Reciprocal TH17 and Regulatory T Cell Differentiation Mediated by Retinoic Acid

See allHide authors and affiliations

Science  13 Jul 2007:
Vol. 317, Issue 5835, pp. 256-260
DOI: 10.1126/science.1145697


The cytokine transforming growth factor–β (TGF-β) converts naïve T cells into regulatory T (Treg) cells that prevent autoimmunity. However, in the presence of interleukin-6 (IL-6), TGF-β has also been found to promote the differentiation of naïve T lymphocytes into proinflammatory IL-17 cytokine-producing T helper 17 (TH17) cells, which promote autoimmunity and inflammation. This raises the question of how TGF-β can generate such distinct outcomes. We identified the vitamin A metabolite retinoic acid as a key regulator of TGF-β–dependent immune responses, capable of inhibiting the IL-6–driven induction of proinflammatory TH17 cells and promoting anti-inflammatory Treg cell differentiation. These findings indicate that a common metabolite can regulate the balance between pro- and anti-inflammatory immunity.

Helper T cells perform critical functions in the immune system through the production of distinct cytokine profiles. In addition to T helper-1 (TH1) and TH2 cells, a third subset of polarized effector T cells characterized by the production of interleukin-17 (IL-17) and other cytokines (and now called TH17 cells) is associated with the pathogenesis of several autoimmune conditions (1). Transforming growth factor (TGF)–β is a suppressor of TH1 and TH2 cell differentiation and drives the conversion of T cells to those with a regulatory phenotype [so called regulatory T (Treg) cells]. In contrast to the suppression of TH1 and TH2 cells, in vitro activation of naïve T cells by dendritic cells (DCs) and TGF-β, together with proinflammatory cytokines including IL-6, leads to the differentiation of TH17 cells (24). These observations indicate that the priming of T cells by DCs in the presence of TGF-β might lead to opposing immune consequences.

Previous studies have indicated that mucosal DCs are immune suppressive (5), leading us to compare the ability of gut-associated DCs and peripheral DCs in driving TH17 cell differentiation. Mesenteric lymph node (MLN) and spleen DCs were thus used to stimulate ovalbumin (OVA) peptide (OVAp)–specific, OT-II T cell receptor (TCR) transgenic CD4 T cells (Fig. 1A) or α-CD3–stimulated polyclonal CD4 T cells (fig. S1A) (6). In the presence of IL-6 and TGF-β, MLN DCs displayed a reduced capacity to induce TH17 cell differentiation, as compared with their splenic counterparts. Mucosal DCs have been shown to imprint primed T cells with gut tropism, and the molecular mechanism for this has been attributed to DC-derived retinoic acid (RA) (7). To investigate if the reduced capacity of MLN DCs to drive TH17 cell differentiation might also be controlled by RA, we included the RA receptor (RAR) antagonist LE135 (8) during the in vitro activation of T cells under conditions that promote TH17 cell differentiation. In these experiments, the relative inefficiency of MLN DCs to mediate TH17 cell differentiation was reversed such that they primed T cells at levels similar to those of spleen DCs. In comparison, the addition of all-trans RA to the cultures inhibited the TH17 cell differentiation by spleen DCs (Fig. 1, A and B; fig. S1A; and SOM text 1). In addition to CD4+ TH17 cells, we observed that wild-type (Fig. 1C) or OT-I TCR+ (fig. S1C) cytotoxic CD8+T lymphocytes activated in the presence of TGF-β and IL-6 also generated IL-17+T cells and that RA was again able to inhibit this process. This finding indicated that T cells can become IL-17+cells, regardless of their effector phenotype, and that RA specifically suppresses IL-17 expression. TH17 cells can also be generated in the absence of DCs if additional cytokines, such as IL-1 and TNF-α, are included in the culture conditions. Under such antigen-presenting cell (APC)–free conditions, RA also inhibited the generation of IL-17+ T cells, demonstrating that RA targets T cells directly (fig. S1D).

Fig. 1.

