No death without life: vital functions of apoptotic effectors

Abstract

As a result of the genetic experiments performed in Caenorhabditis elegans, it has been tacitly assumed that the core proteins of the ‘apoptotic machinery’ (CED-3, -4, -9 and EGL-1) would be solely involved in cell death regulation/execution and would not exert any functions outside of the cell death realm. However, multiple studies indicate that the mammalian orthologs of these C. elegans proteins (i.e. caspases, Apaf-1 and multidomain proteins of the Bcl-2 family) participate in cell death-unrelated processes. Similarly, loss-of-function mutations of ced-4 compromise the mitotic arrest of DNA-damaged germline cells from adult nematodes, even in a context in which the apoptotic machinery is inoperative (for instance due to mutations of egl-1 or ced-3). Moreover, EGL-1 is required for the activation of autophagy in starved nematodes. Finally, the depletion of caspase-independent death effectors, such as apoptosis-inducing factor (AIF) and endonuclease G, provokes cell death-independent consequences, both in mammals and in yeast (Saccharomyces cerevisiae). These results corroborate the conjecture that any kind of protein that has previously been specifically implicated in apoptosis might have a phylogenetically conserved apoptosis-unrelated function, most likely as part of an adaptive response to cellular stress.

Main

‘Programmed cell death’ (PCD), an expression that was admirably coined by Richard Lockshin in the 1960s,¹ is often resolved by apoptosis, a modality of cell death that is defined by characteristic biochemical and morphological features such as chromatin condensation (pyknosis) and nuclear fragmentation (karyorrhexis).^{2, 3, 4, 5} Perhaps, owing to our bilateral symmetry or our intrinsic intellectual limitations, we tend to categorize elements in dualistic terms. According to this one-dimensional and simplistic logic, cell biologists, biochemists and geneticists have assumed that the processes ensuring cell survival and those involved in cell death would rather be diametrically opposed than overlapping.

In apparent accord with this conjecture, genetic screens in Caenorhabditis elegans revealed the existence of genes that are required for developmental cell death of the nematode (and hence named ced genes, for ‘C. elegans death’), yet had no discernible function in its ordinary life.⁶ The concept was born that the apoptotic machinery of C. elegans, as built up by CED-3 (a caspase), CED-4 (an adaptor molecule), CED-9 (a Bcl-2/Bcl-X_L ortholog) and EGL-1 (a Bcl-2 homology domain (BH3)-only protein), would solely exist for the regulation and execution of cell death, and would have no important roles in normal life. This concept was tacitly transferred to the mammalian system when effector caspases (i.e. a group of cysteine-aspartyl-proteases that account for the execution of cell death) were presumed to take part only in cell death and not to be involved in any death-unrelated process.⁷ Similarly, for a long time, it has been assumed that proapoptotic proteins from the Bcl-2 family as well as apoptosis protease-activating factor 1 (Apaf-1) (the mammalian equivalent of CED-4) would play a part only in self-destructive phenomena.

During recent years, however, a paradigm shift has occurred as it becomes clear that proteins involved in the induction of apoptosis or in the apoptotic dismantling of cells also exhibit cell death-unrelated functions.⁸ Indeed, it seems plausible that evolution has not ‘invented’ proapoptotic factors ex nihilo but rather has ‘appropriated’ molecules that already had a function in vital processes (such as adaptation to stress) into the service of PCD. This is the topic of the present review article.

Vital Functions of Mammalian Caspases

The first mammalian caspase to be cloned was caspase-1, which was initially named interleukin-1β-converting enzyme and hence constituted the first example of a ‘proinflammatory caspase’.⁹ Such enzymes (also known as group I caspases) participate in the maturation of cytokines and include caspase-1, -4 and -5 in humans, as well as caspase-1, -11 and -12 in mouse.¹⁰ For a long while, all other caspases have been classified either as ‘initiator’ or ‘executioner’ caspases,¹¹ and it was initially thought that they would only contribute to cell death,¹² a concept that nowadays is amply invalidated (Table 1).

