Migrating Type 2 Dendritic Cells Prime Mucosal Th17 Cells Specific to Small Intestinal Commensal Bacteria Free

Dendritic cells (DCs) are innate immune effectors that regulate many aspects of adaptive T cells responses. Conventional DCs are classified as type 1 DCs (DC1s) or type 2 DCs (DC2s) with tissue-specific distribution and maturation patterns (1, 2). DC1s express Xcr1 and depend on BATF3 and irf8 for lineage development (3, 4). Insights gained from infectious disease models and tumor models revealed that DC1s are critically important for generating CD8 CTLs, CD4 Th1 cells, and T regulatory (Treg) cell responses (5, 6). DC2s are CD11b⁺Sirpα⁺ cells with a developmental requirement for Notch2 and Irf4 (7, 8). This DC subset shares many features with macrophages and is mainly involved in Th2 and Th17 responses in the skin and mucosal tissues (9–11). In the small intestine, both DC subsets can be detected as CD103⁺ cells, and each is specialized in priming different types of T cells. DC1s activate CD8 T cells and imprint T cell gut homing by inducing the CCR9 receptor through a retinoic acid–dependent mechanism (12). The absence of DC2s reduces homeostatic Th17 cells in the lung, small intestine, and colon (10, 13). A significant decrease of mucosal Th17 cells postinfection with Aspergillus fumigatus or vaccination with i.p. OVA indicates that DC2s control the priming of CD4 T cells for these antigenic challenges (10, 13). However, the role of DC2s in the context of commensal bacteria–induced Th17 responses remained relatively unexplored.

Th17 cells are induced by many gut commensal bacteria and segmented filamentous bacteria (SFB) represents the single most potent inducer of intestinal Th17 cells identified (14). Similar to the classically defined Th17 cells (15, 16), SFB-induced Th17 cells are characterized by expression of the master transcription factor retinoic acid–related orphan receptor (ROR)γt and the signature cytokines IL-17A, IL-17F, and IL-22. As a nonpathogenic inhabitant of the small intestine, SFB does not breach the epithelium, and hence the exact mechanism of bacterial Ag uptake by host DCs remained enigmatic. Adhesion of bacterium to the host epithelial cells and a CDC42-dependent endocytosis process by epithelial cells are necessary for the induction of Th17 cells (17, 18). Downstream of Ag acquisition, it has been demonstrated that small intestinal lamina propria (SILP) CCR2-expressing, CX3CR1⁺ macrophages are indispensable for the intestinal Th17 response to SFB (19). However, macrophages are unlikely to migrate to draining lymph nodes, and naive CD4 T cells do not enter SILP due to the lack of gut-homing receptors. It is unclear how Th17 cell differentiation is accomplished in mesenteric lymph nodes (MLNs). One possibility is that the CCR2-expressing phagocytic cells contain a minor subset of emigrating DCs (20) that enter MLNs to prime naive CD4 T cells. Alternatively, Ag transfer could take place between CX3CR1⁺ macrophages to migratory DCs via gap junction (21).

The availability of an SFB peptide-specific transgenic CD4 T cell model has made studying Th17 differentiation at the clonal level possible (22). In addition, advanced classification of DC subsets is made possible by the recent developments in single-cell RNA sequencing (scRNA-seq) technology. In the current study, we focused on the priming stage of naive CD4 T cells in MLNs to identify the migratory DC subset that activates and drives Th17 differentiation. Results from genetic models targeting different myeloid cell populations indicated that the Zbtb46-expressing DCs, but not macrophages, present SFB Ags in the MLNs. CCR7-dependent homing of DCs to the MLNs is essential for naive T cell activation, and IL-6 accounts for a significant fraction of the RORγt expression in T cells. scRNA-seq of MLN migratory DCs revealed an increase in two transcriptionally distinct DC2 populations after SFB colonization, that is, Cd40/Ccl22- and Cd1d1/Tnfrsf4-expressing subsets. IL-6 can be detected in the Cd40/Ccl22 DC2 cluster. Our findings provide insights on the DC2s responsible for the mucosal Th17 cell response.

