cDC1 Vaccines Drive Tumor Rejection by Direct Presentation Independently of Host cDC1

Dendritic cell (DC) vaccines for cancer therapy have been developed using cells generated from culturing peripheral blood monocytes or bone marrow (BM) with GM-CSF (12), referred to as either Mo-DCs or BM-DCs, or collectively as GMDCs (3, 4). GMDCs derived from peripheral blood monocytes are able to strongly stimulate T cells in mixed lymphocyte reactions (1). Similar potency for T-cell activation was found in BM-derived GMDCs (5) and in cells derived from human CD34⁺ BM stem cells cultured with GM-CSF and TNFα (6). Later, addition of IL4 to GM-CSF cultures was reported to enhance and maintain antigen presentation of human DCs (7).

Cancer vaccines based on monocyte-derived GMDCs were considered to function as antigen-presenting cells (APC) by directly presenting tumor-derived antigens to tumor-specific T cells in vivo (8). Initial clinical studies indicated that the monocyte-derived GMDC vaccine formulation Sipuleucel-T (Provenge) could activate T cells specific for prostatic acid phosphatase (2), suggesting possible value in treating prostatic cancer (9). However, only modest survival benefit was reported for Sipuleucel-T in a double-blind, placebo-controlled trial of 512 patients with metastatic castration-resistant prostate cancer (10). A similar DC vaccine formulation, in which monocyte-derived GMDCs were pulsed with autologous tumor lysate for use in patients with relapsed osteosarcoma, showed antitumor responses in only 2 of 12 vaccinated patients and resulted in little evidence of clinical benefit (11). Monocyte-derived GMDC vaccines pulsed with peptides identified as neo-antigen candidates are able to induce specific CD8⁺ T-cell responses in patients with melanoma (12). A subsequent clinical trial showed a slightly increased 5-year survival in patients with metastatic melanoma in response to monocyte-derived GMDC vaccination (13). A smaller trial suggests that progression-free survival correlates with magnitude of the immunologic response (14). A 12-year follow-up study of monocyte-derived GMDC vaccines in patients with melanoma showed a 19% survival similar to treatment with ipilimumab (15). A clinical trial of monocyte-derived GMDCs in bone and soft-tissue sarcoma showed increases in serum IFNγ and IL-12, but resulted in an improvement of the clinical outcome in only a small number of patients (16). In summary, to date, there has been only slight clinical benefits realized by use of monocyte-derived GMDC cancer vaccines (17).

One feature of GMDC vaccines that may reduce their effectiveness is their reliance on host DCs for T-cell activation (18–20). Initially, GMDCs were thought to directly stimulate host T cells through in vivo presentation of tumor antigens. However, CD4⁺ T-cell activation by GMDC vaccination was discovered to require expression of MHC class II molecules (MHC-II) by host conventional DCs (cDC; ref. 18). In mice globally lacking MHC-II expression, CD4⁺ T-cell responses induced by GMDCs were restored by selective reexpression of MHC-II on DCs, but not on B cells. Likewise, in vivo activation of OT-I T cells was induced by GMDCs, produced from C57BL/6 (B6) cells pulsed with OVA257–264 peptide, when used as a vaccine in wild-type (WT) mice, but not in mice whose DCs expressed H-2K^bm1, which cannot activate OT-I (19). Another study also supports the interpretation that endogenous cDC1 are required for vaccination with antigen-loaded monocytes (20). OT-I T-cell activation induced by OVA-peptide bearing monocytes was seen in WT mice, but not in Batf3^–/– mice that lack cDC1 development. All of these results indicate that GMDCs do not present antigens directly to host T cells, but act as a source of antigen that must be transferred and processed by host cDCs that are responsible for the direct antigen presentation to both CD8⁺ and CD4⁺ T cells (21).

DCs include lineages besides those derived from GM-CSF treatment of monocytes or BM. cDCs found in vivo rely on Flt3-ligand for their development and include at least two major branches, called cDC1 and cDC2, which appear to perform different functions in vivo (22–24). In particular, cDC1 are capable of cross-presentation (25) and associate with induction of antitumor CD8⁺ T-cell responses (21, 26). cDC1 are more efficient in acquiring antigens from tumors compared with other DC subsets (27), and vaccination using splenic cDC1 loaded with tumor antigens induce robust tumor-specific CD8⁺ T-cell responses (28). cDC1 vaccines are superior to GMDC vaccines in limiting tumor growth and enhanced response to immune checkpoint blockade (29). However, one reported function of DCs is to carry antigens from sites of tumors to the draining lymph nodes (LN), where they can distribute these antigens to resident DCs (30). If cDC1 vaccines function only to carry antigens to LNs, then they may provide no greater clinical benefit than that of GMDCs. For this reason, we sought to test whether cDC1 vaccines, unlike GMDC vaccines, can function independently of host DCs to present tumor antigens directly to host T cells.

