Determinants for Antitumor and Protumor Effects of Programmed Cell Death

The maintenance of organismal development and tissue homeostasis requires a balance between the production of new cells and clearance of stressed and dying cells. Programmed cell death is a highly regulated process used to not only clear unneeded cells, but also to communicate homeostatic disturbances, such as danger and tissue injury, to the immune system. Hence, cell death can elicit immune responses against malignancies (1, 2) and pathogens (3). The interaction of dying cancer cells with the immune system has been intensively studied across a variety of model systems, revealing that programmed cell death can elicit an antitumor (4–9) or protumor immune response (10–12).

Studies investigating the role of cell death in the initiation of anticancer immunity started with the exploration of how apoptotic cells communicate with dendritic cells (DC; refs. 13–19). Subsequent studies using cytotoxic anticancer agents, such as cytotoxic chemotherapies (20), radiation (21), oncolytic viruses (22), photodynamic therapy (23), extracorporeal photochemotherapy (24, 25), facilitated the rapid identification of mechanistic details associated with the immunogenicity of cell death and its contribution to anticancer immunity (20, 26–29). Two decades later, we have gained a remarkable understanding of the types of cell stress and programmed cell death that contribute to the emission of molecules (Fig. 1) and their role in shaping anticancer immunity (Fig. 2).

Danger signals, cytokines, and chemokines emitted during immunogenic cell death (ICD) promote the activation, maturation, and trafficking of innate immune cells. Specifically, surface-expressed “eat me” signals, such as calreticulin, attract antigen-presenting cells (mainly DCs) to take up dying cells, along with a variety of immunomodulatory secretomes that stimulate DCs to migrate to the lymph nodes to present antigens to naïve T cells for generating cytotoxic T-cell responses (refs. 20, 26–29; Fig. 2A). It is worthy of mention that T-cell responses after ICD are suspected to spread across multiple neoantigens, creating a broad T-cell repertoire and thus, making it attractive to target tumors with high heterogeneity (21, 30, 31). In addition to effective CD8⁺ T-cell activation, ICD-associated secretomes, mainly chemokines, can facilitate their trafficking into non–T cell–inflamed tumors (32–34). On the contrary, in some tumor types, such as pancreatic and prostate cancer, the secretomes emitted after gemcitabine or oxaliplatin promote the recruitment of immunosuppressive cells, such as myeloid-derived suppressor cells (MDSC; refs. 10, 11) and IgA⁺PD-L1⁺IL10-producing plasmocytes (12), thereby promoting disease progression (Fig. 2B).

In line with the antitumor effects of ICD, chemoradiotherapies exert immune-mediated anticancer effect in patients (4–9) and can synergize with immune checkpoint inhibitors (35–40); although, biomarkers to select patients and predict response are poorly defined (41, 42). To fill these gaps, mechanistic understanding of ICD processes, specifically the individual types of premortem stress and programmed cell death that contributes to anticancer immunity, are highly desired. Unfortunately, many types of chemotherapies simultaneously activate multiple types of premortem stress and cell death in neoplastic cells and actively dividing immune cells. Hence, the specific processes and mechanisms, from the dying cancer cell and the immune cell compartment, that contribute to therapeutic efficacy of ICD remain largely unknown.

During cytotoxic anticancer treatments, cell death takes place in the context of underlying stress [genotoxic, endoplasmic reticulum, and mitochondrial stress, as well as reactive oxygen species (ROS) overload] that also facilitates the emission of secretomes, thus, complicating the direct assessment of cell death–associated immunologic events. There is a surge of genetically engineered cell death systems that specifically activate each type of cancer cell death in isolation, thereby facilitating the discovery of how specific cell death types influence the anticancer immunity cycle (43–50). Overall, past studies have largely focused on secretome-mediated effects of cell death; as a result, its indirect effects on the tumor microenvironment remain understudied. Depending on tumor type, cancer cell death can indirectly affect anticancer immunity by remodeling the tumor architecture and vasculature, thereby influencing the trafficking of antitumor and protumor immune cells and modulating the function of supporting stroma and fibroblasts (51). In the next sections, we will describe the individual types of programmed cell death and their effect on anticancer immunity.

