Cutting Edge: Inhaled Antigen Upregulates Retinaldehyde Dehydrogenase in Lung CD103⁺ but Not Plasmacytoid Dendritic Cells To Induce Foxp3 De Novo in CD4⁺ T Cells and Promote Airway Tolerance

CD11c⁺ DCs are essential for de novo induction of Foxp3 expression in naive CD4⁺ T cells and for airway tolerance

Because DCs are key APCs responsible for both Th cell and Treg development, we investigated their role in iTreg generation using an established airway tolerance model (10, 11). CD4⁺ T cells from DO11×RAG2^KO mice (devoid of Tregs; Supplemental Fig. 1) were transferred into naive BALB/c recipients, which were then repeatedly exposed to a low dose of inhaled OVA. This resulted in a significant increase in Foxp3 expression in the donor cells compared with cell transfer in recipient mice that were left untreated (Fig. 1A). To address the role of DCs in Foxp3 induction, CD11c-DTR-eGFP mice were similarly tested for their ability to induce Foxp3 in adoptively transferred naive CD4⁺ T cells after OVA exposure under DC-sufficient or DC-deficient (by treating with DT) conditions. Foxp3 induction was significantly reduced in the CD4⁺ T cells adoptively transferred into DT-treated mice (Fig. 1B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

CD11c⁺ DCs are essential for de novo induction of Foxp3 expression in CD4⁺ T cells. A total of 1 × 10⁶ DO11×RAG2^KO CD4⁺ splenic T cells was adoptively transferred i.v. into BALB/c (A) or CD11c-DTR-eGFP (B) recipients (DC sufficient or deficient) that were tolerized or not to inhaled OVA (n = 4–6 mice/group). Data shown are mean ± SEM and representative of three independent experiments. (C and D) OVA-tolerized DC-sufficient and DC-deficient CD11c-DTR-eGFP mice were immunized with OVA/CT and subsequently challenged with OVA. Twenty-four hours after the last challenge, various parameters of inflammation were analyzed (n = 3 or 4 mice/group). Data shown are mean ± SEM and are representative of two independent experiments. (D) Histological examination of lung sections stained with periodic acid–Schiff (original magnification ×20). Scale bar, 100 μm. **p<0.01, ***p< 0.005.

Next, when tested for the ability to mount tolerance by immunizing with OVA and CT (13, 14), DC-sufficient CD11c-DTR-eGFP mice showed signs of tolerance: the total cell numbers and numbers of granulocytes were low compared with nontolerized mice and were similar to the low numbers found in nonchallenged or naive mice (Fig. 1C). However, the DC-depleted mice failed tolerization, as evident upon sampling bronchoalveolar lavage fluid (BALF) (Fig. 1C) and examining lung histology (Fig. 1D).

De novo Foxp3 expression in CD4⁺ T cells depends on CD103⁺ DCs and not plasmacytoid DCs

The pulmonary CD11c⁺ population is heterogeneous and has been divided into different subsets based on the expression of specific cell surface molecules and anatomical location within the airways and the lung parenchyma (15). The four well-characterized respiratory DCs are CD11c^lo MHCII^lo B220 plasmacytoid DCs (pDCs), CD11c⁺MHCII^hiCD103⁺ DCs, CD11c⁺CD11b⁺MHCII^hi DCs, and CD11c⁺CD11b⁺MHCII^lo monocytic resident DCs (MoRDCs) (16) (Supplemental Fig. 2A). To identify which resident DCs are responsible for inducing Foxp3 expression in CD4⁺ T cells in our model of tolerance, we first examined DC populations in the lungs of DC-sufficient and DC-deficient mice. CD103⁺ DCs were completely depleted by DT treatment, whereas the frequency of CD11b⁺MHCII^hi DCs was reduced (Fig. 2A). In contrast, the number of pDCs and MoRDCs increased 20- and 4-fold, respectively, upon DT treatment (Fig. 2A). Because DT treatment abolished tolerance induction, these results suggested that pDCs are not involved in de novo Foxp3 induction in CD4⁺ T cells during tolerization, and the other subsets also were not efficient in Foxp3 induction in CD4⁺ T cells. Given that both pDCs and CD103⁺ DCs have been associated with tolerance in the gut, we further examined ex vivo the ability of sorted DC populations (Supplemental Fig. 2B) to induce Foxp3 in naive CD4⁺ T cells. In agreement with the results of the in vivo data, pDCs were totally incapable of inducing Foxp3 in naive CD4⁺ T cells, whereas CD103⁺ DCs displayed this ability (Fig. 2B). MoRDCs and CD11b⁺MHCII^hi DCs were also deficient in inducing Foxp3 like pDCs (data not shown), as were lung macrophages (Fig. 2B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

