In vitro-generated CD103⁺ cDC1s29 were cultured (2.5–4.5×10⁶/mL) in Roswell Park Memorial Institute (RPMI) 1640 medium containing 10% heat-inactivated fetal bovine serum (FBS) (Atlanta Biologicals, Atlanta, Georgia, USA), 1% penicillin-streptomycin, 1 mM sodium pyruvate, and 50 µM β-mercaptoethanol (RPMI 1640 complete medium) ± 20 µg/mL poly I:C (Sigma, St Louis, Missouri, USA), 5% XG-3 supernatant or 20 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF) and the surrogate tumor antigens ovalbumin (OVA; 100–400 µg/mL) (Sigma) or K7M3 tumor lysate for 4 hours. Cells were washed twice with phosphate-buffered saline (PBS) and injected (0.5–3×10⁶) into melanoma or osteosarcoma tumors (intratumoral (i.t.)), one or two times 4–7 days following tumor implantation, when tumor sizes reached 10–20 mm². In bilateral tumor assays, poly I:C-treated and tumor antigen-loaded CD103⁺ cDC1s (2–3×10⁶ cells) were injected into left-side tumors; right-side tumors remained untreated. For metastasis assays, poly I:C-treated and tumor antigen-loaded CD103⁺ cDC1s (1–2×10⁶ cells) were injected intravenously (i.v.) approximately 30 days prior to metastasis challenge. In some metastasis assays, mice were injected i.v. with CD45.1⁺ CD45.2⁺ OT-I CD8⁺ T cells (1×10⁶ cells) 1 day prior to CD103⁺ cDC1 delivery. For combination treatments with checkpoint inhibitors, DC-vaccinated animals were injected intraperitoneally (i.p.) with PD-1 antibody (RMP1-14; 200 µg/mouse) or CTLA-4 antibody (9H10; 200 µg/mouse for the first treatment, 100 µg/mouse for subsequent treatments) 4 days following tumor implantation, with two antibody treatments per week for 3 weeks. PD-1 and CTLA-4 antibodies were purchased from BioXCell company (West Lebanon, New Hampshire, USA).

Murine bone marrow (BM) cultures were initiated at 0.5×10⁶ cells/mL in RPMI 1640 complete medium containing 20 ng/mL GM-CSF. Cells were collected following 6–7 days of culture and MoDCs (CD11c⁺ cells) were purified using a FACSAria II or III. MoDCs were pretreated with lipopolysaccaride (LPS) (100 ng/mL), OVA (400 µg/mL), and GM-CSF (20 ng/mL) for 4 hours prior to i.t. delivery.

Murine B16-F10 melanoma cells expressing OVA (B16-OVA cells) were cultured in Dulbecco’s modified Eagle medium (DMEM) medium (Gibco, Grand Island, New York, USA) containing 10% FBS and 1% penicillin-streptomycin. For single primary tumors, C57BL/6J mice were shaved on the abdomen and injected subcutaneously (s.c.) with 4×10⁵ melanoma cells on one side. Tumor size (length x width) was measured every 2–3 days using a caliper; mice were euthanized when tumor sizes reached a maximum of 20 mm in any direction or tumor ulceration occurred. To establish bilateral tumors, C57BL/6J mice were injected s.c. with 3×10⁵ melanoma cells on both sides of the abdomen; tumor sizes were monitored every 2–3 days and animals euthanized on one tumor reaching 15 mm in any direction or tumor ulceration. For metastasis assays, C57BL/6J mice were injected with 7.5×10⁵ B16-OVA melanoma cells i.v., approximately 1 month after CD103⁺ cDC1s were administered. Mice were euthanized 14 d after i.v. melanoma delivery; lung metastases and immune subsets were quantified.

