Recombinant mouse SCF and Flt3-L were purchased from Immunotools (Friesoythe, Germany), recombinant mouse GM-CSF from Peprotech (Rocky Hill, NJ, USA). Opti-MEM Medium (Thermo Fisher Scientific, Waltham, MA, USA) was supplemented with glutamine, penicillin/streptomycin, 50 µM β-mercaptoethanol (Complete Opti-MEM medium). Complete Opti-MEM medium was not supplemented with any serum, except in the cultures with OT-1 and OT-2 cells, as indicated. RPMI Medium 1640 (Sigma-Aldrich, Milan, Italy) was supplemented as above, plus 10% heat-inactivated fetal calf serum (FCS) (complete RPMI medium). Opti-MEM is an optimized version of MEM containing insulin and transferrin, but does not contain GM-CSF, Flt3-L, SCF, or other cytokines (personal communication from Thermo Fisher Scientific Technical Support).

Female C57BL/6J (B6) and OT-2 TCR transgenic mice were purchased from Charles River and housed at the animal facility of Istituto Superiore di Sanità of Rome (ISS), according to institutional guidelines (DL116/92 and 26/2014). Female OT-1 TCR transgenic mice were kindly provided by Dr. M. R. Castrucci (ISS). The OT-1 transgenic TCR recognizes the Kb-restricted OVA 257–264 peptide (35), while the OT-2 transgenic TCR recognizes the I-Ab-restricted OVA 323-339 peptide (36). CX3cr1^gfp/+ and CX3cr1^gfp/gfp B6 mice were purchased from JAX Mice and Services (Bar Harbor, ME, USA) (37). Mice were sacrificed at 5–16 weeks of age and spleen, peripheral, and mesenteric LNs and BM obtained as we previously described (38, 39). In some experiments, CD11c⁺ cells were enriched from either spleen or BM with anti-CD11c magnetic microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany).

We generated DCs from BM cells as previously described (40, 41), with few modifications. Briefly, 10–15 × 10⁶ BM cells were cultured in complete RPMI medium with 20 ng/ml of GM-CSF in Petri dishes (BD Falcon, BD Biosciences, San Jose, CA, USA). After 3 days, fresh medium with GM-CSF was added. At day 7, we collected non-adherent and slightly adherent cells after detachment with PBS 3 mM EDTA. CD11c⁺ cells were purified with anti-CD11c magnetic microbeads (Miltenyi Biotec), thus obtaining BMdDCs. In some experiments, DCs were generated by culturing BM cells with Flt3-L at 100 ng/ml for 8 days, as previously described (42).

OT-1 and OT-2 mice were sacrificed and single cell suspensions were obtained from spleen and LNs. OT-1 CD8⁺ and OT-2 CD4⁺ cells (≥87% pure) were obtained after incubation with anti-CD8β.2 fluorescein isothiocyanate (FITC) monoclonal antibody (mAb) (BD) and CD4 FITC mAb (BD), respectively, followed by positive selection with anti-FITC magnetic microbeads (Miltenyi Biotec). Cells were labeled for 8 min at room temperature with 2.5 µM carboxyfluorescein diacetate succinimidyl ester (CFSE, Molecular Probes, Eugene, OR, USA). For antigen presentation assays, BMdDCs were kept for 2 days in complete Opti-MEM medium with GM-CSF at 20 ng/ml, then incubated for 5 h with Ovalbumin (OVA, Hyglos GmbH, Resenburg, Germany) at 0.2 mg/ml, in the presence or not of SCF at 100 ng/ml. After extensive washings, BMdDCs (200–250 × 10³ cells/well) were cocultured in flat-bottom 96-well plates in complete Opti-MEM medium with 5% FCS with either purified CFSE-labeled OT-1 CD8⁺ (700–750 × 10³ cells/well) or purified CFSE-labeled OT-2 CD4⁺ (200 × 10³ cells/well) cells for 3 days. CFSE dilution by TCR⁺ CD8⁺ and TCR⁺ CD4⁺ cells was evaluated by flow cytometry (43–45).

