all-trans retinoic acid, aTreg: adaptive regulatory T cells
forkhead box protein P3
retinoic acid receptor-α
IL-17-secreting T helper
Autoimmunity is thought to reflect an imbalance between regulatory T helper lymphocytes (Treg) and pathogenic, IL-17-secreting T helper (Th17) cells. Induction of both adaptive Treg and Th17 cells requires signalling from TGF-β. We now show that, in the context of TGF-β signalling, all-trans retinoic acid (ATRA) leads to increased induction of CD4+ T cells expressing the Treg specification factor forkhead box protein P3 (FoxP3) and decreased frequency of cells expressing IL-17, even in the presence of IL-6. Using a specific agonist and antagonist, as well as retroviral over-expression, we also provide evidence that the effects of ATRA are likely to be at least partially mediated by the nuclear retinoic acid receptor-α (RARα). These findings indicate that signalling through a specific nuclear retinoic acid receptor can favour the decision to adopt the Treg fate at the expense of Th17 fate. Specific agonists of RARα could, therefore, be considered candidates for the treatment of autoimmunity.
Autoimmune diseases are a major cause of morbidity in the developed world and include conditions such as inflammatory bowel disease, type 1 diabetes, multiple sclerosis and rheumatoid arthritis. CD4+ helper T lymphocytes play an essential role in the defence against pathogens, but are also responsible for the pathogenesis of autoimmunity. Naive CD4+ (helper) T cells can differentiate into several distinct, non-overlapping subsets. Helper T cells secreting IL-17 (Th17 cells) have recently been implicated in autoimmune diseases 1. The orphan nuclear receptor RORγt is the key transcription factor involved in the differentiation of Th17 cells 2. CD4+ regulatory T cells (Treg), which express the transcription factor forkhead box protein P3 (FoxP3), are able to suppress autoimmune processes 3.
Induction of both adaptive Treg (aTreg) and Th17 cells from antigen-activated naive T cells seems to require signalling from the pleiotropic cytokine TGF-β. The differentiation of Th17 cells and aTreg, however, is typically non-compatible with cytokines such as IL-6 favouring induction of Th17 cells at the expense of aTreg in the setting of TGF-β signalling 4, 5. Whether signals exist that can mediate induction of aTreg at the expense of Th17 cells in cooperation with TGF-β is not yet known.
Vitamin A (retinol) and its metabolite, all-trans retinoic acid (ATRA), have been implicated in immune homeostasis. Vitamin A and its derivatives are capable of ameliorating various models of autoimmunity, including inflammatory bowel disease, rheumatoid arthritis, type 1 diabetes, and experimental encephalomyelitis 6–9. Deficiency of vitamin A leads to exacerbation in experimental colitis 10. Supplementation of vitamin A results in decreased prevalence of diarrhoea in children 11, and retinoids are used in the therapy of rosacea and psoriasis 12.
Dietary vitamin A is processed into ATRA by dendritic cells (DC) from mesenteric lymph nodes and Peyer's patches. It has been suggested that ATRA can influence T cell function insofar as it induces the gut homing receptors integrin α4β7 and CCR9 13. Whether ATRA influences the lineage decisions of helper T cells, however, has not been thoroughly explored. Herein, we show that the presence of ATRA during activation of CD4+ T cells with TGF-β favours the development of the aTreg lineage at the expense of T cells secreting IL-17, even in the presence of IL-6. Furthermore, using a highly selective agonist for retinoic acid receptor-α (RARα), we suggest that the effect is at least in part due to the activation of RARα, thereby identifying a plausible new target for the therapy of autoimmunity.
Results and discussion
All-trans retinoic acid induces FoxP3 and represses IL-17 in helper T cells
To investigate the role of retinoids in the induction of Treg and Th17 cells, we activated CD4+ T cells in the presence of TGF-β and ATRA. Three days after T cell stimulation in the presence of TGF-β, we found that the frequency of cells expressing FoxP3 was augmented by the addition of ATRA (Fig. 1A). As it has recently been found that Treg and Th17 cells develop in reciprocal pathways, we wanted to examine the effect ATRA has on Th17 cells. In T cells stimulated in conditions favouring the development of Th17 cells (TGF-β plus IL-6), addition of ATRA led to a substantial decrease in the number of IL-17-expressing cells (Fig. 1B). ATRA thus diverts cells toward the aTreg lineage and away from the Th17 lineage, even in the presence of Th17-inducing cytokine conditions.
RARα activation causes induction of FoxP3
One of the receptors for ATRA is RARα 14. We therefore examined the potential role of RARα in the induction of FoxP3. Western analysis showed no detectable RARα expression in naive T cells. Upon activation for 3 days, RARα expression was induced in T cells cultured in TGF-β, with or without IL-6 (Fig. 2A). Thus, we examined the effect of AM580, a highly specific small-molecule agonist for RARα 15. We found that CD4+ T cells activated in the presence of TGF-β showed increased frequency of FoxP3-expressing cells, when AM580 was added. The frequency of FoxP3-expresssing cells was decreased in the presence of Ro 41-5253, a specific inhibitor of RARα 16 (Fig. 2B). The suppressive action of the inhibitor can be explained by a background level of RARα activators present in the T cell media. Conversely, when we activated T cells in Th17 conditions, the frequency of IL-17 expression was decreased by AM580 (Fig. 2C).