Irf8 +32^–/− and Batf3^−/− mice have been described previously (26, 31). OT-I (C57BL/6-Tg(TcraTcrb)1100Mjb/J), OT-II (C57Bl/6-Tg(TcraTcrb)425Cbn/J), and CD45.1⁺ (B6.SJL-Ptprca Pepcb/BoyJ) mice were purchased from Jackson Laboratory and bred to generate CD45.1⁺ OT-I and CD45.1⁺ OT-II mice, respectively. K^b−/−D^b−/−β2m^−/− mice (MHCI-TKO; ref. 32) were a gift from Herbert W. Virgin IV and Ted Hansen (Washington University in St. Louis, Missouri). Mice were maintained on the C57BL/6 background. All in vivo experiments were performed in Washington University's specific pathogen-free facility, and both sexes were used between the ages of 8 and 16 weeks. All animals were maintained on 12-hour light cycles and housed at 70°F (∼21°C and 50% humidity). All experiments were performed in accordance with procedures approved by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC)-accredited Animal Studies Committee of Washington University in St. Louis and were in compliance with all relevant ethical regulations. All mice were maintained in a specific pathogen-free animal facility following institutional guidelines and with protocols approved by the Animal Studies Committee at Washington University in St. Louis, an Institutional Animal Care and Use Committee.

Femurs, tibias, and pelvises of naïve C57BL/6 mice were crushed using a mortar and pestle in FACS buffer [10% BSA (Sigma-Aldrich) and 2 mmol/L EDTA in PBS] and filtered through a 70-μm strainer. Red blood cells (RBC) were lysed using ACK Lysis Buffer [150 mmol/L NH₄Cl (Sigma-Aldrich), 10 mmol/L KHCO₃ (Sigma-Aldrich), 0.1 mmol/L EDTA (Sigma-Aldrich)]. After RBC lysis, cells were brought up in complete Iscove's modified Dulbecco's medium (IMDM; Gibco) supplemented with 10% FCS, 55 μmol/L 2-mercaptoethanol (Sigma-Aldrich), 1% penicillin/streptomycin solution (Gibco), 1 mmol/L sodium pyruvate (Corning), 1% MEM nonessential amino acid (Sigma-Aldrich), and 200 μmol/L glutamine (Gibco) and kept at 4°C until plated for culture. GMDCs were generated by plating BM cells at a density of 5×10⁵/mL in complete IMDM supplemented with 20 ng/mL GM-CSF (PeproTech) and 20 ng/mL IL4 (PeproTech) for 4 days. The cells were then isolated and spun at 1,500 RPM for 5 minutes, and the cells were plated again in complete IMDM supplemented with 20 ng/mL GM-CSF and 20 ng/mL IL4 for an additional 4 days. Loosely adherent cells were collected by gentle pipetting. Cells were then sorted as SiglecH⁻B220^– CD11c⁺MHCII⁺ into complete IMDM and kept at 4°C until used for experiments. Flt3L-cDCs were generated by plating BM cells at a density of 2 to 2.5×10⁶/mL in complete IMDM supplemented with 5% Flt3L-Fc conditioned media for 8 days. Loosely adherent cells were isolated by gentle pipetting. Cells were then sorted as SiglecH⁻B220^–CD11c⁺MHCII⁺XCR1⁺Sirpa^– (CD172a^–) for cDC1 and SiglecH⁻B220^–CD11c⁺MHCII⁺XCR1^–Sirpa⁺ (CD172a⁺) for cDC2 into complete IMDM kept at 4°C until used for experiments. Cells were sorted using a FACS Aria Fusion instrument (BD), as indicated in the flow cytometry section, and gated as indicated above.

The methylcholanthrene (MCA)-induced fibrosarcomas 1956 and 1969 were gifts from Robert Schreiber obtained 2012 (Washington University School of Medicine). They were generated in female C57Bl/6 mice, tested for Mycoplasma, and banked at low passage as previously described (21). The immunogenic fibrosarcoma expressing membrane ovalbumin (mOVA) was generated from the MCA-induced progressor fibrosarcoma 1956 (1956 mOVA) as previously described (21) and used for all tumor experiments described herein. Tumor cells derived from frozen stocks were propagated for 1 week, with one intervening passage in vitro in Roswell Park Memorial Institute (RPMI1640, Gibco) media supplemented with 10% FCS (HyClone), washed 3 times with PBS, and suspended at a density of 6.67 × 10⁶ cells/mL in endotoxin-free PBS (Sigma-Aldrich).

