CD8⁺ CD205⁺ splenic dendritic cells are specialized to induce Foxp3⁺ regulatory T cells

Abstract

Foxp3⁺CD25⁺CD4⁺ regulatory T cells (T reg) mediate immunological self-tolerance and suppress immune responses. A subset of dendritic cells (DCs) in the intestine is specialized to induce T reg in a transforming growth factor (TGF)-β and retinoic acid (RA) dependent manner to allow for oral tolerance. Here we compare two major DC subsets from mouse spleen. We find that CD8⁺ DEC-205/CD205⁺ DCs, but not the major fraction of CD8⁻ dendritic cell inhibitory receptor-2 (DCIR2)⁺ DCs, induce functional Foxp3⁺ T reg from Foxp3⁻ precursors in the presence of low doses of antigen but without added TGF-β. CD8⁺ CD205⁺ DCs preferentially express TGF-β, and the induction of T reg by these DCs in vitro is blocked by neutralizing antibody to TGF-β. In contrast, CD8⁻ DCIR2⁺ DCs better induce Foxp3⁺ T reg when exogenous TGF-β is supplied. In vivo, CD8⁺ CD205⁺ DCs likewise preferentially induce T reg from adoptively transferred, antigen-specific, DO11.10 RAG^−/− Foxp3⁻ CD4⁺ T cells, whereas the CD8⁻ DCIR2⁺ DCs better stimulate natural Foxp3⁺ T reg. These results indicate that a subset of DCs in spleen, a systemic lymphoid organ, is specialized to differentiate peripheral Foxp3⁺ T reg, in part through the endogenous formation of TGF-β. Targeting of antigen to these DCs might be useful for inducing antigen-specific Foxp3⁺ T reg for treatment of autoimmune diseases, transplant rejection and allergy.

Introduction

Naturally occurring Foxp3⁺CD25⁺CD4⁺ T regulatory cells (natural Foxp3⁺ T reg), which express the Foxp3 transcription factor and high affinity IL-2 receptor (CD25), derive from the thymus and sustain self-tolerance (1). Foxp3⁺ T reg can also be differentiated or induced from conventional Foxp3⁻CD25⁻CD4⁺ T cells in the periphery with some stimuli such as TGF-β supplementation (2–6). Natural and induced Foxp3⁺ T reg suppress autoimmunity as well as allergy, graft rejection, and immune responses to microbes and tumors (1, 5, 7–10). It is important to understand the generation of antigen-specific Foxp3⁺ T reg to be able to suppress immunity in an antigen-specific manner and avoid global immune suppression by polyclonal T reg.

T cell responses are controlled by dendritic cells (DCs). DCs are antigen presenting cells (APCs) specialized to capture and process antigens for presentation on MHC products and then to control the subsequent differentiation of T cells (11–13). Two such specializations are the expression of numerous receptors that mediate antigen uptake and processing (14, 15), and localization to the T cell rich areas of peripheral lymphoid organs (16, 17). DCs initiate T cell immunity but can also induce tolerance, as is desirable in the case of harmless self and environmental antigens (18–20). Tolerance can develop by different pathways, such as deletion (21, 22), induction of CD5 (23), or both induction and expansion of T reg (5, 24–31). We have recently shown that relative to bulk spleen cells, DCs are much more effective inducers of functional Foxp3⁺ T reg from Foxp3 negative peripheral CD4⁺ T cells (9)

Classical DCs in mouse spleen are comprised of two major subsets that express distinct markers and functions (12, 32, 33). One subset is CD8⁺ and DEC-205/CD205⁺, and the second is CD8⁻ CD205⁻ and dendritic cell inhibitory receptor-2 (DCIR2)⁺, the latter is recognized by the 33D1 mAb (32, 34, 35). Splenic DC-subsets can have different functions in T cell differentiation e.g. CD8⁺ CD205⁺ DCs can induce IFN-γ producing Th1 T cells while CD8⁻ DCIR2⁺ DCs induce Th2 responses (36–39). DC subsets, marked by the presence or absence of the CD103 integrin, are also evident in the intestine and intestine-associated lymphoid organs. It has recently been shown that the CD103⁺ subset is active in inducing Foxp3⁺ T reg from Foxp3⁻ T cells in the presence of endogenous TGFβ, and that the DCs metabolize vitamin A to retinoic acid as an enhancing cofactor (40, 41). These reports found that CD103⁺ DCs from both mesenteric LN and lamina propria could induce a small fraction of Foxp3⁺ cells (2.5–9%) from Foxp3⁻ precursors.

Here, we investigate the capacity of spleen DC subsets to induce ovalbumin (OVA)-specific Foxp3⁺ T reg. We find that CD8⁺ spleen DCs are selectively active and produce the required endogenous TGF-β, whereas CD8⁻ spleen DCs require exogenous TGF-β but then become more proficient than CD8⁺ DCs at inducing T reg. We will also show that targeted in vivo antigen-delivery to CD8⁺ CD205⁺ DCs but not CD8⁻ DCIR2⁺ DCs likewise preferentially induces Foxp3⁺ T reg, even though CD8⁻ DCIR2⁺ DCs more efficiently form peptide-MHC II complexes (35) and better expand preformed natural T reg in vivo. These results indicate that the endogenous differentiation of T reg is controlled by select subsets of DCs in lymphoid tissues, and not only DC subsets in the intestine.