De novo Foxp3 expression in CD4⁺ T cells and airway tolerance depend on CD103⁺ DCs and not pDCs. (A) Representative flow cytometry data (left panel) and numbers (right panel) of various DC subsets in the lungs of CD11c-DTR-eGFP mice (with or without DT). (B) Sorted lung CD11c⁺ macrophages, pDCs, and CD103⁺ DCs from OVA-tolerized mice were cocultured with Treg-depleted CD4⁺ KJ1-26⁺ splenic T cells in the presence of OVA peptide. Representative flow cytometry data (left panel) and average frequency of KJ1-26⁺ Foxp3⁺ CD4⁺ T cells (right panel) induced by each cell subset. (C) DC–T cell coculture assay was repeated with CD103⁺ lung DCs sorted from naive, OVA-tolerized (OVA Tol) and OVA/CT-sensitized (OVA/CT) mice. Representative flow cytometry analysis (left panel) and average frequency (right panel) of Foxp3⁺ cells induced by CD103⁺ DCs from the respective conditions. (D) Number of CD103⁺ DCs in lung-draining LNs of naive and OVA Tol mice (n = 3). (E) MHCII, CD80, and CD86 expression on CD103⁺ DCs from lungs and LNs of naive (black), OVA Tol (red), and OVA/CT (blue) mice. Numbers shown represent median fluorescence intensities. Data shown are mean ± SEM and are representative of three (B, C) or two (D, E) independent experiments. *p< 0.05, ***p< 0.005, ****p< 0.001.

Our next question was whether CD103⁺ DCs were inherently tolerogenic or were rendered so under tolerogenic conditions. We observed that DCs isolated from the lungs of tolerized mice were the most efficient in inducing Foxp3 expression in DC–T cell coculture assays (Fig. 2C). Fig. 2D shows that exposure to inhaled Ag augments CD103⁺ DC numbers in the lung-draining lymph nodes (LNs) of the mice compared with those observed in naive mice. The DCs from tolerized mice expressed higher levels of MHCII than did those from naive mice, which rendered them more competent for Ag presentation (Fig. 2E), although the cells maintained low levels of CD80 and CD86 expression compared with those from OVA/CT-immunized mice that minimized their capacity to induce differentiation of naive T cells (Fig. 2E).

Lung CD103⁺ DCs, and not pDCs, upregulate aldh1a2 in tolerized mice

The results of the experiments described above showed that tolerizing conditions endow CD103⁺ DCs with a special ability to induce Foxp3 in CD4⁺ T cells. TGF-β (17, 18) and retinoic acid (RA) (8, 9, 19–25) together promote Foxp3 induction in CD4⁺ T cells. Lung CD103⁺ DCs from OVA-tolerized mice were cocultured with Treg-depleted CD4⁺ T cells from either wild-type (WT) mice or mice expressing a dominant-negative (DN) mutant of TGF-βRII on CD4⁺ T cells (CD4-TGF-βDNRII mice). We observed that the absence of TGF-β signaling in CD4⁺ T cells appreciably blunted the induction of Foxp3 expression by the DCs (Fig. 3A). We also found increased expression of tgfb1 in lung CD103⁺ DCs isolated from tolerized mice compared with DCs isolated from naive mice (Fig. 3B). Interestingly, blocking RA signaling by the selective RA receptor antagonist LE135 severely compromised their ability to induce Foxp3 expression (Fig. 3C). Reciprocally, RA improved the ability of DCs isolated from OVA/CT-immunized mice to induce Foxp3 (Fig. 3C). We next investigated the mRNA levels of retinaldehyde dehydrogenase 2 (aldh1a2), a key enzyme involved in retinal metabolism (26), in different lung DC subsets, as well as in CD11c⁺ macrophages isolated from OVA-tolerized mice. As shown in Fig. 3E, CD103⁺ DCs expressed higher aldh1a2 mRNA, whereas the pDCs and macrophages showed minimal expression. Although CD103⁺ DCs have been linked to iTreg development, these DCs have been also associated with Th2 (27) and Th17 differentiation (28). In this regard, it was interesting to observe that the relative abundance of aldh1a2 mRNA was lower in CD103⁺ DCs isolated from naive or OVA/CT-immunized mice compared with DCs isolated from tolerized mice (Fig. 3F).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

TGF-β and RA regulate CD103⁺ lung DC–mediated de novo Foxp3 expression in CD4⁺ T cells. (A) CD103⁺ lung DCs sorted from tolerized C57BL/6 (WT) mice were cocultured with WT or CD4-DN-TGF-βRII splenic CD4⁺ T cells. (B) Expression of tgfb1 in CD103⁺ lung DCs sorted from naive and OVA-tolerized (OVA Tol) mice. Expression of aldh1a2 and tgfb1 was analyzed using the 2^−∆Ct method. (C) DC–T cell coculture assay was repeated in the presence or absence of LE135 (1 μM). (D) CD103⁺ lung DCs sorted from OVA/CT-sensitized (OVA/CT) mice were used for DC–T cell coculture assay in the presence or absence of RA (10 nM). In (A), (C), and (D), representative flow cytometry data (left panel) and frequency (right panel) of induced CD4⁺ Foxp3⁺ T cells are shown. (E) Expression of aldh1a2 mRNA in CD11c⁺ macrophages, pDCs, and CD103⁺ DCs sorted from the lungs of OVA-tolerized mice. (F) Expression of aldh1a2 in CD103⁺ lung DCs sorted from naive, OVA Tol, and OVA/CT mice. Data in (A), (C), and (D) are mean ± SEM and are representative of three independent experiments. Data in (B), (E), and (F) are mean ± SD and are representative of two independent experiments. *p< 0.05, **p<0.01, ***p< 0.005.

