Double-stranded RNA directly acts on fibroblast and myeloid lineages to induce necroptosis as in TNFα. Here, we investigated whether this type of cell death occurred in cancer cells in response to polyinosinic–polycytidylic acid (polyI:C) and the pan-caspase inhibitor z-Val-Ala-Asp fluromethyl ketone (zVAD). We found that the colon cancer cell line CT26 is highly susceptible to necroptosis, as revealed by staining with annexin V/propidium iodide. CT26 cells possess RNA sensors, TLR3 and MDA5, which are upregulated by interferon (IFN)-inducing pathways and linked to receptor-interacting protein kinase (RIP) 1/3 activation via TICAM-1 or MAVS adaptor, respectively. Although exogenously added polyI:C alone marginally induced necroptosis in CT26 cells, a combined regimen of polyI:C and zVAD induced approximately 50% CT26 necroptosis in vitro without secondary effects of TNFα or type I IFNs. CT26 necroptosis depended on the TLR3–TICAM-1–RIP3 axis in the tumor cells to produce reactive oxygen species, but not on MDA5, MAVS, or the caspases/inflammasome activation. However, the RNA-derived necroptosis was barely reproduced in vivo in a CT26 tumor–implanted Balb/c mouse model with administration of polyI:C + zVAD. Significant shrinkage of CT26 tumors was revealed only when polyI:C (100 μg) was injected intraperitoneally and zVAD (1 mg) subcutaneously into tumor-bearing mice that were depleted of cytotoxic T lymphocytes and natural killer cells. The results were confirmed with immune-compromised mice with no lymphocytes. Although necroptosis-induced tumor growth retardation appears mechanistically complicated and dependent on the injection routes of polyI:C and zVAD, anti-caspase reagent directed to tumor cells will make RNA adjuvant immunotherapy more effective by modulating the formation of the tumoricidal microenvironment and dendritic cell–inducing antitumor immune system. Cancer Immunol Res; 3(8); 902–14. ©2015 AACR.
A cell death response frequently occurs in malignantly transformed cells accompanied with infection or inflammation. Viral infection also induces programmed cell death, which is believed to be a host defense to restrict viral spread (1). However, recent advance in cell death studies may offer an alternative interpretation of cell death in tumor cell biology: The products of dead cells profoundly modulate the immune system and microenvironment around the established tumor, which is largely relied on the mode of cell death (2, 3). Programmed cell death was categorized as apoptosis, pyroptosis, and necroptosis, based on the difference of caspases involved (4). In general, apoptosis is induced by activation of caspase-8 or caspase-9. Pyroptosis is a form of caspase-dependent cell death initiated by the activation of caspase-1 or caspase-11 (4). Although necroptosis is usually triggered by death receptors, including Fas receptor (FasR), tumor necrosis factor (TNF) receptor, and TNF-related apoptosis-inducing ligand receptor, Toll-like receptors (TLR) can also induce necroptosis when caspase-8 is inhibited by gene depletion or pharmacologic inhibitors (4). Caspase inhibitors, such as z-Val-Ala-Asp fluoromethyl ketone (zVAD), block TNF-induced apoptosis in many cell lines, whereas some cell lines respond to TNF+zVAD by activating necroptosis pathways (5). Necroptosis, but not apoptosis, is believed to cause modulation of the tumor microenvironment (TME) that affects tumor progression and invasion (6). Exosomes, proteins, and nucleic acids, released from dying cells, may be an important extracellular source of the environmental effectors (7). Double-stranded (ds) and stem-structured RNAs representing host and viral patterns of innate immunity also serve as modifiers for inflammatory environment and host immune response (8–10).
Polyinosinic–polycytidylic acid (polyI:C), a synthetic analogue of dsRNA, has been known to have direct cytolytic activity on fibroblast and macrophage lineages. Myeloid cells in tumor are highly susceptible to dsRNA compared with normal cells (7, 11), whereas antigen-presenting dendritic cells (DC) mature in response to dsRNA followed by immune activation (8, 10), suggesting the presence of a cell type–specific RNA-sensing machinery that determines the life or death fate in the cell. PolyI:C appear to trigger both apoptosis and necroptosis (8, 10). In vitro studies on innate immunity suggested that several signaling pathways could be involved in polyI:C–derived cell death, although there are cell type–dependent variations in the resultant cell death (12, 13). However, in tumor cells, the signal that induces cell death by polyI:C remains largely undetermined in in vivo models.