Materials and methods

Mice

6–8 week, specific pathogen free, female, C57BL/6 (B6) and BALB/c were purchased from Taconic (Germantown, NY). DO11.10 RAG^−/− mice were obtained through Taconic, the NIAID Exchange Program (NIH) (42), while DO11.10 RAG^+/+ mice were kindly provided by Dr. P. Marrack (National Jewish Medical and Research Center). We received generous gifts of OT-II mice from Dr. F. Carbone (University of Melbourne, Australia), Foxp3-IRES-RFP (FIR) knockin mice from Dr. R. Flavell (Yale University, CT) (43) and BALB/c Thy1.1⁺ mice from Drs. M. Lafaille and J. Lafaille (New York University, NY). All mice were used according to guidelines of the institutional animal care and use committee of the Rockefeller University.

Antibodies and reagents

PE, FITC, PerCPCy5.5 or APC conjugated anti-CD25 (7D4), CD25 (PC61), CD4 (H129.19), CD8, DX5, B220, CD3, CD11c, streptavidin (SA) PE, SAPECy7, purified anti-CD16/CD32 (2.4G2), anti-IL-10 and Isotype rat Abs were from BD PharMingen (San Diego, CA). Anti-CD11c, FITC, streptavidin, PE, DX5, CD19 and CD90.2 microbeads were from Miltenyi Biotec (Gladbach, Germany). Anti-DEC-205 (NLDC-145) and 33D1(anti-DCIR2) were purified and labeled with biotin or Alexa647 at Memorial Sloan-Kettering Cancer Center (MSKCC), Rockefeller Monoclonal Antibody Core Facility. Carboxyfluorescein diacetate succinimidyl ester (CFSE), live dead fixable aqua, TOPRO-3, and streptavidin Pacific blue was from Molecular Probes (Eugene, OR). Anti-mouse Foxp3 (FJK-16s) staining kit and anti-CD4-Alexa700 were from eBioscience (San Diego, CA). Human TGF-β1, anti-mouse TGF-β mAb (1D11) and isotype control were from R&D systems (Minneapolis, MN). Human IL-2 was from Chiron Corp (Emeryville, CA). Anti-KJ1.26 mAb to the clonotypic receptor on OVA-specific DO11.10 TCR transgenic T cells was from Caltag/Invitrogen (Carlsbad, CA). Poly IC was from InVivoGen (San Diego, CA).

T cells and DCs

CD4⁺ T cells were first negatively separated by MACS beads from lymph nodes and spleen cell suspensions (>90%)(Miltenyi Biotech) and further purified by flow cytometry (>99%). Spleen CD11c⁺ DCs were selected with anti-CD11c beads, and further purified by flow cytometry into CD8⁺ or CD8⁻ CD11c⁺ B220⁻ CD3⁻ DX5⁻ DCs (>95%). The DCs (2×10⁴) were irradiated (15 Gy) prior to co-culture with T cells (2×10⁴). The purified T cells and APCs were cultured in 96-well plates (Corning Costar) with or without exogenous TGF-β (2 ng/ml) in the presence of OVA peptide (323–336). Live cell numbers per culture were calculated from the yields from flow cytometry.

Flow cytometry

Cultured cells were first stained with anti-CD4, CD25, CD11c, live dead aqua with or without the other Abs, and after fixation, intracellular Foxp3 was stained. BDLSRII, FACSort, FACSAria or FACSdiva flow cytometers were used (Becton Dickinson). To assess cell numbers, all cells were acquired from each culture and analyzed with FlowJo software (Tree Star Inc, OR).

ELISA for TGF-β

Spleen CD8⁺ or CD8⁻ DCs were purified by flow cytometery from naïve BALB/c mice or BALB/c mice that had been injected with 50 μg poly IC 12 hours before. The purified DCs were cultured in serum free Nutridoma medium (Invitrogen) for 15–40 h. The concentrations of TGF-β in the supernatants were measured by TGF-β ELISA kit (R&D). Following the manufacturer’s instructions, we measured the TGF-β with or without activation of the latent form of TGF-β.

Adoptive transfer of DO11.10 RAG^−/− CD4⁺ T cells or natural T reg

CFSE-labeled OVA-specific, CD4⁺ T cells from DO11 RAG^−/− mice (2×10⁶) were adoptively transferred into BALB/c Thy1.1⁺ mice. The recipients were stimulated with indicated doses of DEC-OVA, 33D1-OVA or Iso-OVA i.p. one day later. After 10–14 days, a mixture of spleen and lymph node cells was analyzed with FACS gating on the donor Thy1.2⁺ cells. For natural T reg, CD25⁺ CD4⁺ T cells from Thy1.2⁺ DO11.10 transgenic mice were purified by flow cytometry, and expanded with DCs, peptide and IL-2 for 1 week (25). The expanded T reg or freshly isolated T reg were CFSE-labeled, and adoptively transferred into BALB/c Thy1.1⁺ recipients 1 day before injection with DEC-OVA or 33D1-OVA. After 3 d and 14 d, mixtures of spleen and lymph nodes were evaluated with FACS to monitor cell division by CFSE dilution and to enumerate Thy1.2⁺ cells.

Production of chimeric mAbs

Chimeric mAbs (DEC-OVA, 33D1-OVA, and Iso-OVA) were produced by transient transfection as described (21, 35). All mAbs were tested for LPS contamination (Fisher-Cambrex) and were decontaminated when necessary (Pierce).

