Monocyte-Derived Dendritic Cells (moDCs) Differentiate into Bcl6⁺ Mature moDCs to Promote Cyclic di-GMP Vaccine Adjuvant–Induced Memory T_H Cells in the Lung

Vaccination is the most cost-effective approach to fight infectious diseases. Approved vaccines mainly elicit Ab-mediated protection and do not generate strong mucosal responses. Vaccines that induce T cell–mediated protection in lung mucosa would provide major health and economic benefits. Memory T cells include central memory T cells, effector memory T cells, and tissue-resident memory T (T_RM) cells. T_RM cells comprise a majority of memory T cells in the lung and play a crucial role in maintaining long-term protective immunity in the lung mucosa (1, 2). Lung T_RM cells include CD4⁺ T_RM and CD8⁺ T_RM cells.

Lung CD4⁺ T_RM cells are crucial for protection against influenza virus and Streptococcus pneumoniae infection (3, 4). Swain’s group showed that lung memory CD4⁺ T cells protect against influenza through multiple synergizing mechanisms (5). Memory CD4⁺ T cells can accelerate primary CD8⁺ T cell responses (6), activate innate immune cells to combat infections (7), and secret T_H cytokines (8). The dendritic cell (DC) subset that promotes the induction of lung CD4⁺ T_RM cells is unknown.

Lung T_RM cells (both CD4⁺ and CD8⁺ T_RM cells) require local Ag presentation (9–12). In contrast, T_RM cells in the intestine, genital tract, or skin do not require tissue Ag recognition (13, 14). Swain’s group (11) first reported that CD4⁺ T cell memory formation depends on the re-engagement of CD4⁺ T cells with APCs at their effector stage (“Signal 4”), day 5–8 of their response (9, 11). Lung CD8⁺ T_RM cell establishment also requires cognate Ag recognition in the lung (10). It is unclear which lung APCs provide the “Signal 4” for lung CD4⁺ T_RM cell induction at this stage of vaccination or infection.

As professional APCs, DCs comprise developmental and functional diverse populations. Common DC progenitors generate preconventional DCs (precDCs) (15). In the lung, precDCs develop into conventional DC (cDC)–1 and cDC2. cDC1 expresses CD103, whereas cDC2 expresses CD11b and CD24 but are CD64 negative. cDC1 are specialized in Ag cross-presentation and priming cytotoxic CD8⁺ T cells (16). cCD2 prime CD4⁺ T cells and generate T_H responses (17). Circulating monocytes also give rise to lung DCs (18). Upon infection or inhalation of allergens, lung epithelial cells produce chemokines CCL2 that recruit CCR2^hi monocytes (19, 20). In the lung, these infiltrating CCR2^hi monocytes upregulate MHC class II (MHCII) and CD11c and become monocyte-derived DCs (moDCs). moDCs are found virtually in any disease with substantial inflammation (21).

Bona fide moDCs are highly capable of Ag uptake, processing, presentation, and priming CD4⁺ and CD8⁺ T cells (21). For example, tumor-infiltrating moDCs prime CD8⁺ T cells and induce antitumor immunity (22). CCR2⁺ moDCs are critical for T_H17 induction and the development of experimental autoimmune encephalomyelitis (23). During cutaneous Listeria major infection, moDCs could induce T_H1 polarization (20). moDCs were also sufficient to induce T_H2 immunity by high-dose house dust mite asthma (24). Last, we recently showed that moDCs induce T follicular helper (T_FH) cells in the lung by mucosal adjuvant cyclic di-GMP (CDG) (25). Thus, there is a significant degree of plasticity and heterogeneity in moDCs in vivo. The underlying mechanism for moDCs plasticity and heterogeneity is unknown (21).

In this report, we study the role of lung moDCs in generating lung mucosal memory T_H cells. We found that moDCs are specifically required for the generation of memory T_H cells in lung mucosa but not in the systemic compartments. Furthermore, we identified a new, to our knowledge, differentiated moDCs subpopulation expressing the transcriptional factor (transcriptional repressor B cell chronic lymphocytic leukemia/lymphoma 6) Bcl6 that is responsible for memory T_H cells generation in lung mucosa.

