Introduction

All mammalian cells (including normal and neoplastic cells) respond to relatively mild perturbations of homeostasis by activating signal transduction cascades aimed at repairing macromolecular and/or organellar damage and restoring normal cellular functions [1,2,3,4]. When successful, such stress responses fully re-establish cellular homeostasis, hence preserving organismal fitness [5, 6]. Conversely, failed adaptation to stress generally elicits regulated cell death (RCD) as a means to preserve organismal homeostasis in the context of cellular loss [7,8,9].

Importantly, most (if not all) cellular responses to stress are hard-wired to immune signaling [10]. Thus, even when normal cellular functions are ultimately restored, stressed cells pre-alert the immune system of a potential danger by: (1) altering their surface properties, and (2) releasing cytokines, chemokines and so-called damage-associated molecular patterns (DAMPs) [11,12,13]. Generally, these signals support the establishment of an inflammatory response that recruits innate immune effector cells to sites of cellular stress, but per se fail to elicit antigen-specific adaptive immunity [10]. Such an immune engagement, however, serves as a platform for the potential initiation of adaptive immune responses if stressed cells fail to recover homeostasis and ultimately undergo RCD [2, 14]. Whether RCD ultimately promotes or inhibits antigen-specific immune responses depends on several critical determinants [15, 16].

Here, we discuss key determinants of immunogenic cell death (ICD) and provide a brief overview of accumulating data on the prominent implications of ICD for cancer (immuno)therapy.

Core ICD determinants

Five core features are required for RCD to elicit antigen-specific immune responses (over mere innate immune signaling coupled to inflammation) of relevance for cancer (immuno)therapy (Fig. 1). As discussed here below, the absence of any of these determinants converts ICD into immunologically silent or even tolerogenic variants of RCD.

Fig. 1
figure 1

Core requirements for the initiation of adaptive immune responses by dying cells. For cell death to drive bona fide adaptive immune responses: (1) cell death must occur in the context of adaptive stress responses; (2) cell death must ultimately occur, as opposed to successful adaptation to stress; (3) dying cells must present antigens that are not covered by thymic tolerance; (4) regulated cell death (RCD) must be accompanied by the emission of endogenous molecules that operate as immunological adjuvants; and (5) microenvironmental conditions must be permissive for antigen-presenting cell (APC) recruitment, maturation and migration to lymph nodes (or other sites of antigen presentation), as well as for cytotoxic T lymphocyte (CTL) infiltration and activation. Depending on which of these conditions is lacking, cell death can drive innate immune signaling coupled with local inflammation, actively promote immunological tolerance and/or result in antigen-specific CTL priming and expansion but no effector immune response. ACD, accidental cell death; DAMP, damage-associated molecular pattern; ICD, immunogenic cell death

Full size image

Stress

Cell death is not immunogenic when it occurs as an accidental, unregulated process that does not involve adaptation to stress, as in the presence of very harsh physicochemical or mechanic conditions (which can be modeled in experimental settings, but are quite rare in human pathophysiology) [17]. In line with this notion, cancer cells succumbing to a variety of therapeutic agents including selected chemotherapies [18], targeted anticancer agents [19] and radiation therapy (RT) [20] can be successfully used to elicit prophylactic anticancer immunity upon inoculation in immunocompetent, syngeneic hosts. However, the same does not hold true when the same cells are killed instantaneously by freeze-thawing cycles [21, 22]. Interestingly, although accidental cell death (ACD) occurring in the absence of stress responses results in a necrotic morphology that has been consistently associated with inflammation in patients affected by a variety of conditions, rapid ACD may turn out to be considerably less inflammatory than stress-driven regulated instances of necrosis such as necroptosis or pyroptosis [23]. Indeed, many of the immunostimulatory signals underlying inflammatory responses to necrotic cells are actively synthesized during stress responses (e.g., cytokines, chemokines) or released along with (failing) adaptation to stress (e.g., DAMPs) [13].