C57BL/6, 7B8Tg, zDC-DTR, CCR7^−/−, CCR2^−/−, BATF3^−/−, CX3CR1^gfp/gfp, IL-6^−/−, MHC class II (MHC II)^fl/fl, STAT3^fl/fl, IL-6Rα^fl/fl, CD4^Cre, Lysm^Cre, and CD11c^Cre mice were purchased from The Jackson Laboratory. Mice were bred and maintained at the specific pathogen-free animal facilities of the Medical University of South Carolina and later at the The Ohio State University. All experiments were performed under protocols approved by an Institutional Animal Care and Use Committee at both universities. Mouse colonies were screened regularly for the presence of fecal SFB. Mice contaminated with SFB were treated with ampicillin in drinking water (1 g/ml) for 2 wk and allowed to recover before being used for experiments. SFB donor mice included C57BL/6 mice sourced from Taconic and in-house SFB-positive mice.

A single fecal pellet (∼12 mg) was crushed in 2 ml of PBS, filtered through a 40-μm cell strainer, and homogenized using 0.1-mm zirconia/silica beads (BioSpec Products) in a FastPrep-24 instrument (MP Biomedicals). Bacterial DNA was isolated from cell lysate supernatant using a DNA Clean & Concentrator kit (Zymo Research) according to the manufacturer’s protocol. Quantitative PCR was performed using SsoAdvanced Universal SYBR Green supermix (Bio-Rad) with published primer sequences (SFB736F, 5′-GACGCTGAGGCATGAGAGCAT-3′; SFB844R, 5′-GACGGCACGGATTGTTATTCA-3′; UniF340, 5′-ACTCCTACGGGAGGCAGCAGT-3′; UniR514, 5′-ATTACCGCGGCTGCTGGC-3′). The Cq (quantification cycle) value of SFB-specific 16S rRNA was normalized against the Cq value of eubacteria 16S rRNA of the same sample.

A single fecal pellet from SFB-positive donor mice was crushed in 2 ml of PBS, filtered through a 40-μm cell strainer, and used for 10 recipient mice. One week after SFB oral gavage, MACS beads–isolated naive 7B8Tg cells were transferred via tail vein injection at a concentration of 2 × 10⁵ cells in 200 μl of PBS per mouse. 7B8Tg cells were labeled with CellTrace Violet (Thermo Fisher Scientific) according to the manufacturer’s protocol. Three days after transfer, recipient mice were sacrificed and MLNs were harvested for flow cytometry staining. Donor 7B8Tg cells were detected by the expression of congenic marker CD45.1.

Bone marrow cells were collected from Zbtb46-DTR donor mice, and RBC lysis was carried out with ACK (ammonium, chloride, potassium) lysis buffer. Two million cells were injected into lethally irradiated C57BL/6 recipient mice. Two months later, mice were bled to check for donor cell engraftment. Chimeras were orally gavaged with SFB-positive fecal material, and 1 wk later, adoptively transferred with 2 × 10⁵ naive 7B8Tg cells. A day before and after 7B8Tg cell transfer, diphtheria toxin (DT) was administered via i.p. injection to deplete DCs. Three days after 7B8Tg transfer, mice were sacrificed and MLNs were harvested for analysis.

MHC II tetramer detecting SFB3340-specific endogenous CD4 T cells (I-A^b SFB3340 200–210 QFSGAVPNKTD, PE and allophycocyanin labeled) was generated by the National Institutes of Health Tetramer Core Facility. Briefly, a single-cell suspension was incubated with tetramer (1:100 dilution) at 37°C for 30 min. Cells were then stained with surface Abs and viability dye. Samples were recorded with the BD LSRFortessa immediately without fixation.

Spleens were crushed, filtered through a 100-μm cell strainer, and subjected to RBC lysis using ACK buffer. SILP cells were collected as follows: the distal half of the small intestine was harvested, Peyer’s patches were removed, and the small intestine was cut open longitudinally before excising into ∼1-cm segments. Tissue segments were washed in PBS and incubated with DTT/EDTA for 20 min in a 37°C shaker to disrupt the epithelium. After three washes, tissue segments were digested with collagenase D/DNase1/dispase for 30 min in a 37°C shaker. Immediately after the digestion step, the enzymatic reaction was stopped by filling the reaction tube with cold RPMI 1640 complete medium. Tissue segments were then crushed and filtered through 100-μm cell strainers. A single-cell suspension was partitioned on a 40/80% Percoll gradient. Lymphocytes at the interphase were collected and washed prior to flow cytometry staining. For the processing of lung lymphocytes, mice were perfused with PBS before harvesting lungs. Lung tissues were excised into smaller fragments and digested with EDTA for 30 min in a 37°C shaker. After two washes, tissue fragments were digested with collagenase D/DNase1/dispase for 45 min in a 37°C shaker. Immediately after the digestion step, the enzymatic reaction was stopped by filling the reaction tube with cold RPMI 1640 complete medium. A single-cell suspension was obtained by following the same procedure as for SILP cells.

