A New Subset of CD103⁺CD8α⁺ Dendritic Cells in the Small Intestine Expresses TLR3, TLR7, and TLR9 and Induces Th1 Response and CTL Activity

The gastrointestinal tract is constantly exposed to food proteins and commensal bacteria. Although the intestinal immune system has evolved mechanisms that maintain immunological tolerance to food Ags and commensal organisms, it also recognizes invasive pathogens and induces appropriate protective immune responses to eliminate them. Dendritic cells (DCs) are thought to play a key role in discriminating between commensal microorganisms and potentially harmful pathogens and in maintaining the balance between tolerance and active immunity. DCs in the intestine are present not only in GALT, such as the Peyer’s patches and isolated lymphoid follicles, but also in the lamina propria (LP) (1). CD103⁺ DCs and CX3CR1⁺ DCs are representative DCs in the intestinal LP (LPDCs). Recent reports have shown that CD103⁺ LPDCs and CX3CR1⁺ LPDCs are derived from two distinct DC lineages and that they serve separate immune functions in the intestine (2, 3). CX3CR1⁺ DCs in the LP are known to penetrate epithelial tight junctions to sample luminal bacteria (4, 5). However, CX3CR1⁺ DCs are nonmigratory and display poor T cell stimulatory capacity (6). However, CD103⁺ DCs were shown to migrate from the LP to the mesenteric lymph nodes (MLNs) in a CCR7-dependent manner and to induce Foxp3⁺ regulatory T cells (Tregs) via the dietary metabolite retinoic acid (RA) (7–10). RA-producing CD103⁺ DCs in MLNs induce α4β₇ integrin and CCR9 expression on naive lymphocytes to establish gut tropism (11). Previously, we reported that low-density cells in the LP could be classified into four subsets on the basis of their different CD11c/CD11b expression patterns: CD11c^hiCD11b^lo DCs, CD11c^hiCD11b^hi DCs, CD11c^intCD11b^int macrophages, and CD11c^intCD11b^hi eosinophils (12). The CD11c^hiCD11b^hi subset, which is CD103⁺, expresses TLR5 and TLR9 and produces proinflammatory cytokines such as IL-6 and IL-12 in response to flagellin and CpG oligodeoxynucleotide (ODN) (12, 13). The CD11c^hiCD11b^hi subset specifically expresses mRNA of retinal dehydrogenase isoform 2 (Raldh2), which catalyzes the conversion of retinal to RA (12, 13). The ability to produce RA enables the CD11c^hiCD11b^hi DC subset to induce a flagellin-mediated T cell-independent IgA class-switch recombination of B cells (12, 13). Furthermore, the CD11c^hiCD11b^hi subset promotes Th1 and Th17 cell differentiation in response to TLR ligands (12, 14).

We show that CD103⁺ LPDCs in the small intestine are divided into a small subset of CD8α⁺ cells and a larger subset of CD8α⁻ cells. According to the flow cytometry analysis, CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs were equivalent to CD11c^hiCD11b^lo and CD11c^hiCD11b^hi subsets, respectively. We analyzed the novel subset of CD8α⁺ LPDCs to elucidate their immunological functions. The CD103⁺CD8α⁺ LPDCs expressed TLR3, TLR7, and TLR9 and produced IL-6 and IL-12p40 but not TNF-α, IL-10, or IL-23 in response to their respective TLR ligands. In contrast to CD103⁺CD8α⁻ LPDCs, CD103⁺CD8α⁺ LPDCs did not express the gene encoding the RA-converting enzyme, Raldh2, and were not involved in T cell-independent IgA synthesis and Foxp3⁺ Treg induction. We further analyzed immunogenicity in Ag-loaded CD103⁺CD8α⁺ LPDCs in vivo. CD103⁺CD8α⁺ LPDCs induced Ag-specific IgG in serum, Th1 response, and CTL activity. In contrast, CD103⁺CD8α⁻ LPDCs induced Ag-specific IgG in serum as well as IgA in stool samples and also induced Th1 and Th17 responses and strong CTL activity. All of these results suggest that CD103⁺CD8α⁺ LPDCs have a different function to CD103⁺CD8α⁻ LPDCs in active immunity.

Flagellin and CpG ODN (ODN1668) were purified as previously described (13). Polyinosinic-polycytidylic acid (poly I:C) and R-848 were purchased from Invivogen. Mouse MHC class I (K^b)-binding peptide OVA_257–264 (SIINFEKL) and MHC class II-binding OVA_323–339 (ISQAVHAAHAEINEAAGR) were obtained from Hokkaido System Science. Recombinant human TGF-β1 was purchased from R&D Systems.

