Antigen presentation requirements for effective cDC1-based cancer immunotherapy Authors: Josué E Pineda, Tomoyuki Minowa, Li Shen, Yifan Zhou, Allison Dyevoich, Bhakti Patel, Sarah M Schneider, Sunita Keshari, Akata Saha, Morgan N Riba, Jing Wang, Stephanie S Watowich, Matthew M Gubin Published: 2026 ABSTRACT AbstractType 1 conventional dendritic cells (cDC1s) are important for generating and sustaining antitumor immunity. Accordingly, the abundance of cDC1s in human tumors correlates with improved outcomes in cancer. Capitalizing on this role, we previously demonstrated that vaccination with murine cDC1s, generated in culture from bone marrow cells (termed here “in vitro–derived cDC1s”), elicits durable tumor control in multiple preclinical models; however, the immunological mechanisms underlying the efficacy of cDC1 vaccination remain unclear. Here, we examined whether in vitro–derived cDC1s resemble tumor-infiltrating DC populations and whether MHC-I and MHC-II antigen presentation contribute to cDC1-mediated tumor control following vaccination in melanoma. As expected, MHC-I or MHC-II deficiency had minimal impact on the transcriptional state of cDC1s in homeostasis or following stimulation with the adjuvant poly dI:dC. Moreover, in vitro–derived cDC1s cultured under steady-state conditions closely resembled tumor-infiltrating cDC1s, whereas their poly dI:dC–stimulated counterparts resembled CCR7+ tumor-infiltrating DC populations, also referred to as mregDCs or LAMP3+ DCs. Our data further show that both MHC-I and MHC-II contribute to tumor control upon cDC1 vaccination and that coexpression of MHC-I and MHC-II on the same cDC1 is necessary for a robust vaccine response. We also identified an important function for host cDC1s in supporting the efficacy of vaccination with in vitro–derived cDC1s, as judged by impaired tumor control in Irf8 + 32−/− mice, which lack endogenous cDC1s. Overall, these results indicate that effective antitumor responses depend on MHC-I and MHC-II antigen presentation by vaccine-delivered cDC1s, with additional contributions from host cDC1s.Keywords: dendritic cells, cancer vaccines, antigen presentation, MHC class I, MHC class II Introduction Dendritic cells (DCs) have gained attention in cancer immunotherapy research because of their exceptional capacity to present tumor antigens and coordinate T cell–mediated immunity.1–6 Several clinical strategies have explored DC-based vaccines, including those using monocyte-derived DCs (MoDCs) and in vivo antigen-targeting approaches that exploit DC-specific receptors.4,7–11 One of the earliest personalized neoantigen vaccine clinical trials utilized MoDCs, finding that these vaccines could elicit de novo tumor antigen–specific T-cell responses and boost preexisting responses in patients with melanoma.7 However, despite measurable immunogenicity across these different approaches, DC-based cancer vaccines have achieved limited clinical efficacy, emphasizing the need to improve DC-based therapeutic strategies.12,13 Among DC populations, type 1 conventional DCs (cDC1s) are a specialized subset with optimal ability to cross-present exogenous antigens on MHC class I (MHC-I) to prime CD8+ T cells, a process essential for tumor rejection.6,14–16 The development of cDC1s depends on the transcription factors IRF8, ID2, and BATF3, as well as signaling through the Fms-like tyrosine kinase 3 receptor (FLT3) in response to its ligand Flt3L, which are critical for cDC1 differentiation.15 Under homeostatic conditions, cDC1s reside within lymphoid organs and peripheral tissues such as the skin, lung, and intestine, where they continuously sample and present antigens. Tissue-resident cDC1s are capable of undergoing homeostatic or pathogen-induced migration to lymph nodes (LNs), enabling cDC1s to regulate tolerogenic or adaptive immune responses via antigen presentation to T cells.15 cDC1s act as central coordinators of antitumor immunity through multiple complementary mechanisms. In tumor-draining LNs (TdLNs), cDC1s provide canonical antigen presentation and costimulatory signals to activate naïve antitumor T-cell populations. cDC1s are also proficient in cross-presentation of exogenous tumor antigens on MHC-I to stimulate CD8+ T-cell responses. Simultaneously, cDC1s present antigen on MHC-II to engage CD4+ T cells. Prior studies show that upon CD4+ T-cell engagement, CD4+ T cells license cDC1s through CD40 ligand (CD40L)–CD40 interactions, which enhances the cDC1’s ability to prime cytotoxic CD8+ T cells and sustain productive antitumor immunity.17 In addition to their antigen-presenting functions, cDC1s secrete cytokines such as interleukin 12 (IL-12) and type I interferons that promote T-cell activation, survival, and effector differentiation, driving polarization of IFN-γ+ CD4+ T cells (Th1 cells) and CD8+ T-cell cytotoxic responses.18–20 Within the tumor microenvironment (TME), cDC1s facilitate the recruitment and localization of effector T cells through the production of chemokines such as CXCL9 and CXCL10.20–22 Accordingly, the abundance and activation of cDC1s within tumors correlates with enhanced antitumor immune activity, increased responsiveness to immune checkpoint blockade, and improved clinical outcomes across multiple cancer types.1,3,4,15,16,23–27 However, many tumors exhibit a profound deficiency or dysfunction in cDC1s, contributing to limited T-cell infiltration and inadequate immune control.25,28 Therefore, strategies that enhance cDC1 activation, antigen presentation, and persistence within the TME hold considerable potential to improve cancer immunotherapy outcomes. We previously established a cDC1-based vaccine using cells generated in culture from murine bone marrow (termed here “in vitro–derived