RA suppresses TH17 cell differentiation. (A) IL-17 and IFN-γ staining of gated TCR Vβ5+CD4+ spleen cells from OT-II TCR transgenic mice. CD4+ CD25cells were stimulated with OVAp and MLN or spleen (SPL) DCs and, where indicated, with exogenous cytokines and LE135 or all-trans RA. (B) IL-17 enzyme-linked immunosorbent assay (ELISA) of the culture supernatants from (A) and also with 9-cis-RA (9-cis) (mean ± SD). (C) Intracellular IL-17 and IFN-γ staining of gated TCRβ+CD8+ cells. Total CD8+spleen T cells were stimulated with anti-CD3ϵ, and spleen APCs were stimulated with the indicated cytokines and RA. (D) RORγt mRNA analyzed at various times by polymerase chain reaction in CD4 T cells stimulated with anti-CD3ϵ and anti-CD28. Where indicated, IL-17–inducing cytokines (IL-17 cond.) (black) or these cytokines plus RA (white) were included (mean ± SD). Representative data from four experiments are shown. (E) Intracellular IL-17 and IFN-γ staining of CD4+T cells from small intestine lamina propria 5 days after oral infection with Lm. Data are representative of three to four mice per group.

RORγt is an orphan nuclear receptor that has been implicated in the gene transcription of TH17 cells (9). To test whether RA controls RORγt, we activated CD4 T cells under TH17 culture conditions, with or without RA. In the presence of inflammatory cytokines, TGF-β induced high levels of RORγt, whereas added RA greatly reduced its expression (Fig. 1D).

To investigate the apparent suppressive effect of RA on TH17 cell development in vivo, mice were orally infected with Listeria monocytogenes (Lm) and treated with RA or an RAR inhibitor (LE540) (8). A measurable reduction of TH17 mucosal T cells was seen in the animals that received RA, whereas RAR inhibitor–treated mice showed no apparent difference as compared to the controls (Fig. 1E). Collectively, these experiments suggest that RA acts to oppose TH17 cell development in vitro and in vivo, and that this appears to operate directly on T cells via the reduction of RORγt.

The inefficiency of MLN DCs in promoting TH17 cell differentiation and the reciprocal TGF-β–dependent conversion to either TH17 or Treg cells led us to investigate whether mucosal DCs might drive enhanced TGF-β–dependent Treg cell differentiation (fig. S2, A and B, and SOM text 1). Because TGF-β–dependent Treg cells can be identified by expression of the forkhead/winged-helix transcription factor (Foxp3) (1013), we tested for Foxp3 induction in OT-II CD4 cells cultured with spleen or MLN DCs in the presence of TGF-β and OVAp. In contrast to TH17 cell differentiation that was seen with splenic DCs, MLN DCs were able to induce higher frequencies of Foxp3+cells (Fig. 2A). In the presence of an RAR inhibitor, TGF-β–dependent Foxp3 induction by MLN DCs was reduced, but it was enhanced by the addition of RA (Fig. 2A). In addition, RA promoted the expression of cytotoxic T lymphocyte antigen 4 (CTLA-4), a cell-surface receptor typically expressed by Treg cells, on most TGF-β–generated Foxp3+cells (Fig. 2B and SOM text 2). The synergistic effect of RA on TGF-β–dependent Foxp3+T cell differentiation was also apparent with CD8+T cells (Fig. 2C), suggesting that the RA-mediated increase of Foxp3+T cell differentiation might not be limited to CD25+CD4+Treg cells (fig. S2, C and D, and SOM text 3). Overall, these data indicate that RA controls the reciprocal differentiation of TGF-β–dependent Treg and TH17 cells.

Fig. 2.

RA with TGF-β induces Foxp3 and gut-homing receptors. (A) Intracellular staining for Foxp3 and surface staining for CD103 of gated TCR Vβ5+CD4+cells from OT-II TCR transgenic mice. CD4+CD25T cells were stimulated with OVAp and MLN or SPL DCs and, as indicated, with TGF-β1 and LE135 or RA. (B) Intracellular Foxp3 and CTLA-4 staining of OT-II TCR CD4+CD25T cells stimulated as above, except with spleen APCs instead of DCs. (C) CD8+T cells from OT-I TCR transgenic mice were stimulated with OVAp and spleen DCs with TGF-β1 and RA. Intracellular staining of gated TCRβ+ cells for Foxp3 is shown. (D) Cell-surface staining of gated TCRβ+CD4+cells for CD103, α4β7, and CCR9. CD4+CD25T cells were stimulated with soluble anti-CD3ϵ and spleen APCs plus TGF-β1, RA, or TGF-β1 and RA. Isotype controls are indicated with solid gray histograms. Representative data from three experiments are shown. (E) Percentage of Foxp3+CD4+ cells in CD4+TCRβ+lymphocytes from the small intestine lamina propria 5 days after oral infection with Lm (left) or in naïve controls (right). Veh, vehicle. Asterisks indicate P < 0.05 (Student's t test). Horizontal bars indicate averages of the mice that are represented individually.

