional assay, total IgG antibody levels in supernatants from the co-cultures were measured by ELISA, as described previously (24, 25).

Adoptive transfer of Tfr cells

Tfr cells (CXCR5⁺GITR⁺CD4⁺T cells) from BXD2-Il21^−/− mice were sorted using a FACSAria II sorter and injected intravenously into BXD2 mice (8×10⁵ cells in 200 µl PBS per mouse). Recipients were sacrificed 17 days later and splenic responses were analyzed.

Statistical analysis

All results are shown as mean ± standard deviation (SD). A two-tailed t test was used when two groups were compared for statistical differences. An ANOVA test was used when more than two groups were compared. P values less than 0.05 were considered significant.

Results

IL-21 promotes Tfh, but inhibits Tfr development in BXD2 mice

We previously showed that autoimmune BXD2 mice exhibited an increase in autoantibody-producing B cells and an accumulation of Tfh cells in the spleens (23, 24). Consistent with the previous reports that increased serum levels of IL-21 were associated with the development of autoimmune diseases (16, 17) and that there were lower levels of autoantibodies, significantly reduced spontaneous spleen GCs, and alleviated kidney disease in BXD2-Il21^−/− mice compared with WT-BXD2 mice (24), we found that there was an age-dependent increase in serum levels of IL-21 in BXD2 mice, leading to a 6-fold higher levels of IL-21 in 5-mo-old BXD2 mice compared with B6 mice. As expected, IL-21 was not detectable in BXD2-Il21^−/− mice (Fig. 1A). A decrease in Treg cells has been demonstrated in lupus prone mice and SLE patients (15) and IL-21 has been shown to negatively regulate the number of conventional Treg in Il21^−/− mice (21). To determine the role of Treg cells in autoimmune BXD2 mice, we evaluated the Foxp3⁺CD4⁺ Treg cells in B6, BXD2, and BXD2-Il21^−/− mice at 2.5–4 months of age. There was no significant difference in the percentage of Treg cells among the three stains of mice (Fig. 1B). Surprisingly, the total number of Treg cells was increased in the spleens of BXD2 mice and the ratios of Treg versus non-Treg CD4 T cells were nearly equivalent among the three stains of mice (Fig. 1B), despite that high levels of serum autoantibodies in BXD2 mice require the presence of IL-21 (24). Together, these results suggest that the total number of Treg cells is not correlated with suppression of GC development or autoantibody production in BXD2 mice.

Fig. 1. Selective increase in Tfr/Tfh but not conventional Treg/CD4 T cell in IL-21 deficient autoimmune BXD2 mice.

A. ELISA assay of serum IL-21 levels in BXD2, BXD2-Il21^−/− and B6 mice. Unless specified, all samples were collected from 5-mo-old mice (mean ± SD, n ≥ 5 per group). B. Flow cytometry analysis of conventional Foxp3⁺ Treg gated on lymphocytes; bar graph showing the total number of conventional Treg (Foxp3⁺ CD4 T) cells and the ratio of Treg to non-Treg CD4 T cells. C. Flow cytometry analysis of CXCR5⁺ICOS⁺ CD4 T cells and Foxp3 expression by these cells in the spleens of mice; bar graph showing the ratio of Foxp3⁺CXCR5⁺ICOS⁺ Tfr cells versus Foxp3⁻CXCR5⁺ICOS⁺ Tfh cells in the spleens (top) and PBMC (bottom). D. Flow cytometric analysis of GL-7⁺Fas⁺ germinal center (GC) B cells in spleens of 5-month-old mice gated on CD19⁺ B cells. Right, Ratio of Tfr to GC B cells. In flow cytometry plots, values are the mean percent ± SD (n = 5 or more mice per group) (B, C, D); *p<0.05, **p<0.01, ***p<0.005 between the indicated groups or between BXD2 and BXD2-Il21^−/− mice.

