Immunodominant AH1 Antigen-Deficient Necroptotic, but Not Apoptotic, Murine Cancer Cells Induce Antitumor Protection Free

This article is featured in In This Issue, p.731

Immunogenic cell death (ICD) and its relevance for cancer therapy relies on the induction of an effective tumor–host interplay, in which the dying cancer cells mediate an adjuvant-accompanied and Ag-specific antitumor immune response. The combination of tumor adjuvanticity and antigenicity is achieved by the cross-presentation of tumor-associated Ags (TAAs) by APCs like dendritic cells. During ICD, the dying cells release damage-associated molecular patterns (DAMPs) but will also, depending on the cell type, secrete immunostimulatory molecules such as chemokines and cytokines that contribute to the activation of the host’s innate immune cells. In turn, the activated dendritic cells prime CD8⁺ CTLs, which eradicate the tumor mass upon TAA recognition (1–5). This concept of ICD is mainly based on experimental work involving the use of murine cancer cell lines, which, following cell death induction, are studied for their tumor protection efficiency in a prophylactic vaccination model (6). Besides adjuvanticity, the vaccination model requires TAA processing and recognition by the host (i.e., mouse) immune system and hence allows studying CTL cross-priming and activity (7, 8). Conveniently, many stable cancer cell lines express endogenous virally derived peptides that are recognized by the immune system as TAAs (9–11). CT26, a murine colon carcinoma cell line expressing the murine leukemia retrovirus (MuLV)–derived protein Gp70, is such an example. The Gp70 glycoprotein is one of two MuLV envelope-encoded gene products and harbors an immunodominant epitope corresponding to Gp70_423–431/H-2L^d that is translated into the AH1 peptide sequence SPSYVYHQF (12–14). Gp70 is expressed and known to give rise to TAA-specific T cell responses in several murine cell lines, such as 4T1 mammary carcinoma, EL4 lymphoma, and B16 melanoma cells (15), but especially in CT26 cells, Gp70 is found to be the most abundant transcript (16).

In ICD studies, apoptosis is often induced in cancer cells by the use of chemotherapeutics (mitoxantrone, doxorubicin, oxaliplatin, etc.), photodynamic therapy, UV light, or gamma radiation (17, 18). However, especially in the case of chemotherapeutic triggers, it is difficult to distinguish between the immunogenic effects deriving from the cell death process and the direct effects of the drug on the immune system (19–24), although these can be of therapeutic relevance. To avoid such drug-mediated bystander effects, we used ligand-free Tet-ON–inducible constructs that allow studies of the immunogenic impact deriving only from the cell death process itself. To compare the impact of different cell death modalities on TAA expression and recognition, we used our previously reported immunogenic necroptosis system (25) and a similar Tet-ON–inducible apoptosis system. To date, inducible apoptosis has been shown with the active form of caspase-3 (26, 27), a dimerizing form of caspase-9 (28), caspase-8 (29), and a proteolytically activated form of Bid (tBid) (26) and Bim (30). However, common to all of these studies is that the TAA-specific immunogenicity was described only using cells overexpressing OVA, which is an exogenous Ag fulfilling the role of a model tumor Ag. Acknowledging the vast amount of literature on the immunodominant role of Gp70 and AH1 (31–36) in combination with our previous findings on immunogenic necroptosis using a ligand-free inducible system in the CT26 cell line (25), we aimed to study the immunogenic role of apoptosis in the same ligand-free inducible manner and compared it to necroptosis while taking into account that the AH1 peptide expression in these cells is likely to play an immunodominant role as an endogenously derived TAA.

In this study, we show that CT26 cells receiving an apoptosis stimulus give rise to a less prominent adaptive anti-AH1 immune response than in the case of necroptosis. Using AH1-deficient CT26 cells, we demonstrated that AH1 expression dictated the immunogenicity of cancer cells that had received an apoptosis stimulus. Although both apoptotic and necroptotic CT26 cells would efficiently protect mice against challenge with another AH1-expressing cancer cell line, 4T1 mammary carcinoma, this immunogenicity was lost in apoptotic, but not necroptotic, cells upon knockout (KO) of AH1, demonstrating tumor Ag immunodominance. Additionally, immunization of mice with necroptotic CT26 cells revealed the involvement of CD4⁺ and CD8⁺ T lymphocyte–specific neoepitopes likely to explain why necroptotic cells would induce an Ag-specific immune response independent of the AH1 expression in the dead cell vaccine. Our findings highlight necroptosis as a potentially more immunogenic form of cell death because of its capacity to give rise to a broader antitumor immune response against tumor-intrinsic neoepitopes.

The murine colon carcinoma CT26 cells and the murine breast cancer 4T1 cells were cultured in DMEM (Life Technologies) supplemented with 10% heat-inactivated FCS (Life Technologies), l-glutamine (0.03%), and sodium pyruvate (0.4 mM); all products were LPS free. Cells were cultured at 37°C in the presence of 5% CO₂ and were tested for the absence of mycoplasma contamination on a regular basis. The CT26 Fas-associated death domain (DD) protein (FADD) cell line was obtained by lentiviral transduction of CT26 wild-type (WT) cells with pDG2-flag-rtTA3-puro-myristoylation-2FKBP12-l-mFADD plasmid. This is a tetracycline-inducible derivative of pLenti6 (Life Technologies) containing a puromycin selection marker, in which the coding sequence for a myristoylation-targeting peptide, two tandem repeats of FKBP12 (Takara Bio), a short linker peptide (L, GSGGGS), and the full-length murine cDNA of FADD (BC021400; Open Biosystems) was cloned. Upon lentiviral transduction, the stably transduced cells were selected with 9 μg/ml puromycin. Clones were obtained by limited dilution. Apoptosis was induced by 1 μg/ml doxycycline (doxy) and 10 nM of B/B Homodimerizer (Takara Bio). The CT26 FADD DD and receptor-interacting protein kinase (RIPK) 3 co-overexpressing CT26 cells (DDR3) cell line was generated as previously described (25), and necroptosis of this cell line was induced by stimulation with 1 μg/ml doxy alone or in combination with 10 nM B/B Homodimerizer (Takara Bio).

