C57BL/6 mice were purchased from Orient Bio (Seongnam, Korea), and OT-1, OT-2, and GM-CSF^−/− mice were from the Jackson Laboratory (Bar Harbor, ME, USA). FLT3^−/− mice were a gift from Professor Jae-Hoon Choi in Hanyang University. All animals were maintained and bred in specific pathogen-free facilities in the Yonsei University College of Medicine. Female mice between 8 and 12 wk of age were used in all experiments. Animal care and experiments were performed according to the guidelines and protocol (#2016-0040) approved by the Institutional Animal Care and Use Committees of the Yonsei University College of Medicine.

Fluorochrome-conjugated mAbs to CD3, CD19, Ly6G, intercellular adhesion molecule 2 (ICAM2), MHCI, MHCII, CD11b, CD11c, CD115, XCR1, CD172a, 33D1, F4/80, CD14, CD206, CD24, CD8, CD4, Va2, CD25, CD69, CD44, CD62L, IFN-γ, IL-4, IL-17a, forkhead box P3 (FOXP3), CX3CR1, CD64, CD226, CD68, CD207, CD103, CD209a, CD209b, Ly6C, Clec9a, CD205, CD117, CD135, B220, PD-1, PD-L1, CCR2, CD40, CD80, CD86, CCR7, sialic acid binding Ig-like lectin F, and high-affinity IgE receptor were purchased from BioLegend (San Diego, CA, USA). PE-conjugated anti-CD207 and PE-Cyainine7-conjugated anti-proto-oncogene tyrosine-protein kinase MER were purchased from eBioscience (San Diego, CA, USA). CellTrace™ CFSE or CellTrace™ violet (CTV) Cell Proliferation Kits and LIVE/DEAD™ Fixable Yellow or Far Red Dead Cell Stain Kits were purchased from Thermo Fisher Scientific Korea (Seoul, Korea) and were used according to the instructions provided by the manufacturers. Ovalbumin (OVA; Sigma-Aldrich, St. Louis, MO, USA), FITC-conjugated OVA (FITC-OVA; Thermo Fisher Scientific Korea), and Fluoresbrite® yellow green microspheres (YGM) 1.00 μm beads (Polysciences, Warrington, PA, USA) were purchased respectively. LPS and all trans-retinoic acid were purchased from Sigma-Aldrich; human TGF-β1 from Peprotech (Rocky Hill, NJ, USA); mouse IL-4, mouse IL-6, and anti-mouse IFN-γ neutralizing mAb from BioLegend; PMA and ionomycin from LC Laboratories (Woburn, MA, USA). Cells were cultured in DMC7 medium (23) composed of DMEM containing L-glutamine, high glucose, and pyruvate (GE Healthcare Life Sciences, Logan, UT, USA) and 7% fetal calf serum (GE Healthcare Life Sciences) supplemented with 1× solutions of non-essential amino acids (GE Healthcare Life Sciences) and antibiotic-antimycotic (GE Healthcare Life Sciences). Mouse M-CSF, GM-CSF, and FLT3L were produced and purified in house as described previously (232425).

Peritoneal exudate cells were harvested by lavage with 5 ml cold PBS containing 3% fetal calf serum. Single cell suspensions were blocked with neat culture supernatant of 2.4G2 (Fc blocking) mAb hybridoma and washed with FACS buffer (2% FCS, 2 mM EDTA, 0.1% sodium azide). Then, the samples were incubated with the appropriate mixture of fluorochrome-conjugated mAbs and dead cell staining dye for 30 min at 4°C and washed twice with FACS buffer. The mixture of mAbs against lineage (Lin) makers of differentiated cells consisted of anti-CD3, anti-CD19, and anti-Ly-6G. Surface-stained cells were resuspended in Fixation/Permeabilization solution (BioLegend) prior to staining intracellular cytokines. Each sample was analyzed with FACSVerse™ and LSRFortessa™ flow cytometer (BD Biosciences, San Jose, CA, USA) or sorted with FACSAria™ II cell sorter (BD Biosciences) at the Flow Cytometry Core of the Yonsei Biomedical Research Institute in the Yonsei University College of Medicine. Flow cytometric data were analyzed with FlowJo software (FlowJo, Ashland, OR, USA).