To confirm the role of CD103⁺ DCs in inducing tolerance, Batf3^−/− mice were first analyzed for the presence of CD103⁺ DCs and macrophages in the lung. Batf3 deficiency was previously shown to result in loss of CD103⁺ DCs in tissues and LNs in mice (29). While no CD103⁺ DCs were detected in the Batf3^−/− mice, lung macrophage numbers were unchanged (Fig. 4A). WT and Batf3^−/− mice were next subjected to the tolerance protocol with inhaled OVA, and the mice were then tested for tolerance (13, 14). When BALF was examined 24 h after the last OVA challenge, unlike WT mice, Batf3^−/− mice showed signs of inflammation (i.e., increased total cell numbers and numbers of granulocytes compared with WT mice) (Fig. 4B). Representative lung histologies corroborated the conclusion that the absence of CD103⁺ DCs results in a failure to induce tolerance (Fig. 4C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

CD103⁺ DCs are essential for inducing tolerance. (A–C) OVA-tolerized WT and CD103⁺ DC–deficient Batf3^−/− mice were immunized with OVA/CT and subsequently challenged with OVA. Twenty-four hours after the last challenge, various parameters of inflammation were analyzed. (A) Flow cytometry analysis of CD103⁺ DCs and macrophages in the lungs of WT and Batf3^−/− mice. (B) Total (top panels) and differential (bottom panels) cell counts in BALF cytospins under various conditions (n = 3 mice/group). Data shown are mean ± SEM. (C) Histological examination of lung sections stained with periodic acid–Schiff (original magnification ×10). Scale bar, 100 μm. **p<0.01, ****p< 0.001.

In summary, our study shows the superior ability of pulmonary CD103⁺ DCs to induce Foxp3 expression de novo in CD4⁺ T cells, which is essential for establishment of airway tolerance. In the current study, we establish that deletion of CD103⁺ DCs totally abolishes de novo Foxp3 expression in adoptively transferred naive CD4⁺ T cells, despite the presence of substantial numbers of pDCs, MoRDCs, and CD11b⁺MHCII^hi DCs in the lungs. An earlier study showed a role for lung pDCs in the inhibition of an inflammatory response to Ag (7) that was not observed in our own studies, which could have been due to differences in the Ag dose and mode of Ag delivery between the two studies. The model used by us mimics natural inhalation of aeroantigens (10), which was estimated to deposit <30 μg of OVA in the tissue/d (11), in contrast to the 800 μg of OVA that was delivered intratracheally/d in the previous study (7). The different subsets of pulmonary DCs exhibit phenotypic and functional differences and are anatomically segregated in the respiratory mucosa (15, 16). pDCs are normally the least abundant subset in the tissue under steady-state conditions (16). CD103⁺ DCs line the conducting airways with their dendrites extended into the airway lumen (15), allowing them to readily capture inhaled Ags or pathogens. Our results also show that the tolerogenicity in CD103⁺ DCs is dependent on TGF-β and RA.

Recently, a study showed that lung tissue macrophages can induce Foxp3 in intratracheally transferred naive CD4⁺ T cells (30). The lung macrophages did not migrate to lung-draining LNs in response to Ag, which was also observed in this study (data not shown). Because Ag provocation results in increased numbers of CD103⁺ DCs in lung-draining LNs (Fig. 2D), where naive CD4⁺ T cells normally reside and can be primed with Ag presented by DCs, we feel that de novo Foxp3 induction in naive T cells occurs in the LN in which CD103⁺ DCs play an important role. However, lung macrophages may help to maintain Foxp3 expression in the CD4⁺ T cells once the cells populate the lung tissue. Importantly, despite harboring macrophages and pDCs in the lung, Batf3^−/− mice, which lack CD103⁺ DCs, failed to be tolerized by inhaled Ag, demonstrating an essential role for these DCs in de novo Foxp3 induction in naive CD4⁺ T cells.

The results also suggest that uptake of soluble Ag in the absence of adjuvant is sufficient for upregulation of aldh1a2 expression, which, in turn, is required for RA production, thereby increasing Foxp3 induction. It is interesting to note that when the same Ag was combined with an adjuvant, the expression of aldh1a2 was attenuated in the CD103⁺ DCs. Concurrently, the addition of RA enhanced the ability of lung CD103⁺ DCs from mice immunized for inflammation to induce Foxp3 in naive CD4⁺ T cells, suggesting that, just as aldh1a2 expression is important for inducing Foxp3 expression, its inhibition is also important for the induction of inflammation. In aggregate, it appears that DCs possess the dual capacity to induce immune activation or immune suppression, depending upon the context; the latter involves de novo Foxp3 expression to generate iTregs. This adaptable immunoregulatory function makes DCs ideal candidates for immunotherapy that requires their appropriate activation for the treatment of cancer or suppression to combat autoimmune or allergic diseases and transplant rejection.