Double agents of cell death: novel emerging functions of apoptotic regulators

Abstract

Apoptosis is a highly regulated form of cell death that is required for many homeostatic and pathological processes. Recently, alternative cell death pathways have emerged whose regulation is dependent on proteins with canonical functions in apoptosis. Dysregulation of apoptotic signaling frequently underlies the pathogenesis of many cancers, reinforcing the need to develop therapies that initiate alternative cell death processes. This review outlines the convergence points between apoptosis and other death pathways with the purpose of identifying novel strategies for the treatment of apoptosis-refractory cancers. Apoptosis proteins can play key roles in the initiation, regulation, and execution of non-apoptotic death processes that include necroptosis, autophagy, pyroptosis, mPTP-mediated necrosis and ferroptosis. Notably, recent evidence illustrates that dying cells can exhibit biochemical and molecular characteristics of more than one different type of regulated cell death. Thus, this review highlights the amazing complexity and interconnectivity of cell death processes and also raises the idea that a top-to-bottom approach to describing cell death mechanisms may be inadequate for fully understanding the means by which cells die.

Graphical Abstract

Apoptosis is a form of regulated cell death which is critical for many developmental and physiological processes, however, dysregulation of apoptosis underlies cancer pathogenesis. Recent evidence indicates that many proteins involved in the regulation of apoptosis also have key functions in other regulated cell death pathways, revealing potential novel targets for the treatment of apoptosis-refractory cancers.

Introduction

Over 50 years have passed since Richard Lockshin and Carroll Williams first described the sequential destruction of intersegmental muscles in the silkmoth as a process called “programmed cell death” [1]. This developmentally regulated process of cell elimination was also simultaneously being characterized in vertebrates, resulting in the establishment of the term “apoptosis” by Kerr, Wyllie and Currie to describe the series of morphological events leading to controlled cell removal [2]. Ever since these initial observations, apoptosis has now become the most well understood cell death pathway in the molecular era. Apoptosis is a process in which cells undergo discrete morphological and biochemical steps leading to cell demise in response to developmental or pathological cues [3,4]. Recently there has been a renaissance in the field, such that the number of distinct regulated cell death modalities tally to more than a dozen distinct death pathways [5]. While the core components of the apoptotic pathway have been extensively described in the context of apoptosis [3,4,6], a large body of literature has emerged revealing that apoptotic proteins actively engage with and regulate other forms of cell death. The purpose of this review is to bring to light the extensive crosstalk between proteins well known for their function in apoptosis with emerging nonapoptotic cell death pathways (Figure 1). It is becoming increasingly clear that these pathways are not strictly linear, but form a network of entwined physical and functional interactions. While some of these players simply inhibit one pathway while facilitating activation of another, some apoptosis proteins are directly involved in the activation and amplification of nonapoptotic death signals. In an era in which small molecule inhibitors of apoptosis are in clinical use [7–9], the consequences of pharmacologically modulating apoptosis must be assessed for their influence on these entwined death pathways, and how understanding the crosstalk between apoptotic and nonapoptotic death pathways may reveal potential new therapeutically druggable targets.

Figure 1: Apoptosis regulators as double agents of cell death.

Several classes of proteins with key functions in apoptosis also participate in the execution and/or regulation of other death pathways. Colored lines indicate a documented linkage between each class of apoptotic protein (black) with nonapoptotic death processes.

A brief review of the canonical apoptosis pathway

Apoptotic cell death can be initiated either in response to death receptor (DR) signaling from the plasma membrane or mitochondrial signaling following intracellular stress (Figure 2). DR-mediated apoptosis is initiated upon ligand binding (ie. Fas, tumor necrosis factor (TNF), TNF-related apoptosis-inducing factor (TRAIL)) to their cognate receptor. Ligand binding results in DR clustering and assembly of intracellular signaling complexes through homotypic interactions with cytoplasmic death domain (DD) and death effector domain (DED)-containing proteins [10,11]. Activation of CD95/Fas or TRAIL receptors lead to the formation of the death-inducing signaling complex (DISC), consisting of the adaptor molecule Fas-Associated Death Domain (FADD) and procaspase-8/−10, whereas activated TNF receptors recruit distinct complexes which mediate either cell death or survival (described in more detail in the section “Crosstalk between apoptosis and necroptosis”, below). DISC assembly mediates the recruitment and homo-oligomerization of additional procaspase-8 molecules to form a macromolecular filament [12] that facilitates self-cleavage to yield active caspase-8. From there, caspase-8 possesses dual functions by either directly activating downstream executioner caspases or feeding into the intrinsic (mitochondrial) apoptosis pathway.

Figure 2: Crosstalk between apoptosis and necroptosis pathways.

