4T1 and MDA-MB-231 were maintained in DMEM; BT-549 in RPMI, all supplemented with 10% heat inactivated FBS+100 U/mL penicillin and 100 µg/mL streptomycin (P/S). All cell lines were purchased from ATCC in the past year and were verified to be mycoplasma free and show appropriate microscopic morphology at time of use. VSVd51 expressing GFP was propagated on Vero cells and purified using Opti-Prep purification methods. Viral titers were determined by a standard plaque assay as previously published.19 Viral cytotoxicity was assessed on the indicated cell lines, and cell viability was carried out as described previously.19

Female BALB/c mice (6–8 weeks old, 20–25 g) were purchased from Charles Rivers (Quebec). Animals were housed in pathogen-free conditions at the Central Animal Facility of the Université de Sherbrooke with access to food/water ad libitum. Animals were euthanized by cervical dislocation under anesthesia. All studies were conducted in accordance with university guidelines and the Canadian Council on Animal Care.

We have previously developed and validated a mouse model of spontaneous metastasis and surgical resection of primary TNBC.36 37 At day 0, 1×10⁵ 4T1 cells in 100 µL (>98% viability) were injected orthotopically into the fourth mammary fat pad of BALB/c mice. Following tumor implantation, mice were monitored daily by palpation of the injection site, the volume of palpable tumors were measured by a Vernier caliper and the tumor volume was calculated via the equation (width²×length)/2. At days 12–14, a complete resection of the primary tumor was performed (tumor volume=75–80 mm³). During surgery, all mice were kept under anesthesia (3% induction, 1.5% maintenance of isoflurane with 2% O₂). For perioperative pain management, all mice were injected with 0.05 mg/kg of buprenorphine 1 hour before and 4 hours following surgery. Mice were randomized into different cohorts for treatment. At 1 day and 3 days following surgery, irradiated 4T1 cells, infected cell vaccine (4T1-ICV), VSVd51 alone or sterile PBS was injected subcutaneously into the cleared mammary fat pad. For the in vivo depletion of immune cell populations, 6 doses of depletion antibodies (1 dose 24 hours before surgery, followed by 5 additional doses 2–3 days apart) were administered by intraperitoneal injection at 20 µg/dose for anti-Asialo (GM1; Life Technologies) and 250 µg/dose for anti-CD8α (53-6.7; BioXCell). For anti-PD-1 treatment, mice were injected intraperitoneally with 1 dose of anti-PD-1 (RMP1-14; BioXCell) at 100 µg/dose, followed by 1 additional dose 3 days later.

TNBC tissue from patients were collected after surgery and placed in DMEM supplemented with 10% heat inactivated FBS+P/S. Tumors were dissociated using the human tumor dissociation kit (Miltenyi biotec) according to the manufacturer’s recommendations. Briefly, tumors were cut into small pieces (<2 mm³), then treated with enzymes and placed into the gentle MACS OctoDissociator (Miltenyi biotec). Macroscopic pieces were removed using 70 µm cell strainers. Tumor cells were washed twice in DMEM. Cells were viably frozen down or freshly used for downstream experiments.

ICV using VSVd51 was prepared as previously published.19 Viable single cell suspensions (5×10⁶) of 4T1 primary tumors were γ-irradiated at 50 Gy. This range of cell number and irradiation has been previously determined to create a non-proliferating but intact whole-cell vaccine.22 VSVd51 was added to the cells at 5×10⁷ plaque-forming units (PFU) and further incubated at 37°C for 24 hours. This preparation was injected subcutaneously in mice at 100 µL, giving each mouse 1:10 γ-irradiated cells to virus per dose (10 multiplicity of infection, MOI).