PolyI:C is a ligand for both endosomal TLR3 and cytoplasmic melanoma differentiation-associated protein-5 (MDA5) and induces the activation of NF-κB and interferon regulatory factor (IRF) 3 transcription factors followed by production of inflammatory cytokines and type I/III interferons (IFN; refs. 8, 14). TLR3 and MDA5 are upregulated by polyI:C or IFN stimulation, suggesting that they are IRF3- and IFN-inducible factors (14). MDA5 and TLR3 recruit different adaptors, mitochondrial antiviral signaling protein (MAVS), or Toll-interleukin 1 receptor domain (TIR)–containing adaptor molecule (TICAM)-1, respectively (15, 16), which confers distinct functional properties on the two pathways. In response to polyI:C, TLR3/TICAM-1 activates IRF3 as well as receptor-interacting protein kinase (RIP) 1/3 in a cell type–dependent manner (17). Upon malignant transformation, cells usually express high levels of TLR3 and MDA5, which sense polyI:C and initiate RNA-sensing signals that are sometimes linked to cell death or live output, including IFN/cytokine production (13, 18). What factors discriminate between the death and live signal is yet unknown, but both IRF3-derived and IFN receptor (IFNAR)–derived cell death have been reported (13, 18).
Here, we found that polyI:C and zVAD induced cell death in the mouse colon carcinoma cell line CT26. Because the dead cells were stained propidium iodide (PI)/annexin V–double positive and a necroptosis inhibitor (necrostatin, nec-1) blocked the cell death (12), we concluded that the polyI:C/zVAD–induced CT26 cell death was necroptosis. This form of cell death was abrogated in CT26 cells after depletion of Ticam-1 or Ripk3 or treatment with nec-1. In contrast, Mavs knockdown barely affected tumor cell death. Notably, blocking IFNAR or TNFα hardly affected the degree of polyI:C–induced CT26 necroptosis. Thus, necroptosis was induced in CT26 colon cancer cells in vitro directly by polyI:C and zVAD through the TLR3–TICAM-1–RIP3 pathway, independent of IFN or TNFα. Neither the cytoplasmic RNA-sensor pathway (15) nor the TICAM-1–mediated inflammasome-caspase activation (19) participates in this type of tumor necroptosis.
In wild-type mice, polyI:C adjuvancy promotes cross-priming of CD8+ T lymphocytes, activation of natural killer (NK) cells, and IFN/cytokine production by DCs (14, 20). We detected NK/cytotoxic T lymphocyte (CTL)/cytokine-independent tumor-necroptotic shrinkage in CT26-bearing immune-compromised Balb/c mice by injection of polyI:C and zVAD in vivo.
Materials and Methods
Cell culture and reagents
Cell lines EG7 (lymphoma) and C1498 (acute myeloid leukemia) were obtained from the American Type Culture Collection; 3LL (Lewis lung carcinoma), YAC-1 (lymphoma), and colon26 (CT26, colon carcinoma) from Summit Pharmaceuticals International Corporation; L929 (fibroblast) from RIKEN Cell Bank; EL4 (lymphoma) from Dr. N. Sato (Sapporo Medical University School, Sapporo, Japan); B16F1 (melanoma) and B16F10 (melanoma) from Dr. O. Hazeki (Hiroshima University, Hiroshima, Japan); G1 (hepatocellular carcinoma) and G5 (hepatocellular carcinoma) from Dr. Y. Saeki (Osaka Medical Center for Cancer and Cardiovascular Diseases, Osaka, Japan); Renca (renal adenocarcinoma) from Dr. Y. Matsushita (Iwate Medical University School of Medicine, Iwate, Japan); and MC38 (colon adenocarcinoma) from Dr. H. Tahara (University of Pittsburgh Medical Center, Pittsburgh, PA). A B16 subline, B16D8, was characterized as NK sensitive in our laboratory (21). All cell lines were confirmed as Mycoplasma free.
YAC-1, EL4, B16F1, B16F10, B16D8, Renca, G1, G5, 3LL, L929, EG-7, C1498, and CT26 cells were maintained in RPMI-1640 supplemented with 10% heat-inactivated fetal bovine serum and antibiotics. MC38 cells were maintained in RPMI-1640 supplemented with 10% heat-inactivated fetal bovine serum, antibiotics, 2 mmol/L glutamine, 50 μmol/L 2-mercaptoethanol, 1 mmol/L sodium pyruvate, and nonessential amino acid. Necroptosis-resistant CT26 cells were cells that survived after stimulation with 25 μg/mL polyI:C and 25 μmol/L zVAD, and maintained in RPMI-1640 supplemented with 10% heat-inactivated fetal bovine serum and antibiotics. For induction of bone marrow–derived dendritic cells (BMDC), bone marrow cells from C57B6/J WT mice were cultured in RPMI-1640 with 10% heat-inactivated fetal bovine serum and antibiotics containing J558 supernatant for 7 days with medium replenished every other day.
