Aldh1a2 Regulatory Dendritic Cells Expressing IL-4 and Retinoic Acid

The Journal of Immunology

IL-4 and Retinoic Acid Synergistically Induce Regulatory Dendritic Cells Expressing Aldh1a2

Bing Zhu,* Thomas Buttrick,* Ribal Bassil,* Chen Zhu,* Marta Olah,* Chuan Wu,* Sheng Xiao,* William Orent,* Wassim Elyaman,* and Samia J. Khoury*,†

Although activated inflammatory monocytes (IMCs) and inflammatory dendritic cells (IDCs) are potent T cell suppressors, nonactivated IMCs and IDCs promote T cell activation and Th1/Th17 cell differentiation. In this study, we investigated how to reduce the proinflammatory properties of IMCs and IDCs and further convert them into immune regulatory dendritic cells (DCs). We found that IL-4 and retinoic acid (RA) cotreatment of GM-CSF–differentiated IDCs synergistically induced the expression of aldehyde dehydrogenase family 1, subfamily A2, a rate-limiting enzyme for RA synthesis in DCs. IL-4 plus RA–treated IDCs upregulated CD103 expression and markedly reduced the production of proinflammatory cytokines upon activation. IL-4 plus RA–treated IDCs strongly induced CD4+Foxp3+ regulatory T cell differentiation and suppressed Th1 and Th17 differentiation. Mechanistically, the transcription factors Stat6 and RA receptor b play important roles in aldehyde dehydrogenase family 1, subfamily A2, induction. In addition, IL-4 and RA signaling pathways interact closely to enhance the regulatory function of treated DCs. Adoptive transfer of IL-4 plus RA–treated DCs significantly increased regulatory T cell frequency in vivo. Direct treatment with IL-4 and RA also markedly suppressed actively induced experimental autoimmune encephalomyelitis. Our data demonstrate the synergistic effect of IL-4 and RA in inducing a regulatory phenotype in IDCs, providing a potential treatment strategy for autoimmune diseases. The Journal of Immunology, 2013, 191: 3139–3151.