Age- and sex-matched mice (2–3 mo old) were used for immunization. C57BL/6J, B6.CD45.1 (no. 002014), Bcl6^fl/fl (no. 023727), CCR2^–/– (no. 004999), RelA^fl/fl (no. 024342), IL-21–VFP (no. 030295), CD11c^cre (no. 008068), and LysM^cre (no. 004781) mice were purchased from The Jackson Laboratory. Mice were housed and bred under pathogen-free conditions in the Animal Research Facility at the University of Florida. Cre-negative littermates were used as controls for experiments. All mouse experiments were performed by the regulations and approval of the Institutional Animal Care and Use Committee at the University of Florida, Institutional Animal Care and Use Committee number 201909362.

Eα-OVA (ASFEAQGALANIAVDKA-OVA) fusion protein was produced by GeneCust. TNF-IgG2a-Fc fusion proteins were produced by Creative Biolabs. Wild-type (WT) mouse TNF (aa80–aa235) or mutant mouse TNF (TNF_D221N/A223R) was fused with the Fc portion of the mouse IgG2A. The following reagents were from Invivogen: endotoxin-free OVA (vac-pova) and vaccine-grade CDG (vac-nacdg). The following reagent was from eBioscience: YAE mAb (14-5741-82). The following reagent was from Biosearch Technologies: NP₆CGG (N-5055A).

The following reagents were obtained from BioLegend: allophycocyanin-Mouse IgG2A (clone: MOPC173, catalog no. 981906), anti-mouse CD19–PerCP/Cy5.5 (clone: 1D3/CD19, catalog no. 152405), anti-mouse LAP (TGF-β1) (clone: TW7-16B4, catalog no. 141407, 141413), anti-mouse CD45.1-allophycocyanin(clone: A20, catalog no. 110713), anti-mouse CD69–allophycocyanin (clone: H1.2F3, catalog no. 104578), anti-mouse CD80-FITC (clone: 16-10A1,catalog no. 104705), anti-mouse CD86–allophycocyanin-Cy7 (clone: GL-1, catalog no. 105029), anti-mouse Bcl6–allophycocyanin (clone: 7D1, catalog no. 358505), anti-mouse Bcl6–PE (clone: IG191E/A8, catalog no. 648304), anti-mouse CD4-PE/Cy7 (clone: GK1.5, catalog no. 100422), anti-mouse IL-4–allophycocyanin (clone: 11B11, catalog no. 504106), anti-mouse IL-17a–PE (clone: TC11-1810.1, catalog no. 506903), anti-mouse IL-4–allophycocyanin (clone: 11B11, catalog no. 504105), anti-mouse CD45-PercP/Cy5.5 (clone: 30-F11, catalog no. 103131), anti-mouse MHCII(I-A/I-E)–Brilliant Violet 421 (clone: M5/114.15.2, catalog no. 107636), anti-mouse MHCII(I-A/I-E)–Alexa Fluor (clone: M5/114.15.2, catalog no. 107622), anti-mouse CD11c-allophycocyanin/Cy7 (clone: N418, catalog no. 117323), anti-mouse/human CD11b-PE/Cy7 (clone: M1/70, catalog no. 101216), anti-mouse/human CD11b–Brilliant Violet 605 (clone: M1/70, catalog no. 101237), anti-mouse CD64-PerCP/Cy5.5 (clone: X54-5/7.1, catalog no. 139307), anti-mouse TNFR2-PE (clone: TR75-89, catalog no. 113405), anti-mouse PD1 (clone: 29F.1A12, catalog no. 135214, no. 135223), anti-mouse IL-10–allophycocyanin (clone: JESS-16E3, catalog no. 505016), anti-mouse CXCR–allophycocyanin (clone: L138D7, catalog no. 145506), anti-mouse CD49A–PE (clone: HMa1, catalog no. 142604).

Additional reagents were from Invitrogen (anti-mouse IL-12p35–PE [clone: 27537, catalog no. MA5-23559], anti-mouse IL-23–FITC [clone: fc23cpg, catalog no. 53-7023-80]), Miltenyi Biotec (anti-mouse TNFR2-allophycocyanin [clone: REA228, catalog no. 130-104-698]), R&D Systems (anti-mouse CXCL13 [catalog no. AF470]), anti-mouse CCL20–allophycocyanin-Cy7 [clone: 114906, catalog no. IC760N]), Cell Signaling Technologies (anti-mouse/human pRelB-PE [clone: D41B9, catalog no. 13567], anti-mouse pRelA–allophycocyanin [clone: 93H1, catalog no. 4887S]).