Gene array

We further investigated our previous gene array data (GEO Series accession number GSE6259) (35). Briefly, CD11c⁺ DCs were enriched from spleens of B6 mice by depletion of lymphocytes with antibodies to DX5, CD19 and CD90.2 on MACS beads (Miltenyi Biotech). Cells were stained with DEC-205-biotin, then with CD11c-PE, 33D1(anti-DCIR2)-Alexa-647 and SAPECy7, and purified by cell sorting into CD11c^High DCIR2^High CD8⁻and CD11c^High DEC-205^High CD8⁺ on a FACS Vantage. All cell samples were purified to more than 99% homogeneity. Total RNA was prepared using Qiagen RNeasy Mini kit (Qiagen). DNA microarray analysis of gene expression was performed at gene array facility (MSKCC, Genomics Core Laboratory, New York). Fluorescent images of hybridized microarrays (Affymetrix, MOE-430 2.0) were obtained using an Affymetrix Genechip Scanner. Microarray data was analyzed using Affymetrix GeneSpring 7.0 software. All samples were repeated at least three times with individually sorted cells and averaged.

Results

Peripheral differentiation of Foxp3⁺ T reg in vitro is influenced by the DC subset

Recently, it has been shown that DCs from mesenteric lymph nodes induce some Foxp3⁺ T reg using endogenous TGF-β (9). Given the specialized role of the CD103⁺ DC subset from the intestinal environment (40, 41), we hypothesized that a subset of DCs in systemic lymphoid tissue might also have a role in inducing Foxp3⁺ T reg in the periphery. We therefore evaluated two major CD8⁺ and CD8⁻ DC subsets from spleen, purifying these on the FACS and adding the DCs to Foxp3⁻ CD25⁻ CD4⁺ T cells from OVA-specific, DO11.10 RAG^−/− mice in the absence of exogenous TGF-β. After 5 days of culture, we observed that CD8⁺ DCs but not CD8⁻ DCs differentiated some Foxp3⁺ T reg at 2 doses of OVA peptide (Fig. 1A,B top). In contrast, after addition of TGF-β into the culture, CD8⁻ DCs were more potent inducers if the total numbers of newly formed Foxp3⁺ T reg were measured (Fig. 1A,B bottom). In the absence of antigen (Fig. 1B) or in the presence of a nonspecific MHC II binding peptide from influenza virus hemagglutinin (HA) (data not shown), both CD8⁺ DCs and CD8⁻ DCs did not induce Foxp3. These results indicate that the minor subset of spleen CD8⁺ DCs is able to differentiate Foxp3⁺ T reg in an antigen-dependent manner.

Fig. 1. CD8⁺ spleen DCs differentiate Foxp3⁺ CD25⁺CD4⁺ T reg from Foxp3⁻ CD25⁻ CD4⁺ T cells in the absence of exogenous TGF-β in vitro.

(A) Freshly isolated Foxp3⁻CD25⁻CD4⁺ T cells from DO11.10 RAG^−/− mice (2×10⁴) were cultured with the indicated doses of peptide and CD8⁻ or CD8⁺ DCs (2×10⁴) in the absence or presence of TGF-β (2 ng/ml). After 5 days of culture, cells were analyzed by FACS. Representative gates of CD4⁺ CD11c⁺ cells are shown on the left.

(B) As in (A), but the number of Foxp3⁺ cells, the percentage of Foxp3⁺ T cells/CD4⁺ T cells, and the number of CD4⁺ T cells per culture in the presence or absence of TGF-β are shown. The number of Foxp3⁺ cells at 0 μg/ml peptide was < 10². A summary of 5–7 separate experiments. P values were provided by student-t test.

CD8⁺ DCs use endogenous TGF-β to differentiate Foxp3⁺T reg in vitro

The differentiation of Foxp3⁺ T reg from Foxp3⁻ cells occurs by stimulation with TGF-β (2–5, 9). To investigate if endogenous TGF-β allows CD8⁺ DCs to differentiate Foxp3⁺ T reg, we performed blocking experiments with neutralizing anti-TGF-β and control monoclonal antibodies (mAb) (Fig. 2A). Anti-TGF-β blocked the induction of Foxp3⁺ T reg by CD8⁺ DCs, as measured by the percentage (Fig. 2A) and also the total numbers of induced Foxp3⁺ T reg in the culture (Fig. 2B). In addition to TGF-β, we investigated the role of IL-10, as IL-10 from pulmonary DCs helps to induce IL-10 producing Tr1-type T reg (44). However, the induction of Foxp3⁺ T reg by spleen CD8⁺ DCs was not blocked by the presence of neutralizing anti-IL-10 mAb in the culture (Fig. 2C and D). Therefore, it is unlikely that IL-10 has a role in the observed induction of Foxp3⁺ T reg by CD8⁺ spleen DCs.

Fig. 2. CD8⁺ DCs differentiate Foxp3⁺ CD25⁺ CD4⁺ T reg using endogenous TGF-β.

(A) Freshly isolated Foxp3⁻ CD25⁻ CD4⁺ T cells from DO11.10 RAG^−/− mice (2×10⁴) were cultured with 0.03 μg/ml peptide with CD8⁺ DCs (2×10⁴) in the presence of anti-TGF-β mAb or isotype matched control mAb (10 μg/ml). After 5 days of culture, cells were analyzed with FACS by gating on CD4⁺ CD11c⁻ T cells.

(B) As in (A), but the number of Foxp3⁺ cells, the percentage of Foxp3⁺ T cells/CD4⁺ T cells, and the number of CD4⁺ T cells per culture are shown. A summary of 3 separate experiments. P values were provided by student-t test.