Apoptosis is a type of cell death that is integral to the maintenance of tissue homeostasis, and organism development, as well as numerous pathologies, including cancer. Apoptosis can be initiated by the intrinsic or extrinsic pathway, both of which can result in distinct anticancer immune outcomes (Fig. 1A). Apoptotic cells activate diverse immune-mediated anticancer outcomes, depending on the type of cell death stimuli, upstream stress (26, 28, 52, 53), and cell death signaling and tumor type (34, 54). Unlike chemoradiotherapy-induced immunogenic apoptosis (20), induction of apoptosis by dimerizing C-terminal caspase 8 or caspase 9 has demonstrated less-immunogenic outcomes (43, 44). This is not surprising given apoptotic caspases dampen IFN secretion after the sensing of mitochondrial DNA (55–57). Despite that, apoptosis can be immunogenic when it takes place after mitochondrial membrane permeabilization in a caspase-independent fashion (58) or when the cytotoxic agent concurrently activates endoplasmic reticulum (ER) stress and autophagic response that contributes to the surface expression of calreticulin, promoting phagocytosis (26, 59) and/or natural killer (NK) cell–mediated killing (60), as well as the emission of danger signals, ATP, high mobility group 1 (HMGB1), annexin A1, type I IFNs, and C-X-C motif chemokine ligand 10 (CXCL10; refs. 20, 26–28). These danger signals are sensed by pathogen recognition receptors on DCs and, in concert with cytokines and chemokines, stimulate DCs to undergo activation, maturation, and migration into the draining lymph node to efficiently prime T cells (Fig. 2A). The cytokines and chemokines produced during immunogenic apoptosis are also crucial to allow T-cell activation, differentiation, and migration into malignant lesions (32).

Upstream pyroptotic signaling activates caspases and granzymes (46, 47) to cleave gasdermin (GSDM) proteins, freeing the pore-forming N-terminal domain from the autoinhibitory C-terminal domain. The pores formed by N-terminal gasdermin protein activate the formation of higher-order ninjurin-1 polymers that mediate plasma membrane rupture (refs. 61–65; Fig. 1B). Many chemotherapeutics, including doxorubicin, cisplatin, actinomycin-D, CPT-11, etoposide, and mitoxantrone, activate gasdermin E-mediated pyroptosis (45). Moreover, photodynamic therapy activates pyroptosis in pancreatic tumors (66). Furthermore, oncolytic viruses (48), as well as the combination of TNFα, cycloheximide, and navitoclax, activate pyroptosis (67). In addition, granzyme B from cytotoxic CD8⁺ T cells (46) and granzyme A from NK cells and T cells (47) can cleave gasdermin E and B, respectively, to induce tumor pyroptosis. Unlike their natural T-cell counterpart, chimeric antigen receptor–loaded T cells secrete higher amounts of granzyme B, thereby activating gasdermin E–mediated tumor pyroptosis and the associated cytokine release syndrome (68).

Pyroptosis is highly inflammatory, although detailed investigation on the nature of secretomes that mediate anticancer immunity remain to be identified (63). For example, the N-terminal domain of gasdermin A and E can induce mitochondrial oxidative stress and pore formation prior to plasma membrane rupture, which allows the release of danger signals and intracellular contents that contributes to the immunogenicity (69–71). Pyroptosis allows the release of cytokines, such as IL1β and IL18, as well as danger signals, HMGB1, and ATP (63). These immunomodulatory secretomes stimulate a variety of immune cells, including NK cells, macrophages, DCs, and the subsequent generation of antigen-specific CD8⁺ and CD4⁺ T-cell responses (46, 47, 49, 66). In tumors undergoing pyroptosis, there is decreased FoxP3⁺CD4⁺ regulatory T cells, monocytes, neutrophils, and MDSCs (46, 47, 49). Thus, pyroptosis can induce robust anti-immune responses and improve the survival of tumor-bearing mice (46, 47, 49).