uring an immune response, monocytes play an important 10). We therefore explored how to reduce the proinflammatory role in regulating adaptive immunity and also function as properties of IMCs and IDCs and further convert them into im- immune effector cells (1, 2). In mice, emergency mye- mune regulatory dendritic cells (DCs). D + lopoiesis is represented by the significant increase of CD11b Ly- Regulatory or tolerogenic DCs function to delete or suppress 6ChighLy-6G2 inflammatory monocytes (IMCs) in the blood, spleen, pathogenic T cells (10, 11). A major focus of research has been to and inflammatory sites (3–5). IMCs can be readily differentiated generate regulatory DCs that induce Ag-specific Foxp3+ regula- into inflammatory dendritic cells (IDCs) with GM-CSF treatment tory T cells (Tregs) for the prevention and treatment of autoimmune in vitro, and a similar differentiation process occurs in vivo (6–8). diseases, graft-versus-host disease, and rejection in transplantation Nonactivated IMCs and IDCs are able to present Ags to T cells (12, 13). In the steady state, CD103+ DCs isolated from the intes- and produce IL-1b, IL-6, and IL-12 to promote Th1 and Th17 cell tinal lamina propria and mesenteric lymph nodes efficiently induce differentiation (8, 9). In contrast, signals from activated T cells, Tregs through the production of retinoic acid (RA) and TGF-b (14, including GM-CSF, IFN-g, TNF-a, and CD40L, collectively ac- 15). These DCs have enriched mRNA and protein expression of tivate IMCs and IDCs, which in turn induce T cell death through aldehyde dehydrogenase family 1, subfamily A2 (Aldh1a2), a rate- the production of NO (8). Nonetheless, activated IMCs and IDCs limiting enzyme for RA synthesis in DCs (14, 16). In addition, continue to be a major source of proinflammatory mediators (7, 9, CD103+ DCs play an important role in oral tolerance (17). How- ever, intestinal inflammation may reduce Aldh1a2 and TGF-b expression in CD103+ DCs, and impair their ability to induce Tregs, *Center for Neurologic Diseases, Brigham and Women’s Hospital, Harvard Medical † but favor the emergence of IFN-g–producing T cells (18). In ad- School, Boston, MA 02115; and Abou Haidar Neuroscience Institute, American + University of Beirut, Beirut 1007 2020, Lebanon dition, there is significant loss of CD103 DCs during colonic inflammation (19). Received for publication February 1, 2013. Accepted for publication July 13, 2013. RA has been shown to induce Treg differentiation in the presence This work was supported by National Institutes of Health Grants RO1AI058680 and RO1AI067472 (to S.J.K.) and National Multiple Sclerosis Society Grants RG-4490 of TGF-b and to reciprocally suppress Th17 differentiation (20– (to S.J.K.) and RG-4278 (to B.Z.). 24). RA also sustains the stability of naturally occurring Tregs Address correspondence and reprint requests to Dr. Samia J. Khoury and Dr. Bing (nTregs) in the presence of IL-6 (25). Mechanistically, RA was Zhu, Harvard Medical School, Room 641, New Research Building, 77 Avenue Louis shown to enhance TGF-b–induced Smad3 expression and phos- Pasteur, Boston, MA 02115. E-mail addresses: [email protected] (S.J.K.) and [email protected] (B.Z.) phorylation and to reduce the expression of IL-6Ra, IFN regula- The online version of this article contains supplemental material. tory factor-4, and IL-23R in T cells (26). In vivo, RA treatment suppresses Th1- and Th17-mediated immunopathology in various Abbreviations used in this article: 7-AAD, 7-aminoactinomycin D; Aldh1a2, aldehyde dehydrogenase family 1, subfamily A2; BMDC, bone marrow–derived dendritic autoimmune disease models, including experimental autoimmune cell; ChIP, chromatin immunoprecipitation; CT, threshold cycle; DC, dendritic cell; encephalomyelitis (EAE) (27–29), uveoretinitis (30), myositis (31), EAE, experimental autoimmune encephalomyelitis; IDC, inflammatory DC; IMC, inflammatory monocyte; L-NIL, N6-(1-iminoethyl)-L-lysine; MOG, myelin oligo- colitis (32–34), rheumatoid arthritis (35), and type 1 diabetes (36, dendrocyte glycoprotein; nor-NOHA, Nv-hydroxy-nor-arginine; nTreg, naturally oc- 37). curring regulatory T cell; PT, pertussis toxin; RA, retinoic acid; RAR, RA receptor; Vitamin A (retinol) is converted by alcohol dehydrogenase into RXR, retinoid X receptor; Treg, regulatory T cell. retinal, which is further converted by Aldh1a2 into RA in DCs (38). Copyright Ó 2013 by The American Association of Immunologists, Inc. 0022-1767/13/$16.00 It has been reported that Aldh1a2 mRNA expression in GM-CSF– www.jimmunol.org/cgi/doi/10.4049/jimmunol.1300329 3140 INDUCTION OF REGULATORY DCs EXPRESSING Aldh1a2 cultured splenic or bone marrow–derived DCs (BMDCs) could be T cell differentiation ∼ enhanced by 5- to 10-fold with IL-4 or RA treatment (39, 40). The 2D2 CD4+ T cells were isolated by CD4-positive selection kit, and Because IDCs are present in inflammatory sites at high frequen- CD4+CD62LhighCD442 naive T cells were further purified by cell sorting. cies, and are different from resident DCs (41), we explored how to CD4+ T cells (1 3 105 cells/well) were cultured with the same number of convert proinflammatory IDCs into regulatory DCs through the IDCs and 20 mg/ml MOG35–55 peptide (neutral Th0 condition). For Th17 induction of Aldh1a2. Although separate treatment with IL-4 or differentiation, cells were treated with 20 ng/ml IL-6 and 3 ng/ml TGF-b. For Treg differentiation, cells were treated with 5 ng/ml TGF-b. After RA only weakly induced Aldh1a2 mRNA and enzyme activity in 3 d of culture, cells were collected for intracellular cytokine staining and IDCs, we found that a combination of IL-4 and RA treatment FACS analysis. synergistically increased Aldh1a2 mRNA expression by ∼300- Flow cytometry fold and strongly induced its protein expression as well as enzyme activity. IL-4 plus RA treatment also markedly inhibited produc- For surface staining, isolated cells were blocked with 10 mg/ml mouse Fc tion of proinflammatory cytokines in IDCs. We further character- Block (BD Biosciences) at 4˚C for 5 min and labeled with fluorochrome- conjugated Abs and 7-aminoactinomycin D (7-AAD), including isotype ized the signaling events, revealing the close interaction between controls, for 15 min at 4˚C. After washing, cells were analyzed on the IL-4 and RA signaling pathways. In addition, strong regulation by FACSCalibur (BD Biosciences). Intracellular cytokine and Foxp3 staining IL-4 plus RA–treated DCs, as well as by direct IL-4 plus RA treat- was performed according to the BD Biosciences protocol. Data analysis 2 ment in vivo suggests potential clinical implications for autoim- was performed by FlowJo software and gated on 7-AAD live cells. mune diseases. ALDEFLUOR staining Aldehyde dehydrogenase activity was quantified using ALDEFLUOR kit Materials and Methods (Stemcell Technologies). Briefly, 2 3 105 cells from each sample were Animals and reagents stained with ALDEFLUOR substrate for 30 min at 37˚C. Negative control samples were pretreated with ALDH inhibitor diethylaminobenzaldehyde 2/2 Female C57BL/6 mice and Stat6 mice (B6 background) were obtained for 15 min before substrate staining. After treatment with mouse Fc Block from The Jackson Laboratory. The 2D2 myelin oligodendrocyte glyco- (BD Biosciences), cells were stained with cell surface markers and 7- protein (MOG) TCR transgenic mice and Foxp3-GFP knock-in mice were AAD, and analyzed by flow cytometry. provided by V.K. Kuchroo (Harvard Medical School, Boston, MA), and were then crossed to generate 2D2 Foxp3-GFP mice. Animals were used at Real-time PCR (RT-PCR) 6–10 wk of age. All animal procedures performed in this study were ap- proved by the Institutional Animal Care and Use Committee of Harvard Total RNA was extracted using Absolutely RNA miniprep kit (Agilent Medical School. All-trans retinoic acid, retinol, and retinal were obtained Technologies). cDNA was synthesized using SuperScript III first-strand from Sigma-Aldrich. LE135, N6-(1-iminoethyl)-L-lysine (L-NIL), and synthesis system for RT-PCR (Invitrogen). mRNA expression of Aldh1a2, Nv-hydroxy-nor-arginine (nor-NOHA) were obtained from Cayman Rara, Rarb, Rarg, Il4ra, Il13ra, Il5ra, Arg1, Retnla, Chi3l3,andGapdh was Chemicals. Recombinant cytokines were obtained from R&D Systems. examined using TaqMan gene expression assays (Applied Biosystems). FACS Abs and agonistic anti-CD40 were purchased from BD Biosciences, Triplicate samples were examined in each condition. A comparative threshold eBioscience, or BioLegend. cycle (CT) value was normalized for each sample using the following for- mula: DCT =CT (gene of interest) 2 CT (Gapdh), and the relative expression was then calculated using the Equation 22DDCT. IMC isolation and differentiation B6 mice were immunized with an emulsion of 100 ml PBS and 100 mlCFA Immunoblotting containing 0.5 mg heat-inactivated Mycobacterium tuberculosis (H37Ra; Cells were lysed with radioimmunoprecipitation assay buffer supplemented Difco Laboratories). Each animal also received 200 ng pertussis toxin (PT; with protease and phosphatase inhibitors. Protein concentrations were List Biological Laboratories) i.v. on days 0 and 2 postimmunization. On quantified with a bicinchoninic acid assay kit (Pierce). A total of 20 mg 2 + day 10, splenic Ly-6G cells were purified through depleting Ly-6G cells protein from each sample was loaded into each lane of 10% NuPAGE Bis- with anti–Ly-6G biotin (clone 1A8) and anti-biotin microbeads (Miltenyi Tris gels and then separated by electrophoresis. Proteins were transferred + Biotec). CD11b cells were then purified by anti-CD11b microbeads onto a polyvinylidene difluoride membrane. The blot was blocked in (Miltenyi Biotec). After staining with anti–Ly-6C FITC (clone AL-21), StartingBlock buffer (Thermo Scientific) and then incubated overnight + high 2 CD11b Ly-6C Ly-6G IMCs were purified by FACS sorting. To dif- with the primary Abs. After washing, the blot was incubated with HRP- ferentiate IMCs into IDCs, cells were loaded onto 0.4 mg/ml collagen gel labeled goat anti-mouse or anti-rabbit IgG (Sigma-Aldrich), and developed (BD Biosciences) and cultured with 20 ng/ml GM-CSF for 48 h (8). with an Immobilon chemiluminescent HRP substrate (Millipore). To im- DMEM containing 10% FBS, L-glutamine, 2-ME, sodium pyruvate, non- prove staining specificity, RA receptor (RAR)b immunoblotting was car- essential amino acid, and antibiotics (BioWhittaker) was used for culture. ried out using nuclear protein extracts, which were prepared with NE-PER To harvest differentiated IDCs, the collagen gel was digested with 1 mg/ml nuclear and cytoplasmic extraction reagents (Thermo Scientific). Anti- + collagenase IV (Sigma-Aldrich) for 10 min at 37˚C. CD11c IDCs were mouse Aldh1a2 (N-20) and anti-RARb (C-19) Abs were obtained from further purified with CD11c microbeads (Miltenyi Biotec) or FACS sorting Santa Cruz Biotechnology. Anti-Stat6, anti–phospho-Stat6 (Tyr641), and to .95% in purity. anti-histone H3 Abs were obtained from Cell Signaling Technology. Anti–b-actin Ab was obtained from Sigma-Aldrich. Culture of BMDCs Chromatin immunoprecipitation Bone marrow cells were harvested from femur and tibia of naive B6 mice and cultured with 20 ng/ml GM-CSF for 4 d. To mimic the IDC differ- Forty-million BMDCs were treated with IL-4 and RA for 3, 6, 9, and 12 h. entiation process, Ly-6C+Ly-6G2 cells were purified by MACS beads and Chromatin immunoprecipitation (ChIP) assays were carried out using then further cultured with 20 ng/ml GM-CSF on the collagen gel for 2 d. SimpleChIP enzymatic chromatin immunoprecipitation kit (Cell Signal- CD11c+ DCs were then purified by CD11c microbeads (Miltenyi Biotec) ing). Briefly, cells were fixed with 1% formaldehyde and then digested with or FACS sorting. micrococcal nuclease to induce optimal genomic DNA fragmentation. After sonication, samples were subjected to immunoprecipitation with anti-Stat6 T cell proliferation and cytokine assays (Cell Signaling; D3H4), anti-RARb (Santa Cruz; C-19), or control rabbit IgG. DNA samples were then purified, and SYBR Green RT-PCR was For Ag-specific proliferation, CD4+ T cells were purified from 2D2 MOG performed to quantify the amount of specific DNA sequences. Genomic TCR transgenic mice using CD4-positive selection beads (Miltenyi Bio- DNA sequences 1 kb upstream of the transcription start site of Aldh1a2, tec). CD4+ T cells (1 3 105 cells/well) were cultured with the same Il4ra, and Rarb genes were obtained from Genome Browser (University of number of IDCs and 20 mg/ml MOG35–55 peptide. After 48 h of incu- California, Santa Cruz, Santa Cruz, CA), and primers were designed at site 1 bation, 1 mCi [3H]thymidine was added into each well, and cells were (between 21 ∼ 2500 bp) and site 2 (between 2500 ∼ 21000 bp) using harvested 16 h later for the proliferation assay. Cytokine concentrations PrimerQuest (Integrated DNA Technologies). Primer sequences for in the culture supernatants were examined using the Milliplex cytokine/ Aldh1a2 promoter are as follows: site 1, 59-GGCCGTATGCAAATGTC- chemokine immunoassay kit (Millipore). CTCCT-39 (forward) and 59-TCGCTGTATATAGGCAGGTGTCAAGC-39 The Journal of Immunology 3141