(C) As in (A), but cells were cultured in the presence of anti-IL-10 mAb or isotype control mAb (10 μg/ml). After 5 days of culture, cells were analyzed with FACS by gating on CD4⁺ CD11c⁻ T cells.

(D) As in (C), but the number of Foxp3⁺ cells, the percentage of Foxp3⁺ T cells/CD4⁺ T cells, and the number of CD4⁺ T cells per culture are shown. These data are representative of 3 experiments.

To verify that TGF-β was being produced by DCs in the DC-T cell cultures, and in particular by CD8⁺ CD205⁺ DCs, we did microarray analyses on RNA from CD8⁺ CD205⁺ CD11c⁺ and CD8⁻ DCIR2⁺ CD11c⁺ subsets. We found that CD8⁺ CD205⁺ DCs selectively expressed TGF-β and the TGF-β binding protein (Ltbp2) (Fig. 3A), which is important for the efficient secretion and appropriate localization of latent TGF-β (45). On the other hand CD8⁻ DCIR2⁺ DCs express TGF-β receptor 1 and some members of the SMAD family of transcription factors (Fig. 3A) that are activated by TGF-β receptors through phosphorylation (46). These results imply that CD8⁺ DCs can use endogenous TGF-β to induce T reg from Foxp3⁻ precursors, whereas CD8⁻ DCs are ready to respond to TGF-β that is provided from other types of cells.

Fig. 3. CD8⁺ DCs produce TGF-β.

(A) Total RNA was isolated from splenic CD8⁺ DEC-205⁺ or CD8⁻ DCIR2⁺ DCs. Affymetrix gene array analysis showing relative levels of mRNAs associated with the TGF-β and TGF-β receptor signaling pathways expressed by CD8⁻ DCIR2⁺ (left panel) and CD8⁺ DEC-205⁺ DCs (right panel) purified from B6 mice. Each lane consists of three independent microarrays of mRNAs from different cell sorts.

(B) CD8⁺ or CD8⁻ DCs from naïve BALB/c mice or BALB/c mice that had been injected with poly IC 12 hours before were cultured for 15–40 hours in serum free medium and the concentrations of TGF-β in the supernatants were measured by ELISA. The concentration of TGF-β in the serum free medium was 0. A summary of 5 separate experiments. P value was provided by student-t test.

(C) CD8⁺ or CD8⁻ DCs from naïve BALB/c mice or BALB/c mice that had been injected with poly IC 12 hours before were purified by flow cytometry. Pre-sort (top) and post-sort (middle) plots are shown as gated on CD11c⁺ cells. The purified cells were further stained with CD86 (black line) and isotype control Abs (shaded gray line) (bottom).

(D) The purified CD8⁺ or CD8⁻DCs (2×10⁴) from naïve or poly IC injected mice as in (C) were cultured with freshly isolated Foxp3⁻ CD25⁻ CD4⁺ T cells from DO11.10 RAG^−/− mice (2×10⁴) in the presence of 0.03 μg/ml peptide. After 5 days of culture, cells were analyzed with FACS by gating on CD4⁺ CD11c⁻ T cells.

(E) As in (D), but the frequency of Foxp3⁺ T cells/CD4⁺ T cells per culture is also shown. These data are representative of 3 independent experiments with each experiment done in triplicate. P value was provided by student-t test.

To confirm the production of TGF-β at the protein level, we cultured CD8⁺ or CD8⁻ spleen DC subsets and measured TGF-β in the supernatants by ELISA (Fig. 3B). We compared TGF-β concentrations with or without activation of the latent form of TGF-β. Without activation, TGF-β was not detected (data not shown). With activation, we observed a significant amount of TGF-β in the supernatants, with CD8⁺ spleen DCs producing higher amounts of TGF-β than CD8⁻ spleen DCs (Fig. 3B). To evaluate if DC maturation affects the production of TGF-β by CD8⁺ or CD8⁻ DCs, we chose one maturation stimulus, poly IC, as it acts as an effective adjuvant for DC-targeted vaccine (47). When BALB/c mice were injected i.p. with 50 μg of poly IC 12 hours earlier, CD8⁺ and CD8⁻ DCs showed evidence of maturation with higher CD86, CD40 and class II expression (Fig. 3C and data not shown). When the purified DCs were cultured, we found the supernatants from the mature CD8⁺ DCs from poly IC injected mice contained less TGF-β compared to the immature CD8⁺ DCs from naïve mice (Fig. 3B). The mature CD8⁻ DCs from poly IC injected mice produced the same amounts of TGF-β as immature CD8⁻ DCs (Fig. 3B). To evaluate if the reduction of TGF-β production by mature DCs affects the induction of Foxp3, CD8⁺ DCs from poly IC injected mice or non-injected naïve mice were cultured with DO11.10 RAG^−/− CD4⁺ T cells. The frequency of the induced Foxp3⁺ T reg by the mature CD8⁺ DCs was much less than by the immature CD8⁺ DCs (Fig. 3D and E). This indicates that spleen CD8⁺ DCs in the steady state produce a higher amount of a latent form of TGF-β than spleen CD8⁻ DCs, which is essential for induction of Foxp3⁺T reg and that one TLR ligand causes CD8⁺ DCs to reduce TGF-β production.