Cell death was analyzed on a FLUOstar Omega Microplate Reader (BMG LABTECH, Offenburg, Germany). Six or ten thousand cells were seeded in a transparent 96-well plate. Two micromolar SYTOX Green Nucleic Acid Stain (S7020; Molecular Probes) and 33 μM Ac-DEVD-7-amino-4-methylcoumarin (AMC) (3171-V; PeptaNova) were added to the cells, and the cells were then stimulated with doxy and the homodimerizer B/B (doxyBB) as described above. Maximum cell death was obtained by treatment with Triton X-100 (0.05%). This allows the expression of cell death as percentage of maximum SYTOX Green fluorescence (excitation 485 nm, emission 520 nm). If the executor caspases-3/-7 are activated during cell death, they are released upon plasma membrane rupture and cleave Ac-DEVD-AMC present in the culture medium, resulting in the release of fluorescent AMC (excitation 355 nm, emission 460 nm). The DEVD activity is expressed as fold induction compared with the maximum fluorescence intensity value.

Cells were seeded in an eight-well chamber (ibidi), with 15,000 cells per well in 200 μl complete growth medium. Cell death was induced just before imaging, and cell death was monitored by SYTOX Green (10 nM; Molecular Probes). Live cell imaging was performed on a Leica Sp5 AOBS confocal microscope (Leica, Mannheim, Germany), using a 40× HCX PL Apo UV 1.25 numerical aperture oil objective. Cells were imaged for bright-field and SYTOX Green using the 488 line of a Multiline Argon laser. Images were acquired in a sequential mode. Images were captured in three frames per well every 20 min, and the pinhole was opened to 2 AU to prevent phototoxicity and photobleaching; likewise, scanner was set at a speed of 700 Hz. Different focal planes were set at 1-μm intervals. From each image stack, maximum intensity projections for SYTOX Green and montages of the Multi-TIFF time series and two-channel overlays were made by using ImageJ 1.49m, a public domain imaging software.

The 1.25 × 10⁶ CT26 FADD cells were seeded per 10-cm petri dish and stimulated with doxy for 18 h. Twenty micromolar zVAD-fmk and 10 nM B/B Homodimerizer were added for the last 0.5–1.5 h. After treatment with the indicated triggers, the cells were washed with PBS and lysed in NP-40 lysis buffer (150 mM NaCl, 1% NP-40, 10% glycerol, and 10 mM Tris-HCl [pH = 8]) supplemented with cOmplete, EDTA-free Protease Inhibitor Mixture Tablets (11873580001; Roche), and PhosSTOP Phosphatase Inhibitor Mixture Tablets (04906837001; Roche). Immunoprecipitation was done overnight using protein G–coupled anti-FADD (sc-6036; Santa Cruz Biotechnology) or protein A–coupled anti–caspase-8 (homemade Ab). After immunoprecipitation, the beads were washed three times in NP-40 lysis buffer, and the immunoprecipitated proteins were eluted by adding 60 μl of 2× Laemmli buffer to the beads. Complexes were subsequently analyzed by immunoblotting (20 μl protein loaded per lane).

Samples were lysed in Laemmli buffer containing 50 mM Tris-HCl (pH 6.8), 2% SDS, and 10% glycerol. The protein concentration for each sample was determined using a NanoDrop 1000 spectrophotometer according to the manufacturer’s recommendations. The proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. Western blotting was performed with anti-FKBP12 (PA1-026A; Thermo Fisher Scientific), anti-RIPK3 (R4277; Sigma-Aldrich), anti–caspase-8 (MAB3429; Abnova), anti-cleaved caspase-8 (9429; Cell Signaling Technology), anti–caspase-3 (9662; Cell Signaling Technology), anti-cleaved poly(ADP-ribose) polymerase (9544; Cell Signaling Technology), anti-actin (69100; MP Biomedicals), anti–phospho–IκB-α (9246; Cell Signaling Technology), anti–IκB-α (sc-371; Santa Cruz Biotechnology), anti-RIPK3 (2283; ProSci), anti-tubulin HRP (ab21058; Abcam), anti-FADD (sc-6036; Santa Cruz Biotechnology), anti-RIPK1 (610459; BD Biosciences), anti-MLKL (MABC604; MilliporeSigma), and anti–phospho-MLKL S345 (ab196436; Abcam). Secondary Abs included Amersham ECL HRP-linked sheep anti-mouse IgG (NA931; GE Healthcare), donkey anti-rabbit IgG (NA934; GE Healthcare), donkey anti-goat IgG (SC-2020), or goat anti-rat IgG (NA935; GE Healthcare). After incubation with the appropriate secondary Abs conjugated to HRP, blots were revealed using ECL Western blotting substrate (32106; Pierce Biotechnology) or Western Lightning Plus-ECL (NEL105001EA; PerkinElmer).

Cells were washed in annexin V binding buffer (556454; BD Biosciences), followed by staining with SYTOX Blue Nucleic Acid Stain (S11348; Molecular Probes) and allophycocyanin annexin V (550474; BD Biosciences) or annexin V–Alexa Fluor 488 conjugate (A13201; Molecular Probes). Samples were run on a BD FACSVerse or BD LSR II flow cytometer. The data were analyzed using FlowJo 10.6.1 (Becton Dickinson).

CT26 cells were collected after the indicated stimulation (with mitoxantrone or doxyBB) and washed once in ice-cold FACS buffer (PBS [Gibco Dulbecco PBS (DPBS), 14190-094; Life Technologies] plus 0.5% FCS [Life Technologies]). Calreticulin staining was performed in two steps: first with a rabbit anti–calreticulin Ab (92516; Abcam) for 30 min at 4°C and next with a donkey anti-rabbit Ab coupled with DyLight 488 for another 30 min at 4°C. Before acquisition on a BD FACSVerse flow cytometer (BD Biosciences), the cells were resuspended in FACS buffer with SYTOX Blue Dead Cell Stain (S11348; Molecular Probes). The data were analyzed with FlowJo 10.6.1 (Becton Dickinson).