(reverse); site 2, 59-GGTGTGGATGGGAAGAGGAAAGGAAA-39 (for- Aldh1a2 gene expression in IDCs. CD11b+Ly-6Chigh IMCs were ward) and 59-CATCTGCCTTGGGTTGCCTGGATATT-39 (reverse). Primer isolated from CFA/PT-immunized mice and differentiated into sequences for IL4ra promoter are as follows: site 1, 59-TCTGGCACAG- IDCs with GM-CSF treatment for 48 h (8). IDCs were then treated ACGCTGAGTAAACA-39 (forward) and 59-GCCTGATCGCTG GCTTA- TTTGCAT-39 (reverse); site 2, 59-TGGGCAAGGGTCATACAATCCAGT-39 with IL-4 and RA either separately or in combination for 24 h. RT- (forward) and 59-AGCCACCACATCTTGCTGTCTTCT-39 (reverse). Primer PCR shows that whereas separate treatment with IL-4 or RA sequences for Rarb promoter are as follows: site 1, 59-TTGGTGATGTC- significantly induced Aldh1a2 gene expression, combined IL-4 AGACTGGTTGGGT-39 (forward) and 59-AGCAGCTCACTTCCTACC- plus RA treatment had a much stronger synergistic effect (Fig. ACTTCT-39 (reverse); site 2, 59-TCTGTGACTGCTCCAGGTGTTTCT-39 (forward) and 59-TTAGCTGAGCTCGCTGAAGGGATT-39 (reverse). The 1A). Titration of different doses of IL-4 and RA showed that specificity of genomic DNA amplification was confirmed by in silico PCR Aldh1a2 was optimally induced at 20 ng/ml IL-4 in combination (University of California, Santa Cruz), and by single peaks in PCR melting of 1 mM RA (Supplemental Fig. 1A, 1B). Aldh1a2 gene expres- curves. In these time course–based ChIP PCR experiments, positive data sion increased almost linearly during the first 24 h of treatment were consistently obtained from samples harvested 3 h after treatment, and and then reached a plateau (Supplemental Fig. 1C). In addition, were presented in Results. Primers 59-ATGGTTGCCACTGGGGATCT-39 and 59-TGCCAAAGCCTAGGGGAAGA-39 were used in negative control Aldh1a2 protein expression was only detectable in IL-4 plus RA- PCR (42). They amplify a genomic region that is ∼6 kb upstream of tran- treated IDCs (Fig. 1B), along with markedly increased enzyme scription start site of GAPDH, and is not associated with transcription factor activity of aldehyde dehydrogenase (Fig. 1C). binding. ChIP PCR data were analyzed by percent input method. We asked whether other Th2 cytokines and RA precursors would Luciferase assay function similarly as IL-4 and RA. We found that IL-13 and RA also synergistically induced Aldh1a2, whereas IL-5 did not, con- Aldh1a2 promoter reporter plasmid was provided by O. M. Martinez- Estrada (Universitat de Barcelona) and was designed as previously re- sistent with the almost undetectable Il5ra expression in IDCs ported (43). Briefly, Aldh1a2 upstream region (830 bp) was amplified by (Supplemental Fig. 1D, 1E). The weaker effect of IL-13 compared PCR from mouse genomic DNA and cloned into the corresponding sites with IL-4 also correlated with significantly lower expression of upstream of the firefly luciferase construct in the pGL4.10 vector (43). For Il13ra1 than Il4ra (Supplemental Fig. 1D, 1E). Furthermore, RA luciferase assay, RAW264.7 cells (American Type Culture Collection) were transfected with mouse Stat6 and/or Rarb cDNA overexpression precursors retinol and retinal synergistically induced Aldh1a2 plasmids (Thermo Scientific), Aldh1a2 promoter reporter plasmid, and expression when combined with IL-4, and there was no significant pRL-CMV Renilla luciferase reporter plasmid (Promega), using FuGENE difference in effectiveness among RA, retinol, and retinal (Sup- HD transfection reagent (Promega). DNA concentrations were equalized plemental Fig. 1F). among all samples by supplementing with pcDNA3.1 plasmid (Invitrogen). To test the stability of Aldh1a2 expression, IDCs were first After culture for 48 h, luciferase assays were carried out using Dual Lu- ciferase Reporter Assay System (Promega). Data are presented as the ratio treated with IL-4 and RA for 24 h. After several washes to remove of luminescence intensity between firefly and Renilla signals for each ex- IL-4 and RA, we activated these IDCs with IFN-g and agonistic perimental condition. anti-CD40 treatment for another 24 h in the presence of GM-CSF. EAE induction and in vivo treatment Aldh1a2 expression was not significantly altered by activation (Fig. 1D). In contrast, the production of IL-6, IL-12p70, and TNF- Each B6 mouse was immunized with an emulsion made of 75 mg MOG35– a from IL-4 plus RA–treated IDCs was significantly reduced, 55 peptide (MEVGWYRSPFSRVVHLYRNGK; from New England Pep- tide) and CFA. Each animal also received 200 ng PT on days 0 and 2 suggesting that IL-4 plus RA treatment strongly suppressed the postimmunization. For in vivo treatment, carrier-free IL-4 was adminis- proinflammatory properties of IDCs (Fig. 1E). tered s.c. at 100 ng/mouse daily from day 0–15. RA was dissolved in We examined IDC phenotype after 24 h of IL-4 and/or RA DMSO at 20 mg/ml, mixed with Tween 20 and PBS at a volume ratio at treatment. IL-4 treatment increased CD11c expression, but RA 1:2:7, and administered i.p. at 150 mg/mouse daily from day 0–15 (44). mildly reduced CD11c expression. IL-4 plus RA–treated IDCs Vehicle controls were treated with the same volume and composition in the absence of RA. The EAE clinical score was determined as follows: 0, no retained a high level of CD11c expression that is comparable to disease; 0.5, partial tail paralysis; 1, complete tail paralysis; 2, partial hind IL-4–treated IDCs (Fig. 1F). Interestingly, whereas separate IL-4 limb paralysis; 3, complete hind limb paralysis; 4, complete hind limb and or RA treatment did not induce CD103 expression, ∼70% of IL-4 partial front limb paralysis; and 5, moribund or dead animals. plus RA–treated IDCs were positive for CD103 (Fig. 1F). In ad- Histology dition, a majority of IL-4 plus RA–treated IDCs expressed MHC class II, CD80, PD-L1, and PD-L2 (Supplemental Fig. 2). CD86 Animals were sacrificed and perfused with PBS. The entire spinal cord was cut into nine segments, and 20-mm–thick spinal cord cross-sections were expression was increased in IL-4 plus RA–treated IDCs. B7-H3 prepared on a cryostat. The sections were processed for H&E and Luxol and B7-H4 were expressed on very few cells across groups. These fast blue staining following standard protocols. For quantification, in- data suggest that IL-4 plus RA–treated IDCs express CD11c and flammatory foci were identified as focal areas with the aggregation of 20 or molecules critical for Ag presentation and specifically upregulated more cells in H&E-stained sections, and the total numbers of inflammatory foci were counted from nine levels of spinal cord sections. To quantify CD103 expression. myelin damage, the percentage of white matter area with myelin damage was assessed in Luxol fast blue–stained sections, and then averaged from T cell regulation by IL-4 plus RA–treated IDCs nine levels of the sections for each mouse. Tissues from at least three mice To examine whether Aldh1a2 expression in IDCs regulates T cell were included for each group. function, we cultured various IDCs with CD4+ T cells from 2D2 Data analysis MOG TCR transgenic mice, in the presence of 20 mg/ml MOG35– Data in the text represent the mean 6 SEM, and the error bars in figures 55 peptide. Whereas RA-treated IDCs had reduced ability to in- also represent SEM. Unpaired two-tailed t tests were used to analyze the duce T cell proliferation, IL-4 and IL-4 plus RA–treated IDCs statistical difference between two groups, and a one-way ANOVA followed induced similar levels of T cell proliferation as control IDCs (Fig. by Bonferroni test was used to analyze data with more than two groups. A 2A). However, T cells cultured with IL-4 plus RA–treated IDCs , p value 0.05 was considered significant. had a dramatically reduced production of IFN-g and IL-17, whereas IL-2 production was unchanged (Fig. 2B). We further examined Results the function of various IDCs in T cell differentiation. Under neutral Induction of Aldh1a2 in IL-4 plus RA–treated IDCs (Th0) condition, both IL-4–treated and RA-treated IDCs reduced To explore an efficient treatment that could induce IDCs to produce the frequency of IFN-g– and IL-17–producing T cells, but IL-4 RA, we first examined how IL-4 and RA treatment would change plus RA–treated IDCs had the strongest effect (Fig. 2C). Under 3142 INDUCTION OF REGULATORY DCs EXPRESSING Aldh1a2

FIGURE 1. IL-4 and RA cotreatment synergistically induces Aldh1a2 expression in IDCs. (A) IDCs were treated with IL-4 (20 ng/ml) and/or RA (1 mM) for 24 h, and Aldh1a2 mRNA was quantified by RT-PCR. (B) Aldh1a2 protein expression was examined by immunoblotting in nontreated and IL-4 and/or RA–treated IDCs at 24 h. (C) Aldehyde dehydrogenase activity was examined by ALDEFLUOR staining in IDCs treated for 24 h. Data quantifying the median fluorescence intensity are shown in the right graph.(D) IDCs were treated with IL-4 and RA for 24 h. After washing, cells were either nonactivated or activated with IFN-g (20 ng/ml) and agonistic anti-CD40 (20 mg/ml) for 24 h. Aldh1a2 expression was quantified by RT-PCR. (E) Culture supernatants collected from experiments described in (D) were analyzed by Milliplex assay for cytokine production. (F) Phenotype of IDCs was analyzed by flow cytometry after treatment for 24 h. Expression of CD11c and CD103 is shown in solid lines, and isotype control staining is shown in tinted gray areas. Labels over individual columns indicate the statistical significance compared with the first column on left (nontreated group). Labels on the connecting lines show the statistical significance among other columns. Data are representative of two to three independent experiments. *p , 0.05, #p , 0.01. n.s., Not significant.