Cell membrane staining was performed with fluorochrome-conjugated mAbs, after blocking with anti-FcγR (clone 2.4G2) mAb. The following mAbs were used (clone indicated in parentheses): anti-CD11c (HL3), anti-I-Ab or MHCII (M5/114.15.2) anti-CD40 (HM40-3), anti-CD11b (M1/70), anti-CD8α (53-6.7), anti-CD3 (145-2C11), anti-NK1.1 (PK136), anti-c-kit (2B8), anti-TCRβ (H57-597); anti-CD4 (RM4-4); anti-CD8β.2 (53-5.8); anti-CXCR4 (2B11) [from BD Biosciences; Biolegend, San Diego, CA, USA; Miltenyi Biotec; eBioscience; conjugated with FITC, phycoerythrin (PE), peridinin chlorophyll protein (PerCP)-Cy5.5, PE-Cy7, Alexa 647, APC, APC-Vio770; APC-H7]. Dead cells were excluded with Propidium Iodide (PI, Sigma-Aldrich). Samples were analyzed by FACSCanto I and II (BD Biosciences). Data were analyzed using FlowJo software, v.9.7.6 (FlowJo, Ashland, OR, USA).

Total RNA was extracted by TriReagent (Sigma-Aldrich). One microgram of total RNA was used for cDNA first-strand synthesis according to the manufacturer’s protocol for Moloney MLV reverse transcriptase (Promega, Madison, WI, USA). Real-time PCR was performed using the ABI Prism 7900 sequence detection system (Applied Biosystems, Foster City, CA, USA). cDNAs were amplified in triplicate with primers for c-kit (Mm00445212_m1), SCF (Mm00442972_m1), and hprt-1 (Mm00446968 m1) (Applied Biosystems, Thermo Fisher Scientific), all conjugated with fluorochrome FAM. Relative expression of each gene versus hprt-1 was calculated according to the 2^−∆∆Ct method.

Immunoblotting was performed as we previously described, after cell lysis with 1% Triton X-l00, 0.1% SDS Tris buffer containing protease inhibitor cocktail (Sigma-Aldrich) and the phosphatase inhibitors NaF, Na₃VO₄, phenylmethylsulfonyl fluoride (46). Equal amounts of proteins were separated by 10% SDS-PAGE, transferred onto PVDF membrane by Trans-Blot TurboTM Transfer System (Bio-Rad Laboratories), and probed with anti-phospho-AKT (Ser 473), anti-AKT and anti-β-actin antibodies (Cell Signaling Technology, Danvers, MA, USA). Chemiluminescence detection was performed by a CCD camera gel documentation system (ChemiDocXRS, Bio-Rad Laboratories, Hercules, CA, USA).

BM-derived DCs were lysed as above without the phosphatase inhibitors. BMdDC culture supernatants (100 µl/well) and BMdDC lysates (25 µg of cell lysate/well) were tested by mouse SCF ELISA kit (Boster Immunoleader, Pleasanton, CA, USA).

Stem cell factor silencing was performed by transfecting BMdDCs with Lipofectamine^® RNAiMax reagent (Invitrogen, Thermo Fisher Scientific) using the following siRNAs: SCF-siRNA (sc-39735, consisting of a pool of three target-specific 19–25 nt siRNA), scrambled control-A siRNA (sc-37007), fluorescein conjugated control-A siRNA (sc-36869), all from Santa Cruz (Dallas, TX, USA).

Stem cell factor- or control-silenced BMdDCs were cultured for 2 days in 96-well plates at 2 × 10⁵/well in complete Opti-MEM medium with or w/o GM-CSF at 20 ng/ml. Cells were then collected, stained with Annexin V FITC and PI, and analyzed by flow cytometry. In experiments with blocking anti-c-kit mAb, BMdDCs were cultured as above in the presence of 10 µg/ml of either anti-c-kit (ACK2) or isotype control (RTK4530) mAb (both Low Endotoxin Azide Free, Biolegend). After 2 days, cells were collected and analyzed by Annexin V FITC/PI as above.

BM-derived DC numbers were determined by the CyQUANT Cell Proliferation Assay kit from Invitrogen according to manufacturer’s instructions. Optical density was determined using a VICTOR X Multilabel Plate Reader.