The staining of ex vivo T cells began with 30 min of incubation with surface Abs and viability dye on ice. Transcription factors were stained using a Foxp3 transcription factor staining buffer set (eBioscience, Thermo Fisher Scientific). Briefly, fixation was done on ice for 30 min. RORγt, Foxp3, T-bet, Batf, Irf4, and cMaf Abs were added to the permeabilization buffer and cells were stained at room temperature for 30 min. For intracellular cytokine staining, in vitro–differentiated Th17 cells were stimulated with PMA and ionomycin in the presence of brefeldin A for 4 h. Cells were stained with surface Abs and viability dye as described above before fixation with 4% paraformaldehyde for 5 min at room temperature. Abs for IL-17a and TNF-α were added to saponin permeabilization buffer and cells were stained for 30 min on ice. Flow staining for DCs started with a FcR-blocking step for 10 min at room temperature, followed by surface staining for 20 min at room temperature. Live cells were recorded immediately or fixed with 4% paraformaldehyde until acquisition. Flow cytometer instruments used for this study include BD LSRFortessa and Cytek Aurora. Analyses were done using FlowJo v10 software (Tree Star).

C57BL/6 mice were orally gavaged with SFB-positive fecal material. Five days later, 15 naive mice and 15 SFB-positive mice were sacrificed and MLNs were collected. T cells and B cells were depleted using biotinylated anti-mouse CD3e and CD19 followed by anti-biotin MACS beads (Miltenyi Biotec). Negatively selected cells were FACS sorted for live migratory (CD11c⁺MHC II^hi) and resident (CD11c^hiMHC II⁺) cells. cDNA library generation followed established techniques using a Chromium Single Cell 3′ Library v2 kit (10x Genomics; https://assets.ctfassets.net/an68im79xiti/RT8DYoZzhDJRBMrJCmVxl/6a0ed8015d89bf9602128a4c9f8962c8/CG00052_SingleCell3_ReagentKitv2UserGuide_RevF.pdf). Briefly, cells in a single-cell suspension were loaded onto a 10x Genomics chip A and emulsified with 3′ single-cell GEM beads using a Chromium controller (10× Genomics), and libraries were constructed from the barcoded cDNAs (Translational Science Laboratory at the Medical University of South Carolina). RNA-seq was performed on each sample (∼ 50,000 reads/cell) using a NovaSeq S4 flow cell (Illumina) at the VANTAGE facility (Vanderbilt University Medical Center). Resulting data were processed using Cell Ranger software (10x Genomics) at the sequencing core, followed by analysis in our laboratory using the R-based package Seurat according to the standard workflow laid out in a scRNA-seq integration vignette (https://satijalab.org/seurat/reference/seurat-package).

An unpaired, equal variance Student t test was used to calculate statistical significance in Excel and Prism software (*p < 0.05 and **p < 0.01).

The data discussed in this publication have been deposited in Gene Expression Omnibus (GEO) and are accessible through GEO accession number GSE184423 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi).

Host immune responses to SFB are well studied. Specifically, the induction of gut CD4 Th17 cells and the availability of transgenic CD4 T cells recognizing a known bacterial epitope (22) through a natural oral infection route made this model an excellent system to identify important factors in homeostatic mucosal T cell responses. In mice, fecal and epithelial SFB can be detected at ∼1 wk after oral gavage, peaked at 2 wk, and maintained a low but detectable level 1 mo later (Fig. 1A). The growth and contraction of SFB suggest that the host immune system is capable of controlling the bacteria at a nonpathogenic level. The host TCR affinity for the defined SFB3340_568–580 epitope was predicted to have a K_D value of 89 nM (22), which falls within the K_D range of the anti-pathogen TCR–peptide MHC (23). This made reliable detection of host SFB-specific CD4 T cells with MHC II tetramer feasible. In the SILP, Ag-specific CD4 T cell expansion reached a maximal level at 2 wk after SFB oral gavage (Fig. 1B). We next examined the kinetics of adoptively transferred 7B8Tg cell activation, which recognize the same epitope that was used for generating the MHC II tetramer. Similar to endogenous T cells, expansion of 7B8Tg cells peaked at ∼2 wk (Fig. 1C). Expression of the RORγt transcription factor spiked transiently in MLN 7B8Tg cells within the first week of T cell activation but continued to maintain high expression in SILP 7B8Tg cells up until 1 mo later (Fig. 1D). The expansion curve of T cell activation suggests that the initial differentiation of Th17 cells takes place in the draining lymph nodes, but complete differentiation requires microbial elements and microenvironment factors in the small intestine. These analyses clearly show that the kinetic of host primary CD4 Th17 cell response to gut commensal bacteria is consistent with pathogenic bacteria–induced CD4 T cells (24, 25).