Suppressive function of DC-induced Foxp3⁺ T reg in vitro

To determine if CD8⁺ and CD8⁻ DCs induce Foxp3⁺ T reg with suppressive function, we used Foxp3-internal ribosomal entry site-linked monomeric red fluorescent protein (Foxp3-IRES-RFP) knockin mice that had been crossed to OVA transgenic OT II B6 mice (FIR-OTII) (43), so that we could then purify induced FIR⁺, i.e., Foxp3⁺ cells, to test for suppression in an in vitro assay. Foxp3⁻CD25⁻CD4⁺ T cells from FIR-OTII mice were purified by flow cytometry and cultured with CD8⁺ DCs without TGF-β or CD8⁻ DCs with TGF-β. After 5 days of culture, RFP⁺ CD25⁺ and RFP⁻ CD25⁻CD4⁺ T cells were purified by flow cytometry as in Fig. 4A and added to an in vitro suppression assay comprised of CFSE-labeled CD25⁻ CD4⁺ responder T cells stimulated with spleen APCs and anti-CD3 mAb (Fig. 4B,C). The proliferation of CD25⁻ CD4⁺ T cells, as indicated by CFSE dilution as well as live responder cell counts, was suppressed by the RFP⁺ CD25⁺ CD4⁺ T cells induced by CD8⁺ DCs (Fig. 4B,C). In contrast, the response of the CFSE-labeled CD25⁻ CD4⁺ cells was not inhibited by the addition of RFP⁻ CD25⁻CD4⁺ T cells from the same culture (Fig. 4B,C). Similarly, the proliferation of CD25⁻ CD4⁺ T cells was suppressed by the RFP⁺ CD25⁺CD4⁺ T cells induced by spleen CD8⁻ DCs with TGF-β (Fig. 4B,C). Therefore, the T reg induced by spleen DC subsets have suppressive function.

Fig. 4. Suppressive function of Foxp3⁺ CD25⁺ CD4⁺ T reginduced by CD8⁺ or CD8⁻DCs.

(A) Foxp3⁻ CD25⁻ CD4⁺ T cells (2×10⁴) from FIR-OTII mice were cultured with spleen CD8⁺ DCs (2×10⁴) in the presence of peptide (0.03–0.1 μg/ml) without TGF-β. After 5–6 days, the induced Foxp3⁺ CD25⁺ CD4⁺ T reg and Foxp3⁻ CD25⁻ CD4⁺ T cells were purified by flow cytometry. Similarly, Foxp3⁻ CD25⁻ CD4⁺ T cells from FIR-OTII mice were also cultured with spleen CD8⁻ DCs in the presence of peptide (0.03–0.1 μg/ml) with TGF-β (2 ng/ml), and the induced Foxp3⁺ CD25⁺ CD4⁺ T reg were purified by flow cytometery. The square in pre-sort indicates the gate for sorting. The purity of sorted cells is also shown (post-sort).

(B) For the suppression assay, CD25⁻ CD4⁺ responder T cells (1×10⁴) from B6 mice were CFSE-labeled and stimulated with spleen APCs (5×10⁴) and anti-CD3 mAb. To these, the suppressors (1×10⁴) as above were added. After 3 days, CFSE-dilution was analyzed by flow cytometry. Dead cells were gated out by TOPRO-3 iodide. These data are representative of 3 independent experiments.

(C) As in (B), but the number of live CFSE⁺ CD25⁻CD4⁺ responder cells per culture was shown. The indicated suppressors were added in the culture at the indicated ratio. This data is representative of 2 independent experiments.

Proliferation of differentiating T reg upon stimulation with spleen DC subsets

To investigate if cell proliferation took place during the induction of T reg, CD4⁺ T cells were isolated from DO11.10 RAG^−/− mice, CFSE-labeled and cultured with spleen CD8⁺ DCs without exogenous TGF-β, or CD8⁻ DCs with exogenous TGF-β. In both instances, the newly formed Foxp3⁺ T reg were almost entirely derived from T cells that had undergone 2–5 cycles of cell division (Fig. 5, see arrows). This result indicates that proliferating cells are induced to express Foxp3 upon antigen stimulation with DCs.

Fig. 5. T reg proliferate during differentiation by DC subsets.

Foxp3⁻ CD25⁻CD4⁺ T cells from DO11.10 RAG^−/− mice (2×10⁴) were CFSE labeled and cultured with CD8⁺ DCs (2×10⁴) without TGF-β (2 ng/ml) or CD8⁻ DCs (2×10⁴) with TGF-β in the presence or absence of OVA peptide (0.03 μg/ml). At day 5, cells were stained with anti-Foxp3 mAb. Foxp3 expression and CFSE dilution are shown gated on live CD11c⁻ CD4⁺ T cells. This data is representative of 3 independent experiments.

OVA delivery to the CD8⁺ CD205⁺ DC subset in vivo selectively differentiates Foxp3⁺ T reg