The following reagents were obtained through Biological and Emerging Infections Resources, National Institute of Allergy and Infectious Diseases, and the National Institutes of Health: S. pneumoniae Family 2, Clade 3 Pneumococcal Surface Protein A (PspA UAB099) with C-Terminal Histidine Tag, Recombinant from Escherichia coli, NR-33179.

Groups of mice were intranasally (i.n.) vaccinated with CDG (5 µg), adjuvanted Ag (2 µg), or Ag alone (26). For i.n. vaccination, animals were anesthetized using isoflurane in an E-Z Anesthesia system (Euthanex, Palmer, PA). Ag with or without CDG was administered in 30 µl saline. Sera were collected at the indicated time points after the last immunization. The Ag-specific Abs were determined by ELISA. The secondary Abs used were anti-mouse IgG-HRP (1033– 05; SouthernBiotech), and anti-mouse IgA-HRP (1040–05; SouthernBiotech). To determine Ag-specific T_H response, splenocytes and lung cells from Ag or CDG + Ag-immunized mice were stimulated with 5 µg/ml Ag for 4 d in culture. T_H cytokines were measured in the supernatant by ELISA.

Cells were isolated from the lung, as previously described (25). The lungs were perfused with ice-cold PBS and removed. Lungs were digested in DMEM containing 200 μg/ml Dnase I (10104159001; Roche Diagnostics), 25 μg/ml liberase (05401119001; Roche Diagnostics) at 37°C for 2 h. RBCs were then lysed and a single-cell suspension was prepared by filtering through a 70-µm cell strainer.

For transcription factor Bcl6 staining of murine and human cells, cells were fixed and permeabilized with the Foxp3-staining buffer set (catalog no 00-5523-00; eBioscience). The intracellular cytokine staining was performed using the Cytofix/Cytoperm Kit from BD Biosciences (catalog no. 555028). The single lung cell suspension was fixed in Cytofix/perm buffer (BD Biosciences) in the dark for 20 min at room temperature. Fixed cells were then washed and kept in the Perm/Wash buffer at 4°C. Golgi-plug was present during every step before fixation.

Mouse Ly6C^hi monocytes were purified from the bone marrow of I mice (C57BL/6J or CD45.1) following the protocol according to the manufacturer (19861; StemCell Technologies). Mice were i.n. vaccinated with CDG (5 µg) and Ag. Ly6C^hi monocytes (1.5 million per mouse) were administered i.n. into immunized mice at 30 mins, 2 h, and 4 h postimmunization.

Single-cell suspensions were stained with fluorescent dye–conjugated Abs in PBS containing 2% FBS and 1 mM EDTA. Surface stains were performed at 4°C for 20 min. Cells were washed and stained with surface markers. Cells were then fixed and permeabilized (catalog no. 00-5523-00; eBioscience) for intracellular cytokine stain. Data were acquired on a BD LSRFortessa and analyzed using the FlowJo software package (FlowJo). Cell sorting was performed on the BD FACSAriaIII Flow Cytometer and Cell Sorter.

Data exclusion was justified when the positive or negative control did not work. All experiments will be repeated at least two times. All repeats are biological replications that involve the same experimental procedures on different mice. Experiments comparing different genotypes and adjuvant responses are designed with individual treatments being assigned randomly. When possible, treatments will be assigned blindly to the experimenter by another individual in the laboratory. When comparing samples from different groups, samples from each group will be analyzed in concert, thereby preventing any biases that may arise from analyzing individual treatments on different days.

All data are expressed as mean ± SEM. Statistical significance was evaluated using Prism 6.0 software. One-way ANOVA was performed with post hoc Tukey multiple comparison test or Student t test applied, as appropriate for comparisons between groups. A p value <0.05 was considered significant.