Murine K7M3 cells (a subline of K7M2)30 were cultured in DMEM medium (Gibco) containing 10% FBS, 1% non-essential amino acids, 1% minimal essential medium vitamin solution, 1 mM sodium pyruvate, and 1% penicillin-streptomycin. Balb/c mice were shaved on the abdomen and injected s.c. with 2×10⁶ K7M3 cells. Tumor size (length x width) was measured every 2–3 days using a caliper; mice were euthanized when tumor sizes reached a maximum of 15 mm in any direction or tumor ulceration occurred. For metastasis assays, Balb/c mice were injected with 2×10⁶ K7M3 cells i.v., approximately 1 month after CD103⁺ cDC1s were administered. Mice were euthanized 28 days later; lung metastases and immune subsets were quantified.

Statistical analyses were performed using GraphPad Prism V.7 (La Jolla, California, USA). Data are presented as mean±SEM. Statistical significance was determined by unpaired Student’s t-test, one-way analysis of variance (ANOVA), two-way ANOVA with multiple comparisons, or log-rank (Mantel-Cox) test as indicated in the Figure legends. Differences were considered significant when p=0.05.

Mice, CD103⁺ cDC1 culture, immune cell profiling, T cell proliferation, DC proliferation and survival, K7M3 tumor lysate preparation, cytokine detection, and quantitative PCR.

These methods are described in the online supplementary material.

We applied a previously described culture system29 to produce murine nonlymphoid organ CD103⁺ cDC1s (figure 1A and online supplementary figure S1A). In vitro-generated CD103⁺ cDC1s expressed cDC1 markers XCR1, CLEC9A, and CD24, while expression of the cDC2 marker CD172α, or myeloid markers Ly6G and Ly6C, was negligible (figure 1A and online supplementary figure S1B). Moreover, the CD103⁺ cDC1s expressed canonical cDC1 transcriptional regulators including Batf3, Id2, and Irf8. By contrast, expression of pDC (Tcf4, Zeb2) or cDC2 (Irf4) transcription factors was minimal (online supplementary figure S1C).31 In vitro-generated CD103⁺ cDC1s efficiently cross-presented OVA-derived peptide to activate OT-I CD8⁺ T cells on stimulation with poly I:C and GM-CSF (figure 1B and online supplementary figure S1D). In addition, poly I:C-treated CD103⁺ cDC1s produced IL-12p70 and CXCL10, and upregulated MHC-I and cell surface costimulatory molecules CD80, CD86, and CD40 (figure 1C and online supplementary figure S1E,F). MHC-II and toll-like receptor 3 (TLR3) were highly expressed on stimulated and untreated CD103⁺ cDC1s (online supplementary figure S1F). In vitro-generated CD103⁺ cDC1s demonstrated the ability to stimulate IFN-γ-producing OT-II CD4⁺ T cells in culture, suggesting T helper 1 (Th1) induction, but did not promote IL-17⁺ cell (Th17) generation to a significant degree (online supplementary figure S1G). Collectively, our data indicate in vitro-generated CD103⁺ cDC1s resemble the nonlymphoid organ cDC1 subset found in vivo in several parameters important for effective Th1 and CD8⁺ T cell activation.

On i.t. vaccination, poly I:C-activated, OVA-loaded CD103⁺ cDC1s restrained melanoma growth and prolonged mouse survival more effectively than controls (PBS treatment) or CD103⁺ cDC1s stimulated with poly I:C alone (figure 1D and online supplementary figure S2A). Furthermore, i.t. CD103⁺ cDC1 delivery associated with increased amounts of IFN-γ-producing CD8⁺ T cells in tumor-draining LNs (TdLNs) at maximum tumor burden (figure 1E,F). Tumor-infiltrating CD8⁺ T cells were also enhanced by CD103⁺ cDC1 vaccination, although the difference from controls was not significant (figure 1G). CD103⁺ cDC1 vaccination increased CD4⁺ CD25⁺ Foxp3⁺ T regulatory cells (Tregs) in TdLNs, while TdLN pDCs were significantly reduced on poly I:C and OVA-activated CD103⁺ cDC1 treatment (online supplementary figure S2B,C). Other TdLN immune subsets analyzed were unaffected (online supplementary figure S2B,C). These results suggest CD103⁺ cDC1 vaccination inhibits melanoma growth by promoting CD8⁺ T cell responses in vivo.

