Continuous retinoic acid induces the differentiation of mature regulatory monocytes but fails to induce regulatory dendritic cells

Given that RA is a regulator of mucosal immunity and influences myelopoiesis, we hypothesized that RA would induce a population of mature MC_regs. Day 6–7 BM cells differentiated with GM-CSF in the presence of RA were able to suppress the proliferation of responder immune cells and this suppression was markedly greater than either control or E3 treated cells (Figure 1A). The ability of RA differentiated cells to suppress proliferation was apparent regardless of whether responder immune cells were stimulated with either peptide or anti-CD3. Interestingly, cells treated with E3 suppressed proliferation after stimulation with peptide but not anti-CD3 (Figure 1A). We next determined whether the RA differentiated cells remained regulatory when exposed to the inflammatory stimulus LPS. Figure 1B shows that RA differentiated cells maintained their ability to suppress proliferation even after exposure to LPS challenge and that this was present following stimulation of co-cultures with either peptide or anti-CD3. This effect was entirely lost in E3 treated cells. These results suggest that RA differentiated cells are more potent and stable than E3 differentiated cells and that RA differentiated cells maintain their regulatory ability following exposure to an inflammatory stimulus.

Given that increased IL-10 is seen in E3 DC_regs[35] and other MC_reg populations [50, 55] we next evaluated whether RA induced an increase number of IL-10⁺ cells. Figure 1C shows that RA differentiated cells had an increased percentage of IL-10-producing cells compared to either media or E3 control cells. We next evaluated whether RA differentiated cells could increase T_reg numbers. We found that RA differentiated cells were able to induce a significant increased percentage of FoxP3⁺ cells following a 5 day culture with naïve immune cells (Figure 1D). Cells differentiated in vitro in the presence of E3 failed to significantly increase either IL-10⁺ cells or induce T_reg cells (Figures 1C, D). These results show that RA differentiated cells suppressed the proliferative abilities of responder immune cells and induced FoxP3⁺ (T_reg) cells.

To determine whether these RA differentiated cells were mature, we evaluated the cell surface expression of maturation markers CD80, CD86 and MHC class II and inhibitory markers PD-L1 and PD-L2. RA differentiated cells demonstrated an increased percentage of CD80⁺, CD86⁺ and MHC class II⁺ (Figure 2A), indicating that an increased proportion of the cells were mature and/or activated in comparison to E3 or control cells. Additionally, there were increases in the mean fluorescence intensity (MFI) of CD80, CD86 and MHC class II in RA differentiated cells as depicted in Figures 2C and D, indicating that the relative expression levels on a per cell basis were increased in RA differentiated cells. Although E3 differentiated cells had mildly increased expression levels of CD80, CD86 and MHC class II, RA differentiated cells had consistently higher levels than either E3 differentiated or control cells. To confirm that RA differentiated cells demonstrated an “activated regulatory” phenotype as previously described for E3, we evaluated the expression of inhibitory co-stimulatory molecules PD-L1 and PD-L2 [35]. RA increased the percentage of PD-L1⁺ cells (but not PD-L2⁺) (Figure 2B) and the MFI of both PD-L1 and PD-L2 (Figure 2C, D) compared to E3 or media controls. These results demonstrate that during differentiation RA induces a population of mature activated MC_regs that suppress the proliferation of responder immune cells even in the face of inflammatory challenge. Additionally, our data shows that although both RA and E3 may induce MC_regs which suppress proliferation (Figure 1A) RA MC_regs appear to have superior regulatory abilities compared to E3 MC_regs.