Mucosal immunization with the adjuvant CDG elicits protective immunity in the systemic and mucosal compartments (25–30). However, how CDG promotes mucosal immunity is unknown. We set out to define the mechanisms by which CDG targets pulmonary DCs to generate lung mucosal immunity. Pulmonary DCs comprise three unique subsets: cDC1, cDC2, and moDCs (Fig. 1A). We have previously shown that moDCs promote lung mucosal T_FH cells and IgA responses (25, 31). To further evaluate the role of moDCs on lung mucosal immunity, we used CCR2^–/– mice, as CCR2-deficient mice have reduced numbers of moDCs in the lung at a steady state (Fig. 1B). Mice were immunized i.n. twice with CDG and the Ag (pneumococcal surface protein A) PspA at 2-wk intervals, and the immune responses were examined on day 28 (Fig. 1C). We found that CCR2^–/– mice had unaltered anti-PspA IgG in the serum but lacked anti-PspA IgA in the bronchoalveolar lavage fluid (BALF) (Fig. 1D). We then examined memory T_H1/2/17 responses in the spleen and lung from immunized CCR2^–/– mice by the ex vivo recall. Again, CCR2^–/– mice retained memory T_H responses in the spleen but not in the lung (Fig. 1E, 1F). Thus, CDG-induced lung mucosal adjuvant responses can be uncoupled from systemic responses. moDCs may be specifically needed for promoting adjuvant responses in lung mucosa.

moDC activation is required for efficient CDG-induced immune responses (22). Lung moDCs require the NF-κB transcription factor RelA for optimal moDC activation (Supplemental Fig. 1A) (25). We generated the RelA^fl/flCD11c^cre mice to examine the role of moDCs activation in CDG adjuvanticity in the lung mucosa. RelA^fl/flCD11c^cre mice had no Ag-specific IgA in the BALF despite maintaining Ag-specific IgG production in the serum (Supplemental Fig. 1B, 1C). Similar to the CCR2^–/– mice, RelA^fl/flCD11c^cre mice did not generate memory T_H responses in the lung, despite having normal memory T_H responses in the spleen (Supplemental Fig. 1D, 1E).

To further establish that moDCs were responsible for the lack of mucosal immune responses, we performed a monocyte adoptive cell transfer experiment. moDCs differentiate from Ly6C^hi monocytes (32). We isolated bone marrow Ly6C^hi monocytes from C57BL/6J mice and i.n. transferred the cells into RelA^fl/flCD11c^cre mice. The recipient RelA^fl/flCD11c^cre mice were then immunized with CDG/OVA. On day 14, the recipient RelA^fl/flCD11c^cre mice received a second dose of Ly6C^hi monocytes from C57BL/6J mice and were subsequently immunized with CDG/OVA (Supplemental Fig. 1F). Lung mucosal immune responses were analyzed on day 28. RelA^fl/flCD11c^cre mice that received WT monocytes restored the lung memory T_H responses and IgA production in the BALF (Supplemental Fig. 1G, 1H). Together, the data indicate that lung moDCs mediate CDG-induced lung mucosal–specific IgA and memory T_H cells.

How do moDCs promote mucosal IgA and memory T_H responses in the lung but is dispensable in the spleen? We hypothesized that moDCs might differentiate into a lung-specific subpopulation to induce lung mucosal vaccine responses.

We first examined moDCs in the lungs and mediastinal lymph nodes (medLNs) of WT mice following CDG/PspA immunization (Fig. 2A). We noticed that on day 14 postimmunization, the lungs contained a population of moDCs that expressed the transcription factor Bcl6 (Fig. 2B). This population was unique to the lung, as moDCs in the medLNs of CDG-immunized mice did not express Bcl6 (Fig. 2C). Bcl6 is upregulated in CXCR5⁺PD1⁺ T_FH cell as well as in B cells prior to entry into the germinal center (33, 34). Bcl6 expression in lung moDCs is lower than those observed in lung Bcl6⁺ B cells and CXCR5⁺PD1⁺ T cells (Fig. 2D). Bcl6 controls the chemokine receptor CXCR5 expression. The new Bcl6⁺ moDCs express CXCR5⁺, but the expression is lower than those observed in lung Bcl6⁺ B cells and CXCR5⁺PD1⁺ T cells (Fig. 2D).

To confirm that the Bcl6⁺ moDCs were indeed monocyte-derived, we isolated Ly6C^hi monocyte from CD45.1 mice. The monocytes were then i.n. transferred into C57BL/6J mouse. The recipient WT mice were immunized with CDG/PspA (Fig. 2E). On day 14 postimmunization, we identified CD45.1⁺ Bcl6⁺ moDCs (Fig. 2F). Thus, these Bcl6⁺ moDCs are indeed monocyte-derived. As alveolar macrophages (AMs) are also monocyte-derived, we examined AMs from the BALF of immunized mice. AMs did not express Bcl6 (Fig. 2G). Thus, the Bcl6⁺ monocyte-derived cells were not AMs.