Irf8 +32^–/− or C57BL/6 WT mice were subcutaneously injected into the flanks with 10⁶ 1969 tumor cells, 10⁶ 1956 mOVA tumor cells or with 5×10⁵ 1956 mOVA tumor cells for the experiments in which mice with inoculated on both flanks. Tumor growth was measured every 3 to 4 days with a caliper, and tumor area was calculated by the multiplication of two perpendicular diameters. In accordance with our IACUC-approved protocol, maximal tumor diameter was 20 mm in one direction, and in no experiments was this limit exceeded. For intratumoral DC injections, 1×10⁶ GMDC or Flt3L-cultured DCs were injected intratumorally or intravenously on days 1, 4, and 8 after tumor implantation and tumor growth was measured as described above. All tetramer analysis (as described below) was performed on day 15 post tumor inoculation.

For dual tumor inoculation, Irf8+32^–/– mice were inoculated with 5×10⁵ 1956 mOVA tumor cells in both flanks. DC injections of 1×10⁶ GMDCs or Flt3L-cultured DCs were injected intratumorally on days 1, 4, and 8 after tumor implantation, and tumor growth was measured as described above. All tetramer analysis (as described below) was performed on day 15 post tumor inoculation.

For memory tumor experiments, Irf8+32^–/– mice were inoculated with 10⁶ 1956 mOVA tumor cells on one flank. DC injections of 1×10⁶ Flt3L-cultured cDC1s (gated as described above) were injected intratumorally on days 1, 4, and 8 after tumor implantation and tumor growth was measured as described above. After primary 1956 mOVA challenge was cleared at day 16, mice were rested and rechallenged with 10⁶ 1956 mOVA tumor cells and followed for tumor growth as described above.

For DC traffic experiments, CD45.2⁺Irf8+32^–/– mice were injected subcutaneously in the flanks with 1 × 10⁶ 1956 mOVA cells. On day 4 after tumor implantation, 10⁷ GMDCs or Flt3L-cDCs were intratumorally injected. Forty hours after DC tumor implantation, tumors and tumor-draining LNs were harvested and digested in collagenase B (0.25 mg/mL) and DNaseI (30 U/mL) in complete IMDM for 30 minutes at 37°C. Tumors and LNs were then gently passed through a 19-gauge syringe until dispersed and filtered through a 70-μm strainer. Cells were then stained with CD40, B220, CD11c, MHCII, XCR1, Sirpa (CD172a), CD45.1, and CD45.2 fluorescent antibodies for 30 minutes at 4°C for analysis, as indicated in the flow cytometry section.

Spleens were harvested 15 days after tumor transplantation, mashed through a metal strainer, ACK lysed, and filtered through a 70-μm strainer. SIINFEKL-H2-K^b biotinylated monomers were purchased from the immune-monitoring core lab at the Bursky Center for Human Immunology and Immunotherapy Programs at Washington University in St. Louis. The peptide–MHC class I complexes refolded with an ultraviolet-cleavable conditional ligand were prepared as described with modifications as described previously (33-35). Ultraviolet-induced ligand exchange and combinatorial encoding of MHC class I multimers was performed as described (36). The peptide–MHC multimers were then incubated with APC (eBioscience), PE (BioLegend), BV605 (BioLegend), or BV710 (BioLegend) conjugated streptavidin (SA) at a concentration of 1:5 for 30 minutes at 4°C protected from light in separate reactions. SA-labeled tetramers were then incubated with 25 μmol/L D-biotin (Sigma-Aldrich) for 20 minutes at 4°C protected from light to quench free fluorochrome-labeled SA. 3 × 10⁶ splenocytes from tumor-bearing mice were incubated with FACS buffer supplemented with 10% of supernatant containing the Fc-blocking antibody produced from 2.4G2 cells for 5 minutes at 4°C. Fluorochrome-conjugated tetramers were added to the splenocytes at a concentration of 3:50 and incubated at 37°C for 30 minutes. TCRβ, CD8α, CD4, B220 antibodies, and the 7-aminoactinomycin D (7AAD, BioLegend) stain were added without washing and stained for another 30 minutes at 4°C, as indicated in the flow cytometry section. Cells were gated as 7AAD^–B220^–TCRβ⁺CD8α⁺.