Pieces of spongy bone, which would otherwise have been discarded, were obtained from systemically healthy patients undergoing hip replacement at either Centro Traumatologico Ortopedico Andrea Alesini Hospital or Policlinico Tor Vergata, Rome, after approval by Institutional Ethics Committee, and informed consent from patients in accordance with the Declaration of Helsinki (study no. 129.14, prot 76699 and study no. 156.15, prot. 0030053/2015). BM mononuclear cells were isolated by Lympholyte-H (Cedarlane, NC, USA) density-gradient centrifugation. Flow cytometry was performed as above, using FcR Blocking Reagent and Dead Cell Discrimination Kit (Mitenyi Biotec) before addition of the following fluorochrome-conjugated mAbs: CD1c or BDCA-1 (AD5-8E7), CD141 or BDCA-3 (AD5-14H12), CD45 (2D1) CD14 (TÜK4), CD19 (LT19), c-kit (104D2) [from BD Biosciences; Miltenyi Biotec; Exbio, Prague, Czech Republic; conjugated with phycoerythrin (PE), PerCP-Vio700, PE-Cy7, APC, APC-H7]. Cells were fixed with PBS formaldehyde 1.2% before flow cytometric analysis.

We performed a two-tailed paired Student’s t-test using Prism v.6.0f, GraphPad Software (La Jolla, CA, USA). Differences were considered significant when *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

We analyzed membrane c-kit expression by BM and spleen DCs freshly obtained from untreated C57BL6/J (B6) mice, gating on CD11c^hi MHC-II⁺ DC as previously described (47). Typical flow cytometry profiles are shown in Figure 1A (see Figure S1A in Supplementary Material for our gating strategy, based on forward and side scatter plots and exclusion of CD3⁺, NK1.1⁺, and dead cells). While the great majority of spleen DCs expressed membrane c-kit, only a small percentage of BM DCs was c-kit⁺ (Figures 1A,B). Results were similar after about 100-fold enrichment of CD11c⁺ cells by immunomagnetic selection (Figures S2A,B in Supplementary Material). c-kit⁺ DCs were characterized by intermediate levels of membrane c-kit, that is <10⁴ fluorescence intensity, whereas other cells in spleen and BM were c-kit^hi (Figure 1C). We observed that MHC-II median fluorescence intensity (MFI) was higher in BM DCs than in spleen DCs, in agreement with previous results (47). Furthermore, MHC-II MFI was higher in c-kit⁺ BM DCs than in c-kit⁺ spleen DCs (Figure 1D).

It was previously shown that most BM DCs are characterized by high CX3CR1 expression, using a mouse model in which green fluorescence protein (GFP) replaces CX3CR1 by gene targeting (47). We analyzed BM DCs from CX3cr1^gfp/+ mice by flow cytometry to investigate c-kit expression in respect to GFP intensity. We found that c-kit was expressed by the few intermediate green but not by the large population of bright green BM DCs from CX3cr1^gfp/+ mice (Figure 1E), whereas in the spleen most CX3cr1^gfp/+ spleen DCs were c-kit⁺ and had intermediate green fluorescence intensity (Figure 1E). Furthermore, CX3CR1-deficient CX3cr1^gfp/gfp BM DCs contained a significantly higher percentage of c-kit⁺ cells than WT BM DCs, while the percentage of c-kit⁺ cells within spleen DCs was similar in WT, CX3cr1^gfp/+, and CX3cr1^gfp/gfp mice (Figure 1F; Figures S2C,D in Supplementary Material). These results suggest that c-kit expression and high levels of CX3CR1 are inversely correlated in BM DCs and that the presence of c-kit⁺ DCs in the BM does not depend on CX3CR1 expression.

We then performed flow cytometric analysis of cDC subsets from spleen and BM of WT mice, by gating on either cDC1s or cDC2s, based on CD8α and CD11b expression (14, 48). We found that virtually all cDC1s and a small proportion of cDC2s expressed c-kit in the BM, whereas both cDC subsets were c-kit⁺ in the spleen (Figures 2A,B). The difference in c-kit percentages between BM cDC1s and cDC2s was statistically significant (Figure 2C). In the BM, c-kit⁺ DCs expressed intermediate levels of CD11b, while c-kit⁻ DCs were CD11b^hi (Figure 2A, right top panel).

