Antigen-specific transforming growth factor β–induced Treg cells, but not natural Treg cells, ameliorate autoimmune arthritis in mice by shifting the Th17/Treg cell balance from Th17 predominance to Treg cell predominance

Authors


Abstract

Objective

Transferred CD4+CD25+FoxP3+ Treg cells can prevent autoimmune disease, but generally fail to ameliorate established disease. This study was undertaken to compare the effects of antigen-specific Treg cells induced with interleukin-2 (IL-2) and transforming growth factor β (TGFβ) ex vivo (induced Treg [iTreg] cells) to the effects of equivalent expanded thymus-derived natural Treg (nTreg) cells on established collagen-induced arthritis (CIA).

Methods

CIA was induced in DBA/1 mice by immunization with type II collagen (CII), and before or shortly after immunization, mice were treated with iTreg or nTreg cells that were generated or expanded in vitro. Clinical scores were determined. Inflammatory responses were determined by measuring the levels of anti-CII antibody in the serum and examining the histologic features of the mouse joints. The Th1/Th17-mediated autoreactive response was evaluated by determining the cytokine profile of the draining lymph node (LN) cells of the mice by flow cytometry.

Results

Following transfer, nTreg cells exhibited decreased FoxP3 and Bcl-2 expression and decreased suppressive activity, and many converted to Th17 cells. In contrast, transferred iTreg cells were more numerous, retained FoxP3 expression and their suppressive activity in the presence of IL-6, and were resistant to Th17 conversion. Notably, 10 days after the transfer of donor iTreg cells, predominance was shifted from Th17 cells to Treg cells in the draining LNs of recipient mice.

Conclusion

These findings provide evidence that transferred TGFβ-induced iTreg cells are more stable and functional than nTreg cells in mice with established autoimmunity. Moreover, iTreg cells can have tolerogenic effects even in the presence of ongoing inflammation. The therapeutic potential of human iTreg cells in subjects with chronic, immune-mediated inflammatory diseases should be investigated.

CD4+CD25+FoxP3+ Treg cells are crucial in maintaining immune homeostasis (1). The numbers and/or functions of Treg cells have been reported to be abnormal in many autoimmune diseases, including rheumatoid arthritis (RA) (2–5). CD4+FoxP3+ Treg cells are heterogeneous and can be divided into 3 populations: thymus-derived naturally occurring Treg (nTreg) cells, Treg cells induced in vivo, and Treg cells induced ex vivo with interleukin-2 (IL-2) and transforming growth factor β (TGFβ), with or without retinoic acid or rapamycin (1, 6–8). Although some groups have reported that exogenous polyclonal TGFβ–induced Treg (iTreg) cells are unstable (9), we and others have observed notable protective effects of this subset in animal models of autoimmune disease (10–13) and demonstrated that, unlike nTreg cells, these iTreg cells were resistant to conversion to Th1, Th2, Th17, and follicular helper T (Tfh) cells under inflammatory conditions (14–20).

Collagen-induced arthritis (CIA) has been recognized as a useful animal model of human RA since it mimics the syndromes, pathogenesis, and progression of RA (21). Polyclonal nTreg cells can alter the development and progress of CIA but are ineffective in controlling established disease, although they became effective after treatment with retinoic acid (16, 22). Since antigen-specific Treg cells have more potent protective effects than polyclonal Treg cells (23), the objective of this study was to compare the relative effectiveness of collagen peptide–specific, IL-2–expanded nTreg cells and iTreg cells induced with IL-2 and TGFβ in mice with established disease. We observed that transferred nTreg cells failed to suppress established CIA, but iTreg cell infusion notably ameliorated severity and suppressed progression. This was because in these mice with established inflammation, nTreg cells in vivo lost suppressive activity and many converted to Th17 cells in vivo. In contrast, FoxP3+ iTreg cells were stable, more numerous, and had tolerogenic effects.

MATERIALS AND METHODS

Mice.

Eight-week-old female DBA1/J mice were purchased from The Jackson Laboratory. CII TcR–transgenic FoxP3gfp reporter mice were produced by backcrossing DBA/1J mice with C57BL/6 FoxP3gfp knockin mice for 11 generations to develop FoxP3 reporter mice on the DBA/1 background, and then intercrossing with CII TcR–transgenic mice. All mice were housed and treated according to the National Institutes of Health guidelines for the use of experimental animals with approval obtained from the University of Southern California Committee for the Use and Care of Animals.

Induction and assessment of arthritis.

CIA was induced in the mice by subcutaneous injection of 50 μl of emulsion containing bovine type II collagen (CII) and Freund's complete adjuvant (CFA) at a 1:1 ratio. For assessment of arthritis, animals were scored for clinical signs every 2–3 days as follows: 0 = normal joints, 0.5 = swelling of ≥1 digit, 1 = erythema and mild swelling of the ankle joint, 2 = mild erythema and mild swelling involving the entire paw, 3 = erythema and moderate swelling involving the entire paw, and 4 = erythema and severe swelling involving the entire paw. The clinical score for each mouse was the sum of the scores for the 4 limbs; the maximum possible score for each mouse was 16.