To address whether the CD103⁺ cDC1 vaccine elicited systemic tumor immunity, mice bearing bilateral melanoma tumors were treated with poly I:C-stimulated, OVA-pulsed CD103⁺ cDC1s i.t. on the left-side, while right-side tumors remained untreated. Growth of both tumors was restrained by unilateral CD103⁺ cDC1 vaccination, and survival of vaccinated mice was prolonged over controls (figure 1H,I). Earlier vaccination of smaller tumors also correlated with improved tumor control (figure 1D,H). Together, our data indicate immunization with in vitro-generated CD103⁺ cDC1s drives systemic immunity and controls local and distal melanoma growth effectively.

Since MoDC-based vaccines have been used clinically, we compared the antitumor efficacy of in vitro-generated CD103⁺ cDC1s with MoDCs. Mice bearing bilateral melanoma tumors were injected i.t. with OVA-loaded and poly I:C-treated CD103⁺ cDC1s, or OVA-loaded and LPS-treated MoDCs, on the left-side while right-side tumors remained untreated (figure 2A). We used distinct TLR agonists to activate each DC population, as expression of poly I:C-responsive TLR3 is enriched in CD103⁺ cDC1s, while LPS-responsive TLR4 is abundantly expressed by MoDCs.4 32 Vaccination with MoDCs induced systemic antitumor responses, however, CD103⁺ cDC1s showed superior activity in suppressing bilateral tumor growth (figure 2B). In addition, survival of CD103⁺ cDC1 vaccinated mice was prolonged significantly compared with MoDC- or PBS-treated animals (figure 2C), indicating CD103⁺ cDC1s elicit more effective antitumor immunity versus MoDCs.

To evaluate potential mechanisms for the improved response to CD103⁺ cDC1 vaccination, we compared DC amounts in vaccinated and unvaccinated tumors and corresponding TdLNs 40 hours following i.t. delivery. Congenic CD45.1⁺ CD103⁺ cDC1s or CD45.1⁺ MoDCs were used in these assays to distinguish from endogenous populations. CD45.1⁺ CD103⁺ cDC1s were detected in tumors and TdLNs on the side of i.t. injection only (figure 2D–F). In addition, CD45.1⁺ CD103⁺ cDC1s were more abundant in TdLNs compared with endogenous cells, as judged by relative frequencies of CD45.1⁺ versus CD45.2⁺ CD103⁺ cDC1s on the treatment side (figure 2F). By contrast, the majority of CD45.1⁺ MoDCs remained within injected tumors; CD45.1⁺ MoDCs were rarely observed in TdLNs (figure 2G–I). The chemokine receptor CCR7, which mediates DC migration to LNs, was expressed at higher amounts on tumor-associated CD45.1⁺ CD103⁺ cDC1s compared with CD45.1⁺ MoDCs, while TdLN-associated populations expressed similar CCR7 amounts (online supplementary figure S2D). In addition, despite i.t. injection of equivalent DC numbers, CD45.1⁺ MoDC amounts were approximately 10% of CD45.1⁺ CD103⁺ cDC1s within tumors (figure 2D,G). These results suggest CD103⁺ cDC1s persist in tumors longer, express greater amounts of CCR7, and have superior TdLN accumulation versus MoDCs on vaccination.