Antigen delivery via DEC-205 mAb can induce Foxp3⁺ T reg from Foxp3⁻ T cells (48), but it is not known if antigen delivery to the other DC subset via 33D1 mAb can also induce Foxp3. To compare the induction of T reg by the two major DC subsets in vivo, we adoptively transferred Foxp3⁻ CD25⁻ CD4⁺ T cells from Thy1.2⁺ DO11.10 RAG^−/− transgenic mice into Thy1.1⁺ BALB/c mice and stimulated them with OVA delivered by mAbs to DEC-205 (DEC-OVA) or DCIR2 (33D1-OVA) (35). The cells from DO11.10 RAG ^−/− transgenic mice were also CFSE-labeled prior to transfer to follow cell proliferation in response to DEC-OVA or 33D1-OVA (Fig. 6A). 3 days later, 33D1-OVA induced more effective proliferation of transferred Thy1.2⁺ CD4⁺ T cells than DEC-OVA (Fig. 6B). When expression of Foxp3 and CFSE dilution were assessed on these cells at day 3, proliferating cells expressed Foxp3 as indicated by arrows, although 33D1-OVA induced a greater proportion of proliferating Foxp3⁻ cells (Fig. 6B). 10–14 days later, the numbers of recovered transferred Thy1.2⁺CD4⁺ T cells showed a sizeable reduction following the 3 μg dose of the antibody-OVA proteins, suggesting T cell deletion as a result of stimulation with either DEC-OVA or 33D1-OVA (Fig. 6C) as previously reported (21, 35, 49). When Foxp3 expression and CFSE-dilution were evaluated on the remaining Thy1.2⁺ T cells, Foxp3 was induced on undivided cells, as well as T cells that had undergone cell division, but mainly in response to DEC-OVA (3 μg) rather than 33D1-OVA (Fig. 6C). Nevertheless, 33D1-OVA induced strong proliferation of CD4⁺ T cells by the cells that did not express Foxp3 (Fig. 6C). When a control isotype-matched mAb fused with OVA (Iso-OVA) (35) was used, DO11.10 T cells did not induce Foxp3 (Fig. 6D). The percentages and numbers of the induced Foxp3⁺ T cells were higher with 3 μg of DEC-OVA than all tested doses of 33D1-OVA (Fig. 6E), even though the CD8⁻ DCIR2⁺ DC subset more rapidly produces peptide-MHC II complexes (35). Thus, antigen delivery to CD8⁺ CD205⁺ DCs in vivo selectively induced more Foxp3⁺ T reg from Foxp3⁻ precursors.

Fig. 6. DC targeting of OVA via DEC-205 leads to the development of Foxp3⁺ CD25⁺ CD4⁺ T cells in vivo.

(A) Thy1.2⁺ DO11.10 RAG^−/− mice lack Foxp3⁺ CD25⁺ CD4⁺ T cells as shown in the plots. The Foxp3⁻ CD25⁻ CD4⁺ T cells were CFSE-labeled (2×10⁶) and injected i.v. into Thy1.1⁺ BALB/c mice. The recipients were stimulated with indicated doses of DEC-OVA or 33D1-OVA i.p. one day later. After 3 or 10–14 days, spleen and lymph node Thy1.2⁺ cells were analyzed by FACS.

(B) As in (A), but the frequencies of donor Thy1.2⁺ DO11.10 RAG^−/−CD4⁺ T cells within gates at day 3 are shown in the top row. Foxp3, isotype control staining and CFSE dilution gated on Thy1.2⁺ transferred DO11.10 RAG^−/−CD4⁺ T cells are shown in the middle and bottom rows. This data is representative of 2 independent experiments.

(C) As in (A), but the frequencies of donor Thy1.2⁺ DO11.10 RAG^−/− CD4⁺ T cells within gates at day 13 are shown in the top row. Foxp3, isotype control staining and CFSE dilution gated on the Thy1.2⁺ transferred DO11.10 RAG^−/−CD4⁺ T cells are shown in the middle and bottom rows.

(D) As in (A), but recipient mice were stimulated with PBS, 3 μg DEC-OVA or Iso-OVA i.p. one day later. After 13 days, a mixture of spleen and lymph nodes was analyzed by FACS. Foxp3 (left), isotype control staining (right) and CFSE dilution gated on the Thy1.2⁺ transferred DO11.10 RAG^−/−CD4⁺ T cells are shown. Data are representative of 3 independent experiments.

(E) Cell recoveries of Foxp3⁺ Thy1.2⁺ T cells from transferred DO11.10 RAG^−/− CD4⁺ T cells at day10–14 are shown as percentages per Thy1.2⁺ T cells (left) and absolute numbers from each mouse (right). A summary of 5 separate experiments where each data point is a separate experiment. P value is from student t-test.

OVA delivery to the CD8⁻ DCIR2⁺ DC subset in vivo better expands natural Foxp3⁺ T reg

It is not known how already-existing natural Foxp3⁺ T reg behave after the targeting of antigens to DC-subsets in vivo. To address this question, we compared the proliferation of Foxp3⁺ T reg by DEC-OVA and 33D1-OVA in vivo (35). Natural Foxp3⁺ T reg from Thy1.2⁺ DO11.10 RAG^+/+ OVA CD4 transgenic mice were purified by flow cytometry, and expanded with DCs, peptide and IL-2 for 1 week (25). The expanded natural T reg (>90% Foxp3⁺) were CFSE-labeled, and adoptively transferred into Thy1.1⁺ BALB/c recipients one day before injection with DEC-OVA or 33D1-OVA (Fig. 7A). After 3 d and 10–14 d, mixtures of spleen and lymph node cells were evaluated for proliferation of Foxp3⁺ T reg. We used DC-expanded Foxp3⁺ T reg for in vivo transfer because of the larger number of cells after in vitro expansion. At day 3, more transferred Thy1.2⁺ DO11 Foxp3⁺ T reg were observed after 3 μg of 33D1-OVA than 3 μg of DEC-OVA (Fig. 7B). At day 10, the T reg stimulated by both Abs were deleted compared to day 3; however, when antigen was delivered by 33D1, the transferred Thy1.2⁺ DO11 T reg were more numerous (Fig. 7B). When the expression of Foxp3 and CFSE dilution were investigated, antigen delivery by 33D1, in contrast to antigen delivery by DEC-205, induced >3 cell divisions, and importantly, the divided T reg maintained high expression of Foxp3 (Fig. 7B). When iso-OVA was used, neither the DC-expanded T reg nor freshly isolated DO11.10 T reg divided at day 3 as in PBS controls (freshly isolated T reg are shown in Fig. 7C). Also the absolute numbers of Thy1.2⁺ Foxp3⁺ T reg at day 10–14 were higher after stimulation with 3 μg of 33D1-OVA (Fig. 7D). Thus, antigen delivery to CD8⁻ DCIR2⁺ DCs, relative to CD8⁺ CD205⁺ DCs in vivo, results in better stimulation of natural Foxp3⁺ T reg.