Signaling from the TNF receptor (TNFR) can induce distinct cell death outcomes through the dual activity of common factors such as caspase-8 and the IAPs. Factors and pathways with known functions in apoptosis or necroptosis are indicated in black or blue shaded boxes/arrows, respectively. Those factors with dual functions (such as caspase-8, cIAP1/2 or XIAP) are highlighted with black and blue boxes. Pharmacological inhibitors of apoptosis or necroptosis are indicated in red. The ubiquitination state of RIPK1 is regulated by cIAP1/2, which keeps the kinase inactive and promotes a survival signal through NFκB. Inhibition of cIAPs by Smac mimetic compounds coupled with TNFR signaling liberates RIPK1 from the receptor complex, allowing it to interact with downstream factors leading to apoptosis (FADD, caspase-8) or necroptosis (RIPK3, MLKL). While caspase-8 and IAPs possess a direct role in the regulation of RIPK1, other apoptosis regulators, including BAX, BAK and several BH3-only proteins, have been shown to influence necroptotic signaling via mechanisms that are not fully understood.

The central regulators of the mitochondrial apoptosis pathway are members of the BCL-2 (B-cell lymphoma-2) family, a collection of pro- and anti-death proteins whose engagement with each other dictates cell fate. The founding member of the family, BCL-2, was identified at a chromosomal breakpoint t(14,18) common in patients with B-cell follicular lymphoma [13], in which the BCL-2 gene was translocated downstream of the immunoglobulin heavy chain locus (IgH) resulting in constitutive expression of the protein. By characterizing the cell cycle profile of BCL-2 expressing cells, David Vaux, Suzanne Cory and Jerry Adams made the seminal observation that BCL-2 was sufficient to mediate survival of the cells after a death signal (interleukin-3 withdrawal), and did not simply promote growth as was previously believed to be solely responsible for tumorigenesis [14]. Since then, the BCL-2 family has expanded to include members that can either promote survival (BCL-2, BCL-xL, MCL-1, A1, BCL-w, and others) or apoptotic cell death (BAX, BAK, BOK) [15]. Additionally, a third branch of the apoptotic BCL-2 protein family tree is the BCL-2-homology 3 (BH3)-only proteins (Bid, Bim, Bad, NOXA, PUMA and others), a class of sensitizer and activator proteins charged with the task of integrating signals from the intracellular environment to modulate interactions between the survival and death-promoting family members [16].

Activation of the intrinsic apoptosis pathway can integrate with extrinsic apoptotic signaling through caspase-8-mediated cleavage of the BH3-only protein, BID. Truncated BID (tBID) directly engages prodeath BCL-2 family members, BAX and BAK, leading to their activation by inducing conformational changes, oligomerization, and mitochondrial outer membrane permeabilization (MOMP) [17]. The release of mitochondrial factors, including cytochrome c and Smac/DIABLO (second mitochondria-derived activator of caspases/direct IAP-binding protein with low pI), leads to the initiation of several downstream consequences including DNA fragmentation, loss of mitochondrial function, inhibition of anti-death proteins, and assembly of the apoptosome complex. Cytochrome c released from mitochondria stimulates the assembly of the apoptosome by complexing with Apaf-1 and procaspase-9, leading to proteolytic activation of caspase-9 and subsequent activation of the executioner caspases-3/7. Regardless of whether directly activated by caspase-8 (extrinsic) or through the apoptosome (intrinsic), activation of the executioner caspases-3/−7 executes downstream events leading to the morphological and biochemical features of apoptosis.

A number of excellent reviews are available for more detailed discussions of the canonical apoptosis pathway [4][3][6][15][18][19].

Targeted approaches for inducing apoptosis in cancers

Therapeutic manipulation of apoptotic pathways has been a long sought out strategy for the treatment of cancers. Most chemotherapeutic and radiotherapeutic interventions interfere with tumor cell proliferation by inducing DNA damage, interfering with the cell cycle or halting DNA replication, the downstream consequence of which leads to apoptotic cell death. However, many cancers contain inactivating mutations or upregulate antiapoptotic proteins that prevent damaged/injured cells from undergoing apoptosis [20]. Recent clinical strategies for activating apoptosis in human cancers have focused on the development of bioactive small molecules, including Smac mimetics/IAP antagonists and inhibitors of antiapoptotic BCL-2 family members (BH3 mimetics). Upon its release from the mitochondria during apoptosis, Smac potentiates an apoptotic signal by directly binding and inactivating endogenous inhibitor of apoptosis proteins (IAPs) [21], a family of negative regulators of intrinsic and extrinsic apoptotic signaling [22]. Although eight family members have been identified, the IAPs predominantly involved in apoptosis inhibition include X-chromosome-linked IAP (XIAP), cellular IAP1 (cIAP1), and cellular IAP2 (cIAP2). Consequently, IAPs are frequently upregulated in human cancers [23], with their expression correlating with chemoresistance and poor prognosis. Monovalent and bivalent small molecule mimetics contain four key amino acids of the Smac N-terminus (Ala-Val-Pro-Ile) that engage the baculoviral IAP repeat (BIR) domains of IAPs. As a result, Smac mimetics functionally disrupt the inhibitory interaction of XIAP with caspases or stimulate the E3 ubiquitin ligase activity of cIAPs leading to a reduction in canonical NF-κB-mediated survival signals and increased noncanonical NF-κB signaling resulting in autocrine TNF secretion [23,24]. Clinical use of Smac mimetics are being explored for the treatment of multiple cancers, both as a single agent or in combination with existing chemotherapeutic agents [23,25]. Although well tolerated in Phase I clinical trials [26], Smac mimetics possessed limited efficacy as a single agent therapeutic in patients with a variety of solid tumors and hematological malignancies, including triple-negative breast cancer, ovarian, primary peritoneal, fallopian tube cancer, and multiple myeloma [27]. Upregulation of circulating inflammatory cytokines was also observed in a Phase I study after Smac mimetic treatment in patients with solid tumors, likely due the effects of cIAP1/2 inhibition activating noncanonical NF-κB signaling [28]. However, IAP inhibition appears to sensitize tumors to standard chemotherapeutic and radiologic interventions [27].