Necroptosis is initiated by a variety of stimuli, including the engagement of death receptors (TNFR1 and Fas/FasL), Toll-like receptors, or intracellular nucleic acid sensors under a caspase-8 inhibited state (72). Necroptosis execution involves the kinases receptor-interacting serine/threonine-protein kinase 1 (RIPK1) and RIPK3 to activate NFκB-mediated inflammation and to phosphorylate mixed lineage kinase domain-like (MLKL), which executes cell lysis (73, 74). Several chemotherapies such as etoposide, givinostat, obatoclax, and 5-fluorouracil trigger necroptosis (75–77). Furthermore, ionizing radiotherapy and targeted receptor therapy also activate necroptosis (75, 78–80). Along similar lines, certain oncolytic virus monotherapies can induce necroptosis and combine favorably with chemotherapies to potentiate anticancer effect (refs. 34, 81, 82; Fig. 1C).

Necroptotic fibroblasts, induced after dimerization of RIPK3, release danger signals HMGB1, ATP, and proinflammatory cytokines (IFNγ, IL6) and chemokines (CCL3, CCL4, CCL5; refs. 43, 44). Intratumoral inoculation of necroptotic fibroblasts elicits a CD103⁺ cell–mediated T-cell response (43, 44). Furthermore, necroptosis induces an appreciable increase in intratumoral neutrophils and macrophages (34). It is worthy of mentioning that RIPK3 lacking the RIP homotypic interaction motif domain fails to elicit NFκB-mediated inflammation and tumor control (43). Meanwhile, another study reports that necroptotic cell death induced by intratumoral injection of MLKL mRNA exerts durable antitumor effects, suggesting that necroptotic cell death in the absence of NFκB activation can also elicit antitumor immunity (82, 83).

Ferroptosis is driven by iron-dependent, excessive phospholipid peroxidation that compromises plasma membrane integrity and cell lysis (84). It is regulated by a variety of metabolic pathways, including redox homeostasis, iron handling, mitochondrial activity, and the metabolism of lipids and amino acids (ref. 85; Fig. 1D). Phospholipid peroxidation relies on iron, ROS, and phospholipids containing polyunsaturated fatty acid chains. An important regulator of ferroptosis is the micronutrient selenium, which is required for the biosynthesis of selenoproteins that scavenge ROS. A key inhibitor of phospholipid peroxidation is glutathione peroxidase 4 (GPX4; ref. 86). In addition, a variety of factors can influence the induction of ferroptosis by regulating cellular metabolism and ROS levels. Despite a wealth of knowledge in the mechanisms regulating ferroptosis, the exact molecular events and terminal effectors of ferroptotic cell death are unknown.

Ferroptosis is activated in cancer cells after treatment with inhibitors of the cystine/glutamate antiporter system, such as erastin (87, 88) and sulfasalazine (89, 90), tyrosine kinase inhibitors (91), radiotherapy (92), and neratinib (a potent, irreversible pan-tyrosine kinase inhibitor; ref. 93). Ferroptosis inducers ML161, RSL, and erastin, as well as inducible genetic depletion of GPX4 in cancer cells, promotes the surface expression of calreticulin and emission of danger signals ATP and HMGB1 and cytokines (50, 94). However, there is conflicting evidence on the effects of ferroptotic cancer cells on DC activation and cross-priming (50, 94). A recent study shows that ferroptotic cancer cells suppress DC function and fail to protect against tumor growth (50). However, previous work using RSL3 reports that early ferroptosis has the opposite effect of potentiating DC function and the associated antitumor effects (94). Interestingly, both studies used the MCA205 tumor model, hence the conflicting findings could be attributed to animal facility-associated microbiome differences (95). There are also two instances where ferroptosis is shown to promote tumor progression via immunosuppressive intratumoral macrophage infiltrate (96, 97).

