Toll-like receptor 2-dependent induction of vitamin A-metabolizing enzymes in dendritic cells promotes T regulatory responses and inhibits autoimmunity

Abstract

Immune sensing of a microbe occurs via multiple receptors. How signals from different receptors are coordinated to yield a specific immune response is poorly understood. We show that two pathogen recognition receptors, Toll-like receptor 2 (TLR2) and dectin-1, recognizing the same microbial stimulus, stimulate distinct innate and adaptive responses. TLR2 signaling induced splenic dendritic cells (DCs) to express the retinoic acid metabolizing enzyme retinaldehyde dehydrogenase type 2 and interleukin-10 (IL-10) and to metabolize vitamin A and stimulate Foxp3(+) T regulatory cells (T(reg) cells). Retinoic acid acted on DCs to induce suppressor of cytokine signaling-3 expression, which suppressed activation of p38 mitogen-activated protein kinase and proinflammatory cytokines. Consistent with this finding, TLR2 signaling induced T(reg) cells and suppressed IL-23 and T helper type 17 (T(H)17) and T(H)1-mediated autoimmune responses in vivo. In contrast, dectin-1 signaling mostly induced IL-23 and proinflammatory cytokines and augmented T(H)17 and T(H)1-mediated autoimmune responses in vivo. These data define a new mechanism for the systemic induction of retinoic acid and immune suppression against autoimmunity.

Figure 1

Mechanism of induction of vitamin-A metabolizing enzymes in splenic DCs. (a) Expression of Raldh (Aldh1a) mRNA in splenic DCs cultured in vitro for 24 h with (black bars), or without (grey bars) zymosan. Expression of Raldh mRNA relative to expression of mRNA encoding glyceraldehyde phosphate dehydrogenase (GAPDH) was analyzed by quantitative RT-PCR in this and all figures below. (b) Induction of Raldh2 (Aldh1a2) in splenic DCs cultured in vitro with TLR ligands or curdlan for 24 h. (c) Induction of Raldh2 in splenic DCs in vivo, in C57BL/6 mice injected with zymosan. (d) Western blot analysis of expression of Raldh2 protein in splenic DCs in vivo, 5 h after C57BL/6 mice were injected with zymosan. (e) Immunofluorescence microscopy of frozen tissue section of spleens of C57BL/6 mice injected with PBS or Zymosan, fixed and stained with antibodies specific for mouse CD11c (red), Raldh (green) and B220 (blue). (f) Induction of Raldh2 is dependent on TLR2. Left panel: Splenic DCs from WT or Tlr2^−/− mice were cultured in media alone or with zymosan and, 24 h later expression of Raldh2 mRNA was analyzed. Right panel: C57BL/6 mice or Tlr2^−/− were injected with zymosan and at various time points, splenic DCs were isolated, and expression of Raldh2 mRNA was analyzed. (g) Raldh2 induction is not dependent on dectin-1. Splenic DCs from wild type or dectin-1^−/− mice were cultured in media alone or with zymosan, and 24 h later expression of Raldh2 mRNA was analzed. Difference between wild type and dectin-1^−/− mice is not significant. (h) Induction of Raldh2 is dependent on ERK MAPK signaling. Splenic DCs were cultured as described above, with or without U0126. (i) Induction of Raldh2 in splenic DCs by Candida albicans is dependent on TLR2. Splenic-DCs from WT or Tlr2^−/− mice were cultured in media alone or with Candida albicans. In all figures results are means + s. d. of 2 – 3 mice per group in one representative experiment out of two or three. *,P < 0.05; **,P < 0.005; ***P < 0.0001 in all figures.

Figure 2

RA and IL-10 synergize to induce Foxp3⁺ Treg cells. (a) Zymosan stimulates splenic DCs to induce Treg cells. Splenic DCs were stimulated with zymosan or curdlan or LPS for 12 h, and washed and cultured with naïve (CD4⁺CD62L⁺) OT-II T cells with OVA_323–339 peptide (OVA) in the presence or absence of TGF-β. After 4 d, OT-II cells were restimulated for 6 h with plated bound antibodies to CD3 and CD28. Foxp3 expression and, intracellular production of IL-17, IFN-γ and IL-10 by CD4⁺ T cell were assessed by intracellular staining and flow cytometry. Data are from one representative experiment of three. (b) Induction of Treg cells by zymosan stimulated splenic DCs is dependent on RA. Splenic DCs were stimulated with zymosan in the presence of disulphiram or vehicle for 10 h, and washed and cultured with naïve OT-II T cells with OVA and TGF-β in the presence or absence of retinol. After 4 d, Foxp3 expression by CD4⁺ T cell was assessed by intracellular staining and flow cytometry. Data are representative of one experiment of three. (c) Induction of Treg cells by zymosan stimulated splenic DCs, is dependent on TLR2. Splenic DCs from wild type or Tlr2^−/− mice were stimulated with zymosan for 10h, and washed and cultured with naïve OT-II T cells with OVA and TGF-β in the presence or absence of retinol. After 4 d, Foxp3 expression by CD4⁺ T cell was assessed by intracellular staining and flow cytometry. Data are representative of one experiment of three. (d) Effect of blocking ERK activation in zymosan stimulated DCs on Treg induction. Data are representative of one experiment of three.