The in vitro differentiation of bone marrow cells with GM-CSF is a commonly used protocol to produce large numbers (>80%) of highly enriched CD11c⁺ DCs [38, 56] that, as a population, are considered immature DCs. However, our data demonstrated that while approximately 80-90% of the cells were CD11c⁺, the remaining 10-20% were CD11c^- but still CD11b⁺ (Additional file 1: Figure S1A). To determine whether the MC_regs induced by RA were DCs, we purified CD11c⁺ cells from day 7 differentiated cells and cultured them with responder immune cells. Although RA induction of mucosal “DC_regs” have been described [9, 36, 57], we found that RA-treated CD11c⁺ cells were not the suppressive cell population (Figure 3A). In all experiments, RA-treated CD11c⁺ cells failed to suppress proliferation and had variable to no effect on proliferation with some experiments actually demonstrating enhanced proliferation (data not shown). Phenotypic evaluation of these CD11c⁺ cells showed no difference in percentage (Figure 3B) or expression levels of CD80, CD86, MHC class II, PD-L1 and PD-L2 compared to media controls. To determine the source of the suppressive MC_regs, we evaluated the CD11c^- population and found that the RA CD11c^- cells suppressed proliferation of responder cells (Figure 3C). These CD11c^- cells had a marked (>30%) increase in the percentage of CD80⁺, CD86⁺, MHC class II⁺ and PD-L1⁺ cells (with no differences in PD-L2⁺ cells) (Figure 3D) when differentiated with RA, consistent with an activated regulatory phenotype in these cells described previously [35]. In contrast, levels of CD80, MHC class II and PD-L1 did not change, remaining consistently high (>80%) in RA versus control MCs. (Figure 3B). These data suggest that RA present during GM-CSF differentiation increased an activated regulatory phenotype in the CD11c^- (non-DC) populations.

Both differentiated and precursor populations within the bone marrow are predominantly but not completely CD11b⁺ (>90%) (Additional file 1: Figure S1A). To definitively isolate the effects of CD11b⁺ CD11c^- cells, we serially purified CD11b⁺ cells from the CD11c^- fraction and evaluated their phenotype and function. As expected, the increases in the percentage of CD80⁺, CD86⁺, MHC class II⁺ and PD-L1⁺ cells seen in Figure 3D was also seen in the CD11b⁺ CD11c^- serially purified population (Figure 4A). We then went on to evaluate the ability of these cells to influence CD4 and CD8 responses. We found that the CD11b⁺ CD11c^- population was able to suppress the proliferation of responder immune cells (Figure 4B) and could modify the cytokine profile of T cells. The proliferating CD4⁺ responder immune cells cultured with RA CD11b⁺ CD11c^- cells were also shown to have reduced expression of IL-17 IFN-gamma (Figure 4C) and IL-10 (Additional file 2: Figure S2) but no change in IL-4 production as determined by intracellular cytokine staining (Figure 4C). Intracellular IL-10 and FoxP3⁺ cells were also increased as expected (Additional file 2: Figure S2A and S2B, respectively). We also evaluated the ability of RA CD11b⁺ CD11c^- cells to influence CD8⁺ T cell responses. Figure 4D demonstrates reduced cytotoxicity in CD8⁺ T cells cultured with RA CD11b⁺ CD11c^- cells. Taken together, these results suggest that RA induced an activated regulatory population of CD11b⁺ CD11c^- cells that were able to suppress both CD4⁺ and CD8⁺ adaptive immune responses.

Although used primarily to induce large numbers of DCs, differentiation with GM-CSF can potentially promote the differentiation of a mixture of granulocytes, monocytes, macrophages and DCs [42, 54, 56, 58‐60] In our GM-CSF cultures, we found that Ly-6G⁺ granulocytes were no longer present in CD11b⁺ cells at day 7 of differentiation (Additional file 1: Figure S1B), indicating that granulocytes were not responsible for the suppression seen [61, 62]. To determine whether monocytes were present and may be responsible for the suppressive effects, we evaluated day 7 non-adherent cells sorted based on their relative expression of the monocyte marker Ly-6C. Ly-6C expression levels have been shown to correlate with cellular function and maturation level where Ly-6C^high monocytes are inflammatory and Ly-6C^low monocytes are steady-state or regulatory [7, 63]. Figure 5A shows that the presence of RA during differentiation increased the percentage of cells expressing low to intermediate levels of Ly6C. To determine whether the increase in these cells was responsible for the suppression seen in the CD11b⁺ CD11c^- population, we sorted cells based on Ly-6C^low, Ly-6C^intermediate, and Ly-6C^high expression patterns. Figure 5B demonstrates that both Ly-6C^low and Ly-6C^intermediate cells were able to suppress up to a 6-fold decrease in proliferation of responder cells following antigenic stimulation while Ly-6C^high cells failed to influence peptide-specific proliferation. Similarly, Ly-6C^low, Ly-6C^intermediate cells maintained their ability to suppress proliferation (Figure 5C) even when co-cultures were stimulated with LPS. In contrast, Ly-6C^high actually significantly increased the proliferation of responder immune cells (Figure 5C) following stimulation with LPS. These results demonstrate that RA Ly-6C^low and Ly-6C^intermediate cells are the suppressive population and are able to maintain suppressive abilities even in the presence of inflammatory (LPS) challenge. Phenotypically, RA Ly-6C^low cells showed the most marked increase in the percentage of PD-L1⁺, as well as, CD86⁺ and MHC class II⁺ cells, with over 90% of the Ly-6C^low cells expressing PD-L1 (Figure 6). A similar but less dramatic phenotype was seen in the Ly-6C^intermediate cells (data not shown). Taken together, these data show that RA induces a small but potent population of CD11b⁺ CD11c^- Ly-6C^{low/intermediate} MC_regs consistent with an activated regulatory monocyte phenotype that are able to suppress immune cell proliferation.