Mice were injected intraperitoneally (i.p.) with 200 μg of FITC-OVA or 200 μl of 0.0027% YGM beads for 1 h, and peritoneal exudate cells were harvested as described above. Then, peritoneal exudate cells were stained for surface markers with fluorochrome-conjugated mAbs followed by the analysis with a flow cytometer as described above. Geometric mean fluorescence intensity index of the FITC channel is defined as below.

Mice were injected i.p. with 1.5 mg of soluble OVA for 1 h, and peritoneal exudate cells were harvested as described above. Then, peritoneal exudate cells were stained with fluorochrome-conjugated mAbs before purifying OVA-laden APCs by sorting with FACSAria™ II cell sorter. Splenic T cells from OT-1 and OT-2 mice were enriched by excluding CD19⁺, CD49b⁺, MHC II⁺, CD11b⁺, F4/80⁺, B220⁺, and CD4⁺ (for OT-1) or CD8⁺ (for OT-2) splenocytes using appropriate biotinylated Abs and anti-biotin Dynabeads^® (ThermoFisher Scientific Korea) and labeled with 5 mM CFSE or CTV Cell Proliferation Kits for 10 min at 37°C. Then, CFSE or CTV labeled T cells were added into round bottom 96-well plates at 2.5×10⁴/well and mixed with isolated APCs at an indicated APC: T cell ratio in the media containing 57.2 μM β-mercaptoethanol (Sigma-Aldrich). After co-culture with APCs for 3–4 days, the proliferation of live T cells was evaluated by the dilution of CFSE or CTV (626). For in vitro T-cell polarization assays, the mixture of APCs and T cells (APC:T cell = 1:10) included additional cytokines, neutralizing Abs, and reagents as follows: Th0 (media alone), Th1 (1 μg/ml LPS) (27), Th2 (10 ng/ml IL-4) (2829), Th17 (3 ng/ml TGF-β, 20 ng/ml IL-6) (30), iTreg (3 ng/ml TGF-β, 1 nM all-trans retinoic acid (ATRA); (3132), type 0 cytotoxic T cell (Tc0) (media alone), Tc1 (1 μg/ml LPS) (33), Tc2 (10 μg/ml anti-IFN-γ, 20 ng/ml IL-4) (33), Tc17 (10 μg/ml anti-IFN-γ, 5 ng/ml TGF- β, 20 ng/ml IL-6) (33). After culture for 3–4 days, cells were stimulated with PMA (12 nM), ionomycin (1 μM), and brefeldin A (5 μg/ml) for 4 h before analysis.

Mice were subcutaneously injected daily with 10 μg of either mouse M-CSF, GM-CSF, or FLT3L for up to 5 days. The effect of mouse GM-CSF was significant enough after 3 consecutive daily injections. Peritoneal exudate cells were harvested and analyzed 24 h after the last injection.

Peritoneal exudate cells were prepared, stained, and subjected to flow cytometric isolation with FACSAria™ II cell sorter as described above. Then, we isolated RNA samples from the respective Lin^-MHCII⁺ peritoneal cell subsets using MiniBEST Universal RNA Extraction kit (TaKaRa Bio, Shiga, Japan). Subsequent procedures and analysis were performed by Macrogen, Inc (Seoul, Korea) as follows: reverse transcription of mRNA and generation of cDNA libraries were carried out with SMARTer Ultra low input RNA library kit and sequenced with Illumina NovaSeq. The relative abundances of gene were measured in Fragments Per Kilobase of exon per Million fragments mapped using StringTie. The result was statistically analyzed to find differentially expressed genes using the estimates of abundances for each gene in samples. Multidimensional scaling method was used to visualize the similarities among samples. The larger the dissimilarity between 2 samples, the further apart the points representing the experiments in the picture should be. We applied to the Euclidean distance as the measure of the dissimilarity. Hierarchical clustering analysis also was performed using complete linkage and Euclidean distance as a measure of similarity to display the expression patterns of differentially expressed transcripts which are satisfied with |fold change|≥2. All data analysis and visualization of differentially expressed genes was conducted using R 3.5.1 (https://www.r-project.org). RNA-Seq data have been deposited in NCBI's Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE130424 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE130424).

Experiments with multiplicate samples were presented as mean±SEM from at least 3 independent experiments. Statistical comparisons between different groups were analyzed using unpaired Student's t-test using SigmaPlot (Systat Software, San Jose, CA, USA). Statistical significance is denoted by the p values equal or below 0.05 (^*), 0.01 (^**), 0.001 (^***), and 0.0001 (^****). Data were plotted for graphs with PRISM6 (GraphPad Software, La Jolla, CA, USA).