The percentage of CXCR5⁺ICOS⁺ CD4⁺ Tfh cells was low in the spleens of BXD2-Il21^−/− mice compared with WT-BXD2 mice (Fig. 1C left top), which was consistent with reports that IL-21 plays a fundamental role in Tfh development (3, 18). However, the percentage of Foxp3⁺ cells within the CXCR5⁺ICOS⁺CD4 T subset was significantly increased in BXD2-Il21^−/− mice compared with BXD2 mice. Thus, almost half of the CXCR5⁺ICOS⁺CD4⁺ cells in BXD2-Il21^−/− mice displayed a Tfr phenotype (Foxp3⁺CXCR5⁺ICOS⁺CD4⁺) (Fig. 1C left bottom). The ratio of Foxp3⁺ Tfr cells to Foxp3⁻ Tfh cells (Foxp3⁻CXCR5⁺ICOS⁺CD4⁺) was significantly higher in both the spleens and the peripheral blood (PBMCs) of BXD2-Il21^−/− mice compared with BXD2 mice (Fig. 1C right). Although it is not known whether the Tfr cells act directly on B cells or indirectly through inhibition of Tfh cells, Tfr cells have been shown to locate in nascent GCs and suppress further GC B cell differentiation in vivo (12, 13). Consistently, we found that the percentage of Fas⁺GL-7⁺ GC B cells was significantly lower in BXD2-Il21^−/− mice compared with WT-BXD2 mice (Fig. 1D left). The ratio of Tfr to GC B cells was also dramatically increased in BXD2-Il21^−/− mice (Fig. 1D right). These results indicate that IL-21 promotes Tfh development and negatively affects the formation of Tfr cells but not the conventional Treg population in BXD2 mice.

Increased Tfr cells in the GCs of BXD2-Il21^−/− mice

Tfr cells share features of Tfh and Treg cells and are located in the GCs (12, 13). We determined the expression of molecular markers that are associated with Tfh and Treg cells in Foxp3⁺ and Foxp3⁻ CXCR5⁺ICOS⁺ CD4 T cells obtained from the spleens of BXD2-Il21^−/− mice. Compared with Foxp3⁻ Tfh cells, Foxp3⁺ Tfr cells expressed higher levels of Treg markers including CTLA-4, membrane TGF-β1, GITR and CD25, whereas they shared similar high levels of PD-1, CD44 and Bcl6 (Fig. 2A). In addition to the significantly increased percentage of CXCR5⁺ICOS⁺CD4 T cells expressing Foxp3 (Fig. 1C left bottom), the expression of CTLA-4, membrane TGF-β1 and GITR in CXCR5⁺ICOS⁺ CD4 T cells were also higher in BXD2-Il21^−/− mice compared with WT-BXD2 mice (Fig. 2B). The results further confirm the Foxp3⁺ CXCR5⁺ICOS⁺ CD4 T as Tfr cells and this population is increased in BXD2-Il21^−/− mice.

Fig. 2. Phenotype and location of Tfr cells in the spleens of BXD2-Il21^−/− mice.

Unless specified, all samples were collected from 5-mo-old mice (Values are mean ± SD, n = 5 or more per group). A, Flow cytometric analysis of the indicated molecules expressed by Tfh and Tfr cells from the spleens of BXD2-Il21^−/− mice. The broken line indicates gating. B, Flow cytometric analysis of the indicated Treg cell markers expressed on CXCR5⁺ICOS⁺ CD4 T cells from the spleens of BXD2-Il21^−/− mice and wild-type BXD2 mice. Broken lines indicate gating. C, Top, Confocal imaging analysis of FoxP3⁺ CD4 T cells in the spleen follicles of wild-type BXD2 mice and BXD2-Il21^−/− mice. Broken line indicates the GC border. Boxed areas in the middle panels are digitally magnified equally in the right panels to show the staining pattern of FoxP3 in CD4 T cells. Original magnification × 20 in left and middle panels; Bottom, Number of FoxP3⁺ cells per 100 CD4 T cells in the GC area. Bars show the mean ± SD. * = P< 0.05; ** = P<0.01, versus Tfh cells in A and versus BXD2 mice in B and C. TGFβ1 = transforming growth factor β1; GITR = glucocorticoid-induced tumor necrosis factor receptor; PD-1 = programmed death 1 (see Figure 1 for other definitions).