Th17-polarizing condition, IL-4 plus RA–treated IDCs decreased arginase inhibitor) (Fig. 2D), suggesting that reduction in Th17 IL-17–producing T cells by ∼90% (Fig. 2C). The inhibition of differentiation was dependent on RA production from IL-4 plus Th17 differentiation with IL-4 plus RA–treated IDCs was fully RA–treated IDCs. When IL-4 plus RA–treated IDCs were cultured reversed by the treatment with LE135 (an RAR antagonist), but together with 2D2 T cells and control IDCs, Th17 polarization was not with L-NIL (a NO synthase 2 inhibitor) or nor-NOHA (an also reduced by ∼90%, and this reduction was again fully reversed The Journal of Immunology 3143

FIGURE 2. IL-4 plus RA–treated IDCs inhibit T cell production of IFN-g and IL-17. (A) CD4+ T cells from 2D2 MOG TCR transgenic mice were cultured with nontreated IDCs or IDCs treated with IL-4, RA, or IL-4 plus RA in the presence of MOG35–55 peptide (20 mg/ml). T cell proliferation assay was performed after 48 h. (B) IL-2, IFN-g, and IL-17 concentrations in the culture supernatants of above experiments were measured. (C) Naive 2D2 CD4+ T cells were cultured with various IDCs in neutral (Th0) or Th17-polarizing condition for 72 h. Intracellular staining for IFN-g and IL-17 was examined in gated CD4+ T cells. (D) Intracellular staining of 2D2 CD4+ T cells after cultured with control IDCs and/or IL-4 plus RA–treated IDCs in Th17-polarizing condition. LE135 (RAR antagonist), L-NIL (NO synthase 2 inhibitor), and nor-NOHA (arginase I inhibitor) were used at 1 mM, 1 mM, and 0.5 mM, respectively. In these experiments, T cells were used at 1 3 105/well, and IDCs were used at 5 3 104/well in a round-bottom 96-well plate. Data are representative of three independent experiments. *p , 0.05, #p , 0.01. n.s., Not significant. 3144 INDUCTION OF REGULATORY DCs EXPRESSING Aldh1a2 by LE135 treatment, suggesting that RA production from IL-4 plus from 2D2 Foxp3KI mice and cultured them with control IDCs or RA–treated IDCs dominantly suppresses Th17 differentiation in- IL-4 plus RA–treated IDCs in the presence of MOG35–55 peptide duced by control IDCs (Fig. 2D). FACS analysis of 7-AAD staining and TGF-b for 48 h (Fig. 3D). IL-4 plus RA–treated IDCs induced shows that coculture with IL-4 plus RA–treated IDCs did not in- much stronger expansion of nTregs than control IDCs, and IDCs crease T cell death (data not shown), consistent with the observed themselves did not proliferate. normal T cell proliferation (Fig. 2A). Under Treg-polarizing condition, IL-4 plus RA–treated IDCs Signaling mechanism for Aldh1a2 induction in IDCs consistently showed the strongest effect in inducing CD4+Foxp3+ Because Stat6 activation is a major pathway downstream of IL-4 T cells, which was largely reversed by LE135 treatment (Fig. 3A). and IL-13 signaling (45), we investigated the role of Stat6 in IL-4 This suggests that RA production from IL-4 plus RA–treated IDCs plus RA–treated IDCs in Aldh1a2 induction. Immunoblotting shows plays a critical role in enhanced Treg induction. Interestingly, IL-4 that from 30 min to 2 h after IL-4 and RA treatment, there was plus RA–treated IDCs induced more Foxp3+ T cells than IL-17– strong Stat6 phosphorylation on tyrosine 641, whereas the total producing T cells under the Th17 condition (Fig. 3B), suggesting amount of Stat6 remained unchanged (Fig. 4A). Using IDCs from that IL-4 plus RA–treated IDCs can block IL-6 signaling in T cells. Stat6 knockout mice, we show that Aldh1a2 expression was ab- Because increased Treg differentiation could be due to enhanced rogated in IL-4–treated IDCs, not affected in RA-treated IDCs, conversion from non-Treg cells and/or enhanced expansion of nTregs, and reduced to a level similar to RA-treated IDCs in IL-4 plus RA– we tested these possibilities in our model. CD4+Foxp32 non-Treg treated IDCs (Fig. 4B). These data suggest that Stat6 is necessary cells were purified from 2D2 Foxp3KI mice and cocultured with in IL-4–induced Aldh1a2 expression, but not in RA-induced Aldh1a2 various IDCs in Treg-polarizing condition (Fig. 3C). The data show expression. that IL-4 plus RA–treated IDCs most efficiently increased conver- To test whether Stat6 directly binds to the promoter region of sion from non-Treg cells to Treg cells. To test the function of IL-4 Aldh1a2, we performed chromatin immunoprecipitation (ChIP) plus RA–treated IDCs on Treg expansion, we purified Foxp3+ nTregs PCR assays. BMDCs were used due to the insufficient cell num-

FIGURE 3. IL-4 plus RA–treated IDCs promote Treg differentiation through RA production. (A) CD4+ T cells from 2D2 Foxp3 knock-in mice were cultured with various IDCs in Treg-polarizing condition for 72 h. Foxp3+GFP+ cell frequency was examined in gated CD4+ T cells by ﬂow cytometry. (B) Frequencies of Foxp3+GFP+ and IL-17–producing CD4+ T cells were determined after coculturing with control or IL-4 plus RA–treated IDCs in Th17- polarizing condition. (C) CD4+Foxp32 cells were isolated from 2D2 Foxp3 knock-in mice and cultured with various IDCs under Treg-polarizing condition. Efﬁciency of conversion from Foxp32 to Foxp3+ T cells was examined. (D) CD4+Foxp3+ nTregs were isolated from 2D2 Foxp3 knock-in mice, and cultured with control or IL-4 plus RA–treated IDCs under Treg-polarizing condition. Proliferation assay was performed after 48 h. In these experiments, T cells were used at 1 3 105/well, and IDCs were used at 5 3 104/well in a round-bottom 96-well plate. Data are representative of three independent experiments. #p , 0.01. The Journal of Immunology 3145