Supernatant was collected and cleared from cells by centrifugation (1500 rpm for 5 min in the case of HMGB1). ATP was quantified using an ATP Bioluminescent Assay Kit (Sigma-Aldrich) and measured with the GloMax96 Microplate Luminometer (Promega). HMGB1 was measured with an ELISA kit (IBL International, Hamburg, Germany) and was quantified using the iMARK Microplate Absorbance Reader (Bio-Rad Laboratories), followed by analysis with a Four Parameter Logistic Curve Fit (MyAssays).

Cells were seeded in six-well plates, left to adhere for 18 h, and then treated with doxy or doxyBB for another 18 h. Supernatant was collected and cleared from dying tumor cells by centrifugation (400 × g for 5 min). Cytokine quantification was performed using a bead-based multiplex immunoassay (ProcartaPlex; eBioscience) in accordance with the manufacturer’s instructions. Measurements were performed with Bio-Plex 200 Systems (Bio-Rad Laboratories) and analyzed with the Bio-Plex Manager software.

Streptococcus pyogenes WT Cas9 (pSpCas9(BB)-2A-GFP) was obtained from Addgene (plasmid no. 48138). The single guide RNAs (sgRNAs) targeting the AH1 region of gp70 were designed using the CRISPR Design Tool (http://crispr.mit.edu) and were ordered with Thermo Fischer Scientific. The sgRNA sequences are listed in Table I. The Cas9 vector was digested with BpiI (BbdI), and the sgRNA oligo was annealed and cloned intro the BpiI-digested Cas9 vector. The cloning protocol by F. Ann Ran was largely followed (37). The sgRNA Cas9 plasmid was transfected into CT26 cells via jetPRIME transfection reagent (catalog no. 114-15, Polyplus-transfection). For each transfection, 2 μg of plasmid was added per 250,000 cells. The medium was replaced after 4 h, and 4 d after transfection, the cells were harvested. Single-cell sorting was performed on a FACSAria III from BD Biosciences. Effective genomic deletion was confirmed with genomic PCR and with a pair of primers flanking the deletion, followed by Sanger sequencing, and the allele editing was analyzed using TIDE (http://tide-calculator.nki.nl). The PCR and sequencing primers used are listed in Table II.

RNA was isolated using the RNeasy Plus Mini Kit according to the manufacturer’s instructions (74134; QIAGEN). RNA concentration was measured on the NanoDrop 1000 spectrophotometer. For the inflammatory cytokine profile analysis, the RNA quality was assessed using Agilent RNA 6000 Nano Reagents and Nano Chips (5067-1511; Agilent Technologies) and analyzed with the Agilent 2100 bioanalyzer (Agilent Technologies). cDNA synthesis was performed with SensiFAST cDNA Synthesis Kit (BIO-65054; Bioline) or with iScript cDNA Synthesis Kit (170-8891; Bio-Rad Laboratories). AH1 gene expression was assessed with four Ah1 gene-specific primers (Table II) using SensiFAST SYBR No-ROX kit (BIO-98020; Bioline) on a LightCycler 480 instrument (Roche). Ptgs1 and Ptgs2 gene expression was assessed with TaqMan probes (Mm00477214_m1 and Mm00478374_m1; Applied Biosystems) using TaqMan Fast Advanced Master Mix (4444558; Applied Biosystems) on a LightCycler 480 instrument (Roche). Data were analyzed with qbase+ 3.1 (Biogazelle) generating normalized relative quantities. The results shown represent the fold increase compared with the values obtained in the untreated or nontransfected control samples.

Cells were harvested, and DNA was extracted with QuickExtract DNA Extraction Solution 1.0 (QE09050; Epicentre) according to the manufacturer’s instructions. One microliter of the lysate was used per 10 μl PCR reaction, in which a 629-bp region spanning the AH1-encoding sequence was amplified using Kapa HiFi HotStart ReadyMixPCR Kit (KK2602; Kapa Biosystems). The PCR product was purified using CleanPCR beads (CPCR-0050; CleanNa). Four hundred nanograms of purified DNA was used for the Surveyor Mutation Detection Kit for Standard Gel Electrophoresis according to the manufacturer’s instructions (706020; Integrated DNA Technologies). The samples were analyzed by separation on a 2% Tris-acetate-EDTA–agarose gel and imaged by staining with ethidium bromide. To sequence the mutant clones, the cleaned PCR products were analyzed by sequencing using a nested primer. Primer sequences are given in Table II.

Bone marrow–derived dendritic cells (BMDCs) were differentiated from the femurs and tibias of 7-wk-old BALB/c WT mice for 8 d using RPMI 1640 culture medium (Life Technologies) supplemented with 5% heat-inactivated FCS, l-glutamine (0.03%), sodium pyruvate (0.4 mM), 2-ME (50 μM), and murine GM-CSF (20 ng/ml). Fresh culture medium was added on day 3, and on day 6, the medium was refreshed. On day 8, only the suspension cells were collected to be used in the phagocytosis and coculture assays.