The location of Tfr cells in the spleen follicles of BXD2 and BXD2-Il21^−/− mice was determined by confocal microscopy analysis. The GC, indicated by PNA staining, was much smaller in BXD2-Il21^−/− mice; however, there was a significant increase in the percentage of Foxp3⁺ CD4 T cells in the GC area of BXD2-Il21^−/− mice compared with BXD2 mice (Fig. 2C). A higher magnification view including the area of GC showed Foxp3 (green) as nuclear staining in CD4 (red) T cells. There was no reduction in the number of Foxp3⁺ CD4 T cells in the non-GC area in the spleens of BXD2 mice, compared with BXD2-Il21^−/− mice (Fig. 2C). Together, these results suggest that there is an increase in the percentage of Foxp3⁺ Tfr cells in the GC of BXD2-Il21^−/− mice. These cells predominantly share characteristics of Treg and Tfh cells.

IL-21 suppresses Tfr development in vivo

The aforementioned results suggest that deficiency of IL-21 leads to increased Tfr cells and decreased Tfh cells in BXD2-Il21^−/− mice, and that high levels of IL-21 may reduce Tfr cells in the GCs. To determine if exogenous IL-21 suppresses Tfr cell but promotes Tfh cell development in BXD2-Il21^−/− mice, adenovirus vectors producing IL-21 (Ad-IL21) or control Ad-LacZ were administered and mice were analyzed 7 days later. There was a significant induction of IL-21 but suppression of TGF-β1 in the sera of mice administered with Ad-IL21 compared with Ad-LacZ administered control mice (Fig. 3A). In the spleens of Ad-IL21 administered mice, there was a significant increase in the percentage of CXCR5⁺ICOS⁺CD4 T cells (Fig. 3B left top) and a significant decrease in the percentage of Foxp3⁺ Tfr cells (Fig. 3B left bottom). The decreased ratio of Tfr to Foxp3⁻ Tfh cells in Ad-IL21 administered mice (Fig. 3C) was associated with a significantly increased percentage of Fas⁺GL-7⁺ GC B cells (Fig. 3D). Consistent with the decreased serum levels of TGF-β1, the expression of membrane TGF-β1 in CD4⁺Foxp3⁺CXCR5⁺ICOS⁺ Tfr cells was reduced in Ad-IL21 administered mice (Fig. 3B right panel).

Fig. 3. High level of IL-21 inhibits Tfr and promotes GC formation in vivo.

(A–D) IL-21 producing adenovirus (Ad-IL21) or control Ad-LacZ was administered to 5-mo-old BXD2-Il21^−/− mice. Mice were analyzed 7 days later. A. ELISA assay of serum IL-21 and TGF-β1 levels in the mice. B. Flow cytometry analysis of the CXCR5⁺ICOS⁺ Tfh cells (left top), the frequency of Foxp3⁺ in the CXCR5⁺ICOS⁺ CD4 T subset (left bottom) and the expression of membrane TGF-β1 by Tfr cells (right panel) in the spleens of mice; cells were first gated on CD4 T cells; dash line indicates the gating; numbers in the plots indicate mean percentage ± SD. C. Bar graphs showing the numbers of Tfr cells and the ratios of splenic Foxp3⁺CXCR5⁺ICOS⁺ Tfr cells versus Foxp3⁻CXCR5⁺ICOS⁺ Tfh cells. D. Flow cytometry analysis of GL-7⁺Fas⁺ GC B cells in the spleens of mice; cells were first gated on CD19⁺ B cells Results are representative of 2 experiments; *p<0.05, **p<0.01, ***p<0.005 between the two different treatment groups. See Figure 1 for other definitions.

IL-21 down regulates Foxp3 via inhibiting p-AKT in CXCR5⁺ICOS⁺CD4 T cells

To further confirm that IL-21 suppresses Tfr commitment, we analyzed the effect of in vitro IL-21 stimulation (50 ng/ml) on CXCR5⁺ICOS⁺CD4 T cells from BXD2 and BXD2-Il21^−/− mice. In the absence of IL-21 stimulation, there was an increase in the Foxp3⁺CXCR5⁺ICOS⁺CD4 T population in BXD2-Il21^−/− mice compared with BXD2 mice (Fig. 4A top). This Foxp3⁺ population from BXD2-Il21^−/− mice was significantly reduced following a 16-hour co-culture with IL-21 (Fig. 4A bottom). Similarly, IL-21 treatment also led to decreased expression of membrane TGF-β1 and GITR by CXCR5⁺ICOS⁺CD4 T cells (Fig. 4B).

Fig. 4. Inhibition of Tfr programing by IL-21 in vitro is associated with down-regulation of p-Akt.