Small molecule mimetics have also been developed that inhibit anti-apoptotic BCL-2 family members, with the goal of triggering an apoptotic response in tumor cells. Designed to mimic the BH3 domain conserved among many BCL-2 family members, BH3 mimetics engage the hydrophobic groove of anti-apoptotic BCL-2 proteins with high affinity and specificity, relieving inhibition of pro-apoptotic family members, BAX and BAK, and resulting in a caspase-dependent cell death [29]. The first FDA approved BCL-2 selective inhibitor, venetoclax, is in clinical use for the treatment of chronic lymphocytic leukemia (CLL) [8,30]. More recently, venetoclax was approved in combination with demethylating chemotherapeutic agents for the treatment of acute myeloid leukemia (AML) [31]. However, the recent emergence of venetoclax-refractory tumors have reinforced the need for the development of additional treatment strategies [32–34]. Venetoclax resistance has been observed due to acquired mutations within the BCL-2 protein, by upregulation of other anti-apoptotic family members (such as MCL-1) or by altering other signaling pathways (JNK, autophagy) [32,35]. Thus new strategies for the treatment of apoptosis-refractory cancers are needed to complement existing therapeutics, perhaps through the pharmacological manipulation of alternative, non-apoptotic forms of regulated cell death.

Apoptosis proteins as double agents of cell death

Historically, apoptosis was the only known form of regulated cell death, both in developmental and pathological contexts, with any other modes of cell death described as accidental (necrotic). However, an abundance of evidence has identified several additional forms of regulated cell death, many of which are beginning to be understood in mechanistic detail [36–39]. Quite striking is the fact that many apoptotic proteins play key roles in the regulation of these alternative cell death pathways. Here, this review provides brief descriptions of each of these emerging pathways, and highlights how apoptotic proteins function within those pathways.

Crosstalk between apoptosis and necroptosis.

Over a decade ago, Junying Yuan and colleagues described a novel caspase-independent, death receptor-induced form of inflammatory cell death they termed necroptosis [40]. Since then, the cell death field has experienced a renaissance, during which a detailed mechanism of this regulated necrosis pathway has begun to emerge. Unexpectedly, several key regulators of apoptosis also have important functions in the necroptosis pathway, suggesting a sophisticated mechanism for cross-regulation (Figure 2). The pathway of regulated necroptosis centers around the activity of the kinases Receptor-Interacting Protein Kinase 1 (RIPK1) and RIPK3, the activity of which are regulated in response to TNF-mediated trimerization of TNFR1. While the default outcome of TNFR1 activation leads to survival signaling through NF-κB, TNFR1 can also initiate distinct apoptotic or necroptotic pathways depending on the faction of regulatory components that are engaged with the receptor’s cytoplasmic domain [36]. As one of many components of TNFR1 complex-1, RIPK1 undergoes distinct subsets of posttranslational modifications that dictate whether the complex potentiates a survival signal by activating NF-κB or initiates a death pathway. In their canonical antiapoptotic role, cIAP1/2 prevents TNFR1-mediated apoptosis by constitutively ubiquitinating RIPK1 to inactivate the kinase and facilitate cell survival [41,42]. Loss of these inhibitory signals following a death stimulus allow RIPK1 to dissociate from complex-1 and form complex-2a (ripoptosome) with FADD and procaspase-8, which leads to death by the extrinsic apoptotic pathway [42]. Additionally, active caspase-8 functions as a negative regulator of necroptosis by cleaving RIPK1 within its kinase domain (Asp324), thereby ensuring that the cell proceeds by an apoptotic death pathway [43].

Necroptosis was initially observed only in the absence of caspase activity (either by treatment with caspase inhibitors or genetic deletion) [40]. When caspases are inhibited during TNFR1 activation, RIPK1 is relieved from inhibitory caspase-8 cleavage, allowing it to heterodimerize and activate RIPK3. Together, the RIPK1/RIPK3 necrosome (complex-2b) phosphorylates the pseudokinase, mixed-lineage kinase domain-like protein (MLKL), inducing a conformational change that liberates an N-terminal four-helix bundle (4HB) domain from the molecule that favors MLKL oligomerization and formation of higher order amyloid-like structures [44]. Activated MLKL subsequently translocates to the plasma membrane, leading to membrane permeabilization by a mechanism that has not been fully delineated [45–47]. While many initially questioned the physiological relevance of a death pathway that was apparently observed only after apoptosis inhibition, it has become clear that necroptosis represents a bona fide cell death pathway with clear roles in neurodegenerative disease pathology and neuroinflammation [48,49].

