A retinoic acid–dependent checkpoint in the development of CD4⁺ T cell–mediated immunity

To determine whether RA production and signaling occurs during immune activation, mice were administered immune activators, and WBI was performed. The distribution and intensity of RA signaling after immune activation were determined using a RARE-Luciferase reporter mouse (DR5-Luciferase), which transcriptionally up-regulates luciferase expression upon signaling by RA (Svensson et al., 2008). Constitutive luciferase signal was routinely observed in the abdominal area of mice because of signaling emanating from the area of small and large intestines. Surprisingly, the intradermal administration of α-CD40, α-CD3, CFA, and LPS elicited a robust, enhanced luciferase signal that was localized to the site of administration as early as 24 h after immunization (Fig. 1 A). In addition, after the i.p. administration of CFA, a locally and temporally restricted luciferase signal was first induced over controls in 24 h, but the signal intensified by the following days, with a gradual decrease over the ensuing 20 d (Fig. 1 B). A more in-depth analysis of the temporal and spatial induction of RA reporting and identity of the cells reporting was undertaken after the i.p. administration of CFA and α-CD3. In this case, the administration of CFA triggered RA signaling on CD4⁺ T cells (Fig. 1 C), but to a lesser extent than T cell activation induced by α-CD3, where a heightened activity was observed when compared with IgG control (Fig. 1 D). The in vivo administration of a RA antagonist before T cell isolation ablated the RA reporting, establishing the possibility that the enhanced reporting was specific for RA (Fig. 1, C and D). Therefore, different immune activators stimulate RA signaling in a spatially and temporally restricted manner.

In addition, RA reporting by CD4⁺ T cells was studied in a model of allogeneic skin graft rejection. DR5-Luciferase CD4⁺ T cells were adoptively transferred into RAG^−/− mice before syngeneic or allogeneic skin graft transplantation. After transfer, WBI was used to assess RA reporting by the DR5-Luciferase T cells. As shown in Fig. 2 A, no signal was found at early time points (first 5 d after transplantation); however, between 7 and 21 d after transplant, enhanced RA signaling was detected in allogeneic grafts versus the syngeneic grafts. By week 3 after transplant, RA reporting declined to near baseline levels in both groups. Signaling from the transplanted allograft was associated with the presence of the allograft because RA reporting disappeared when the graft was fully rejected. In Fig. 2 B, quantified photon flux is presented for both mice transplanted with syngeneic and allogeneic grafts. Therefore, alloreactive CD4⁺ T cells that enter an allograft are triggered by RA.

Our findings establish that RA signaling in T cells may be linked with the development of inflammation. The temporally restricted and regionally controlled signaling by RA suggests that RA regulates immunity within the inflammatory site over a prescribed period of time. In this study, polyclonal T cell activation in vivo augments RA signaling (or sensing) by CD4⁺ T cells. Although gut-resident DCs (Coombes et al., 2007; Sun et al., 2007) have been identified as the major source of immune-derived RA, it is likely that RA production by DCs, other APCs, and even nonhematopoietic cells may be responsible for the reporting that is observed in intact mice. These observations are supported by several studies, in which not only gut-resident DCs are able to synthesize RA but also draining LN (dLN)– and skin-resident DCs (Guilliams et al., 2010), as well as nonhematopoietic cells (Hammerschmidt et al., 2008; Molenaar et al., 2009).

To analyze the functional impact of intrinsic RA signaling on CD4⁺ T cell population dynamics during graft rejection, RA signaling was selectively ablated in graft-specific T cells. A dominant-negative form of the RAR-α, RAR403 (hereafter denoted dnRAR-α; Rajaii et al., 2008), was overexpressed in T cells by interbreeding the dnRAR-α with CD4^Cre mice.