(A–B) Spleen cells from WT-BXD2 and BXD2-Il21^−/− mice (5-mo-old) were cultured in media only or with rmIL-21 for 16 hours. Flow cytometry analysis of the expression of Foxp3 (A) or membrane TGF-β1 and GITR (B) on CXCR5⁺ICOS⁺CD4 T cells; dash lines indicate the gating; numbers in the plots indicate mean percentage ± SD. C. Quantitative RT-PCR analysis of the indicated gene expression in sorted Tfr cells from BXD2-Il21^−/− mice after rmIL-21 stimulation for 4 hours. D. Left: Flow cytometry analysis of the phosphorylation of Stat3 (top) and Akt (bottom) in CXCR5⁺ICOS⁺CD4 T cells after rmIL-21 stimulation for 15 minutes, numbers in the plots indicate mean percentage ± SD; Right: Flow cytometry analysis of Foxp3 expression by CXCR5⁺ICOS⁺CD4 T cells after treatment with the indicated doses of LY294002 (a p-Akt inhibitor) for 16 hours (top), dash lines indicate the gating; bar graph showing the frequency of Foxp3 after the treatment (bottom). Data are representative or mean values ± SD (n=6, 5-mo-old mice), *p<0.05, **p<0.01, ***p<0.005 between IL-21 stimulated versus control group or between the indicated groups.

The suppressive effect of IL-21 on Tfr cells was also confirmed by co-culture of sorted Tfr cells (CXCR5⁺GITR⁺CD4 T) either with rmIL-21 (50 ng/ml) or media alone. RT-PCR analysis showed that, except Il10, there was significant decrease in expression of Treg-cell-related markers Tgfβ1, Il2, and Gitr in cells cultured with IL-21 (Fig. 4C). In contrast, the inflammatory-cell-related cytokines Il21 and Il6 were significantly elevated in cells stimulated with IL-21. The expression of Tfh cells signature surface markers Cxcr5, Icos and Pd1 was also increased in IL-21 treated Tfr cells (Fig. 4C).

IL-21 promotes differentiation of Th17 and Tfh through Stat3 (21, 26) and mediates CD8 T cells proliferation through the PI3K-Akt pathway (27), which is involved in Foxp3 expression and accessibility (28–30). To determine the underlying mechanism by which IL-21 suppresses Tfr cells commitment, CD4 T cells from spleens of BXD2 and BXD2-Il21^−/− mice were cultured with either IL-21 or media alone for 15 minutes. Cells were then analyzed for intracellular expression of Stat3 (pY705) and p-Akt (Serine 473). IL-21 induced significant upregulation of p-Stat3 in CXCR5⁺ICOS⁺CD4 T cells from both BXD2-Il21^−/− mice and WT-BXD2 mice (Fig. 4D left top). Akt phosphorylation was higher in the majority of unstimulated CXCR5⁺ICOS⁺CD4 T cells from BXD2-Il21^−/− mice than in cells from BXD2 mice, indicating that IL-21 may inhibit p-Akt in this CD4 T subset. Consistent with this, IL-21 stimulation significantly down-regulated p-Akt in CXCR5⁺ICOS⁺CD4 T cells from BXD2-Il21^−/− mice. The p-Akt level in cells from WT-BXD2 mice was inhibited to a lesser extent by IL-21 (Fig. 4D left bottom). To determine whether p-Akt regulates the expression of Foxp3 by CXCR5⁺ICOS⁺CD4 T cells, CD4 T cells were cultured with different doses of the p-Akt inhibitor, LY294002. There was a significant dose-dependent decrease in the frequency of Foxp3⁺ cells in CXCR5⁺ICOS⁺CD4 T population (Fig. 4D right). These results suggest that IL-21 prevents Foxp3 expression and Tfr cell commitment by inhibiting the phosphorylation of Akt while activating Stat3.

IL-21 acts directly on Tfr cells to counteract their suppression of Tfh cells and IgG production from B cells

To determine the suppressive function of Tfr cells by IL-21, Tfr cells were sorted from BXD2-Il21^−/− mice (Fig. 5A) and then co-cultured with only Tfh, B cells or with both Tfh and B cells. Tfr cells from BXD2-Il21^−/− mice suppressed the proliferation of Tfh cells. This suppressive effect was counteracted by the addition of a high dose of IL-21 (50 ng/ml) (Fig. 5B). Tfr cells from WT-BXD2 mice not only did not suppress but also promoted the proliferation of Tfh cells.