Antibodies are listed in online supplementary table S1. To analyze mouse spleen and blood lymphocyte populations, an initial incubation was done in ACK lysis buffer for 5 min to lyse red blood cells. A total of 1×10⁶ splenocytes or blood cells were then added to each tube. Fc block was added prior to antibody staining for 20 min at 4°C. Samples were washed twice with flow cytometry buffer (PBS+2% FBS) and acquired on a CytoFLEX 30 (Beckman Coulter). Data were analyzed with CytExpert software. For assessment of NK and T-cell functionality, fresh blood or spleen lymphocytes were cultured with VSVd51-infected 4T1 cell lysates for 4 hours in the presence of brefeldin A (1 µL/mL) at 37°C. After 4 hours, cells were washed twice with PBS, and then stained for NK and T-cell markers (online supplementary table S1). Cells were then fixed and permeabilized using BD Cytofix/Cytoperm kit, according to the manufacturer’s protocol, and intracellular staining for granzyme B and IFNγ was performed. For CD107a, the antibody was added to cells in the beginning for cell-surface staining of CD107a on degranulation, as described previously.

Conditioned media (CM) was obtained by seeding 5×10⁵ cells in 24-well plates in their corresponding media for 24 hours followed by infection with VSVd51 at the indicated PFU for the indicated time points. Flow cytometry: infected cells were harvested and processed as described above. Antibodies are listed in online supplementary table S1. Bioimaging was performed using a fluorescence microscope (Leica). Western blot: proteins from CM (HMGB1) or cell lysates (for other proteins) were resolved by SDS-PAGE and transferred to Immun-Blot-PVDF membranes (BioRad) for immunoblotting. Protein expression was detected using specific primary antibodies (1:1000) and corresponding HRP-conjugated secondary antibodies (1:10,000). Protein expression was visualized by chemiluminescence detection (ChemiDoc; BioRad). All antibodies are listed in online supplementary table S1. For ATP detection, the relative concentration of ATP in the CM was measured with the ENLITEN-ATP kit (Promega). Briefly, 100 µL of CM was transferred to 96-well opaque plates. Then 100 µL of reconstituted rLuciferase/Luciferin reagent was added to each well followed by measurement of luciferase using a luminescence microplate reader (Fusion V.3.0).

Seventy-two hours following infection with VSVd51, TNBC cells were harvested and provided to the histological platform (Université de Sherbrooke) for processing and imaging. Briefly, cells were fixed in 2.5% cacodylate-buffered glutaraldehyde for a minimum of 2 hours. Cells were then centrifuged and the pellet resuspended in sodium cacodylate buffer (pH 7.2), post-fixed in 2% osmium tetroxide for 2 hours and dehydrated up to absolute ethanol concentrations. A final dehydration in pure acetone was followed by three changes in Spurr’s resin and a final embedding at 65°C. Then 80 nm sections were cut using a Leica Ultracut UCT and stained with uranyl acetate and lead citrate. Grids were screened on a Hitachi 7500 Transmission EM and the images were digitally captured.

Total RNA from infected mouse and human TNBC cell lines and single cell suspensions of patient TNBC tissue were extracted using Trizol (Invitrogen) and cDNA was prepared using the QuantiTect Reverse Transcription Kit (Qiagen) from 1.2 µg of total RNA in the presence of 20 IU of RNAseOUT (Invitrogen). Target transcripts were quantified by qPCR by cDNA amplification with Taq Polymerase (Feldan Bio) and SYBR Green I (Millipore Sigma) according to the manufacturer’s instructions. All primers were purchased from Integrated DNA Technologies and are listed in online supplementary table S2. Target gene expression was normalized to housekeeping genes 18S ribosomal proteins for human cell lines and tissue samples or to Rplp0 ribosomal protein for mouse cell lines. Transcript levels were calculated using the ΔΔCT (cycle threshold) method and results were displayed as fold change of infected samples relative to uninfected samples. Genes were additionally visualized on 2% agarose gel to confirm their expression if basal levels were not detected.

Culture supernatants were diluted fivefold. ELISA kits for detecting mouse CXCL10, CCL5, CCL2 and CCL4 were purchased from Peprotech and performed according to the manufacturer’s instructions.