Table 1 Apoptosis-regulatory and vital functions of caspases

Caspase-2 is the only caspase found in the nucleus and can be directly activated by DNA damage.¹³ Some data suggest that caspase-2 is involved in DNA repair. Thus, caspase-2^−/− mice age prematurely, a phenotype that is often associated with deficient DNA repair.¹⁴ Caspase-3 activation has also been implicated in the differentiation of multiple cell types. For instance, caspase-3 can participate in osteoblastic differentiation, and caspase-3-deficient mice exhibit delayed ossification and decreased bone mineral density.¹⁵

The involvement of caspases in differentiation was initially thought to be restricted to pathways involving partial self-destructive processes, in which entire organelles are eliminated. During terminal differentiation, some cell types, including erythroblasts, keratinocytes and lens epithelial cells, lose their nuclei and cytoplasmic organelles.^{16, 17} For example, erythropoiesis involves the sequential formation in the bone marrow of a series of red blood cell precursors (proerythroblasts, basophilic, polychromatophilic and orthochromatic erythroblasts), which extrude their nuclei and enter the circulation as mature, anucleate erythrocytes.¹⁸ Caspase inhibitors arrest the maturation of erythroid progenitors at early stages of differentiation, well before nuclear shrinkage and chromatin condensation occur.¹⁹ Effector caspases such as caspase-3 are transiently activated through the mitochondrial pathway during erythroblast differentiation.¹⁹ In this context, caspase-3 cleaves proteins involved in nuclear envelope integrity (e.g. lamin B) and chromatin condensation (e.g. acinus), but spares other potential substrates (such as the major erythroid transcription factor GATA-1), presumably because these proteins are protected by the interaction with HSP70.^{19, 20} Interestingly, it has been suggested that caspase-3 involvement in erythropoiesis is restricted to the early phases of proerythroblast differentiation, with no major roles in the nuclear substructure reorganization and nuclear extrusion that characterize the late phases of the process.²¹

Caspase activation is also involved in thrombopoiesis. Platelets are formed from mature megakaryocytes and arise from the budding of thin, long cytoplasmic extensions called proplatelets. This process requires both localized caspase-3 and -9 processing and mitochondrial membrane permeabilization (MMP) and cytochrome c (Cyt c) release within the central body of the cell.²² Caspase inhibitors, as well as Bcl-2 overexpression, block platelet formation in vitro.²³ Importantly, in the context of proplatelet generation, the activation of caspases is confined to granular, perinuclear structures. Conversely, widespread caspase activation is associated with megakaryocyte death and inefficient thrombopoiesis, as it occurs in myelodysplastic syndromes.^{23, 24}

Furthermore, caspases play a role in the differentiation of human blood monocytes into macrophages, a process exhibiting no morphological signs of apoptosis.¹⁹ In this context, caspase activation involves Cyt c release from mitochondria and leads to the cleavage of a specific subset of caspase substrates, including the protein acinus but not poly (ADP-ribose) polymerase 1 (PARP-1).²⁵ Monocytic-macrophagic differentiation is repressed by pharmacological inhibitors of caspases as well as by overexpression of Bcl-2.²⁵ Apparently, the apical caspase involved in the macrophagic maturation of monocytes is caspase-8, which acts by cleaving the serine/threonine kinase receptor-interacting protein 1 (RIP1), thereby preventing sustained NF-κB activation and setting off downstream caspases.²⁶ Notably, caspase-8 has also been implicated in the differentiation of the human placental trophoblast, by mediating the syncytial fusion of the cytotrophoblast that accounts for the generation of the syncytiotrophoblast.²⁷