CD45.1⁺ OT-II TCR transgenic mouse LNs and spleens were harvested and dispersed into single-cell suspensions by mechanical separation. All cells were combined and then stained with biotinylated Ter119, CD8β, I-A/I-E, and Ly6G antibodies in FACS buffer for 20 minutes at 4°C. Cells were washed and incubated with MagniSort SAV Negative Selection Beads (Invitrogen) according to the manufacturer's protocol. Cells were magnetically separated using an EasyEights EasySep Magnet (STEMCELL Technologies) and sorted as B220^–CD8^–TCRβ⁺ CD4⁺ CD45.1⁺ Vα2⁺ (OT-II). Cells were sorted using a FACS Aria Fusion instrument (BD) as indicated in the flow cytometry section. After sorting, purified cells were labeled with 1 μmol/L CellTrace Violet (CTV) proliferation dye (Thermo Fisher Scientific). Two million labeled OT-II were transferred intravenously into C57BL/6 and Irf8 +32^–/– mice bearing 1956 mOVA tumors on one flank on day 2 after tumor implantation. Tumor-draining LNs were harvested on day 5 after tumor implantation (3 days after T-cell transfer) and assayed for dye dilution of CD45.1⁺ OT-II. Cells were stained for CD45.1, CD45.2, Vα2, 7AAD, CD4, CD44, and TCRβ in FACS Buffer. Cells were gated as CD4⁺TCRβ⁺CD45.1⁺Vα2⁺CD44⁺ cells and analyzed for proliferation dye dilution of CD45.1⁺ OT-II on a FACS CANTO II as indicated in the flow cytometry section.

CD45.1⁺ OT-I TCR transgenic mouse LNs and spleens were harvested and dispersed into single-cell suspensions by mechanical separation. All cells were combined and then stained with biotinylated Ter119, CD4, I-A/I-E, and Ly6G antibodies in FACS buffer for 20 minutes at 4°C, as indicated in the flow cytometry section. Cells were then incubated with MagniSort SAV Negative Selection Beads (Invitrogen) according to the manufacturer's protocol. Cells were magnetically separated, as described for OT-II cells, and sorted as B220^– CD8⁺ TCRβ⁺ CD4^– CD45.1⁺ Vα2⁺ (OT-I). Cells were sorted using a FACS Aria Fusion instrument (BD), as indicated in the flow cytometry section. After sorting, purified cells were labeled with 1 μmol/L CTV (Thermo Fisher Scientific) proliferation dye. In vitro–generated GMDCs or Flt3L-cDCs were sorted as described above, and 2.5×10⁴ DCs and 2.5×10⁴ labeled OT-I cells were plated into each well of a 96-well round bottom plate. Log2 dilutions of OVA-loaded splenocytes (cell-associated OVA) were plated with the DCs and T cells. Cell-associated OVA was produced by isolating MHCI TKO splenocytes and osmotically loading with 100 mg/mL soluble OVA (sOVA; Worthington Biochemical Corporation). Cells were x-ray irradiated at 1,350 rad and plated. After 3 days, cells were stained for TCRβ, CD8α, CD45.1, CD44, and Vα2. CD8⁺ TCRβ⁺CD45.1⁺ Vα2⁺CD44⁺ cells and analyzed for proliferation dye dilution of CD45.1⁺ OT-I, as described for OT-II.

Flow cytometry and cell sorting were completed on a FACS Canto II or FACS Aria Fusion instrument (BD) and analyzed using FlowJo analysis software (Tree Star). Staining was performed at 4°C in the presence of 10% Fc block (derived from 2.4G2 cells) in magnetic-activated cell-sorting (MACS) buffer (PBS + 0.5% BSA + 2 mmol/L EDTA). The following antibodies to the following markers were used from BD Biosciences: 7AAD, CD4 (RM4–5), CD8α (53–6.7), CD8β (53–5.8), CD11b (M1/70), B220 (RA3–6B2), CD64 (X54–5/7.1), CD19 (1D3), CD95 (Jo2), CD3 (145–2C11), CD45 (30-F11); from Tonbo Biosciences: MHCII (M5/114.15.2), CD44 (IM7), CD45.1 (A20), CD45.2 (104), CD11c (N418); from BioLegend: SA-PE, SA-BV605, SA-711, XCR1 (ZET), Ter119 (Ter-119), Ly6G (1A8), TCRβ (H57–597), CD8α (53–6.7), CD4 (RMA4–5), CD44 (IM7), CD40 (1C10); from eBiosciences: SA-APC, TCRVα2 (B20.1), CD45.1 (A20), F4/80 (BM8); from Invitrogen: CD172a(P84), CD45 (30F11).