C57BL/6 mice were purchased from CLEA Japan. OT-I transgenic (Tg) mice and OT-II Tg mice (C57BL/6) were provided by W.R. Heath (Department of Microbiology and Immunology, University of Melbourne, Parkville, VIC, Australia) (12). All animal experiments were carried out with the approval of the Animal Research Committee of the Research Institute for Microbial Diseases at Osaka University.

All Abs and reagents were obtained from BD Pharmingen unless otherwise stated. Before staining, FcRs were blocked for 15 min at 4°C. LPCs were stained with the following biotinylated mAbs: CD11b (M1/70), CD11c (HL3), CD103 (M290), and CD8α (53-6.7). Cocultured B cells or LP leukocytes were surface-stained with PerCP-Cy5.5–labeled anti-B220 (RA3-6B2). Cells were then fixed and permeabilized with Cytofix/Cytoperm and incubated with biotin-conjugated IgA (C10-1) followed by intracellular staining with allophycocyanin-labeled streptavidin. The surface of cocultured T cells was stained with PerCP-Cy5.5–labeled anti-CD4 (L3T4). Cells were then fixed and permeabilized with Cytofix/Cytoperm before intracellular staining with allophycocyanin-labeled anti-Foxp3 (FJK-16s; eBioscience) (Fig. 4), PE-labeled anti–IFN-γ (XMG1.2), allophycocyanin-labeled anti–IL-17 (TCC11-18H10.1) (Fig. 5B), PE-labeled anti–T-bet (4B10; eBioscience), or PE-labeled anti-retinoic acid-related orphan receptor γT (RORγT) (AFKJS-9; eBioscience) (Supplemental Fig. 1). Data were acquired with an FACSCalibur or FACSCanto II (BD Biosciences) and analyzed using the software FlowJo 8.6 (Tree Star).

Small intestinal segments were treated with PBS containing 10% FCS, 20 mM HEPES, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, 10 mM EDTA, and 10 μg/ml polymyxin B (Calbiochem) for 30 min at 37°C to remove epithelial cells, followed by extensive washing with PBS. Small intestinal segments were digested with 400 Mandl units/ml collagenase D (Roche) and 10 μg/ml DNase I (Roche) in RPMI 1640/10% FCS with continuous stirring at 37°C for 45–90 min. EDTA was added (10 mM final concentration), and the cell suspension was incubated for an additional 5 min at 37°C. Cells were passed through a 17.5% Accudenz (Accurate Chemical & Scientific) solution to enrich the DCs. The obtained cells were incubated with FITC-conjugated anti-CD103, PE-conjugated anti-CD8α, PE-Cy7–conjugated anti-CD11c, and allophycocyanin-Cy7–conjugated anti-CD11b after FcR blocking. CD11c^hi DC subsets were sorted on the basis of their expression of CD103 and CD8α using an FACSAria (BD Biosciences). The purity of the sorted DCs was routinely >95%. Naive CD4⁺ T cells from the spleens of OT-II Tg mice were purified by magnetic sorting using mouse anti-CD4 beads (Miltenyi Biotec). Naive CD8⁺ T cells from the spleens of OT-I Tg mice were purified by magnetic sorting using the CD8a⁺ T Cell Isolation kit II (Miltenyi Biotec). Peritoneal cells from C57BL/6 mice were incubated with FITC-conjugated anti-IgD (11-26c.2a) and PE-Cy7–conjugated anti-IgM (R6-60.2; both BD Pharmingen) after FcR blocking. Naive B cells were sorted on the basis of their expression of IgD and IgM using an FACSAria system (BD Biosciences). The purity of the sorted cells was routinely >95%. EL-4 cells were obtained from American Type Culture Collection (TIB-39).

RNA (1 μg) was reverse-transcribed using Superscript II Reverse Transcriptase (Invitrogen) according to the manufacturer’s instructions with random hexamers as primers. The sequences of primers specific for Tlr2, Tlr3, Tlr4, Tlr5, Tlr7, Tlr9, or Actb have been described previously (12). The primer pairs and Taq polymerase (Takara Shuzo) were used for PCR as follows: 25 cycles at 97°C (30 s), 57°C (30 s), and 72°C (30 s); the products were then separated by agarose gel electrophoresis (12). Quantitative real-time PCR was carried out in a final volume of 25 μl containing the cDNA (amplified as described above), 2× PCR Master Mix (Applied Biosystems), and primers specific for 18S rRNA (Applied Biosystems) as an internal control or primers specific for Aldh1a1, Aldh1a2, or Aldh1a3 (Applied Biosystems), using a 7700 Sequence Detector (Applied Biosystems). After incubation at 95°C for 10 min, products were amplified using 35 cycles of 95°C (15 s), 60°C (60 s), and 50°C (120 s).