The potential impact of dnRAR-α gene expression on T cell ontogeny in the dnRAR-αCD4^Cre was addressed. Analysis of the T cell development indicated that overexpression of the dnRAR-α gene impacted the frequencies of thymic single-positive CD8⁺ T cell populations, as previously described (Zhou et al., 2008), but had little effect on T cell populations in the periphery (Fig. S1). Overexpression of the dnRAR-α did not affect α-CD3–induced T cell proliferation in vitro or in vivo (Fig. S2) but, as anticipated, did inhibit RA-induced TGF-β–dependent FoxP3 conversion (Fig. S3; Benson et al., 2007; Coombes et al., 2007; Sun et al., 2007). To study the impact of RA signaling on graft rejection, the dnRAR-α gene was also overexpressed in OVA-specific OT-II TCR transgenic T cells, and mice were grafted with skin that expresses OVA as a transplantation rejection antigen (Ehst et al., 2003). C57BL/6 mice received dnRAR-α–OT-II–Ly5.2 or dnRAR-αCD4^Cre–OT-II–Ly5.2 cells by i.v. injection, and the next day, mice were transplanted with either syngeneic (B6) or allogeneic (actin-OVA) grafts. After 7 d, skin grafts and dLNs were removed and analyzed for OT-II–CD4⁺ T cell infiltration. Fig. 3 (A and B) shows that the alteration of RA signaling in CD4⁺ T cells impairs accumulation of T cells at the site of inflammation (skin graft). No differences in the expression of either activation markers (CD44, CD69, and CD62L) or chemokine receptors that are up-regulated upon inflammation and are associated with migration to inflamed sites (CCR4, CCR5, CCR6, CXCR3, and CCR10) were observed (Fig. S4). Therefore, early TCR-mediated events and expansion of graft-specific T cells appeared intact in the absence of RA signaling, but tissue accumulation was impaired.

It appeared that clonal T cell proliferation was intact in the absence of RA signaling, and so additional experiments were designed to evaluate the impact of RA signaling on T cell differentiation. Because of the difficulty of recovering sufficient numbers of transferred CD4⁺ T cells when the host is on the WT C57BL/6 background, skin-grafted RAG^−/− mice were used as recipients to ensure a high recovery and purity of OT-II–CD4⁺ T cells. dnRAR-α–OT-II–Ly5.2 or dnRAR-αCD4^Cre–OT-II–Ly5.2 cells were transferred to RAG^−/− recipient mice and grafted the next day. dLN-resident CD4⁺ T cells were isolated after 7 d and co-cultured with unpulsed or ISQ-pulsed DCs. After 72 h, cytokine and chemokines were quantified. As shown in Fig. 4 A, the production of Th1-type cytokine IFN-γ and IL-17 was significantly reduced (∼60% for both, with P ≤ 0.022 [IFN-γ] and P ≤ 0.028 [IL-17]) when dLN-resident dnRAR-αCD4^Cre–OT-II–Ly5.2 T cells were recalled with their cognate antigen, as compared with their counterpart dnRAR-α–OT-II–Ly5.2 T cells. Chemokine production by recalled dnRAR-αCD4^Cre–OT-II–Ly5.2 and dnRAR-α–OT-II–Ly5.2 T cells was also evaluated. As depicted in Fig. 4 B, the interruption of RAR-α signaling in CD4⁺ T cells blocks the production of Th1-linked chemokines such as MIP-1α (P ≤ 0.014) and MIP-1β (P ≤ 0.008), including a down-regulation in the production of RANTES (Fig. S5; Cook, 1996). Moreover, we observed a tendency to a Th2-skewed/suppressive phenotype characterized by the up-regulation in the production of IL-4 (P ≤ 0.037), IL-9 (P ≤ 0.026; Fig. 4 C), and IL-10 (P ≤ 0.128), although for IL-10 this difference was not statistically significant (Fig. S5). The impairment of Th1-linked chemokines by graft-specific T cells may contribute to the transcriptional program required for Th1/Th17 cell differentiation and the development of CD4⁺ T cell–mediated inflammation.

To evaluate the functional significance of RA signaling to graft rejection, T cell–mediated graft rejection was studied in mice in which RA signaling was ablated chemically and genetically. Mice receiving syngeneic (B6) or allogeneic (actin-OVA) skins were treated with a RAR pan-antagonist (NRX 194310). As shown in Fig. 5 A, the blockade of RAR signaling resulted in complete acceptance of the skin allograft (open red circles). Experiments then evaluated the involvement of RA signaling in CD4⁺ T cells during graft rejection. WT, dnRAR-α, and dnRAR-αCD4^Cre were grafted with syngeneic or allogeneic skin, and the kinetics of graft rejection was evaluated over time. As depicted in Fig. 5 B, dnRAR-α mice reject the allograft following the same kinetics previously described for C57BL/6 (Ehst et al., 2003). However, rejection in the dnRAR-αCD4^Cre mice was completely abolished. Therefore, experiments with an RAR pan-antagonist and overexpression of the dnRAR-α gene in T cells demonstrate that RA signaling is required for T cell–mediated graft rejection.