To visualize the early phase of Th17 differentiation, naive CD4 7B8Tg T cells were labeled with CellTrace dye and adoptively transferred into SFB-colonized C57BL/6 mice. Three days later, T cell proliferation versus expression of lineage-determining transcription factors and homing receptors were examined. Although SFB is localized to the gut, 7B8Tg cell activation and proliferation can be observed in distal organs such as the spleen and lung (Fig. 2A), suggesting that SFB Ags are disseminated systemically either in free form or carried by APCs. However, only in the MLNs that activated but undivided 7B8Tg cells and their early progeny upregulated RORγt (Fig. 2B), indicating that cellular machinery that supports Th17 differentiation is restricted to the draining lymph nodes. Activated 7B8Tg cells leaving MLNs may then accumulate at spleen and lung via systemic circulation. APCs driving Th17 differentiation are likely limited in numbers because increasing 7B8Tg donor cell numbers resulted in slower proliferation and reduced RORγt expression (Fig. 2C). As these early Th17 cells progressively strengthen their identity, the expression levels of core transcription factors such as IRF4, BATF, and cMaf (26) are amplified, and they displayed no sign of plasticity toward the Treg or Th1 cell lineage (Fig. 2D). Lymphoid organ–homing receptor CD62L was downregulated on activated Th17 cells, and peripheral tissue–homing receptor CCR4 was upregulated (Fig. 2D). Surprisingly, gut-homing receptors CCR6 and CCR9 (27, 28) were not significantly upregulated at this early time point (Fig. 2D). Gut homing of these early Th17 cells is likely dependent solely on α₄β₇ integrin (29). PD1 was upregulated on Th17 cells (Fig. 2D) but did not seem to have inhibitory effects on either Th17 cell proliferation or differentiation at this early stage (data not shown). Taken together, the limited numbers of Th17-promoting APCs that prime the first wave of Th17 cells are localized to MLNs.

Myeloid cells at the gut mucosal system are phenotypically and functionally heterogeneous. In tissue-draining lymph nodes, DCs are generally divided into migratory (activated and mature) versus resident (steady state). Both populations contain DC1s (Xcr1⁺) and DC2s (Sirpα⁺CD11b⁺) that are thought to preferentially home to a specific region of the lymph node and induce selected subsets of T cells (9). Both CD103⁺ DC2s and macrophages have been reported in the induction of gut Th17 cells. Specifically for SFB, CCR2-expressing CX3CR1⁺ macrophages, but not CD103⁺ DCs, support the induction of SILP Th17 cells (19). However, a small subset of CCR2⁺ cells derived from the DC precursors and are thought to be bona fide DCs involved in Th17 cell activation (20). It remains unclear whether Th17 cell–inducing APCs are of macrophage or DC lineage. We took advantage of the SFB-specific 7B8Tg model to address this question during T cell priming. The Zbtb46 transcription factor distinguishes classical DCs from macrophages, monocytes, and plasmacytoid DCs (30). We generated a Zbtb46-DTR bone marrow chimera, orally gavaged mice with SFB, and administered DT to the mice before 7B8Tg cell adoptive transfer. A single dose of DT effectively depleted both the migratory (MHC II^hiCD11c⁺) and resident (MHC II⁺CD11c^hi) DCs in the MLNs, but SILP DCs were not significantly impacted (Fig. 3A). Naive 7B8Tg cells failed to activate in these hosts (Fig. 3B), suggesting that classical DCs are indispensable for the priming step of Th17 cells. In MHC II^fl/flLysM^Cre (Supplemental Fig. 1A, 1B), CX3CR1^gfp/gfp (Supplemental Fig. 1C), and CCR2^−/− (Supplemental Fig. 1D, 1E) recipients, however, where APCs of the macrophage or monocyte lineages were targeted, Th17 differentiation was not affected. Next, in CCR7^−/− mice, where most of the migratory DCs are absent (Fig. 3C), 7B8Tg cells failed to be activated (Fig. 3D), confirming that the sources of SFB Ags are gut-derived activated, migratory DCs. CCR7 deficiency impairs migration of all classes of gut-derived DCs and is insufficient for the specific DC/T cell axis. To further examine the role of DC subtypes in Th17 induction, we made use of the DC lineage-specific model. The MLN migratory DC population consists of DC1s (Xcr1⁺CD103⁺CD11b⁻) and DC2s (Sirpα⁺CD11b⁺). In Batf3^−/− mice, Xcr1⁺ DC1s are absent, but the DC2 compartment remained intact (Fig. 3E). CD11b⁺CD103⁺ DC2s are present but with reduced CD103 expression (Fig. 3E, lower panel). We found that DC1s are not involved in SFB-induced Th17 cells (Fig. 3F), consistent with the previous report (19). Altogether, these findings suggest that migratory DC2s are the APCs that prime CD4 Th17 cells in the intestinal draining lymph nodes.