Generation of CD4+ iTreg cells ex vivo.

Naive CD4+CD62L+CD25−CD44low T cells were isolated from the spleen cells of DBA/1 CII TcR–transgenic mice using a naive CD4+ T cell isolation kit (Miltenyi Biotec). Cells were cultured in 48-well plates and stimulated with CII245–270 (50 μg/ml) in the presence of γ-irradiated (30 Gy) antigen-presenting cells (APCs) and 40 units/ml of IL-2 (R&D Systems), either with 2 ng/ml of TGFβ (R&D Systems) (to generate iTreg cells) or without of TGFβ (to generate control) for 4 days. RPMI 1640 medium supplemented with 100 units/ml penicillin, 100 mg/ml streptomycin, 10 mM HEPES (Invitrogen Life Technologies), and 10% heat-inactivated fetal calf serum (HyClone) was used for all cultures. FoxP3 expression was determined by flow cytometry. The suppressive activity of these cells against T cell proliferation was examined using a standard in vitro suppressive assay as previously described (24). Cells (3 × 106) were transferred to DBA/1J mice on day 0, 14, or 28 after immunization with CII and CFA.

Generation of nTreg cells.

CD4+CD25+ cells sorted from the thymus of CII TcR–transgenic mice were expanded with CII245–270 (50 μg/ml) and γ-irradiated APCs or anti-CD3/CD28 beads for 7 days. IL-2 (300 units/ml) was renewed every 3 days. After culturing, cells were harvested and beads were removed. The percentage of FoxP3+ cells was examined by flow cytometry before and after 7 days of expansion. Cells (3 × 106) were transferred to DBA/1J mice on day 0, 14, or 28 after immunization with CII and CFA.

Differentiation of Th17 cells by IL-6 and TGFβ.

Naive CD4+ cells were isolated from the splenocytes of normal mice as described above and cultured in 96-well plates. Cells were stimulated with 1 μg/ml soluble anti-CD3, 1 μg/ml anti-CD28, and 10 μg/ml anti–interferon-γ (anti-IFNγ) and anti–IL-4 monoclonal antibodies (mAb), irradiated APCs (1:1 ratio of APCs to naive CD4+ cells), and 10 ng/ml IL-6, with or without 2 ng/ml TGFβ, for 3 days. Cells were harvested and stained with anti–IL-17A mAb using the intracellular flow cytometry staining protocol as described below.

Proliferation assay.

The iTreg cells generated as described above or the nTreg cells expanded as described above were added to fresh naive T cells at the indicated ratios and were stimulated with anti-CD3 mAb (0.025 μg/ml) and irradiated APCs (30 Gy; 1:1 ratio of APCs to naive CD4+ cells) for 3 days. 3H was added to cultures for the last 16–18 hours, and T cell proliferation (3H-thymidine incorporation) was measured using a scintillation counter.

Histologic assessment.

For histologic examination, mice were anesthetized after the final assessment for arthritis index score. One limb from each mouse was removed and preserved in 10% buffered formalin, decalcified, and subsequently trimmed so as to render a longitudinal section through the limb and digits. The specimens were processed, blocked, sectioned, and stained with hematoxylin and eosin. Lesions in each mouse joint were evaluated in a blinded manner, using a scale of 0–3, as previously described, where 0 = normal and 3 = severe (25). This global histologic score reflects both synovitis (synovial proliferation and inflammatory cell infiltration) and joint destruction (bone and cartilage thickness and irregularity and presence of erosions).

Enzyme-linked immunosorbent assay (ELISA) for anti-CII antibodies.

Blood was collected from each mouse on day 14 after adoptive transfer, clotted at room temperature for 1 hour, and incubated at 4°C overnight. Sera were frozen at −20°C. Anti-CII antibodies were measured by ELISA.

Intracellular staining for flow cytometry.

For intracellular staining of cytokines, cultured cells were stimulated with phorbol myristate acetate (0.25 μg/ml) and ionomycin (0.25 μg/ml) for 5 hours, and with brefeldin A (5 μg/ml) for 4 hours. Cells were then stained with surface markers such as CD4 and CD25 (eBioscience) and further fixed, permeabilized, and stained with FoxP3, IL-17A, IFNγ, IL-4, IL-2, and IL-10 (BioLegend).

Real-time polymerase chain reaction.