In addition to caspase-8 and cIAPs, other apoptotic regulators have also been shown to modulate the necroptotic death process. The other major IAP family member, XIAP, can also play dual roles in preventing apoptosis or necroptosis. To inhibit apoptosis, XIAP engages specific subsets of effector caspases in the cytoplasm either by direct binding or by ubiquitinating them for proteasomal degradation. Necroptosis inhibition by XIAP occurs downstream of TNFR1 by ubiquitination of RIPK1 in the context of complex-2b [50]. Moreover, BCL-2 family proteins have also been identified as both accomplices and adversaries of the necroptotic death process. Mouse embryonic fibroblasts (MEFs) deficient for both BAX and BAK are resistant to necroptosis induced by TNF, cycloheximide, and zVAD treatment [51]. In a separate study, BAK deficient cells underwent delayed necroptosis in response to IAP inhibition (with Smac mimetic) combined with glucocorticoid treatment [52]. Necroptosis induction also depletes mitochondrial localized MCL-1 in a RIPK3-dependent manner [53], thus leading to mitochondrial dysfunction. Very recently, endogenous MLKL was shown to interact with BCL-2 via a putative BH3 domain, inhibiting RIPK3-dependent MLKL phosphorylation and oligomerization after necroptosis stimulation [54]. In addition, knockout MEFs of the BH3 protein Bmf were mildly resistant to death induced by cadmium treatment, a known inducer of necroptosis [55], suggesting that BH3 proteins may also be involved in RIPK1-mediated death. Another group discovered that the proapoptotic BH3 protein, PUMA is transcriptionally upregulated upon necroptosis initiation by the p65 subunit of NF-κB, which amplifies necroptosis through PUMA-mediated signaling back to RIPK1 and MLKL [56]. Of note, there is mixed evidence as to whether mitochondria are required for necroptotic cell death. While mitophagy-mediated depletion of mitochondria did not alter the kinetics of RIPK3-mediated necroptotic cell death in 3T3 cells [57], other studies have demonstrated that mitochondrially-generated reactive oxygen species (ROS) potentiates the necroptotic signal [58]. Whether or not mitochondria are required for necroptosis, many nonapoptotic functions of BCL-2 proteins occur at non-mitochondrial sites throughout the cell (ie. endoplasmic reticulum, cytoplasm, nucleus) [59], thus opening up the possibility that BCL-2 proteins may exert direct influence on the necroptotic pathway.

Small molecule modulators of the apoptotic pathway have been shown to stimulate necroptotic cell death in apoptosis-deficient cells. Smac mimetics, which were designed to functionally inhibit IAP proteins to stimulate apoptosis, have been shown to cause necroptotic cell death in apoptosis-deficient cells [60–62]. In AML mouse models and patient-derived xenografts, combined pharmacological inhibition of IAPs and caspase-8 induced necroptotic cell death of leukemic cells, resulted in reduced tumor burden and enhanced survival [63], thus paving the way toward selective activation of necroptosis for the treatment of hematological malignancies. Many commonly used anti-cancer agents can also stimulate RIPK1-dependent necroptosis [64], but cancer cells also frequently epigenetically silence RIPK3 leading to tumor resistance [65,66].

The intersection of apoptosis and necroptosis provides an opportunity for the development of therapeutics to selectively target both pathways, yet assessing the prudency of altering the activity of one pathway to alternatively stimulate the other is required. While initiating an inflammatory response may be an effective way to combat cancer cells that have become apoptosis resistant, inflammatory responses would require precise control to avoid extensive pathology due to acute or chronic inflammation [67,68]. Paradoxically, necroptosis-induced inflammatory responses have been implicated in promoting tumor progression and metastasis [69]. For example, high RIPK1 expression in glioblastoma patients correlates with a worse prognosis, likely due to increased NF-κB signaling leading to an upregulation of mdm2 and loss of p53 tumor suppressor activity [70]. Pharmacological inhibition of RIPK1 with necrostatin-1 (a small molecule inhibitor that locks RIPK1 in its inactive conformation [71]) or tissue specific ablation of RIPK3 in mouse endothelial cells reduced metastasis that occurs in response to endothelial necroptosis caused by tumor cells [72]. Thus, necroptosis activation may prove advantageous for the treatment of apoptosis-resistant tumors, but whether engaging an immune response would be beneficial or deleterious has yet to be determined in patients.

Crosstalk between apoptosis and autophagy.