Myeloid mononuclear cells of the mouse peritoneal exudate cells were analyzed for the expression of various surface markers of DCs and macrophages using multi-parameter flow cytometry (Fig. 1A). Peritoneal myeloid mononuclear cells were recognized following the exclusion of other hematopoietic cells expressing CD3, Ly6G, and CD19. Then, the cells with ICAM2⁻MHCII⁻ phenotypes (peritoneal eosinophils, mast cells, and monocytes) were further excluded (17). LPMs were identified as ICAM2⁺low-level MHC class II (MHCII^lo) population (P-I in Fig. 1A), which also expressed F4/80^hi (Fig. 1B) and comprised the vast majority of peritoneal myeloid mononuclear cells (Fig. 1C).

Unlike the most abundant LPMs, the other peritoneal myeloid mononuclear cells expressed high levels of MHC II, a typical feature of APCs. These MHCII⁺ cells were further divided into 3 populations based on the expression of CD11c and CD115: CD11c⁺CD115⁻ DCs (P-II), CD11c⁻CD115⁺ SPMs (P-III), and CD11c⁺CD115⁺ population (P-IV) (Fig. 1A). The identity of CD11c⁺CD115⁺ P-IV subset in the peritoneal cavity has been ambiguous and classified as either DCs (162021) or SPMs (1922) because it expresses both markers of DCs (CD11c) and macrophages (CD115). To clarify the nature of P-IV cells, we further examined the levels of various markers expressed by DCs and macrophages, including F4/80, XCR1, 33D1, CD11b, CD14, CD24, CD172α, CD206, CD226, CD301b, and others (Fig.1B; not all data shown). Similar to DCs which consisted of a minor subset of XCR1⁺CD24⁺ DC1 and a major subset of XCR1⁻CD24⁻ DC2, the result clearly indicated that P-IV subset was also composed of 2 phenotypically heterogeneous cells (Fig. 1B). Therefore, we further classified P-IV subset into 2 subpopulations, i.e., CD14⁻CD206⁻ P-IV.I and CD14⁺CD206⁺ P-IV.II. The number of P-IV.II cells was almost twice the number of P-IV.I cells but similar to the number of DCs in the peritoneal cavity (Fig. 1C).

DCs and macrophages can be distinguished by their peculiar morphological characteristics (62134). Mouse peritoneal exudate cells were separated into 5 subsets of myeloid mononuclear cells, as described above, by flow cytometric sorting and were then cultured overnight before examination. It was obvious that, out of these 5 populations, 3 populations, i.e., LPMs, SPMs, and P-IV.II cells exhibited a typical macrophage morphology, or were observed as large adherent cells with irregular shapes (Fig. 2A). Meanwhile, DCs and P-IV.I cells displayed a dendritic morphology, or were observed as non-adherent cells with long cellular processes (Fig. 2A).

DCs and macrophages are also distinguished by their different ability to take up Ags (1734). Endocytic and phagocytic capacities of the 5 myeloid mononuclear cell subsets we classified above were evaluated within the peritoneal cavity. We injected either soluble (FITC-OVA) or particulate (YGM) forms of fluorescently labeled Ags into the peritoneal cavity. An hour after injection, endocytosis of FITC-OVA or phagocytosis of YGM beads by the 5 myeloid mononuclear cell subsets were assessed respectively and compared to one another (Fig. 2B and C). As expected, LPMs were the most active in both endocytosis and phagocytosis. Compared to LPMs, the other populations were much poor in endocytosis and phagocytosis. However, the activities of both endocytosis and phagocytosis by SPMs and P-IV.II cells were similar and significantly higher than those by DCs and P-IV.I cells (Fig. 2B and C).