Polarization: human monocytes were isolated from peripheral blood (Human CD14⁺ isolation kit; Stemcell). Then 5×10⁵ monocytes were seeded in 24-well plates in complete RPMI and incubated overnight at 37°C and 5% CO₂. Twenty-four hours later, the monocyte media was replaced with the CM of infected human cell lines. For controls, monocytes were co-cultured with recombinant human IL-10, IL-4 and TGFβ (BioBasic) all at a final concentration of 20 ng/mL for differentiation to M2-like macrophages; and with lipopolysaccharides (50 ng/mL) (Millipore Sigma) and recombinant human IFNγ (20 ng/mL) (BioBasic) for M1-like macrophages. Undifferentiated monocytes remained in complete media as M0. Following overnight incubation, cells were harvested and processed for flow cytometry as described above. Antibodies are listed in online supplementary table S1. Proliferation: monocytes from the polarization assay were further incubated with carboxyfluorescein succinimidyl ester (CFSE)–labeled (3 µM; BD Horizon) human peripheral blood mononuclear cells (PBMCs) for 5 days in the presence of plate-bound anti-CD3 (OKT3, 5 µg/mL) and anti-CD28 (CD28.2, 2 µg/mL). Cells were harvested and stained with anti-CD3 (HIT3a), anti-CD8 (SK1) for flow cytometric analysis of CFSE dilution. Migration: 200 µL of CM was placed in the lower well of Boyden chambers, separated from the top well by a 5 mm pore polycarbonic membrane (Neuro Probe). Then 6×10⁵ human PBMCs was added to the top chamber, followed by incubation at 37°C, 5% CO₂ for 45 min. Next, the media in the top of the chamber was aspirated and the membrane removed with forceps. This was followed by harvesting of media in the bottom chamber and quantification of migrated cells by Trypan Blue exclusion. The cells were stained and acquired by flow cytometry as described above.

Lungs from treated mice were harvested and fixed in 4% formaldehyde for 24 hours and kept in ethanol for analysis. Lungs were then embedded in paraffin and 5 µm sections were used for immunofluorescence staining using CD3, IFNγ and granzyme B. All antibodies are listed in online supplementary table S1. Slides were scanned using a digital slide scanner (Nanozoomer-XR C12000; Hamamatsu) provided by the Histology Platform (Université de Sherbrooke). Percentage staining of marker-positive areas were quantified using ImageJ software (NIH).

All statistical analyses were generated using Prism V.7 (GraphPad). Unpaired two-tailed t-tests were used for comparing uninfected or infected cells or differentially treated mice. Survival differences of tumor-bearing and treated mice were assessed using Kaplan-Meier curves and analyzed by log-rank testing. P value <0.05 was considered as statistically significant.

We previously demonstrated that an autologous rhabdovirus ICV elicited profound anti-tumor immune responses in B16 melanoma and CT26 peritoneal carcinomatosis preclinical models.19 Given the lack of therapeutic options for poor-prognosis TNBC, we proposed to develop an adjuvant ICV to prevent relapse and reduce metastases in this aggressive disease. We used rhabdoviral VSVd51 expressing enhanced green fluorescence protein (GFP) and first assessed its cytotoxic activity in mouse and human TNBC cells. VSVd51 was able to infect mouse 4T1 and human MDA-MB-231 and BT-549 cells as shown by GFP expression following 72 hours of infection with 10 MOI (figure 1A) and induce cellular cytotoxicity over a range of increasing MOI as measured by a MTT assay (figure 1C). Given the importance of the mode of tumor cell death in initiating anti-tumor immune responses,38 39 we investigated cell death features following infection of TNBC cells with VSVd51. We first examined cellular morphology using transmission EM (figure 1B). Condensed nuclear structures, cytoplasmic vacuoles and ruptured cellular membranes were observed. Next, we detected high mobility group box 1 (HMGB1) protein (figure 1D) and ATP (figure 1E) in the supernatant of VSVd51-infected cells at various time points post-infection, suggesting passive release from necrotic cells. Another feature of necrosis is the presence of cell surface externalized calreticulin. Following VSVd51 infection, we observed an increase in the percentage of necrotic (calreticulin⁺/DAPI⁺) cells in all tested cell lines at 48 and 72 hours post-infection (figure 1F). Together, the presence of these danger-associated molecular patterns (DAMPs) suggest a necrosis-like phenotype of TNBC cells following VSVd51 infection. Features of classical apoptosis (Annexin V⁺/DAPI⁻, Caspase-3 and PARP cleavage) were minimally or not observed (online supplementary figure 1A,B). In addition, the autophagic flux was blocked by bafilomycin treatment and no differences in the conversion of LC3-I to LC3-II was observed following VSVd51 infection in all cell lines tested. This suggests that VSVd51 infection of TNBC cells does not lead to autophagic cell death (online supplementary figure 1C). By comparison, treatment of TNBC cells with doxorubicin, a clinically relevant neoadjuvant chemotherapeutic for TNBC, revealed that VSVd51 induced greater release of HMGB1 and calreticulin exposure (online supplementary figure 1D,E).