Loss-of-function mutations in the human caspase-8 gene cause defects in the activation of T, B and NK cells, culminating in immunodeficiency.²⁸ Similarly, knockout of caspase-8 (or that of its cytosolic adaptor FADD, i.e. Fas-associating death domain-containing protein) in mice results in impaired heart muscle development and defects in the immune system, particularly the hematopoietic precursor and T-cell progenitor compartments.²⁹ Moreover, mice carrying a T-cell-specific inactivation of caspase-8 develop impaired activation-induced expansion of peripheral T cells and an inability to clear lymphocyte choriomeningitis virus.²⁹ Caspase activation linked to the activation of T or B lymphocytes reportedly results in the intracellular proteolysis of a restricted panel of substrates, including PARP-1, lamin B and the kinase Wee1 (but not the DNA fragmentation factor subunit of 45 kDa, i.e. DFF45, nor replication factor C 140, i.e. RFC140, both of which are frequently cleaved in apoptosis).³⁰ In mice, liver-specific inactivation of caspase-8 attenuates the first wave of hepatocyte proliferation after partial hepatectomy.³¹ However, depending on the mouse genetic background, this may prompt an inflammatory response that eventually leads to enhanced proliferation and hepatomegaly.³¹ While caspase-8 deletion in bone marrow cells results in the functional impairment of hematopoietic progenitors, caspase-8 loss in cells of the myelomonocytic lineage leads to an arrest of macrophagic differentiation and cell death.³²

In humans, genetic alterations in caspase-10 may be causative or protective in type II autoimmune lymphoproliferative syndrome, most likely due to its role in the initiation of the extrinsic apoptotic pathway.³³ Furthermore, caspase-10 has been implicated in the activation of NF-κB-dependent pro-survival signaling pathways, by a mechanism that may not require its catalytic domain.³⁴

Caspase-11 was first described as an obligate activator of caspase-1.³⁵ More recently, caspase-11 has been reported to interact physically and functionally with actin-interacting protein 1 (Aip1), an activator of cofilin-mediated actin depolymerization.³⁶ This interaction is mediated by the caspase-recruitment domain (CARD) of caspase-11 and the C-terminal WD40 propeller domain of Aip1.³⁶ Thus, cells lacking Aip1 or caspase-11 exhibit similar defects in actin dynamics.³⁶

Murine caspase-12 was initially reported to play a role in apoptosis induced by endoplasmic reticulum (ER) stress including disruption of Ca²⁺ homeostasis and accumulation of misfolded proteins.³⁷ Human caspase-12 may attenuate the inflammatory and innate immune response to endotoxins, and hence the loss of caspase-12 function may constitute a risk factor for developing sepsis.³⁸ In humans, a single nucleotide polymorphism in the caspase-12 gene is responsible for the synthesis of either a truncated (Csp12-S) or a full-length proenzyme (Csp12-L). Interestingly, the read-through polymorphism resulting in the production of Csp12-L is confined to populations of African descent, and the frequency of the Csp12-L allele is particular high in African-American individuals characterized by severe septic responses.³⁸

The activation of caspase-14 (whose expression is constitutively high during embryonic development but almost exclusively restricted to the suprabasal layers of the epidermis and the hair follicles in adult mice and humans) has been associated with the terminal differentiation of human keratinocytes and cornification.³⁹ Caspase-14^−/− mice exhibit an altered composition of the stratum corneum, presumably due to an aberrant processing of filaggrin.⁴⁰ This results in reduced skin-hydration levels and enhanced sensitivity of the skin to UVB-induced photodamage and apoptosis.⁴⁰

Altogether, these examples (Table 1) illustrate that most if not all caspases have cell death-unrelated functions.

Vital Functions of Mitochondrial Death Effectors

Mitochondria are crucial organelles in a cell's life and death. On the one hand, they act as major regulators of mitochondrial apoptosis, by integrating pro-survival and pro-death signals and ultimately sealing the cell's fate via MMP.^{3, 41, 42} On the other hand, they generate the bulk of intracellular ATP via oxidative phosphorylation, and participate in multiple biosynthetic pathways.⁴³ Most mitochondrial death effectors exert also cell death-unrelated functions (Table 2).