The concentrations of IFN-γ, IL-17, IL-4, TNF-α, IL-6, IL-10, and IL-12p40 were measured using the Bio-plex system (Bio-Rad) following the manufacturer’s instructions. The levels of IgA and IL-23 were determined by ELISA (R&D Systems and eBioscience, respectively).

Peritoneal IgM⁺IgD⁺ cells (1 × 10⁵) were cultured in medium supplemented with BAFF (50 ng/ml) together with CD11c⁺CD103⁺CD8α⁺ LPDCs (2 × 10⁴) in the presence or absence of poly I:C (50 μg/ml), R-848 (100 nM), CpG ODN (1 μM), or CD11c⁺CD103⁺CD8α⁻ LPDCs (2 × 10⁴) in the presence or absence of flagellin (1 μg/ml) or CpG ODN (1 μM) for 5 d. After the depletion of DCs by anti-CD11c microbeads (Miltenyi Biotec), cells were analyzed by flow cytometry, and the concentration of IgA in the culture supernatants was determined by ELISA.

OT-II Tg CD4⁺ T cells (1 × 10⁵) were cultured in medium supplemented with TGF-β (1 ng/ml) and OVA protein (10 μg/ml) together with CD11c⁺CD103⁺CD8α⁺ LPDCs (2 × 10⁴) in the presence or absence of poly I:C (50 μg/ml), R-848 (100 nM), CpG ODN (1 μM), or CD11c⁺CD103⁺CD8α⁻ LPDCs (2 × 10⁴) in the presence or absence of flagellin (1 μg/ml) or CpG ODN (1 μM) for 4 d. After the depletion of DCs by anti-CD11c microbeads (Miltenyi Biotec), CD4⁺ cells expressing Foxp3 were analyzed by flow cytometry.

OT-II Tg CD4⁺ T cells (1 × 10⁵) were cultured in medium supplemented with OVA protein (10 μg/ml) together with CD11c⁺CD103⁺CD8α⁺ LPDCs (2 × 10⁴) in the presence or absence of poly I:C (50 μg/ml), R-848 (100 nM), CpG ODN (1 μM), or CD11c⁺CD103⁺CD8α⁻ LPDCs (2 × 10⁴) in the presence or absence of flagellin (1 μg/ml) or CpG ODN (1 μM) for 4 d. The concentrations of IL-17, IFN-γ, and IL-4 in the culture supernatants were determined using the Bio-plex system (Bio-Rad). After the depletion of DCs by anti-CD11c microbeads (Miltenyi Biotec), CD4⁺ cells expressing T-bet and RORγT were analyzed by flow cytometry. Cocultured cells were restimulated for 4 h with PMA (50 ng/ml; Sigma-Aldrich) and ionomycin (500 ng/ml; Calbiochem) in the presence of GolgiStop (BD Pharmingen) after the depletion of DCs by anti-CD11c microbeads. Then, CD4⁺ cells producing IL-17 and IFN-γ were analyzed by flow cytometry.

OT-I Tg CD8⁺ T cells (1 × 10⁵) were cultured in triplicate with the indicated numbers of CD11c⁺CD103⁺CD8α⁺ LPDCs in the presence or absence of poly I:C (50 μg/ml), R-848 (100 nM), CpG ODN (1 μM), or CD11c⁺CD103⁺CD8α⁻ LPDCs (2 × 10⁴) in the presence or absence of flagellin (1 μg/ml) or CpG ODN (1 μM) in medium supplemented with OVA protein (10 μg/ml) for 96 h. Following this, 1 mCi [³H]thymidine (Amersham Biosciences) was pulsed for the last 8 h, and then [³H] uptake was measured in a scintillation counter (Packard Instruments).

OT-I Tg CD8⁺ T cells (1 × 10⁶) were cultured in medium supplemented with OVA protein (10 μg/ml) together with CD11c⁺CD103⁺CD8α⁺ LPDCs (1 × 10⁵) in the presence or absence of poly I:C (50 μg/ml), R-848 (100 nM), CpG ODN (1 μM), or CD11c⁺CD103⁺CD8α⁻ LPDCs (1 × 10⁵) in the presence or absence of flagellin (1 μg/ml) or CpG ODN (1 μM). On day 4 of culture, 10 U/ml recombinant human IL-2 (rhIL-2; R&D Systems) was added, and activated cells were further expanded in the presence of rhIL-2. On day 7 of culture, activated OT-I CTL were used as effectors in in vitro cytotoxicity assay.