cDC1s”).23,25,26,29 Our studies demonstrated that delivery of adjuvant-activated, antigen-loaded cDC1s to the TME elicits systemic and durable antitumor immunity; moreover, intravenous delivery of the cDC1 vaccine restrains outgrowth of distal metastases. We also showed that cDC1 vaccination has superior tumor control relative to MoDC vaccination and demonstrated efficacy against a variety of murine solid tumors, suggesting tumor-agnostic activity.23,25,26,29 Recent reports from others have confirmed the efficacy of cDC1 vaccination and demonstrated its superiority to vaccination with type 2 cDCs (cDC2s).30 Despite these advances, the immunological mechanisms by which cDC1 vaccination elicits robust tumor control remain unknown. Here, we investigate whether MHC-I and MHC-II antigen presentation and the presence of endogenous cDC1s contribute to the efficacy of adjuvant-activated, tumor antigen–loaded cDC1 vaccines. Materials and methods Mouse strains All mice used were obtained from The Jackson Laboratory (JAX, Bar Harbor, ME, USA) and were on the C57BL/6 background. These include C57BL/6J wild-type (WT; JAX Stock No. 000664), B6.129P2-H2-K1b-tm1Bpe H2-D1b-tm1Bpe/DcrJ (MHC-IKO; JAX Stock No. 019995), B6.129S2-H2dlAb1-Ea/J (MHC-IIKO; JAX Stock No. 003584), C57BL/6-Rr172em1Kmm/J (Irf8 + 32−/−; JAX Stock No. 032744), and C57BL/6-Tg(TcraTcrb)1100Mjb/J (OT-I; JAX Stock No. 003831) mice.31–36 To generate MHC-IKO mice, B6.129P2-H2-K1b-tm1Bpe H2-D1b-tm1Bpe/DcrJ sperm was retrieved from cryopreservation and implanted into C57BL/6J mice by JAX. Progeny were obtained and bred in-house to generate double-knockout mice lacking H2-K1 and H2-D1 (H2-Kb and H2-Db). All experiments were conducted using 8- to 12-week-old mice. Mice were maintained under specific pathogen–free conditions at The University of Texas MD Anderson Cancer Center (Houston, TX, USA) and used in accordance with Institutional Animal Care and Use Committee (IACUC)–approved protocols. Generation of cDC1s in culture Culture conditions to generate cDC1s were adapted from a protocol developed by Mayer et al.37 and performed using methods previously described by our laboratory.23,25,26,29 Murine bone marrow cells (1.5 × 106 cells/mL) were cultured in RPMI 1640 containing 10% heat-inactivated fetal bovine serum (FBS; Atlanta Biologicals, Atlanta, GA, USA), 1% penicillin-streptomycin, 1 mM sodium pyruvate, and 50 µM β-mercaptoethanol (complete RPMI), and supplemented with 2 ng/mL murine granulocyte-macrophage colony-stimulating factor (mGM-CSF; PeproTech, Rocky Hill, NJ, USA) and 50 ng/mL human Flt3L (hFlt3L; PeproTech). On day 9, nonadherent cells were collected and replated in fresh complete RPMI at 0.3 × 106 cells/mL with 2 ng/mL mGM-CSF and 50 ng/mL hFlt3L. On days 15 to 17, nonadherent cells were collected and cDC1s (CD11c+CD45R−CD24+CD172α−CD103+) were purified by fluorescence-activated cell sorting (FACS) using a FACSAria or FACSAria Fusion cell sorter (BD Biosciences, Palo Alto, CA, USA). Preparation and administration of cDC1 vaccines FACS-purified cDC1s were cultured (2 to 4.5 × 106 cells/mL) in complete RPMI supplemented with 20 ng/mL mGM-CSF, 20 µg/mL poly(deoxyinosinic-deoxycytidylic) acid (poly dI:dC; Sigma-Aldrich, St. Louis, MO, USA), and 200 µg/mL ovalbumin (OVA; Sigma-Aldrich) for 4 hours at 37 °C. Following stimulation, cells were washed 3 times with phosphate-buffered saline (PBS) and resuspended in endotoxin-free PBS. For intratumoral vaccination, 3 × 106 cDC1s were injected directly into established tumors of mice bearing OVA-expressing B16-F10 melanoma (B16-OVA) on days 4 and 7 after tumor implantation, as previously described.26 For mixed vaccines (MixKO), mice received a total of 6 × 106 cDC1s per treatment, composed of equal numbers of MHC-IKO cDC1s and MHC-IIKO cDC1s (3 × 106 each). For experiments assessing CD40 stimulation, cDC1s were stimulated with 20 ng/mL mGM-CSF, 20 µg/mL poly dI:dC, and 200 µg/mL OVA, in the presence or absence of 20 µg/mL agonistic anti-CD40 antibody (αCD40; Leinco Technologies, Fenton, MO, USA). Cells were washed with PBS and prepared for intratumoral injection as described above. Tumor transplantation and monitoring Murine B16-OVA melanoma cells, originally generated by transfection of B16-F10 cells with a plasmid encoding full-length OVA under a constitutive promoter,38 were maintained in Dulbecco’s modified Eagle medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 10% FBS and 1% penicillin-streptomycin (complete DMEM). Cells were passaged 3 to 5 times prior to use. WT C57BL/6J or Irf8 + 32−/− mice were shaved and injected subcutaneously on the flank with 3 × 105 B16-OVA cells in 150 µL of endotoxin-free PBS. Tumor length and width were measured every 2 to 3 days using digital calipers. Mice were euthanized when tumor diameters reached 15 mm in any direction, or when ulceration occurred, in accordance with IACUC-approved protocols. Activation and phenotypic analysis of cDC1s FACS-purified cDC1s were cultured at 2 to 4.5 × 106 cells/mL in complete RPMI supplemented with 20 µg/mL poly dI:dC, 20 ng/mL mGM-CSF, and 200 µg/mL OVA ± 20 µg/mL αCD40 for 4, 16, or 24 hours at 37 °C, or were left untreated. Following culture, cells were washed 3 times with PBS and incubated with Fc block (anti-mouse CD16/32; clone 2.4G2, Tonbo Biosciences, San Diego, CA, USA) in FACS buffer (PBS supplemented with 2% FBS and 2 mM EDTA) for 15 minutes at 4 °C. Cells were then stained with master mixes comprising fluorescent antibodies specific for murine surface markers for 15 to 30 minutes at 4 °C, washed in FACS buffer, and prepared for flow cytometric analysis. Antibodies used for phenotypic analysis included APC-conjugated CD70 (FR70) and CD45R/B220 (RA3-6B2); BV605-conjugated CD103 (2E7); BV711-conjugated CD40 (3/23), CD24 (M1/69), and MHC-II (I-A/I-E; M5/114.15.2); BV785-conjugated CD86 (GL-1); eFluor 450–conjugated CD11c (N418); FITC-conjugated CD172α/Sirpα (P84) and H-2Kb (AF6-88.5); PE-Cy7–conjugated CD70 (FR70) and CD80 (16-10A1); PerCP-Cy5.5–conjugated CD40 (3/23) and CD24 (M1/69); and BV650-conjugated MHC-II (I-A/I-E; M5/114.15.2). All antibodies were purchased from Tonbo Biosciences, BD Biosciences, Thermo Fisher Scientific (Waltham, MA, USA), or BioLegend (San Diego, CA, USA). Dead cells were excluded using Ghost Dye Violet 510 (Tonbo Biosciences). Samples were acquired on a BD LSRFortessa flow cytometer (BD Biosciences), and data were analyzed using FlowJo version 10 software (FlowJo, Ashland, OR, USA). Antigen presentation assays FACS-purified cDC1s were stimulated with 20 µg/mL poly dI:dC and 20 ng/mL mGM-CSF, and pulsed with either 200 µg/mL OVA or 200 µg/mL of OVA-I-SIINFEKL peptide (Peptide 2.0, Chantilly, VA, USA, custom order) for 2 hours at 37 °C. Cells were then washed 3 times with PBS. OVA-specific CD8+ T cells were isolated by FACS from single-cell suspensions of spleens from OT-I mice and labeled with 5-(and-6)-carboxyfluorescein diacetate succinimidyl ester (CFSE; Tonbo Biosciences) according to the manufacturer’s instructions for 10 to 20 minutes at room temperature in the dark. Labeled OT-I CD8+ T cells were washed twice with endotoxin-free PBS and co-cultured with poly dI:dC–stimulated and antigen-loaded cDC1s for 48 or 60 hours at 37 °C in complete RPMI. Following co-culture, cells were washed 3 times with PBS, incubated with Fc block for 15 to 30 minutes at 4 °C, and stained with fluorescently conjugated antibodies for murine surface markers for 15 to 30 minutes at 4 °C. Antibodies used included APC-conjugated CD45R/B220 (RA3-6B2) and CD8α (53-6.7); BV605-conjugated CD103 (2E7); BV711-conjugated CD24 (M1/69); eFluor 450–conjugated CD11c (N418); FITC-conjugated CD172α/Sirpα (P84); PerCP-Cy5.5–conjugated CD24 (M1/69); BV785-conjugated CD3 (17A2); PE-594–conjugated CD152 (UC10-4B9); and PE-conjugated CD279 (29F.1A12). All antibodies were purchased from Tonbo Biosciences, BD Biosciences, Thermo Fisher Scientific, or BioLegend. Dead cells were excluded with Ghost Dye Violet 510. OT-I splenocytes and cDC1s were sorted using a FACSAria or FACSAria Fusion. Samples were acquired on a BD LSRFortessa flow cytometer and analyzed using FlowJo version 10 software. RNA isolation, library preparation, and sequencing Total RNA was isolated from FACS-purified cDC1s using the Qiagen RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Purified RNA was stored at −80 °C until shipment for sequencing. Individual biological samples were submitted to Novogene Corporation Inc. (Sacramento, CA, USA) for quality assessment, library preparation, and sequencing. RNA concentration and purity were evaluated by NanoDrop spectrophotometry, 1% agarose gel electrophoresis, and RNA integrity was assessed using an Agilent 2100 Bioanalyzer to determine RNA integrity numbers, a standardized quantitative measure of RNA degradation based on electrophoresis profiles.39 Libraries were generated using a poly(A)-enriched mRNA workflow and sequenced on an Illumina NovaSeq X-Plus platform to produce 150-bp paired-end reads. Bioinformatics processing and differential expression analysis For data processing, raw sequencing reads were processed using the nf-core/rnaseq pipeline v3.14,40 a community-curated workflow built upon the nf-core framework.41 The workflow was executed with Nextflow v23.10.142 under reproducible software environments provided through Bioconda43 and BioContainers.44 Reads were aligned to the GRCm39 reference genome using STAR, and transcript quantification was performed with RSEM to generate gene-level count estimates. Default nf-core parameters were used unless otherwise specified. Differentially expressed genes (DEGs) were identified using the DESeq2 package, which models count data with a negative binomial framework. DESeq2 estimates gene-wise dispersions and log2 fold changes using empirical priors to improve stability, particularly for low-count genes. Statistical significance was assessed using Wald tests and resulting P values were adjusted for multiple testing using the Benjamini–Hochberg procedure to control the false discovery rate. Projection of bulk RNA sequencing data to single-cell RNA sequencing Processed single-cell RNA sequencing (scRNA-seq) datasets of tumor-infiltrating human and mouse DCs were used as reference datasets for projection analyses.45 Projection from bulk RNA sequencing (bulk RNA-seq) results from in vitro–derived cDC1s onto this reference was performed using the projectLSI pipeline (version 0.1.0).46 Bulk RNA-seq data were first converted into pseudo-single-cell data using the CreateSeuratObject function and normalized with the NormalizeData function in Seurat (version 5.1.0). For projection onto the human scRNA-seq reference, mouse gene symbols were converted to their human orthologs using the convert_mouse_to_human_symbols function in nichenetr (version 2.1.5). Both the scRNA-seq reference and the pseudo-single-cell dataset were processed by computing latent semantic indexing (LSI) embeddings using the calcLSI function, followed by generation of a low-dimensional representation using the CreateDimReducObject function within the projectLSI workflow. Uniform manifold approximation and projection (UMAP) visualization was performed using the umap function (n_neighbors = 20) implemented in uwot (version 0.2.2). Then, the reference scRNA-seq and pseudo-single-cell datasets were merged, and the joint UMAP embedding was used for visualization. DEGs in the reference scRNA-seq dataset were identified using the FindAllMarkers function in Seurat and visualized with