Antibodies used were: anti-RIP1 (BD Biosciences), anti-RIP3 (QED Bioscience), anti-FLAG monoclonal (Sigma), anti-FLAG polyclonal (Sigma), anti-tubulin (BioLegend), anti-HA monoclonal (Covance), anti-β-actin (Sigma), allophycocyanin (APC) anti-TLR3 (BioLegend), and fluorescein isothiocyanate (FITC) anti-TLR4 (MBL) antibodies. PolyI:C was purchased from Amersham Biosciences, zVAD, butylated hydroxyanisole (BHA), and nec-1 were from Sigma. Anti-IFNAR and anti-TNFα antibodies were from BioLegend.
Balb/c AJcl and C57B6/J WT female mice were purchased from CLEA Japan. Rag-2/Jak3 double-KO mice in Balb/c background (22) were housed and monitored in our animal research facility according to institutional guidelines. All mice were maintained under specific pathogen-free conditions in the Animal Facility in Hokkaido University Graduate School of Medicine (Sapporo, Japan) and used when they were 7 to 9 weeks of age. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH (Bethesda, MD). The protocol was approved by the Committee on the Ethics of Animal Experiments in the Animal Safety Center, Hokkaido University (Hokkaido, Japan). All mice were used according to the guidelines of the Institutional Animal Care and Use Committee of Hokkaido University, approval no. 13-0043. All efforts were made to minimize suffering.
Tumor challenge and antibody treatment
Mice were shaved at the back and injected 200 μL of 1 × 105 or 5 × 105 CT26 cells in PBS. Tumor size was measured using caliper. Tumor volume was calculated using the following formula: tumor volume (cm3) = (long diameter) × (short diameter)2 × 0.4. PolyI:C (usually 100 μg) was injected into mice intraperitoneally (i.p.) or subcutaneously (s.c.), whereas zVAD (1 mg) was s.c. injected per mouse. For depletion of CD8+ T cells and NK cells, anti-murine CD8β antibody prepared from H35.17.2 hybridoma that was kindly provided by Dr. Toshitada Takahashi (Aichi Cancer Center, Aichi, Japan; ref. 23) and anti-asialo-GM1 antibody (WAKO) were i.p. injected before polyI:C treatment (21). Optimal doses of the antibodies were determined in preliminary studies (anti–asialo-GM1 30 μL/body), and the same lot of anti-CD8β antibody was used in the experiment.
Water-soluble tetrazolium salts-1 assay
A water-soluble tetrazolium (WST)-1 Cell Counting kit (Dojindo) was used following the manufacturer's instructions. Cells (2 × 104) were plated in a 96-well plate. The following day, cells were stimulated with 25 or 50 μg/mL of polyI:C, 25 μmol/L of zVAD, and 50 μmol/L of nec-1. After 24 hours, 10 μL of WST-1 reagent was added to each well and incubated at 37°C for 1 to 4 hours. The absorbance at 450 nm was measured by a microplate reader.
Detection of reactive oxygen species
CT26 cells (2 × 104) were plated in a 96-well plate. The following day, cells were stimulated with 50 μg/mL of polyI:C, 25 μmol/L of zVAD, 100 μmol/L of BHA, and 50 μmol/L of nec-1. After 6 hours, CT26 cells were incubated with 5 μmol/L CM-H2DCFDA (Invitrogen) at 37°C for 15 minutes. Cells were washed with culture medium and incubated at 37°C for 15 minutes. Cells were washed with FACS buffer and analyzed by flow cytometry.
Measurement of high-mobility group protein B1 (HMGB1)
An HMGB1 ELISA kit (Shino test) was used as per the manufacturer's instructions. CT26 cells (2 × 104) were plated in a 96-well plate. The following day, cells were stimulated with 25 μg/mL of polyI:C, 25 μg/mL of zVAD, and 50 μmol/L of nec-1. After 24 hours, HMGB1 in culture supernatants was quantified with the HMGB1 ELISA kit (Shino test) following the manufacturer's instructions.
Cytometric bead array assay and ELISA
The production of cytokines was measured by a cytometric bead array (CBA) assay (BD Biosciences). Culture supernatants or sera were incubated with capture beads for 1 hour at room temperature following incubation with phycoerythrin (PE)-labeled detection reagents. The intensity of beads bound to cytokines was detected by flow cytometry. Data analysis was performed by FCAP Array Software. Culture supernatants of CT26 cells or concanavalin A–stimulated splenocytes (5 × 105) were analyzed for IFNγ levels using ELISA. An IFNγ ELISA kit was purchased from eBiosciences. The assay was performed according to the manufacturer's instructions. IFNβ levels in sera were measured by a mouse IFN beta ELSA kit (PBL) following the manufacturer's instructions.