FIGURE 4. Role of Stat6 activation in the induction of Aldh1a2 expression. (A) IDC protein samples were collected after IL-4 plus RA treatment for 0–2 h. Stat6 phosphorylated at Tyr641 and total Stat6 were quantified by immunoblotting. (B) IDCs from wild-type or Stat62/2 mice were treated with IL-4, RA, or IL-4 plus RA for 24 h. Aldh1a2 expression in treated IDCs was compared with nontreated IDCs by RT-PCR. (C) BMDCs derived from wild-type B6 mice were treated with IL-4 plus RA for 3 h. ChIP analysis was performed with control IgG or anti-Stat6 Ab, and PCR was performed with primers specific for two sites within 1 kb upstream of Aldh1a2.(D) ChIP PCR was performed, as above, with BMDCs derived from Stat62/2 mice. Data are representative of two independent experiments. #p , 0.01. n.s., Not significant. bers of IDCs, and we found that IL-4 plus RA treatment also 6B), whereas no binding was seen in Stat62/2 BMDCs (Fig. 6C). synergistically induced Aldh1a2 expression in BMDCs (Supple- Furthermore, RA treatment was found to induce Il4ra gene ex- mental Fig. 3A). ChIP PCR shows that Stat6 directly binds to the pression (Fig. 6D). In support of this, ChIP PCR data show that promoter region of Aldh1a2 in wild-type BMDCs (Fig. 4C), whereas RARb binds to the Il4ra promoter (Fig. 6E). Thus, IL-4 enhances there was no binding in Stat62/2 BMDCs (Fig. 4D). Additional RA signaling by inducing RARb expression, and RA enhances IL- negative control PCR assays were performed with primers specific 4 signaling by inducing IL4ra expression. for a chromosome DNA region not associated with transcription To directly test the role of Stat6 and RARb in Aldh1a2 ex- factor binding (42). The amplification was the same between control pression, we performed a luciferase assay (Fig. 7). RAW264.7 cells Ig and anti-Stat6 immunoprecipitated DNA from Stat6+/+ BMDCs were transfected with plasmid vectors expressing Stat6, RARb, (data not shown), suggesting that the changes in Aldh1a2 ChIP PCR or both. A significant increase in Aldh1a2 promoter activity was are specific. detected only in the cells overexpressing both Stat6 and RARb All-trans RA specifically binds to RAR, which form hetero- (Fig. 7A). In addition, when Stat6 and RARb-transfected RAW264.7 dimers with retinoid X receptor (RXR) to modulate gene tran- cells were activated by IL-4 and RA treatment, more robust ac- scription (38, 46). We examined the expression of RAR subtypes tivation of the Aldh1a2 promoter was observed (Fig. 7B), sug- in IDCs using RT-PCR (Fig. 5A). Whereas IL-4 plus RA treatment gesting that activation of transcription factors Stat6 and RARb synergistically induced Rarb expression, Rara and Rarg expres- by IL-4 and RA treatment further enhances the Aldh1a2 promoter sion was significantly reduced (Fig. 5A). A time course study activity. shows that Rarb induction mostly occurred within the first 3 h of IL-4 and RA treatment (Fig. 5B). To examine RARb protein ex- In vivo IL-4 and RA treatment strategies pression, we harvested nuclear protein extracts to enhance the Ab To study the function of IL-4 plus RA–treated DCs in vivo, we specificity in immunoblotting. The data show that IL-4 plus RA– treated BMDCs with IL-4 and RA for 48 h, and then pulsed them treated IDCs expressed the highest levels of RARb (Fig. 5C). with MOG35–55 peptide for 4 h before transferring to naive 2D2 Because IL-4 plus RA treatment also synergistically induced Rarb mice at 107 cells/mouse on days 0 and 4. On day 7, significantly expression in BMDCs (Supplemental Fig. 3B), we used these cells increased frequencies of CD4+Foxp3+ Tregs were observed in the in RARb ChIP PCR assays. The data show that RARb binds to the spleen and blood of IL-4 plus RA–treated DC recipients, but not in promoter regions of both Rarb and Aldh1a2 (Fig. 5D, 5E), sug- control DC recipients (Fig. 8A). The absolute numbers of CD4+ gesting that RARb not only functions as positive feedback in Foxp3+ Tregs were also significantly increased in the spleens of inducing its own expression, but also directly drives Aldh1a2 IL-4 plus RA–treated DC recipients (3.81 6 0.13 million in IL-4 expression. Negative control PCR again shows the same amplifi- plus RA–treated DC recipients versus 1.43 6 0.41 million in cation for control Ig and anti-RARb immunoprecipitated DNA nontreated mice and 1.44 6 0.1 million in control DC recipients, samples (data not shown). p , 0.05). Whereas transfer of control or IL-4 plus RA–treated In addition, we found that IL-4 and RA signaling pathways DCs induced few IL-17–producing CD4 T cells in the spleen, closely interact and mutually enhance each other (Fig. 6). Syn- IFN-g–producing T cells were mildly induced in both groups, with ergistic induction of Rarb by IL-4 and RA treatment was lost in no significant difference between them (Fig. 8B). Stat62/2 IDCs (Fig. 6A). ChIP PCR shows that Stat6 binds to Both IL-4 and RA treatments have been reported to be protec- Rarb promoter in IL-4 plus RA–treated Stat6+/+ BMDCs (Fig. tive in EAE (27, 29, 47). Based on our results in IDCs, we asked 3146 INDUCTION OF REGULATORY DCs EXPRESSING Aldh1a2

FIGURE 5. Expression and role of RARb in IL-4 plus RA–treated IDCs. (A) Expression of Rara, Rarb, and Rarg in control IDCs and those treated with IL-4, RA, or IL-4 plus RA for 24 h was examined by RT-PCR. (B) Time course of Rarb expression in IL-4 plus RA–treated IDCs was determined by RT- PCR. (C) Expression of RARb was measured in nuclear protein extracts from control and treated IDCs at 24 h by immunoblotting. (D) BMDCs were treated with IL-4 plus RA for 3 h. ChIP PCR was performed with control IgG or anti-RARb Ab, and PCR primers were specific for two sites within 1 kb upstream of Rarb.(E) ChIP experiment was the same as in (D), with PCR primers specific for two sites within 1 kb upstream of Aldh1a2. Data are representative of two to three independent experiments. *p , 0.05, #p , 0.01. n.s., Not significant. whether IL-4 and RA cotreatment might offer even better EAE In summary (Fig. 10), our data show that IL-4 and RA treatment protection. We treated MOG35–55-immunized mice with IL-4, synergistically induces the expression of Aldh1a2 in IDCs, and RA, or IL-4 plus RA from day 0 to day 15 after immunization treated IDCs promote Treg differentiation while suppressing Th1 (Fig. 9). Mean maximal disease score was significantly reduced in and Th17 cell differentiation through induced RA production. the IL-4 and RA treatment groups, and it was further significantly Mechanistically, Stat6 and RARb are important transcription fac- reduced in IL-4 plus RA–treated group (Fig. 9A, Supplemental tors in driving Aldh1a2 expression. IL-4 and RA signaling pathways Table I). Histology on spinal cord tissues harvested on day 20 interact closely, with Stat6 inducing Rarb expression and RARb shows that cotreated mice had the lowest number of inflammatory inducing Il4ra expression. RARb also positively induces its own foci and the lowest percentage of white matter area with myelin expression. In vivo, adoptive transfer of IL-4 plus RA–treated IDCs damage (Fig. 9B). Splenocytes were harvested on day 20 and increased Treg frequency in an Ag-specific fashion, and direct treat- stimulated with 20 mg/ml MOG35–55 for 48 h, and the cytokine ment with IL-4 and RA also suppressed actively induced EAE. production was examined by Luminex assay (Fig. 9C). IL-17 was significantly reduced in all three treated groups, but IFN-g, Discussion as well as IL-6 and GM-CSF (data not shown) were only signifi- We and others have reported that IMCs are present at high fre- cantlyreducedinIL-4plusRA–treated group. The production quencies in inflammatory tissues and readily differentiate into IDCs of these proinflammatory cytokines was significantly lower in upon GM-CSF signaling (4, 8, 9). Nonactivated IDCs efficiently IL-4 plus RA–treated group than in separately treated groups. present Ags to T cells and promote Th1/Th17 cell differentiation In addition, splenic CD11b+Ly-6Chigh cells isolated from IL-4 (8, 9). In contrast, CD103+ DCs in the gut express Aldh1a2 and plus RA–treated group on day 20 had the strongest increase in the produce RA to regulate T cell differentiation (14, 15). They play expression of Aldh1a2 and Rarb (Fig. 9D). However, Foxp3+ Treg a critical role in oral tolerance by inducing Tregs (48). In this frequency in the spleen was not significantly different among study, we discovered the synergistic role of IL-4 and RA treatment these groups on day 9 postimmunization, probably due to the on the induction of Aldh1a2 expression in IDCs at the mRNA, polyclonal T cell repertoire in these mice or due to migration of protein, and enzyme activity levels. Yokota et al. (40) reported that Ag-specific Tregs to the immunization sites or CNS (data not GM-CSF–differentiated BMDCs and splenic DCs had increased shown). Aldh1a2 expression compared with Flt-3 ligand–differentiated The Journal of Immunology 3147

FIGURE 6. Interaction of IL-4 and RA signaling pathways. (A) IDCs were derived from wild-type and Stat62/2 mice, and treated with IL-4, RA, or IL-4 plus RA for 24 h. Rarb expression was compared with nontreated IDCs by RT-PCR. (B) BMDCs from wild-type mice were treated with IL-4 plus RA for 3 h. ChIP PCR was performed with control IgG and anti-Stat6 Ab, and PCR primers were specific for the two sites within 1 kb upstream of Rarb.(C) ChIP experiment was the same as in (B), with BMDCs derived from Stat62/2 mice. (D) IDCs from wild-type mice were treated with RA for 24 h. Il4ra expression was compared with nontreated IDCs by RT-PCR. (E) BMDCs from wild-type mice were treated with IL-4 plus RA for 3 h. ChIP PCR was performed with control IgG and anti-RARb Ab, and PCR primers were specific for two sites within 1 kb upstream of Il4ra. Data are representative of two to three independent experiments. *p , 0.05, #p , 0.01. n.s., Not significant.

