The molecular mechanism of the extrathymic generation of adaptive, or inducible, CD4+Foxp3+ regulatory T cells (iTregs) remains incompletely defined. We show that exposure of splenic CD4+CD25+Foxp3− cells to IL-2, but not other common γ-chain cytokines, resulted in Stat5 phosphorylation and induced Foxp3 expression in ∼10% of the cells. Thus, IL-2/Stat5 signaling may be critical for Foxp3 induction in peripheral CD4+CD25+Foxp3– iTreg precursors. In this study, to further define the role of IL-2 in the formation of iTreg precursors as well as their subsequent Foxp3 expression, we designed a two-step iTreg differentiation model. During the initial “conditioning” step, CD4+CD25−Foxp3− naive T cells were activated by TCR stimulation. Inhibition of IL-2 signaling via Jak3–Stat5 was required during this step to generate CD4+CD25+Foxp3− cells containing iTreg precursors. During the subsequent Foxp3-induction step driven by cytokines, IL-2 was the most potent cytokine to induce Foxp3 expression in these iTreg precursors. This two-step method generated a large number of iTregs with relatively stable expression of Foxp3, which were able to prevent CD4+CD45RBhigh cell–mediated colitis in Rag1−/− mice. In consideration of this information, whereas initial inhibition of IL-2 signaling upon T cell priming generates iTreg precursors, subsequent activation of IL-2 signaling in these precursors induces the expression of Foxp3. These findings advance the understanding of iTreg differentiation and may facilitate the therapeutic use of iTregs in immune disorders.
CD4+CD25+ regulatory T cells (Tregs) are able to suppress various immune responses against self and foreign Ags. The transcription factor Foxp3 is predominantly expressed in CD4+CD25+ Tregs and plays a central role in establishing and maintaining the Treg lineage. Deficiency of a functional Foxp3 gene in both humans and mice leads to ablation of Tregs, severe autoimmunity, and early mortality (1, 2). Therefore, elucidating the factors that control Foxp3 expression will advance our understanding of Treg biology and its therapeutic application for immune diseases.
IL-2 has long been considered a major T cell growth factor optimizing immune responses, as signaling through its high-affinity IL-2R (consisting of the IL-2Rα [CD25], IL-2Rβ, and common γ-chain [γc] subunits) is essential for the expansion of recently activated effector T (Teff) cells (3). Therefore, it was somewhat unexpected that mice deficient in IL-2, IL-2Rα, or IL-2Rβ developed autoimmune diseases, often with lethal consequences (4, 5). Further studies revealed that Tregs constantly express high-affinity IL-2R. Indeed, because a constant supply of IL-2 is critical for Treg homeostasis, the lethal autoimmunity was eventually associated with an IL-2 signaling defect in Tregs (3, 6). Thus, IL-2 proved to be an essential cytokine not only in Teff-mediated immunity but also in Treg-maintained immune tolerance (3, 7). Studies by our group and others have further shown that Tregs exert their suppressive effect on Teff cell responses at least partially through creating and modulating an IL-2–deprived environment (8, 9).
Foxp3-expressing Tregs are either derived from the thymus as natural regulatory T cells (nTregs) or generated de novo from peripheral mature CD4+Foxp3− T cells in response to TCR stimulations as adaptive, or inducible, regulatory T cells (iTregs) (10). The low frequency of CD4+CD25+Foxp3+ cells in the thymus of IL-2Rβ−/− mice led to the assumption that IL-2 signaling is also critical for thymic nTreg generation (6). Moreover, mice deficient in both IL-2 and IL-15 (also binding to IL-2Rβ) exhibit marked deficiency in thymic nTregs (11). Recently, a “two-step model” of thymic nTreg generation suggested that TCR–ligand interactions result in elevated CD25 (IL-2Rα) expression on Foxp3– CD4 single positive (CD4SP) thymocytes, followed by an IL-2–directed and TCR-independent step that subsequently induces Foxp3 expression in these CD25+Foxp3– CD4SP nTreg precursors (12, 13). Some other γc-dependent cytokines (IL-7 and IL-15) less effectively induced Foxp3 expression in nTreg precursors (14).
The essential polarizing cytokines involved in the differentiation of iTregs appear to be TGF-β and IL-2. In vitro activation of CD4+CD25−Foxp3− T cells in the presence of exogenous TGF-β results in a substantial percentage of Foxp3-expressing iTregs (15). Importantly, IL-2 was required for TGF-β–mediated iTreg differentiation, as addition of IL-2 neutralizing Ab to cultures or using IL-2–deficient T cells abrogated iTreg generation induced by exogenous TGF-β. In addition, only IL-2, but not other γc cytokines, was able to restore TGF-β–mediated Foxp3 expression in IL-2–deficient T cells (16, 17). Thus, IL-2 plays an essential and nonredundant role in TGF-β–mediated iTreg generation.