Total RNA was extracted with an RNeasy Mini kit (Qiagen). Complementary DNA was generated using an Omniscript reverse transcriptase kit (Qiagen). FoxP3 and Bcl-2 messenger RNA (mRNA) expression was quantified with ABsolute SYBR Green ROX Mix (Thermo). Samples were run in triplicate, and the relative expression of FoxP3, Bcl-2, suppressor of cytokine signaling 1 (SOCS-1), or SOCS-3 was determined by normalizing the expression of each target to hypoxanthine guanine phosphoribosyltransferase (HPRT). Primer sequences used (forward and reverse, respectively) were as follows: for FoxP3, 5′-CCC- AGG-AAA-GAC-AGC-AAC-CTT-3′ and 5′-TTC-TCA-CAA-CCA-GGC-CAC-TTG-3′; for Bcl-2, 5′-CCT-GGC-TGT-CTC-TGA-AGA-CC-3′ and 5′-CTC-ACT-TGT-GGC-CCA-GGT-AT-3′; for SOCS-1, 5′-ACC-TTC-TTG-GTG-CGC-GAC-3′ and 5′-AAG-CCA-TCT-TCA-CGC-TGA-GC-3′; for SOCS-3, 5′-CCT-TCA-GCT-CCA-AAA-GCG-AG-3′ and 5′-GCT-CTC-CTG-CAG-CTT-GCG-3′; and for HPRT, 5′-TGA-AGA-GCT-ACT-GTA-ATG-ATC-AGT-CAA-C-3′ and 5′-AGC-AAG-CTT-GCA-ACC-TTA-ACC-A-3′.

Statistical analysis.

Results were calculated using GraphPad Prism 4.0 software and are presented as the mean ± SEM. Student's t-test was used to assess whether differences between 2 groups were statistically significant, and one-way analysis of variance and/or nonparametric tests were used to assess whether differences among multiple groups were statistically significant. P values less than 0.05 were considered significant.

RESULTS

Comparable properties of antigen-specific iTreg cells induced ex vivo and expanded nTreg cells.

We and others previously reported that antigen-specific iTreg cells can be generated in the presence of IL-2 and TGFβ (10, 11). In this study, we used CII-specific TcR-transgenic mice that express an I-Aq–restricted CII260–267–specific T cell receptor (TCR) composed of Vα11.3 and Vβ8.3. When naive CD4+CD25−FoxP3−CD44low cells were stimulated with CII245–270, few CD4+ cells expressed FoxP3, a key Treg cell marker. Stimulation with exogenous TGFβ enabled ∼50% of the iTreg cells to express FoxP3. (Results are available online at http://keck.usc.edu/Education/Academic_Department_and_Divisions/Department_of_Medicine/Our_Divisions/Division_of_Rhematology/Research/Publications.aspx.) CD4+CD25+ cells sorted from the thymus (nTreg cells) of CII TcR–transgenic mice were expanded with CII245–270 and IL-2 for 1 week, and ∼75% expressed FoxP3. Expression of FoxP3 mRNA and protein was comparable in iTreg cells and expanded nTreg cells.

Following induction and expansion, these FoxP3+ cells sustained expression of Vβ8.3 as an indication of antigen specificity. (Results are available online at http://keck.usc.edu/Education/Academic_Department_and_Divisions/Department_of_Medicine/Our_Divisions/Division_of_Rheumatology/Research/Publications.aspx.) Both antigen-specific iTreg cells and nTreg cells exhibited reduced proliferation when restimulated with CII245–270, but proliferation was restored with exogenous IL-2, demonstrating that these cells had become anergic, a hallmark feature of Treg cells. Accordingly, both subsets also produced little IL-2 or IFNγ, but produced increased levels of IL-10 relative to CD4+ control cells (http://keck.usc.edu/Education/Academic_Department_and_Divisions/Department_of_Medicine/Our_Divisions/Division_of_Rheumatology/Research/Publications.aspx). Unstimulated iTreg cells and unstimulated nTreg cells had equivalent suppressive activity (Figures 1A–C).

Differences between antigen-specific iTreg cells and nTreg cells in suppressing T cell proliferation and Th17 differentiation in vitro.

Studies have revealed that the suppressive activity of nTreg cells can be abolished with IL-6 (26), and this finding was confirmed in the present study. While both nTreg and iTreg cells displayed excellent suppressive activity against T cell proliferation, only iTreg cells maintained their suppression in the presence of exogenous IL-6 in vitro (Figure 1). We conducted 2 separate proliferation assays using either 5,6-carboxyfluorescein succinimidyl ester (CFSE) labeling to detect proliferation cytometrically (Figure 1A) or 3H-thymidine incorporation to measure DNA synthesis (27) (Figure 1B). Different ratios of Treg cells to T responder cells were used (Figure 1C), and both assays had similar results.

Figure 1.