The phagocytosis assay was performed entirely according to the protocol as previously described (25, 38). Cell death was induced in the CT26 FADD cells for 18 h. The dead cells were collected, washed in RPMI 1640 culture medium, and cocultured with 50,000 BMDCs in three different ratios, 1:1, 1:10, and 1:20 (BMDC/CT26 cells). Coculture was performed for 18 h at 37°C in six-well plates with 2.5 ml RPMI 1640 culture medium. Control BMDCs were left untreated or stimulated with 100 ng/ml LPS (L2630; Sigma-Aldrich). After coculture, all cells were collected, spun down (400 × g for 6 min at 4°C), and washed once in PBS (Gibco DPBS, 14190-094; Life Technologies) plus 0.5% FCS. Dead cells were excluded from the flow cytometry analysis by staining with SYTOX Blue (S11348; Molecular Probes). Maturation of BMDCs were analyzed by immunostaining using allophycocyanin–anti-CD11c (550261; BD Pharmingen), FITC–anti–MHC class II (MHCII) (11-5321; eBioscience), PE-CyTM7–anti-CD86 (560582; BD Pharmingen), PerCP-Cy5.5–anti-CD80 (A14895; Molecular Probes), and mouse Fc block (553142; BD Pharmingen). Gating strategy is as follows: single cells were obtained by gating on forward scatter area versus forward scatter height, followed by side scatter area versus side scatter height. Dead cells were excluded by SYTOX Blue. Mature BMDC cells were identified as a percentage of CD11c⁺MHCII⁺CD86⁺ and CD11c⁺MHCII⁺CD80⁺. All samples were acquired on the BD FACSVerse flow cytometer and analyzed with FlowJo 10.6.1 (Becton Dickinson).

PGE₂ was measured directly in conditioned medium from CT26 FADD AH1 KO or CT26 DDR3 AH1 KO cells that were left untreated, treated with 50 μM ibuprofen (710280; Cayman Chemical), and/or induced to cell death with doxy/doxyBB for 18 h in total. PGE₂ quantification was performed using a PGE₂ ELISA Kit (514010; Cayman Chemical) in accordance with the manufacturer’s instructions. Measurements were performed with the iMark Microplate Absorbance Reader (Bio-Rad Laboratories).

Female BALB/c WT mice (6–7-wk-old) were purchased from Harlan Laboratories or from Charles River Laboratories and female BALB/c Foxn1nu mice (6-wk-old) were bought from Janvier Labs. All mice were housed in specific pathogen-free conditions, and all experiments were performed according to guidelines of the local Ethics Committee of Ghent University.

CT26 cells were seeded on 15-cm petri dishes or in tissue culture-treated flasks, and cell death was induced via the inducible systems for 18–24 h. Where indicated, the cells were pretreated for 30 min prior to cell death induction with 5 μM TPCA-1 (2559; Tocris Bioscience) or for 1 h with 50 μM ibuprofen (710280; Cayman Chemical). After cell death induction, the cells were collected, washed once in PBS (Gibco DPBS, 14190-094; Life Technologies), and resuspended in the desired concentration in PBS. Mice were inoculated s.c. with the indicated amount of CT26 cells or with PBS on the left flank side. Accidental necrosis was induced by subjecting CT26 cells, already resuspended in the desired PBS volume, to 3× freeze-thaw (F/T) cycles (dry ice/37°C). On day 8 after vaccination, the mice were challenged s.c. on the opposite flank with 5 × 10⁵ untreated CT26 WT or 4T1 WT cells. Tumor growth on the challenge site was evaluated using an electronic caliper for up to 4 wk after the challenge. All mice, including the Foxn1nu mice, vaccinated with F/T necrotic cells were tumor free on the vaccination site. In the Foxn1nu mice, all other vaccinations would give rise to tumor formation on the vaccination site. Despite the high cell death percentage, vaccination with necroptotic cells would, in most cases (up to 80–100%), develop a tumor on the vaccination site. Apoptotic FADD AH1^KO cell immunizations would rarely (0–25%) give rise to a vaccination tumor. Apoptotic CT26 FADD immunizations would form a tumor on the vaccination site in 50–75% of the mice, except for Fig. 1C, in which all the mice were tumor free on the vaccination site. Because of the variations of tumor growth on the vaccination site and no obvious correlation between vaccination tumor growth and challenge tumor protection, all mice were included in the Kaplan–Meier curves throughout the paper despite the tumor growth on the vaccination site. Mice were sacrificed when a tumor on the challenge site (or, in some cases, on the vaccination site) became necrotic or reached a volume (V) of 2000 mm³; if tumor free on the challenge site on the day of sacrifice, this is shown as a symbol on the horizontal part of the Kaplan–Meier curve. The volume was calculated using the following formula: V = the numerical value of π × 1/6 × tumor length × tumor height × tumor width (39).

Mice were injected s.c. with 3 × 10⁶ apoptotic CT26 FADD cells, necroptotic CT26 DDR3 cells, or PBS at the tail base. Twenty-one days after immunization, the mice were sacrificed, and the spleens and inguinal draining lymph nodes were isolated. The spleens were treated twice with ACK Lysing Buffer (10-548E; Lonza), and, like the lymph nodes, the cells were spun down (400 × g, for 7 min, at 4°C) before resuspension in PBS (GibcoDPBS, 14190-094; Life Technologies). Following lysis of erythrocytes, 1 × 10⁵ lymph node cells from the draining lymph nodes or 2.5 × 10⁵ splenocytes were seeded per well on an IFN-γ capture Ab-coated 96-well PVDF bottomed plate (MSIPS4510; MilliporeSigma), and the cells were restimulated for 24 h with either 5 μg/ml of AH-1 peptide (SPSYVYHQF, AS-64798; tebu-bio), 5 μg/ml of neoepitopes (a mixture consisting of CT26-M20 [PLLPFYPPDEALEIGLELNSSALPPTE], CT26-M26 [VILPQAPSGPSYATYLQPAQAQMLTPP], CT26-M03 [DKPLRRNNSYTSYIMAICGM-PLDSFRA], CT26-M37 [EVIQTSKYYMRDVIAIESAWLLELAPH], and CT26-M27 [EHIHRAGGLFVADAIQVGFGRIGKHFW]), or 4 μg/ml Con A type IV (C2010; Sigma-Aldrich), which served as a positive control. The murine IFN-γ ELISpot assay was performed according to the manufacturer’s instructions (862.031.010; Diaclone). Spots were imaged using an A.EL.VIS 4 Plate ELISPOT Reader V2.1 and quantified manually. The presented values (each dot) are the mean values of technical triplicates. For each mouse, spots counted in wells with AH-1 peptide (or neoepitope)–stimulated cells had only the background RPMI 1640 culture medium stimulation subtracted. The overall mean (black horizontal bar) and 95% confidence interval (error bars) for each immunization is calculated for all of the mice from two independent experiments.