Next, we sought to determine if VSVd51-induced necrosis is immunogenic in nature. To accomplish this, we examined a panel of genes related to pro-inflammatory, anti-inflammatory, antigen presentation and immune differentiation markers by qPCR. Following overnight infection with VSVd51, we detected a general upregulation of genes related to immune cell recruitment and activation in mouse and human TNBC cell lines tested. Notably, mouse CCL2, CCL4, CCL5, CXCL10, IL-6 and MHC-I related genes showed an increase in expression in 4T1 cells following infection compared with non-infected controls (figure 2A). The expression of several top immunogenic genes (CCL2, CCL4, CCL5 and CXCL10) at the protein level were measured by ELISA in 4T1 cells following infection to support the gene expression data (online supplementary figure 2A). In human MDA-MB-231 and BT-549 cell lines, CCL5, CXCL2, IRF-1 and MHC-I genes were upregulated (figure 2B,C). Genes were additionally visualized on 2% agarose gel to confirm their expression if basal levels were not detected. Notably, IFNγ, IL-2 and PD1 genes were induced following infection in 4T1 cells (online supplementary figure 2B), while CSF-1, CCL4 and CXCL10 were induced in BT-549 and MDA-MB-231 infected cells (online supplementary figure 2C,D). These data suggest that an immunogenic gene signature is present in TNBC cells following VSVd51-induced necrotic cell death.

To determine if the observed in vitro ICD features (figure 1) and gene signatures (figure 2) translate to enhanced immune function in vivo, we tested the ICV in the adjuvant setting in BALB/c mice bearing orthotopic 4T1 tumors following primary tumor resection (figure 3A, timeline). This model makes use of an aggressive mouse stage IV TNBC from the BALB/c strain that spontaneously metastasizes from the mammary glands to multiple distant sites, in particular the lungs. We included the following treatment cohorts (PBS, irradiated 4T1 cells, VSVd51 alone and ICV) to delineate the role of the vaccine’s constituent parts. At early and late time points following vaccination, we observed that postoperative vaccination of mice with 2 doses of ICV significantly enhanced the proportion of blood IFNγ⁺, granzyme B⁺ (cytotoxicity) and CD107a⁺ (degranulation) NK cells compared with administration of virus alone, non-infected cells or PBS (figure 3B). Similar results were observed in CD11c⁺ conventional dendritic cells (DCs) in terms of their overall proportion and activation status (CD86⁺) (figure 3C). Analysis of both blood and lung CD3⁺/CD8⁺ T cells showed enhanced IFNγ, granzyme B and CD107a degranulation in ICV-treated mice over controls (figure 3D,E). Immunofluorescence staining of mice lungs bearing metastatic 4T1 tumors treated with ICV showed increased presence of CD3⁺ T cells, granzyme B and IFNγ expression compared with lungs from mice treated with irradiated 4T1 cells (figure 3F,G). By comparison with ICV, the addition of a treatment cohort receiving systemic doxorubicin injection, a known ICD inducer, resulted in decreased CD8⁺ T-cell CD107a degranulation and IFNγ production compared with ICV treatment. Importantly, survival of ICV-treated mice significantly surpassed those cohorts treated with irradiated 4T1 cells or doxorubicin (online supplementary figure 3A–C). Taken together, these in vivo data demonstrate the innate and adaptive immune activating capacity of the ICV approach.