2.4g2 cells were generated previously (Unkeless, JEM 1979). For 2.4g2 supernatant, 2.4g2 cells were incubated for 3 weeks at 37°C in 2 L complete IMDM. The cells were centrifuged at 1,500 RPM, and the supernatant was isolated, filtered using a 20-μm filter (Corning), and titrated for blocking of Fc receptors. Flt3L-Fc was produced from J558 myeloma cells engineered to overexpress the extracellular portion of human Flt3 L fused to human Fc using the pCD4-Hg1 vector (kindly provided by Dr. Marina Cella at Washington University in St. Louis). Cells were incubated in complete IMDM in 1 L roller bottles until concentration was at ∼2×10⁶ cells/mL. The cells were spun at 1,500 RPM, and the supernatant was isolated, filtered using a 20-μm filter (Corning), and used in BM culture to generate DCs. The Flt3L-Fc cultured supernatant was compared and equivalent to 100 ng/mL Flt3L (PeproTech) for in vitro DC generation.

Statistical analysis was performed using GraphPad Prism software version 8. Unless otherwise noted, one-way ANOVA was used to determine significant differences between samples, and all center values correspond to the mean. P ≤ 0.05 was considered statistically significant. Randomization was performed as comparisons were done across mice of the same genotypes receiving different treatments. No formal randomization was done across all other samples. Investigators were blinded to the treatments of the mice during sample preparation and data collection.

The data generated in this study are available upon request from the corresponding author.

We previously generated mice harboring deletions in the +32 kb enhancer of Irf8 that engages IRF8:BATF3 complexes to stabilize Irf8 transcription in the specified pre-cDC1 progenitor (31). Irf8 +32^–/– mice lacked cDC1 in all peripheral lymphoid tissues (31), including the spleen (Fig. 1A). The Irf8 +32^–/– enhancer deletion does not affect Batf3 expression (31), but provides a superior model of cDC1 deficiency compared with Batf3^–/– mice, in which germline Batf3 deficiency can impact other immune lineages (37), and in which cDC1 development is restored in certain conditions by compensation from Batf and Batf2 (38). We previously described the immunogenic fibrosarcoma 1956 mOVA, which is rejected by WT B6 mice by mechanisms involving licensing of cDC1 by CD4⁺ T cells (21). Here, we confirmed that Irf8 +32^–/– mice did not reject 1956 mOVA (Fig. 1B), indicating that rejection of this tumor relies on cDC1. Therefore, we used 1956 mOVA tumors and Irf8 +32^–/– mice to test whether cell-based vaccines could directly present antigens to host T cells because any CD8⁺ T-cell response induced by DC vaccines in Irf8 +32^–/– mice must be generated by the vaccine itself, and not antigen transfer to host cDC1.

Current strategies generate DC vaccines by culturing monocytes or BM with GM-CSF, with or without IL4, producing large numbers of CD11c⁺MHCII⁺ cells (GMDC; Fig. 1C). By contrast, culturing BM with FL3L generates CD11c⁺MHCII⁺ cells containing XCR1⁺Sirpα^– cells (cDC1) and XCR1^–Sirpα⁺ cells (cDC2; Fig. 1C). GMDCs, like cDC2, expressed Sirpα but expressed XCR1 at levels intermediate between cDC1 and cDC2, and none expressed CD40 in culture (Fig. 1D). GMDCs, cDC1, and cDC2 were all able to process and present sOVA to OT-I CD8⁺ T cells. In contrast, GMDCs and cDC1, but not cDC2, were able to process and present cell-associated OVA to OT-I CD8⁺ T cells (Fig. 1E and F).

We previously reported that the fibrosarcoma 1956 mOVA is rejected by C57BL/6 mice, but not by Irf8 +32^–/– mice (21). To test whether vaccines based on Flt3L-cDCs could directly prime host T cells in vivo, we generated cDC1, cDC2, and GMDCs in vitro and injected these cells subcutaneously into tumors growing in Irf8 +32^–/– mice (Fig. 2A) and monitored tumor growth (Fig. 2B). Injection of GMDCs and cDC2 into tumors did not affect tumor growth compared with the PBS negative controls. By contrast, injection of cDC1 led to regression of tumors in 9 of 11 mice (Fig. 2B). As a control, we showed that B6 mice inoculated with 1956 mOVA and administered intratumoral GMDC injections also rejected tumors (Supplementary Fig. S1A), as expected.