We have recently shown that in the absence of exogenous TGF-β and IL-2, TCR stimulation of neonatal T cells converted them into stable Tregs (18). This finding led us to reinvestigate the role of TGF-β and IL-2 in generating adult iTregs. In the current study, we first demonstrate that addition of IL-2 alone induces Stat5 phosphorylation and Foxp3 expression in ex vivo isolated peripheral CD4+CD25+Foxp3− iTreg precursors. Next, we show in a two-step model (including a TCR-directed “conditioning” stage and a cytokine-driven Foxp3-induction stage) that IL-2 plays a dynamic dual role in the differentiation of iTregs. At the initial conditioning stage upon TCR stimulation, inhibition of IL-2 signaling promotes the generation of iTreg precursors. Subsequently, IL-2 alone added at the Foxp3 induction phase induces Foxp3 expression in these iTreg precursors. This two-step process of iTreg differentiation does not require exogenous TGF-β, although direct blocking of TGF-β signals impaired such iTreg generation. The iTregs generated by this method exhibit relatively stable expression of Foxp3 and demonstrate potent suppressive function both in vitro and in vivo.
Materials and Methods
C57BL/6 (B6), B6.129P2-Il2tm1Hor/J (IL-2−/−), B6.129S7-Rag1tm1Mom/J (Rag1−/−), B6.Cg-Foxp3tm2Tch/J (Foxp3/GFP), and B6.Cg-Tg(Cd4-TGFBR2)16Flv/J (dnTGFBRII) mice were purchased from The Jackson laboratory (Bar Harbor, ME). B6.SJL-Ptprca/BoyAiTac (B6.SJL) congenic mice were obtained from Taconic Farms (Hudson, NY). Animals were maintained at the University of Toledo specific pathogen–free animal facility according to institutional guidelines.
Abs, cytokines, and reagents
Fluorescence-conjugated Abs were purchased from BD Biosciences and eBioscience. Purified anti-CD3 and anti-CD28 mAbs, BD Phosflow Lyse/Fix Buffer I, BD Phosflow Perm/Wash Buffer I, and FITC Annexin V Apoptosis Detection Kit were purchased from BD Biosciences. Functional-grade anti–IL-2 mAbs (clone S4B6 or JES6), anti-CD25 (PC61) mAb, and anti-Foxp3–PE and intracellular staining kit were purchased from eBioscience. CFSE was obtained from Molecular Probes. TGF-β neutralizing Abs (9016 or 1D11), murine recombinant thymic stromal lymphopoietin, and IL-7 were purchased from R&D Systems. Other murine recombinant cytokines were purchased from Peprotech. An inhibitor of TGF-β superfamily type I activin receptor–like kinase receptors (SB-431542) was purchased from Sigma-Aldrich. Jak3 inhibitor (CP690550) was purchased from Axon Biochemicals BV, whereas the Stat5 inhibitor [N′-((4-Oxo-4H-chromen-3-yl)methylene)nicotinohydrazide] and the Stat3 inhibitor (WP1066) were obtained from EMD Bioscience. Abs specific for Smad3 (C67H9) and phospho-Smad3 (C25A9) were purchased from Cell Signaling Technology. Anti–β-actin and lumin B mAbs were obtained from BD Biosciences and Santa Cruz Biotechnology, respectively.
Cell preparation, cultures, and flow cytometry
Single-cell suspensions were obtained from thymus and spleens of Foxp3/GFP, wild-type B6, IL-2−/−, B6.SJL, or dnTGFBRII mice. To determine the existence of Treg precursors in thymus and spleen, CD4+CD8–CD25+Foxp3/GFP– thymocytes, CD4+CD8–CD25–Foxp3/GFP– thymocytes, CD4+CD25+Foxp3/GFP– splenocytes, and CD4+CD25–Foxp3/GFP– splenocytes were obtained from Foxp3/GFP mice by cell sorting, and were cultured with 10 U/ml IL-2 or 10 ng/ml other cytokines (IL-4, IL-7, or IL-15, as indicated in the text) for 3 d. The percentage of GFP+ T cells in the cultures was analyzed by flow cytometry. The phospho-Stat5 expression was determined by flow cytometry using anti-Stat5 (pY694)–PE according to the BD phosphoflow protocol.
For generating peripheral iTregs by the two-step method, 5 × 104 CD4+CD25–Foxp3/GFP– or CD4+CD25– splenocytes from different mice were cultured in 96-well flat-bottom plates, and were stimulated with 4 μg/ml plate-bound anti-CD3 mAb and 2 μg/ml soluble anti-CD28 mAb for 3 d. During this TCR-directed conditioning phase, 10 μg/ml anti–IL-2 mAb, 10 μg/ml anti-CD25 mAb, 5 μg/ml anti–TGF-β mAb, 1 μM TGF-β inhibitor, 50 nM JAK3 inhibitor, 150 μM STAT3 inhibitor, 50 μM STAT5 inhibitor, or 2 ng/ml TGF-β was added into the cultures, respectively. The cells were then washed and cultured again, this time in the presence of 10 U/ml IL-2 or 10 ng/ml other cytokines for 3 d (the Foxp3-induction phase). Additional cultures were performed with various concentrations of plate-bound anti-CD3 mAb, soluble anti-CD28, TGF-β, anti–TGF-β mAb, or TGF-β inhibitor, as indicated in the text.
Some CD4+CD25– splenocytes were CFSE labeled prior to cultivation (8). Foxp3 expression was determined by Foxp3/GFP expression or by intracellular Foxp3 staining. CFSE dilution and the expression of CD25, CD69, GITR, CTLA4, Helios, Bcl-2, phospho-Stat5, and Annexin V were analyzed by a FACSCalibur flow cytometer (Becton Dickinson).