Recent studies demonstrate that T cell–mediated tumor killing involves ferroptosis (98, 99). In vitro IFNγ alone fails to induce ferroptosis but synergizes with arachidonic acid to activate ACSL4-dependent tumor ferroptosis (99). However, in established tumors, IFNγ secreted by cytotoxic CD8⁺ T cells inhibits the glutamate–cystine antiporter system x_c⁻ in tumors to promote lipid peroxidation and ferroptosis (98). Furthermore, forcing tumors to undergo immunogenic ferroptosis by cyst(e)inase, an engineered enzyme that degrades both cystine and cysteine, synergizes with immune checkpoint inhibitor therapy (98). Overall, because ferroptosis is a relatively new type of cell death, additional studies are required to clarify the immunologic outcomes of inducing ferroptosis in distinct tumor types with variable immune landscapes and microenvironments.

Tumors evolve to evade anticancer immunity via acquisition of genetic or epigenetic aberrations in pyroptosis and necroptosis effector proteins (refs. 46, 100–105; Fig. 3). Mutations in RIPK3 and MLKL are well documented in several tumor types (102, 106–108). RIPK3 expression is epigenetically regulated at different stages of tumor progression (100, 109), and treatment with a hypomethylating agent, decitabine, can restore RIPK3 expression and significantly improve responses to chemotherapy in human tumor xenografts (100).

Lung adenocarcinoma, colon adenocarcinoma, and uterine corpus endometrioid carcinoma possess the most frequent gasdermin E linker region mutations, with lung and colon adenocarcinoma commonly containing loss-of-function mutations (46, 110). In addition, selection pressure to downregulate gasdermin E coincides with emerging data suggesting it can regulate anticancer immunity (46). Tumors express five spliced variants of gasdermin B, whereby N-terminal fragments of isoforms 3 and 4 induce pyroptosis, but isoforms 1, 2, and 5 lack the motif that allows gasdermin B to insert in the membrane. These mutated gasdermin B 3/4 isoforms block pyroptosis caused by other cytotoxic gasdermin B isoforms (111). Gasdermin proteins can also undergo cancer-specific methylation. Methylation of gasdermin A, C, and E promotor sites has been observed in both gastric cancer and esophageal squamous cell carcinoma (104, 112). Treatment with decitabine successfully restores gasdermin A expression and induction of pyroptosis in gastric cancers (113, 114). Furthermore, gasdermin E is hypermethylated at the CpG islands near the transcription site in over half of breast cancers (105). Human tumors frequently mutate gasdermin E and C in the linker region between the N- and C-terminal domain, abrogating cleavage and activation of their pore-forming activities (46).

Tumor genetics and associated oncogenic factors can establish immunosuppressive intratumoral niches, as has been comprehensively reviewed elsewhere (115). In addition to these noncell autonomous effects, oncogenic pathways involving oncogenes and tumor suppressors can also interfere with the cancers own cell death machinery to either provide a proliferative advantage or allow escape from immune-mediated attack (100, 116–121). In many cancers, distinct combinations of oncogenes and loss of tumor suppressors drive tumorigenesis and regulate signaling, leading to apoptosis. In most instances, tumor suppressors p53 (122–125), Rb (126–128), and PTEN (129) stimulate proapoptotic signaling to clear premalignant cells, whereas oncogenic Kirsten rat sarcoma virus (KRAS) positively regulates cancer cell survival and proliferative capacity by downregulating apoptotic pathways (130–132). It is not known whether inhibition of KRAS or any other antiapoptotic oncogenic pathway results in immunostimulatory apoptosis.