In many human cancers, neo-antigens may be difficult to identify, particularly for immunodominant neo-epitopes (39). Therefore, to extend our results to a second tumor model system, we examined the ability of cDC1 vaccines to induce rejection of the 1969 fibrosarcoma, which does not exogenously express OVA. As above, cDC1, cDC2, and GMDCs were generated in vitro and injected subcutaneously into 1969 tumors growing in Irf8 +32^–/– mice. Again, in this model, cDC1 vaccination induced tumor rejection of 1969 fibrosarcomas growing in Irf8 +32^–/– mice (Supplementary Fig. S1B). By contrast, GMDCs or cDC2 vaccination failed to induce tumor rejection, which was in agreement with results for the 1956 mOVA tumor model (Fig. 2B).

CD8⁺ T-cell responses to 1956 mOVA can be monitored using an H-2K^b-SIINFEKL MHC-I tetramer (21). Injection of GMDCs, cDC2, and PBS failed to generate an endogenous CD8⁺ T-cell response, whereas injection of cDC1 expanded tetramer⁺CD8⁺ T cells substantially (Fig. 2C). These findings agree with previous reports that monocyte-derived GMDC vaccines rely on host cDCs for induction of antitumor immunity (18–20). However, we now show for the first time that cDC1 vaccines are able to directly present tumor-derived antigens to host CD8⁺ T cells in vivo in the absence of all endogenous cDC1.

We next asked whether vaccination with cDC1 and GMDCs could activate antitumor CD4⁺ T cells. WT or Irf8 +32^–/− mice were inoculated with 1956 mOVA tumor cells and vaccinated with cDC1s, cDC2s, GMDCs, or with PBS as a negative control. After 2 days, OT-II CD4⁺ T cells labeled with CTV were administered intravenously and analyzed 3 days later (Fig. 3A). In WT B6 mice, significant OT-II proliferation occurred, even without vaccination, consistent with the spontaneous tumor rejection in these mice. In contrast, in Irf8 +32^–/− mice receiving PBS virtually had no OT-II proliferation. This result is consistent with our previous report showing that cDC1 are required for early CD4⁺ T-cell activation by 1956 mOVA (21). However, robust OT-II proliferation was induced in tumor-bearing Irf8+32^–/− mice with cDC1 vaccination, but not with GMDC vaccination. For CD8⁺ T-cell responses, we demonstrated that vaccination with Flt3L-cDC1 led to direct presentation of tumor-derived antigens to host T cells without transfer to host DCs (Fig. 2).

Because GMDC vaccination failed to activate OT-II T cells in draining LNs, we asked whether they could migrate from the injection site to local LNs (Fig. 3B). To test this, we generated Flt3L-cDC1 or GMDCs from CD45.1⁺ C57BL/6 mice and vaccinated CD45.2⁺Irf8 +32^–/− mice previously inoculated with 1956 mOVA. After 2 days, we assessed migration of CD45.1 cells into LNs (Fig. 3B). As a positive control, we were able to identify CD45.1⁺ cells remaining at the tumor vaccination site after vaccination with cDC1 and GMDCs. After cDC1 vaccination, we identified CD45.1⁺ cDC1 in tumor-draining inguinal and axillary LNs, but not in the non-draining brachial LNs or contralateral inguinal LNs. In contrast, after GMDC vaccination, we could not identify CD45.1⁺ GMDCs in any LNs, suggesting reduced migratory capacity relative to cDC1. cDC1 remaining at the tumor site did not express CD40, whereas cDC1 that migrated to LNs expressed CD40 while retaining XCR1, MHCII, and CD11c expression (Fig. 3C).

A study suggests that cDC1 may be required within the tumor microenvironment (TME) in order to recruit tumor-specific T cells (40). We, therefore, asked whether cDC1 were required within the TME in the setting of endogenous T-cell responses and if cDC1 vaccination induced abscopal rejection of unvaccinated tumors in Irf8+32^–/– mice (Fig. 4A). 1956 mOVA was inoculated into both flanks of Irf8+32^–/– mice on day 0, followed by intratumoral vaccination on only one side with GMDCs, Flt3L-cDC1, Flt3L-cDC2, or PBS on days 1, 4, and 8. As before, vaccination with cDC1, but not GMDCs or cDC2, caused rejection of vaccinated tumors. cDC1 vaccination on one side also caused rejection of tumors on the contralateral, unvaccinated flank. Again, vaccination with cDC1, but not GMDCs or cDC2s, expanded H-2K^b-SIINFEKL tetramer⁺CD8⁺ T cells (Fig. 4B and C). cDC1 were not present in unvaccinated tumors of Irf8+32^–/– mice, suggesting that induction of T-cell responses was sufficient for mediating effective antitumor immunity without a requirement for cDC1 in the TME.