CD4+CD25– B6 splenocytes stimulated with 4 μg/ml plate-bound anti-CD3 mAb and 2 μg/ml soluble anti-CD28 mAb were cultured in the presence or absence of anti–IL-2 mAb (JES6, 10 μg/ml), JAK3 inhibitor (CP690550, 50 nM), or STAT5 inhibitor (50 μM). After 3-d activation, the levels of TGF-β and IL-2 in culture supernatants were assessed by ELISA, using commercial kits according to the manufacturer’s instructions (R&D Systems).
Western blot analysis
CD4+CD25– B6 splenocytes stimulated by 4 μg/ml plate-bound anti-CD3 mAb and 2 μg/ml soluble anti-CD28 mAb were cultured in the presence or absence of anti–IL-2 mAb (JES6, 10 μg/ml) or TGF-β (2 ng/ml). Nuclear and cytoplasmic extracts were prepared from 3-d cultured cells, using a commercial lysis buffer (NE-PER Nuclear and Cytoplasmic Extraction Reagents; Thermo Fisher Scientific). Protein concentration was determined by Bradford assay using the Coomassie Protein Assay Reagent (Pierce) and a Bio-Rad spectrophotometer. Equal amounts of protein (30 μg) were loaded in each lane and separated by SDS-polyacrylamide gel (10%) electrophoresis. Fractionated proteins were blotted onto a nitrocellulose membrane (Bio-Rad). Blots were probed with anti-Smad3 (C67H9) rabbit mAb, followed by HRP-labeled goat anti-rabbit IgG (Santa Cruz Biotechnology), and visualized by SuperSignal West Pico Chemiluminescent Substrate (Pierce) and the FluorChem 890 Alpha Innotech) detection system. To detect the phospho-Smad3 expression, the probed membrane was stripped (2% SDS, 62.5 mM Tris-HCl, 100 mM 2-ME, and pH adjusted to 6.5) and reprobed with Phospho-Smad3 (Ser423/Ser425) (C25A9) (Cell Signaling Technology), followed by HRP-labeled goat anti-rabbit IgG (Bio-Rad), and visualized as above. The β-actin and lumin B served as internal loading controls for cytoplasmic and nuclear proteins, respectively.
Treg restimulation and in vitro suppression assay
CD4+CD25–Foxp3/GFP– cells were stimulated by anti-CD3/anti-CD28 mAbs for 3 d in the presence of 2 ng/ml TGF-β and 10 U/ml IL-2 (TGF-β–iTreg group), 10 μg/ml S4B6 anti–IL-2 mAb (anti–IL-2–iTreg group), or 50 nM CP690550 (CP690550-conditioned Foxp3/GFP+ adaptive regulatory T cell [CP-iTreg] group). Cultured cells were washed and rested with 10 U/ml IL-2 for another 3 d prior to sorting out Foxp3/GFP+ cells. Ex vivo sorted CD4+Foxp3/GFP+ splenocytes served as the nTreg group. These sorted Tregs were recultured in 96-well flat-bottom plates and stimulated by 4 μg/ml plate-bound anti-CD3 mAb and 2 μg/ml soluble anti-CD28 mAb for 3 d, followed by analysis of the Foxp3/GFP expression by flow cytometry.
For the in vitro suppression assay, 5 × 104 per well B6.SJL CD4+CD25– T cells were CFSE labeled, as previously described (8), and were used as responder cells. Suppressors were sorted CP-iTregs or nTregs and were seeded with B6.SJL CD4+CD25– T cells in various ratios. Cells were stimulated for 3 d by 1.5 × 105 per well CD3– syngeneic splenocytes and 0.5 μg/ml soluble anti-CD3 mAb. The proliferation of responder cells was determined by CFSE dilution in CD45.1+ T cells measured by flow cytometry.
Induction and prevention of colitis
Single-cell suspensions were obtained from spleen and mesenteric lymph node of B6 mice, and were labeled with anti-CD45RB–FITC, anti-CD25–PE, and anti-CD4–PE-Cy5. CD4+CD45RBhigh T cells and CD4+CD45RBlowCD25+ nTregs were sorted out with a FACSAria cell sorter (Becton Dickinson). Foxp3/GFP+ CP-iTregs were sorted from the 6-d cultures with CP690550 conditioning and IL-2 resting. A total of 4 × 105 CD4+CD45RBhigh T cells were suspended in 0.2 ml PBS and injected i.v. into each RAG1−/− mouse. Some RAG1−/− mice were also injected with 4 × 105 CD4+CD45RBlowCD25+ nTregs or CP-iTregs. Mice were monitored weekly for body weight and were euthanized 8 wk after T cell transfer. Cecum, proximal colon, midcolon, and distal colon were removed and fixed in 10% buffered formalin, paraffin-embedded, sectioned, and stained with H&E. Inflammation was scored in a blinded fashion, using a previously described scoring system (19).
Statistical analysis was performed using an unpaired, two-tailed, Student t test to calculate p values. Calculated p < 0.05 was considered statistically significant.