Tumors isolated from CT26-bearing Rag2−/−/Jak3−/− mice were fixed with 4% paraformaldehyde/PBS for 30 minutes at 4°C. Fixed tissues were impregnated with 15% sucrose/PBS for 2 hours following 30% sucrose/PBS for overnight at 4°C. Tissues were then embedded in O.T.C. compound (Sakura Finetek Japan), and the frozen tissue blocks were sectioned by using a cryotome (LEICA CM1850). Terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) staining of frozen sections was performed using an in situ cell death detection kit (Roche) following the manufacturer's instructions. Stained sections were monitored at ×20 or ×40 magnification using LSM510 META microscopy (Zeiss).
FACS analysis of dead cells
For the detection of dead cells, 1 × 105 CT26 cells were plated in 24-well plates. The following day, cells were stimulated with 25 or 50 μg/mL of polyI:C, 25 μmol/L of zVAD, and 50 μmol/L of nec-1. After 24 hours, cells were stained with PI and annexin V–FLOUS staining kit (Roche) following the manufacturer's instructions. Cells were stained by intracellular staining with APC anti-TLR3 (11F8), FITC anti-TLR4 (UT49), or isotype control antibody with or without permeabilization. Stained cells were analyzed by flow cytometry.
Necroptosis induced by polyI:C and zVAD
Necoptosis is induced in macrophages by TLR stimulation when caspase-8 is inhibited by caspase inhibitors, such as zVAD (4). To determine whether polyI:C or combination of polyI:C/zVAD could induce cell death in tumor cell lines, we added these reagents to the culture of mouse tumor cell lines, including YAC-1, EL-4, B16F1, B16F10, B16D8, Renca, G1, G5, 3LL, MC38, and CT26. The L929 fibroblast cell line was used as a positive control for cell death (24). Cell viability was measured by the WST-1 assay (Supplementary Table S2). Cell death was detected after polyI:C and zVAD treatment only in the CT26 cell line of all the tumor cell lines tested. A negligible level of cell death was detected by treatment with either polyI:C or zVAD alone (Fig. 1A and B). PolyI:C/zVAD–dependent cell death was observed in the CT26 as well as the control L929 cell line (Supplementary Table S2), suggesting that polyI:C–induced factors other than caspases are involved in polyI:C–derived cell death in most tumor lines. Although polyI:C activates NALP3 inflammasome via TICAM-1 and caspase-11 in bacterial infections (19), no zVAD-affecting factor participates in caspase-11 or IL1β levels in CT26 cells (Supplementary Fig. S1).
Comparing CT26 cells treated with polyI:C and zVAD with CT26 cells treated with polyI:C alone, the population of cells stained annexin V–positive (10.5% vs. 0.3%) and double-positive for annexin V and PI (23.4% vs. 3.7%) were increased by the addition of zVAD to polyI:C (Fig. 1C). Although dead cells were stained with PI (Fig. 1D), the rate of single positive for annexin V was increased after polyI:C and zVAD treatment (Fig. 1C), indicating that polyI:C/zVAD–induced cell death was involved, in part, in apoptosis. HMGB1, a necrosis marker, was produced by polyI:C/zVAD stimulation (Fig. 1E). Hence, both necrosis and apoptosis are induced in CT26 cells by treatment with polyI:C independent of caspases.
The properties of CT26 necroptosis induced by polyI:C and zVAD
To further analyze the features of cell death after polyI:C and zVAD stimulation, we examined whether cell death induced by polyI:C and zVAD could be prevented by treatment with nec-1, an inhibitor of RIP1 kinase in the necroptosis pathway. Consistent with previous observations (25), when 50% of cell death was induced by polyI:C and zVAD, nec-1 treatment fully restored cell viability (Fig. 2A and B). The population of PI-positive cells was reduced with nec-1 treatment (Fig. 2C). Moreover, treatment with nec-1 reduced the population of cells that were double positive for annexin V and PI (Fig. 2D). By imaging analysis with PI, we confirmed that nec-1 inhibited cell death (Fig. 2E). Initiation of pyroptosis is mediated by caspase-11 expression that is dependent on TLR4–TICAM-1 (26), but neither expression of caspase-11, production of IL1β (Supplementary Fig. S1), nor TLR4 protein was induced by treatment with polyI:C and zVAD in CT26 cells (the latter compared with control JAWSII cells; Supplementary Fig. S2). No evidence thus far endorses that cell death induced by polyI:C and zVAD involves NALP3-inflammasome- or caspase-mediated pyroptosis.