EL-4 cells were pulsed with or without 10 μg/ml OVA_257–264 peptide. After 60 min, targets were washed with 2% FCS/PBS to remove excess peptide and resuspended in PBS. Targets were labeled with 0.1 μM CFSE (Sigma-Aldrich) for 10 min at room temperature and washed twice. Targets were then plated out at 10⁵ cells/well in a 96-well U-bottom plate in a 100 μl volume. Media alone or 2 × 10⁶ cells with effectors (activated OT-I CTL) were added and incubated for 5 h at 37°C/5% CO₂. After incubation, cells were stained with Kusabira-Orange–conjugated anti-Annexin V Ab for 20 min. Cells were then analyzed by flow cytometry. A gate was set on forward scatter versus CFSE to exclude effectors and measure cell death and apoptosis on targets. The percentage-specific lysis was calculated using the following formula: cytotoxicity (%) = ([ET − T₀]/[100 − T₀]) × 100, in which ET was the percentage of Annexin V⁺ cells in CFSE⁺ cells after the culture of targets and effectors, and T₀ was the percentage of Annexin V⁺ cells in CFSE⁺ cells after the culture of targets alone.

CD11c⁺CD103⁺CD8α⁺ LPDCs or CD11c⁺CD103⁺CD8α⁻ LPDCs were cultured for 12 h with OVA protein (100 μg/ml) in the presence of CpG ODN (1 μM). Ag-loading cells (5 × 10⁴ per mouse) were injected on days 0 and 14 into the peritoneal cavities of naive Tlr9^−/− mice; control mice were treated with PBS.

One week after the final immunization, serum and fecal extracts were obtained from the immunized and control mice. For determination of OVA-specific IgG titers in the sera or IgA titers in the fecal extracts, 96-well microtiter plates (Corning) were coated overnight at 4°C with 200 μg/ml OVA in 50 mM carbonate buffer (pH 9.6). Plates were blocked with PBS containing 1% fraction V BSA (Serologicals Proteins) at 37°C for 1 h. Test sera were serially diluted with PBS containing 0.2% BSA and 0.02% Tween 20 and incubated at 37°C for 2 h. After incubation, HRP-conjugated goat anti-mouse IgG or IgA (Zymed), diluted 1:2000 in PBS containing 0.2% BSA and 0.02% Tween 20, was added and incubated at 37°C for 1 h. Wells were washed five times with PBS containing 0.02% Tween 20 between each step. Plates were developed at room temperature following the addition of o-phenylendiamine (0.4 mg/ml) and hydrogen peroxide (0.012%) in 7 mM citrate buffer (pH 5). Finally, 1 M H₂SO₄ was added, and absorbance was measured at 490 nm with a microplate reader (model 550; Bio-Rad). Preimmunized serum was used as a negative control. The average extinction in negative control wells, to which three times the SD was added, provided the reference point for determination of the titer in the test sera. Ab titers were expressed as the reciprocal of the last dilution yielding an extinction value higher than the reference value.

One week after the final immunization, splenocytes were collected from the immunized and control mice and were cultured for 4 d with the OVA_257–264 peptide (10 μg/ml) or OVA_323–339 peptide (10 μg/ml). The concentration of IFN-γ (OVA_257–264 and OVA_323–339), IL-17 (OVA_323–339), and IL-4 (OVA_323–339) in the culture supernatants was measured using the Bio-plex system (Bio-Rad).

C57BL/6 splenocytes were labeled with either 0.5 or 5 μM CFSE for 15 min at room temperature and washed twice. CFSE^bright cells (M2) were subsequently pulsed with 0.5 μg/ml OVA_257–264 for 90 min at 37°C. The CFSE^dull cells (M1) remained unpulsed. Cells were mixed at a 1:1 ratio, and then a total of 5 × 10⁶ cells was injected i.v. into the immunized or control mice.

Statistical significance was evaluated using an unpaired two-tailed Student t test in all experiments. A p value <0.05 was considered significant.

DCs comprise a heterogenous population of APCs, which include plasmacytoid DCs and CD11c^hi conventional DCs (cDCs) (15). Mouse cDCs in lymphoid tissues include CD8α⁺ and CD8α⁻ subsets that have distinct functional properties. CD103⁺ DCs are a major cDC subset in the intestinal LP (7). Similar to cDCs in lymphoid tissues, we found that CD11c⁺CD103⁺ cells in the small intestinal LP were divided into CD8α⁺ and CD8α⁻ subsets (Fig. 1A). Our previous report showed that CD11c⁺ LPCs in the small intestine consisted of four subsets: CD11c^hiCD11b^lo DCs, CD11c^hiCD11b^hi DCs, CD11c^intCD11b^int macrophages, and CD11c^intCD11b^mid eosinophils (12). Flow cytometry analysis indicated that CD11c⁺CD103⁺CD8α⁺ cells converged on the CD11c^hiCD11b^lo subset, whereas CD11c⁺CD103⁺CD8α⁻ cells were predominantly in the CD11c^hiCD11b^hi subset (Fig. 1A). Although both CD11c^hiCD11b^lo and CD11c^hiCD11b^hi DC subsets are CD103⁺ in the intestinal LP (Fig. 1B) (12), only CD11c^hiCD11b^lo LPDCs expressed CD8α (Fig. 1C). These findings suggest that CD103⁺ LPDCs are divided into CD8α⁺ and CD8α⁻ subsets, which were equivalent to CD11c^hiCD11b^lo and CD11c^hiCD11b^hi LPDCs, respectively.