Fig. 5. Inhibition of Tfh proliferation and B cell function by Tfr cells is counteracted by IL-21 in vitro.

(A) FACS sorting strategy of Tfr cells from the spleens of BXD2-Il21^−/− mice (5-mo-old). Values are the mean percent (n = 3 experiments with pooled spleen cells from 5 mice per group). (B–D) Sorted Tfr cells were co-cultured with responder CD19⁺ B cells or Tfh cells from the spleens of WT-BXD2 mice with or without rmIL-21 (50 ng/ml) for 3 to 5 days under B cell or T cell stimulating conditions (abbreviated as Sti.). Unless stated, the ratio of B cells: Tfh: Tfr was 3:1:1. Tfh cells from BXD2-Il21^−/− or Tfr cells from WT-BXD2 or BXD2-Il21r^−/− were used as the controls. B. FACS histogram (left) and bar graph (right) showing the percent of divided Tfh, indicated by CFSE dilution on day 3; numbers in the plots indicate mean percentage. C, D. ELISA analysis of IgG levels in culture supernatant on day 5 following B cell (C) or T cell stimulation (D) in vitro. Bars show the mean ± SD from 3 experiments with pooled spleen cells from 5 mice per group, *p<0.05 compared with all other stimulation groups or between the indicated groups.

In the Tfr and B cell co-culture assay, B cells were stimulated with anti-IgM and anti-CD40, the presence of Tfr cells from BXD2-Il21^−/− inhibited IgG secretion whereas addition of IL-21 restored IgG secretion from B cells (Fig. 5C). Interestingly, Tfr cells from WT-BXD2 mice did not suppress IgG production from B cells. Together, these results indicate that Il21^−/− Tfr cells inhibit Tfh cells proliferation and antibody production by B cells. The presence of high levels of IL-21, however, counteracts the suppressive effects of Il21^−/− Tfr cells.

To further determine whether Tfr cells inhibit Tfh-assisted B cell IgG production, Tfh-B cell-co-culture assays were performed in the presence or absence of IL-21 and Tfr cells. In the absence of IL-21, Tfr from WT-BXD2 did not affect IgG levels; in contrast, compared with the control group, Il21^−/− Tfr cells suppressed IgG production and the effect was positively associated with Tfr cell number (Fig. 5D). Addition of IL-21 restored the IgG levels despite the presence of the Il21^−/− Tfr cells. To investigate if IL-21 acts directly on Tfr cells to counteract their suppressive effect, Tfr cells from BXD2-Il21r^−/− mice were added to replace the Il21^−/− Tfr cells in the co-culture system. Intriguingly, Il21r^−/− Tfr cells suppressed IgG production even in the presence of IL-21 (Fig. 5D). The result suggests that IL-21 can act directly on Tfr cells to “convert” these cells into non-suppressive cells, which thereby enables antibody production in the co-culture system.

Suppression of GC formation by transfer of Tfr cells into BXD2 mice

The above results suggest that Tfr from BXD2-Il21^−/− mice might inhibit IgG response under conditions in which IL-21 levels are not high enough to convert Tfr cells into Tfh cells. To demonstrate that Tfr cells from BXD2-Il21^−/− mice could suppress GC development and autoantibody production in vivo, Tfr cells were sorted from BXD2-Il21^−/− mice as described in Figure 5A and transferred into 2-mo-old BXD2 mice (Fig. 6A), the IL-21 levels of which were much lower compared with those in older BXD2 mice (Fig. 1A). By 17 days after transfer, a significant decrease in Fas⁺GL-7⁺ GC B cells and CXCR5⁺ICOS⁺Tfh was found in recipient mice that received Tfr cells (Fig. 6B). Compared with the control group, transfer of Tfr cells reduced the size of the spleens and PNA⁺ GCs in the spleens (Fig. 6C top). The localization of transferred Tfr cells in the GCs was confirmed by confocal microscopy analysis which showed the co-localization of Foxp3⁺ cells in the vicinity of PNA⁺ cells (Fig. 6C bottom). There was a significant decrease in B cells that produced IgM (Fig. 6D top) and IgG (Fig. 6D bottom) autoantibodies to DNA, rheumatoid factor (RF) and BiP in Tfr transferred recipients compared with control mice. The results demonstrate that the Tfr (CXCR5⁺GITR⁺) subpopulation of CD4 T cells, which were increased in BXD2-Il21^−/− mice, exhibit the regulatory properties of being able to home to the GC, inhibit Tfh cells, and suppress GC development and autoantibody production in vivo.