Because reactive oxygen species (ROS) generation is a major inducer of necroptosis, we examined the possibility that ROS being involved in polyI:C/zVAD–derived necroptosis. ROS generation was induced in CT26 cells by polyI:C and zVAD treatment (Fig. 2F). Nec-1 treatment suppressed the ROS generation induced by polyI:C and zVAD (Fig. 2F). To investigate the effect of ROS generation on cell death induced by polyI:C and zVAD, CT26 cells were pretreated with BHA, a ROS scavenger. Both cell death and ROS generation induced by polyI:C and zVAD were suppressed with BHA (Fig. 2F and G). These data indicate that ROS production is crucial for necroptosis triggered by polyI:C and zVAD in tumor cells.
Type I IFN, IFNγ, and TNFα are inducers of necroptosis (27, 28). To examine the participation of type I IFN and TNFα in necroptosis induced by polyI:C and zVAD, CT26 cells were pretreated with anti-IFNAR antibody or anti-TNFα antibody (Fig. 3A). Blocking IFNAR or TNFα signaling by specific antibody did not affect cell viability (Fig. 3A). The levels of TNFα protein in the culture supernatant of stimulated CT26 cells were below the detection limit (Fig. 3B). CT26 cells were resistant to IFNγ-induced cell death (Fig. 3C) and they barely produced IFNγ following polyI:C treatment (Fig. 3D).
The signaling pathway for necroptosis in CT26 cells
To determine the pathway involved in polyI:C– and zVAD-induced necroptosis, genes encoding molecules that might participate in polyI:C recognition were silenced by siRNA in CT26 cells with knockdown efficiencies higher than 50% (Fig. 4A). Tlr3 or Ticam-1 knockdown resulted in a recovery of cells from necroptosis induced by polyI:C and zVAD (Fig. 4B). RIPK3 is the key molecule that interacts with mixed lineage kinase domain-like protein (MLKL) and RIP1 to transmit necroptosis signal (25, 28). Cell viability was recovered in Ripk3 knockdown cells (Fig. 4B). Mavs knockdown did not affect cell viability in polyI:C/zVAD–induced cell death (Fig. 4B). Collectively, necroptosis initiated by polyI:C and zVAD critically depends on the TLR3–TICAM-1–RIP3 pathway. In confirmation experiments, CT26 cells expressed TLR3 protein (Fig. 4C) and Tlr3 knockdown cells did not induce Ifn-β mRNA in response to polyI:C (Fig. 4D), suggesting that TLR3 mainly signals the presence of polyI:C through the TICAM-1 pathway in CT26 cells.
To identify the molecular mechanism of necroptosis induced by polyI:C and zVAD, we examined the physical interaction among TICAM-1, RIP1, and RIP3 in cells stimulated with polyI:C. CT26 cells were transfected with HA-tagged TICAM-1 and FLAG-tagged RIP3. Immunoprecipitation assays using anti-FLAG antibody after 24 hours of transfection showed that RIP3 interacted with TICAM-1 but not RIP1 in steady state conditions (Fig. 5A). Upon polyI:C stimulation, RIP3 interacted with both TICAM-1 and RIP1 in CT26 cells (Fig. 5A). However, in necroptosis-resistant B16D8 cells, RIP3 did not bind TICAM-1 after polyI:C stimulation, and little necroptosis was induced after treatment with polyI:C and zVAD (Fig. 5B and Supplementary Table S1). We sometimes found a minute amount of TICAM-1–RIP3 interaction in the absence of polyI:C stimulation in B16D8 cells (Fig. 5B), although the reason is as yet unknown. Even when the RIP3 protein was overexpressed, extrinsic RIP3 did not interact with TICAM-1 in B16D8 cells stimulated with polyI:C (Fig. 5B), and cell death was not induced by polyI:C and zVAD in RIP3-overexpressing cells (Supplementary Fig. S3).
To identify the molecules that define sensitivity to cell death induced by polyI:C and zVAD, we established necroptosis-resistant CT26 cells by culturing in medium with polyI:C and zVAD (Fig. 5C). Expression of RIP3 was decreased in necroptosis-resistant CT26 cells (Fig. 5D and E). The expression levels of negative regulatory molecules of necroptosis, including cellular FLICE (FADD-like IL1β-converting enzyme)–inhibitory protein (cFILP)s, cFLIPL, and inhibitor of apoptosis (cIAP; ref. 4), remain unchanged between parent and death-resistant CT26 cells (Supplementary Fig. S4). Expression of 5-azacytidine–induced 2 (Azi2), dynamin-related protein (Dnm1l), Mlkl and phosphoglycerate mutase family member 5 (Pgam5), which are critical necroptosis factors downstream of RIP1 and RIP3 (4), was unaltered in necroptosis-resistant CT26 cells (Fig. 5D). RIP3-mediated cell death was observed with mouse bone marrow–derived macrophages (BMDM; ref. 25) and L929 (24), but not with EL4 (Supplementary Table S2). Tlr3 mRNA and Ifn-β mRNA expressions were not induced in EL4 cells in response to polyI:C (Takaki, Unpublished Data), suggesting unresponsiveness of our EL4 cells to polyI:C, the mechanism of which remains to be determined. Ripk3 mRNA was hardly expressed in B16D8, B16F10, 3LL, MC38, C1498, and Renca cell lines that were resistant to cell death by polyI:C and zVAD (Fig. 5F and Supplementary Table S2). Thus, the interaction between TICAM-1 and RIP3 is a regulatory step in polyI:C–induced necroptosis in some types of tumor cells.