We analyzed DC proliferation and survival in vaccine conditions, to further understand the improved efficacy of CD103⁺ cDC1s. These assays revealed modest but non-significant differences in Ki67 status prior to vaccination, and no detectable differences 40 hours following i.t. delivery, suggesting similar proliferation rates (figure 2J). By contrast, the activated CD103⁺ cDC1 vaccine contained fewer apoptotic and dead cells, and a greater proportion of viable cells prior to i.t. delivery, relative to the MoDC vaccine (figure 2K). In addition, CD103⁺ cDC1s showed enhanced viability following delivery to melanoma tumors versus MoDCs, although the viability of both populations 40 hours after vaccination was low (figure 2K and online supplementary figure S2E). These results indicate CD103⁺ cDC1s have improved survival following antigen and TLR agonist stimulation, as well as on exposure to melanoma tumors, compared with MoDCs, suggesting enhanced CD103⁺ cDC1 viability contributes to the efficacy of this population as a tumor vaccine.

To further examine mechanisms by which the CD103⁺ cDC1- and MoDC-based vaccines mediate distinct antitumor efficacies, we evaluated their ability to stimulate T cell proliferation in vitro. Poly I:C-activated CD103⁺ cDC1s showed modestly enhanced ability to induce CD8⁺ T cell proliferation versus poly I:C-stimulated MoDCs, while LPS-treated DCs demonstrated indistinguishable activity (online supplementary figure S3A). CD103⁺ cDC1s, however, were significantly inferior to MoDCs in their ability to stimulate CD4⁺ T cell proliferation in vitro (online supplementary figure S3A).

To evaluate whether differential T cell responses were elicited by each DC population in vivo, we investigated immune responses in tumors and TdLNs during the predicted T cell expansion phase,17 14 days after i.t. vaccination. CD8⁺ T cell amounts were increased in tumors vaccinated with CD103⁺ cDC1s, compared with MoDC-vaccinated tumors, PBS-treated controls and distal untreated tumors (figure 3A). Moreover, OVA-specific CD8⁺ T cells within treated as well as distal untreated tumors were increased significantly on vaccination with CD103⁺ cDC1s, as judged by analysis of SIINFEKL/H-2Kb pentamer⁺ CD8⁺ T cells (figure 3B,C). By contrast, tumors in MoDC-vaccinated mice or controls did not accumulate OVA-specific CD8⁺ T cells (figure 3B,C). In addition, vaccine-origin CD45.1⁺ CD103⁺ cDC1s isolated 40 hours following i.t. delivery showed improved ability to stimulate OT-I CD8⁺ T cell proliferation in vitro, versus vaccine-origin CD45.1⁺ MoDCs (figure 3D).

The CD103⁺ cDC1 vaccine also promoted an increase in tumor-infiltrating CD4⁺ T cells, including Treg and T effector subsets, in vaccinated tumors but not distal untreated tumors, compared with MoDC vaccination or PBS treatment (figure 3E–G). The relative frequency of Tregs within the tumor-infiltrating CD4⁺ T cell population was reduced in DC vaccine-treated mice, however, compared with controls (figure 3H). Furthermore, vaccine-origin CD45.1⁺ CD103⁺ cDC1s and CD45.1⁺ MoDCs isolated from tumors were equivalently effective at stimulating CD4⁺ T cell proliferation in vitro, although CD4⁺ T cell proliferation was notably reduced compared with CD8⁺ T cell proliferation in these conditions (figure 3D,I). Tumor-infiltrating B cells, monocytes, macrophages, cDC1s, and cDC2s were not significantly different in treated or untreated tumors among the three groups, while neutrophils were induced in tumors on CD103⁺ cDC1 but not MoDC vaccination (online supplementary figure S3B).