Cytokines are critical drivers of CD4 Th cell differentiation. Ample evidence established that IL-1β, IL-6, IL-21, IL-23, and TGF-β are all implicated in Th17 cell differentiation and pathogenicity (31, 32). Temporally speaking, IL-6 acts at an earlier window because IL-6Rα expression is detected on naive CD4 T cells (Supplemental Fig. 1F). IL-1β, IL-21, and IL-23 are likely to act on differentiated Th17 cells because IL-1R, IL-21, and IL-23R expression levels are sequential to IL-6 stimulation and RORγt expression (33–35). We transferred 7B8Tg cells into SFB-colonized IL-6^−/− recipients and found that, indeed, IL-6 is important for early Th17 differentiation (Fig. 3G). Despite normal proliferation, activated 7B8Tg cells showed suboptimal RORγt upregulation in IL-6^−/− mice (Fig. 3G). IL-6 sensing by CD4 T cells is generally thought to be achieved through classical signaling, acting on the IL-6Rα/gp130 complex on T cells. We generated IL-6Rα^flfCD4^Cre 7B8Tg mice and confirmed that IL-6Rα expression on naive 7B8Tg CD4 T cells was absent (Supplemental Fig. 1G). IL-6R-null 7B8Tg cells showed no deficiency in early Th17 differentiation (Supplemental Fig. 1H). On the contrary, T cell–intrinsic IL-6R signaling is indispensable for in vitro Th17 differentiation with direct IL-6 and TGF-β stimulation (Supplemental Fig. 1I, upper panel). The lack of IL-6R signaling selectively affects IL-17A production but not a non-Th17 cytokine TNF-α (Supplemental Fig. 1I, lower panel). Interestingly, deletion of a IL-6 downstream signaling molecule Stat3 in APCs significantly decreased 7B8Tg cell proliferation and Th17 differentiation (Fig. 3H). Stat3 deficiency did not impact DC1 or DC2 development and the ratio in the MLN migratory DC compartment (Supplemental Fig. 1J). Our findings indicate that in vivo, commensal bacteria–induced Th17 cells required IL-6, which indirectly promotes Th17 differentiation through Stat3 signaling in APCs.