IL-2 induces Stat5 phosphorylation and Foxp3 expression in peripheral CD4+CD25+Foxp3− iTreg precursors
The discovery of nTreg precursors within CD25+Foxp3− CD4SP thymocytes and peripheral iTreg precursors within CD4+CD25+Foxp3− cells greatly advanced our understanding of Treg development (12, 20). We confirmed these findings in Foxp3/GFP reporter mice, as ∼ 15% of CD4+CD25+ splenocytes were Foxp3/GFP− cells (Fig. 1A). Ex vivo exposure of purified CD4+CD25+Foxp3/GFP− thymocytes or splenocytes to 10 U/ml IL-2 alone in 3-d cultures resulted in conversion of > 30% (Fig. 1Bii, 1C) and 10% (Fig. 1Biv, 1C) of cultured cells into Foxp3/GFP+ cells, respectively. In contrast, only ≤ 1% of CD4+CD25−Foxp3/GFP− thymocytes or splenocytes upregulated Foxp3 expression in response to IL-2 (Fig. 1Bi, iii, 1C). Therefore, the CD25+Foxp3− population, among both CD4SP thymocytes and CD4+ splenocytes, was always enriched for Treg precursors, which readily expressed Foxp3 upon exposure to IL-2.
Other γc-dependent cytokines, including IL-4, IL-7, and IL-15, were much less effective than IL-2 in inducing Foxp3 expression in the cultures of CD4+CD25+Foxp3− splenocytes (Fig. 1D, upper panels, 1E). A putative explanation may be that IL-2 can potently induce Stat5 activation in CD4+CD25+Foxp3− splenocytes, as shown by flow cytometric analysis of phosphorylated Stat5 (Stat5pY694) (Fig. 1F), whereas none of the remaining tested γc cytokines were as effective as IL-2 in activating Stat5 in CD4+CD25+Foxp3− splenocytes (Fig. 1F). All tested γc cytokines, including IL-2, were ineffective at inducing Foxp3 expression (Fig. 1D, lower panels, 1E) and Stat5 phosphorylation (Fig. 1F) in CD4+CD25−Foxp3− splenocytes. Taken together, these results show that IL-2, but not other γc cytokines, effectively induces Stat5 activation and Foxp3 expression in CD4+CD25+Foxp3− iTreg precursors.
IL-2 deprivation upon TCR stimulation generates iTreg precursors
To further reveal the mechanism of iTreg differentiation, we established a two-step method to delineate the conditions for the formation of iTreg precursors (Fig. 2A). In the initial “conditioning” step, CD4+CD25−Foxp3/GFP− splenocytes isolated from Foxp3/GFP mice were stimulated with plate-bound anti-CD3 mAb and soluble anti-CD28 mAb, and exposed to various cytokines, mAbs, and inhibitors. At 3 d later, cells were washed and rested with IL-2 (10 U/ml) for an additional 3 d; this served as the “Foxp3-induction” step to determine the presence of putative iTreg precursors after conditioning (Fig. 2A).
To assess the role of IL-2 in generating iTreg precursors, we added exogenous IL-2 or anti–IL-2 neutralizing mAbs during the initial conditioning step. As shown in Fig. 2B and 2D, few Foxp3/GFP+ cells could be detected in the 6-d culture of CD4+CD25−Foxp3/GFP− splenocytes, which were treated with or without exogenous IL-2 during conditioning. In contrast, neutralizing IL-2, by addition of an anti–IL-2 mAb (clone S4B6 or JES6) during an initial 3-d conditioning step followed by an IL-2–driven Foxp3 induction step, resulted in a high frequency (40%) of Foxp3/GFP+ cells (Fig. 2C, 2D). The Foxp3/GFP expression was induced during the IL-2–driven Foxp3-induction step (days 4 to 6), as very few Foxp3/GFP+ cells could be detected after 3-d conditioning by neutralizing IL-2 (Supplemental Fig. 1). The results from these experiments suggest that an initial IL-2 deprivation upon TCR stimulation generates iTreg precursors, which then express Foxp3 upon subsequent exposure to IL-2.
To confirm the above finding, IL-2−/− CD4+CD25− T cells were labeled with CFSE and cultured using the same two-step method. After 6 d of culture, the Foxp3 expression was assessed by intracellular staining. To exclude many IL-2−/− T cells that were not activated and that were not proliferating, the flow cytometric analysis of cultured cells was gated only on dividing cells based on the CFSE dye dilution. As shown in Fig. 2E and 2F, ∼ 35% of the divided IL-2−/− CD4+CD25− T cells became Foxp3 positive. Addition of increasing concentrations of exogenous IL-2 during the conditioning step prevented Foxp3 induction in IL-2−/− T cells (Fig. 2E, 2F). Thus, IL-2 deprivation during the TCR-directed conditioning step is critical for the generation of iTreg precursors.
Induction of Foxp3 expression in IL-2−/− CD4+CD25− T cells was prevented when they were cocultured with B6.SJL (CD45.1 congenic mice) CD4+CD25− T cells that can produce IL-2 upon TCR stimulation. This inhibitory effect was dependent on the concentration of B6.SJL cells (Fig. 2G, upper panels, 2H). Moreover, a significant frequency of Foxp3-expressing cells was also detected within the B6.SJL population in the coculture group with the smallest fraction of IL-2–producing B6.SJL cells (Fig. 2G, lower panels, 2H). Therefore, sufficient IL-2 production from T cells upon TCR stimulation inhibits the generation of iTreg precursors.