Table 2 Apoptosis-related and -unrelated functions of non-caspase cell death effectors in mammals

Although Bcl-2 family proteins were initially characterized as cell death regulators, it has recently become clear that several members also control autophagy, either as part of a cell death or cell survival program.^{5, 44, 45} Thus, antiapoptotic proteins such as Bcl-2, Bcl-X_L, Bcl-w and Mcl-1 inhibit autophagy, presumably because they bind to the BH3 domain of Beclin 1,⁴⁶ an essential autophagy protein, thereby inhibiting the capacity of Beclin 1 to activate the phosphoinositide-3-kinase Vps34 (which participates in phagophore nucleation).^{45, 47, 48, 49} Conversely, proapoptotic BH3-only proteins from the Bcl-2 family such as BNIP3, Bad, Noxa, p53-upregulated modulator of apoptosis (Puma), Bim_EL and Bik can induce autophagy, likewise because they competitively disrupt the above-mentioned inhibitory interaction between their antiapoptotic relatives (e.g. Bcl-2, Bcl-X_L, and so on) and Beclin 1.^{48, 50, 51} Importantly, both antiapoptotic and proapoptotic proteins from the Bcl-2 family contribute to the modulation of Ca²⁺ signaling at the ER.⁵² In this context, Bcl-2 and Bcl-X_L have been reported to lower the luminal steady-state concentration of Ca²⁺, by directly promoting Ca²⁺ leak into the cytosol,⁵³, by gating the response of the inositol-1,4,5-trisphosphate receptor (IP₃R) to inositol-1,4,5-trisphosphate,⁵⁴ or by destabilizing the sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA).⁵⁵ Irrespective of their specific mechanism of action, Bcl-2 and Bcl-X_L act at the ER by dampening IP₃R-mediated Ca²⁺ release.⁵⁶ Conversely, Bax and Bak enhance the release of Ca²⁺ from ER stores, via a mechanism that implicates the SERCA.⁵⁷ Owing to the established role of Ca²⁺ overload in MMP,⁵⁸ the control of Ca²⁺ fluxes by members of the Bcl-2 family has been principally studied in the context of apoptosis.⁵⁷ However, since Ca²⁺ signaling impacts so many areas of cell biology, it seems appropriate to consider the relationship between Bcl-2-like proteins and Ca²⁺ from a broader point of view, which includes a plethora of apoptosis-unrelated processes (e.g. differentiation, regeneration, autophagy).^{59, 60} A similar consideration holds true for the role of proapoptotic Bcl-2 family members in the regulation of mitochondrial morphology and dynamics.⁶¹ Thus, whereas the link between Bak, Bax and Bik and components of the mitochondrial fission/fusion machinery (e.g. dynamin-like protein 1, i.e. Drp1, hFis1, mitofusins)^{62, 63} has been extensively characterized during cell death, it cannot be excluded that this interaction might as well account for apoptosis-unrelated changes in mitochondrial dynamics. As a final example, Bid, another BH3-only protein involved in the crosstalk between the extrinsic and intrinsic apoptotic pathways,⁶⁴ has also been suggested to take part in the DNA damage response activated by ATM.⁶⁵

The prototype of non-caspase apoptotic effectors characterized by well-defined vital roles is Cyt c.⁸ In healthy cells, Cyt c is associated with the outer surface of the mitochondrial inner membrane (IM), where it functions as an electron shuttle between complex III and complex IV of the respiratory chain.⁶⁶ This activity of Cyt c is necessary for life, as indicated by the fact that knockout mice embryos die in utero by midgestation.⁶⁷ After apoptosis-related mitochondrial outer membrane permeabilization (MOMP), Cyt c is released into the cytosol, where it interacts with Apaf-1 and dATP to form a large heptameric complex (the so-called apoptosome) that recruits and allosterically activates caspase-9 and hence sets off the caspase cascade.^{68, 69} However, cytosolic Cyt c has been implicated also in a number of cell death-unrelated processes including the fragmentation of mature megakaryocytes,²³ monocytic-macrophagic differentiation,²⁵ Drosophila melanogaster sperm cell differentiation,⁷⁰ and B-cell homeostasis.⁷¹ In this context, whereas the sustained release of Cyt c following irreversible MOMP activates caspase-dependent apoptosis, lower amounts of cytosolic Cyt c may promote limited caspase activation, and hence the cleavage of a restricted subset of substrates involved in cell death-unrelated processes.⁷²