Autophagy is a homeostatic process in which cytoplasmic contents and/or dysfunctional organelles are self-consumed and recycled into components to be utilized by the cell [73]. Most typically thought of as a recycling mechanism in the face of nutrient depletion, autophagy is morphologically characterized by the de novo formation of a double-membrane vesicle containing cytoplasmic contents that fuses with the lysosome [74]. Dual roles for apoptosis proteins in canonical autophagy regulation have been described with proteins engaged at nearly every stage of the apoptotic pathway, including death receptor ligands, BCL-2 family members, BH3 proteins, and caspases, emphasizing the importance of precise regulation between these pathways (Figure 3). Inhibition of the inflammatory factor and death receptor ligand TNF resulted in protection of photoreceptors after retinal detachment, presumably due to a transient increase in autophagy and a corresponding reduction in apoptosis [75]. Anti-apoptotic BCL-2 proteins also bind to the autophagy protein Beclin 1, preventing it from interacting with the VPS34 PI3K complex responsible for autophagy initiation [76–78]. In addition to binding Beclin 1, BCL-2 has also been reported to directly bind to and regulate the lipidation of GABARAP, a protein important for autophagosome formation [79], but this interaction has not been independently confirmed. BCL-xL has been shown to inhibit autophagosome:lysosome fusion and bind Beclin 1:UVRAG to suppress bacterial internalization in a model of Group A Streptococcus infection [80]. Atg12, which when complexed with Atg5 and Atg16 at sites of autophagosome assembly, binds and inhibits anti-apoptotic BCL-2 proteins through a novel BH3 domain [81]. Selective degradation of mitochondria, known as mitophagy, is regulated by BCL-2 through a direct interaction with Parkin [82]. A BH3 containing protein, BNIP3, functions as a mitophagy receptor that recruits autophagosomes by directly binding LC3 [83]. Beclin 1 also engages at the mitochondria with the other major anti-apoptotic BCL-2 family member, MCL-1, leading to inhibition of autophagy initiation [84,85]. Under nutrient rich conditions, siRNA depletion of MCL-1 or Beclin-1 results in stabilization of the other, and their degradation occurred through competitive binding to the USP9X deubiqutinase [84]. A number of BH3 proteins have been shown to regulate the apoptosis/autophagy rheostat, including BIM, BMF, and NOXA, and which serve to either mislocalize autophagy components or are themselves targets of autophagy [86–89].

Figure 3: Crosstalk between apoptosis and autophagy.

Apoptosis proteins can engage with the autophagy pathway throughout all steps of autophagosome formation, either by sequestering autophagy proteins away from active complex components, by associating with critical autophagy factors or cargo, by caspase-mediated cleavage of autophagy regulators, or by directly influencing autophagosome/lysosome function. Autophagy proteins are indicated in orange, with dual function apoptotic proteins highlighted in orange boxes with black shading.

Cleavage of autophagy proteins by caspases is another mechanism by which apoptosis and canonical autophagy are suggested to be differentially regulated, and may perhaps represent additional therapeutic targets for manipulating autophagy. Caspases have been shown to cleave Beclin 1 and Atg4D to inhibit autophagy and promote apoptosis [90,91]. Interestingly, a Crohn’s-disease associated mutation in Atg16L1 generates a new caspase-3 cleavage site, leading to compromised xenophagy of Yersinia entercolitica [92]. In rice, the glucosidase II β subunit Gas2 is downregulated to initiate autophagy in response to mild ER stress, while severe ER stress results in Gas2 cleavage by caspase-3 to participate in cell death [93]. Further down the apoptotic pathway, caspase-9 forms a complex with Atg7 to promote early events leading to autophagosome formation independently of its protease activity [94]. Remarkably, knockdown of caspase-9 expression in BAX^−/− HCT116 cells (BAX being the predominant pro-death protein in this cell type [95]) conferred sensitivity to death after autophagy induction by TRAIL treatment or nutrient deprivation, and caspase-9 knockout MEFs showed reduced autophagosome formation under basal and stimulated conditions, strongly indicating that caspase-9 possesses nonapoptotic functions in regulating autophagy [94].

The complex crosstalk between autophagic processes and apoptosis is apparent from observations that autophagy can act as a bystander witness to a cell’s demise (autophagy-associated), an accomplice (autophagy-mediated) or even as the murderer itself (autophagy-dependent) [96,97]. In autophagy-associated cell death, an autophagic response occurs concurrently with a cell death process (usually apoptosis) but does not directly participate in the death pathway. Such responses are typically interpreted as the cell’s effort to maintain essential functions in the face of cellular stress [98]. Autophagy-mediated cell death requires the activation or utilization of autophagic processes/proteins to directly promote a subsequent apoptotic or necroptotic response, such as through the degradation of survival factors or by facilitating the assembly of cell death complexes. Finally, autophagy-dependent cell death (ADCD) requires the establishment of causality, such that cell death is a direct consequence of autophagy and is completely independent of other cell death processes. Accurately characterizing the involvement of autophagy in any death process is impossible simply by assessment of cellular morphology or autophagic flux, as all three autophagy-related death processes frequently display increased autophagic activity. Instead, precise genetic, biochemical, and pharmacological manipulation of essential autophagy processes are required for fully establishing the involvement of autophagy in regulated cell death [5,97].