Next, we examined the kinetics of moDCs differentiation in the lung during CDG immunization. The activation of moDCs, indicated by pRelA and CD80, peaked on days 6 and 9, respectively, postimmunization (Fig. 2H, 2I). Intriguingly, Bcl6⁺ moDCs differentiation had different kinetics from moDCs activation. Bcl6⁺ moDCs appeared early on day 2 and peaked on day 14, even when moDCs activation was waning (Fig. 2J). On day 14 postimmunization, ∼40% of lung moDCs expressed Bcl6 (Fig. 2J).

To understand the functional significance of Bcl6 expression in moDCs populations in vivo, we generated Bcl6^fl/flLysM^cre and Bcl6^fl/flCD11c^cre mice. Bcl6^fl/flLysM^cre mice delete Bcl6 in myeloid cells. The cDC population in the Bcl6^fl/flLysM^cre mice were unaltered (Fig. 3A, 3B). Bcl6^fl/flLysM^cre mice had expanded lung neutrophils and Ly6C^hi monocytes populations (Fig. 3C), as previously reported (35). However, Bcl6^fl/flLysM^cre mice had reduced lung moDCs (Fig. 3A, 3B), similar to the CCR2^–/– mice (Fig. 1B), suggesting that Bcl6 expression in myeloid cells is crucial for the differentiation of lung moDCs.

Next, Bcl6^fl/flLysM^cre mice were immunized with two doses of CDG/NP₆CGG at 2-wk intervals. Immune memory responses were determined on day 120 (Fig. 3D). Bcl6^fl/flLysM^cre mice did not produce IgA in the BALF (Fig. 3E) or lung memory T_H responses (Fig. 3G). Similar to the CCR2^–/– mice, spleen memory T_H responses and serum Ag-specific IgG in Bcl6^fl/flLysM^cre mice were comparable to the Bcl6^fl/fl mice 4 mo post–CDG immunization (Fig. 3F, 3H).

As monocytes develop into moDCs, the expression of CD11c is increased. To bypass the influence of Bcl6 on lung moDCs development, we generated Bcl6^fl/flCD11c^cre mice to ablate Bcl6 after moDCs have developed. Different from the Bcl6^fl/flLysM^cre mice, Bcl6^fl/flCD11c^cre mice had increased numbers of lung moDCs at steady state (Fig. 4A, 4B).

Bcl6^fl/flCD11c^cre mice will delete Bcl6 gene in CD11c⁺ cells, including cDCs and AMs. Watchmaker, et al. (36) first showed that CD103⁺/CD8⁺ cDC1, not Sirpα⁺IRF4⁺ cDC2, from gastrointestinal tract and spleen specifically express Bcl6, which is important for cDC1 development. Later, Zhang, et.al., found that Bcl6 protein expression is elevated in peripheral precDCs, especially cDC1 (37). We observed that Bcl6^fl/flCD11c^cre mice lack cDC1s in the lung (Fig. 4A, 4B), confirming that Bcl6 expression in DCs is required for lung cDC1 development. The numbers of lung cDC2 also reduced in Bcl6^fl/flCD11c^cre mice, although was not statistically significant (Fig. 4B).

cDC2 are essential for CDG mucosal adjuvanticity whereas cDC1 are dispensable (25). We examined the function of lung cDC2 in the Bcl6^fl/flCD11c^cre mice. Mice were treated (i.n.) with CDG for 16 h. Lung cDC2 activation was determined by CD86 and CCR7 expression. CD86 and CCR7 expression on CDG-activated lung cDC2 were comparable between Bcl6^fl/fl and Bcl6^fl/flCD11c^cre mice (Fig. 4C, 4D). However, cDC2 from the Bcl6^fl/flCD11c^cre mice had elevated basal CD86 and CCR7 expression (Fig. 4C, 4D). To further determine the function of cDC2 in Bcl6^fl/flCD11c^cre mice, we measured CDG adjuvanticity in the systemic compartments. Three months after CDG immunization, Bcl6^fl/flCD11c^cre mice had unaltered Ag-specific serum IgG (Fig. 4E) and memory T_H responses in the spleen (Fig. 4F). Intracellular Bcl6 stain confirmed the lack of Bcl6 upregulation in lung moDCs from Bcl6^fl/flCD11c^cre (Fig. 4G). The phenotypes of the conditional Bcl6 mice were summarized in Table I.