To further evaluate the role of RARα, we constructed a bicistronic retrovirus containing the coding region of RARα followed by the codons of GFP. When AM580 was added, GFP-expressing T cells transduced with RARα retrovirus had a higher frequency of FoxP3 expression than those transduced with the control virus (Fig. 3). We found that the augmentation in the frequency of FoxP3-expressing cells resulting from over-expression of RARα was only moderate (Fig. 3), and conclude that this might be attributable to the high endogenous levels of RARα in activated T cells (Fig. 2A). The only two other subtypes of retinoic acid receptor (RARβ and RARγ) are also activated by ATRA. We cannot exclude the possibility that their activation or even that of other transcription factors is important in the induction of FoxP3 by ATRA. We have shown, however, that RARα specifically induces FoxP3 upon activation.
The mechanism by which RARα induces FoxP3 could be by direct promoter activation or involving intermediate factors. Retinoids are known to induce a family of proteins called CCAAT/enhancer binding proteins 17. A binding site for CCAAT/enhancer binding proteins has recently been identified in the FoxP3 promoter 18. Therefore, it might be useful to investigate agonists of RARα for their potential as anti-inflammatory and anti-autoimmune drugs.
ATRA is known to induce gut homing receptors. Activation of RARα by its natural ligand ATRA might provide an explanation for tolerance in the gut, where the concentration of ATRA is particularly high 13. It is possible, for instance, that T cells reacting to foreign food antigens or commensal bacteria could be diverted from effector differentiation by RARα-mediated induction of aTreg.
Herein, we have provided evidence that ATRA induces expression of FoxP3 and suppresses the expression of IL-17 in CD4+ T cells in the presence of TGF-β. The effect is at least in part due to the activation of RARα based on the use of a highly specific ligand. These results may explain the control of autoimmunity by retinoids that has previously been reported. Ligands for RARα could therefore be lead compounds for drugs against autoimmune diseases. While this manuscript was in preparation, another report indicated that ATRA induces expression of FoxP3 and represses expression of IL-17 19. That study, however, did not implicate RARα as the factor responsible for this important regulatory switch.
Materials and methods
Mice and cell culture
All animal work was done with approval of University of Pennsylvania's Institutional Animal Care and Use Committee. MACS depletion (Miltenyi, Bergisch Gladbach, Germany) was carried out according to manufacturer protocol.
T cell stimulation and retroviral gene transfer were performed as described 20. Briefly, cells from BALB/c mesenteric lymph node and spleen were CD8-depleted by MACS column. Cells were stimulated for 3 days with soluble anti-CD3 (1 µg/mL), anti-CD28 (1 µg/mL), IL-2 (100 U/mL), anti-IFN-γ (10 μg/mL), anti-IL-4 (10 μg/mL). Where indicated, rTGF-β (3 ng/mL) with or without rIL-6 (20 ng/mL) was added to the culture. ATRA (Sigma, St. Louis, MO) in DMSO was used at a final concentration of 1 µM. AM580 (Biomol, Plymouth Meeting, PA), a highly specific agonist for RARα, was used at a final concentration of 1 µM. The specific RARα antagonist Ro 41-5253 was a used at a final concentration of 5 µM. Four hours before harvesting, cells were treated with 50 ng/mL PMA and 500 ng/mL ionomycin, with brefeldin A (10 µg/mL) added for the last 2 h. A mouse stem cell virus-based retroviral vector containing the codons of RARα followed by an internal ribosomal entry sequence and the codons for GFP was utilized as previously described 20 under the conditions indicated in legend to Fig. 3.
Antibodies, flow cytometry and Western blotting
Intracellular staining with antibodies against IL-17 (BD, Franklin Lakes, NJ) and CD4 (Caltag, Carlsbad, CA) was performed according to manufacturer protocol. Intranuclear anti-FoxP3 (eBioscience, San Diego, CA) staining was performed by fixation of cells with 4% paraformaldehyde in PBS, followed by permeabilisation and staining in 0.1% Triton X, 1% FBS in PBS. For immunoblot analysis, protein was isolated and separated in a 4–20% SDS polyacrylamide gel (Invitrogen) and transferred to a polyvinylidene difluoride membrane (Invitrogen). After incubation with the primary antibody RARα (sc-551; Santa Cruz Biotechnology) or the ubiquitously expressed GTPase RAN (BD Biosciences, San Jose, CA), a secondary horseradish peroxidase-conjugated antibody (Invitrogen) was added, and an enhanced chemiluminescent substrate kit (Amersham, Little Chalfont, UK) was used for detection.
We are grateful to K. Gonnella for assistance. These studies were supported by the NIH (AI42370 to S.L.R. and DK43806 to M.A.L.) and the Abramson Family. M. S. was supported by a mentored fellowship Award from the American Diabetes Association.