Sustained suppressive activity of induced Treg (iTreg) cells, but not natural Treg (nTreg) cells, in the presence of exogenous interleukin-6 (IL-6). A, Thymic CD4+CD25+ cells (nTreg cells) or transforming growth factor β (TGFβ) and IL-2–treated naive CD4+CD25− cells (iTreg cells) from mice were added to 5,6-carboxyfluorescein succinimidyl ester (CFSE)–labeled T responder cells at a ratio of 1:4 in a T cell proliferation assay. Cells were left unstimulated or were stimulated with IL-6, and were gated on the CD8+ T responder cells. Values are the number of CFSE+ (undivided CD8+) T cells. Results are representative of 4 separate experiments. Baseline indicates T responder cells alone. B, CII TcR–transgenic FoxP3gfp reporter mice were used to generate nTreg cells and iTreg cells. FoxP3gfp+ cells were sorted and added in a T cell proliferation assay at a ratio of 1:4 Treg cells to T effector (Teff) cells. Cells were left unstimulated or were stimulated with IL-6, and 3H-thymidine incorporation was measured. Baseline indicates Teff cells alone. C, Experiments were conducted as described in A, with various ratios of Treg cells to T responder cells. D, Naive T cells isolated from DBA/1J mice were stimulated with soluble anti-CD3/CD28 antibodies and IL-6 with or without TGFβ for polarizing Th17 cells (baseline). CFSE-labeled nTreg cells or iTreg cells were added to some culture wells (at a ratio of 1:4), and the percentages of IL-17+CD4+ CFSE− cells were determined. Results are representative of 3 separate experiments. cIgG = control IgG. E, The percentages of IL-17+ cells at baseline and after the addition of nTreg cells or iTreg cells are shown. F, Experiments were conducted as described in D. The supernatants were harvested, and soluble IL-17A was assayed by enzyme-linked immunosorbent assay. NIL = T cells stimulated with T cell receptor without IL-6 or TGFβ. Values in B, C, E, and F are the mean ± SEM of 3 independent experiments. In B, E, and F, ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001, versus baseline. In C, ∗∗ = P < 0.01 versus iTreg cells. NS = not significant.

Given the close relationship between Th17 cells and nTreg cells, and the critical role played by each in the initiation or prevention of many autoimmune diseases, we next considered the effect of each Treg cell population in controlling Th17 differentiation. In a standard Th17-polarizing culture system, we observed that both IL-6 and TGFβ induced intracellular IL-17A expression in ∼25% of TCR-stimulated naive CD4+ cells. The addition of nTreg cells (at a ratio of 1 nTreg cell to 4 naive T cells) actually slightly enhanced the percentage of CD4+ cells expressing IL-17 (Figures 1D and E) and did not inhibit production of this cytokine by CD8+ cells (results not shown). Importantly, iTreg cells markedly suppressed the frequency of CD4+IL-17+ cells (Figures 1D and E) and CD8+ IL-17+ cells (data not shown), and reduced the concomitant in vitro production of soluble IL-17 (Figure 1F). These results provide further evidence that relative to nTreg cells, iTreg cells have a superior ability to down-regulate Th17 cell differentiation even in the presence of IL-6.

Protective effects of iTreg and nTreg cells against arthritis in mice when transferred before or shortly after CII immunization in vivo.

Previous studies revealed that nTreg cells could limit the progression of CIA (22, 28). To investigate the role of iTreg cells in this process and compare their functional characteristics to those of nTreg cells, we injected mice with 3 × 106 iTreg cells or nTreg cells at the time of CIA challenge. This dose was previously shown to control experimental autoimmune encephalomyelitis (EAE) development (29). CIA incidence and severity were monitored every 2–3 days after cell injection. While CD4+ control cell infusion did not affect disease incidence or severity, we observed that both nTreg cells and iTreg cells markedly decreased the incidence of CIA (Figure 2A). Among those mice that did eventually develop CIA, nTreg cells and iTreg cells similarly delayed the onset of disease and decreased the clinical scores compared to control groups (Figure 2B). Both Treg cell subsets also suppressed IgG2a complement-fixing antibodies, and iTreg cells suppressed total IgG and IgG1 levels as well (Figure 2C).

Figure 2.

Prevention of the development of autoimmune arthritis in mice by both natural Treg (nTreg) cells and induced Treg (iTreg) cells. DBA/1 mice were immunized with type II collagen (CII) and Freund's complete adjuvant (CFA), and 3 × 106 nTreg cells or iTreg cells generated as described in Materials and Methods were transferred intravenously into the mice at the time of immunization. A and B, Incidence (A) and severity (B) of collagen-induced arthritis (CIA) in untreated mice (CII/CFA), control mice treated with CD4+ cells (CD4con), mice treated with nTreg cells, and mice treated with iTreg cells. C, Anti-CII–specific IgG subsets in sera harvested from untreated mice (black bars), mice treated with nTreg cells (red bars), and mice treated with iTreg cells (blue bars) on day 45 after immunization and measured by enzyme-linked immunosorbent assay. Values are the mean ± SEM of combined data from 2 independent experiments (n = 5 or more mice per group in each experiment). ∗ = P < 0.05; ∗∗ = P < 0.01, versus untreated mice with CIA. OD = optical density. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

We next determined the effects of each of the Treg cell subsets on CIA when the cells were transferred to mice just prior to disease onset. Since we observed that mice produce substantial levels of proinflammatory cytokines on day 14 following CIA challenge (data not shown), we injected mice with 3 × 106 Treg cells at this time point. Forty-five days after cell injection, the levels of anti-CII IgG1, IgG2a, and IgG2b were significantly lower in the mice injected with iTreg cells than in the controls, while only anti-CII IgG2b levels were lower in the mice injected with nTreg cells (Figure 3C). At this time point the incidence of arthritis had peaked. We observed significantly decreased CIA incidence in mice that received iTreg cells but not in mice that received nTreg cells (Figure 3A). Nonetheless, among the mice that did develop disease, those treated with nTreg cells and those treated with iTreg cells had similar disease severity, which was significantly lower than that in controls (Figure 3B).