Apoptosis-inducing factor (AIF) is a phylogenetically ancient protein essential for survival.⁷³ Murine AIF is synthesized from a nuclear gene as an immature precursor with a ∼100 amino-acid long N-terminal mitochondrial localization signal (MLS). Upon import into mitochondria, the MLS is removed and AIF inserts into the IM via an N-terminal transmembrane region. The rest of the protein, which refolds and incorporates flavine adenine nucleotide as a prosthetic group required for its NADH oxidase activity,^{74, 75} faces the mitochondrial intermembrane space (IMS).⁷⁶ After MOMP, AIF translocates from mitochondria to the cytosol and eventually is imported into the nucleus (together with its obligate cofactor cyclophilin A), where it participates in chromatin condensation and DNA degradation.^{77, 78} Mutational and biochemical analysis of AIF indicate that its apoptotic and redox functions reside in distinct domains of the protein.⁷⁹ Importantly, murine aif knockout causes a defect in oxidative phosphorylation, mainly due to the downregulation of components of complex I of the respiratory chain.^{80, 81} This explain why, in the Harlequin mutant mouse, reduced AIF expression (due to a retroviral insertion into the first intron of the aif gene) leads to cerebellar and retinal neurodegeneration.⁸² Indeed, cells expressing little or no AIF are particularly vulnerable to oxidative stress, in line with the notion that AIF has an antioxidant function.⁸² Tissue-specific deletion of aif in the muscle or liver provokes increased glycolytic rates, insulin hypersensitivity and significant resistance to diabetes and high lipid diet-induced obesity,⁸³ underscoring a major role for AIF in apoptosis-unrelated mitochondrial metabolism.

Endonuclease G (EndoG) is a nuclear DNA-encoded nuclease that normally resides in IMS. After MOMP, EndoG translocates to the nucleus, where it mediates oligonucleosomal DNA fragmentation independently of caspases.⁸⁴ Importantly, EndoG is also involved in DNA recombination, and its deficiency reduces cell proliferation and favors the arrest of cells in the G₂ phase of the cell cycle.⁸⁵ In accord with the observation that recombination-dependent DNA repair is essential for the survival of tetraploid cells,⁸⁶ EndoG knockdown causes the death of tetraploid tumor cells.⁸⁷

The stress-activated endoprotease Omi stress-regulated endoprotease (Omi) (also known as high temperature requirement protein A 2, i.e. HtrA2) belongs to a family of serine proteases that is well conserved from bacteria to humans.⁸⁸ In bacteria, HtrA2 is localized within the periplasmic space and determines thermotolerance,⁸⁹ whereas in healthy eukaryotic cells Omi/HtrA2 is confined to the IMS. After MOMP, the protease is released into the cytosol and promotes apoptosis via caspase-dependent and -independent mechanisms.⁹⁰ Thus, Omi/HtrA2 indirectly favors the activation of caspases by sequestering and cleaving inhibitor of apoptosis proteins (IAPs),^{91, 92} but also contributes to the execution of apoptosis via the cleavage of caspase-unrelated substrates like cytoskeletal proteins.⁹³ However, Omi/HtrA2 has also been suggested to act as a negative regulator of cell cycle progression during interphase, presumably through the proteolytic processing of the mitotic kinase WTS/large tumor-suppressor 1 (WARTS).⁹⁴ Interestingly, a loss-of-function mutation in omi (Ser276Cys) has been shown to underlie the pathology of the mnd2 (for motor neuron degeneration 2) mouse strain, which exhibit early onset neurodegeneration with parkinsonian features and juvenile lethality.⁹⁵ The same phenotype is observed in omi^−/− mice, suggesting that the most important function of Omi/HtrA2 in vivo relates more to protection against stress than to apoptosis.⁹⁶ Mitochondria purified from mnd2 mice are more susceptible to Ca²⁺-mediated permeabilization in vitro than control organelles purified from wild-type animals. Thus, almost paradoxically, the loss of Omi/HtrA2 enhances the susceptibility to apoptosis, thereby provoking the degeneration of striatal neurons in mnd2 mice.⁹⁵ Moreover, it has been recently found that Omi/HtrA2 is activated through phosphorylation by PTEN-induced putative kinase 1, a putative mitochondrial protein kinase that is mutated in some cases of familial early-onset Parkinson's disease.⁹⁷ Finally, it has also been shown that the protease activity of Omi/HtrA2 is necessary for the physiological processing of β-amyloid precursor protein at mitochondria, thus pointing to Omi/HtrA2 as a possible therapeutic target for Alzheimer's disease.⁹⁸ Thus, multiple links exist between reduced Omi/HtrA2 activity and neurodegeneration.