Fig. 7. DC targeting of OVA via 33D1 better sustains naturally occurring Foxp3⁺ CD25⁺CD4⁺ T cells in vivo.

(A) CD25⁺ CD4⁺ Foxp3⁺ T cells from Thy1.2⁺ DO11.10 transgenic mice were cultured with mature DCs, peptide and IL-2. After 7 days, the expanded T reg were > 90% CD25⁺Foxp3⁺. The expanded T reg (0.5~1×10⁶) were CFSE-labeled and adoptively transferred into Thy1.1⁺ BALB/c mice. One day later, the mice were i.p. injected with 3 μg of 33D1-OVA, DEC-OVA Abs or PBS. Spleen and lymph node cells were mixed and analyzed by FACS at day 3 or 14.

(B) Mixtures of spleen and lymph node cells at day 3 or 10 after injection of 33D1-OVA, DEC-OVA Abs or PBS were stained with anti-CD4 and anti-Thy1.2 (Top). Foxp3, isotype control staining and CFSE dilution gated on Thy1.2⁺ transferred DO11.10 T reg are shown in the middle and bottom rows.

(C) As in (A), but freshly isolated CFSE-labeled Thy1.2 DO11.10 CD25⁺ CD4⁺ T reg were transferred into Thy1.1⁺ BALB/c mice. One day later, the mice were i.p. injected with 3 μg of 33D1-OVA, DEC-OVA Abs, Iso-OVA or PBS. Spleen and lymph node cells were mixed and analyzed FACS at day 3. Data are representative of 2 independent experiments

(D) As in (A), but cell recovery of Thy1.2⁺ transferred DO11.10 T reg at day 10–14 is shown. A summary of 5 separate experiments. P value is from student t-test.

Discussion

CD103⁺ DCs in the intestinal environment are specialized to differentiate Foxp3⁺ T reg and this contributes to the induction of oral tolerance in the gut (40, 41). However, self-tolerance to tissue-specific antigens also needs to be induced elsewhere in the body (18). We wondered whether DCs in systemic lymphoid organs could be involved in inducing T reg in the periphery. We therefore investigated the capacity of two major spleen DC subsets to differentiate Foxp3⁺ T reg from Foxp3⁻ precursors. We found that CD8⁺ CD205⁺ DCs were selectively active in vitro and in vivo. Therefore, CD8⁺ CD205⁺ DCs from the spleen can differentiate Foxp3⁺ regulatory T cells from Foxp3⁻ precursors in the periphery, as CD103⁺ DCs can induce Foxp3 in the intestinal environment.

Mechanistic studies indicated that spleen CD8⁺ CD205⁺ DCs selectively produced the requisite TGF-β required for T reg differentiation. Interestingly, CD8⁺ CD205⁺ DCs also are specialized to take up dying cells (50, 51). Presentation of allogeneic apoptotic cells by CD8⁺ DCs resulted in deletion of alloreactive T cells and induction of Foxp3⁺ T reg (52). Apoptotic cell death is associated with secretion of anti-inflammatory cytokines such as TGF-β (53). Therefore, it is possible that CD8⁺ CD205⁺ DCs in the steady state take up apoptotic cells and this triggers endogenous TGF-β secretion.

CD103⁺ DCs in the intestinal environment can differentiate Foxp3⁺ T reg with endogenous TGF-β (40). TGF-β is a cytokine that is highly expressed in mucosal tissues (54, 55). CD103⁺ DCs in the intestine have retinal dehydrogenase to convert Vitamin A or retinal into retinoic acid, which is an essential co-factor for the induction of Foxp3⁺ T reg by TGF-β (40, 41) and for the induction of gut tropism for T cells and IgA secreting B cells (56, 57). Lamina propria macrophages also play a role in inducing Foxp3⁺ T reg, although TGF-β needs to be added to the inductive cultures (58). These findings suggest that the intestinal environment is specialized for inducing Foxp3⁺ T reg for intestinal antigens.

When we examined the induction of Foxp3⁺ T reg by DEC-targeting in lymph nodes, we found that DEC-OVA induced Foxp3 in lymph nodes (data not shown). CD8⁺ CD205⁺ DCs from lymph nodes also induced Foxp3⁺ T reg in vitro (Juliana Idoyaga, SY, RMS, unpublished data). However, CD8⁻ DCs from lymph nodes appear more complex than spleen DCs because they can be divided into more subsets (12, 33). Other types of DCs might participate in inducing Foxp3⁺ T reg. More mature DCs in the human thymus, following stimulation by thymic stromal lymphopoietin (TSLP), are responsible for generating natural Foxp3⁺ T reg from CD8⁻ CD4⁺ CD25⁻ human thymocytes (59). Also, plasmacytoid DCs are required for antigen-specific regulatory tolerance following donor splenocyte transfusion plus anti-CD40 ligand mAb (60).