Next, Bcl6^fl/flCD11c^cre mice were immunized with CDG/PspA and the vaccine responses were examined 14 d later. Unlike Bcl6^fl/flLysM^cre mice, the Bcl6^fl/flCD11c^cre mice had unaltered lung IgA (Fig. 5A). However, Bcl6^fl/flCD11c^cre mice did not generate lung memory lung T_H responses (Fig. 5B). We examined CD4 T_RM cells, as these cells are a unique population of lung-specific memory T cells. CD4⁺ T_RM cells were characterized based on the expression of CD69 and CD49a. Bcl6^fl/flCD11c^cre mice did not generate CD4⁺ T_RM cells compared with the Bcl6^f/f mice following immunization (Fig. 5C). Thus, CDG induces lung memory T_H responses but not IgA in Bcl6^fl/flCD11c^cre mice.

To further demonstrate that Bcl6 expression in moDCs, not other CD11c⁺ cells, are responsible for the loss of lung memory T_H responses in Bcl6^fl/flCD11c^cre mice, we adoptively transferred WT moDCs into Bcl6^fl/flCD11c^cre mice (Fig. 5D). We sorted out lung moDCs frInaive C57BL/6 mouse and i.n. transferred the moDCs into Bcl6^fl/flCD11c^cre mice. Recipient Bcl6^fl/flCD11c^cre mice were immunized with CDG/NP₆CGG. We examined lung memory T_H responses on day 14 postimmunization. WT moDCs restored memory T_H1 responses and lung CD4⁺ T_RM cells in the immunized Bcl6^fl/flCD11c^cre (Fig. 5E, 5F). Together, the data indicate that Bcl6 expression in moDCs is required for generating lung memory T_H responses, but not IgA.

To further understand the role of Bcl6 expression in moDCs, we examined the ability of Bcl6⁺ moDCs to stimulate T cells. We first asked if lung Bcl6⁺ DCs were able to present Ag at the effector stage (day 14) in vivo. We immunized C57BL/6J mice with CDG plus a fusion protein of Eα peptide (aa52–68) with OVA (Eα-OVA) (Fig. 6A). The YAE mAb will detect cells that express the Eα peptide on I-A^b of MHCII. On day 14 postimmunization, Bcl6⁺ moDCs were able to present the Eα peptide on MHCII (Fig. 6B). Furthermore, the YAE⁺Bcl6⁺ moDCs were activated and marked by CD86 expression (Fig. 6B). Thus, Bcl6⁺ moDCs were mature APCs and were capable of activating T cells.

We next examined the ability of Bcl6⁺ moDCs to produce T cell–promoting cytokines (the Signal 3). We immunized C57BL/6J mice with CDG/OVA and characterized the moDCs on day 14. We measured the production of prototypic T_H1 (IL-12p70)-, T_H2 (IL-4)-, and T_H17 (IL-23)-promoting cytokines by moDCs (Fig. 6C–E). Bcl6⁺ moDCs were proficient at producing all these cytokines. Besides, Bcl6⁺ moDCs were capable of producing IL-10 and TGF-β1, two cytokines that are critical for the induction of T_RM cells (Fig. 6F, 6G) (38, 39). Together, the data indicate that the lung-specific Bcl6⁺ moDCs are mature on day 14 postimmunization and may promote T_H responses in the lung.

IL-21 is a major T_FH cell–inducing cytokine. Using an IL-21 reporter mouse, we found that lung moDCs were the main IL-21–producing lung DCs in the lung on day 14 postimmunization (Fig. 6H). However, the IL-21⁺ moDCs were Bcl6^– moDCs (Fig. 6H). CXCL13 is essential for germinal center formation. We found that on day 14 postimmunization, moDCs produced CXCL13 (Fig. 6I). However, most CXCL13⁺ moDCs were Bcl6^– moDCs (Fig. 6I). Thus, the lung-specific Bcl6⁺ moDCs do not produce IL-21 or CXCL13.