Because direct inoculation of cDC1 into tumors allowed sufficient antigen capture and transport to LNs to induce effective antitumor immunity, we asked whether intravenous vaccine infusion would be as effective. For this, we delivered GMDCs, Flt3L-cDC1s, and Flt3L-cDC2s intravenously into Irf8+32^–/– mice harboring 1956 mOVA (Fig.5). However, none of these conditions were sufficient to induce responses that could eliminate tumors (Fig. 5A), and only a small increase in H-2K^b-SIINFEKL tetramer⁺CD8⁺ T cells was observed (Fig. 5B and C).

Previously, DCs (41), and specifically cDC1 (42), were found to be required for optimal memory T-cell responses. However, whether tumor rejection requires optimized T-cell memory, or alternately can be achieved without cDC1-dependent restimulation is unknown. Therefore, we asked whether cDC1 were also required for antitumor memory responses. We inoculated Irf8+32^–/– mice with 1956 mOVA and administered the cDC1 vaccine as above (Fig. 2). As before, tumors were eliminated in all mice, which were then rested. After 3 weeks, these mice were reinoculated with 1956 mOVA on the contralateral flank and followed for tumor growth (Supplementary Fig. S1C). No residual cDC1 were present, and mice were not reinjected with the cDC1 vaccine. Nonetheless, the previously vaccinated Irf8+32^–/– mice showed complete rejection of secondary 1956 mOVA tumors. As controls, unvaccinated WT B6 mice, but not unvaccinated Irf8+32^–/– mice, also rejected 1956 mOVA tumors (Supplementary Fig. S1D). These results suggest that in the setting of 1956 mOVA fibrosarcomas, the effector memory population is sufficient to reinitiate direct tumor rejection independently of endogenous cDC1. These results do not indicate how long such antitumor immunity might last, but do suggest that some persistence of effector memory T-cell memory for at least 3 weeks.

Studies have identified several important differences between DCs derived from in vitro GM-CSF stimulation and conventional DCs developing naturally in vivo or from Flt3L-treated BM cultures (3, 4, 43, 44). Once considered homogeneous, work show that GM-CSF–derived DC systems actually produce heterogeneous populations, containing cells with features of both macrophages and DCs (3). The behavior of these cells differed from that of cDC1 in several ways. Although cDC1 require Irf8 and Batf3 for their development, GM-CSF–derived DCs require Irf4 and are independent of Batf3 (4). cDC1 rely on Rab43 to support cross-presentation, whereas GM-CSF–derived DCs do not (43). Likewise, Wdfy4 is required for cross-presentation of cell-associated antigens by cDC1, but not by GM-CSF–derived DCs (44). In summary, substantial differences exist in the biologic behavior of GM-CSF–derived DCs and native or cultured cDC1.

DC vaccination using GM-CSF–derived cells was originally thought to be mediated by direct presentation to host T cells. However, accumulating evidence has shown that host cDCs are required to mediate the effect of GMDC vaccines (18–20). First, a requirement for MHC-II expression on host cDCs was recognized as being required for responses induced by GMDC vaccines (18), indicating an indirect action at least for CD4⁺ T-cell activation. Later, a similar requirement was found for CD8⁺ T cells in a model using OT-I T cells (19). A study using OVA-loaded monocytes as a vaccine showed that CD8⁺ T-cell activation induced by OVA-peptide bearing monocytes was impaired in Batf3^–/– mice (20), which lack cDC1 development (26). A previous study used Batf3^–/– mice as a platform to test a vaccine's reliance on host cDC1 cells (20). However, cDC1 development in Batf3^–/– mice can be restored under certain conditions due to compensation by Batf and Batf2, as we previously reported (38). This indicates that Batf3^–/– mice are not an optimal system to determine a response's reliance on endogenous cDC1 because a restored host cDC1, rather than vaccine, could be responsible for direct T-cell activation. We previously developed an alternative method to eradicate cDC1 development. Deletion of the Irf8 +32kb enhancer eliminates BATF3-mediated Irf8 autoactivation that normally occurs in the specified pre-cDC1 progenitor and is required for commitment to the cDC1 lineage (31). cDC1 are completely missing in Irf8 +32^–/– mice and are not restored in any of the conditions that could restore cDC1 development in Batf3^–/– mice (31). Thus, Irf8 +32^–/– mice represent a robust platform to test the requirement for host cDC1 in the action of DC vaccines. Our results clearly demonstrated that the cDC1 used for vaccination could enter the tumor, capture antigens, and migrate to local LNs to directly prime CD8⁺ T cells. Using a system of Xcr1-Cre mediated gene targeting in vivo, we previously found that for cell-associated tumor antigens, cDC1 are not only responsible priming CD8⁺ T cells, but are also the predominant APC responsible for priming CD4 T cells (21). In agreement, here, we found that cDC1 used for vaccination could migrate to tumor-draining LNs and stimulate OVA-specific CD4⁺ T cells. By contrast, in this system, GMDCs did not migrate efficiently to tumor-draining LNs.