Blockade of IL-2 signaling upon TCR stimulation generates iTreg precursors that become Foxp3+ iTregs upon exposure to IL-2
The IL-2Rαβγ complex associates with the intracellular tyrosine kinases: Jak1, which binds to the IL-2Rβ-chain, and Jak3, which binds to the IL-2Rγ-chain. Binding of IL-2 to high-affinity IL-2R results in autophosphorylation and activation of Jak1 and Jak3. We used an anti–IL-2 mAb (clone S4B6 or JES6), an anti-CD25 (IL2Rα) mAb (clone PC61), or a Jak3 inhibitor (CP690550) to define the role of IL-2 signaling in the formation of iTreg precursors. CD4+CD25−Foxp3/GFP− splenocytes isolated from Foxp3/GFP mice were cultured using the two-step method described in Fig. 2A. Addition of anti–IL-2 mAb, anti-CD25 mAb, or CP690550 during the conditioning step did not induce Foxp3-expressing cells in 3-d cultures (Supplemental Fig. 1). However, subsequent exposure of these cells to 10 U/ml IL-2 resulted in an increase in Foxp3/GFP+ cells to 30–50% in the 6-d cultures (Fig. 3A, 3C). Thus, initial blockade of IL-2 signaling by different methods upon TCR stimulation generates iTreg precursors, which in turn express Foxp3 upon IL-2 exposure.
During the TCR-directed conditioning step, anti–IL-2 mAb, anti-CD25 mAb, or CP690550 only marginally downregulated the phosphorylation of Stat5Y694 (Supplemental Fig. 2A). Moreover, in contrast to 250 μM Stat5 inhibitor, 50 μM Stat5 inhibitor did not decrease the phosphorylation of Stat5Y694 (Supplemental Fig. 2B). However, addition of 50 μM Stat5 inhibitor (but not Stat3 inhibitor) during the conditioning step was sufficient to generate 30–50% Foxp3/GFP+ cells in the 6-d culture of CD4+CD25−Foxp3/GFP− cells (Fig. 3B, 3C). Further studies are required to define how Stat5 and other transcription factors influence the programming of iTreg precursors.
We also investigated the effects of cytokines on the second Foxp3-induction step (Fig. 3D). Following an initial 3-d conditioning step of CD4+CD25−Foxp3/GFP− cells with Jak3 inhibitor (CP690550), the culture was washed and supplemented with 12 different cytokines. As shown in Fig. 3D, IL-2 alone was the most potent cytokine for inducing Foxp3, whereas IL-7 and IL-15 were much less effective and other cytokines were not effective; IL-4 inhibited Foxp3 expression.
Various concentrations of plate-bound anti-CD3 mAb and soluble anti-CD28 mAb were also investigated in the two-step generation of iTregs. As shown in Fig. 3E, in the presence of CP690550 during the conditioning step, 4 μg/ml anti-CD3 mAb induced more iTregs than 0.5 μg/ml anti-CD3 mAb (Fig. 3Eii). Thus, proper TCR stimulation is required to induce iTregs in this system. In the absence of CP690550 conditioning, none of the anti-CD3/CD28 mAb concentrations used was able to generate iTregs. Thus, conditioning of naive T cells by TCR stimulation and IL-2 signaling blockade transforms them into iTreg precursors, which are robustly induced by IL-2 to become Foxp3-expressing iTregs.
Characterization of CD4+ T cells during iTreg generation
By CFSE labeling of CD4+CD25− cells prior to cultivation, using the two-step method, we showed that a significant cell proliferation was retained in cultures from day 2 to day 6 (Fig. 4A), regardless of the conditioning treatment with 10 μg/ml anti–IL-2 mAb or 50 nM CP690550. Foxp3 expression in cells could not be significantly induced unless IL-2 was neutralized or IL-2 signaling was blocked during the initial conditioning step (Fig. 4A–C). Addition of IL-2 into the conditioned cultures at day 3 then promptly induced Foxp3 expression within 12 h, and further enhanced the frequency (Fig. 4A, 4B) and total number (Fig. 4C) of Foxp3-expressing cells in the following 3 d. Thus, these two events must happen in sequence for an efficient induction of precursors and a timely switch to iTregs. Moreover, a similar frequency of Foxp3+ cells was detected in each divided cell population (Supplemental Fig. 3A, 3B). It is possible that during the Foxp3 induction step, IL-2 upregulates Foxp3 expression equally in each dividing cell population.
Significant cell proliferations were observed during iTreg generation (Fig. 4A), suggesting that this two-step process did not inhibit T cell activation. Indeed, the evaluation of T cell activation markers CD25 and CD69 after stimulation of CD4+CD25− wild-type T cells showed that T cell activation was not dramatically impaired by the conditioning treatment with 10 μg/ml anti–IL-2 mAb or 50 nM CP690550 (Fig. 4D). During the conditioning step, 10 μg/ml anti–IL-2 mAb was sufficient to neutralize IL-2 in the 3-d cultures. Nevertheless, blocking IL-2 signaling with 50 nM Jak3 inhibitor or 50 μM Stat5 inhibitor reduced (but did not completely inhibit) IL-2 production (Supplemental Fig. 3C).