DoHeatmap, enabling annotation of DC subsets including cDC1, cDC2, and CCR7+ DCs. Statistical analyses All statistical analyses were performed using GraphPad Prism version 10 (GraphPad Software, San Diego, CA, USA). The statistical tests applied to each dataset, including multiple comparison and survival analyses, are specified in the corresponding figure legends. Results were considered significant when P < 0.05. Results Transcriptional similarities between in vitro–derived cDC1s and intratumoral DCs To test the importance of MHC-I and MHC-II expression in vaccine-delivered cDC1s, we used a previously established culture system to generate cDC1s from the bone marrow of WT C57BL/6J mice, H2-Kb−/− H2-Db−/− (MHC-IKO) mice, and B6.129S2-H2dlAb1-Ea/J (MHC-IIKO) mice (Fig. 1A).31–35,37 Flow cytometric analysis confirmed the expected loss of MHC-I or MHC-II surface expression on MHC-IKO or MHC-IIKO cDC1s, respectively (Fig. 1B). Comparable numbers of cDC1s were derived from mice of all genotypes, with high purity, and stimulation with poly dI:dC enhanced surface expression of CD80 and CD86 similarly between WT and MHC-IKO or WT and MHC-IIKO cDC1s (Fig. S1A–C). In addition, we did not detect differences in the expression of MHC-I or MHC-II on WT and MHC-IIKO or WT and MHC-IKO cDC1s, respectively (Fig. S1C). These data indicate that loss of MHC-I or MHC-II does not impair cDC1 differentiation, expansion, or poly dI:dC–induced maturation. Characterization and functional evaluation of WT, MHC-IKO, and MHC-IIKO cDC1s. (A) Schematic of the workflow used to generate cDC1s from WT, MHC-IKO, or MHC-IIKO mice in vitro. Bone marrow cells were cultured in medium containing Flt3L and GM-CSF, followed by FACS purification, poly dI:dC stimulation, and incubation with OVA. (B) Flow cytometry histogram plots showing cell surface expression of MHC-I (H2-Kb) and MHC-II (I-A/I-E) on cDC1s after 16 hours under the indicated conditions. n = 6 (WT), n = 2 (MHC-IKO), n = 4 (MHC-IIKO); representative data from at least 2 independent experiments. (C) Bulk RNA-seq analysis of cDC1s in steady-state conditions (no stim) or following stimulation with poly dI:dC with or without αCD40, as indicated. The heatmaps show log2(TPM + 1) gene expression, grouped according to cDC1 functions. (D) Mapping of bulk RNA-seq profiles from in vitro–derived cDC1s onto mouse or human tumor-infiltrating DC scRNA-seq reference datasets using ProjectLSI. Reference scRNA-seq cells are displayed as dots, and bulk RNA-seq samples are overlaid as shapes, corresponding to individual genotypes, as indicated. To evaluate whether MHC-I or MHC-II deficiency intrinsically alters the transcriptional identity or activation of cDC1s, we performed RNA-seq using in vitro–derived WT, MHC-IKO, and MHC-IIKO cDC1s cultured under steady-state conditions or following stimulation with poly dI:dC, either alone or in combination with agonistic CD40 antibody (αCD40). As expected, MHC-IKO cDC1s displayed defects in H2-D1 and H2-K1 transcript expression, while MHC-IIKO cDC1s showed loss of major MHC-II genes (Fig. 1C). In contrast, WT, MHC-IKO, and MHC-IIKO cDC1s expressed comparable levels of genes involved in antigen processing and presentation, including proteasome activators (Psme1 and Psme2), TAP transporters (Tap1 and Tap2), and tapasin (Tapbp) (Fig. 1C), suggesting that protein degradation and peptide-loading pathways relevant to antigen presentation remained intact. Stimulation with poly dI:dC ± αCD40 induced similar transcriptional responses among WT, MHC-IKO, and MHC-IIKO cDC1s, with no major differences detected in genes associated with costimulation, survival, or migration (Fig. 1C). Examination of the top 20 most variable genes across all conditions further indicated that transcriptional patterns were broadly similar between WT, MHC-IKO, and MHC-IIKO cDC1s (Fig. S1D). A notable exception was Btnl2, which encodes butyrophilin-like 2 (BTNL2); Btnl2 was upregulated in MHC-IIKO cDC1s relative to WT or MHC-IKO cDC1s. Given prior reports implicating BTNL2 in the dampening of T-cell activation and proliferation,47,48 we tested the ability of MHC-IIKO cDC1s to induce CD8+ T-cell proliferation. In co-culture assays with OT-I CD8+ T cells, MHC-IIKO cDC1s induced CD8+ T-cell proliferation to a similar extent as WT cDC1s, suggesting that Btnl2 upregulation does not impair their ability to promote CD8+ T-cell division (Fig. S1E, F). Consistent with this observation, the proportion of PD-1+ OT-I CD8+ T cells and the magnitude of PD-1 expression were comparable between CD8+ T cells from co-cultures with WT or MHC-IIKO cDC1s (Fig. S1G). Together, these data indicate that WT and MHC-IIKO cDC1s are similarly capable of inducing CD8+ T-cell activation and proliferation under in vitro culture conditions. To assess whether in vitro–derived cDC1s resemble naturally occurring DC populations, we compared their transcriptional profiles to human and mouse scRNA-seq reference datasets of tumor-infiltrating DCs derived from multiple cancer types and independent studies (Fig. 1D).45 Bulk RNA-seq profiles from in vitro–derived cDC1s were embedded together with scRNA-seq data into shared UMAP space, revealing close alignment with bona fide intratumoral DC populations. Canonical cDC1, cDC2, and CCR7+ DC subsets within the UMAP were annotated based on their DEG signatures (Fig. S1H). In this analysis, in vitro–derived cDC1s in homeostatic conditions were transcriptionally similar to tumor-infiltrating cDC1s, whereas cDC1s stimulated with poly dI:dC, with or without αCD40, aligned with intratumoral CCR7+ DCs. The CCR7+ DCs are also referred to as mature DCs enriched in immunoregulatory molecules (mregDCs) or lysosomal-associated membrane protein 3 (LAMP3) DCs.49–51 Collectively, these data