Murine second mitochondria-derived activator of caspase (Smac) and its human ortholog direct IAP-binding protein with a low pI (DIABLO) are encoded by the nuclear genome and synthesized as an immature precursor that harbors an N-terminal MLS.^{99, 100} After mitochondrial import, MLS is proteolytically removed to yield a mature polypeptide of 23 kDa localized in the IMS and exposing an IAP-binding domain.¹⁰⁰ After MOMP, Smac/DIABLO is released into the cytosol, homodimerizes and promotes apoptosis by sequestering different members of the IAP family, thereby favoring caspase activation.^{101, 102} Notably, cell death-unrelated functions for Smac/DIABLO have not been unambiguously identified yet. Smac^−/− mice are viable, grow normally and exhibit no major phenotypic alterations.¹⁰³ Moreover, smac^−/− cells responded normally to multiple apoptotic stimuli. These observations suggest the existence of a redundant molecule compensating for Smac/DIABLO loss in vivo, with regards to both its lethal and vital functions.¹⁰³ Ectopic overexpression of Smac/DIABLO in the cytosol has been reported to induce the arrest of cells at the G₀/G₁ cell cycle transition.¹⁰⁴ However, the actual significance of this observation in a physiological context remains to be established.

Vital Functions of Cell Death-Related Adaptor Molecules

In both the extrinsic (death receptor-mediated) and intrinsic (mitochondrial) apoptotic pathways, supramolecular complexes are assembled to facilitate the interaction (and hence the activation) of transducers of the lethal signal with molecules responsible for upstream (initiation) or downstream (execution) phases.³ So far, a number of adaptor proteins have been characterized that assist in the assembly of these complexes, including (i) Apaf-1, which contributes to the formation of the ‘apoptosome’ to activate caspase-9;⁶⁸ (ii) FADD and tumor necrosis factor receptor (TNFR)-associated death domain protein (TRADD), which recruit pro-caspase-8 at the plasma membrane within the death-induced signaling complex (DISC), thus transducing proapoptotic signals from the extracellular to the intracellular milieu;¹⁰⁵ (iii) p53-induced protein with a death domain (PIDD) and caspase-2 and RIPK1 domain containing adaptor with death domain (CRADD, also known as RIP-associated ICH-1/CED-3 homologous protein with a death, i.e. RAIDD), both participating in the assembly of the ‘PIDDosome’, a protein complex implicated in caspase-2 activation following genotoxic stress.^{106, 107} These factors are implicated also in several processes distinct from cell death (Table 2).

Knockdown of Apaf-1 in human cells as well as knockout of apaf-1 in mice implicated Apaf-1 in DNA damage-induced cell cycle arrest. Thus, Apaf-1 loss compromises the DNA damage checkpoints elicited by ionizing irradiation or chemotherapy.¹⁰⁸ Apaf-1 depletion also reduces the activating phosphorylation of the checkpoint kinase Chk1 provoked by DNA damage,^{108, 109} and knockdown of Chk1 abrogates Apaf-1-mediated cell cycle arrest. Morevoer, epistatic analyses revealed that Chk1 operates downstream of Apaf-1 to mediate the intra-S-phase DNA damage checkpoint.¹⁰⁸ Finally, the nuclear translocation of Apaf-1, which in vitro can be induced by exogenous DNA damaging agents in an ATM- or ATR-dependent fashion, correlates in vivo with the endogenous activation of Chk-1, as assessed in biopsies from non-small cell lung cancer patients.¹⁰⁸ The influence of Apaf-1 on the cell cycle is not modulated by pharmacological inhibitors of caspases like Z-VAD.fmk nor by antiapoptotic proteins such as the baculovirus-encoded IAP p35 and Bcl-2. Moreover, Apaf-1 mutants that lack the N-terminal CARD can replace endogenous Apaf-1 in the control of DNA damage-induced cell cycle-blockade. This indicates that the cell cycle-arresting function of Apaf-1 is independent of its caspase-activating (proapoptotic) role.¹⁰⁸