Physiological ADCD in nonmammalian model organisms has become widely accepted to occur during development [39,99], while experimentally identifying ADCD processes in mammalian cells has been challenging due to higher organismal complexity and the interconnected nature of cell death pathways [100]. ADCD has been observed during midgut regression in Drosophila [101] and germ cell and ventral cord neuron development in C. elegans [102]. Lacking caspases and other apoptotic machinery, the slime mold Dictyostelium discoideum requires ADCD for development of the fruiting body [103]. Although analogous evidence in mammalian cells is highly context dependent, several compelling examples of ADCD have been described. Recently, a novel form of ADCD known as autosis, was described that occurs in response to autophagy-inducing peptides, starvation, or hypoxia/ischemia [104,105]. Autosis requires core components of the autophagy pathway (knockdown of Beclin 1, Atg13 and Atg14 inhibited death), and cell death proceeds via a mechanism independent of apoptosis or necroptosis [104–106]. Interestingly, autosis was also inhibited by cardiac glycosides, suggesting a yet to be characterized role for the Na⁺,K⁺-ATPase in mediating this form of ADCD [104]. In MEFs, abolishing apoptosis by genetic knockout of BAX and BAK lead to nonapoptotic cell death after treatment with etoposide or staurosporine [107]. This nonapoptotic cell death was inhibited after treatment with an autophagy inhibitor (3-methyl adenine) or by knockdown of the autophagy genes Atg5 or Beclin 1, suggesting that the death process in BAX/BAK double knockout (DKO) cells was due to ADCD [107]. In the context of mouse development, genetic deletion of Atg5, BAX, and BAK caused early embryonic lethality with an increased occurrence of brain malformations (exencephaly), indicating that ADCD partially compensates for the lack of developmental cell death in apoptosis-deficient mice [108]. Compared with DKO mice, TKO mice also had a significant delay in interdigital web elimination, a hallmark trait of mice lacking pro-apoptotic proteins BAX and BAK, confirming that ADCD plays a role in the embryonic development of DKO mice [108].

Can autophagy pathways be therapeutically manipulated to tip the balance from a survival function toward ADCD in apoptosis-deficient cancers? This remains under investigation as a screen of over 14000 compounds failed to identify any single agent that could enhance autophagic flux past a toxic threshold to induce ADCD in U2OS cells [109]. Furthering our understanding of the detailed mechanisms of ADCD, particularly any differences between ADCD and canonical autophagy, may yield novel strategies for bypassing the survival function of autophagy and specifically stimulate a death activity. For example, blocking autophagosome fusion with the lysosome does not prevent death due to autosis [104], indicating that this death process does not proceed along the canonical downstream autophagy pathway. Despite the developmental requirement for ADCD in the Drosophila midgut, the autophagic death process does not require components of the Atg8 lipid conjugation pathway, Atg7 and Atg3, instead relying on the E1 ubiquitin ligase Uba1 [110]. Anticancer drugs that possess ADCD-inducing activities include the BH3 mimetics obatoclax and gossypol, which disrupt the inhibitory interaction between Beclin 1 and anti-apoptotic BCL-2 proteins allowing Beclin 1 to initiate autophagosome formation as part of the VPS34 PI3K complex (Figure 3) [111].

Given that autophagy processes have been shown in some cases to promote tumor survival, contribute to metastasis, and mediate resistance [112], it might be somewhat counterintuitive to activate an autophagic pathway for the treatment of cancers. Autophagy inhibition is also a therapeutic strategy being tested in cancer patients, with moderate responses observed in cases of glioblastoma, melanoma, and certain types of brain cancer after treatment with the lysosomal inhibitor, chloroquine, in combination with radiation and/or alkylating chemotherapeutics [112]. Thus, selective regulation of ADCD and autophagy-related processes may provide a novel therapeutic approach for cancer therapy.

Crosstalk between apoptosis and pyroptosis.

Another emerging death pathway that has received much attention is the inflammatory process known as pyroptosis. Pyroptosis is a form of cell death most commonly occurring as an innate immune response to pattern recognition receptor (PRR) activation, leading to an inflammatory response [113]. The presence of intracellular toxins, bacterial or viral infections leads to the assembly of a complex of factors collectively known as inflammasomes, which act as a molecular scaffold to mediate the cleavage and activation of the inflammatory caspase-1. Alternatively, sensing of bacterial LPS challenge leads to an inflammasome-independent activation of other inflammatory caspases-4,−5,or −11 [114]. Active inflammatory caspases lead to cleavage of gasdermin D, that when cleaved assembles into pores on the plasma membrane [115]. Plasma membrane poration by gasdermin D leads to cell swelling and lysis, which may mediate the release of the proinflammatory cytokine IL-1β (although gasdermin independent IL1-β release has been described [116]). Although the inflammatory caspases have no known role in the canonical apoptosis pathway, new evidence has brought to light that cleavage of gasdermin family members by apoptotic caspases initiates pyroptosis. The activation of pyroptosis by caspase-3 is dependent on the expression level of endogenous gasdermin E, and gasdermin E knockout mice are protected from lung and intestinal toxicity that results after treatment with cisplatin, a chemotherapeutic believed to induce caspase-3 dependent apoptosis in tumors. Thus, toxicity associated with chemotherapy drugs may be due to a secondary pyroptotic death process in cells expressing gasdermin E [117,118]. Additionally, many cancers epigenetically silence gasdermin E expression as another means to bypass endogenous cell death processes [119–121], suggesting that therapeutic reactivation of pyroptosis may hold promise in the treatment of tumors refractory to conventional chemotherapeutics. Caspase-8 has also been reported to cleave gasdermin D and gasdermin E in response to Yersinia infection [122,123]. Finally, endogenous gasdermin D directly binds BCL-2 through a putative BH3 motif, effectively altering the specificity of its cleavage by the inflammatory caspases to generate a fragment less competent to induce pyroptotic cell death [54].

Crosstalk between apoptosis and mPTP-mediated necrosis.