demonstrate that in vitro–derived murine cDC1s closely recapitulate transcriptional states observed in tumor-infiltrating DC populations in both mice and humans and that loss of MHC-I or MHC-II does not substantially alter cDC1 identity, maturation, or transcriptional activation. MHC-I– and MHC-II–dependent antigen presentation contribute to the therapeutic efficacy of cDC1 vaccines We next tested whether MHC-I or MHC-II expression on vaccine-delivered cDC1s is required for tumor control following cDC1 vaccination. WT mice were implanted with B16-OVA melanoma cells; OVA contains both MHC-I– and MHC-II–restricted antigens. Using our established vaccination approach,26 mice were then treated intratumorally on days 4 and 7 with poly dI:dC–activated, OVA-loaded WT, MHC-IKO, or MHC-IIKO cDC1s or with PBS (Fig. 2A). Consistent with our prior studies,26 vaccination with WT cDC1s significantly restrained tumor growth and prolonged mouse survival compared to PBS-treated controls (Fig. 2B–D). By contrast, vaccination with MHC-IKO or MHC-IIKO cDC1s modestly controlled tumor growth, relative to the PBS treatment group, yet this was suboptimal relative to WT cDC1s. These data indicate that effective tumor control by cDC1 vaccination requires MHC-I and MHC-II antigen presentation by vaccine-delivered cDC1s. Loss of MHC-I or MHC-II antigen presentation impairs cDC1 vaccine efficacy. (A) Schematic of the experimental design. WT C57BL/6J mice were implanted with B16-OVA tumors on day 0 and treated intratumorally on days 4 and 7 with either PBS, WT, MHC-IKO, or MHC-IIKO cDC1 vaccines. (B) Representative mean tumor area (mm2) over time. (C) Individual tumor growth curves corresponding to mice in (B). (B, C) n = 5 mice per group; representative data from one of 3 independent experiments. (D) Cumulative mouse survival representing time to humane endpoint, defined as tumors reaching 15 mm in any direction. n = 35 (PBS), n = 35 (WT), n = 25 (MHC-IKO), and n = 35 (MHC-IIKO) mice; cumulative data from at least 5 independent experiments. (B) Data shown are the mean ± SEM and analyzed by 2-way ANOVA. (D) Survival was analyzed by log-rank (Mantel–Cox) test. ***P < 0.001; ****P < 0.0001. Expression of MHC-I and MHC-II by the same cDC1 is critical for cDC1 vaccine efficacy Given that loss of either MHC-I or MHC-II on vaccine-delivered cDC1s compromised tumor control, we next asked whether coexpression of both MHC molecules on the same vaccine-delivered cDC1 is important for efficacy. To test this, B16-OVA tumor-bearing mice were treated intratumorally with vaccines consisting of WT cDC1s, MHC-IKO cDC1s, MHC-IIKO cDC1s, or a 1:1 mixture of MHC-IKO cDC1s and MHC-IIKO cDC1s (MixKO). Treatment with the MixKO cDC1 vaccine was unable to enhance tumor control relative to vaccination with MHC-IKO or MHC-IIKO cDC1s alone and, similar to vaccines lacking MHC-I or MHC-II only, demonstrated diminished tumor control compared to WT cDC1 vaccination (Fig. 3A–C). Together, these results support the concept that coexpression of both MHC-I and MHC-II by the same cDC1 is required for robust tumor control following cDC1 vaccination.17,52–54 Coexpression of MHC-I and MHC-II on the same cDC1 is required for full cDC1 vaccine efficacy. WT C57BL/6J mice were implanted with B16-OVA tumors on day 0 and treated intratumorally on days 4 and 7 with either PBS, WT, MHC-IKO, MHC-IIKO, or a 1:1 mixture of MHC-IKO and MHC-IIKO (MixKO) cDC1 vaccines. For MixKO vaccines, mice received 6 × 106 cDC1s per injection, composed of equal numbers of MHC-IKO cDC1s and MHC-IIKO cDC1s (3 × 106 each), while individual vaccines (WT, MHC-IKO, or MHC-IIKO) consisted of 3 × 106 cells each. (A) Representative mean tumor area (mm2) over time. (B) Individual tumor growth curves corresponding to mice in (A). (A, B) n = 5 mice per group; representative data from one of 2 independent experiments. (C) Cumulative mouse survival representing time to humane endpoint, defined as tumors reaching 15 mm in any direction. n = 10 mice per group; cumulative data from 2 independent experiments. (A) Data shown are the mean ± SEM and analyzed by 2-way ANOVA. (C) Survival was analyzed by log-rank (Mantel–Cox) test. ***P < 0.001; ****P < 0.0001. CD40 ligation by agonistic antibody fails to enhance cDC1 vaccine-mediated tumor control Since our data indicate that MHC-II is required by vaccine-delivered cDC1s for effective tumor control, we next asked whether a CD40-mediated licensing signal could enhance vaccine efficacy. We found that both WT and MHC-IIKO cDC1s expressed detectable levels of cell surface CD40 at steady-state and following 4 hours of poly dI:dC stimulation in culture (Fig. 4A), suggesting that both populations are capable of signaling via CD40 in the conditions that we used for activation and vaccination. Notably, CD40 expression remained detectable at 24 hours poststimulation with poly dI:dC (Fig. S2A), suggesting that cDC1s maintain CD40 responsiveness beyond our initial 4-hour stimulation timepoint. Furthermore, stimulation with poly dI:dC + αCD40 drove increased expression of the T-cell costimulatory molecule CD70, with this effect being most evident at 16 and 24 hours poststimulation (Fig. S2B), consistent with prior reports linking CD40 signaling to CD70 induction.52,55 We therefore tested whether exogenous activation of CD40 could bypass the need for MHC-II expression on vaccine cDC1s in vivo and potentially enhance WT cDC1 vaccine efficacy. B16-OVA tumor-bearing mice were vaccinated intratumorally with WT or MHC-IIKO cDC1s that had been treated with poly dI:dC alone or with poly dI:dC + αCD40 prior to administration (Fig. 4B). While vaccination with WT cDC1s significantly delayed tumor progression compared to MHC-IIKO cDC1 vaccination or PBS-treated controls, as