We also investigated the possible proapoptotic effect produced in the conditioning step. As shown in Supplemental Fig. 4A and 4B, the conditioning treatment with anti–IL-2 mAb, anti-CD25 mAb, or Jak3 inhibitor (CP690550) at the indicated concentrations did not affect the proliferation of cells during iTreg generation, and it did not increase the frequency of apoptotic cells (annexin V positive) in the cultures. In fact, TCR stimulation enhanced expression of the antiapoptotic molecule Bcl-2, and the conditioning treatment with 10 μg/ml anti–IL-2 mAb or 50 nM CP690550 did not inhibit Bcl-2 expression (Fig. 4E). These results demonstrate that in response to TCR stimulation, CD4+CD25− cells receiving diminished IL-2 signaling can be activated to proliferate, but their differentiation is directed into iTreg precursors. The described culture conditions do not diminish the survival of T cells despite the use of mAbs and inhibitors at the indicated concentrations.
Inhibition of TGF-β signaling upon TCR stimulation reduces iTreg differentiation
Addition of exogenous TGF-β during T cell activation potently upregulates Foxp3 expression in T cells. We did not add exogenous TGF-β to the cultures during the process of iTreg generation, and the TGF-β concentration in the cultures was low (between 40 and 55 pg/ml), as shown in Fig. 5A. Addition of 2 ng/ml exogenous TGF-β upon TCR stimulation increased the expression level of phosphorylated Smad3 both in the cytoplasmic fraction of T cells (Fig. 5B, left panel) and more predominantly in the nucleus (Fig. 5B, right panel). In contrast, neutralizing IL-2 in the absence of exogenous TGF-β did not upregulate the expression of phosphorylated Smad3 in both locations (Fig. 5B). Thus, enhanced TGF-β signaling via activation of Smads is dispensable for the IL-2 deprivation–mediated generation of iTreg precursors.
To test whether TGF-β signaling was required for the formation of iTreg precursors, CD4+CD25− T cells were stimulated with anti-CD3/anti-CD28 mAbs in the presence of conditioning treatment with 10 μg/ml anti–IL-2 mAb alone, or anti–IL-2 mAb plus either an anti–TGF-β mAb (9016 or 1D11; 5 μg/ml) or a TGF-β signaling inhibitor (SB 431542; 1 μM). At 3 d later, cells were washed and recultured with 10 U/ml IL-2. We found that neutralization of TGF-β or inhibition of TGF-β signaling resulted in a significant reduction in the frequency of Foxp3-expressing cells in the 6-d cultures when compared with the conditioning treatment with anti–IL-2 mAb alone (Fig. 5C, 5D). Furthermore, we assessed iTreg differentiation using CD4+CD25− splenocytes isolated from dnTGFβRII mice, which express a dominant-negative form of the human TGF-β receptor II under the direction of the mouse CD4 promoter. Again, only 5–7% of Foxp3-expressing cells were present in the 6-d cultures that received conditioning treatment with anti–IL-2 mAb, anti-CD25 mAb, or Jak3 inhibitor followed by exposure to IL-2 (Fig. 5E, 5F). These results reveal that although significant TGF-β signaling via activation of Smads was not observed, the blockade of TGF-β signaling reduced IL-2 deprivation–mediated generation of iTreg precursors.
We further investigated whether the levels of TGF-β signaling affect the two-step generation of iTregs. CD4+CD25− T cells were stimulated with anti-CD3/anti-CD28 mAbs in the presence of conditioning treatment with 50 nM CP690550. At 3 d later, cells were washed and recultured with 10 U/ml IL-2. As shown in Fig. 5Gi, addition of anti–TGF-β mAb or TGF-β inhibitor during the CP690550-conditioning step inhibited iTreg generation. By contrast, addition of exogenous TGF-β during the conditioning step dose-dependently increased the iTreg generation and could induce ∼ 90% Foxp3+ cells in the day 6 cultures (Fig. 5Gi). However, addition of exogenous TGF-β or blockade of TGF-β signaling during the IL-2–exposing step only mildly modulated the iTreg generation (Fig. 5Gii). Therefore, TGF-β signaling promotes iTreg generation at the early stage after TCR stimulation.
Two-step–generated iTregs prevent experimental colitis
To characterize the phenotype and function of the CD4+Foxp3+ iTregs generated under IL-2 signal deprivation conditions in response to TCR stimulation, we examined additional membrane and intracellular markers, the stability of Foxp3 expression, and the suppressive function of these cells. With our standard procedure, CD4+CD25–Foxp3/GFP– T cells were stimulated for 3 d with anti-CD3/anti-CD28 mAb in the presence of CP690550, followed by exposure to IL-2 for an additional 3 d. Flow cytometric analysis showed that these CP-iTregs in the 6-d cultures expressed high levels of CD25, GITR, and CTLA4 on their surface, a cell surface phenotype similar to that of ex vivo sorted CD4+Foxp3/GFP+ nTregs (Fig. 6A).