Mucosal DC-derived RA also mediates the induction of gut homing receptors, including the integrin α4β7 and CCR9, specific for homing to the small intestine (7), whereas TGF-β has been found to promote the induction of CD103, the αE subunit of the αEβ7 integrin (14). We therefore tested whether TGF-β and RA might synergize to induce these receptors. Consistent with synergy, RA greatly enhanced TGF-β–mediated CD103 expression. In contrast, TGF-β partially antagonized RA-induced CCR9 (Fig. 2D). These results show that the combination of RA and TGF-β results in CCR9+ Treg cells with likely tropism for the small intestine and CCR9 Foxp3+cells with potentially different homing capacities.

To further test the influence of RA in vivo, we treated Lm-infected mice with RA or an RAR inhibitor. Although RA alone did not measurably enhance the differentiation of Foxp3+Treg cells in vivo, the inhibition of RAR significantly reduced the number of mucosal Foxp3+Treg cells in Lm-challenged mice (Fig. 2E). No effect of RA or RAR on spleen Foxp3+CD4 cells was observed (fig. S3C). The finding that RA combined with TGF-β, but not alone, can drive differentiation of Foxp3+T cells in vitro (Fig. 2A) suggests that TGF-β might be a limiting factor in the lack of Treg cell differentiation induced by exogenous RA in vivo (fig. S3D and SOM text 4).

To examine whether in vitro–generated Foxp3+CD4 T cells could function to suppress effector T cells in vivo, we performed cotransfer experiments using naïve CD45RBhiCD4 T cells that induce colitis in immune-deficient mice (15), in combination with CD4 T cells that were previously cultured under different conditions. Mice that received CD4 T cells activated without cytokine conditioning, combined with naïve CD45RBhiCD4 cells, developed severe colitis (Fig. 3). In contrast, mice that were cotransferred with CD4 T cells activated in the presence of TGF-β were partially protected from disease, and mice cotransferred with naïve T cells and CD4 T cells activated in vitro in the presence of both TGF-β and RA showed no apparent signs of disease (Fig. 3). Furthermore, fewer mucosal CD4 T cells isolated from these animals produced IL-17 and interferon-γ (IFN-γ) (fig. S4). These results suggest that Foxp3+T cells generated in vitro with RAand TGF-β have a measurable regulatory capacity and are able to control inflammation upon transfer in vivo. In addition, whereas Treg cells generated in vitro by TGF-β alone lost Foxp3 expression upon restimulation, most of the RA+TGF-β–differentiated Foxp3+T cells remained Foxp3+ after restimulation, indicating that RA drives the differentiation of a stable Treg cell lineage (Fig. 3D). Conversely, RA and TGF-β also suppressed committed TH17 cells in secondary cultures, whereas TGF-β alone did not (Fig. 3E). The experiments thus far suggest that RA might directly counteract the activity of IL-6, and to test this hypothesis, we analyzed TGF-β–dependent T cell differentiation in the presence of RA, together with IL-6. CD4 (Fig. 4A) or CD8 (Fig. 4B) T cells cultured with RA under conditions that otherwise promote TGF-β–dependent TH17 cell differentiation converted to Foxp3+cells with a decrease in TH17 cell differentiation. The antagonistic effect of RA on IL-6 was dose dependent (fig. S5A), suggesting that under physiological conditions, the RA-driven TGF-β–dependent Treg cell differentiation might override the IL-6–promoted TGF-β–dependent TH17 cell generation.

Fig. 3.

TGF-β plus RA in vitro–differentiated Treg cells regulate in vivo. (A) Hematoxylin and eosin staining of the distal colon of mice deficient in recombination activating gene 1 (RAG-1–/–) 6 to 7 weeks after cotransfer of 5×105 CD4+CD45RBhi cells with 2.5 × 105 CD4+T cells stimulated in vitro with anti-CD3ϵ alone (none) or with TGF-β1 and RA. Original magnification, 40×. Representative data from four mice in each group are shown. (B) Body weight of RAG-1–/– mice after transfer of 5 × 105 CD4+CD45RBhicells with 2.5 × 105 anti–CD3ϵ-stimulated CD4+T cells with no additions (squares), TGF-β 1 (triangles), or TGF-β1 and RA (diamonds). The mean ± SD weight of four animals per group is shown; data are representative of three experiments. (C) Histological scores of the groups described in (B). Horizontal bars indicate averages of the mice that are represented individually. (D) Foxp3 intracellular staining of naïve TCRβ+CD4+ cells that were initially stimulated with soluble anti-CD3ϵ and spleen APCs with the indicated cytokines. The cells were rested for 2 days with IL-2 and then restimulated in the absence of exogenous cytokines before analysis. (E) Intracellular staining for IL-17 of naïve CD4+T cells that were initially stimulated and rested as described in (D), but in the presence of TGF-β and IL-6, and then restimulated in the indicated conditions. The percentage of IL-17+cells in the gated TCRβ+CD4+cells is depicted. Representative data from three experiments are shown.