Figure 3.

Suppressive effects of nTreg cells and iTreg cells on autoimmune arthritis in mice. On day 14 after immunization of DBA/1 mice with CII and CFA, 3 × 106 nTreg cells or iTreg cells generated as described in Materials and Methods were adoptively transferred into the mice. A and B, Incidence (A) and severity (B) of CIA in untreated mice, mice treated with nTreg cells, and mice treated with iTreg cells. ∗ = P < 0.05 versus untreated mice. C, Anti-CII–specific IgG subsets in sera harvested from mice on day 45 after immunization and measured by enzyme-linked immunosorbent assay. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001. In B and C, values are the mean ± SEM of combined data from 2 independent experiments (n = 5 or more mice per group in each experiment). D, Interferon-γ (IFNγ) and interleukin-17 (IL-17) production. On day 60 after immunization, mouse spleen and draining lymph node (LN) cells were harvested and stimulated with phorbol myristate acetate, ionomycin, and brefeldin A. IFNγ and IL-17 production by CD4+ cells was analyzed by flow cytometry. Results are representative of ≥5 mice per group. NS = not significant (see Figure 2 for other definitions). Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

Both Treg cell types suppressed IL-17+ cells in the lymph nodes (LNs) and spleens of mice with CIA; however, iTreg cells displayed superior efficacy (Figure 3D). Interestingly, although iTreg cells decreased the frequencies of splenic IFNγ+ cells, nTreg cells actually increased them. Thus, as nTreg cells lose suppressive activity they may develop helper activity, since IFNγ promotes B cell differentiation toward plasma cells (30). These results suggest that iTreg cells that are induced ex vivo are at least as effective as nTreg cells in the prevention of CIA.

Loss of the protective effects of nTreg cells, but not iTreg cells, in established CIA.

To consider clinical relevance, we explored the therapeutic effect of each Treg cell population in established CIA. Consistent with the findings of previous studies (16), transfer of 3 × 106 nTreg cells to mice with evident arthritis 28 days after collagen immunization did not significantly decrease disease severity (Figure 4A), autoantibody production (Figure 4B), or joint damage, bone erosion, or inflammatory cell infiltration (Figures 4C and D). Conversely, injection of iTreg cells into the mice almost completely suppressed the progress of disease for 2 weeks, and afterward the severity never reached the levels observed in controls (Figure 4A). In addition, iTreg cell treatment significantly suppressed IgG2a and IgG2b, but not IgG or IgG1, autoantibody production (Figure 4B). Moreover, iTreg cell treatment markedly reduced articular cartilage and joint pathology, reduced inflammatory cell infiltration, and left the joint space substantially intact (Figures 4C and D).

Figure 4.

Significant suppression of the development of established autoimmune arthritis in mice treated with iTreg cells but not in mice treated with nTreg cells. DBA/1 mice were immunized, and 3 × 106 nTreg cells or iTreg cells were adoptively transferred to the mice when they had developed typical arthritis (around day 28 after immunization). Arthritis development was monitored after Treg cell therapy. A, Clinical scores in untreated mice, mice treated with iTreg cells, and mice treated with nTreg cells. B, Anti-CII–specific IgG subsets in the indicated groups of mice 2 weeks after cell transfer, as measured by enzyme-linked immunosorbent assay. ∗ = P < 0.05; ∗∗ = P < 0.01, ∗∗∗ = P < 0.001. C, Histologic alteration of the joints of mice from each group on day 56 after immunization. Joints were removed, fixed, and stained with hematoxylin and eosin. Original magnification × 100. D, Inflammation and destruction scores in untreated mice, mice treated with nTreg cells, and mice treated with iTreg cells. ∗∗∗ = P < 0.001 versus untreated mice. NS = not significant. Values in A, B, and D are the mean ± SEM of combined data from 2 independent experiments (n = 5 mice per group in each experiment). Results in C are representative of ≥5 mice per group. See Figure 2 for other definitions.

Decreased stability of nTreg cells, but not of iTreg cells, in mice with established arthritis.

Given that nTreg cells, unlike iTreg cells, lose suppressive activity in vitro when stimulated with proinflammatory IL-6, we believed they might be unstable in vivo. To address this possibility, we labeled Treg cells of both subsets with CFSE to distinguish them from recipient cells and transferred them to mice with established arthritis. Mice were killed 1, 10, or 20 days after cell transfer, and FoxP3 expression in donor cells isolated from the spleens and LNs of recipient mice was examined by fluorescence-activated cell sorting (FACS). The percentage and total number of donor nTreg cells that had survived in the spleens of recipient mice 10 or 20 days after cell transfer (Figures 5A and B) and in the LNs of recipient mice 10 or 20 days after cell transfer (Figure 6) were significantly lower than those of donor iTreg cells, although their levels were the same 1 day after transfer (Figures 5A and B).

Figure 5.