Tumor shrinkage induced by polyI:C/zVAD in CT26-implanted mice
To investigate whether polyI:C/zVAD–induced necroptosis was involved in tumor retraction in vivo, CT26 cells were implanted in Balb/c mice. Constant tumor growth was observed, as expected (Fig. 6A). PolyI:C i.p. injection effectively and dose dependently suppressed tumor growth (Fig. 6A). Injection of 100 μg polyI:C resulted in >80% regression by day 20 of CT26-implanted tumors compared with injection of a PBS control, and tumoricidal action of 1 mg zVAD only was negligible (data not shown). Tumor regression by polyI:C treatment was partially abrogated when NK cells or CD8+ T cells were depleted by anti-asialo GM1 or CD8β-specific antibodies (Fig. 6B), suggesting that tumor growth retardation by polyI:C was mediated by NK cells and CD8+ T cells. Of note, DCs did not directly induce cytotoxicity in CT26 cells (Supplementary Fig. S5). On the basis of these results, we assessed the effect of zVAD on polyI:C–treated, NK/CD8β-depleted, CT26-bearing mice. Mice were s.c. inoculated with 1 mg zVAD, which suppresses death receptor–mediated liver injury in vivo (29). Injection with s.c. zVAD and i.p. polyI:C resulted in tumor regression in immune effector–depleted mice (Fig. 6C). In the presence of CD8 T and NK cells, the zVAD effect was less than that expected from the in vitro results. Therapeutic use of polyI:C promotes DC maturation but combination with zVAD tends to decrease the viability of BMDMs (25). Subcutaneous administration of polyI:C alone effectively induced growth retardation of CT26 tumors, but s.c. administration of polyI:C together with zVAD hampered the antitumor activity in NK/CD8 T cell–depleted tumor-bearing mice (Fig. 6D). In our setting (Fig. 6D), NK and CD8 T cells preferentially kill the tumor cell population that can be targeted by polyI:C/zVAD. Although the reason why the polyI:C/zVAD tumor suppression is abrogated by s.c. administration is unknown, these reagents might target epidermal myeloid and fibroblastic cells in this route, which alters the TME to barely promote tumor retardation (Fig. 6D).
To confirm the antitumor effect of zVAD and polyI:C in the absence of immune cells, CT26-bearing Rag2−/−/Jak3−/− mice, which lacks T, B, and NK cells (22), were injected with i.p. polyI:C and s.c. zVAD (Fig. 7A). Although injection of polyI:C alone did not decrease tumor volume, additional s.c. zVAD treatment significantly suppressed tumor growth (Fig. 7A). Serum IFNβ and TNFα levels in polyI:C/zVAD–treated mice were comparable with those treated with polyI:C alone (Fig. 7B). IFNγ was not detected in sera of mice treated with polyI:C or polyI:C/zVAD (Fig. 7B), indicating that IFNβ, TNFα, or IFNγ is dispensable for tumor retardation in vivo as is in vitro. Upregulation of Ripk3, Ripk1, Mlkl mRNA, which is a marker of necroptosis in vivo (30), was observed in tumors prepared from polyI:C/zVAD–treated mice as compared with those of polyI:C alone (Fig. 7C). TUNEL-positive cells also increased in tumors in polyI:C/zVAD–treated mice as compared with those treated with polyI:C alone (Fig. 7D and Supplementary Fig. S6). Taken together, polyI:C/zVAD treatment induces cell death in CT26 tumors in mice without participation of immune cells, resulting in tumor retardation.
Here, we demonstrate that administration of a TLR3 agonist and pan-caspase inhibitor, zVAD, results in tumor regression in mice secondary to tumor cytolysis. Notably, direct action of these reagents on tumor cells induces a tumoricidal event. In the literature, activation of caspases usually accelerates programmed cell death, and prior polyI:C–mediated priming of TICAM-1 is crucial in promoting caspase-mediated inflammasome activation in lipopolysaccharide (LPS) signaling (19, 26). In contrast with the TICAM-1–inflammasome axis, which involves caspase-11 and effector caspases-3 and -7 (19, 31), the necrotic cell death observed in fibroblasts and macrophages is mainly induced through the TICAM-1–RIP3 pathway that involves no caspases (13): This pathway is activated in the absence of caspase-8 activity. This cell death process fits the definition of necroptosis, in which cell death is RIR1/3-dependent and can be inhibited by nec-1, and based on the production of ROS (4). We found this type of tumor death evidenced in the CT26 colon cancer cell line. Because similar RIP1/3-mediated necroptosis has been reported with neuroblastoma cell lines, most of which are caspase-8 deficient (32), the RIP1/3-mediated tumor shrinkage we observed is not an isolated phenomenon in tumors.