Our data demonstrate that RA signaling to CD4⁺ T cells is essential for the inflammatory responses that mediate the rejection of allogeneic skin grafts. The absence of RA signaling in CD4⁺ T cells alters T cell polarity, resulting in a shift from a Th1/Th17 to a Th2 phenotype. The selective impact of RA on T cell differentiation and not clonal expansion is consistent with the well-described role of RA as a differentiation factor in developmental biology (Maden, 2007). With the present data considered, it is clear that RA exerts both proinflammatory and antiinflammatory effects on cell-mediated immunity. Numerous studies have implicated RA in facilitating immune suppression via the enhanced differentiation of aT_reg cells (Benson et al., 2007; Coombes et al., 2007; Sun et al., 2007) and the suppression of Th17 cell differentiation (Mucida et al., 2007; Elias et al., 2008). In fact, it has been shown that RA agonists can suppress several murine autoimmune diseases, including systemic lupus erythematosus and experimental autoimmune encephalomyelitis (Yamauchi et al., 2005). In the latter, reduced differentiation of Th17 was attributed to the resolution in disease. Furthermore, a major emphasis for the role of RA in immunity has been placed on its role in gut homeostasis. Studies have shown that CD103⁺ DCs from the gut can potentiate the differentiation of aT_reg cells through their capacity to express RALDH enzymes required for RA synthesis (Coombes et al., 2007; Sun et al., 2007). Although its role in the gut is no doubt critical, the findings reported herein provide a completely new perspective of the role of RA in immunity. In this regard, our preliminary results show that stimuli like T cell activation by α-CD3 antibody treatment increases the frequencies of DCs expressing RALDH. In the skin transplantation model, although we observed heightened RA signaling in CD4⁺ T cells during the rejection process, we did not see a significant difference in the numbers of RALDH⁺ DCs between the syngeneic and allogeneic groups (either dLNs or skin grafts; unpublished data). Thus, signaling of RA to CD4⁺ T cell may not be correlated with differential production from other cell types (APCs and stromal cells). It is known that RA signaling can be regulated by the expression of RA transport proteins expressed by the target cell. Alternatively, RA concentrations in vivo can be regulated by its catabolism by the catabolic enzyme, Cyp26. In this regard, future experiments involving Cyp26, the protein which degrades RA in the cytosol, will be pursued to understand its role in immunity. In contrast, the molecular mechanisms underlying the ability of RA to control the lineage commitment of T cells in vivo are unresolved. Our data show that ablation of RA signaling in vivo reduces the expression of Th1- and Th17-associated cytokines and chemokines. This effect is not seen in vitro when dnRAR-αCD4^Cre T cells are stimulated with α-CD3 (unpublished data). The role of RA in eliciting Th1 and Th17 inflammatory T cells may be through its capacity to influence the expression of master transcription factors that regulate the fate of Th cells, such as T-bet or ROR-γt. Alternatively, RA may control T cell plasticity in vivo, and dnRAR-αCD4^Cre may have impairments in lineage commitment. Studies are underway to evaluate these hypotheses.

In addition, our data show that the disruption of RA signaling in the CD4⁺ T cell compartment impairs T cell infiltration of the skin graft. Although the expression of the major skin/inflammation-related chemokine receptors was not compromised, the added effect of reduced Th1 cytokines and chemokines may cooperate for the poor recruitment of CD4⁺ T effector cells to the allograft. We observed that dnRAR-αCD4^Cre CD4⁺ T cells produce less Th1-linked chemokines, which can explain in part the lack of CD4⁺ T cell accumulation within the graft if chemokine migration by infiltrating T cells amplifies the recruitment of additional T cells to the inflammatory site.

A recent publication (Hall et al., 2011) has highlighted the role of RA signaling of T cells in the mucosal immune response to Toxoplasma gondii. These investigators have shown that VAD impairs IFN-γ recall responses in immune mice. Furthermore, they show that T cells derived from an RAR-α^−/− mouse display abnormalities in cell growth and Ca²⁺ mobilization and produce less IFN-γ in response to vaccination with OVA protein and Escherichia coli toxin. In contrast to these findings, we show that T cells that are unresponsive to RA via overexpression of a dnRAR-α proliferate in response to antigen in vivo and in vitro. An apparent explanation to the differences is not clear at this time. It is possible that T cells derived from RAR-α^−/− are developmentally impaired. Alternatively, it is possible that the signaling function of the RAR-α molecule may still be intact in the dnRAR-α–expressing T cells. Future studies with conditional deletion of each of the RAR will help resolve these issues and shed insights into the role of other RAR in immunity. Nonetheless, collectively, the study by Hall et al. (2011) and our study on the role of RA in immunity provide new insights into the function of this mediator in inflammation.