Oncogenic drivers suppress the expression of necroptotic and pyroptotic effector proteins and ablation of oncogenic signaling results in the enhancement of ICD, thereby reigniting antitumor immunity. For example, c-Myc interacts with RIPK3 and RIPK1 to prevent necrosome formation (119). Depletion of c-MYC reinstates necroptosis-promoting antitumor immunity (119). Yet another oncogene that compromises the ICD machinery is BRAF. BRAF/AXL ablates RIPK3 expression in patient-derived xenografts (120). In a separate study, the combined inhibition of BRAF and MEK in mutant melanoma has proven successful at promoting pyroptosis, with the hallmark of HMGB1 release and expansion of tumor-specific CD8⁺ T cells (121, 133). Finally, inhibition of KRAS^G12C regresses patient-derived colorectal, lung, and pancreatic xenografts through activation of ICD (116). Indeed, such an outcome has been translated into patients and has garnered clinical efficacy (116). Yet it is not known what type of specific cell death is activated upon KRAS^G12C inhibition after treatment with AMG510. Cancer cell ferroptosis is also regulated by oncogenic Ras (134–137) and the tumor suppressor p53 (138, 139), yet it remains unknown whether this directly influences immune outcome. In summary, all of these studies highlight the need for comprehensive and systematic studies to elucidate how distinct oncogenic pathways influence cell death–mediated immunity and antitumor effects (1). The findings are foundational to apply chemoradiotherapies in combination with oncogenic inhibitors to stimulate patient antitumor responses.

Specific cell death types have been associated with natural and therapy-induced antitumor or protumor effects. Tumors that endogenously express higher levels of immunostimulatory cell death effectors proteins, such as RIPK3, gasdermins, and ACSL4, show an elevated CD8⁺ T-cell infiltrate and prolonged patient survival (8, 9, 46, 47, 99, 111). In the context of ICD-inducing treatments, the nature of the lethal stimuli, along with tumor type, its underlying genetics, and the microenvironment, dictate the type of immunostimulatory secretomes emitted to predict the initiation of anticancer or protumor immunity. As reviewed elsewhere (140, 141), additional host factors, such as germline mutations in pathogen recognition receptors and host microbiota, affect immune response and survival response after chemoradiotherapies.

There are many clinical trials combining ICD-inducing treatments with immune checkpoint inhibitors (35–40), although there is lack of defined clinical biomarkers to select patients and to predict treatment outcomes. Regarding biomarker development, care must be taken in correlating the level of transcripts or full-length cell death proteins with immune cell infiltration, functionality, and patient survival. In addition to widely known discrepancies between transcript and protein expression patterns, cell death effector proteins have multiple functions, and their specific role in programmed cell death requires active protein modifications, such as cleavage or phosphorylation. In addition, tumors express dominant-negative isoforms of cell death effector proteins to evade ICD. For example, gasdermin B has five isoforms, and tumors preferentially overexpress isoforms with a dominant-negative effect to escape NK cell– and T cell–mediated attack (111); thus, correlating the totality of gasdermin B isoforms with immune function would lead to incorrect interpretation. To add to this complexity, there is extensive cross-talk among cell death pathways, with many of the upstream signaling and effector proteins shared between distinct cell death modalities and other inflammatory processes. Hence, careful analysis is warranted when correlating the extent of cell death with immune infiltration and survival outcomes. To this end, we recommend future studies to comprehensively quantify all the forms of cell death to define the major modality of cell death induced by specific ICD-inducing treatments in a specific cancer type and how that shapes the tumor immune landscape. In parallel, preclinical studies using genetic systems to activate specific cell death modalities will provide fundamental insight of the transcriptional, immunologic, and microenvironmental signatures for further biomarker development and integration into existing clinical biomarkers.