A key functional difference between DCs and macrophages is determined by the outcome of Ag uptake; i.e., DCs are efficient in processing and presenting Ags to responding T cells whereas macrophages are much less efficient (3435). To evaluate the difference in Ag-presenting ability, soluble OVA protein was injected into the peritoneal cavity, and the peritoneal exudate cells were harvested after an hour and were sorted by flow cytometry into the subsets of myeloid mononuclear cells. Then, the subsets of cells isolated from the OVA-pulsed peritoneal cavity were assessed for their ability to stimulate naive T cells in vitro using OVA-specific OT-1 (CD8) and OT-2 (CD4) T cell receptor transgenic mice (3637). First, OVA-pulsed LPMs, DCs, SPMs, and P-IV cells were separated and examined for their ability to stimulate CD8⁺ OT-1 and CD4⁺ OT-2 T cells. Although all the subsets of peritoneal myeloid mononuclear cells showed similar levels of MHC I expression (Supplementary Fig. 1), it was evident that DCs were most capable of cross-presenting OVA Ag and thus strongly inducing the proliferation of CD8⁺ OT-1 T cells (Fig. 3A, left panels). P-IV cells were also efficient in stimulating OT-1 T cells although weaker than DCs. Meanwhile, both LPMs and SPMs were poor in cross-presenting OVA Ag to CD8⁺ OT-1 T cells. Similarly, CD4⁺ OT-2 T cells were used to evaluate the Ag presentation of OVA via MHC II molecules (Fig. 3A, right panels). Among the MHCII⁺ subsets, DCs were the most potent in stimulating responding OT-2 T cells and P-IV cells were weaker but SPMs were very poor. However, LPMs were completely incompetent to induce the proliferation of OT-2 T cells even in the highest APC to T cell ratio, likely because of the MHCII^−/lo phenotype of LPMs. These data imply that P-IV cells might contain a DC-like subpopulation. Therefore, we compared the Ag-presenting ability between P-IV.I and P-IV.II subsets (Fig. 3B). To our surprise, P-IV.I cells were able to stimulate both CD8⁺ OT-1 and CD4⁺ OT-2 T cells as efficiently as or better than peritoneal DCs. Meanwhile, like other peritoneal macrophage subsets, P-IV.II cells were unable to induce the proliferation of responding T cells.

We investigated the functional difference between 2 robust APCs, i.e., DCs and P-IV.I cells by analyzing the activation status of proliferated T cells following co-culture with the respective APCs. Activated T cells are considered to upregulate CD25, CD69, and CD44 and to downregulate CD62L (35). Compared to peritoneal DCs, P-IV.I cells induced both OT-1 and OT-2 T cells with less activated phenotypes (Fig. 4A). In particular, Ag-specific T cells stimulated by P-IV.I cells contained fewer fractions of CD25⁺, CD69⁺, or CD62L⁻ cells than those stimulated by peritoneal DCs.

DCs can modulate the type of immune response through mediating the activation and differentiation of naïve T cells, and the type of functionally differentiated T cells are classified on the profiles of cytokines they secrete (383940414243). To check whether peritoneal DCs induce a specific type of T cell response, we performed an intracellular staining of cytokines. When OVA-laden peritoneal APCs were cultured with CD8⁺ OT-1 T cells, DCs were able to generate a higher frequency of IFN-γ producing Tc1 cells than P-IV.I cells (Fig. 4B). However, both DCs and P-IV.I cells induced neither IL-4 producing Tc2 nor IL-17a producing Tc17 cells. In the co-culture with CD4⁺ OT-2 T cells, a higher frequency of IFN-γ⁺ Th1 cells was produced by DCs than P-IV.I cells, but IL-4 and IL-17a, the signature cytokines of Th2 and Th17 cells, were not detected (Fig. 4C). It is known that when Ag-laden APCs and responding naïve T cells are cultured under specifically conditioned media, T cells are preferentially converted to particular types of Tc or Th cells (28303344). Hence, we tested the culture of peritoneal APCs and naïve T cells under differently conditioned media (Supplementary Fig. 2). With the addition of LPS, both DCs and P-IV.I cells produced comparable levels of IFN-γ⁺ Tc1 and Th1 cells. Similarly, both DCs and P-IV.I cells exhibited a comparable ability to induce IL-4⁺ Tc2/Th2 and IL-17a⁺ Tc17/Th17 under pertinent conditions.

Our data indicated that T cells stimulated by P-IV.I cells possessed less activated phenotypes than those stimulated by DCs. Therefore, we examined whether DCs and P-IV.I cells had different ability to promote immunological tolerance by generating iTreg cells. DCs are known to induce FOXP3⁺ iTreg cells in vitro when cultured with naïve T cells in the media conditioned with TGF-β and/or ATRA (3132). Both peritoneal DCsand P-IV.I cells rarely produced iTreg cells in the culture medium without any supplements (Fig. 4D). However, when TGF-β and/or ATRA were added to the culture media, both DCs and P-IV.I cells strongly iTreg cells. Especially when both TGF-β and ATRA were included in the culture, P-IV.I cells produced iTreg cells more potently and in a higher frequency than DCs. With both TGF-β and ATRA supplemented, OVA-laden P-IV.1 cells could make nearly 50% of proliferated OT-2 T cells differentiate into iTreg cells, which was a frequency almost twice as higher than peritoneal DCs iTreg cells in the same condition (Fig. 4D). Interestingly, under these conditions, more of the proliferated high level of CD25 OT-2 T cells induced by P-IV.I cells were FOXP3⁺ than those induced by peritoneal DCs (Fig.4D).