Figure 3

RA and IL-10 exert autocrine effects on DCs to induce Socs3 which regulates activation of p38 MAPK and pro-inflammatory cytokines. (a) Expression of RA nuclear receptors (RARs and RXRs) in splenic DCs by western blot. (b, c) Cytokines secreted in supernatants obtained after culture of splenic DCs from wild-type mice (b), and Il-10^−/− mice (c), with zymosan for 24 h, in the presence or absence of antibodies against IL-10 receptor (aIL-10R) IL-10R, retinol or retinol plus LE135/ LE540, or IL-10. Representative of four experiments. (d, e) RA dependent induction of Socs3 mRNA expression in splenic DCs from wild type (d) or IL-10^−/− mice (e) stimulated with zymosan. (f, g) RA-dependent induction of Socs3 in splenic DCs in vivo. C57BL/6 mice were injected with zymosan or zymosan plus disulphiram (g), or zymosan plus LE135/ LE540 (g), and spleens were harvested at different time points. RNA was isolated from purified splenic DCs, and expressions of Socs1 and Socs3 mRNA were analyzed by quantitative RT-PCR. (h) Induction of Socs3 in splenic DCs in vivo is dependent on TLR2. Wild type or Tlr2^−/− mice were injected with zymosan, splenic DCs isolated and mRNA expression of Socs3 evaluated by RT-PCR. Representative of three experiments. *P < 0.01; **P < 0.001; ***P < 0.0001 in all figures.

Figure 4

Induction of antigen specific IL-10⁺ Tr1 and Treg cells in vivo. (a) B6.PL (Thy1.1⁺) mice reconstituted with OT-II TCR transgenic T cells were injected i.v. with class II–restricted OVA_323–339 peptide (OVA) alone, or OVA plus LPS, OVA plus zymosan, or OVA plus curdlan. Four days after challenge, the splenocytes were isolated and expression of Foxp3, IL-17, IFN-γ and IL-10 by CD4⁺ T Thy1.2⁺ cell was assessed by intracellular staining and flow cytometry. Data are from one experiment representative of two. (b) C57BL/6 or Tlr2^−/− or Il-10^−/− mouse were reconstituted with OT-II TCR transgenic T cells, and on the following day, injected with OVA or OVA plus zymosan. Five days later, splenocytes were isolated and induction of OVA specific Foxp3⁺ T cells was assessed by intracellular staining and flow cytometry. Means + s. d. of 3 or 4 mice per group. (c) Splenocytes from immunized mice described in (b) were restimulated with OVA in culture for 48 h and cytokines in the supernatants were analyzed by ELISA. Means + s. d. of 3 or 4 mice per group. *P < 0.01; **P < 0.001; ***P < 0.0001 in all figures.

Figure 5

Zymosan suppresses IL-23 and T_H-17 mediated EAE. (a) In wild type mice (left panel), immunization with MOG + zymosan resulted in substantially reduced EAE, relative to immunization with MOG + CFA. In Tlr2−/− mice, immunization with MOG + zymosan resulted in enhanced disease, relative to wild type mice (P < 0.0001). Representative experiment of two. (b) Wild type or Tlr2^−/− mice were immunized s.c with MOG + CFA, MOG + CFA plus zymosan (i.v) or MOG + CFA plus curdlan (i. v), and monitored for disease. Injection of zymosan suppressed the severity of disease, relative to immunization with MOG + CFA alone, in wild type mice (left panel), but not in Tlr2^−/− mice (right panel). Representative experiment of two. (c) Mononuclear cells were isolated from CNS tissue on day 18 after immunization and induction of IFN-γ, IL-17, IL-10 and Foxp3 was assessed by intracellular staining and flow cytometry, as described in Supplementary Methods, online. Representative experiment of two. (d) Wild type or Tlr2^−/− mice were injected with zymosan, and expression of Il23p19 mRNA in splenic DCs analyzed by quantitative RT-PCR. Representative of three experiments. (e) IL-23 induction in the serum of the mice described in (d) was assayed by ELISA. (f) TLR2 regulates IL-23 production in splenic DCs in response to zymosan. IL-23 secretion by splenic DCs from wild type or Tlr2^−/− mice cultured in vitro with zymosan. *P < 0.01; **P < 0.001; ***P < 0.0001.

Figure 6

Mechanism of induction of Raldh2 in DCs. Innate sensing of zymosan via TLR2 efficiently (thick arrows) induces ERK activation and Raldh2. Dectin-1 does not play a major role in Raldh2 induction (thin arrow), although signaling via syk is critical, suggesting the involvement of an additional syk-dependent receptor (X). Thus, the combinatorial activation of TLR2 dependent ERK and syk, likely orchestrates induction of Raldh2. This results in the conversion of retinal to RA, which then exerts an autocrine effect on DCs via RAR/RXR to induce SOSC3, which suppresses activation of p38 MAPK and pro-inflammatory cytokines. In contrast, dectin-1 promotes induction of pro-inflammatory cytokines.