Within TdLNs from the vaccine-treated side, MoDC vaccination was associated with induction of CD4⁺ and CD8⁺ T cells, as well as IFN-γ⁺ CD4⁺ Th1 and IL-4⁺ CD4⁺ Th2 subsets, compared with PBS or CD103⁺ cDC1 vaccination (online supplementary figure S3C). TdLNs on the untreated side did not show appreciable changes in T cell amounts on DC vaccination (online supplementary figure S3C). Furthermore, we did not detect differences in SIINFEKL/H2Kb pentamer⁺ CD8⁺ T cells, Tregs, B cells, myeloid subsets or DC populations in TdLNs among the three treatment groups (online supplementary figure S3C–F). Vaccine-origin CD45.1⁺ MoDCs isolated from TdLNs, however, were superior to TdLN-derived vaccine CD45.1⁺ CD103⁺ cDC1s in eliciting CD4⁺ T cell proliferation in vitro, while both DC populations from TdLNs stimulated CD8⁺ T cell proliferation in vitro similarly (online supplementary figure S3G). Collectively, our data suggest a propensity for CD103⁺ cDC1s to elicit CD8⁺ T cell responses, while MoDCs promote greater CD4⁺ T cell proliferation in vitro. Moreover, our results indicate in vitro-generated CD103⁺ cDC1s enhance systemic immunity, as well as i.t. accumulation of total CD8⁺ and tumor antigen-specific CD8⁺ T cells, more effectively than MoDC vaccination.

Metastasis treatment remains a major clinical challenge33; thus, we investigated whether the CD103⁺ cDC1 vaccine would control metastatic tumors. We transferred tumor antigen (OVA)-specific CD45.1⁺ CD45.2⁺ OT-I CD8⁺ T cells to C57BL/6J mice (CD45.2⁺) and vaccinated animals with poly I:C activated and OVA-pulsed CD103⁺ cDC1s 1 day later by i.v. delivery. After approximately 1 month, to encompass initial T cell expansion and memory generation phases,34 animals were challenged i.v. with B16-OVA cells to establish experimental lung metastasis (figure 4A). This strategy allowed us to examine whether the in vitro-generated CD103⁺ cDC1 vaccine elicits long-lasting antitumor immunity on combination treatment with adoptive T cell transfer. Our experiments revealed significant restraint of lung metastases in CD103⁺ cDC1 vaccine-treated animals versus controls (figure 4B,C), indicating CD103⁺ cDC1s delivered i.v. stimulate a durable antitumor immune response that suppresses pulmonary metastases.

To evaluate immune responses induced on T cell transfer and i.v. CD103⁺ cDC1 vaccination, we measured IFN-γ-producing CD8⁺ T cells in lymphoid organs 14 days following i.v. challenge with B16-OVA. These assays revealed increased amounts of CD45.1⁺ IFN-γ⁺ OT-I CD8⁺ T cells in spleen and inguinal LNs relative to controls (figure 4D,E; online supplementary figure S4A). Importantly, CD45.1⁺ IFN-γ⁺ OT-I CD8⁺ T cells were elevated approximately 6 weeks after OT-I T cell transfer and CD103⁺ cDC1 vaccination, suggesting long-term maintenance of the OVA antigen-specific T cell response. CD103⁺ cDC1 vaccination also enhanced endogenous CD45.1^- IFN-γ⁺ CD8⁺ T cell amounts in spleen, while their amounts in inguinal and lung-draining LNs showed an increased trend (figure 4D,F and online supplementary figure S4A,B). Furthermore, IFN-γ⁺ CD4⁺ Th1 cells were increased in spleen, and inguinal and lung-draining LNs, following CD103⁺ cDC1 vaccination (figure 4G,H and online supplementary figure S4C,D). CD103⁺ cDC1 vaccination did not affect other CD4⁺ T cell subsets, B cells, or myeloid cells in lymphoid organs appreciably (figure 4H; online supplementary figure S4C–G). These results suggest CD103⁺ cDC1 vaccination induces CD8⁺ T cell and Th1 responses without eliciting global inflammatory or immune system activity.

We next treated mice by i.v. CD103⁺ cDC1 delivery alone, and challenged animals approximately 1 month later with melanoma cells i.v., to examine whether cotransfer with antigen-specific OT-I CD8⁺ T cells was required for metastasis suppression. These assays revealed CD103⁺ cDC1 vaccination alone controlled lung metastases (online supplementary figure S4H,I), suggesting activation of endogenous immune responses by the CD103⁺ cDC1 vaccine is sufficient to inhibit metastatic tumors. Collectively, our data imply CD103⁺ cDC1 vaccination confers effective protection against primary melanoma tumors as well as melanoma metastases by eliciting long-lasting, tumor antigen-specific CD8⁺ T cell-mediated and Th1-mediated immunity.