In CT26 necroptosis induction, the TLR3–TICAM-1 pathway plays a pivotal role without the involvement of TNFα and IFNs in the process of RIP3 activation followed by cytolysis. Thus, exogenously added polyI:C acts on TLR3 and, together with zVAD, induces necroptosis in tumor cells. Exogenous polyI:C activates not only TLR3 but also MDA5, as shown by the IFNβ reporter assay (33): Type I IFN represents an output of the live signal induced by RNA sensors. Virtually, no upregulation of IL1β is detected in CT26 cells by stimulation with polyI:C/zVAD (Supplementary Fig. S1B), suggesting that inflammasome activation is again dispensable for the CT26 cell death. Upon polyI:C stimulation, on the other hand, TICAM-1 immediately and transiently interacts with RIP1 and RIP3 to initiate necroptosis signaling, resulting in the production of ROS in CT26 cells. Although we cannot yet define what mechanism is responsible for the switch from live to death signal in tumor cells by RNA stimulation, ROS production might reflect mitochondrial oxidative stress induced by the TICAM-1 signal, leading to the PGAM5–DRP-1 axis (4).
The function of the TICAM-1–RIP3 pathway has been shown to be involved in the activation of the GTPase DRP1 that is translocated from cytosol to mitochondria to drive mitochondrial damage in macrophages, which markedly modifies inflammation (34). Because phosphorylation controls the RIP3–DRP1 activation, the phosphatase–kinase balance in tumor cells needs to be further investigated.
In CT26 tumor–bearing mice, tumor growth is abrogated with i.p. injection of polyI:C, which causes activation of antitumor NK cells and CD8+ CTLs (21, 23). These immune effectors must be depleted in mice in order to detect an alternative mode of tumor growth retardation by the direct action of polyI:C and zVAD on tumor cells. There appears no additive effect on immune activation and direct tumor killing induced by polyI:C and zVAD in vivo, suggesting that the cell death system supports the immune system in tumor clearance. Notably, the direct tumoricidal activity by polyI:C/zVAD can be observed only when polyI:C is administered by the i.p. route. We surmised that s.c. injection of polyI:C and zVAD directly affected the viability of skin fibroblasts and Langerhans cells; the decreased viability of these lineages after polyI:C and zVAD treatment is consistent with that reported in the literature (25). Taken together, tumor necroptosis induced by polyI:C makes a small contribution to CT26 growth suppression in wild-type Balb/c mice with sufficient immune effectors in vivo. Yet, the tumor necroptosis activity and tumor regression by polyI:C/zVAD become evident in Rag2−/−/Jak3−/− double-deficient mice (Fig. 7). This study clarifies that the RNA-induced tumor clearance system works in tumors, which is engaged in necroptosis but is independent of the immune effectors or IFN/cytokines.
Although necrotic cells prepared by freeze–thaw cycles, formaldehyde fixation, or osmotic shock provoke no protective immune response in a tumor vaccination model (35–37), heat-killed necrotic cells stimulate antigen-presenting cells to increase production of IL12 and TNFα (38). The inconsistent results among studies might be explained by the differences in the composition and properties of damage-associated molecular patterns (DAMP)–containing RNA released from necrotic cells that depends on type of stimulation, resulting in induction of diverse immune responses. In this context, self-RNA with incomplete stems that activates TLR3 (39) would be involved in inflammation-mediated tumor cell death.
Although RIP1 and RIP3 are required to initiate necroptotic signaling, necroptosis is induced in the absence of RIP1 (40, 41), which has been reported to protect some tissue/organ cells from necroptosis, reflecting the complex arrays of necroptosis. Similar to L929 cells, CT26 cells can be sensitized by polyI:C to induce necroptosis in the presence of zVAD (24). However, this result cannot always be generalized for the other tumor cell lines (Supplementary Table S2 and Fig. 5F). The expression levels of RIP3 is lower in B16D8 cells (resistant to the polyI:C–induced cell death) than in wild-type CT26 and L929 cells (Fig. 5F), in which necroptosis can be induced (24). The RIP3 protein expression levels are clearly different between necroptosis-resistant CT26 cells and the parent CT26 cells (Fig. 5B). In L929 cells, which are used for in vitro necroptosis studies (24), the Ripk3 mRNA level is high as well as in CT26 and EL4 cells. Except for EL4, these RIP3 profiles imply that RIP3 expression correlates with sensitivity to polyI:C–induced necroptosis in most tumor lines.