DCs, and IL-4 treatment further increased Aldh1a2 mRNA ex- Aldh1a2 promoter. Another possibility is that the formation of a pression by ∼6- to 7-folds. Similar levels of Aldh1a2 induction more complex transcription activation machinery at later time were reported in IL-4–treated mesenteric lymph node DCs (49). In points may prevent the Ab from recognizing the specific epitope addition, RA treatment has also been shown to increase Aldh1a2 on Stat6. However, these negative data serve as additional speci- mRNA expression by 5- to 10-folds in BMDCs (39). In IDCs, we ficity controls, in addition to the results from Stat62/2 control and found that Aldh1a2 mRNA was weakly induced by separate IL-4 negative control PCR. or RA treatment, but the protein expression or enzymatic activity All-trans RA binds to RARa, RARb, and RARg, which form of aldehyde dehydrogenase was not induced. In contrast, the com- heterodimers with RXRa, RXRb, and RXRg (46). RA binding bined IL-4 and RA treatment in IDCs increased Aldh1a2 mRNA induces a conformational change of the heterodimers, resulting expression by ∼300-folds, and markedly induced Aldh1a2 protein in the dissociation of corepressor complexes and the recruitment expression and enzymatic activity. of coactivator complexes, and subsequent changes in target gene Our data suggest that Stat6 is necessary for IL-4–induced expression (46). We found that Rarb mRNA expression was Aldh1a2 expression. It is well known that Stat6 is a major IL-4 synergistically induced by IL-4 and RA treatment, in parallel with signaling molecule and may bind to specific DNA motifs in the the induction of Aldh1a2. Rarb induction reached peak level at 3 h promoters of target genes (45). Bioinformatics analysis with Ex- after treatment, earlier than the peak Aldh1a2 induction (24 h). It plain 3.0 (Biobase) suggests that multiple Stat6 binding sites may has been reported that Rarb promoter contains a RA response exist within 1 kb upstream of Aldh1a2 transcription start site (data element, which may bind RARb (50, 51). In tumor cells, RA not shown). ChIP PCR assay clearly shows the binding of Stat6 treatment induces the expression of RARb, which functions as to Aldh1a2 promoter region 3 h after treatment. Interestingly, no a tumor suppressor gene (52). Our ChIP analysis showed that positive binding was detected when ChIP PCR assay was per- RARb was able to bind directly to its own gene promoter, pro- formed at 6, 9, and 12 h. It is possible that 3 h posttreatment is the viding further evidence for the positive feedback gene induction. most active time point when phosphorylated Stat6 binds to the Also using ChIP PCR, RARb was shown to bind to the promoter

FIGURE 7. Role of Stat6 and RARb in activating Aldh1a2 promoter. (A)RAW264.7 cells were transfected with Stat6 and/or Rarb expression plasmids, Aldh1a2 promoter reporter plasmid, and Renilla luciferase plasmid. Aldh1a2 promoter activity was examined by luciferase assay after 48 h. (B) The same luciferase assay was performed as above, and RAW264.7 cells were treated with IL-4 plus RA after plasmid transfection. Data are representative of three independent experiments. #p , 0.01. n.s., Not signiﬁcant. 3148 INDUCTION OF REGULATORY DCs EXPRESSING Aldh1a2

FIGURE 8. Transfer of IL-4 plus RA–treated BMDCs increases Treg frequency in vivo. (A) Control DCs and IL-4 plus RA–treated DCs were pulsed with MOG35–55 peptide (20 mg/ml) for 4 h and transferred into naive 2D2 mice at 107 cell/mouse on days 0 and 4. On day 7, splenocytes and blood leukocytes were harvested, and Foxp3+ cell frequency in CD4+ T cells was determined by flow cytometry. Quantification data are shown in graphs on the right.(B) IFN-g and IL-17 production from splenic CD4+ T cells was examined by intracellular cytokine staining and flow cytometry. Quantification data are shown in the graph on the right. Data are representative of two independent experiments with three mice per group in each experiment. *p , 0.05, #p , 0.01. n.s., Not significant.

region of Aldh1a2, explaining the synergistic effect with IL-4– portant markers for M2 cells, such as arginase I, Ym-1, resistin-like a, induced Stat6 signaling. In support of this, analysis with Explain and insulin-like growth factor (data not shown), suggesting that 3.0 software shows two potential RAR binding sites within 1 kb IL-4 plus RA treatment could be an effective approach to polarize upstream of Aldh1a2 transcription start site (data not shown). type 2 myeloid cells in general (55, 56). Although characterization Luciferase assay further confirms the synergistic effects of Stat6 of the mechanisms behind these findings is out of the scope of this and RARb in the activation of Aldh1a2 promoter. RARb is known study, we found that IL-4 and RA signaling pathways closely in- to play a unique role in RA treatment for cancer and skin diseases teract. IL-4 synergistically induced Rarb expression with RA, and (53) and in RA-induced neuronal differentiation (54). However, RA treatment induced the expression of Il4ra expression. Close our data do not exclude the possible role of RARa and RARg in interactions between two signaling pathways may help explain Aldh1a2 induction. The relative contribution of these RAR subtypes the strong effect of IL-4 and RA treatment in inducing the alter- will need to be studied in specific strains of gene knockout mice. native activation of IDCs and the strong and stable induction of Colonic inflammation was shown to reduce Aldh1a2 expression Aldh1a2 expression. in CD103+ DCs (18). Therefore, the stability of Aldh1a2 expression Along with the strong induction of Aldh1a2, IL-4 plus RA– appears to be a critical factor in generating functional immuno- treated IDCs have a significantly enhanced ability in promoting regulatory DCs. We found that the expression of Aldh1a2 was Treg and suppressing Th1/Th17 differentiation, and this was de- sustained in IL-4 plus RA–treated IDCs even after strong activa- pendent on RA production. Even under strong Th17-polarizing tion with a combination of IFN-g and agonistic anti-CD40 for condition, IL-4 plus RA–treated IDCs still induced significantly 24 h. In addition, the production of proinflammatory cytokines such more Treg cells than IL-17–producing T cells. Furthermore, IL-4 as IL-6, IL-12p70, and TNF-a was markedly reduced in IL-4 plus plus RA–treated IDCs dominantly suppressed Th17 differentiation RA–treated IDCs after activation. We found that IL-4 and RA induced by control IDCs. Although similar effects were observed treatment in IDCs also synergistically induced expression of im- with direct RA treatment during T cell differentiation (20–24), our The Journal of Immunology 3149

FIGURE 9. Cotreatment with IL-4 and RA effectively suppresses EAE. (A) B6 mice were immunized with MOG35–55 on day 0 to induce EAE, and received daily treatment from day 0 to day 15. Five groups were set up, as follows: nontreated, vehicle treated, IL-4 treated, RA treated, and IL-4 plus RA treated. Data were pooled from two independent experiments showing similar results. (B) Spinal cord tissues were harvested on day 20 from three mice per group, and processed for H&E and Luxol fast blue staining. The number of inflammatory foci and the percentage of white matter area with demyelination were quantified. (C) Splenocytes were harvested on day 20 from three mice per group, and activated with 20 mg/ml MOG35–55 for 48 h in vitro. Cytokine production in the supernatant was examined with Milliplex assays. (D) Splenic CD11b+Ly-6Chigh cells were isolated on day 20, and mRNAs of Aldh1a2 and Rarb were quantified by RT-PCR. *p , 0.05, #p , 0.01. data demonstrate the role of IL-4 plus RA treatment in converting pulsing with MOG35–55, IL-4 plus RA–treated BMDCs signifi- otherwise proinflammatory DCs to immune regulatory DCs. cantly increased Treg frequency when adoptively transferred into In this study, we show that combination treatment with IL-4 and MOG TCR transgenic mice. Due to the cell number limitation, RA induced CD103 expression on the majority of IDCs, which we were not able to observe consistent EAE suppression by adop- share the strong Aldh1a2 expression as the gut CD103+ DCs. tive transfer of IL-4 plus RA–treated DCs, which may need to Although RA and GM-CSF have been suggested to induce CD103 be performed at multiple time points during EAE priming. Al- expression on GALT DCs, our data suggest that a Th2 cytokine though we focused on DCs in this study, it is possible that IL-4 environment may synergistically induce the expression of CD103 plus RA treatment may induce Aldh1a2 expression in other cells as well as Aldh1a2. In contrast to the typical tolerogenic DCs that as well, such as macrophages or mesenchymal stem cells, which are phenotypically immature (57), IL-4 plus RA–treated IDCs are being considered for cell therapy in autoimmune diseases (58, express high levels of CD11c, MHC-II, and other costimulatory 59). molecules. This is consistent with their efficient Ag-presenting Our data show that direct treatment of IL-4 and RA significantly function in vitro. In addition, in vivo IL-4 plus RA treatment in suppressed actively induced EAE disease. This is manifested by EAE model did not suppress MOG35–55-induced T cell prolif- reduced CNS inflammation and reduced production of a number eration in the recall (data not shown). of proinflammatory cytokines in the recall assay. In addition, RA treatment in vivo has shown strong effects in suppressing Aldh1a2 and Rarb mRNA expression in the splenic Ly-6Chigh autoimmune diseases in various animal models (27–37). An al- myeloid cells was significantly increased. In addition to their effects ternative strategy is to induce RA production from endogenous on IMCs and IDCs, IL-4 plus RA treatment may directly target cells, especially in cells actively migrating to the inflammatory T cells in vivo, as well as modulating the immune function of sites, such as IMCs and IDCs. We show in this study that after resident DCs, macrophages, and microglia. Therefore, the combined IL-4 and RA treatment may represent a novel treatment strategy for Th1- and Th17-mediated autoimmune diseases. Re- cently, IL-15 was shown to convert RA from an immunoregulatory to a proinflammatory function on DCs (60). It remains to examine whether IL-4 cotreatment could reverse such an effect. In summary, our findings could have important clinical implications in treating autoimmune diseases and inducing immune tolerance in transplantation settings.