Statistical analysis was performed in GraphPad Prism (v.8.2.1). All prophylactic tumor vaccination experiments and the corresponding Kaplan–Meier survival curves showed the timeline for tumor occurrence on the challenge site. Calculations to compare two groups were done using the log-rank Mantel–Cox test. The phagocytosis assay was analyzed by a Mann–Whitney nonparametric t test. The BMDC activation and maturation assay was analyzed using a two-way ANOVA test. Gp70 real-time quantitative PCR (RT-qPCR) results were analyzed by log transformation of the normalized values and two-way ANOVA Tukey multiple comparisons tests. The IFN-γ ELISpot assays were all analyzed with an ordinary one-way ANOVA with Tukey multiple comparisons test. The DAMP measurements (ATP, HMGB1, and cytokines) were analyzed using a two-tailed Mann–Whitney nonparametric t test. PGE₂ ELISA and Ptgs1/2 RT-qPCR results were analyzed with one-way ANOVA test with Holm-Sidak multiple comparisons test. Significant p values are indicated as follows: *p < 0.0332, **p < 0.0021, ***p < 0.0002, and ****p < 0.0001. No exclusion criteria were used.

It has been reported that overexpression and oligomerization of FADD can result in necrotic cell death in conditions in which caspase-8 is absent or blocked (40). However, depending on the cell line used, FADD oligomerization also serves as a platform for apoptosis induction (41). The BALB/c mouse–derived colon carcinoma cell line CT26 is commonly used in ICD studies (17, 25, 42, 43), so we chose to stably transfect this cell line with a FADD-dimerizing construct (CT26 FADD). For this purpose, we used a Tet-ON doxy-inducible construct of Flag-tagged FADD coupled to two copies of FKBP (FK506-binding protein), which allows oligomerization of FADD upon the addition of a dimerizing compound (B/B) (Supplemental Fig. 1A). Upon costimulation with doxyBB, the cells undergo apoptotic cell death, which is associated with the activation of the apoptotic executioner caspases as measured by DEVD-AMC proteolysis and which could be inhibited by pretreatment with a pan-caspase inhibitor, zVAD-fmk (Supplemental Fig. 1B). We also confirmed, as previously partially shown (25), that molecular characteristics of apoptosis (44–47) were present in the cell lysates following doxyBB stimulation; these included cleaved caspase-8, cleaved caspase-3, and cleaved poly(ADP-ribose) polymerase (Supplemental Fig. 1C). By time-lapse microscopy, we observed clear morphological features of apoptosis in the form of apoptotic bodies (48) prior to plasma membrane permeabilization (Supplemental Fig. 1D, Supplemental Video 1). And finally, we detected complex IIb formation, which has been shown to consist of FADD DD interaction with RIPK1 and FADD death effector domain interaction with caspase-8 (45, 49–51) but only upon FADD oligomerization by B/B (Supplemental Fig. 1E, 1F). Therefore, we conclude that oligomerization of FADD induces apoptosis in CT26 cells.

Upon efficient ICD induction, cross-priming and activation of CTLs will prevent tumor growth on the challenge site of mice that received an ICD vaccination (6). We induced apoptosis in CT26 FADD cells and collected them after 18 h when the majority of cells showed an apoptotic or secondary necrotic phenotype (Fig. 1A). We injected the cells as a vaccine into syngeneic naive BALB/c mice, which 8 d later, were challenged with untreated CT26 WT cells on the opposite flank (Fig. 1B). Vaccination with 3 × 10⁶ FADD-apoptotic CT26 cells proved to be highly immunogenic. All the vaccinated mice, which were also tumor free on their vaccination site, were completely protected against tumor challenge on the opposite flank side (Fig. 1C). The immunogenicity of the FADD-apoptotic CT26 cells was confirmed in vitro in coculture experiments with BMDCs (52). BMDCs engulfed the apoptotic CT26 cells to a significantly higher degree than untreated CT26 FADD cells (Fig. 1D, 1E), and overnight coculture of apoptotic CT26 FADD cells with BMDCs resulted in an upregulation of DC activation markers CD80 and MHCII (53), whereas the untreated CT26 FADD cells had no significant influence on the BMDC activation status (Fig. 1F, 1G). The release and exposure of DAMPs such as HMGB1, ATP, and calreticulin have been proposed as crucial key factors for ICD (17, 42, 54, 55). We confirmed the release of the former two but did not find evidence for calreticulin exposure in our apoptotic model (Supplemental Fig. 1G–I).

Altogether, these results show that oligomerization of FADD overexpressed in the murine CT26 colon carcinoma cell line induces apoptosis as characterized by caspase-8 and -3 cleavage (25) and apoptotic blebbing. After cell death induction, coculture with BMDCs induces an efficient uptake of the apoptotic cells and upregulation of BMDC maturation and activation markers. Finally, the apoptotic cells are also immunogenic in an in vivo setting, in which they protect mice with a high efficiency against tumor challenge in a prophylactic tumor vaccination model using syngeneic BALB/c WT mice.

RIPK1 and NF-κB signaling have been reported to mediate cross-priming of CD8⁺ T cells in response to immunization with apoptotic mouse embryonic fibroblasts (29). Apoptosis induced by FADD oligomerization in CT26 cells caused an increase in the expression of several NF-κB–regulated inflammatory genes (Fig. 2A), many of which were similar to those induced by necroptosis in CT26 DDR3 cells (25). Previously, we have shown that DDR3 induced to undergo necroptosis, were equally immunogenic, both under conditions with pronounced (+B/B) and minor (−B/B) NF-κB activity (25). Among its many gene targets, NF-κB drives the transcription of two chemokines CXCL1 and CXCL2, of which we measured a substantial increase in the supernatant of mainly apoptotic but also necroptotic CT26 cells (Fig. 2B, 2C). To test the concept of NF-κB activity contributing to the immunogenicity of dying CT26 cells, we used a selective inhibitor of IκB kinase 2 (56) prior to cell death induction in CT26 cells and used these cells in the prophylactic antitumor vaccination model. TPCA-1 pretreatment efficiently inhibited NF-κB activity in both apoptotic (Fig. 2D) and necroptotic (Fig. 2E) conditions. However, in the cell death vaccination model, blocking NF-κB activity showed no effect on the immunogenic properties of neither apoptotic nor necroptotic cancer cells (Fig. 2F, 2G).