Death

As mentioned above, cell death must occur for perturbations of cellular homeostasis to ultimately results in adaptive immune responses [15]. Thus, while successful adaptation to stress may still ignite local inflammatory responses, cancer cells must die for their corpses to be efficiently taken up by antigen-presenting cells (APCs), especially dendritic cells (DCs), and processed for antigen presentation [24, 25]. At least in part, this reflects the notion that immature DCs are highly proficient at (macro)pinocytosis, which involves material of subcellular size, but much less so at engulfing entire cells [76]. Thus, if (ICD-driven) anticancer immunity had relevance for therapeutic outcome, one would expect immunostimulatory agents (including ICD inducers) to be enriched as compared to immunosuppressive (or immunologically neutral) therapies, which currently is the case [10]. Moreover, drug discovery programs have been designed to actively search for ICD inducer and two of such drugs, i.e., lurbinectedin and belantamab mafodotin, have received regulatory approval for use in cancer patients [77, 78].

Third, in line with preclinical findings, a growing number of ICD inducers positively interact with ICIs or other immunotherapeutic approaches in patients with cancer [79, 80]. Notable examples of such successful combinations include (1) nab-paclitaxel plus atezolizumab (an ICI specific for PD-L1), which is currently employed in the management of triple negative breast cancer (TNBC) [81], carboplatin/etoposide plus atezolizumab, which is approved for patients with extensive-stage small cell lung cancer (SCLC) [82], as well nab-paclitaxel/carboplatin plus the programmed cell death 1 (PDCD1, best known as PD-1) blocker pembrolizumab [83].

Altogether, these preclinical and clinical findings suggest that ICD induction plays a major role in the successful control of multiple neoplasms by (immuno)therapy.

Conclusions and future perspectives

In summary, ICD-driven adaptive immunity is mechanistically and conceptually different from both inflammatory reactions driven by non-immunogenic variants of RCD and adaptive immune responses that do not rely on cell stress and death. Importantly, several RCD routines have been characterized in molecular terms and classified based on the mechanistic involvement of specific signal transduction cascades (Table 1) [9]. For instance, apoptosis is currently defined as an RCD variant that is precipitated by the activation of cysteine proteases of the caspase family, while necroptosis involves the activating phosphorylation of receptor interacting serine/threonine kinase 3 (RIPK3) and consequent phosphorylation-dependent oligomerization of the pore-forming protein mixed lineage kinase domain like pseudokinase (MLKL) [84]. That said, once adaptation to stress fails, cells appear to die irrespective of active signaling, largely because of bioenergetic failure and/or irreparable damage to macromolecular structures that underlie cellular homeostasis itself, including (but not limited to) organelles and membranes [9]. The signal transduction cascades elicited during cell death rather seem to determine the kinetic and immunological manifestations of the process, rather than its occurrence sensu stictu [85]. In line with this notion, both pharmacological and genetic interventions targeting so-called “executioners” of cell death invariably delay the cellular demise, but do not prevent it, at least in mammalian systems [9].

Table 1 Key aspects of regulated cell death variants
Full size table

Most importantly, the biochemical cascades underlying RCD in its multiple variants are not necessarily linked to its immunogenicity [85]. As a standalone example, apoptotic cell death as precipitated by caspases is normally an immunologically silent event, largely reflecting the ability of caspase 3 (CASP3) to initiate signaling pathways that promote macrophage-mediated efferocytosis in the absence of active immunostimulatory signaling and the overall implication of apoptosis in development and adult tissue homeostasis [86]. However, multiple caspase-dependent instances of RCD that classify as ICD by all definitions have been reported [22, 87]. Thus, the immunogenicity of a specific RCD instance cannot be determined with certainty based on the molecular pathways that precipitate RCD only, as abundantly discussed herein. Indeed, RCD-independent, host-related factors including antigenicity and microenvironmental parameters stand out as critical determinants of RCD immunogenicity [10].

Despite this and other conceptual (and experimental) caveats, ICD stands out as a major, therapeutically actionable process for cancer immuno(therapy). Future efforts will have to focus on identifying novel, clinically useful ICD inducers (irrespective of the RCD mode they im**e on) as well as biomarkers predicting the likelihood of specific neoplastic lesions to elicit adaptive immune responses downstream of ICD in response to treatment. Alongside, it will be important to devise clinically viable strategies to increase the immunogenicity of otherwise immunologically silent RCD variants, and to investigate novel combinatorial regimens combining ICD inducers and immunotherapy in the clinic, with the ultimate goal to facilitate efficient anticancer immunosurveillance. We surmise that ICD induction will occupy an ever more central stage in modern cancer management.