Despite mounting evidence underpinning the contribution of stress responses in ICD (142), expression of cell death effector proteins MLKL (82, 83) and gasdermin A/E (46, 49) induce robust anticancer immunity, highlighting the immunogenicity of cancer cell death in the absence of underlying stress and inflammatory signaling. However, there is limited knowledge on how the induction of cancer cell death in acute and chronic setting shapes immunity. It is likely that the type, rate, and amount of cell death may dictate immunity by influencing the activation and trafficking of immune cells and the availability of secretomes and tumor antigens. In a recent work, pyroptosis induction in less than 15% the cancer cells eradicates entire established mammary tumors (49). Relating this to a clinical setting, the dose regimen of chemoradiotherapies that preferentially activate ICD and anticancer effects are unknown. However, sublethal doses are suspected to exert better anticancer effects, mainly by enhancing cancer cell genotoxic and ER stress to stimulate T-cell responses (143) and/or activating cancer cell senescence that potently activates T cell–mediated immunity (144). Future clinical studies correlating the type and extent of cancer cell death with T-cell infiltrate and anticancer effects are highly desired.

In a variety of immune-responsive and spontaneously arising murine tumors, chemoradiotherapy-mediated rapid activation of ICD results in immune-mediated tumor elimination (145). On the contrary, the protumor effects of cell death are presented in chronic inflammatory settings, where elevated expression of cell death effector proteins accelerates tumor progression (10, 11, 146–152). For example, necroptosis is frequently reported as immunostimulatory in the context of chemoradiotherapy (34, 153) or when it is activated by expressing dimerizable RIPK3 or MLKL (30, 43, 44, 154). In contrast, pancreatic tumors overexpressing RIPK1/3 accelerate tumor progression by secreting CXCL1 to attract MDSCs. Accordingly, ablating necrosome formation generates a T cell–inflamed pancreatic tumor microenvironment (10, 11). Although previously unexplored, defining the mechanisms by which pancreatic ductal adenocarcinoma (PDA) prevents plasma membrane rupture downstream of pathologically elevated necrosome formation (10) may identify therapeutic opportunities to trigger PDA cell lysis and possibly kickstart the antitumor immunity cycle.

It remains intriguing under what tumor microenvironmental conditions a specific type of cell death becomes antitumor or protumor. Tumor type, genetics, and anatomical location can shape the underlying microenvironment, which in turn dictates the immune-mediated anticancer effect of ICD-inducing therapies. In this respect, a highly immunostimulatory cell death may not overcome an immunosuppressive tumor when administered as a monotherapy (33, 34, 54), and a less-ICD may readily exert durable anticancer immunity in immune-responsive tumors (20, 155). Hence, investigating how cell death affects the overall immune landscape using tumors arising in different anatomical locations with variable intratumoral environments is highly desired. Future studies are needed to investigate how tumors with varying immune landscape respond to a similar lethal stimulus, with the goal to define the type and amount of cell death, the composition of their secretomes, or the recruitment and functionality of innate and adaptive immune cells. Furthermore, given the microenvironment-based immune cell plasticity and distinct immunoregulatory networks in tumor-draining lymph nodes (156), it is possible for tumors and draining lymph nodes to subvert the entry or functionality of cell death–activated immune cells. Elucidating how the tumor microenvironment shapes immune-mediated outcomes of ICD will have profound effects in the design of biomarkers for patient selection and to predict treatment efficacy.

S.T. Workenhe reports grants from Alivid LLC and POINT Biopharma outside the submitted work. No disclosures were reported by the other authors.

S.T. Workenhe is funded by Canadian Institute of Health Research (PJT 185868 and PJT 190071), SickKids New Investigator award (NI23-1064R), Cancer Research Society (ID# 942502 and 1056603), Ontario Veterinary College (OVC) Pet Trust, Ontario Institute for Cancer Research funding from The Joseph and Wolf Lebovic Cancer Genomics and Immunity Program. J.M. Inkol is supported by OVC PhD Scholarship and Ontario Graduate Scholarship. M.J. Westerveld is supported by OVC PhD Scholarship. S.G. Verburg is supported by CGS-M, OVC MSc Scholarship and Ontario Graduate Scholarship. S.M. Worfolk is supported by Queen Elizabeth II Graduate Scholarship in Science & Technology.