Altogether, these results show a significant NF-κB activation and NF-κB–induced gene transcription profile in both apoptotic CT26 FADD and necroptotic (doxyBB-induced) CT26 DDR3 cells. However, upon blocking NF-κB activity with TPCA-1 in the dying cells and using these cells in the prophylactic tumor vaccination model in vivo, we did not observe any change in antitumor immunogenicity, hence demonstrating that the mere apoptotic or necroptotic cells used for prophylactic vaccination are sufficient to mount an antitumor response.

The MuLV-derived glycoprotein, Gp70, and the Gp70-derived antigenic peptide, AH1, have been used in antitumor studies, in which they showed strong vaccination boost potential (33, 34, 57). Interestingly, Gp70 is also one of the most expressed genes in the entire CT26 transcriptome (16). Thus, we speculated that knocking out AH1 from the CT26 genome by means of CRISPR/Cas9 technology (Supplemental Fig. 2A) would affect the immunogenicity of CT26 cells. Following transfection of CT26 FADD and CT26 DDR3 cells with an AH1-targeting Cas9–sgRNA plasmid (Table I), we generated AH1-deficient clones and verified the KO efficiency by measuring AH1 transcript levels (Supplemental Fig. 2B) and indels in the AH1^KO clones (Supplemental Fig. 2C) (Table II). Although our original results showed that FADD-dimerized induction of apoptosis in CT26 cells gave rise to a strong immunogenic antitumor response (Fig. 1C), vaccination with 1.5 × 10⁶ apoptotic cells was no longer immunogenic in the absence of AH1 expression, whereas mice vaccinated with necroptotic AH1^KO cells were protected for a longer time period against a CT26 tumor challenge (Fig. 3A). Increasing the cell number used for vaccination to 3 × 10⁶ cells did not increase the immunogenicity of AH1-deficient apoptotic cells, which remained different from nonimmunized PBS-injected mice (Fig. 3B). Although the immunogenicity was lower in the necroptotic cells that did not express AH1 than in the parental necroptotic cell line, the tumor-free survival of mice that received a vaccination with AH1^KO necroptotic cells was significantly longer than those that received an apoptotic AH1^KO immunization (Fig. 3B). Gp70 and AH1 is expressed by a number of different mouse strains and murine cell lines (15, 58). To check the cross-reactivity of the CT26 prophylactic vaccinations, we immunized BALB/c WT mice with apoptotic (Fig. 3C) or necroptotic (Fig. 3D, 3E) CT26 cells expressing or not (AH1^KO) the immunodominant AH1 tumor Ag, but this time, we challenged the mice with the BALB/c-syngeneic breast cancer and AH1-expressing cell line 4T1. Apoptotic CT26 cells would efficiently protect mice against a 4T1 tumor challenge, but this tumor protection was solely dependent on AH1 expression in the dead cell immunization (Fig. 3C). Immunization with necroptotic CT26 cells induced by doxyBB (Fig. 3D) or doxy (Fig. 3E) also efficiently protected mice against 4T1 tumor challenge. Despite a slight difference in the level of tumor protection, AH1 expression did not seem to fully determine the immunogenicity of the necroptotic cells. Although the AH1-deficient necroptotic cells would induce a significantly stronger adaptive immune response than a cell-free PBS vaccine, a level of significant statistical difference was not reached when compared with the nonimmunogenic AH1^KO F/T vaccination (Fig. 3D, 3E). However, in the absence of AH1 expression, necroptotic doxy-induced CT26 DDR3 cells still protected significantly more mice against 4T1 tumor challenge than in the case of apoptotic CT26 FADD immunization (Fig. 3F).

Recently, it was demonstrated that the therapeutic immunogenic response to necroptosis can be driven by neoepitopes (59, 60). To look further into the efficacy of the inducible necroptotic versus apoptotic cancer cell vaccines, we tested the AH1-deficient CT26 cells for their capacity to elicit an immune response to a mix of CT26-specific neoepitopes, consisting of four mutated MHCII-binding peptides and one MHC class I (MHC I)–binding peptide (60). Vaccination with necroptotic doxyBB AH1-expressing CT26 cells induced an AH1-specific adaptive immune response as measured by the production of IFN-γ from the draining lymph nodes (Fig. 3G) or splenocytes (Fig. 3H) from immunized mice. Apoptotic CT26 cells did not induce significantly increased IFN-γ levels whether in the presence or absence of AH1 expression (Fig. 3G–I). When we stimulated lymphocytes with neoepitopes, only the lymph nodes isolated from mice that had received the doxy-induced necroptotic CT26 vaccination responded in the IFN-γ ELISpot assay (Fig. 3I). AH1 and the tested neoepitopes are presented on MHC I or MHCII molecules (14, 60); thus, to examine the contribution of specific immune cells to the immunogenicity of apoptotic and necroptotic CT26 cells, we vaccinated athymic BALB/c mice that lack the T lymphocyte compartment. Immunization with neither apoptotic nor necroptotic CT26 cells could protect mice that lacked T lymphocytes against a challenge with CT26 WT cells (Supplemental Fig. 2D–G). Athymic mice immunized with doxy-induced necroptotic CT26 DDR3 cells showed a slight delay in tumor growth on the challenge site compared with nonvaccinated mice and mice vaccinated with apoptotic CT26 cells (Supplemental Fig. 2F, 2G); however, eventually all the mice, irrespective of the vaccination, showed the first tumor growth signs within 8 d after tumor challenge.