Fig. 6. Decrease in GC formation and autoantibody-producing B cells in young BXD2 mice that received Tfr cells from BXD2-Il21^−/− mice.

A. FACS sorted Tfr cells from BXD2-Il21^−/− mice were i.v. transferred into 2-mo-old BXD2 mice (8×10⁵ cells per mouse); mice without cell transfer were used as control; Recipient mice were sacrificed on day 17. B. Flow cytometry analysis of GL-7⁺Fas⁺ GC B cells (top) and CXCR5⁺ICOS⁺Tfh cells (bottom) in the spleens of recipient mice; cells were first gated on CD19⁺ B cells or CD4 T cells; numbers in the plots indicate mean percentage ± SD. C. Immunofluorescence staining and confocal imaging analysis of PNA⁺ GC and Foxp3⁺ CD4 T cells in the representative frozen spleen section from each group. Sections were stained with the indicated markers. The GC border in each image is marked by a dashed line. The gross image of the spleen from a representative mouse is shown in the lower right corner (top). D. ELISPOT analysis of IgM (top) and IgG (bottom) autoantibody producing B cells from the spleens of recipient mice. Data are representative results or mean values ± SD (n= 4 per group); *p<0.05, **p<0.01, ***p<0.005 between the 2 groups.

Discussion

An imbalance between Treg and inflammatory cells has been observed in autoimmune diseases (15). Increased Th17 and Tfh cells have been shown in autoimmune patients and lupus prone mice models (4–7, 22, 31). In addition, a deficiency in the number and function of Treg cells in autoimmune conditions has been demonstrated in vivo and in vitro (15, 32–34). Tfr cells are the relatively new regulatory T cell subset that are located in the GCs and thus it is unclear if they prevent GC B cell differentiation as a result of their anatomical location in the GCs or due to their unique effects to inhibit Tfh cells. In addition, how the Tfr cells are regulated in autoimmune conditions remains unexplored. The present study showed that Tfr cells not only could home to the GCs in vivo but also they could inhibit Tfh cell proliferation and IgG production by Tfh-dependent B cells in vitro in the absence of high levels of IL-21.

Our data demonstrates that in BXD2 mice, IL-21 regulates GC formation by preferential expansion of Tfh cells, relative to Tfr cells. These CXCR5⁺ICOS⁺ subpopulations of CD4 T cells are located in the GC as we previously described (24). In BXD2-Il21^−/− mice, the frequency of Foxp3⁺ Tfr cells was increased, whereas there was a significant decrease in the frequency and the number of Foxp3⁻ Tfh cells, resulting in an increase in the ratio of Foxp3⁺/Foxp3⁻ CXCR5⁺ICOS⁺CD4 T cells. This increase in the ratio of T regulatory/T helper cells was restricted to the CXCR5⁺ICOS⁺CD4 T cells, whereas there was no change in the frequency and the ratio of conventional Treg to total CD4 T cells. This is in contrast to a previous study that showed a deficiency of IL-21 enhanced the development of conventional Treg in an in vitro cell culture system (21). This increase in the ratio of Tfr to Tfh cells had a dramatic effect on limiting the development of GCs. Together, these results suggest that IL-21 regulates GC development through an effect that is limited to the relatively small subpopulations of Tfh and Tfr cells, which underscores the important role of these two small subpopulations in regulating GCs. The results also indicate a less important role of other T regulatory cell subpopulations in regulating development of spontaneous autoreactive GCs in BXD2 mice.