Characterization of splenic DCs by single-cell sequencing and flow cytometry revealed considerable heterogeneity within DC2s (36). T-bet⁺Esam⁺ DC2s have an anti-inflammatory signature, whereas RORγt⁺Clec12a⁺ DC2s are proinflammatory and support Th17 cell differentiation in vitro (36). Using Esam and Clec12a markers, we refined DC2s in the spleen, SILP, and MLNs (Fig. 4A, 4B). To distinguish bona fide DCs from macrophages, we used CD24 marker (Supplemental Fig. 2A, 2B), a surface receptor that distinguishes DCs from macrophages and has costimulation function on T cells (13, 37, 38). In the SFB-colonized mice, the MLN migratory population is dominated by Sirpα⁺ DC2s, mirroring the small intestine, whereas MLN-resident DCs are mostly Xcr1⁺ DC1s, similar to the spleen (Fig. 4A, 4B). Secondarily, unlike splenic and resident DCs that contain both Esam⁺ and Clec12a⁺ cells, migratory DCs are largely Esam⁺ and small intestinal DC2s are partially Clec12a⁺ (Fig. 4A, 4C). Hence, MLN migratory DC2s cannot be further subclassified using Esam and Clec12a. Overall, both migratory DC1s and DC2s express a high level of costimulation receptors (CD86, CD40, and CD70) and migration marker (CCR7) compared with MLN-resident DCs, splenic DCs, or F4/80⁺CD24^lo macrophages (Fig. 4D, Supplemental Fig. 2C), highlighting their maturation status and T cell priming capacity. Altogether, currently known costimulatory receptors (Fig. 4D) could not adequately distinguish migratory DC2s from DC1s functionally. Additionally, we were unable to detect the essential Th17-polarizing cytokine IL-6 in the MLN DCs by flow cytometry. To identify the source of IL-6 and further interrogate DC2 heterogeneity in MLNs, we subjected both migratory and resident DCs in SFB newly colonized mice (5 d after oral gavage when fecal SFB was beginning to be detectable) to scRNA-seq.

From SFB⁻ and SFB-positive mice (15 mice in each group), MLNs were harvested and DCs were FACS sorted based on a live lineage (CD3e, CD19, NK1.1)⁻CD11c⁺MHC II⁺ gating scheme. Migratory (MHC II^hiCD11c⁺) and resident (MHC II⁺CD11c^hi) DCs were partitioned, and single-cell cDNA libraries were generated using the 10x Genomics Chromium system. Sequencing was done on an Illumina NovaSeq 6000, and downstream analyses were done using Seurat (39). Integrated analysis performed on the migratory DCs (4,335 cells, 14,023 genes) and resident DCs (4,604 cells, 13,850 genes) from SFB-positive mice resulted in 14 DC clusters (Fig. 5A, Supplemental Fig. 3A). Migratory DCs formed two transcriptionally distinct cell clusters that are nonoverlapping with all resident DC clusters (Fig. 5B). Based on known DC genes (Flt3, Ly86) and DC1/DC2 (Xcr1, Clec9a, Sirpa, Irf4) gene signatures, lineage identity was assigned to each cluster (Fig. 5C). Consistent with flow cytometry data (Fig. 4D), migratory DC1s (C1) and DC2 (C0) were enriched in Ccr7 and CD40. Interestingly, OX40 (Tnfrsf4), typically detected on CD4 T cells, was highly expressed by MLN migratory DCs (Fig. 5D). In addition, selected chemokines (Ccl22, Ccl5, and Cxcl16) were abundant in migratory DCs (Fig. 5D). These and other genes (Apol7c, Serpinb6b, AW112010) that distinguish migratory DC2s from resident DC2s were mostly shared by migratory DC1s (Supplemental Fig. 3B). Nevertheless, we identified genes that are unique to migratory DC2s (Cd1d1, Car2, Gbp2) versus resident DC2s (Ltb, Il1b, Fcer1g) that are minimally expressed by DC1s (Fig. 5E). Altogether, results and analyses from single-cell sequencing data reinforce the general observations that migratory DCs are activated with elevated expression of costimulatory receptors and chemokines.