Acknowledgments We thank Dr. Francisco Quintana and Dr. Linda Ottoboni for helpful discussion.

FIGURE 10. Illustration of IL-4 and RA signaling pathways leading to Aldh1a2 induction. Increased RA production from IDCs promotes Treg Disclosures differentiation and suppresses Th1 and Th17 differentiation. The authors have no ﬁnancial conﬂicts of interest. 3150 INDUCTION OF REGULATORY DCs EXPRESSING Aldh1a2

References 28. Massacesi, L., E. Castigli, M. Vergelli, J. Olivotto, A. L. Abbamondi, F. Sarlo, and L. Amaducci. 1991. Immunosuppressive activity of 13-cis-retinoic acid and 1. Geissmann, F., C. Auffray, R. Palframan, C. Wirrig, A. Ciocca, L. Campisi, prevention of experimental autoimmune encephalomyelitis in rats. J. Clin. In- E. Narni-Mancinelli, and G. Lauvau. 2008. Blood monocytes: distinct subsets, vest. 88: 1331–1337. how they relate to dendritic cells, and their possible roles in the regulation of T- 29. Racke, M. K., D. Burnett, S. H. Pak, P. S. Albert, B. Cannella, C. S. Raine, cell responses. Immunol. Cell Biol. 86: 398–408. D. E. McFarlin, and D. E. Scott. 1995. Retinoid treatment of experimental al- 2. Shi, C., and E. G. Pamer. 2011. Monocyte recruitment during infection and lergic encephalomyelitis: IL-4 production correlates with improved disease inflammation. Nat. Rev. Immunol. 11: 762–774. course. J. Immunol. 154: 450–458. 3. Sunderko¨tter, C., T. Nikolic, M. J. Dillon, N. Van Rooijen, M. Stehling, 30. Keino, H., T. Watanabe, Y. Sato, and A. A. Okada. 2010. Anti-inflammatory D. A. Drevets, and P. J. Leenen. 2004. Subpopulations of mouse blood mono- effect of retinoic acid on experimental autoimmune uveoretinitis. Br.J.Oph- cytes differ in maturation stage and inflammatory response. J. Immunol. 172: thalmol. 94: 802–807. 4410–4417. 31. Ohyanagi, N., M. Ishido, F. Suzuki, K. Kaneko, T. Kubota, N. Miyasaka, and 4. Zhu, B., Y. Bando, S. Xiao, K. Yang, A. C. Anderson, V. K. Kuchroo, and T. Nanki. 2009. Retinoid ameliorates experimental autoimmune myositis, with S. J. Khoury. 2007. CD11b+Ly-6C(hi) suppressive monocytes in experimental modulation of Th cell differentiation and antibody production in vivo. Arthritis autoimmune encephalomyelitis. J. Immunol. 179: 5228–5237. Rheum. 60: 3118–3127. 5. Geissmann, F., S. Jung, and D. R. Littman. 2003. Blood monocytes consist of 32. Bai, A., N. Lu, Y. Guo, Z. Liu, J. Chen, and Z. Peng. 2009. All-trans retinoic two principal subsets with distinct migratory properties. Immunity 19: 71–82. acid down-regulates inflammatory responses by shifting the Treg/Th17 6. Campbell, I. K., A. van Nieuwenhuijze, E. Segura, K. O’Donnell, E. Coghill, profile in human ulcerative and murine colitis. J. Leukoc. Biol. 86: 959– M. Hommel, S. Gerondakis, J. A. Villadangos, and I. P. Wicks. 2011. Differ- 969. entiation of inflammatory dendritic cells is mediated by NF-kB1-dependent GM- 33. Collins, C. B., C. M. Aherne, D. Kominsky, E. N. McNamee, M. D. Lebsack, CSF production in CD4 T cells. J. Immunol. 186: 5468–5477. H. Eltzschig, P. Jedlicka, and J. Rivera-Nieves. 2011. Retinoic acid attenuates 7. Rivollier, A., J. He, A. Kole, V. Valatas, and B. L. Kelsall. 2012. Inflammation switches the differentiation program of Ly6Chi monocytes from antiinflammatory ileitis by restoring the balance between T-helper 17 and T regulatory cells. macrophages to inflammatory dendritic cells in the colon. J. Exp. Med. 209: 139– Gastroenterology 141: 1821–1831. 155. 34. Kang, S. G., C. Wang, S. Matsumoto, and C. H. Kim. 2009. High and low vi- 8. Zhu, B., J. K. Kennedy, Y. Wang, C. Sandoval-Garcia, L. Cao, S. Xiao, C. Wu, tamin A therapies induce distinct FoxP3+ T-cell subsets and effectively control W. Elyaman, and S. J. Khoury. 2011. Plasticity of Ly-6C(hi) myeloid cells in intestinal inflammation. Gastroenterology 137: 1391-1402.e1-6. T cell regulation. J. Immunol. 187: 2418–2432. 35. Kwok, S. K., M. K. Park, M. L. Cho, H. J. Oh, E. M. Park, D. G. Lee, J. Lee, 9. King, I. L., T. L. Dickendesher, and B. M. Segal. 2009. Circulating Ly-6C+ H. Y. Kim, and S. H. Park. 2012. Retinoic acid attenuates rheumatoid inflam- myeloid precursors migrate to the CNS and play a pathogenic role during au- mation in mice. J. Immunol. 189: 1062–1071. toimmune demyelinating disease. Blood 113: 3190–3197. 36. Stosic´-Grujicic´, S., T. Cvjeticánin, and I. Stojanovic´. 2009. Retinoids differen- 10. Steinman, R. M., and J. Banchereau. 2007. Taking dendritic cells into medicine. tially regulate the progression of autoimmune diabetes in three preclinical Nature 449: 419–426. models in mice. Mol. Immunol. 47: 79–86. 11. Thomson, A. W. 2010. Tolerogenic dendritic cells: all present and correct? Am. 37. Van, Y. H., W. H. Lee, S. Ortiz, M. H. Lee, H. J. Qin, and C. P. Liu. 2009. All- J. Transplant. 10: 214–219. trans retinoic acid inhibits type 1 diabetes by T regulatory (Treg)-dependent 12. Pulendran, B., H. Tang, and S. Manicassamy. 2010. Programming dendritic cells suppression of interferon-gamma-producing T-cells without affecting Th17 to induce T(H)2 and tolerogenic responses. Nat. Immunol. 11: 647–655. cells. Diabetes 58: 146–155. 13. Belkaid, Y., and G. Oldenhove. 2008. Tuning microenvironments: induction of 38. Hall, J. A., J. R. Grainger, S. P. Spencer, and Y. Belkaid. 2011. The role of regulatory T cells by dendritic cells. Immunity 29: 362–371. retinoic acid in tolerance and immunity. Immunity 35: 13–22. 14. Coombes, J. L., K. R. Siddiqui, C. V. Arancibia-Ca´rcamo, J. Hall, C. M. Sun, 39. Feng, T., Y. Cong, H. Qin, E. N. Benveniste, and C. O. Elson. 2010. Generation Y. Belkaid, and F. Powrie. 2007. A functionally specialized population of mu- of mucosal dendritic cells from bone marrow reveals a critical role of retinoic cosal CD103+ DCs induces Foxp3+ regulatory T cells via a TGF-beta and ret- acid. J. Immunol. 185: 5915–5925. inoic acid-dependent mechanism. J. Exp. Med. 204: 1757–1764. 40. Yokota, A., H. Takeuchi, N. Maeda, Y. Ohoka, C. Kato, S. Y. Song, and 15. Sun, C. M., J. A. Hall, R. B. Blank, N. Bouladoux, M. Oukka, J. R. Mora, and M. Iwata. 2009. GM-CSF and IL-4 synergistically trigger dendritic cells to ac- Y. Belkaid. 2007. Small intestine lamina propria dendritic cells promote de novo quire retinoic acid-producing capacity. Int. Immunol. 21: 361–377. generation of Foxp3 T reg cells via retinoic acid. J. Exp. Med. 204: 1775–1785. 41. Geissmann, F., M. G. Manz, S. Jung, M. H. Sieweke, M. Merad, and K. Ley. 16. Ahrendt, M., S. I. Hammerschmidt, O. Pabst, R. Pabst, and U. Bode. 2008. 2010. Development of monocytes, macrophages, and dendritic cells. Science Stromal cells confer lymph node-specific properties by shaping a unique mi- 327: 656–661. croenvironment influencing local immune responses. J. Immunol. 181: 1898– 42. Wu, X., Y. Yoo, N. N. Okuhama, P. W. Tucker, G. Liu, and J. L. Guan. 2006. 1907. Regulation of RNA-polymerase-II-dependent transcription by N-WASP and its 17. Peron, J. P., A. P. de Oliveira, and L. V. Rizzo. 2009. It takes guts for tolerance: nuclear-binding partners. Nat. Cell Biol. 8: 756–763. the phenomenon of oral tolerance and the regulation of autoimmune response. 43. Guadix, J. A., A. Ruiz-Villalba, L. Lettice, V. Velecela, R. Munõz-Cha´puli, Autoimmun. Rev. 9: 1–4. N. D. Hastie, J. M. Pe´rez-Pomares, and O. M. Martıńez-Estrada. 2011. Wt1 18. Laffont, S., K. R. Siddiqui, and F. Powrie. 2010. Intestinal inflammation abro- controls retinoic acid signalling in embryonic epicardium through transcriptional gates the tolerogenic properties of MLN CD103+ dendritic cells. Eur. J. activation of Raldh2. Development 138: 1093–1097. Immunol. 40: 1877–1883. 44. Tzimas, G., R. Thiel, I. Chahoud, and H. Nau. 1997. The area under the 19. Strauch, U. G., N. Grunwald, F. Obermeier, S. Gu¨rster, and H. C. Rath. 2010. concentration-time curve of all-trans-retinoic acid is the most suitable pharma- Loss of CD103+ intestinal dendritic cells during colonic inflammation. World J. cokinetic correlate to the embryotoxicity of this retinoid in the rat. Toxicol. Appl. Gastroenterol. 16: 21–29. Pharmacol. 143: 436–444. 20. Benson, M. J., K. Pino-Lagos, M. Rosemblatt, and R. J. Noelle. 2007. All-trans 45. Goenka, S., and M. H. Kaplan. 2011. Transcriptional regulation by STAT6. retinoic acid mediates enhanced T reg cell growth, differentiation, and gut Immunol. Res. 50: 87–96. homing in the face of high levels of co-stimulation. J. Exp. Med. 204: 1765– 46. Theodosiou, M., V. Laudet, and M. Schubert. 2010. From carrot to clinic: an 1774. overview of the retinoic acid signaling pathway. Cell. Mol. Life Sci. 67: 1423– 21. Elias, K. M., A. Laurence, T. S. Davidson, G. Stephens, Y. Kanno, 1445. E. M. Shevach, and J. J. O’Shea. 2008. Retinoic acid inhibits Th17 polarization 47.Racke,M.K.,A.Bonomo,D.E.Scott,B. Cannella, A. Levine, C. S. Raine, and enhances FoxP3 expression through a Stat-3/Stat-5 independent signaling E. M. Shevach, and M. Ro¨cken. 1994. Cytokine-induced immune deviation as pathway. Blood 111: 1013–1020. a therapy for inflammatory autoimmune disease. J. Exp. Med. 180: 1961– 22. Kang, S. G., H. W. Lim, O. M. Andrisani, H. E. Broxmeyer, and C. H. Kim. 1966. 2007. Vitamin A metabolites induce gut-homing FoxP3+ regulatory T cells. J. 48. Scott, C. L., A. M. Aumeunier, and A. M. Mowat. 2011. Intestinal CD103+ Immunol. 179: 3724–3733. dendritic cells: master regulators of tolerance? Trends Immunol. 32: 412– 23. Mucida, D., Y. Park, G. Kim, O. Turovskaya, I. Scott, M. Kronenberg, and 419. H. Cheroutre. 2007. Reciprocal TH17 and regulatory T cell differentiation me- 49.Elgueta,R.,F.E.Sepulveda,F.Vilches,L.Vargas,J.R.Mora,M.R.Bono, diated by retinoic acid. Science 317: 256–260. and M. Rosemblatt. 2008. Imprinting of CCR9 on CD4 T cells requires IL-4 24. Schambach, F., M. Schupp, M. A. Lazar, and S. L. Reiner. 2007. Activation of signaling on mesenteric lymph node dendritic cells. J. Immunol. 180: 6501– retinoic acid receptor-alpha favours regulatory T cell induction at the expense 6507. of IL-17-secreting T helper cell differentiation. Eur. J. Immunol. 37: 2396– 50. de The´, H., M. M. Vivanco-Ruiz, P. Tiollais, H. Stunnenberg, and A. Dejean. 2399. 1990. Identification of a retinoic acid responsive element in the retinoic acid 25. Zhou, X., N. Kong, J. Wang, H. Fan, H. Zou, D. Horwitz, D. Brand, Z. Liu, and receptor beta gene. Nature 343: 177–180. S. G. Zheng. 2010. Cutting edge: all-trans retinoic acid sustains the stability and 51. Hoffmann, B., J. M. Lehmann, X. K. Zhang, T. Hermann, M. Husmann, function of natural regulatory T cells in an inflammatory milieu. J. Immunol. G. Graupner, and M. Pfahl. 1990. A retinoic acid receptor-specific element 185: 2675–2679. controls the retinoic acid receptor-beta promoter. Mol. Endocrinol. 4: 1727– 26. Xiao, S., H. Jin, T. Korn, S. M. Liu, M. Oukka, B. Lim, and V. K. Kuchroo. 2008. 1736. Retinoic acid increases Foxp3+ regulatory T cells and inhibits development of 52. Sun, S. Y., H. Wan, P. Yue, W. K. Hong, and R. Lotan. 2000. Evidence that Th17 cells by enhancing TGF-beta-driven Smad3 signaling and inhibiting IL-6 retinoic acid receptor beta induction by retinoids is important for tumor cell and IL-23 receptor expression. J. Immunol. 181: 2277–2284. growth inhibition. J. Biol. Chem. 275: 17149–17153. 27. Massacesi, L., A. L. Abbamondi, C. Giorgi, F. Sarlo, F. Lolli, and L. Amaducci. 53. Swift, C. B., J. L. Hays, and W. J. Petty. 2008. Distinct functions of retinoic acid 1987. Suppression of experimental allergic encephalomyelitis by retinoic acid. receptor beta isoforms: implications for targeted therapy. Endocr. Metab. Im- J. Neurol. Sci. 80: 55–64. mune Disord. Drug Targets 8: 47–50. The Journal of Immunology 3151