Beyond its role within the DISC, FADD has been implicated in a number of cell death-unrelated processes.¹¹⁰ Fadd^−/− mice die early during embryogenesis, with signs of cardiac failure and abdominal hemorrhage.¹¹¹ To circumvent this problem, transgenic mice expressing a dominant negative mutant of FADD (lacking the caspase-dimerizing death effector domain) have been generated. These animals exhibit retarded thymocyte development and peripheral lymphocyte pools devoid of T cells.¹¹² A similar phenotype is observed in transgenic mice engineered for the conditional, T-cell-specific knockout of fadd.¹¹³ Taken together, these observations highlight an essential role for FADD in the development and homeostasis of peripheral T cells. Several other studies point to the implication of FADD in proliferation and cell cycle progression, mainly in hematopoietic progenitors and cells of the lymphocytic lineage.^{114, 115} As a possibility, such functions of FADD may be accounted for by a nuclear pool of the protein,¹¹⁶ and may be regulated by cell cycle-dependent phosphorylation at multiple serine residues.¹¹⁷ Interestingly, FADD is also involved in multiple innate immunity signaling pathways, either dependent or independent from the Toll-like receptor 4.¹¹⁸ Upon TNFR activation, two sequential signaling complexes are formed, which may account for its dual role in promoting cell death and survival. Whereas complex I (which contains TRADD and RIP1 but not FADD) signals NF-κB to promote cell survival, complex II (involving FADD and caspase-8) triggers cell death.¹¹⁹ siRNA-mediated depletion of TRADD (gene knockout is incompatible with life) suggested that this adaptor is dispensable for necrosis induction along the TNFR–RIP1 axis, but required for the activation of both NF-κB and caspase-8.¹²⁰ Interestingly, a nuclear pool of TRADD has been implicated in both cell death and survival pathways, possibly via the interaction with the transcriptional factor Stat1.¹²¹

According to recent reports, PIDD represents a master switch in the response to DNA damage. On the one hand, PIDD (transactivated by p53) can cooperate with CRADD/RAIDD to assemble the PIDDosome, thereby activating caspase-2 and promoting apoptosis.^{106, 107} As an alternative, PIDD is able to enhance genotoxic stress-induced NF-κB activation by augmenting the sumoylation and ubiquitination of NF-κB essential modulator.¹²² This pro-survival role of PIDD depends on a 51-kDa C-terminal fragment including the death domain (PIDD-C), generated by an auto-proteolysis mechanism. Further processing of PIDD-C would then result in the formation of a 37-kDa fragment (PIDD-CC) responsible for caspase-2 activation. A non-cleavable PIDD mutant fails to translocate from the cytoplasm to the nucleus, thereby losing both activities.¹²³ So far, three isoforms of PIDD have been described, of which all are capable of activating NF-κB upon DNA damage, but only isoform 1 interacts with RAIDD/CRADD and activate caspase-2.¹²⁴ RAIDD/CRADD cell death-unrelated functions are still poorly characterized, but may include the regulation of (at least some) differentiation programs.¹²⁵ Notably, raidd^−/− mice are not viable, yet this presumably depends on RAIDD/CRADD involvement in apoptosis-mediated embryo remodeling rather than in other processes (as assessed by its expression pattern during organogenesis, which correlates with profound morphological changes occurring in the developing embryo).¹²⁶

Vital Functions of Components of the Permeability Transition Pore Complex

In some models of cell death characterized by excessive Ca²⁺ fluxes and reactive oxygen species overproduction, the cascade of events leading to MMP and apoptosis is set off at the IM, due to the activation of the permeability transition pore complex (PTPC), a large multiprotein complex formed at the contact sites between the mitochondrial outer membrane (OM) and IM.^{3, 127} Despite considerable effort to determine its exact molecular structure, the precise composition of the PTPC still remains elusive.¹²⁸ In this context, numerous studies suggest that the PTPC might be formed by the dynamic interaction of several partners, including the adenine nucleotide translocase (ANT, in the IM), the voltage-dependent anion channel (VDAC, in the OM), cyclophilin D (CypD, in the mitochondrial matrix), creatine kinase (CK, in the IMS), the peripheral-type benzodiazepine receptor (PBR, in OM) as well as hexokinase isoforms (HKI and HKII, in the cytosol).^{128, 129}