To gain insight into the cellular source of IL-6, we performed an integrated analysis on SFB⁻ migratory DC (4,220 cells, 13,767 genes) and SFB-positive migratory DC (4,335 cells, 14,023 genes) datasets. Eighteen cell clusters were identified (Fig. 6A), with three major DC1 and two major DC2 clusters (Fig. 6B, 6C). Although the cluster-specific markers of the three DC1 clusters are largely overlapping, the major DC2 clusters can be distinguished by the higher level of Csrp1, Ccl22, and Isg15 in C4 (Supplemental Fig. 4A). In addition to classical DC1/DC2 lineage markers (Irf8 and Fcer1g), these DC subsets can be identified using additional markers as revealed in this analysis, such as Cd81 and Glipr2 versus Cd1d1 and Fabp5 (Fig. 6D, Supplemental Fig. 4B). SFB colonization increased the proportion of DC2s (C0 and C4, (Fig. 6B) that are characterized by Ccl22, Cd40, and Cd1d1 compared with the unchanged DC2 clusters (Ltb, S100a4, and S100a6) (Fig. 6E). IL-6 was detected in both of these DC2 clusters but was much less or nearly absent in DC1 clusters (Fig. 6F). In contrast, the transcripts of TGF-β, another essential cytokine for in vivo Th17 activation (32), are widely distributed across DC1s and DC2s (Fig. 6F, lower panel). We identified several transcripts that are enriched in migratory DC2s and examined their relationship to IL-6 expression. Il6, Cd40, Ccl17, and Ccl22 are mostly detected on the C4 DC2 cluster; the C0 DC2 cluster, in contrast, expressed a higher level of Cd1d1 and Tnfrsf4 (Fig. 7A). However, the transcriptional differences between these DC2 clusters were not resolvable at the protein level using CD1d (Cd1d1) and OX40 (Tnfrsf4), which present an expression spectrum instead of distinct cell populations (Fig. 7B). Nonetheless, we validated that there is an increased expression of these markers compared with lymph node–resident and splenic DCs (Fig. 7B). To identify a gene signature that is unique to IL-6–producing DCs, we performed differential expression analysis using migratory DCs from SFB-positive mice. Dll4, a Notch ligand, was found to be enriched in this DC subset (Fig. 7C), in agreement with the role of this ligand for Th17 cell differentiation (40). Additionally, novel transcripts associated with Il6 expression such as Ifitm1, Htra1, Mylk1, and Ccnd2 were revealed. Altogether, our single-cell sequencing data revealed that IL-6–producing, Th17-polarizing DC2s constitute a small fraction of migratory DC2s that coexpress Cd40, Ccl17, and Ccl22.

Immune cells at the mucosal tissues are charged with the responsibility of host protection and, consequently, have devised a plethora of cellular and molecular mechanisms to balance immunity and tolerance at the barrier tissues. DCs and macrophages are abundant and phenotypically diverse at the intestinal lamina propria (41), making the examination of gut T cell responses challenging.

Historically, CD11b⁺CD103⁺ DC2s were shown to be essential for inducing homeostatic CD4 Th17 cells in the gut (10, 13). More recently, along with the emerging interests in the gut microbiome and the identification of specific Th17-inducing commensal bacteria species (SFB), intestinal macrophages were found to control Th17 and Treg cell responses in the context of microbiota (19, 42). Using the SFB Ag-specific T cell model, our results indicated that DCs control the priming of naive CD4 T cells and the early differentiation of Th17 response (Figs. 1–3). At this early stage of Th17 differentiation, RORγt expression is relatively low, and cell frequency is rare in the small intestine (Fig. 1D). Notably, activated Th17 cells do not persist in the MLNs but instead they home to the small intestine and continue to reinforce RORγt expression and expand (Fig. 1C, 1D). The kinetics of the Th17 response suggest that early priming steps are initiated by MLN DCs and, later, full differentiation and maintenance of the Th17 cell population are likely supported by intestinal macrophages.

Effective T cell differentiation requires proper costimulation and cytokine signals delivered by APCs. Ample evidence suggested that IL-6, IL-1β, IL-21, and IL-23 are all implicated in Th17 responses both in physiological and inflammatory settings (32–35, 43, 44). We found that IL-6 is important for Th17 polarization in CD4 T cells; however, IL-6Rα expression is not essential for the optimal expression of RORγt (Fig. 3G, Supplemental Fig. 1H). Examination of the IL-6R signaling pathway revealed that the expression of downstream signaling molecule Stat3 in T cells is required for the overall gut Th17 response (29). It is possible that deficiency in T cell–intrinsic IL-6R signaling could be compensated by DCs trans-presenting IL-6 to naive T cells, which has been shown to support pathogenic Th17 priming (45). Additionally, other Th17-related cytokines could also compensate for the loss of IL-6 (29). Overall, IL-6 may play an important role in early Th17 differentiation or when it is the sole source of the polarizing factor (Supplemental Fig. 1I) but is not absolutely required in vivo.