We previously showed that CD11c^hiCD11b^hi LPDCs specifically express TLR5 and have unique properties to mediate innate and acquired immune responses induced by flagellin stimulation (12). To examine the immunological function of the newly identified CD103⁺CD8α⁺ LPDC subset, we isolated CD11c^hiCD103⁺CD8α⁺ cells by FACS sorting. CD11c^hiCD103⁺CD8α⁻ cells (CD103⁺CD8α⁻ LPDCs) were used as controls for the following experiments. We first checked the expression patterns of TLR family members. Whereas CD103⁺CD8α⁻ LPDCs expressed TLR5 and TLR9, CD103⁺CD8α⁺ LPDCs expressed TLR3, TLR7, and TLR9 (Fig. 2A). As shown in our previous study, the CD11c^hiCD11b^hi subset, which correspond to CD103⁺CD8α⁻ LPDCs, produced IL-6 and IL-12 p40 in response to flagellin and CpG ODN (12). We stimulated CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs with their respective TLR ligands and checked proinflammatory cytokine production (Fig. 2B). Consistent with the previous results, CD103⁺CD8α⁻ LPDCs produced IL-6 and IL-12 p40 in response to flagellin and CpG ODN. Similarly, CD103⁺CD8α⁺ LPDCs produced IL-6 and IL-12p40 in response to their respective ligands, poly I:C, R-848, and CpG ODN. However, TNF-α, IL-23, or IL-10 were not induced in either cell subset.

We next investigated the role of CD103⁺CD8α⁺ LPDCs in the activation of adaptive immune responses. We previously showed that the CD11c^hiCD11b^hi subset in the LP specifically expresses Raldh2 and is involved in T cell-independent IgA class-switch recombination of B cells (12, 13). We examined the involvement of CD103⁺CD8α⁺ LPDCs in IgA synthesis. We first checked whether CD103⁺CD8α⁺ LPDCs synthesize RA and found that whereas CD103⁺CD8α⁻ LPDCs specifically expressed Raldh2, no expression of RALDH isoforms was detected in CD103⁺CD8α⁺ LPDCs (Fig. 3A). Thus, RA-producing CD103⁺ DCs in the LP belong to the CD8α⁻ subset (CD11c^hiCD11b^hi) and not the CD8α⁺ subset (CD11c^hiCD11b^lo). We further examined whether CD103⁺CD8α⁺ LPDCs could induce T cell-independent IgA class switching, demonstrating that although flagellin- and CpG ODN-stimulated CD103⁺CD8α⁻ LPDCs efficiently induced the differentiation of B220⁻IgA⁺ plasma cells, stimulation of CD103⁺CD8α⁺ LPDCs via TLR3, TLR7, or TLR9 did not (Fig. 3B). Consistent with the FACS data, IgA production was not detected in the supernatants of naive B cells cocultured with CD103⁺CD8α⁺ LPDCs and stimulated with TLR ligands (Fig. 3C). Taken together, these findings suggest that CD103⁺CD8α⁺ LPDCs are not able to induce T cell-independent differentiation of IgA⁺ plasma cells.

Previous reports have shown that CD103⁺ DCs isolated from the LP or from the MLNs promote the differentiation of Foxp3⁺ Tregs, an activity dependent on RA and TGF-β (8, 10). We examined whether CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs could induce Treg conversion. Naive OVA-specific OT-II Tg CD4⁺ T cells were cocultured with CD103⁺CD8α⁺ or CD103⁺CD8α⁻ LPDCs together with or without the indicated TLR ligand in the presence of OVA protein and TGF-β (Fig. 4). CD103⁺CD8α⁻ LPDCs could not induce the conversion of Foxp3⁺ Tregs regardless of TLR ligand stimulation. In contrast, CD103⁺CD8α⁻ LPDCs, which express Raldh2, promoted the differentiation of Foxp3⁺ Tregs. Furthermore, stimulation with both flagellin and CpG ODN reduced Foxp3⁺ Treg induction. However, even CD103⁺CD8α⁻ LPDCs could not induce Foxp3⁺ Tregs in the absence of TGF-β (data not shown). Thus, CD103⁺CD8α⁻ but not CD103⁺CD8α⁺ LPDCs effectively induced Foxp3⁺ Tregs in the presence of TGF-β in the absence of TLR activation.