C57BL/6 (H-2^b) mice (4–8 wk old), C57BL/6-Tg (TcraTcrb)425Cbn/J, C57BL/6-Tg(Tcra2D2,Tcrb2D2)1Kuch/J and reporter Foxp3EGFP (B6.Cg-Foxp3 ^tm2Tch) ) were purchased from Jackson Laboratories (Bar Harbor, ME, USA) or bred in-house. Mice were housed five per cage and maintained on a 12 hr. light/dark cycle, maintained under specific pathogen-free conditions and were housed and cared for according to the institutional guidelines of the Ohio State University’s Institute for Animal Care and Use Committee.

EG7 and EL7 (kindly provided by P. Boyaka, Ohio State University) were used to study the MHC class I-restricted response of CTLs in mice. The EG7 cells have been transfected with plasma to synthesize and constitutively secrete OVA 257–264 peptide and should be cultured in 10% RPMI. The EL4 cells are the non-OVA secreting duplicate of the EG7. Both are commonly found at ATCC but were acquired through Dr. Boyaka. The DC2.4 cell line was kindly provided by K. Rock, University of Massachessetts and as a DC antigen-presenting cell.

Bone marrow (BM) cells were collected from C57Bl/6 mice femurs and tibias. After erythrocyte lysis (AKC or in-house lysis buffer), cells were cultured with RPMI 1640 (Invitrogen) supplemented with 10% FBS, 25 mM HEPES, 2 mM L-glutamine, 50 U/ml penicillin, 50 mg/ml streptomycin, 5 × 10^-5 M 2-mercaptoethanol and 200U/ml recombinant murine GM-CSF (R&D Systems) ± 100 nM of either estriol (E3) or all-trans retinoic acid (RA) (Sigma-Aldrich) for 6–7 days at a density of 2 × 10⁶ cells/ml. Day 6–7 cells were considered differentiated BM-MCs (media control) and BM-MC_regs(RA and E3). Cells were challenged with inflammatory stimulus LPS (1 μg/ml, 055:B5, Sigma-Aldrich) during culture as indicated at day 6 or later for BM-DCs.

Myeloid cells (BM-MCs or BM-MC_regs) were cultured with responder spleen cells from antigen-specific T cell receptor transgenic (TCR Tg; where antigen was either OVA323-339 or MOG35-55) or Foxp3EGFP mice as indicated. To assess T cell proliferation co-cultures were stimulated with anti-CD3 (BD Bioscience), T cell-receptor specific antigen MOG35-55 (Bio Matic) or T cell-receptor specific antigen OVA 323–339 (Anaspec). To assess the effects of myeloid cell activation, co-cultures were stimulated with LPS from Escherichia coli, 055:B5 (Sigma-Aldrich) for 96 hours, pulsed with (H³ thymidine) (Perkin Elmer Life Sciences or MP Biomedicals) in the last 18 hours, harvested and counted, data is expressed as counts per million (cpm) ± SEM [35].

To generate CTL, spleen and lymph nodes (LN) were removed from OT-1 mice and co-cultured with OVA (257–264) pulsed DC2.4 cells (kindly provided by Kenneth Rock, University of Massachussets) for 4 days, removed and cultured with mrIL-2 (R&D Systems) for 2 days. OVA-expressing (EG7) and non-transfected control cells (EL4) were seeded at 2 × 10⁴ cells per well and co-cultured with CTLs (1 × 10⁵) and control or RA treated monocytes (2 × 10⁴) for 6–18 hours [73, 78]. The MTT assay (Sigma-Aldrich) was used to determine quantity of live cells. Briefly, after incubation, cells were centrifuged (1500 RPM for 5 min) media was decanted and 100 ul of fresh media was added. 10ul of 5 mg/ml thiazolyl blue tetrazolium bromide (MTT) was added to each well for 2 hours at 37°C. After incubation cells were centrifuged (1500 RPM for 7 min) and media was decanted, cells were allowed to dry for 15–30 min before 100 ul of DMSO was added, mixed well and read at 570 nm on a Spectra Max 2. The absorbance levels were calculated by averaging the non-specific and specific absorbance levels of five separate data sets. Media control is compared to RA treated cells.

