B16-F1 (B16) murine melanoma cells were purchased from the ATCC. The B16.Kb^loss cell line was a kind gift from Esteban Celis, University of Southern Florida, USA. B16 cells were passaged routinely at 70-80% confluency and cultured in RPMI media (Life Technologies) supplemented with 10% FCS (Sigma-Aldrich), 2 mM L-glutamine, 50 µM 2-mercaptoethanol, 100 µg/mL streptomycin and 100 U/mL penicillin (all Life Technologies) (R10 media) at 37 °C, 5% CO₂. 293T, COS-7 and L929 cells were similarly passaged, in DMEM media (Life Technologies) supplemented with 10% FCS, 2 mM L-glutamine, 100 µg/mL streptomycin and 100 U/mL penicillin (D10 media).

B16-F1 and B16.Kb^loss cells were transduced, as described previously (30), with retroviral vectors containing a full-length membrane-bound form of HSV glycoprotein B (gB) and enhanced GFP (GFP) or CFP, respectively, to generate B16-F1-gB-GFP (B16.gB) and B16.Kb^loss-gB-CFP (B16.Kb^loss.gB) cell lines. Briefly, retroviruses were generated by transfecting the 293T cell lines with pMIG-gB or pMIC-gB, pMD.old.gag.pol, and pCAG-VSVG. B16 cells were next transduced with filtered retrovirus supernatant in the presence of 8 µg/mL polybrene (Sigma-Aldrich). For the generation of B16.Kb^loss.gB_IFN cell lines for vaccination, murine IFNα1, IFNα2, IFNα4, IFNα5, IFNα6, IFNα9, and IFNβ was amplified from the pkCMVint mammalian expression vector (31) and subcloned into pMIG, as described previously (29). B16.Kb^loss.gB cells were then retrovirally transduced with the pMIG-IFN constructs, and GFP⁺ cells sorted using a BD FACSAriaIII cell sorter (BD Biosciences) to select stable B16Kb^loss.gB_IFN cell lines. GFP expression of sorted cell lines was confirmed using a BD LSRFortessa X-20 (BD Biosciences).

Bioactive IFNα/β secretion was confirmed and quantitated using an in vitro IFN bioassay (32). Briefly, L929 cells were exposed to serial dilutions of acid-treated supernatants from the engineered B16.Kb^loss.gB_IFN cell lines or NIH IFNα/β standard (1,000 IU/mL). Cell supernatants were collected from 5x10⁵ irradiated cells after 24 h in culture. After 24 h, encephalomyocarditis virus (EMCV) was added to each well. Following a further 24 h, end-point titers were defined as the dilution producing a 50% reduction in cytopathic effect (CPE) of the L929 cells. Bioactive titers were calculated by comparing the CPE of the B16 or COS-7 cell supernatants to the IFNα/β standard.

C57BL/6 (B6) female mice that express the CD45.2 allele were purchased from the Animal Resources Centre, Murdoch, Western Australia. gB-specific T-cell receptor (TCR) transgenic (gBT.I) mice that express the CD45.1 allele (33), type I IFN knockout mice (IFNAR1^o/o) (34), XCR1-DTRvenus mice (XCR1-DTR) (35), and I-A/E knockout mice (IA/E^o/o) (36) were bred at the Telethon Kids Institute. Mice were typically used at 8-12 weeks. Animals were housed under pathogen-free conditions and all studies were approved by the Telethon Kids Institute’s Animal Ethics Committee (AEC) (AEC#290, AEC#295, AEC#325, AEC#348).

For transfer of precursor gBT.I cells, single cell suspensions were prepared from pooled lymph nodes from naïve gBT.I female mice. Purity of gBT.I CD8⁺ T cells was determined by flow cytometry, and 5 x 10⁴ CD45.1⁺ Vα2⁺ CD8⁺ gBT.I T cells were washed and resuspended in 200 µL RPMI for i.v. injection into recipient mice at least one day prior to whole-cell vaccination.

Mice were vaccinated i.p. with 2.5 x 10⁶ irradiated (200 Gy) B16Kb^loss.gB or B16Kb^loss.gB_IFN cells. Cells were washed in PBS prior to irradiation and resuspended in 300 µL PBS for injection. For recombinant IFN experiments, mice received 2.5 x 10⁶ irradiated (200 Gy) B16Kb^loss.gB cells at the same time as injection with 10⁵ IU IFNα1 or IFNβ, produced in-house as previously described (31, 37).

For XCR1 depletion experiments, XCR1-DTR mice were administered with either PBS control or 25 ng/g weight diphtheria toxin (Sigma) one day prior to vaccination. For NK depletion experiments, mice received control PBS or 200 µg anti-NK1.1 (BioXCell) one day before and after vaccination. For CD40L blocking experiments, mice were administered with either control IgG isotype (BioXCell) or 200 µg anti-CD40L blocking antibody (BioXCell) on the same day as vaccination.