54. Gong, M., Y. Bi, W. Jiang, Y. Zhang, L. Chen, N. Hou, J. Chen, and T. Li. 58. Schwartz, M., I. Cohen, O. Lazarov-Spiegler, G. Moalem, and E. Yoles. 1999. 2013. Retinoic acid receptor beta mediates all-trans retinoic acid facilitation The remedy may lie in ourselves: prospects for immune cell therapy in central of mesenchymal stem cells neuronal differentiation. Int. J. Biochem. Cell Biol. nervous system protection and repair. J. Mol. Med. 77: 713–717. 45: 866–875. 59. Uccelli, A., A. Laroni, and M. S. Freedman. 2011. Mesenchymal stem cells for 55. Murray, P. J., and T. A. Wynn. 2011. Protective and pathogenic functions of the treatment of multiple sclerosis and other neurological diseases. Lancet macrophage subsets. Nat. Rev. Immunol. 11: 723–737. Neurol. 10: 649–656. 56. Mosser, D. M., and J. P. Edwards. 2008. Exploring the full spectrum of mac- 60. DePaolo, R. W., V. Abadie, F. Tang, H. Fehlner-Peach, J. A. Hall, W. Wang, rophage activation. Nat. Rev. Immunol. 8: 958–969. E. V. Marietta, D. D. Kasarda, T. A. Waldmann, J. A. Murray, et al. 2011. Co- 57. Manicassamy, S., and B. Pulendran. 2011. Dendritic cell control of tolerogenic adjuvant effects of retinoic acid and IL-15 induce inﬂammatory immunity to responses. Immunol. Rev. 241: 206–227. dietary antigens. Nature 471: 220–224.