We next assessed the ability of CD103⁺CD8α⁺ LPDCs to induce Ag-specific Th cell differentiation in OVA-specific OT-II Tg CD4⁺ T cells. As shown in our previous study, CD11c^hiCD11b^hi LPDCs induced Th1 and Th17 cells in response to flagellin and CpG ODN (12). We could detect both IFN-γ– and IL-17–producing cells in cocultures of OT-II CD4⁺ T cells and CD103⁺CD8α⁻ LPDCs stimulated by flagellin and CpG ODN (Fig. 5A). However, only IFN-γ–producing cells were detected in cocultures of OT-II CD4⁺ T cells and CD103⁺CD8α⁺ LPDCs stimulated by poly I:C, R-848, and CpG ODN. Both CD103⁺CD8α⁺ LPDCs stimulated by poly I:C, R-848, and CpG ODN and CD103⁺CD8α⁻ LPDCs stimulated by flagellin and CpG ODN induced high expression of T-bet, a lineage-determining factor for Th1 cells in naive CD4⁺ T cells (Supplemental Fig. 1A). Furthermore, expression of RORγT, the key transcription factor in Th17 cells was upregulated in naive CD4⁺ T cells together with CD103⁺CD8α⁻ LPDCs stimulated by flagellin and CpG ODN (Supplemental Fig. 1B). Consistent with the flow cytometry analysis, both IFN-γ and IL-17 were significantly induced in cocultures with CD103⁺CD8α⁻ LPDCs stimulated by flagellin and CpG ODN (Fig. 5B). In contrast, only IFN-γ was significantly induced in cocultures with CD103⁺CD8α⁺ LPDCs stimulated by poly I:C, R-848, and CpG ODN. Taken together, these findings suggest that activated CD103⁺CD8α⁺ LPDCs induced phenotypic and functional Th1 cells but not Th17 cells in vitro.

Soluble and cell-associated Ags can be presented to CD8⁺ T cells in an MHC class I-restricted manner (16), a process referred to as cross-presentation. It is believed that CD8α⁺ DCs but not CD8α⁻ DCs are responsible for cross-presentation in the spleen (17). To examine whether CD103⁺CD8α⁺ LPDCs contribute to the presentation of Ag to CD8⁺ T cells in vitro, we checked their ability to stimulate the proliferation of OVA-specific CD8⁺ T cells from OT-I Tg mice (Fig. 6A). Although the proliferation of CD8⁺ T cells was low when they were cocultured with CD103⁺CD8α⁺ LPDCs, treatment of the DCs with poly I:C, R-848, and CpG ODN strongly stimulated CD8⁺ T cell proliferation. Similarly, flagellin- and CpG ODN-stimulated CD103⁺CD8α⁻ LPDCs potently induced the proliferation of CD8⁺ T cells. Thus, both CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs activated by TLR ligands were capable of stimulating proliferation of CD8⁺ T cells in vitro. We next examined the cytotoxic activity of OT-I CTL induced by CD103⁺CD8α⁺ LPDCs and CD103⁺CD8α⁻ LPDCs in vitro. OT-I CTL induced by CD103⁺CD8α⁺ LPDCs specifically killed the OVA_257–264 peptide-pulsed target cells (Fig. 6B). Treatment of the CD103⁺CD8α⁺ LPDCs with poly I:C, R-848, and CpG ODN further increased cytotoxicity of OT-I CTL. Similarly, OT-I CTL induced by CD103⁺CD8α⁻ LPDCs specifically killed the targets pulsed with OVA_257–264 peptide. Flagellin and CpG ODN treatment enhanced cytotoxic activity in CD8⁺ T cells. Taken together, both CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs activated by TLR ligands strongly induced CTL in vitro.

We finally compared the adaptive immune responses in mice immunized with Ag-loaded CD103⁺CD8α⁺ LPDCs. We isolated CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs from C57BL/6 mice and cultured them with OVA protein (100 μg/ml) overnight. For optimum activation of Ag-loaded DCs, we added CpG DNA (1 μM), a mutual TLR ligand for CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs, to the culture media. Ag-loaded cells (1 × 10⁵) were injected into the peritoneal cavities of Tlr9^−/− mice on days 0 and 14, whereas control mice were treated with PBS. One week after the second injection, the DC vaccination-induced Ag-specific B cell responses were analyzed. Groups of mice immunized with both CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs showed significantly higher levels of serum OVA-specific IgG Ab (Fig. 7A). In addition, high titers of OVA-specific IgA were detected in fecal extracts from CD103⁺CD8α⁻ LPDC-immunized mice. These results indicated that CD103⁺CD8α⁺ LPDCs induced Ag-specific IgG in sera, whereas CD103⁺CD8α⁻ LPDCs induced Ag-specific IgG in sera as well as Ag-specific IgA in the intestinal compartments.