expected, the addition of αCD40 stimulation did not enhance tumor control beyond that achieved with poly dI:dC–stimulated WT cDC1s (Fig. 4C–E). Moreover, αCD40 stimulation failed to improve the efficacy of MHC-IIKO cDC1 vaccines. Together, these results indicate that CD40 ligation provided by a short-term exposure to αCD40 in culture is insufficient to enhance WT cDC1 vaccine efficacy or to compensate for the loss of MHC-II on vaccine-delivered cDC1s. In vitro CD40 ligation by agonistic antibody fails to enhance cDC1 vaccine-mediated tumor control. (A) Representative flow cytometry histograms (left) showing CD40 surface expression on WT and MHC-IIKO cDC1s after 4-hour stimulation in vitro, with quantification of CD40+ frequency and CD40 Mean Fluorescence Intensity (right). n = 4 per group for all conditions; cumulative data from 2 independent experiments. (B) Schematic of the in vivo experimental design. cDC1s were stimulated in vitro with either poly dI:dC or poly dI:dC and αCD40 and pulsed with OVA for 4 hours, then prepared for vaccination as described in the Materials and Methods. WT mice were implanted with B16-OVA tumors on day 0 and treated intratumorally on days 4 and 7 with either PBS or cDC1 vaccines. (C) Representative mean tumor area (mm2) over time. (D) Individual tumor growth curves corresponding to mice in (C). (C, D) n = 5 mice per group; representative of one of 2 independent experiments. (E) Cumulative mouse survival representing time to humane endpoint, defined as tumors reaching 15 mm in any direction. n = 10 mice per group; cumulative data from 2 independent experiments. (A, C) Data shown are the mean ± SEM and were analyzed by 2-way ANOVA, followed by Tukey multiple-comparisons test applied to (A). (E) Survival was analyzed by log-rank (Mantel–Cox) test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Endogenous cDC1s are required for cDC1 vaccine efficacy Previous studies showed that cDC1 vaccination could effectively mediate tumor rejection in the absence of endogenous cDC1s, using a tumor model that shows spontaneous regression in immunocompetent hosts.56 However, whether host cDC1s contribute to cDC1 vaccine-mediated tumor control with less immunogenic tumors remains unclear. To address this, we evaluated cDC1 vaccine efficacy against B16-OVA melanoma in Irf8 + 32−/− mice (Fig. 5A), which lack endogenous cDC1s but retain other DC lineages.34 These studies revealed that vaccination with WT, MHC-IKO, or MHC-IIKO cDC1s resulted in reduced tumor control in Irf8 + 32−/− mice compared to identically treated WT hosts (Fig. 5B, Table S1), indicating that the presence of host cDC1s is required to achieve optimal therapeutic benefit. Notably, within Irf8 + 32−/− mice, all 3 cDC1 vaccines still conferred a modest survival benefit relative to PBS controls, indicating that vaccine-delivered cDC1s retain partial antitumor activity even in the absence of endogenous cDC1s. Together, these data indicate that while intratumorally delivered cDC1s can partially restrain tumor progression on their own, full vaccine efficacy in poorly immunogenic tumors depends on the presence of host cDC1s, which likely provide additional immune support required for robust vaccine-elicited antitumor responses. Lack of endogenous cDC1s reduces therapeutic cDC1 vaccine efficacy. (A) Schematic diagram of the experimental design. WT C57BL/6J or cDC1-deficient (Irf8 + 32−/−) mice were implanted with B16-OVA tumors on day 0 and treated intratumorally on days 4 and 7 with either PBS, WT, MHC-IKO, or MHC-IIKO cDC1 vaccines. (B) Cumulative mouse survival representing time to humane endpoint, defined as tumors reaching 15 mm in any direction. Survival rates of Irf8 + 32−/− mice treated with the indicated cDC1 vaccines are shown as dashed lines, with identically treated WT mice shown as solid lines. Full pairwise survival comparisons are provided in Table S1. n = 10 mice per group per genotype (WT and Irf8 + 32−/−); cumulative data from 2 independent experiments. Survival was analyzed by log-rank (Mantel–Cox) test. Discussion In this study, we define key antigen presentation requirements that govern the efficacy of cDC1-based cancer vaccination. By selectively disrupting MHC-I or MHC-II expression on vaccine-delivered cDC1s, we demonstrate that both antigen presentation pathways contribute to therapeutic efficacy. We further show that coexpression of MHC-I and MHC-II on the same vaccine-delivered cDC1 is required for robust tumor control. These findings suggest that effective vaccination depends on the ability of individual vaccine-delivered cDC1s to engage both CD4+ and CD8+ T cells. Notably, although host cDC1s contribute to the efficacy of cDC1 vaccination in controlling B16-OVA melanoma tumors, cDC1 vaccines still elicit modest antitumor activity in their absence. Collectively, these findings extend prior work establishing cDC1s as central coordinators of antitumor immunity and indicate that their function as a cellular therapy requires the simultaneous engagement of CD8+ and CD4+ T cells via MHC-I and MHC-II antigen presentation. Our data demonstrate that cDC1s generated in culture respond to poly dI:dC stimulation by acquiring a mature phenotype characterized by upregulation of canonical costimulatory molecules and transcriptional alterations such as induction of cytokine and chemokine gene expression. Analyses of cDC1 transcriptional states also indicate that in vitro–derived MHC-IKO and MHC-IIKO cDC1s retain a conserved cDC1 identity, with expression of genes involved in antigen processing, costimulation, and migration remaining comparable to WT cDC1s. Thus, loss of MHC-I or MHC-II does not appear to grossly alter cDC1 differentiation or poly dI:dC–mediated maturation. Furthermore, when mapped onto integrated