In response to ischemic injury or neurodegenerative processes, neurons and cardiomyocytes succumb to a form of cell death known as mitochondrial permeability transition pore (mPTP)-mediated necrosis. Distinct from the outer mitochondrial membrane permeabilization induced by apoptosis, the hallmark characteristic of mPTP-mediated necrosis is the formation of inner membrane pores that allow the free passage of ions and solutes up to 1.5 kD [124–126]. Triggered by mitochondrial calcium overload, this breech of inner mitochondrial membrane integrity leads to disastrous consequences for the cell, ultimately sending the cell into a bioenergetic crisis by disrupting the electrochemical proton gradient and ceasing respiration. After decades of intense research, the molecular composition of the mPTP pore/channel was recently confirmed. Previously, only regulatory components were identified including cyclophilin D (CypD), a chaperone that when inhibited by cyclosporin A (CsA) prevents mPTP-mediated necrosis, and the adenine nucleotide translocase (ANT) [125]. Other candidate pore complex regulators included the voltage-dependent anion channels (VDAC), the phosphate carrier (PiC), and the translocator protein (TSPO), but genetic analysis failed to prove essential functions for these components in mPTP-mediated necrosis [125]. A turning point in the field was the revelation that the F1F0 ATP synthase could function as a component of the mPTP pore in one of two ways, by passing solutes either through the border between two ATP synthase monomers or through the c-subunit ring that spans the inner mitochondrial membrane [127,128], but this remains controversial [129]. In two independent studies, highly purified ATP synthase complexes were recently confirmed to possess channel activity consistent with the biophysical characteristics of the mPTP, strongly supporting the identification of the ATP synthase as the sole component of the mPTP pore [130,131]. Further, monomeric complexes were sufficient to mediate mPTP-like channel activity [131], thus distinguishing the channel functions of ATP synthase monomers from the role of ATP synthase dimers in mitochondrial cristae formation [132].

Since proapoptotic BCL-2 family members are famous for forming pores in mitochondrial membranes, it seemed reasonable to assess whether they play a role in mPTP-mediated necrosis. BAX and BAK have been reported to directly interact with and influence the activity of mPTP regulators ANT and VDAC [133,134]. In mice lacking BAX and BAK, mPTP-mediated cell death and associated mitochondrial damage is inhibited in cardiomyocytes after myocardial infarction [135]. Similar results were described in mouse embryonic fibroblasts genetically deleted for BAX and BAK, which are resistant to OMM permeabilization and cell death after pharmacological stimulation of mitochondrial permeability transition, despite ongoing permeabilization of the inner membrane [136]. While necessary for pore formation in the outer membrane after mPTP, BAX and BAK are not directly involved in the CypD-regulated inner membrane permeabilization.

Conversely, antiapoptotic BCL-2 family members can also influence mPTP activity through direct interactions with regulatory components. While initially believed to exist solely on the outer mitochondrial membrane, a role for BCL-xL on the inner membrane as a regulator of the ATP synthase is becoming widely accepted [137,138]. BCL-xL has also been reported to influence ATP/ADP exchange across the outer mitochondrial membrane by way of ANT and VDAC [139,140]. Pharmacological inhibition of BCL-2 and/or BCL-xL may also sensitize cells to mPTP-mediated necrosis [141,142]. The BH3 protein, BID, has also been reported to functionally interact with components of the mPTP to mediate necrosis [143]. Importantly, BCL-2 proteins are known to bind to calcium channels and calcium channel activity at the endoplasmic reticulum [144–146] or mount an apoptotic response to ER stress [147], and therefore may promote ER calcium release that triggers mPTP-mediated necrosis. Although BCL-2 proteins have not been identified as physical components of the mPTP pore, modulation of their activities may prove beneficial for regulating mPTP-mediated necrosis.

Crosstalk between apoptosis and ferroptosis.

Ferroptosis is another emerging cell death pathway whose molecular mechanisms are just beginning to be delineated. The hallmarks of ferroptotic cell death include the oxidation of polyunsaturated fatty acid (PUFA)-containing phospholipids, loss of the ability to repair and detoxify lipid peroxides, and the accumulation of redox-active iron [148]. Morphological features of ferroptosis that have been reported include mitochondrial shrinkage and rounding up of cells with plasma membrane rupture [37,149,150], but these observations remain controversial. A large number of small molecule inducers and inhibitors of ferroptosis have been identified, and recently it was discovered that monounsaturated fatty acids (MUFAs) can protect membranes by blocking lipid ROS accumulation [151]. Ferroptosis inducers have been used to sensitize cancer cells to traditional chemotherapeutic agents or to overcome drug resistance [152,153]. While several proteins have been implicated in mediating ferroptosis, it is clear that this mode of regulated cell death occurs in response to disruptions of normal cellular homeostatic processes, including but not limited to imbalances in iron metabolism and decreases in cellular antioxidants.