It is important to expand investigation of immunotherapeutic approaches in tumor types that are refractory to current treatments. We selected osteosarcoma for a second model since this tumor presents in pediatric populations and relapsed or metastatic disease has limited therapeutic options. K7M3 osteosarcoma-bearing mice (Balb/c) were vaccinated with Balb/c BM-derived CD103⁺ cDC1s following activation by poly I:C and incubation with K7M3 tumor lysate (tumor antigen). CD103⁺ cDC1s restrained osteosarcoma growth and prolonged mouse survival (figure 5A,B). Moreover, the majority of mice rechallenged with K7M3 cells did not develop tumors (figure 5B), suggesting CD103⁺ cDC1 vaccination elicits long-lasting protection against osteosarcoma. Notably, IFN-γ⁺ CD8⁺ T cell and Th1 infiltration into osteosarcoma tumors was increased on CD103⁺ cDC1 vaccination, as judged by analysis at maximum tumor burden, whereas tumor infiltration of other T cell subsets, B cells, myeloid subsets, or DC populations was not affected (figure 5C and online supplementary figure S5A). Moreover, TdLN T cell subsets were unchanged on CD103⁺ cDC1 vaccination (online supplementary figure S5B). These results are consistent with our findings in melanoma, suggesting CD103⁺ cDC1 vaccination induces IFN-γ⁺ CD8⁺ T cell and Th1 responses, as well as i.t. accumulation of IFN-γ⁺ CD8⁺ T cells, regardless of tumor type.

To examine whether CD103⁺ cDC1 vaccination protected against osteosarcoma pulmonary metastases, we delivered poly I:C-activated and K7M3 tumor lysate-pulsed CD103⁺ cDC1s i.v. and challenged mice 40 days later with K7M3 tumor cells i.v. (figure 5D). After 4 weeks, pulmonary metastatic foci were visible in controls, yet less frequent in CD103⁺ cDC1-vaccinated mice (figures 5E,F; online supplementary figure S6). Lung weights, as well as total tumor and necrotic areas in lung, were lower in mice vaccinated with CD103⁺ cDC1s versus controls (figure 5G–J, online supplementary figure S6). These data indicate CD103⁺ cDC1 vaccination effectively protects animals from experimental osteosarcoma lung metastases.

CTLA-4 and PD-1 checkpoint blockade have modest impact in osteosarcoma.27 28 35 To examine whether combination treatment with CD103⁺ cDC1 vaccination improves their efficacy, mice were treated with the CD103⁺ cDC1 vaccine alone or in combination with CTLA-4 or PD-1 antibody (figure 6A). CTLA-4 or PD-1 blockade alone restrained osteosarcoma growth in some but not all mice (figure 6B). Similar results were found for animals vaccinated with CD103⁺ cDC1s alone or CD103⁺ cDC1 vaccination and PD-1 blockade (figure 6B). By contrast, CD103⁺ cDC1 vaccination and CTLA-4 blockade significantly impaired osteosarcoma growth and led to complete tumor regression in 100% of mice approximately 2 weeks following initiation of therapy (figure 6B,C) These results suggest combination therapy with CD103⁺ cDC1s and CTLA-4 checkpoint blockade effectively restrains osteosarcoma.

To examine whether individual or combination immunotherapies induced long-lasting antitumor immunity, we rechallenged the surviving mice from each therapeutic group with K7M3 tumors. Tumor-free survival was observed in the majority of mice up to 370 days following tumor rechallenge (figure 6D). These data indicate single or combination immune therapy is capable of establishing immunological memory responses against murine K7M3 osteosarcoma, with the most potent response mediated by combining CD103⁺ cDC1 vaccination and CTLA-4 blockade.