mouse and human tumor scRNA-seq datasets, in vitro–derived cDC1s align closely with bona fide intratumoral cDC1s, whereas poly dI:dC–activated cDC1s align with CCR7+ DCs, a DC state that is associated with maturation.49 The expression of CCR7 on DCs is important for their migration from tumors to TdLNs, enabling the transport of tumor antigen and initiation and regulation of adaptive immune responses. Although intratumoral CCR7+ DCs display both maturation and immunoregulatory features, this population has been shown to contribute to activating antitumor immune responses and to associate with improved responses to immunotherapy.57 Our data, therefore, support the notion that cDC1s generated in culture recapitulate transcriptional states that naturally occur within tumors, reinforcing the relevance of cDC1-based vaccines for modeling and studying clinically important DC states. Moreover, given that culture-derived cDC1s transcriptionally align with intratumoral cDC1s, our system may offer a robust approach for further dissecting physiologically relevant functions of DC states in vivo. The induction of cytokine and chemokine gene expression in in vitro–derived cDC1s upon poly dI:dC stimulation raised the possibility that antigen presentation–independent functions might be sufficient to promote therapeutic activity after intratumoral cDC1 vaccination by recruiting and influencing endogenous immune cells within the TME. However, both MHC-IKO and MHC-IIKO cDC1 vaccines showed impaired tumor control, indicating that antigen presentation–independent mechanisms are insufficient to elicit optimal antitumor immunity. Together, our data indicate that both MHC-I and MHC-II antigen presentation are key determinants of cDC1 vaccine efficacy. The reduced efficacy of MHC-IKO cDC1s was consistent with the established role of cDC1-mediated MHC-I cross-presentation in priming tumor antigen–specific CD8+ T cells,16,58 as well as our previous finding that cDC1 vaccination enhances CD8+ T-cell responses.26 MHC-IKO cDC1 vaccines nonetheless retained partial activity, suggesting that MHC-I–independent mechanisms also contribute to vaccine efficacy. One possible contributor is preserved MHC-II–dependent activation of CD4+ T cells, which can support antitumor immunity through cytokine production, cDC1 licensing, and promotion of CD8+ T-cell effector function. In addition, MHC-I–deficient vaccine cDC1s may retain other effector functions, including costimulation, cytokine production, chemokine-mediated immune recruitment, activation of other immune populations within the TME, or cooperation with host cDC1s. However, these mechanisms are insufficient to support optimal tumor control in the absence of MHC-I–dependent antigen presentation by vaccine-delivered cDC1s. We further found that effective tumor control requires MHC-II expression by the cDC1 vaccine, supporting an important role for CD4+ T-cell engagement in antitumor immunity. This finding aligns with prior evidence that indicates cDC1s undergo licensing by CD4+ T cells via a CD40–CD40L interaction, which enhances cDC1-mediated CD8+ T-cell activation.17,52,59 Licensing of cDC1s has been shown to occur within TdLNs, where migratory cDC1s acquire enhanced functional capacity following CD40 engagement.17,52,59 Additionally, CD40 signaling promotes cDC1 survival and induces expression of critical costimulatory ligands such as CD70, which sustain effective CD8+ T-cell priming.52,59 Activated CD4+ T cells have also been shown to promote increases in viability, antigen presentation, and T-cell activation gene programs in peripheral blood cDC1s following co-culture in vitro.60 Furthermore, intratumoral immune triads composed of DCs, CD4+ T cells, and CD8+ T cells have been found to be essential for orchestrating effective antitumor responses.53,54 Together, our data and results from others support the idea that cDC1 vaccination elicits both MHC-I–dependent CD8+ T-cell priming and MHC-II–dependent CD4+ T-cell engagement to drive effective antitumor immunity. Interestingly, we found that vaccines composed of a 1:1 mixture of MHC-IKO and MHC-IIKO cDC1s failed to restore full tumor control, suggesting that CD4+ and CD8+ T-cell responses must be coordinated in cis by the same vaccine-delivered cDC1. In this context, CD4+ T-cell recognition through MHC-II may promote CD40-dependent licensing of the cDC1, leading to increased costimulatory molecule expression and cytokine production that support effective cDC1-mediated priming of CD8+ T cells. CD4+ T-cell engagement by vaccine-delivered cDC1s may also promote CD8+ T-cell activation through mechanisms that remain undefined. One possibility is that presentation of both MHC-I and MHC-II antigens by the same vaccine-delivered cDC1 facilitates localized coordination of CD4+ and CD8+ T-cell interactions, allowing helper signals to be delivered in a sustained and prolonged manner. These interpretations are consistent with established mechanisms in which CD4+ T-cell activation licenses endogenous cDC1s and enhances their ability to support cytotoxic CD8+ T-cell responses, as well as roles for intratumoral immune triads in coordinating effective antitumor immunity.17,52,59 Given the established role of CD40-dependent licensing in mediating CD4+ T-cell help to cDC1s, we examined whether CD40 stimulation could enhance cDC1 vaccine function and compensate for the absence of MHC-II–dependent CD4+ T-cell engagement during vaccination. We found that poly dI:dC stimulation of cDC1s did not fully reproduce activation features associated with CD40-dependent licensing. For instance, short-term poly dI:dC stimulation (4 hours), conditions used to activate our cDC1 vaccine, did not lead to robust expression of CD70, while the addition of