Despite the redundant role of IL-6, we found it to be informative when probing the DC2 subsets that prime Th17 cells. Concluding from the results obtained with CCR7^−/− and BATF3^−/− mice (Fig. 3D, 3F), migratory DC2s are the most likely source of Th17 Ags. DC2s are considerably heterogeneous and can be further subset into Esam⁺ and Clec12a⁺ cells in the spleen, both of which are capable of supporting Th17 differentiation (7, 36). Our characterization of MLN migratory DC2s revealed them to be mostly Esam⁺ and CD24^hi and have and increased level of costimulatory markers compared with splenic or resident DC2s (Fig. 4, Supplemental Fig. S2). On the contrary, small intestine DC2s are partially Clec12a⁺ and devoid of Esam⁺ cells (Fig. 4). To further understand the composition and function of DC2s, we took an unbiased approach and subjected MLN DCs to single-cell sequencing. IL-6 transcripts were found in a minor subset of DC2s, mostly colocalized with Cd40-high DC2s (Fig. 7A). CD40–CD40L crosstalk has been shown to promote Th17 response (46). However, in the MLN CD40 is expressed by both migratory DC1s and DC2s (Fig. 4D), and it is unlikely to be the determining element for CD4 Th17 cell activation in this context. In the lymph nodes, migratory DC1s (with high CCR7 expression) are observed to travel deep into the T cell zone, attracted by the gradient of CCL19/CCL21 (9). This was observed with skin-draining lymph node informed using the Xcr1 marker (47). The T cell/B cell border localization of DC2s due to their lower expression of CCR7 (Fig. 4D and reviewed in Ref. 9) could be responsible for their preferential interaction with naive CD4 T cells. Interestingly, we found that CD1d effectively distinguished migratory DC2s from DC1s at the transcript and protein levels, displaying a completely reversed trend compared with the resident DCs (Figs. 6D, 7B). CD1d is known to present lipid Ags and activate NK T cells (48), but early reports suggest that it can interact with the CD4 coreceptor, contributing to CD4⁺ invariant NKT cell activation (49). It will be interesting to investigate its role in CD4 T cell activation.

In MLNs, chemokines are constitutively expressed by migratory DCs and are generally more abundant than cytokines. Among the chemokines identified, Ccl17 and Ccl22 overlapped with Il6 in the DC2 cluster that was increased by SFB colonization (Fig. 7A). Conversely, CCR4, the receptor for these two chemokines, was detected on newly activated Th17 cells in MLNs (Fig. 2D). CCR4/CCR6-deficient mice showed a suppressed Th17 cell response in experimental autoimmune encephalomyelitis (50). We speculate that the CCL17/CCL22 produced by DC2s may act as a chemoattractant for Th17 cells for sustained interaction in the MLNs, reinforcing Th17 cell differentiation.

Our study identified the pivotal role of migratory DC2s in early Th17 programming at the small intestine. Importantly, however, note that DC2s play functionally diverse roles in adaptive T cell responses. They have been shown to induce CD4 Th1, Th2, Th17, T follicular helper, and Treg cells (10, 21, 51, 52) when responding to different immunizations. In the MLNs, DC1s and DC2s play redundant roles in the Th1 response (52, 53), but DC2s are uniquely required for Th2 and Th17 responses. This was clearly demonstrated by subjecting IRF4^fl/flCD11c^Cre mice to Schistosoma mansoni infection or when examined under steady state, whereby Th2/Th17 cytokines were greatly diminished but IFN-γ production remained unaffected (10, 13, 52). On closer examination using Ag-specific transgenic T cells, it was found that the absence of DC2s did not impair CD4 OT-II cell activation and proliferation but specifically blocked IL-17 production after OVA immunization (10). These results suggest that for Th17 differentiation, DC2s deliver mucosal tissues or pathogen-imprinted molecular factors that are not found on DC1s. IL-6, the key cytokine that directs the early stage of Th2 and Th17 differentiation (35, 54), represents an important clue for identifying the DC2 subset that is specific for this role. Moving forward, the novel markers identified on migratory DC2s could be potential targets for therapeutic intervention in Th17-related inflammatory diseases.

We acknowledge Cynthia Timmers and Marty Romeo of the Translational Science Core, Hollings Cancer Center, Medical University of South Carolina for processing the samples for single-cell RNA sequencing. We are grateful to the Flow Cytometry & Cell Sorting Unit of the Medical University of South Carolina and to the Immune Monitoring and Discovery Platform, Pelotonia Institute of Immuno-Oncology of The Ohio State University for flow cytometry services. We thank Dr. Zihai Li and Tong Xiao for advice and help with the bioinformatics analysis.

The authors have no financial conflicts of interest.