We compared c-kit membrane expression by DCs obtained from mouse BM cells differentiated in vitro with either GM-CSF or Flt3-L (40–42). Both protocols resulted in a small percentage of c-kit⁺ within CD11c⁺ cells (Figures S3B,D in Supplementary Material; gating strategy in Figures S3A,C in Supplementary Material). However, DCs generated with GM-CSF were characterized by increased cell yield and higher cell viability (Figures S3A–D in Supplementary Material and data not shown). Thus, we decided to use GM-CSF for further experiments and to magnetically select CD11c⁺ cells from the fraction including non-adherent and slightly adherent cells at day 7 to obtain highly purified DCs (called “BM-derived DCs or BMdDCs,” see Figures S3E,F in Supplementary Material). On average, 15% of the BMdDCs were MHC-II^hi CD40^hi CD11b^int and 76% MHC-II^int CD40^int CD11b^hi cells (see example in Figure 3A, top panel). c-kit was expressed only by the MHC-II^hi CD40^hi cells, suggesting that it is a marker of the more mature cells (Figure 3B, top panels). The c-kit⁺ BMdDCs were all CD11b^int, whereas the remaining c-kit⁻ cells were CD11b^hi (Figure 3C, top panel).

To increase the proportion of c-kit⁺ cells, we replated BMdDCs in different culture conditions, all with GM-CSF at 20 ng/ml, and then analyzed their membrane phenotype (Figures S4A,B in Supplementary Material and data not shown). After 2 days of culture in complete Opti-MEM medium, the majority of BMdDCs were MHC-II^hi CD40^hi CD11b^int, and expressed c-kit⁺ (Figures 3A–C, bottom panels). Membrane c-kit fluorescence intensity significantly increased within the MHC-II^hi CD40^hi cells, while the MHC-II^int CD40^int cells remained negative for this marker (Figures 3B,D). c-kit mRNA expression increased significantly, indicating that membrane c-kit up-regulation depended on increased transcription (Figure 3E).

We usually collected non-adherent and slightly adherent BMdDCs when we harvested cells 2 days after replating. In some experiments, we separately analyzed adherent cells and found that they comprised a higher proportion of MHC-II^int CD40^int cells compared to the floating fraction of BMdDCs. Nevertheless, a high percentage (>70%) of adherent cells was still MHC-II^hi CD40^hi, confirming that MHC-II^hi CD40^hi cells represented the large majority of BMdDCs after 2 days of culture in our conditions (Figure S4C in Supplementary Material). Notably, as for the non-adherent fraction, the MHC-II^hi CD40^hi cells were the only c-kit⁺ cells in the adherent fraction (Figure S4D in Supplementary Material).

To determine whether c-kit could function as a signal transducing receptor, c-kit⁺ BMdDCs were stimulated with SCF and analyzed for phosphorylated Akt, a serine-threonine kinase located downstream of PI3-kinase with an essential role in cellular survival and proliferation (34). After incubation with SCF, phospho-Akt levels increased rapidly, while total Akt remained unchanged (Figures 3F,G).

To evaluate the effect of SCF on the capacity of c-kit⁺ BMdDCs to present the model antigen OVA, we assessed the proliferation of OVA-specific T cells from TCR transgenic mice OT-1 and OT-2, by CFSE dilution. We observed that both OT-1 (Figures 4A,B) and OT-2 (Figures 4C,D) cells proliferated in response to BMdDCs preincubated with OVA, showing the functionality of both MHC-I and MHC-II presentation pathways. The presence of SCF during preincubation of BMdDCs with OVA did not influence the proliferation of neither OT-1 (Figures 4A,B) nor OT-2 (Figures 4C,D) cells.