Maintenance of the phenotype and function of iTreg cells, but not nTreg cells, after adoptive transfer of 5,6-carboxyfluorescein succinimidyl ester (CFSE)–labeled nTreg or iTreg cells to mice with arthritis. On day 1, 10, or 20 after transfer, mice were killed, and spleen, blood, and lymph node (LN) cells were harvested. FoxP3 expression was identified by flow cytometry in donor (CFSE+) or recipient (CFSE−) cells. A, Total numbers of donor iTreg cells and nTreg cells in the spleens of recipient mice. ∗∗∗ = P < 0.001 versus nTreg cells. B, Percentages of FoxP3+ cells in the spleens of recipient mice before cell transfer and percentages of FoxP3+ donor cells in the spleens of recipient mice after cell transfer. ∗∗∗ = P < 0.001 versus nTreg cells. C, Suppression rates. Donor (CFSE+) cells were sorted on day 10 after infusion and were added to T cells isolated from mice with CIA. These cells were then stimulated with anti-CD3 in the presence of irradiated antigen-presenting cells. Cell proliferation was determined by 3H-thymidine incorporation. Suppression rates of nTreg cells and iTreg cells were calculated using the following formula: ([proliferative value of baseline − sample value]/baseline value) × 100%. ∗∗∗ = P < 0.001 versus baseline. D, Relative Bcl-2 mRNA expression. Donor (CFSE+) cells were sorted on day 10 after infusion, and Bcl-2 mRNA expression was determined by quantitative reverse transcriptase–polymerase chain reaction. ∗∗ = P < 0.01 versus iTreg cells. Experiments were repeated with similar results. Values are the mean ± SEM (n = 5 mice per group). NS = not significant (see Figure 2 for other definitions).

Figure 6.

Increased ratio of FoxP3 to interleukin-17 (IL-17) in the draining lymph node (LN) cells of mice with CIA treated with iTreg cells but not mice with CIA treated with nTreg cells. A, Twenty-eight days after immunization of DBA/1 mice with CII and CFA, 5,6-carboxyfluorescein succinimidyl ester (CFSE)–labeled iTreg cells (3 × 106) or nTreg cells (3 × 106) was transferred to the DBA/1 mice. Mice were killed, and IL-17 expression in donor (CFSE+) and recipient (CFSE−) cells in draining LNs was determined by flow cytometry 10 days after cell transfer. Results are representative of 5 mice per group. B, The total numbers of Th17 cells, total numbers of Treg cells, and ratios of Treg cells to Th17 cells were determined in the draining LNs of untreated mice with CIA, mice treated with nTreg cells, and mice treated with iTreg cells. Values are the mean ± SEM (n = 5 mice per group). ∗∗ = P < 0.01; ∗∗∗ = P < 0.001, versus untreated mice with CIA. Experiments were repeated with similar results. See Figure 2 for other definitions. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

Examination of sorted subsets revealed that donor nTreg cells 10 days after transfer also expressed significantly lower levels of Bcl-2 mRNA than donor iTreg cells (Figure 5D). Bcl-2 plays an important role in the prevention of cell apoptosis (31). Additionally, a significant proportion of donor nTreg cells but not iTreg cells lost expression of FoxP3. Although >75% of the CFSE+ nTreg cells expressed FoxP3 at the time of transfer, 10 days later, FoxP3 was expressed by <10% of donor nTreg cells in the spleens and ∼25% of donor nTreg cells in the draining LNs of the recipient mice. Conversely, donor iTreg cells mostly maintained FoxP3 expression equivalent to that at the time of transfer (∼50%), even 10–20 days after transfer (Figures 5B and 6A). Thus, it was not surprising that the donor nTreg cells that were recovered in these mice now had markedly less suppressive ability than the donor iTreg cells (Figure 5C).

We further demonstrated that >50% of donor nTreg cells in the draining LNs of the mice had become IL-17+ cells (Figure 6A), although few of them converted into IFNγ+ and/or IL-4+ cells (results not shown). Fewer nTreg cells converted in the spleen. (Results are available online at http://keck.usc.edu/Education/Academic_Department_and_Divisions/Department_of_Medicine/Our_Divisions/Division_of_Rheumatology/Research/Publications.aspx.) Conversely, donor iTreg cells were almost completely resistant to Th17 cell conversion in the draining LNs of the mice (Figure 6A) as well as Th1/Th2 conversion under similar conditions (results not shown). The findings of in vitro experiments suggested that nTreg cells expressed substantial levels of IL-6 receptor α chain (CD126) and activated STAT-3 following IL-6 stimulation, whereas iTreg cells expressed undetectable levels of IL-6 receptor and pSTAT-3 even after IL-6 stimulation (http://keck.usc.edu/Education/Academic_Department_and_Divisions/Department_of_Medicine/Our_Divisions/Division_of_Rheumatology/Research/Publications.aspx). Although iTreg cells were less sensitive to IL-6 and resistant to Th17 conversion (20) (data not shown), FoxP3 expression on iTreg cells generated by strong TCR stimulation (plate-bound or coating) was less sustained compared to suboptimal TCR stimulation (beads) (http://keck.usc.edu/Education/Academic_Department_and_Divisions/Department_of_Medicine/Our_Divisions/Division_of_Rheumatology/Research/Publications.aspx).