TGF-β1 are critical cytokines for the induction of T_RM cells (38, 39). We found that on day 14 postimmunization, lung moDCs from immunized Bcl6^fl/flCD11c^cre mice had decreased production of TGF-β1 (Fig. 7A). Lung moDCs from immunized Bcl6^fl/flCD11c^cre mice also produced less T_H-promoting cytokines IL-12p70, IL-4, and IL-23 (Fig. 7B–D). CCL20 (MIP3A), the CCR6 ligand, recruits lymphocytes and DCs to mucosal lymphoid tissues and is critical for inducing mucosal immune responses (40). We found that CCL20 production by lung moDCs was also reduced in CDG-immunized Bcl6^fl/flCD11c^cre mice on day 14 (Fig. 7E). Thus, Bcl6 expression in moDCs is required for the production of T cell–promoting cytokines and chemokines.

The Bcl6⁺ moDCs produce T_H cell– and T_RM cell–promoting cytokines, chemokines and are likely responsible for lung memory T_H responses. How does CDG induce the differentiation of lung Bcl6⁺ moDCs in vivo? We previously showed that the lung TNFR2⁺ cDC2 population induced lung memory T_H responses (Fig. 8A) (25). Furthermore, moDCs did not take up i.n. administered CDG (25). We hypothesized that the CDG directly activated TNFR2⁺ cDC2, which drives moDCs differentiation into Bcl6⁺ moDCs.

IRF4^fl/flCD11c^cre mice lack the cDC2 population and do not respond to CDG immunization (25). We performed an adoptive cell transfer to identify whether TNFR2⁺ cDC2 drive moDC differentiation in IRF4^fl/flCD11c^cre mice. Lung TNFR2⁺ cDC2 were sorted from WT mice and transferred i.n. into IRF4^fl/flCD11c^cre mice (Fig. 8B). Recipient IRF4^fl/flCD11c^cre mice were immunized with CDG/PspA. moDCs were characterized on day 14. IRF4^fl/flCD11c^cre that received no cells did not generate Bcl6⁺ moDCs (Fig. 8C). In contrast, IRF4^fl/flCD11c^cre received WT TNFR2⁺ cDC2 population had lung Bcl6⁺ moDCs (Fig. 8C).

How does the lung TNFR2⁺ cDC2 population stimulate the differentiation of Bcl6⁺ moDCs in the lung? We previously discovered that TNF is essential for CDG mucosal adjuvant activity (26). cDC2 is the main source for i.n. CDG-induced lung TNF (25). We hypothesize that TNFR2⁺ cDC2 cells promote the differentiation of Bcl6⁺ moDCs via TNF secretion. Indeed, the i.n. administration of recombinant soluble TNF (sTNF)/PspA-induced lung Bcl6⁺ moDCs in C57BL/6J mice on day 14 postimmunization (Fig. 8D).

To demonstrate that lung TNF acts on moDCs directly to promote the differentiation of Bcl6⁺ moDCs, we generated moDCs-targeting TNF fusion proteins. moDCs express the high-affinity FcR, FcγRI, also known as CD64 that is not found on cDCs or lymphocytes (41). FcγRI binds the Fc of IgG2a with the highest affinity (10⁸ M^–1), more than 1,000-fold higher than its next binding partner IgG2b-Fc (42). Indeed, i.n. administration of APC-conjugated mouse IgG2A was taken up exclusively by CD64⁺ lung cells, including MHCII^hi CD64⁺ moDCs (Fig. 8E). The CD64⁺ MHCII^low/int macrophages were also targeted. However, macrophages are dispensable for CDG mucosal adjuvanticity in vivo (25, 29). We thus generated a fusion protein TNF-Fc (IgG2A) to target TNF to moDCs (Fig. 8F).

CDG immunization generates sTNF and transmembrane TNF (tmTNF) (25). We fused TNF with the Fc portion of IgG2A to generate TNF-Fc (IgG2A), which represents sTNF. We also made a TNF_D221N/A223R-Fc (IgG2A) fusion protein that targets tmTNF to moDCs. The TNF_D221N/A223R mutant mimics tmTNF that binds only to TNFR2, not TNFR1 (43).