Altogether, apoptotic CT26 FADD cells efficiently immunize BALB/c WT mice against challenge with two different AH1-expressing cancer cell lines, CT26 and 4T1, but this immunogenicity appears to rely solely on AH1 expression in the apoptotic cells used for vaccination. Vaccination with necroptotic CT26 DDR3 cells provides a stronger immunogenic phenotype as seen by the IFN-γ production from AH1- or neoepitope-stimulated immune cells. In the absence of AH1 expression, the necroptotic CT26 cells also show a more efficient vaccination potential than apoptotic cells against 4T1 tumor challenge.

Classically, apoptosis and caspase activation are linked with an immunosuppressive phenotype (61, 62). We therefore sought to investigate whether the AH1-deficient CT26 cells presented an immunosuppressive cytokine profile, which could explain their reduced immunogenicity and differences between immunogenic AH1^KO apoptotic and necroptotic cells. We were able to detect a mixture of immunosuppressive, proinflammatory, and chemoattractant cytokines in the supernatant of both apoptotic and necroptotic cells (Fig. 4A, Supplemental Table I). Albeit low levels, apoptotic cells secreted more IL-6 but also more CXCL1, CXCL2, and CXCL10 than their necroptotic counterparts, regardless of the AH1 expression status (Fig. 4B–E). Whereas IL-10 was produced in higher concentrations in necroptotic CT26 DDR3 cells, IFN-β secretion was higher in apoptotic CT26 FADD cells (Supplemental Fig. 3A, 3B), suggesting that there is no clear immunosuppressive cytokine profile linked to apoptotic cells in our models. Production of PGE₂ is driven by cyclooxygenases (COX) and is known to suppress immunity and support tumor growth (63); hence, we tested whether blocking COX activity would increase the immunogenicity of apoptotic and necroptotic CT26 tumor vaccines. For this purpose, we used ibuprofen, which is reported to inhibit both the constitutive active COX-1 and the inducible COX-2 (64, 65). PGE₂ secretion was not detectable in response to an apoptosis stimulus, but PGE₂ was secreted in high amounts by doxyBB-induced necroptotic DDR3 cells, and this secretion was efficiently blocked by pretreatment with ibuprofen (Fig. 4F). The PGE₂ secretion was likely to be due to COX2 activity rather than COX1 because we measured a similar increase in Ptgs2, but not Ptgs1, gene expression in the same doxyBB-induced necroptotic lysates (Supplemental Fig. 3C, 3D). Pretreatment with ibuprofen prior to cell death induction did not improve the vaccination potential of apoptotic cells (Fig. 4G). In the case of necroptosis, we also did not observe any difference in tumor protection following immunization with COX-inhibited necroptotic cells (Fig. 4H, 4I); hence, COX inhibition did not result in any significant changes in immunogenicity between apoptotic and necroptotic AH1-deficient CT26 cells (Fig. 4J).

Altogether, these results show that apoptotic CT26 cells, also in the absence of dominant AH1 tumor Ag expression, secrete a mixture of proinflammatory, immunosuppressive, and chemoattractant cytokines and, at times, in higher amounts than their necroptotic counterparts. Additionally, the decreased immunogenicity of AH1-deficient apoptotic CT26 cells cannot be attributed to COX activity in the dying cells.

In this study, we show that ligand-free inducible apoptosis in a murine colon carcinoma cell line can elicit an efficient antitumor immune response. Using a similar cell death induction strategy, we also show that necroptosis, unlike apoptosis, gives rise to an antitumor immunogenic response directed toward not only an endogenous TAA, namely AH1, but also to a mixture of CT26-specific neoepitopes. Our experimental approach includes a model of drug-independent apoptosis in CT26 cells initiated by overexpression and oligomerization of FADD, which induces classical caspase-dependent apoptosis and is associated with the release of DAMPs (ATP and HMGB1), NF-κB activation, and secretion of chemokines (CXCL1, -2, and -10). The apoptotic CT26 FADD cells are phagocytosed by BMDCs and lead to their phenotypic activation as measured by CD80 and MHCII cell surface expression. Immunizing mice with these apoptotic CT26 FADD cells, we observed a complete protection against a challenge with live tumor cells. This adaptive immunity response was independent of NF-κB activity in the dying cells but relied on the expression of an endogenous MuLV-derived tumor Ag, AH1. These results demonstrate that apoptosis as a cell death modality can be immunogenic when induced in a ligand-free manner but that in the absence of an immunodominant epitope, the immunogenicity of these dead cells is lost.

As a comparative cell death model to FADD-induced apoptosis, we used another inducible system, CT26 DDR3, in which necroptosis induction triggers an adaptive anti-AH1 tumor Ag immune response (25). Using these necroptotic CT26 DDR3 cells induced by either doxy (DD and RIPK3 overexpression) or doxyBB (additional oligomerization of DD) (25) in a prophylactic tumor vaccination model, we now show that the antitumor protective immune response was independent of NF-κB activity in the necroptotic cells. As opposed to the FADD-induced apoptosis, vaccination with necroptotic CT26 cells induced not only a strong AH1-specific adaptive immune response but also neoepitope-specific immunogenicity. The neoepitope-induced immune response was also present, albeit to a lower extent, upon vaccination with AH1-deficient necroptotic CT26 cells.

It has been reported that inducible necroptosis in OVA-expressing 3T3 fibroblast cells is more potent than apoptosis in activating an Ag-specific immune response and that cross-priming of T cells relies on RIPK1-driven NF-κB activity in necroptotic cells (29). Contrary to these findings, we were not able to see this same dependency on NF-κB activity in the immunogenicity of apoptotic and necroptotic cells because blocking NF-κB activity in the cancer cells with an inhibitor of IκB kinase 2 did not alter the antitumor immune response toward a challenge with live CT26 cancer cells. The dependency of antitumor efficiency and CTL cross-priming activity on RIPK3–RIPK1–NF-κB signaling in dying cancer cells has been questioned before by us and others (25, 66). Indeed, Ren et al. (66) showed that efficient activation of BMDCs and cross-priming of CTLs in response to necroptotic fibroblasts was achieved regardless of whether the inducible necroptosis construct engaged RIPK1 via the RIP homotypic interaction motif or not. Necroptosis induced by oligomerized RIPK3 constructs was demonstrated mainly in fibroblasts, and the immunogenic studies used OVA as a model tumor Ag (29, 66). Our study differs from these in the sense that we overexpress but do not force oligomerization of RIPK3 to induce necroptosis. Additionally, our study shows advances in the implication of endogenous TAA (AH1) and neoepitopes in a prophylactic tumor vaccination setting.