Recent studies have demonstrated that SOCS, proteins that are usually induced in response to cytokines and other stimuli, can directly regulate the activation of STATs or Treg cells and Th17 cell differentiation (32). We found that SOCS-1 and SOCS-3 expression was elevated in nTreg cells but not in iTreg cells following IL-6 stimulation. (Results are available online at http://keck.usc.edu/Education/Academic_Department_and_Divisions/Department_of_Medicine/Our_Divisions/Division_of_Rheumatology/Research/Publications.aspx.) It is likely that nTreg cells rapidly expressed SOCS in response to IL-6 to regulate STAT-3 activation and promote their conversion to Th17 cells. Indeed, disruption of SOCS-1 expression in T cells has been shown to strongly inhibit Th17 differentiation and diminish disease severity in a model of EAE (32). The precise role of SOCS family members in Treg cells and Th17 cell differentiation and development requires further investigation.

Transfer of iTreg cells to mice with CIA shifts the predominance from Th17 to Treg cells in draining LNs.

Previous studies showed that the transfer of iTreg cells can markedly increase the numbers of FoxP3+ Treg cells in the recipient (24, 33). This was also the case in the present study. As shown in Figures 6A and B, both percentages and total numbers of FoxP3+ cells in the LNs of recipient mice were markedly increased in the mice treated with iTreg cells relative to the untreated mice with CIA. Moreover, the infusion of iTreg cells in mice with established CIA markedly down-regulated IL-17+ cell frequencies. FACS analysis revealed that both the percentages and the total numbers of IL-17+ cells in the LNs of recipient mice were markedly lower in the mice treated with iTreg cells than in the untreated mice with CIA. Before treatment, Th17 cells were twice as numerous as Treg cells in the draining LNs of the mice. Ten days after iTreg treatment, FoxP3+ recipient Treg cells were predominant and were 4 times more numerous than Th17 cells. In contrast, the total numbers of Treg cells and Th17 cells in recipient mice were unchanged in nTreg-treated mice (Figure 6B). Thus, treatment with iTreg cells markedly altered the balance between Treg cells and Th17 cells in the recipient mice, and the disease course was changed.

DISCUSSION

The stability and therapeutic effectiveness of CD4 regulatory cells induced with IL-2 and TGFβ ex vivo is a subject of controversy. We and others have demonstrated that these iTreg cells have protective effects in several experimental models of immune-mediated diseases (10, 12, 13, 29) and are resistant to conversion to Th17 cells (20). In contrast, others have reported that these iTreg cells were unstable in vitro (9) and in vivo following antigen stimulation (34), and lacked the protective activity needed to prevent fatal graft-versus-host disease (9, 35).

To address this controversy, we have conducted a head-to-head comparison of antigen-specific mouse thymus–derived nTreg cells and TGFβ-induced iTreg cells. We used antigen-specific Treg cells since these are more protective than polyclonal Treg cells in autoimmune diseases (23). We chose established autoimmunity because Treg cells are generally therapeutic when transferred before the onset of autoimmunity. In CIA, the model examined in this study, polyclonal nTreg cells can prevent disease but are ineffective in established disease (16, 22). This study clearly demonstrated that antigen-specific iTreg cells are superior to nTreg cells in ameliorating established CIA. This was because iTreg cells remained stable and fully functional following transfer. Moreover, these iTreg cells had tolerogenic effects in the draining LNs of mice that resulted in a shift from Th17 predominance to Treg predominance.

This notable difference between iTreg cells and nTreg cells could not be attributed to differences in the starting populations of the 2 Treg cell subsets. Actually, a greater percentage of nTreg cells expressed FoxP3. They both had equivalent suppressive effects in vitro, and both had equivalent therapeutic effects on mice with CIA when transferred before CII immunization in vivo. Significant differences between the 2 Treg cell subsets began to appear when mice were injected with Treg cells after CII immunization but before the onset of arthritis. In this case, iTreg cells were more effective in reducing disease incidence. Consistent with the findings of a previous study (22), nTreg cells also failed to suppress autoantibody production.

There are several possible explanations for the inability of nTreg cells to ameliorate CIA and other autoimmune diseases. First, proinflammatory cytokines may hamper their suppressive activity. Pasare and Medzhitov have reported that the suppressive activity of Treg cells can be abolished by IL-6 (26). Valencia et al also demonstrated that elevated levels of tumor necrosis factor α may interfere with the suppressive capacity of nTreg cells (36). Second, Th17 cells may be resistant to nTreg cells. This may be the reason that nTreg cells are able to prevent disease before Th17 cells are established. Third, at least some nTreg cells are inherently unstable and can be converted to Th1, Th2, Th17, and Tfh effector cells in an inflammatory milieu (15–19).