With regard to cell differentiation and stability, data from experiments examining the role of RA in aT_reg cell generation have implicated RA as an important differentiation factor for T cells, which may impact their plasticity. Data show that RA-treated natural T_reg cells are lineage committed and virtually irreversible in their phenotype, even in the face of extremely proinflammatory conditions (Zhou et al., 2010). As such, RA controls the plasticity of aT_reg cells (Benson et al., 2007; Nolting et al., 2009) and perhaps Th1 and Th17 cells. The enhanced aT_reg cell differentiation by RA has been recently resolved to show that RA enhances the association of the RAR/RXR receptor to the FoxP3 enhancer, leading to enhanced histone acetylation and enhanced binding of phosphorylated Smad3 (Xu et al., 2010). For nearly a century, it has been clear that VAD compromises immunity, and these new findings on the role of RA in controlling T cell–mediated differentiation provide a rational, scientific basis for the increased susceptibility to infectious disease in vitamin A–deficient individuals.

C57BL/6, RAG^−/−, CD4^Cre, and actin-OVA (Ehst et al., 2003) mice were purchased from the Jackson Laboratory. The DR5-Luciferase reporter (Svensson et al., 2008) and dnRAR-α (Rajaii et al., 2008) have been described previously. Animal experiments were approved by the Institutional Animal Care Use Committee of Dartmouth Medical School. All animals were maintained in a pathogen-free facility at Dartmouth Medical School.

C57BL/6 or DR5-Luciferase reporter mice were immunized with PBS1X, CFA (mixed 1:1 with PBS1X; DIFCO), 2.5 µg control IgG, 2.5 µg α-CD3 (clone 2c11), 50 µg LPS (Sigma-Aldrich), and 50 µg α-CD40 (BioXCell) by i.p. (200 µl) or footpad (20 µl) injections (as indicated).

Mice received 3 mg d-Luciferin (Gold Biotechnology) in 200 µl DPBS (Cellgro) by i.p. injection and were imaged in the IVIS 2000 (Xenogen Corp.) according to the manufacturer’s instructions. Analysis and images were obtained using the Living Image Software (version 2.6.1; Caliper Life Sciences).

Thymus, peripheral LNs, spleen, and mesenteric LNs were removed from WT, dnRAR-α, or dnRAR-αCD4^Cre mice. Cell suspensions were prepared, and staining for CD4 and CD8 T cells was performed and analyzed by flow cytometry.

CD4⁺ T cells were isolated from the spleens of dnRAR-α or dnRAR-αCD4^Cre mice and labeled with 5 µM CFSE (Invitrogen). In parallel, T cell–depleted splenocytes (APC) were prepared, and 0.15 × 10⁵ APCs were co-cultured with 0.5 × 10⁵ CD4⁺ T cells in the presence of soluble α-CD3 (clone 2C11). CFSE dilution was evaluated by flow cytometry at day 3.

CD4⁺ T cells were isolated from the spleens of dnRAR-α or dnRAR-αCD4^Cre mice and stimulated with 10 µg/ml plate-bound α-CD3 (clone 2c11) and 1 µg/ml α-CD28 (clone PV-1) plus the addition of 100 U/ml IL-2, 10 ng/ml TGF-β, and different concentrations of RA (Sigma-Aldrich; as indicated). After 4 d in culture, the expression of FoxP3 was evaluated by intracellular staining (eBioscience) and analyzed by flow cytometry.

CD11c⁺ and CD4⁺ T cells were isolated from the spleens of immunized DR5-Luciferase reporter mice. 0.5 × 10⁶ cells were resuspended in 100 µl DPBS, and the procedure was performed according to the manufacturer’s instructions (SteadyLite kit; Roche).

Skin grafting was performed as described previously (Quezada et al., 2005). For tracking RA signaling on CD4⁺ T cells, splenic CD4⁺ T cells were isolated from the DR5-Luciferase reporter mouse, and 3 × 10⁶ cells were transferred into RAG^−/− recipients by i.v. injection. The next day, syngeneic or allogeneic skin grafts were transplanted. For the antagonist treatment regimen, when indicated, recipient mice were treated three times/week with control vehicle (DMSO) or 50 µg/mouse of the RAR pan-antagonist (NRX 194310; NuRx Pharmaceuticals).