By definition, an adaptive immune response depends on the recognition of tumor Ags, and depending on the type of cancer, different tumor Ags or neoantigens have been identified (67). Previous studies reported the endogenously expressed retrovirus-derived tumor Ag in CT26 cells, Gp70, and its immunodominant epitope (AH1) to be recognized by CTLs when presented on MHC I molecules (14). The immunogenic role of AH1 has been further substantiated in different vaccination studies using recombinant AH1 or AH1-mimicking peptides (8, 33, 34, 57, 68), and Gp70 expression is not confined to CT26 cells only but is also found in other murine cell lines, such as 4T1, B16, and EL4 (9, 10, 15). Consolidating these findings with our immunogenic observations using dying CT26 cells as a prophylactic tumor vaccine led us to demonstrate that the high expression of the immunodominant AH1 was indispensable for the immunogenicity of apoptotic CT26 cells, whereas necroptotic CT26 cells could engage also neoepitopes. Vaccination with both AH1-proficient apoptotic and necroptotic CT26 cells protected against 4T1 tumor challenge in syngeneic BALB/c mice, but under AH1-deficient vaccination conditions, only necroptosis still showed the remaining protection against challenge with CT26 or 4T1 WT cancer cells. This demonstrates the immunodominant phenotype of AH1 expression in these murine cancer cell lines, especially in apoptotic conditions. In necroptotic conditions, cross-presentation of neoepitopes is still able to elicit a persisting antitumor response against CT26 cells. It is quite remarkable that vaccination with AH1-deficient necroptotic CT26 colon carcinoma cells are still able to prevent tumor growth of an unrelated 4T1 breast carcinoma cell line. It can be speculated that either other unknown Ags shared by CT26 and 4T1 are at stake and favored by necroptotic cell death or necroptotic in contrast to apoptotic tumor cells may be associated with elucidation of an Ag-independent antitumor response involving NK cells or innate lymphoid cells (69). This hypothesis is subject to further research.

Our work highlights the importance of immunodominant TAAs in the CT26 cancer cell line and demonstrates the antigenic contribution to ICD, which in addition to DAMPs, induce efficient antitumor immunity. The balance between adjuvanticity and antigenicity in ICD models is an issue that has been previously discussed in a therapeutic tumor treatment setting, in which reported chemotherapeutic ICD inducers showed no involvement of adaptive immune cells in the eradication of spontaneous tumor growth (70). However, our study now illustrates the impact of different intrinsically induced cell death modalities on the antigenicity of cancer cells in a prophylactic antitumor vaccination setting. Our results are in line with a recent report in which it was demonstrated that necroptotic CT26 cells, induced by tumor electroporation with an MLKL-encoding mRNA, gave rise to an efficient antitumor immune response directed against CT26-specific neoepitopes (59, 60). However, in that study, the response of apoptotic cells against the dominant AH1 Ag was not evaluated. Related to ICD adjuvanticity, caspase-3 activation and apoptosis induction by radiation therapy has been shown to stimulate the secretion of PGE₂ (71). Apoptotic caspases were also demonstrated to suppress type I IFN production (72, 73), and recently, it was reported that caspase-3 has an inhibitory effect on the immunogenicity of cancer cells that are dying in response to radiation therapy (74, 75). In our Tet-ON–inducible system, despite the NF-κB and caspase-3 activity measured in the FADD–apoptotic colon carcinoma cell model, we observed neither an increase in COX-2 gene expression nor secretion of PGE₂. We did, however, measure a slight increase in IFN-β production, but not IFN-α secretion, upon caspase-3 activation, but because treatment with a pan-caspase inhibitor (i.e., Z-VAD-fmk) would completely block apoptosis in these cells, we did not further evaluate whether the type I IFN response would become more profound upon caspase-3 inhibition or knockdown as recently published (74). Additionally, as opposed to the radiation-induced models that take days for apoptosis to occur (71, 74), our Tet-ON–inducible apoptosis model is an extremely fast system; it kills the cells within just a few hours following FADD induction and oligomerization by B/B administration. Hence, the added role of type I IFNs to immunogenicity in our system is probably neglectable, yet it is worth further investigation.

It can be speculated that, in the absence of AH1 expression, other (e.g., virally derived) endogenous tumor Ags expressed in CT26 cells account for the Ag-dependent immune responses. Indeed, in the CT26 cell line, other mutations have been identified that can lead to the formation of antigenic peptides, which are predicted to bind to MHC molecules and to be recognized by CD4⁺ and CD8⁺ T cells (16, 60). Identifying a pattern of antigenic epitopes that are induced under various cell death stimuli can become a highly valuable tool for future immunotherapeutic cancer treatments in the clinic. However, already at the discovery phase, our findings emphasize the importance of endogenous immunodominant Ag titers in cancer cell lines used for ICD and cancer immunology studies but also how the cell death modality determines the strength of the immune response to alternative neoepitopes.

We thank the VIB Bioimaging Core, especially Amanda Gonçalves, for assistance with the live cell imaging experiments, VIB Flow cytometry and FACS core, especially Gert Van Isterdael and Hans Demol for synthesizing peptides, Philippe De Groote for making the lentiviral vectors, Francis Santens for helping design the CRISPR/Cas9 AH1-targeting primers, Susanne Løve Aaes-Jørgensen for excellent editorial assistance, and Stefan De Koker, Lien Van Hoecke, and Xavier Saelens for helpful discussions.

The authors have no financial conflicts of interest.