Several factors can explain the therapeutic success of iTreg cells. First, the iTreg cells exhibited reduced levels of IL-6 receptor and subsequent STAT-3 phosphorylation (20). Thus, they were resistant to IL-6 stimulation and maintained their phenotype and function. The iTreg cells, but not the nTreg cells, expressed undetectable levels of SOCS-1 and SOCS-3 in response to IL-6, whereas the SOCS, at least SOCS-1, promote Th17 cell differentiation (32). Increased SOCS expression on IL-6–stimulated nTreg cells may contribute to STAT-3 activation and Th17 conversion. In this study, antigen-specific TGFβ-induced iTreg cells demonstrated similar IL-6 resistance. Furthermore, these iTreg cells even suppressed the Th17 cell differentiation that is induced by IL-6 and TGFβ. It is understandable that nTreg cells lack this ability since IL-6 was included in the cultures for Th17 polarization. Second, antigen-specific iTreg cells, but not nTreg cells, were stable in mice with established CIA. Only the former maintained FoxP3 expression and exhibited suppressive activity when recovered from the draining LNs of the mice. Third, measurement of Bcl-2 gene expression revealed higher expression in recovered iTreg cells, suggesting that nTreg cells were probably more susceptible to apoptosis than iTreg cells. Previous studies have shown that TGFβ increases Bcl-2 expression and decreases T cell apoptosis (7).

The final, and probably most important, reason for the therapeutic success of iTreg cells is that TGFβ-induced iTreg cells shifted the balance between Treg and Th17 cells in the draining LNs of the mice from Th17 predominance to Treg predominance. This is probably because the transferred Treg cells induced immunogenic APCs in the recipient mice to become tolerogenic. Thus, ongoing antigen stimulation resulted in the generation of Treg cells rather than effector T cells (6). Previously, we reported that a single injection of alloantigen-specific iTreg cells followed by continuous alloantigen stimulation steadily increased CD4+CD25+FoxP3+ Treg cells in recipient mice (24). In the present study, we demonstrated that before the transfer of Treg cells, mice with CIA had twice as many Th17 cells as Treg cells in the draining LNs. Ten days after iTreg injection, however, FoxP3+ cells were much more numerous than Th17 cells in the recipient mice, and the clinical scores of these animals had decreased. Thus, even in an inflammatory environment, injected TGFβ-induced iTreg cells can have tolerogenic effects. Recent studies by Nguyen et al (37) indicate that chemokines secreted by antigen-specific TGFβ-induced iTreg cells regulate T cell trafficking and thereby suppress ongoing autoimmune disease. They reported that these iTreg cells were therapeutic in an ongoing autoimmune gastritis model (37).

There are technical reasons that may explain why some investigators have generated unstable, ineffective TGFβ-induced iTreg cells. Some groups used high concentrations of plate-bound anti-CD3 with TGFβ, and stimulated the CD4+ cells for >72 hours. Others have demonstrated that strong, sustained TCR stimulation activates the mammalian target of rapamycin/Akt signaling pathway, which facilitates T effector (Teff) cell differentiation and inhibits FoxP3 expression and Treg cell differentiation (38). Treg cell generation is best established with suboptimal TCR stimulation that facilitates FoxP3 expression (6). We have demonstrated that the use of suboptimal concentrations of anti-CD3– and anti-CD28–coated beads, rather than plate-bound anti-CD3/CD28 stimulation, with IL-2 and TGFβ induces stable, protective FoxP3+ Treg cells. It has been reported that TGFβ is unable to demethylate FoxP3 in its promoter and CpG sites in the +4,201 to +4,500 intronic CpG island in conserved noncoding DNA sequence 3 (CNS3), which affect FoxP3 expression and maintenance by Treg cells (9). However, we have recently observed that all-trans-retinoic acid markedly increases FoxP3 stability, and this does not affect the methylation status in CpG islands in CNS3 of the FoxP3 locus (39). Others have also observed protective human TGFβ-induced Treg cells that exhibit methylated FoxP3 (40).

The results we obtained using these ex vivo TGFβ–induced iTreg cells may not apply to endogenous iTreg cells induced in vivo. Although suboptimal TCR stimulation and the presence of TGFβ are also important in the generation of endogenous iTreg cells, this subset, like nTreg cells, may be susceptible to Th17 conversion. Because of the documented Teff cell resistance to suppression (41), it is likely that all endogenous CD4+FoxP3+ subsets exhibit this characteristic. We suspect that it is the pharmacologic concentrations of the IL-2 and TGFβ used that confer resistance. It is for this reason that we distinguish endogenous nTreg and iTreg subsets from iTreg cells induced ex vivo.

The findings of the present study must be interpreted with caution. The decrease in arthritis severity was transient, and the clinical scores in the treated mice later increased, although they remained significantly below those of control mice. We induced mouse iTreg cells, and the protocols used must be modified considerably to induce similar human iTreg cells. Nonetheless, the notable difference between iTreg cells and nTreg cells observed in this study suggests that the therapeutic potential of human TGFβ-induced iTreg cells should be fully explored.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Drs. Zheng and Zou had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Horwitz, Quesniaux, Ryffel, Brand, Zou, Zheng.

Acquisition of data. Kong, Lan, Chen, Wang, Shi, Liu.

Analysis and interpretation of data. Kong, Wang, Zou, Zheng.

Ancillary