Since P-IV.I cells were potent peritoneal APCs expressing significant levels of CD115 (Supplementary Fig. 3), the receptor for a key hematopoietic cytokine M-CSF, we investigated the role of different hematopoietic cytokines in the development of myeloid mononuclear cells in the peritoneal cavity. First, we examined the effect of M-CSF, GM-CSF, and FLT3L by injecting each hematopoietic cytokine. After 3 to 5 daily injections of the respective cytokines to mice, peritoneal exudate cells were harvested and analyzed (Fig. 5A). Interestingly, the administration of M-CSF to mice did not cause any changes in the numbers of not only CD115⁻ DCs and LPMs but also CD115⁺ SPMs, P-IV.I, and P-IV.II cells. Meanwhile, the injection of either GM-CSF or FLT3L augmented the numbers of DCs and P-IV.I cells, whereas the injection of only GM-CSF, not FLT3L, increased the numbers of LPMs, SPMs, and P-IV.II cells. Especially, the increase in the number of P-IV.II cells by the injection of GM-CSF was quite dramatic (Fig. 5A).

To further verify the roles of hematopoietic cytokines tested above, we also evaluated the number of subpopulations in peritoneal myeloid mononuclear cells from FLT3 knockout (KO) and GM-CSF KO mice (Fig. 5B). In GM-CSF KO mice, no significant changes were observed in the numbers of all subpopulations. In FLT3 KO mice, however, the numbers of LPMs, DCs, and P-IV.I cells decreased while the number of SPMs increased. Unlike the other subpopulations, the number of P-IV.II cells was not affected in FLT3 KO mice.

To systematically analyze the differences and similarities among the 5 subsets of MHCII⁺ peritoneal myeloid mononuclear cells, including DC1 and DC2 subpopulations (Fig. 6A), RNA samples were prepared from these peritoneal cell subsets and subjected to global mRNA-seq. The main principal components analysis identified 4 distinct clusters among the 5 subsets of MHCII⁺ peritoneal myeloid mononuclear cells (Fig. 6B); peritoneal DC1s, DC2s, and P-IV.I cells formed separate clusters, but SPMs and P-IV.II cells clustered indistinguishably close to each other. Meanwhile, P-IV.I cells clustered closer to DC2s than either DC1s or SPMs (Fig. 6B).

As expected from the result of principal components analysis above, the gene expression profile of P-IV.II cells was highly similar to that of SPMs. Especially, most of the macrophage core signature genes (45) were induced in both SPMs and P-IV.II cells (Fig. 6C). In the meantime, P-IV.I cells were found to upregulate the expression of a significant number of conventional DC core signature genes (46) as similarly to peritoneal DC2s (Fig. 6D). To better characterize the relationship between P-IV.I cells and DC subsets, we also analyzed the different expressions of transcription factors involved in DC development. The expression of transcription factors involved in DC2 development such as RELB, IFN regulatory factor 4 (IRF4), and neurogenic locus notch homolog protein 2 (NOTCH2) (46) were elevated in both DC2s and P-IV.I cells compared to DC1s (Fig. 6E). The transcription factors involved in DC1 development such as BATF3 and IRF8 (46) were expressed more in peritoneal DC1s than DC2s or P-IV.I cells (Fig. 6E). Then, we also examined the expression profile of the genes associated with monocytes and/or monocyte-derived DCs (MoDCs) (4748). None of the MHCII⁺ peritoneal myeloid mononuclear cell subsets expressed detectable levels of Ly6C, a monocyte marker, according to RNA-Seq data (data not shown). MoDCs in LPS-inflamed lymph nodes were found to express CD14, CD206, and CD209a (4748). It is noticeable that both DC2s and P-IV.I cells expressed significantly elevated levels of CD209a compared to the other subsets, but the expression of CD14 and CD206 was not particularly upregulated in P-IV.I cells (Fig. 6F).