Since GM-CSF can inhibit c-kit expression by mast cells (49, 50), we next evaluated the effects of GM-CSF in our BMdDC cultures, by comparing c-kit membrane expression by BMdDCs cultured in complete Opti-MEM medium either with or without GM-CSF at 20 ng/ml. Furthermore, considering the possible intercellular competition for GM-CSF, we cultured BMdDCs at two different cell densities, i.e., about 6.3 × 10⁵ and 1 × 10⁵ cells/cm², corresponding to 1.2 × 10⁶ and 0.2 × 10⁶ cells/well in 24-well plate, respectively. We observed that c-kit membrane expression by MHC-II^hi CD40^hi cells was higher when BMdDCs were cultured at higher cell density, and in particular, MHC-II^hi CD40^hi cells had extremely high c-kit fluorescence intensity when cultured without GM-CSF if compared to BMdDCs cultured in the presence of GM-CSF (Figures 5A–C).

Stem cell factor can be produced by a variety of cell types, including DCs (25, 29, 30). We found that BMdDCs produced cell-associated SCF and its levels remained stable after 2 days of culture in complete Opti-MEM medium with GM-CSF. When BMdDCs were cultured in the absence of GM-CSF, cell-associated SCF significantly increased (Figure 5D). SCF concentration in supernatants was 50.33 ± 1.05 pg/ml after 2 days of culture in the presence of GM-CSF and did not increase in the absence of GM-CSF (data not shown). Notably, SCF expression increased at the mRNA level when BMdDCs were cultured for 2 days without GM-CSF (Figure 5E). Thus, BMdDCs can produce SCF, but the presence of GM-CSF interferes with this process.

CXCR4 plays a key role in leukocyte retention in the BM, by interacting with CXCL12 (51). Moreover, DC maturation results in upregulation of CXCR4 (3, 5). We compared CXCR4 membrane expression by BMdDCs cultured in complete Opti-MEM medium either with or without GM-CSF at 20 ng/ml and evaluated whether an overnight addition of SCF to these cultures modulated CXCR4. MHC-II^hi CD40^hi cells had always a higher CXCR4 MFI than MHC-II^int CD40^int cells (Figures 6A,B). CXCR4 expression by MHC-II^hi CD40^hi cells increased in the absence of GM-CSF, and the addition of SCF to the culture did not influence the expression of CXCR4 (Figures 6B,C).

To evaluate whether BMdDC survival could be modulated by SCF autocrine production, SCF was silenced by transfection with SCF-siRNA and cell survival analyzed after 2 days of culture either with or without GM-CSF. In our conditions, transfection efficiency was about 40%, as evaluated in control cells transfected in parallel with fluorescein-conjugated RNA, and SCF protein concentration in lysates from SCF-siRNA-silenced cells was about 70% than that in lysates from scrambled siRNA-silenced control cells (e.g., 489 pg/ml in SCF-silenced cells and 671 pg/ml in control cells, in cultures with GM-CSF; 589 pg/ml in SCF-silenced cells and 714 pg/ml in control cells, without GM-CSF). Although SCF expression was only partially reduced, we observed a statistically significant decrease in cell survival in the cultures with GM-CSF, based on Annexin V/PI staining (Figures 7A,B). Absolute cell numbers were significantly decreased upon SCF silencing in both BMdDC cultures, with and without GM-CSF (Figure 7C). In a different set of experiments, we cultured BMdDCs for 2 days in the presence of 10 µg/ml of either the anti-c-kit blocking mAb ACK2 or its corresponding control mAb and observed a modest but statistically significant decrease in cell survival, based on Annexin V/PI staining (Figure S5 in Supplementary Material). Results were similar in BMdDC culture with and without GM-CSF (Figure S5 in Supplementary Material). Altogether, these findings indicate that DCs promote their own survival both by the autocrine production of SCF and by expressing the receptor c-kit.

We also analyzed cDC from human BM samples, by using high levels of CD141 and expression of CD1c as markers for cDC1s and cDC2s, respectively (14, 52). Typical examples of human BM cytometric profiles are shown in Figures 8A,B (gating strategy in Figure S1B in Supplementary Material). In agreement with mouse data, we found that cDC1s contained a significantly higher percentage of c-kit⁺ cells than cDC2s (Figure 8C). c-Kit histograms suggested that cDC1s represented a homogeneous subset expressing c-kit at low levels, rather than a heterogeneous population containing a negative and a positive fraction (Figure 8A). Further analysis showed that the CD141⁻ fraction in the BM contained cells expressing much higher c-kit levels than cDC1s (Figure 8B).