We used IRF4^fl/flCD11c^cre as CDG does not generate lung TNF in the IRF4^fl/flCD11c^cre (25), which facilitates the TNF-Fc (IgG 2A) complement experiment. We immunized IRF4^fl/flCD11c^cre mice with CDG/NP₆CGG with tmTNF or sTNF (Fig. 8G). On day 14, we examined lung moDCs (Fig. 8G). The addition of sTNF, not mTNF, generated Bcl6⁺ and CCL20⁺ moDCs in the lung (Fig. 8H). In contrast, the addition of mTNF generated CXCL13⁺ moDCs (Fig. 8H). We propose that sTNF directly induces moDCs differentiation into Bcl6⁺ moDCs.

moDCs have a universal presence at the site of inflammation and can promote CD4⁺ and CD8⁺ T cell responses in vivo. Yet, we know very little about how moDCs achieve this plasticity in vivo. Consequently, we have few approaches to control this common and versatile APC population in vivo during inflammation. The most exciting discoveries in this report were the identification of Bcl6⁺ moDCs responsible for lung memory T_H in vivo and a potential method to induce Bcl6⁺ moDC in vivo. Indeed, we recently reported that moDCs-targeting TNF fusion proteins enhanced CDG adjuvanticity in 2-y-old mice (44).

We used the mucosal adjuvant CDG system to study moDCs in vivo. CDG activates mainly the STING pathway (26) and does not induce lung damage (29). Its simplicity uncouples the lung mucosal versus systemic, lung memory T_H versus lung IgA vaccine responses, thus facilitates the extraction of the underlying mechanisms. Lung TNF generates Bcl6⁺ moDCs. TNF is a hallmark of inflammation. We propose that Bcl6⁺ moDCs exist at the site of all inflammation and may play a critical role in driving T_H cell–mediated inflammatory responses.

The role of Bcl6 in DCs is not well understood. Ohtsuka et al. (45) first reported that cDCs in the spleen requires Bcl6 expression for development in total Bcl6 knockout mice. However, Watchmaker, et al. showed that Bcl6 was the transcriptional factor specifically controlling cDC1 development in the gut, lymph nodes, and spleen (36). Zhang, et al. also found that cDC1 have the highest Bcl6 expression (37). In this study, we found that the Bcl6^f/fCD11c^cre mice lack cDC1 in the lung but retain functional cDC2 confirming that Bcl6 is specifically required for cDC1 differentiation. We also found that Bcl6^f/fLysM^cre had reduced moDCs similar to the CCR2^–/– mice. Future study is needed to understand the mechanism by which Bcl6 controls cDC1 and moDCs differentiation.

How do Bcl6⁺ moDCs promote the induction of lung CD4⁺ memory T cells? Lung CD4⁺ T cell memory formation depends on the re-engagement of CD4⁺ T cells with APC at their effector stage (Signal 4) (9, 11). The lung Bcl6⁺ moDCs are a differentiated moDCs population that appears in the lung after the initial priming and peaked on day 14. We propose that the differentiated Bcl6⁺ moDCs are the long-thought lung APCs that provide the “Signal 4” during the effector stage to induce lung memory CD4⁺ T cells. Indeed, lung Bcl6⁺ moDCs bore Ags and expressed coactivators on day 14 postimmunization.

The recall response is a hallmark of immune memory cells, which establishes Ag-specificity and distinguishes T_H1, T_H2, and T_H17 memory CD4⁺ T cells. We used lung ex vivo recall assays to determine the total lung CD4⁺ memory T cells, including memory T_H1, T_H2, and T_H17 cells induced by immunization. We reasoned that although the lung memory T_H responses are mainly from lung CD4⁺ T_RM cells, effector memory T cells, or other unknown memory CD4 T cells in the lung may also contribute to the total lung memory T_H responses (46). Furthermore, growing evidence suggests that T_RM cells, even at the same tissue site (e.g., lung), are heterogeneous (2). Our goal is to generate Ag-specific protection against lung infections for which all memory CD4⁺ T cells in the lung contribute. We believe measuring total lung memory T_H1/2/17 cells by ex vivo lung recall is a better indicator for the vaccine efficacy in lung mucosa than enumerating total cell numbers of lung CD69⁺CD49a⁺ CD4 T_RM cells.

In summary, moDCs can differentiate into the new Bcl6⁺ lung moDCs to promote memory CD4⁺ T_H cells in the lung mucosa. Targeted activation of moDCs by sTNF fusion protein could be a valuable method to enhance vaccine responses in vulnerable populations (e.g., the elderly) (44).

We thank the Center for Immunology and Transplantation at the University of Florida for the assistance.

L.J. and S.M. are coinventors on a patent (PCT/US19/53548) on moDCs-targeting TNF fusion proteins. The other authors have no financial conflicts of interest.