Spleen, lymph nodes and/or tumors were harvested and passed through a 70 µm metal mesh and red blood cell lysed. Resulting single cell suspensions were stained with monoclonal antibodies specific for mouse CD8α (53-6.7), CD45.1 (A20), Vα2 (B20.1), IFNγ (XMG1.2), PD-1 (29F-1412), MHCI (AF6-88.5), CD45 (30-F11), and/or NK1.1 (PK1136) (all from BD Biosciences). For ex vivo cytokine production assays, splenocytes were first restimulated with 1 µM gB_498-505 peptide for 1 h at 37 °C prior to addition of 0.22 mg Brefeldin A (BD GolgiPlug™, BD Biosciences) for a further 4 h. Following surface stain, cells were fixed with 4% paraformaldehyde prior to permeabilization with Permeabilization Buffer (eBioscience) and staining with IFNγ. Cells were stained with Fixable Viability Stain 575V at 1:20,000 (BD Biosciences) prior to surface staining or propidium iodide (PI; Sigma) immediately prior to acquisition to exclude dead cells. Cells were analyzed using the BD LSRFortessa and FlowJo software (BD Biosciences/TreeStar).

Mice were injected subcutaneously with 5 x 10⁵ B16 wildtype or B16.gB cells in 50 µL of RPMI media. Mice were then vaccinated, as described above, three days (for survival experiments) or four days (for T cell infiltration experiments) post-tumor inoculation. Mice receiving CPB were injected i.p. with 200 µg anti-PD-L1 (BioXCell) on days six, nine, and twelve post-tumor inoculation. Tumor size was monitored using calipers and tumor volume was calculated using the following formula: (length (mm) x width (mm)²)/2. Mice with tumors >1000 mm³ were euthanized. Tumor-free mice are defined as mice with no palpable masses.

All statistical analyses were performed using GraphPad (GraphPad Software Inc. v7.0a). Comparison of MFI, IFN titers, MHCI allele expression, and T cell expansion was assessed using one-way or two-way ANOVA. Difference in tumor survival was compared using the Log-Rank Mantel-Cox test. Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.005, and ****p < 0.001, unless otherwise stated.

To compare the adjuvant potential of different type I IFN subtypes, we established a whole-cell vaccination model to systematically interrogate the capacity of individual IFN subtypes to enhance CD8⁺ T cell priming. We engineered B16 melanoma cells deficient in the MHC class I alloantigen H-2K^b haplotype (B16.Kb^loss) (38) to stably express a model tumor antigen (Herpes Simplex Virus (HSV) derived glycoprotein B (gB)). The resulting B16.Kb^loss.gB cell line is defective in direct presentation of K^b-restricted gB epitopes by tumor cells to the CD8⁺ T cell compartment, providing a model system to evaluate cross-priming by DCs. A panel of seven distinct type I IFN subtypes were stably expressed to produce a suite of cell lines for vaccination (B16.Kb^loss.gB_IFNα1, B16.Kb^loss.gB_IFNα2, B16.Kb^loss.gB_IFNα4, B16.Kb^loss.gB_IFNα5, B16.Kb^loss.gB_IFNα6, B16.Kb^loss.gB_IFNα9 and B16.Kb^loss.gB_IFNβ; collectively referred to as B16.Kb^loss.gB_IFN). Cross-priming in our model system can be tracked by measuring the expansion of adoptively-transferred T cell receptor (TCR) transgenic T cells specific for the HSV immunodominant gB_498-505 peptide (39) (gBT.I cells) and/or IFNγ production by CD8⁺ T cells following restimulation with gB_498-505 (33). We used a cytopathic protective effects (CPE) assay (32) to validate transgenic expression of type I IFN subtypes. Quantification of bioactive type I IFN titers in the cell supernatants confirmed IFN production was robust and not significantly different among the different subtypes ( Figure 1A ).