αCD40 treatment led to CD70 upregulation but only after a longer time point of treatment (24 hours). We also found that short-term αCD40 treatment of poly dI:dC–stimulated cDC1s was unable to improve the antitumor efficacy of either WT or MHC-IIKO cDC1s. These results suggest that transient CD40 stimulation of cDC1s is insufficient to substitute for the full spectrum of MHC-II-dependent CD4+ T-cell help. This may reflect constraints in the timing, duration, or physiological context of CD40 signaling provided in culture rather than an absolute inability of CD40 engagement to support the licensing of the cDC1 vaccine. In vivo, sustained interactions with CD4+ T cells may be required to fully induce the transcriptional, metabolic, and survival programs necessary for optimal cDC1 vaccine function.52 These interactions may also include locally delivered cytokines or other factors that directly regulate either cDC1s or CD8+ T cells, which are not fully recapitulated by short-term agonistic CD40 stimulation in vitro. Although transcriptional profiles were largely similar among WT, MHC-IKO, and MHC-IIKO cDC1s, Btnl2 was found to be upregulated in MHC-IIKO cDC1s. BTNL2 has been implicated in modulating T-cell costimulation.47,48 The increased Btnl2 transcript expression we observed in MHC-IIKO cDC1s may be linked to the genomic architecture surrounding the MHC-II locus, as Btnl2 is positioned adjacent to the targeted H2-Ab1 region in the mouse genome.32,36 For instance, the targeted deletion may have altered regulatory elements that normally restrain Btnl2 transcription. Despite the increased expression of Btnl2, we found that MHC-IIKO cDC1s are fully capable of inducing OT-I CD8+ T-cell proliferation and PD-1 upregulation in co-culture assays, eliciting responses comparable in extent and quality to those induced by WT cDC1s. These findings suggest that the increase in Btnl2 does not translate into a functional defect in MHC-IIKO cDC1-mediated T-cell activation, arguing against a major BTNL2-mediated defect as the primary explanation for the reduced efficacy of MHC-IIKO cDC1 vaccines in vivo. Nevertheless, our finding highlights a potentially important consideration when using this mouse strain. Finally, we found that cDC1 vaccine efficacy in B16-OVA melanoma depended on the presence of endogenous cDC1s, as judged by impaired tumor control in Irf8 + 32−/− mice.34 These data indicate that host cDC1s provide important immune functions that aid in mounting effective antitumor responses following cDC1 vaccination. This requirement differs from other studies using a spontaneously regressing tumor model, which have reported that cDC1 vaccines could mediate tumor rejection independently of host cDC1s. One possible explanation for these differences is that spontaneously regressing tumors are inherently highly immunogenic, and thus administration of exogenous cDC1s is sufficient to induce potent effector CD4+ and CD8+ T-cell responses even without endogenous cDC1s. In contrast, B16-OVA melanoma tumors do not spontaneously regress and are considered poorly immunogenic; thus, sustained cDC1 function may be required to maintain antitumor T-cell activity in this setting. Endogenous cDC1s may reinforce vaccine-initiated immune responses by supporting continued T-cell priming, expansion, and intratumoral function. Moreover, early tumor cell killing induced by vaccine-driven T-cell responses may release sterile inflammatory factors that activate host cDC1s, further amplifying immune responses within tumors and TdLNs.61 Host cDC1s may also acquire tumor-derived antigens following vaccination through uptake of tumor material released during immune-mediated tumor cell death or through antigen transfer between vaccine-delivered and host cDC1s. However, the requirement for coexpression of MHC-I and MHC-II on the same vaccine-delivered cDC1 argues against antigen transfer to host cDC1s as the central mechanism underlying vaccine efficacy. This is based on the assumption that MHC-IKO and MHC-IIKO vaccine cDC1s would be expected to show complementary antigen transfer functions in the mixed vaccine (MixKO) experiment, and thereby restore full vaccine efficacy, yet our results indicate the MixKO vaccine was unable to improve tumor control relative to the MHC-IKO or MHC-IIKO vaccines. Thus, our data suggest that direct antigen presentation via MHC-I and MHC-II by vaccine cDC1s, on the same cell, initiates antitumor immune responses, whereas endogenous cDC1s provide suboptimal initiation of antitumor immunity in our tumor model yet remain important for sustaining and potentially amplifying antitumor immunity following vaccination through continued antigen acquisition and T-cell stimulation within the TME or TdLNs. Future studies using cellular imaging or cell-tracing approaches could help further define precise cellular interactions and mechanisms by which vaccine-derived and endogenous cDC1s mediate antigen presentation and immune effector programming. Collectively, our findings identify several key parameters relevant to effective cDC1-based cancer vaccination. Antigen presentation by MHC-I and MHC-II must occur in cis, by the same cDC1, and CD4+ T-cell help cannot be replaced by short-term CD40 stimulation under the conditions used for our vaccination approach. In addition, endogenous cDC1s contribute to vaccine efficacy in the poorly immunogenic B16-OVA tumor setting, yet the precise mechanisms involved remain unknown. Together, insights from our study emphasize the importance of coordinated antigen presentation, CD4+ T-cell engagement, and endogenous cDC1 involvement in supporting durable antitumor immunity, and provide important considerations for the development of cDC1-based immunotherapies. Supplementary Material