While much of the ferroptotic death pathway identified thus far exists independently of the classical apoptotic pathway, initiation of ferroptosis appears to influence the activity or localization of select apoptotic proteins. Notably, the small molecular ferroptosis inducer, erastin, induced caspase-dependent apoptosis in colorectal cancer cell lines [154], perhaps due to mitochondrial translocation of the BH3 protein, BID [155]. HT-22 cells genetically disrupted for BID were resistant to erastin-induced cell death, maintained normal mitochondrial morphology and respiration, and reduced lipid peroxide formation [155]. Crosstalk has also been observed between the tumor suppressor p53 and SLC7A11, a solute carrier receptor necessary for cystine import and subsequent antioxidant response [156]. During the induction of apoptosis, the tumor suppressor p53 becomes stabilized upon integrating cellular stress signals, leading to the transcriptional activation of several apoptotic regulators, including BIM, NOXA and PUMA [157]. P53 can also directly repress SLC7A11 expression, leading to increased sensitivity to erastin, and p53 knockout MEFs were resistant to erastin-induced ferroptosis [156]. Another target of p53-mediated transcriptional regulation, SAT1, was found to induce lipid peroxidation and promote ferroptosis in response to ROS stress [158]. Thus, in addition to regulating apoptosis, p53 appears to coordinate the cellular response to ferroptotic death stimuli by promoting lipid peroxidation and decreasing antioxidant response [159]. Ferroptosis induction also results in upregulation of the BH3 protein, PUMA, independently of p53, but the consequence of PUMA expression is unknown [160,161]. Another p53 family member, ΔNp63α, also inhibits ferroptosis by transcriptionally upregulating genes involved in glutathione metabolism [162]. As the pathways regulating ferroptosis are more precisely delineated, the regulators controlling ferroptosis and apoptosis are likely to become more intertwined.

Conclusions and future perspectives

In the ever expanding and rapidly changing field of cell death, surprises still loom on the horizon, which will again reshape our thinking of the apoptosis pathway. Although apoptosis is commonly believed to occur independently of immune signaling, two recent paradigm-shifting studies have described a novel cell death mechanism in macrophages in which initiation of an apoptotic pathway leads to a downstream inflammatory response [163,164]. In LPS-primed macrophages, genetic or pharmacological inhibition of anti-apoptotic BCL-2 proteins leads to BAX/BAK-mediated MOMP and release of inhibitors of IAP proteins (such as Smac/DIABLO). The resulting IAP depletion leads to the activation of the caspase-8:RIPK1 complex to promote cleavage and release of the inflammatory cytokine IL-1β [163,164]. Surprisingly, BAX/BAK-mediated IL-1β release is independent of the canonical pyroptosis effectors, gasdermin-D and -E; however the role of the NLRP3 inflammasome in this process is disputed [163,164]. The unexpected implication that apoptosis proteins can mediate inflammatory responses awakens a new appreciation for the complexity and interconnectivity of these death pathways. While this seemingly “inflammatory apoptosis” process was observed in LPS-activated macrophages, it remains to be determined whether apoptotic regulators can engage in additional novel cell death pathways in other cell types or tissues.

It is not coincidental that mammalian cells have evolved mechanisms to coregulate independent death pathways by the same molecular factors, surely with the intention of installing a fail-safe mechanism for ensuring the success of an initiated death program or by amplifying PCD signals. As we learn more about the detailed mechanisms of nonapoptotic cell death pathways, more opportunities for crosstalk with apoptotic regulators will likely emerge as possible targets for therapeutic interventions. Since many tumors find ways of inactivating the apoptotic pathway [18], stimulating an alternative death pathway that initiates a localized inflammatory response may more effectively arm the body to combat cancer. Identifying common regulatory factors that participate in multiple death pathways, such as caspase-8 functioning in apoptosis, necroptosis, and pyroptosis, are likely to facilitate the development and implementation of novel strategies that alter their response in favor of progressing down other death pathways.

Abbreviations:

ADCD

autophagy-dependent cell death

AML

acute myeloid leukemia

ANT

adenine nucleotide translocator

ATP

adenosine triphosphate

BCL-2

B-cell lymphoma-2

BH3

BCL-2 homology-3

BIR

baculoviral IAP repeat

cIAP

cellular inhibitor of apoptosis protein

CLL

chronic lymphocytic leukemia

CsA

cyclosporin A

CypD

cyclophilin D

DD

death domain

DED

death effector domain

DISC

death-inducing signaling complex

DKO

double knock out

DR

death receptor

ER

endoplasmic reticulum

FADD

Fas-associated death domain

IAP

inhibitor of apoptosis protein

IgH

immunoglobulin H

MEF

mouse embryonic fibroblasts

MLKL

mixed-lineage kinase-like

MOMP

mitochondrial outer membrane permeabilization

mPTP

mitochondrial permeability transition pore

MUFA

monounsaturated fatty acid

PI3K

phosphatidylinositide 3-kinase

PiC

phosphate carrier

PRR

pattern recognition receptor

PUFA

polyunsaturated fatty acids

RIPK

receptor-interacting serine/threonine kinase

ROS

reactive oxygen species

Smac/DIABLO

second mitochondria-derived activator of caspases/direct IAP-binding protein with low Pi

TNF

tumor necrosis factor

TNFR

tumor necrosis factor receptor

TRAIL

TNF-related apoptosis-inducing factor

TSPO

translocator protein

VDAC

voltage-dependent anion channel

XIAP

X-chromosome-linked IAP