IL-21 could also directly suppress Tfr cells in BXD2 mice. Tfr cells from BXD2-Il21^−/− mice exhibited a higher response to high levels of IL-21 in IL-21-induced down-regulation of Foxp3 in vitro, compared with those from WT-BXD2 mice. In the present in vitro culture condition, rmIL-21 inactivated Akt by down-regulating the phosphorylation of Akt at Ser-473 in CXCR5⁺ICOS⁺CD4 T cells from BXD2-Il21^−/− mice while enhancing p-Stat3 in this cell subset. Inactivation of Akt by LY294002 for 16 hours could down regulate Foxp3 expression in CXCR5⁺ICOS⁺CD4 T cells in a dose-dependent manner. These results extends the scope of previous observations that IL-21 can signal through Stat3 to promote T helper cells differentiation (21, 26) and reverse Treg suppression function (35). IL-21 was shown to mediate proliferation of primary CD8 T cells through PI3K-Akt pathway (27) whereas the effect of Akt phosphorylation in regulating the commitment of T helper cells, including Th17 cells, is controversial (36–38). The role of the PI3K-Akt-mTOR pathway in regulating Foxp3 expression is also controversial and has been shown to either negatively or positively regulate Foxp3 expression (28–30). The present results suggest that high levels of IL-21 can upregulate p-Stat3 to promote Tfh cell resistance to Tfr suppression, as well as down-regulate p-Akt to inhibit Foxp3⁺ Tfr cells commitment.

Previous observations on the plasticity of Treg cells in vivo remain controversial. In the current study, the suppressive function of Tfr cell was demonstrated by in vivo transfer of Tfr cells from BXD2-Il21^−/− into WT-BXD2 mice. Interestingly, the transferred Tfr cells homed to preexisting GCs and reduced Tfh and the number, size and autoantibody formation function of these GCs. This is consistent with previous study by Rubtsov et al. showing that Tfr phenotype is durable in vivo (39). However, Tsuji et al. (40) showed that transferring of Foxp3⁺Treg cells resulted in conversion of Treg into Tfh like cells in Peyer's patches, leading to formation of GCs, and proposed that the pro-inflammatory cytokines of the microenvironment might cause the conversion of Treg cells. The above results of our study suggest that high levels of IL-21 inhibited the Foxp3 expression and the commitment of Tfr cells. IL-21 could also convert Tfr cells into Tfh like cells by inhibiting expression of Treg functional molecules including Tgfb1, Il2, and Gitr but promoting expression of pro-inflammatory Il21, Il6 and Tfh cell markers Cxcr5, Icos and Pd1. Thus, in the presence of high levels of IL-21, such as achieved after administration of Ad-IL21 to BXD2-Il21^−/− mice or addition of high levels of IL-21 to in vitro cultures of Tfr cells from BXD2-Il21^−/− mice, IL-21 exhibited the potential to convert Tfr cells into Tfh like cells. However, due to the absence of IL-21 in BXD2-Il21^−/− mice, the number of Foxp3⁺ Tfr cells was increased. Also, in the in vivo transfer experiment, the IL-21 levels in 2-mo-old BXD2 mice may not be high enough to convert the transferred Tfr cells into Tfh cells, and thus the inhibitory ability of the transferred Tfr cells was maintained to suppress GC development in recipient mice. These results suggest that Tfr cells can retain their ability to suppress GC development if IL-21 levels in vivo are relatively low. We propose that the low levels of IL-21 which are normally found in vivo in early stages of GC development are set at such a level to favor durability of Tfr cells, and lack of this regulatory control in autoimmunity requires higher levels of IL-21 that favor conversion of Tfr into Tfh cells, thereby leading to Tfh-driven GC development.

In summary, the current study shows that IL-21 promotes autoreactive GC reactions in autoimmune BXD2 mice by selectively promoting development of Tfh and, at high levels, inhibiting programing and commitment of Tfr cells as well as compromising the suppressive function of Tfr cells on B cells and Tfh cells. The study suggests the feasibility of Tfr cells transfer together with IL-21 blockade as a possible therapeutic strategy to suppress GC activity in autoimmune disease. Since Tfr cells have been reported in human tonsils and immunized mice (13), the present finding should have broader implication in immune homeostatic regulation. Further studies in human and immunized systems would be important to verify the role of IL-21 in regulating Tfr cells.

Abbreviations used in this paper

B6

C57BL/6

Foxp3

forkhead box protein 3

GC

germinal center

PD-1

programmed cell death-1

PNA

peanut agglutinin

rm

recombinant murine

SA

streptavidin

Tfh

follicular T helper cells

Tfr

follicular regulatory T cells

Treg

regulatory T cells

WT-BXD2

wild-type BXD2