When evaluated by previously described methods (18, 21), nTregs, but not CD4+Foxp3/GFP+ iTregs (generated with exogenous TGF-β; TGF-β–iTreg), stably maintained their Foxp3 expression upon repeated stimulation (Fig. 6B, left two panels). Of interest, upon repeated TCR stimulation, 60–75% of the CD4+Foxp3/GFP+ iTregs obtained by our two-step culture conditions maintained their Foxp3 expression (Fig. 6B, right two panels, anti–IL-2–iTreg and CP-iTreg). Nevertheless, CP-iTregs expressed lower levels of Helios than did nTregs (Fig. 6Ci versus iii). Helios expression was also low in TGF-β–iTregs (Fig. 6Cii), ex vivo isolated CD4+CD25+Foxp3– splenocytes (Fig. 6Civ), and Foxp3+ cells generated from the cultures of ex vivo isolated CD4+CD25+Foxp3– splenocytes after IL-2 exposure (Fig. 6Cv).
To investigate the suppressive function of two-step–generated iTregs, we cultured nTregs or CP-iTregs together with CFSE-labeled syngeneic CD4+CD25– T cells, which were stimulated by syngeneic APCs and soluble anti-CD3 mAb. We found that nTregs and CP-iTregs suppressed CD4+CD25– T cell proliferation in a similar fashion (Fig. 6D). On the basis of these experiments, we concluded that IL-2 signal deprivation–conditioned CP-iTregs exhibit relatively stable expression of Foxp3 and exert suppressive function in vitro similar to that in nTregs, but remain different from nTregs in Helios expression.
To assess the in vivo suppressive function of CP-iTregs, we transferred sorted CD4+CD45RBhigh T cells alone, or together with either CP-iTregs or nTregs, into Rag1−/− mice. Host mice receiving CD4+CD45RBhigh T cells alone dramatically lost body weight within 8 wk after cell transfer, whereas host mice receiving CD4+CD45RBhigh T cells, together with either CP-iTregs or nTregs, remained healthy without losing body weight (Fig. 6E). The average histology scores of nTreg- or CP-iTreg–treated groups were similar and were indicative of a low degree of inflammation compared with scores of mice that did not receive Tregs (Fig. 6F, 6G). Thus, CP-iTregs are suppressive in vivo, and prevent experimental colitis in mice.
IL-2 is the principal T cell growth factor normally produced by Teff cells in response to antigenic stimulation. Binding of IL-2 to high-affinity IL-2R on Teff cells upon Ag stimulation positively regulates the magnitude and duration of Teff cell responses, and is even required for the generation of functional memory T cells (3, 22). Nevertheless, IL-2 also plays a critical role in maintaining tolerance by controlling Treg homeostasis and function (3, 7). In particular, IL-2 is the major growth factor for CD4+CD25+Foxp3+ Tregs. The lethal autoimmune phenotype observed in mice deficient in IL-2, IL-2Rα, or IL-2Rβ is attributable to a Treg defect (4, 5). Moreover, an IL-2:anti-IL-2 mAb complex has been shown to dramatically expand Tregs in vivo, and prevent experimental autoimmune encephalomyelitis as well as islet allograft rejection in mice by expanding Tregs (23, 24). Low-dose IL-2 treatment also reverses the onset of type 1 diabetes in NOD mice through enhancing CD25 and Foxp3 expression in pancreatic Tregs, which may subsequently upregulate Treg function (25). More profoundly, IL-2 is also involved in Treg generation, and induction of Foxp3 expression in CD25+Foxp3– CD4SP thymocytes and peripheral CD4+CD25+Foxp3– T cells (12, 20). Because Treg generation also requires TCR stimulation, we sought to understand how IL-2 signals during T cell priming influence different outcomes (e.g., generating Teff cell versus iTreg responses) in T cells.
In this study, we investigated the role of IL-2 signaling in iTreg generation, using a two-step differentiation model, which includes an initial TCR-driven conditioning step generating iTreg precursors and a subsequent cytokine-driven step inducing Foxp3 expression in those precursors. In the TCR-driven conditioning step, inhibition of IL-2 signaling via Jak3–Stat5 conditions naive CD4+CD25−Foxp3− T cells to become CD4+CD25+Foxp3− cells, which contain iTreg precursors. In contrast, sufficient IL-2/Jak3/Stat5 signals during TCR activation abrogate the generation of iTreg precursors, reflecting the fact that quantitative differences in IL-2 signaling at this differentiation step led to distinct outcomes of either dominant Teff cell responses or iTreg responses. Currently, no specific marker for iTreg precursors is recognized as a CD4+CD25+Foxp3– population, and their bona fide presence may be determined by the consequent expression of Foxp3 upon exposure to cytokines. Among 12 tested cytokines, IL-2 was the most potent cytokine to induce Foxp3 expression in iTreg precursors, whereas IL-7 and IL-15 were less effective. These results may suggest that IL-2 deficiency protects naive T cells from apoptosis by switching them to an iTreg precursor line, but their further development is dependent upon timely exposure to γc cytokines to induce Foxp3 expression.