Skin transplants were removed from animals at the indicated day in the figure legends. The transplants were incubated in HBSS media with 4 mg/ml DNase, 4 mg/ml Liberase, and 10 mg/ml Collagenase D for 3 h at 37°C (all from Roche). After this, the tissue was disintegrated using a mortar and pestle and filtered through a 100-µm nylon strainer. Cell suspension was stained for flow cytometric analysis.

At day 0, RAG^−/− mice received 10⁶ OT-II–dnRAR-α–OT-II–Ly5.2 or dnRAR-αCD4^Cre–OT-II–Ly5.2 cells by i.v. injection. The next day, recipient mice were grafted with syngeneic (C57BL/6) or allogeneic (actin-OVA) skins. After 7 d, CD4⁺ T lymphocytes were isolated from skin graft dLNs (L3T4 kit; Miltenyi Biotec) and co-cultured at a 3:1 ratio with naive or ISQ-pulsed splenic CD11c⁺ cells (10 µg/ml New England Peptide and CD11c microbeads; Miltenyi Biotec).

Supernatants were collected 72 h after culture and analyzed using ELISA (mouse IL-17A kit; BioLegend) or Luminex (Bio-Rad Laboratories), as indicated. The Luminex assay was performed by the Immune Monitoring Core (Dartmouth College).

The following FITC-, PE-, PerCP-, or APC-conjugated antibodies were used: α-Ly5.2 (A20), α-CD4 (L3T4), α-CD8 (53-6.7), α-CD11c (N418), α-CD62L (Mel-14), α-CD11b (M1/70), α-CCR4 (2G12), α-CCR5 (7A4), α-CCR6 (29-2L17), α-CCR10 (248918), α-CXCR3 (CXCR3-173), and αvβ5 (MR9-4). All antibodies except αvβ5 (BD) and α-CCR10 (R&D Systems) were purchased from BioLegend. Data acquisition was performed on a FACScan (BD), and analysis was performed using FlowJo software (Tree Star).

Differences between the means of experimental groups were analyzed using the Mann-Whitney test. Data analysis was performed using Prism software (version 4.0; GraphPad Software). A p-value ≤0.05 was considered significant.

Fig. S1 shows the T cell frequencies and phenotype from different organs in WT, dnRAR-α, and dnRAR-αCD4^Cre mice. Fig. S2 shows that CD4⁺ T cells from the dnRAR-αCD4^Cre mice do not convert to FoxP3 cells in vitro. Fig. S3 shows that the ablation of RA signaling on CD4⁺ T cells does not affect the expression of chemokine receptors or activation markers. Fig. S4 shows that the lack of RAR-α signaling promotes the generation of Th2/suppressive CD4⁺ T cells. Fig. S5 shows that disruption of RAR-α signaling on CD4⁺ T cells skews T cell polarity toward the Th2 phenotype.

We are very grateful to Dr. Pierre Chambon for insightful discussions on data and manuscript preparation. We also thank the Immune Monitoring and Flow Cytometry Shared Resource at Dartmouth College.

The authors acknowledge the support of the National Institutes of Health (grants AT005382, AI048667, and CA062275), the National Cancer Institute (grant 062275), the Medical Research Council Centre for Transplantation, the Biomedical Research Council, and The Wellcome Trust. E. Nowak is supported by a postdoctoral fellowship from the National Multiple Sclerosis Society.

Author contributions: K. Pino-Lagos designed and performed the experiments, interpreted and analyzed data, and wrote the manuscript. Y. Guo contributed on imaging-reporting experiments and discussion on the manuscript. C. Brown contributed with surgeries, imaging and recall experiments, and discussion on the manuscript. M.P. Alexander contributed with imaging experiments and mice screening. R. Elgueta contributed with BM-DC experiments (response to reviewers). K.A. Bennett performed surgeries and maintained mouse colonies. V. De Vries performed surgeries. E. Nowak standardized RAR pan-antagonist administration. S. Sockanathan provided the dnRAR-α mice and advised on screening and use. R. Blomhoff provided the RARE-Luc mice and advised on screening and use of mice. R.A Chandraratna provided us with the RAR pan-antagonist and guidance on in vivo treatments. E. Dmitrovsky advised on analysis of RAR function and assisted with data interpretation and analysis. R.J. Noelle designed the study, interpreted data, and wrote the manuscript.

The authors declare no competing financial interests.