Bone marrow (BM) cells were collected from C57Bl/6 mice femurs and tibias. After erythrocyte lysis, BM cells were cultured with supplemented RPMI (as previously described) for 7 days at a density of 2x10⁶ cells/ml +/− RA. Spleens from mice with reporter Foxp3EGFP (B6.Cg-Foxp3 ^tm2Tch) ) were harvested, passed through cell strainers (70 μm, BD Falcon), collected by centrifugation (1500 RPM for 7 Min at 4°C) and subjected to erythrocyte lysis. Responder cells and MCs or CD11c^- MCs were cultured for 4–6 days and aliquots from cultures assessed for Foxp3 expression by flow cytometry.

In vivo and in vitro derived DCs and MCs were labeled and evaluated by three-color flow cytometry using combination of the following conjugated directly antibodies (clone): CD11c (HL3), CD11b (M1/70), Gr-1 (RB6-8C5), Ly-6G (1A8), Ly-6C (AL-21), MHC class II (AF6-120.1), CD80 (16-10A1), CD86 (IT 2.2), PD-L1 (MIH5), and PD-L2 (YT25) with appropriate isotype controls. (BD Bioscience, eBiosciences or Miltenyi Biotec). Cells were stained with fluorochrome-labeled antibodies or isotype controls for 20 min in the dark at 4°C, washed twice in FACS buffer and re-suspended in 300 μl FACS buffer for flow cytometry analysis.

Intracellular IL-10, IL-4, IL-17 and INF-γ levels were measured after incubating myeloid cells with 1 μg/ml LPS overnight (IL-10) or Ionomyocin (1 mg/ml) and PMA (25 ng/ml) for 4 hours (IL-4, IL-17, and INF-γ). 2 μM of Monensin (eBioscience) added 2–4 hrs. before harvesting cells. Cells were removed from culture, washed with 2 ml of supplemented RPMI and blocked with 0.5 μg/ml Fc block (anti-CD16/CD32) for 15 minutes. Cells that required extracellular markers were re-suspended in FACS buffer and stained with anti-CD4 (0.2 mg/ml) and incubated in the dark at 4°C for 20 min. Cells were washed with FACS buffer (2x with 1 ml) and then fixed and permeabilized using FIX/PERM solution (BD Bioscience), briefly vortexed and incubated in the dark at 4°C for 20 min. Cells were then washed twice with 1 ml of PERM/WASH buffer (BD Bioscience), re-suspended in PERM/WASH buffer and stained with 0.2 mg/ml anti-IL-10 (BD Bioscience) for 30 min. in the dark at 4°C. All flow samples were processed on an Accuri C6 flow cytometer and results analyzed using the Accuri C6 Flow software (BD Biosciences).

Day 6–7 differentiated BM cells were incubated with manufacturer suggested amounts of CD11c/CD11b microbeads (Miltenyi Biotec) for 15 minutes in the dark at 4°C. Cells were washed with running buffer (10% FBS in PBS with 900 mg of NaN₂ per 1 L of PBS), and centrifuged (1500 RPM, 7Min). Cell separation was performed using either the Auto Macs (Miltenyi Biotec) magnetic separation instrument or the FACS Aria III 12 color, 4 laser cell sorter. The Auto Macs was used according to the manufacturer’s instructions. Cell sorting with the FACS Aria III was performed at the OSU Flow Cytometry Core and isotype control antibodies were included to determine detection levels. CD11b⁺ CD11c^- Ly-6C^low monocyte populations were serially gated on CD11c^- cells, followed by CD11b⁺ with gates set around distinct populations of Ly-6C low, intermediate and high. The purity of the cell populations was ≥95%.

Data are represented as mean +/− SEM or fold change. Statistical significance was determined using a Student’s t-test or 1 way ANOVA with a significance level (p-value) < 0.05 and the Wilcoxen signed-rank test. All analyses were performed using Excel and/or GraphPad Prism software (La Jolla, CA).