Differentiation of naive T cells into various Teff cell subsets requires TCR stimulation and distinct polarizing cytokines. For iTreg differentiation, TGF-β appears to be the most potent polarizing cytokine (15, 26). In most in vitro studies, high amounts of exogenous TGF-β (at the nanogram per milliliter level) are needed to convert TCR-stimulated naive CD4+ T cells into Foxp3-expressing iTregs (15, 18). Such high amounts of TGF-β may be rare under physiological conditions. Moreover, in the presence of exogenous TGF-β (at the nanogram per milliliter level), addition of IL-2 during the initial TCR stimulation does not inhibit iTreg generation and even facilitates the generation of high frequencies of iTregs (16, 17). Nevertheless, these TGF-β–iTregs promptly lose Foxp3 expression upon TCR restimulation (Fig. 6B). We demonstrated in our two-step process of iTreg differentiation that exogenous TGF-β is not required, as the need for downstream Smad2/Smad3 signaling seemed to be minimal. However, a low level of TGF-β signaling is needed with a yet undefined TGF-β origin, as neutralizing TGF-β or blocking TGF-βR signals decreased iTreg generation. Addition of exogenous TGF-β during the conditioning step dose-dependently increased the iTreg generation and could induce ∼ 90% Foxp3+ cells in the day 6 cultures (Fig. 5Gi). Therefore, two-step iTreg development is favored when environmental TGF-β levels are relatively high.
The transcription factor Stat5 is part of the downstream IL-2 signaling pathway. Activated Stat5 translocates to the nucleus and binds to the promoter region and an intronic regulatory DNA element within the Foxp3 locus (11, 27, 28), suggesting its direct role in transcriptional regulation of Foxp3. Indeed, Stat5−/− mice that survive for only 6 to 8 wk generate very few Foxp3-expressing CD4 cells, a phenotype similar to that of Jak3−/− and Il2rg−/− mice (27). In addition, expressing a constitutively active Stat5 in mice leads to expansion of Tregs and restores Treg numbers even in the absence of IL-2 (11, 13). Thus, Stat5 is required for Treg development and homeostasis. The intriguing findings of this paper do not contradict these previous reports, as IL-2/Stat5 signaling is needed to induce Foxp3 expression in iTreg precursors. However, the striking observation of our work reveals that TCR stimulation under a low IL-2/Stat5 signaling condition (i.e., IL-2–deficient T cells, IL-2 neutralizing mAb, IL-2Rα blocking mAb, as well as Jak3 or Stat5 inhibition) generates iTreg precursors. Recently, analyzing mice with a deletion of the Stat5a/b amino termini or with a T cell–specific deletion of Stat5 showed that these mice have residual Stat5 function and thus have some Tregs (11, 27), suggesting that a low Stat5 signaling threshold allows for the development of Treg precursors. Further studies should elucidate the delicate balance between the effects of Stat5 signaling on the generation of iTreg precursors and the differentiation into Teff cells.
T cells produce IL-2 in response to TCR stimulation. We now show that this natural process not only optimizes Teff cell function but also restrains the formation of iTreg precursors. This finding implies that iTreg differentiation in vivo is limited under immunogenic Ag stimulation. Indeed, iTreg generation is generally identified under certain defined conditions, such as suboptimal Ag presentation and chronic low-dose exposure to Ags, and in unique environments, such as GALTs (10, 29). In particular, Kretschmer et al. (30) have shown that the conversion of truly naive CD4 T cells into iTregs in vivo was achieved by minute Ag doses with suboptimal dendritic cell activation. More strikingly, in the same host, adoptively transferred IL-2−/− T cells converted into iTregs 2- to 3-fold more efficiently than did IL-2–competent T cells (30). Our study likely provides an insight into the mechanism of their finding, as low IL-2/Stat5 signaling in IL-2−/− T cells during the TCR-directed conditioning step induces the formation of iTreg precursors, which may subsequently receive paracrine IL-2 from IL-2–competent T cells for Foxp3 induction. We thus speculate that when T cells receive suboptimal or chronic Ag stimulation in the absence of robust IL-2 signaling some of these cells develop into Treg precursors that then remain in the environment, where, over time, sufficient levels of environmental IL-2 for Foxp3 induction become available. This two-step iTreg generation still requires low levels of TGF-β and should be favored in unique environments where TGF-β levels are relatively high, such as GALTs (10, 29). Compared with iTreg generation by TCR stimulation in the presence of high amounts of exogenous TGF-β (at the nanogram per milliliter level) and IL-2, our model may mimic and provide insights into the biological conditions favorable for iTreg generation, as high amounts of IL-2 and TGF-β (at the ng/ml level) are not often present at the same time physiologically.
In summary, IL-2/Stat5 signaling during TCR stimulation negatively regulates the formation of iTreg precursors, but positively regulates the subsequent Foxp3 expression in these precursors. The iTregs generated by our two-step method showed relatively stable Foxp3 expression upon TCR restimulation (in comparison with exogenous TGF-β–induced iTregs) and prevented CD4+CD45RBhigh cell–mediated colitis in Rag1−/− mice. Importantly, the number of generated iTregs in the 6-d culture was 5-fold more than the initial number of cultured naive T cells. These aspects of our two-step–generated iTregs may facilitate the clinical application of iTregs as a cell therapy for immune disorders.
This work was supported by National Institutes of Health Grants HL69723 and P30 DK079638, American Heart Association Grant 11SDG7690000, and a University of Toledo Biomedical Research Innovation Award.
The online version of this article contains supplemental material.
Abbreviations used in this article:
CD4 single positive
CP690550-conditioned Foxp3/GFP+ adaptive regulatory T cell
adaptive, or inducible, regulatory T cell
natural regulatory T cell
effector T (cell)
regulatory T cell.
The authors have no financial conflicts of interest.