The capacity of type I IFN subtypes to expand tumor-specific CD8⁺ T cells was investigated in cohorts of C57BL/6 mice inoculated with a single irradiated B16.Kb^loss.gB_IFN cell line as a whole-cell vaccine. To track gB-specific responses, 5 x 10⁴ naïve gBT.I CD8⁺ T cells expressing the congenic marker CD45.1 were adoptively transferred into the mice prior to vaccination. An optimal saturating dose of 2.5 x 10⁶ irradiated cells was selected in a prior experiment by titration of B16.Kb^loss.gB_IFNα4 cells and comparison of gBT.I CD8⁺ T cell expansion ( Supplementary Figure 1A ). Mice vaccinated with irradiated B16.Kb^loss.gB_IFN cells producing 4 out of the 7 IFN subtypes tested (IFNα1, IFNα4, IFNα6 or IFNβ) induced significantly greater expansion of gBT.I CD8⁺ T cells compared to vaccination with control irradiated B16.Kb^loss.gB cells expressing no IFN ( Figure 1B ). Notably, mice receiving B16.Kb^loss.gB_IFNβ cells showed the most striking increase in T cell expansion over and above all other subtypes tested, as well as vaccination with a commonly used adjuvant in the clinic (40), polyinosinic-polycytidylic acid (poly I:C). This trend was also observed when using recombinant doses of two of our highest-performing IFN subtypes, IFNα1 and IFNβ. Vaccination with irradiated B16.Kb^loss.gB cells in combination with 1 x 10⁵ U IFNβ, but not IFNα1, produced significantly enhanced CD8⁺ T cell expansion compared to vaccination with the cell inoculum alone, indicating a subtype-intrinsic effect ( Supplementary Figure 1B ).

Next, we performed our whole-cell vaccination protocol in IFNAR^o/o mice that lack the receptor through which all type I IFNs signal (34). We first verified that transferred transgenic gBT.I cells were not rejected in IFNAR^o/o mice by confirming similar persistence 30 days post-transfer to that observed in wildtype C57BL/6 mice ( Supplementary Figure 1C ). Expansion of transferred naïve gBT.I CD8⁺ T cells in response to vaccination was abrogated in IFNAR^o/o mice, demonstrating a requirement for signaling through IFNAR on host cells for IFN to have adjuvant activity ( Figure 1C ). Interestingly, increased CD8⁺ T cell expansion in IFNAR^o/o mice was observed following vaccination with the B16.Kb^loss.gB line in combination with the adjuvant poly I:C, suggesting IFNAR-independent mechanisms for this adjuvant.

We next asked if enhanced recruitment of gB-specific CD8⁺ T cells during vaccination also occurred in the endogenous T cell compartment, selecting two of our strongest performing adjuvant candidates, IFNα1 and IFNβ, for the remainder of our analyses. Splenocytes from vaccinated C57BL/6 mice were re-stimulated ex vivo with the immunodominant gB_498-505 peptide (41) and the percentage of IFNγ ⁺ CD8⁺ T cells was measured as a marker of vaccine-specific CD8⁺ T cell expansion. Consistent with our results from transgenic T cell experiments, endogenous gB-specific CD8⁺ T cell expansion was significantly enhanced by vaccination with IFNβ compared to IFNα1 or no adjuvant (1.84- and 3.89-fold increase, p=0.004 and p<0.0001 respectively; Figure 1D ), which was abrogated in IFNAR^o/o mice ( Figure 1E ). A sustained increase in gB-specific CD8⁺ T cells was observed 60 days post-vaccination, suggesting the potential for long-lived anti-tumor responses ( Supplementary Figure 2 ).

Considering IFN does not directly act on CD8⁺ T cells to drive T cell expansion, as demonstrated by the lack of expansion of transferred gBT.I cells in IFNAR^o/o mice, we next focused on the role of specific cell types that may be critical for driving enhanced CD8⁺ T cell expansion. Given that the tumor cells comprising the vaccine inoculum are unable to present the K^b-restricted gB antigen directly to CD8⁺ T cells, we evaluated cross-priming by professional antigen-presenting cells. We and others (30, 42–44) have demonstrated previously that XCR1⁺ cross-presenting DCs are the key cell type cross-priming anti-tumor CD8⁺ T cell immunity. To determine whether these cells were responsible for the cross-priming in our vaccination model, we utilized XCR1-DTR mice (35) to selectively deplete XCR1⁺ DCs immediately prior to vaccination with irradiated B16.Kb^loss.gB_ IFNβ cells. Consistent with the critical role XCR1⁺ DCs are reported to play in cross-priming, the enhanced gBT.I expansion observed post-vaccination with IFNβ in control mice was completely abrogated in the XCR1⁺ DC-depleted mice, confirming that these cells are essential for priming the observed CD8⁺ T cell response in this setting ( Figure 2A ).