However, simply overexpressing RIP3 is insufficient to trigger necroptosis: An additional phosphorylation or positive regulator plays a key role in inducing cell death. Recently, sirtuin-2 (SIRT2), MLKL, and PGAM5 were reported to positively regulate the TNF necroptotic pathway (4, 28, 42). MLKL interacts with RIP3 to trigger necrosis in fibroblasts stimulated with TLR ligands (25). The Pgam5 and Mlkl mRNA levels are unchanged between parental and resistant CT26 cells (Fig. 5D). Therefore, tumor necroptotic death represents an output produced by complex signaling involving RNA sensors.
PolyI:C, a synthetic TLR3 agonist, is used as an effective adjuvant for antitumor treatment and vaccines because of its prominent effects on DCs to induce CD8+ T and NK cells (21, 23). In addition, stem-structured RNA from viral replicative intermediates can act as a TLR3-specific ligand (8, 9). A GpC-capped dsRNA named ARNAX, which exclusively targets TLR3, potently induces cross-presentation in CD8α+ DCs without a significant increase of serum cytokines (43). CD8α+ DCs, which have high expression of TLR3, are activated by i.p. or s.c. injection of these TLR3 ligands to promote cross-priming of T cells in a TLR3/TICAM-1–dependent manner (23). This treatment also induces antitumor NK activation through induction of a polyI:C–inducible gene, INAM (IRF3-dependent NK-activating molecule), in DCs (44, 45). Moreover, polyI:C injection induces production of type I IFNs via the MAVS pathway in stromal cells, which suppresses tumor growth (46). However, the polyI:C/zVAD tumor regression was DC unrelated. Recently, polyI:C was found to act on tumor-associated macrophages to facilitate robust production of TNFα, resulting in hemorrhagic necrosis of tumors (47). Tumors usually contain various types of macrophages concomitant with invasive properties (48). The recruited macrophages are obliged to support tumor progression, but often turn tumor suppressive in response to polyI:C via TNFα production (47). Like other myeloid species (48), ROS would be a macrophage-derived antitumor modulator induced via extracellular RNA stimulation.
This study further emphasizes an alternative mode of RNA-mediated tumor suppression that is attributable to the direct effect of RNA on tumor cells independent of DC or macrophage responses. Tumor regresses by direct action of RNA without participation of the products of macrophages or DAMPs. In CT26 cells, tumor retardation by polyI:C depends on NK cells because of lower expression of class I MHC in CT26 cells (49). Hence, in addition to the immune or macrophage activation by polyI:C, tumor cells can be direct targets of polyI:C for the induction of cell death. TLR3 is often expressed in murine and human tumor cells (50, 51). TLR3 levels in tumor cells would be a biomarker for the therapeutic efficacy of dsRNA therapy in renal cell carcinoma and breast cancer (52, 53). We found that tumor necroptosis by TLR3 signaling occurred only under specific conditions, in line with the findings that caspase-8 deficiency, caspase inhibition by zVAD, or the presence of anticaspase viral proteins is required for necroptosis induced by TNFα or death receptors (54, 55). Further elucidation of the molecular composition of RNA-mediated cell death and development of a strategy to deliver an anti-caspase reagent to tumor cells will make RNA immunotherapy more potent in conjunction with dsRNA-mediated DC maturation.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: H. Takaki, M. Matsumoto, T. Seya
Development of methodology: R. Takemura, H. Shime, T. Seya
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Takemura, H. Takaki, S. Okada, H. Shime, T. Akazawa, T. Seya
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Takemura, H. Takaki, T. Seya
Writing, review, and/or revision of the manuscript: H. Takaki, H. Oshiumi, M. Matsumoto, T. Seya
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Seya
Study supervision: T. Teshima, T. Seya
This work was supported, in part, by Grants-in-Aid from the Ministry of Education, Science, and Culture (MEXT), “the Carcinogenic Spiral” a MEXT Grant-in-Project, the Ministry of Health, Labor, and Welfare of Japan, the Takeda Foundation, the Yasuda Cancer Research Foundation, and the Kato Memorial Bioscience Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
H. Takaki is a scholarship member in the Japan Society for the Promotion of Science. The authors thank laboratory members for valuable discussions. The authors are also grateful to Olivier Donze and George Chappuis for their kind gift of the RIR3 plasmid.
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).
- Received November 21, 2014.
- Revision received March 21, 2015.
- Accepted April 7, 2015.
- ©2015 American Association for Cancer Research.