CD205⁺ DCs are able to differentiate Foxp3⁺ T reg in vivo following targeted delivery of antigen within mAb to DEC-205 (48, 61), but it has been uncertain if Foxp3⁺ T reg are induced only by antigen delivery to CD205⁺ DCs. We found that antigen delivery to CD8⁻ DCIR2⁺ DCs in vivo is much less efficient for inducing Foxp3⁺ T reg from Foxp3⁻ precursors. In vitro blocking assays, microarray and ELISA results indicate that CD8⁺ CD205⁺ DCs use endogenous TGF-β, in contrast to CD8⁻ DCIR2⁺ DCs, to induce Foxp3. Kretschmer et al reported that Foxp3⁺⁻T reg were more efficiently induced with low doses of anti-DEC-205-HA peptide fusion mAb, which was a subimmunogenic condition (48). Antigen delivered by the 33D1 mAb to CD8⁻ DCIR2⁺ DCs in the steady state are processed and transferred to the cell surface as MHC-class II peptide complexes more efficiently (35). Therefore, it is also possible that the amounts of MHC-class II peptide complexes after antigen delivery by anti-DEC-205 and 33D1 mAbs were different even though we injected the same amounts of DEC-OVA and 33D1-OVA mAbs.

Our group previously reported that in vivo antigen delivery to CD205⁺ DCs by DEC-205 mAb induced deletion of adoptively transferred transgenic T cells (21, 35, 49). In these reports, we did not investigate the induction of Foxp3⁺ T reg in the remaining T cells, although other groups have shown that T reg could be induced by DEC targeting (48, 61). Although Hawiger et al (23) described that CD25 was not up-regulated on the remaining T cells after DEC targeting, a higher dose of anti-DEC antigen and shorter observation times was used compared to this study, and Foxp3 antibodies were not used to monitor the response. Here we confirmed that in vivo antigen delivery to CD8⁺ CD205⁺ DCs induce Foxp3⁺ T reg on remaining T cells; however, it has not been studied whether T cells stimulated by DEC targeting actively induced apoptosis during the deletion process.

In contrast, antigen delivery to CD8⁻ DCIR2⁺ DCs stimulates natural Foxp3⁺ T reg better than CD8⁺ CD205⁺ DCs in vivo. Natural T reg proliferate in response to their selecting self-peptide in vivo (62–65). CD8⁻ DCIR2⁺ DCs excel in producing MHC class II-peptide complexes compared to CD8⁺ CD205⁺ DCs (35). This might explain the better response by natural Foxp3⁺ CD25⁺CD4⁺ T reg to antigen delivery to CD8⁻ DCIR2⁺ DCs. CD8⁻ DCIR2⁺ DCs might have additional intrinsic differences to stimulate natural Foxp3⁺ T reg relative to CD8⁺ CD205⁺ DCs beyond the preferential processing of MHC class II pathway. For example, the function of natural Foxp3⁺ T reg is maintained by IL-2 (66–71), and IL-2 is known to be produced from DCs stimulated with bacterial products (72). However, we did not see a difference of IL-2 expression by DC subsets in our gene array data (data not shown).

At the mRNA level, our gene array data showed that CD8⁺ CD205⁺ DCs expressed TGF-β and TGF-β binding protein, whereas CD8⁻ DCIR2⁺ DCs expressed TGF-β receptor and the downstream transcription factors, SMAD1 and SMAD4. This suggests that TGF-β from CD8⁺ CD205⁺ DCs can act on Foxp3⁻ precursors to induce Foxp3, in contrast to CD8⁻ DCIR2⁺ DCs, which are ready to respond to TGF-β. DCs express αvβ8 integrins, which activate latent TGF-β, and this process is important for induction and homeostasis of Foxp3⁺ T reg (73). TGF-β could be provided by natural Foxp3⁺ T reg (74–76), and TGF-β regulates the homeostasis and function of Foxp3⁺ T regs (75, 77, 78). T cells from TGF-β receptor II dominant negative mice did not induce Foxp3 after antigen delivery by DEC-205 mAb in vivo (48). Therefore, TGF-β signaling on T cells seems to play a role to induce Foxp3⁺ T reg in the periphery, but it would be important to investigate whether TGF-β signaling on DCs might additionally influence the induction of T reg.

We confirmed that spleen CD8⁺DCs produced higher amounts of TGF-β than spleen CD8⁻ DCs by ELISA (Fig. 3B). DC maturation is a complex subject because of the large number of different maturation stimuli, and their different mechanisms of action. However, we tested one of the stimuli, poly IC, and found that maturing CD8⁺ DCs in poly IC reduced the production of TGF-β. Consistent with the idea that immature DCs are inducing tolerance whereas mature DCs are actively inducing immunity (18), we observed that the frequency of induced Foxp3⁺ T reg was reduced by poly IC matured CD8⁺ DCs (Fig. 3D and E). Further studies are needed to understand how different maturation stimuli affect the capacity of DC to induce T reg in vivo.

The factors that allow DCs to control the different pathways of T cell differentiation need to be identified, since these will permit a better understanding of DC function in vivo and in the design of new immune based treatments. Our findings indicate that endogenous production of TGF-β in the context of a particular DC subset in systemic lymphoid tissues, i.e., CD8⁺ CD205⁺ spleen DCs itself, permits the peripheral induction of Foxp3⁺ T reg even in the steady state. We would like to propose that delivery of antigen to CD8⁺ CD205⁺ DCs provides a means for the induction of therapeutic antigen-specific Foxp3⁺ T reg from Foxp3⁻ T cells, as in autoimmune diseases, allergy or transplant rejection.