NK/DC signaling can occur via Flt3L (45), IFNγ and TNFα (12) and may be a crucial factor in augmenting cross-priming and anti-tumor responses (46–48). Therefore, we hypothesized that NK cells may also contribute to the enhanced cross-priming mediated by IFNβ in our model. To assess this, endogenous NK cells were depleted one day before and after vaccination with irradiated B16.Kb^loss.gB_IFNβ cells ( Supplementary Figures 3A,B ). However, no effect on the expansion of transferred tumor-specific transgenic CD8⁺ T cells was observed in these mice ( Figure 2B ). We next considered the role of CD4⁺ T cells in our model, which can license DCs for successful cross-priming (49). Helper T cell dependence was indicated by the poor induction of endogenous IFNγ-producing gB-specific CD8⁺ T cells in MHC class II-deficient mice (I/AE^o/o) ( Figure 2C ). Whilst significant expansion of tumor-specific T cells remained following vaccination with IFNβ in I/AE^o/o mice, a greater than 3-fold decrease (5.5% vs 1.8%, p < 0.0005) was observed when compared to the percentage of IFNγ ⁺ CD8⁺ T cells in wildtype C57BL/6 mice ( Figure 1D ). We hypothesized that the observed T-helper dependence could reflect a requirement for CD40/CD40L signaling. To this end, we blocked CD40L in mice vaccinated with IFNβ and measured the expansion of transferred gBT.I CD8⁺ T cells. There was approximately a 3-fold decrease in expansion between control isotype- and anti-CD40L-treated mice following whole-cell vaccination with IFNβ (4.8% vs 1.6%, p < 0.0001) ( Figure 2D ), suggesting a dependence on CD40L signaling for optimal CD8⁺ T cell priming.

We next investigated the impact of whole-cell vaccination with IFNβ on circulating T cells and infiltration into the tumor microenvironment (TME) in mice bearing B16.gB tumors. Re-stimulation of lymphocytes ex vivo with gB_498-505 peptide demonstrated an increase in the number of endogenous gB-specific CD8⁺ T cells in the spleen, tumor and lymph nodes of mice vaccinated with IFNα1 or IFNβ compared to vaccination alone ( Figure 3A ). Furthermore, IFNβ-vaccinated mice showed significantly higher numbers of tumor-specific CD8⁺ T cells in the spleen and tumor relative to IFNα1-vaccinated mice suggesting successful tumor-infiltration of functional, tumor-reactive CD8⁺ T cells.

We next sought to determine the therapeutic potential of our vaccination strategy. Vaccination three days post-B16.gB tumor inoculation resulted in a significant increase in survival for the IFNβ-vaccinated cohort (18.4 ± 6.1 days) relative to vaccination with B16.Kb^loss.gB cells without IFN (14.8 ± 3.7 days, p < 0.0323) ( Figures 3B, C and Supplementary Figure 4 ). Strikingly, 40% of mice in the IFNβ-vaccinated group survived to day 21, compared to 0% of mice receiving vaccination alone or with IFNα1. Therefore, whilst both IFNα1- and IFNβ-vaccination has the capacity to increase tumor-specific T cell infiltration into the TME, only IFNβ resulted in a therapeutic benefit.

It is well established that the immunosuppressive TME can induce upregulation of inhibitory markers on infiltrating immune cells, which can be overcome by CPB (50). For example, infiltration of the tumor by T cells expressing the inhibitory receptor PD-1 is associated with response to anti-PD-1 therapy (51). We assessed PD-1 expression on tumor-infiltrating transgenic CD8⁺ T cells seven days post-vaccination and observed a significant upregulation across all groups relative to gBT.I T cells in the spleen ( Figures 4A, B ). These data provided a rationale to investigate whether vaccination with IFNβ would sensitize mice to anti-PD-L1 CPB. Mice were vaccinated three days post-tumor inoculation and dosed with anti-PD-L1 on days three, six, and nine post-vaccination ( Figure 4C ). Both IFNα1- and IFNβ-vaccination, but not vaccination without IFN (no IFN), significantly increased survival when used in combination with anti-PD-L1 relative to vaccination alone ( Figure 4D and Supplementary Figure 5 ). When compared to treatment with anti-PD-L1 in the absence of IFN, the combination of IFNβ-vaccination and anti-PD-L1 appeared to further increase overall survival, which however, did not reach a statistical significance. Notably, 30% of mice receiving IFNβ-vaccination plus anti-PD-L1 displayed complete tumor regression, surviving to at least 100 days post-tumor inoculation. In contrast, only 10% of anti-PD-L1 treated mice displayed this complete regression in the IFNα1-vaccination or vaccine alone groups. Thus, vaccination strategies incorporating IFNβ might function in conjunction with anti-PD-L1 treatment to promote overall survival. Interestingly, when comparing tumors harvested at endpoint from mice treated with or without anti-PD-L1, anti-PD-L1 appears to promote gB antigen downregulation (p=0.0079; Supplementary Figures 6A, B ). Surviving mice were re-challenged with wildtype B16 tumors, with a significant increase in survival observed in mice initially receiving IFNβ-vaccination relative to naïve, vaccine-alone or IFNα1-vaccinated mice ( Supplementary Figures 6C, D ). Taken together, these data suggest that enhanced epitope spreading may occur during IFNβ-vaccination.