To further investigate the critical role of NK and CD8⁺ T cells after ICV administration, we monitored for survival in ICV-treated mice that were pharmacologically depleted singly or of both immune cell populations (figure 4A,B). In support of the in vivo data showing enhanced NK and CD8⁺ T-cell function (figure 3), the protective effect of vaccinated mice with ICV was partially abrogated on depletion of NK cells, but completely abrogated on depletion of CD8⁺ T cells or combination of NK and CD8⁺ T cells (figure 4B). These results suggest that the therapeutic benefit of this treatment strategy is dependent on both NK and CD8⁺ T-cell recruitment, but likely more dependent on CD8⁺ T cells. Given the importance of CD8⁺ T cells and their role in mediating the response to ICV treatment, we examined cell surface expression of exhaustion markers on CD8⁺ T cells at day 9 following ICV treatment and observed augmented levels of PD-1, but not TIM-3 or LAG-3 (figure 4C). In addition, we observed upregulation of PD-L1 expression levels on 4T1 cells following infection with VSVd51 in vitro (figure 4D). These data suggest that the adaptive T-cell response could be modulated to override exhaustion. Therefore, to improve the immune response and survival of vaccinated mice, we combined ICV with anti-PD1 checkpoint inhibitor treatment (figure 4E,F). We observed that combination therapy prolonged survival compared with either monotherapy ICV or anti-PD-1 alone. These preclinical results demonstrate the therapeutic potential of ICV in combination with checkpoint inhibitors to treat TNBC.

To improve the translational potential of our work, we examined the effect of ICV on human primary antigen-presenting cells. In ex vivo co-culture experiments with CD14⁺ human monocytes incubated with cell-free lysates derived from infected human TNBC cells, we observed polarization of monocytes toward an M1-like phenotype that have been previously suggested40 41 to promote anti-tumor immune responses (figure 5A). We additionally examined the activation status of human DC treated ex vivo with the same cell-free lysates (online supplementary figure 4A). ICV-lysate treated DC displayed a more mature phenotype compared with controls. To examine the consequences of ICV-induced M1-like monocytes on effector immune cells, we measured human NK and CD8⁺ T-cell migration and CD8⁺ T-cell proliferation in the ex vivo setting. We observed increased migration of NK cells and increased migration and proliferation (CFSE dilution) of CD3⁺/CD8⁺ T cells in co-cultures with ICV-lysate treated M1 monocytes (figure 5B,C). Taken together, these data using human primary immune cells and human TNBC cell lines demonstrate the immune activating potential of ICV.

To investigate whether the ICV could elicit an immunogenic signature in human TNBC patient tissue, patients with TNBC were enrolled in the VACS study as part of the Sherbrooke Gynecologic Biobank (Ethics no. 2018-2414). Dissociated breast tumor tissue was obtained from two patients with TNBC (BRC1762, BRC1756). The cells were infected with VSVd51 overnight and qPCR for gene expression analysis and assays to measure biomarkers of ICD were conducted. Patient BRC1762 displayed an immunogenic gene expression pattern with enhanced expression of multiple immune genes, notably CCL5, CCL2, CXCL9, CXCL11, CCL3,TGFb, CSF-2, TAP1 and TAP2 (figure 6A). In addition, the genes CCL20, IFNα, IFNβ and GRA that were not basally expressed in uninfected samples showed induced expression following infection with VSVd51. In patient BRC1756, CCL2, CCL5, CXCL2, CCL20, IRF1, TAP1 and TAP2 gene expression were also increased (figure 6B) and the genes CCL20, CTLA-4, CCL3 and CCL4 were induced following infection (online supplementary figure 4B). Biomarkers of ICD including calreticulin cell surface expression (for patient BRC1762) (figure 6C), ATP (figure 6D) and HMGB1 (figure 6E) release for both patients were detected at higher levels in VSVd51-infected cells compared with uninfected controls. These human data demonstrate that an ICD gene signature is present in patient TNBC cells following VSVd51 infection, and this phenotype has the potential of recruiting and activating important immune cells in vivo. Taken together, our translational data highlight the clinical potential of using ICV as adjuvant vaccine to treat patients with TNBC.