We next examined the Th cell responses of mice immunized with Ag-loaded LPDCs. One week after the second injection, spleen cells from the immunized mice were cultured in the presence of OVA class II peptide (OVA_323–339). As shown in our previous study, we detected OVA class II peptide-specific IFN-γ production as well as IL-17 production after injection of CD103⁺CD8α⁻ LPDCs (Fig. 7B) (12). In contrast, only IFN-γ was produced by splenocytes from mice injected with CD103⁺CD8α⁻ LPDCs. These results suggested that CD103⁺CD8α⁺ LPDCs induced Th1 response but not Th17 response in vivo.

To further examine the Ag-specific CD8⁺ T cell response, spleen cells from immunized mice were cultured in the presence of OVA class I peptide (OVA_257–264), and IFN-γ levels were measured. We detected OVA class I peptide-specific IFN-γ production after injection of both CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs (Fig. 7C). However, the class I-restricted IFN-γ level was much lower after CD103⁺CD8α⁺ LPDC immunization than after CD103⁺CD8α⁻ LPDC immunization, despite the injection of the same numbers of Ag-loaded DCs stimulated with the same TLR ligand, CpG ODN. We then evaluated the cytotoxic activity of the Ag-specific CD8⁺ T cells after CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDC immunization. To detect potential OVA-specific cytotoxicity of CTLs in immunized mice, we used an in vivo cytotoxicity assay consisting of an i.v. infusion of OVA_257–264 peptide-pulsed CFSE^bright target cells and nonpulsed CFSE^dull target cells followed by subsequent ex vivo quantification of the remaining OVA_257–264 peptide-pulsed CFSE^bright target cells in spleen cell suspensions by flow cytometry. Partial lysis of CFSE^bright target cells was observed in the mice immunized with CD103⁺CD8α⁺ LPDCs, whereas CFSE^bright target cells were effectively lysed in mice immunized with CD103⁺CD8α⁻ LPDCs (Fig. 7D), correlating with the OVA_257–264 peptide-specific IFN-γ production findings.

CD103⁺ DCs are the major population of APCs present in the intestinal LP. They have been thought of as a monosubset of DCs and were originally identified through their induction of Foxp3⁺ Tregs via their derived RA (8, 10). Therefore, CD103⁺ LPDCs have been considered to be regulators of intestinal immunity, which induce tolerance and maintain intestinal homeostasis. In this study, we found that CD103⁺ LPDCs in the small intestine are divided into distinct subsets: a small CD8α⁺ subset and a large CD8α⁻ subset. Our previous study showed that there existed two subsets of DCs in intestinal LP: CD11c^hiCD11b^low and CD11c^hiCD11b^hi (12). The CD11c^hiCD11b^low and CD11c^hiCD11b^hi subsets had a DEC-205⁺, MHC class II high, CD80⁺CD86⁺ surface phenotype. In addition, the CD11c^hiCD11b^hi subset is moderately F4/80 positive, suggesting that this subset expresses both DC (DEC-205) and macrophage (F4/80) markers (12). According to the flow cytometry analysis, CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs were equivalent to CD11c^hiCD11b^lo and CD11c^hiCD11b^hi subsets, respectively (Fig. 1A).

In this study, we analyzed the immunological function of the newly identified CD8α⁺ conventional DCs in the small intestinal LP for the first time, to our knowledge. CD103⁺CD8α⁺ LPDCs showed different TLR family member expression profiles from CD103⁺CD8α⁻ LPDCs. Whereas CD103⁺CD8α⁻ LPDCs expressed TLR5 and TLR9, CD103⁺CD8α⁺ LPDCs expressed TLR3, TLR7, and TLR9, which recognize dsRNA, ssRNA, and CpG ODN, respectively (Fig. 2A). However, both LPDC types produced proinflammatory cytokines, such as IL-6 and IL-12p40, but not the anti-inflammatory cytokine IL-10, in response to their respective TLR ligands, suggesting that CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs induce inflammatory responses after TLR ligand stimulation (Fig. 2B).

Both DC subsets were CD103⁺; however, only CD8α⁻ LPDCs expressed the mRNA of RA-converting enzyme Raldh2 (Fig. 3A). Consistent with Raldh2 expression, CD103⁺CD8α⁻ LPDCs induced Foxp3⁺ Tregs in the presence of TGF-β (Fig. 4). These data suggested that CD8α⁻ LPDCs but not CD8α⁺ LPDCs are responsible for the regulatory functions of CD103⁺ DCs in the LP. Accordingly, CD103⁺CD8α⁺ LPDCs may not have the ability to induce immunological tolerance in the small intestinal LP.