Fig. 4.

Reciprocal TGF-β–dependent T cell differentiation by IL-6 and RA. (A) Naïve CD4+T cells labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) were stimulated with anti-CD3ϵ, spleen APCs, the indicated cytokines, and RA. TNF-α, IL-1-β, TGF-β, and IL-6 were used to drive IL-17 differentiation. Intracellular staining of gated TCRβ+CD4+cells for Foxp3 and IL-17 is depicted. (B) Intracellular staining for Foxp3 and IL-17 of CD8+T cells stimulated with soluble anti-CD3ϵ and spleen APCs under the indicated conditions. (C and D) Intracellular staining for Foxp3 and surface staining for CD103 (C) or intracellular staining of IL-17 and IFN-γ (D) of naïve CD4+T cells from B7.1/2–/– IL-2+/+ or B7.1/2–/– IL-2–/– mice. Cells were stimulated with soluble anti-CD3ϵ, spleen APCs and the indicated cytokines, and RA and/or a blocking antibody to IL-2 (20 μg/ml). The staining was performed on gated TCRβ+CD4+ cells. (E) ELISA of IL-17 in the supernatant of the cultures in (C) and (D) (mean ± SD). Representative data of two experiments are shown.

IL-2 is required for the production of TGF-β–dependent Foxp3+Treg cells (16), and it was recently shown that exogenously added IL-2 also suppresses TH17 cell differentiation (17). To investigate whether RA-mediated regulation of T cell polarization required IL-2 signaling, we examined RA-mediated effects on T cell differentiation in vitro in the absence of IL-2, using IL-2–specific antibodies (anti–IL-2) or IL-2–/– T cells. In these experiments, the enhanced effect of RA in driving differentiation of Foxp3+cells and the inhibitory effect of RA on TGF-β–dependent TH17 cell differentiation were primarily dependent on IL-2 (Fig. 4, C to E, and fig. S5, B and C). RA-mediated regulation did not require exogenously added IL-2, although when added together, RA and exogenous IL-2 synergized to drive the reciprocal regulation of TGF-β–dependent T cell differentiation (Fig. 4). However, RA- and exogenous IL-2–controlled differentiation appeared distinct, in that RA-mediated TGF-β–dependent Foxp3 differentiation generated mostly CD103+Treg cells, whereas most of the IL-2–driven Foxp3+Treg cells were CD103 (Fig. 4C). Conversely, the mechanism of RA and IL-2 used to suppress TH17 cells also appeared to have some distinctions, because RA measurably decreased IL-17 cytokine secretion, whereas exogenous IL-2 did not (Fig. 4E). Collectively, these experiments show that although both exogenous IL-2 and RA require initial IL-2 signaling for their regulatory function, the cooperation with TGF-β and the further downstream mechanisms they activate might be very different (SOM text 5).

The reciprocal activity of RA in inhibiting TGF-β–dependent TH17 cell generation while promoting Foxp3+Treg cell differentiation might provide a self-correcting mechanism for TGF-β to regulate both pro- and anti-inflammatory immunity. This regulatory capacity has particular relevance for the intestine, where efficient immune protection has to coincide with maintaining the mucosal-barrier integrity.

Vitamin A deficiency causes immune dys-function and increased mortality (18). Our results show that RA-mediated effects might be important in vivo. The immune pathology characteristic of subphysiological levels of RA might in part be attributed to an imbalanced TGF-β function favoring proinflammatory TH17 cells at the expense of anti-inflammatory Treg cells. These insights may stimulate new approaches for the treatment of immune disorders in which the imbalance mediated by TGF-β might be an important contributor to immune pathophysiology.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S5


References and Notes

Stay Connected to Science

Navigate This Article