In the system used here, injection of the cDC1 vaccine directly into growing tumors allowed sufficient antigen capture to drive effective antitumor immunity, which was capable of eliminating not only tumors at the cDC1 vaccine injection site, but also the unvaccinated tumors growing concurrently on the opposite flank. That this abscopal effect occurred in Irf8 +32^–/– mice suggests that cDC1 vaccination generated effective antitumor immunity in draining LNs that is sufficient to achieve elimination without cDC1 being present at the tumor site.

A recent study suggests that cDC1 may be required in the TME to recruit tumor-specific T cells by producing chemokines such as CXCL10 (40). Mixed BM chimeras that either lack or possess endogenous cDC1 were treated with DTA to deplete cDCs, followed by adoptive transfer of in vitro–activated 2C T cells expressing cerulean fluorescent protein. In mice lacking cDC1, very few 2C T cells were found, whereas in chimeras with cDC1, 2C T cells were abundant in the TME (40). In contrast, in our study here, the abscopal rejection of opposite flank tumors occurred in the absence of cDC1 within the TME. Nonetheless, there are many differences between the systems used in the previous study (40) and ours, and it is plausible that the requirement for cDC1 in the TME may vary between different model systems. Intravenous vaccination with cDC1 did not induce sufficient antitumor immune responses for tumor rejection. This result may be due to a dilution of the numbers of cDC1 that reached the tumor, but also could point to an effect of tissue-imprinting on the cDC1’s ability to mount sufficient antitumor immunity. Resolving this issue will require additional studies.

In summary, this study showed that vaccines based on cDC1 and GMDCs differ substantially in their mechanism of action. GMDC vaccines were unable to migrate to LNs from tumors to LNs to directly prime host T cells, and so rely on antigen transfer to host DCs. In contrast, intratumoral cDC1 vaccines were able to capture tumor antigens, migrate to LNs, and directly prime host T cells. This difference supports the further interest in developing vaccines based on Flt3L-derived cDC1 as a possible therapeutic avenue for the future.

S.T. Ferris reports grants from Cancer Research Institute during the conduct of the study. R.D. Schreiber reports grants from NIH (NCI), Parker Institute for Cancer Immunotherapy, SU2C, Asher Biotherapeutics, Jounce Therapeutics, NGM Biotherapeutics; and grants from Sensei Biotherapeutics during the conduct of the study; personal fees from A2 Biotherapeutics, Arch Oncology, Asher Biotherapeutics, Codiak Biosciences, Jounce Therapeutics, NGM Biopharmaceuticals, Sensei Biotherapeutics; and other support from Cancer Immunology Research outside the submitted work. K.M. Murphy reports grants from NIH; and grants from NIH during the conduct of the study. No disclosures were reported by the other authors.

One of the Editors-in-Chief of Cancer Immunology Research is an author on this article. In keeping with AACR editorial policy, a senior member of the Cancer Immunology Research editorial team managed the consideration process for this submission and independently rendered the final decision concerning acceptability.

This publication is solely the responsibility of the authors and does not necessarily represent the official view of the NIH.

S.T. Ferris: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. R.A. Ohara: Resources, formal analysis, validation, investigation, writing–review and editing. F. Ou: Software, formal analysis, investigation. R. Wu: Formal analysis, investigation, writing–review and editing. X. Huang: Formal analysis, validation, investigation. S. Kim: Formal analysis, visualization, methodology. J. Chen: Conceptualization, investigation. T.T. Liu: Conceptualization, resources. R.D. Schreiber: Conceptualization, resources, supervision. T.L. Murphy: Conceptualization, resources, formal analysis, supervision, validation, methodology, writing–review and editing. K.M. Murphy: Conceptualization, resources, data curation, supervision, funding acquisition, visualization, methodology, writing–original draft, project administration, writing–review and editing.

This work was supported by the NIH (R01AI150297, R01CA248919, and R21AI164142 to K.M. Murphy, R01CA240983 TO R.D. Schreiber, and F30CA247262 to R.W.) S.T. Ferris and X. Huang are Cancer Research Institute Irvington Fellows supported by the Cancer Research Institute.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.