Although Ag-specific IgG induction is essential for the resolution of systemic infection, secretory IgA is critical for the protection of mucosal surfaces against viruses, bacteria, and toxins by direct neutralization or prevention of binding to the mucosal surface (18). Ag-specific IgA induction is important in preventing bacterial and viral infection. RA has been shown to have a direct IgA-promoting effect on B cells (12, 19). Both Ag-loaded CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs induced high titers of Ag-specific IgG in serum. However, in accordance with Raldh2 expression, only CD103⁺CD8α⁻ LPDCs could induce T cell-independent IgA synthesis in vitro and Ag-specific IgA in stool samples (Figs. 3B, 3C, 7A). Thus, CD103⁺CD8α⁺ LPDCs are not involved in IgA synthesis in the small intestinal LP.

T cell-dependent immune responses are polarized by the activation of different CD4⁺ T cells, which produce individual patterns of cytokines. Both CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs induced Ag-specific Th1 cells following TLR ligand stimulation (Figs. 5A, 5B, 7B). In addition to Th1 cells, CD103⁺CD8α⁻ LPDCs induced Ag-specific Th17 cells. Thl cells secrete IL-2 and IFN-γ and predominate in cellular immune responses, particularly during intracellular infections. Th17 cells were reported to have important roles in a variety of inflammatory diseases, though it still remains unclear whether Th17 cells contribute to host protection and inflammation (20). Th17 cytokines such as IL-17A, IL-17F, and IL-22 have been shown to be critical for eliminating extracellular pathogens, suggesting that Th17 cells may play crucial roles in host defense against extracellular pathogens (21–24). Therefore, CD103⁺CD8α⁻ LPDC-mediated immunization may be more effective than CD103⁺CD8α⁺ LPDC-mediated immunization in the prevention of infectious diseases due to the induction of both Th1 and Th17 cells.

We finally examined the Ag-specific CD8⁺ T cell response mediated by CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs. CD103⁺CD8α⁺ LPDCs and CD103⁺CD8α⁻ LPDCs induced the proliferation and cytotoxicity of CD8⁺ T cells in response to TLR ligand stimulation in vitro (Fig. 6). Although we injected the same numbers of Ag-loaded DCs stimulated with CpG ODN, class I-restricted IFN-γ levels and cytotoxic activity after CD103⁺CD8α⁻ LPDC immunization were much higher than those after CD103⁺CD8α⁺ LPDC immunization (Fig. 7C, 7D). Haan et al. (17) demonstrated that cell-associated Ag is cross-presented by CD8α⁺ DCs but not CD8α⁻ DCs in the spleen. Furthermore, i.v.-injected soluble Ags were cross-presented by CD8α⁺ DCs in the spleen. In contrast, CD8α⁻CD11b⁺ DCs were generally involved in CD4⁺ T cell responses to soluble Ags (25). Therefore, CD8α⁺ DCs in the spleen are believed to be responsible for cross-presentation to CD8⁺ T cells. However, CD103⁺CD8α⁻ LPDCs induced Ag-specific CTLs more effectively than CD103⁺CD8α⁺ LPDCs. Furthermore, both TLR activated CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs effectively induced CD4⁺ Th responses. A previous study showed that CD8α⁺ and CD8α⁻CD11b^lo DCs in Peyer’s patches take up viral Ags from reovirus-infected intestinal epithelial cells and can present them to CD4⁺ T cells (26). In addition, CD8α⁻CD11b⁻ DCs as well as CD8α⁺ DCs present viral Ags to CD8⁺ T cells during viral infection in the airway (27). Moreover, Chung et al. (28) demonstrated that CD8α⁻CD11b⁺ DCs rather than CD8α⁺ DCs cross-present intestinal Ags to CD8⁺ T cells in MLNs. Hence, within the mucosal-associated lymphoid tissues, cross-presentation may be mediated by CD8α⁻ DCs.

In summary, our data suggest that CD103⁺CD8α⁺ and CD103⁺CD8α⁻ LPDCs have divergent functions in active immunity. Based on the results obtained in this study, from both quantitative and qualitative viewpoints, CD103⁺CD8α⁺ LPDCs may be less suitable targets for oral vaccines than CD103⁺CD8α⁻ LPDCs in the small intestine. Further analysis of these two CD103⁺ DC subsets as well as their functional cross talk is needed for a better understanding of how the intestinal immunity is regulated to finely tune the innate and adaptive systems.

We thank N. Kitagaki for technical assistance and E. Kamada for secretarial assistance.

The authors have no financial conflicts of interest.