However, since ‘housekeeping’ genes (such as those coding for the majority of putative PTPC components) cannot be easily manipulated by genetic means (because their knockout is incompatible with life), PTPC constituents have been rather poorly investigated for their role in lethal processes, with the notable exception of CypD.^{127, 130, 131} Thus, mice knockout for all isoforms of VDAC,^{132, 133} ANT^{132, 134} and other putative PTPC components have not yet been generated, despite the fact that there is firm evidence (obtained in cell lines) that numerous PTPC components do indeed play a role in cell death.³ Nevertheless, all these proteins are known to participate into a number of mitochondrial and extramitochondrial metabolic pathways, reinforcing the concept that proapoptotic effectors also exert vital functions (Table 3). Thus, (i) ANT mediates ATP/ADP exchange between the mitochondrial matrix and the cytosol, thereby ensuring an adequate cytosolic energy supply while maintaining high rates of oxidative phosphorylation (by releasing substrate inhibition);¹³⁵ (ii)VDAC is the most abundant OM protein and, in healthy cells, exists as a large, voltage-gated channel accounting for OM permeability properties;¹³⁶ (iii) CypD exhibits a peptidyl-prolyl cis–trans isomerase activity, which contributes to the correct folding of mitochondrial matrix proteins;¹³⁷ (iv) CK catalyzes the ATP-dependent conversion of creatine into phosphocreatine, to constitute a highly diffusible intracellular energy buffer;¹³⁸ (v) PBR regulates cholesterol transport from OM to IM, the rate-limiting step in steroidogenesis;¹³⁹ and (vi) HK isoforms catalyze the production of glucose-6-phosphate, the first intermediate in glucose metabolism.¹⁴⁰

Table 3 Cell death-regulatory and normal functions of PTPC constituents

Vital Functions of Evolutionarily Distant Death Effectors

The cell death-unrelated functions of apoptotic proteins are so conserved among evolutionarily distant species that it has been possible to predict them in model organisms other than mammals, based on the data that had been obtained in the human and murine systems (Table 4).

Table 4 Lethal and vital functions of evolutionarily distant cell death effectors

Based on the observation that BH3-only proteins induce autophagy in mouse and human cells, we made the prediction that the sole BH3-only protein present in C. elegans, that is EGL-1, would also regulate autophagy. Indeed, gain-of-function mutations of egl-1 induce a maximum degree of autophagy that cannot be further enhanced by starvation, the physiological inducer of autophagy. Conversely, the egl-1 knockout strongly limits starvation-induced autophagic responses.⁵¹ In view of the fact that autophagy is mostly a cytoprotective mechanism,⁴⁴ these results suggest that BH3-only proteins may be required for an optimal adaptation to nutrient depletion. Recently, it has been proposed that EGL-1 also plays a role in the regulation of mitochondrial dynamics, by preventing CED-9 from binding to (and hence inhibiting) components of the mitochondrial fission/fusion machinery.¹⁴¹

Starting from the observation that Apaf-1 translocates to the nucleus of human DNA-damaged cells and participates in the intra-S-phase DNA damage checkpoint upstream of Chk1, we wanted to determine whether the C. elegans Apaf-1 ortholog CED-4 would have a similar function. Indeed, we found that CED-4 translocates to the nuclei of germline cells upon exposure of nematodes to ionizing irradiation, in an ATM- and ATL (the worm ortholog of ATR)-dependent fashion. Moreover, CED-4 was required for the proliferation arrest of C. elegans germline cells induced by γ-irradiation or UVC light.¹⁰⁸ Two independent lines of evidence indicate that the cell cycle-modulatory effect of CED-4 does not require the activation of apoptotic effectors: (i) the absence of CED-4 continues to reduce the DNA damage-induced cell-cycle arrest when the absence of proapoptotic proteins such as EGL-1 or CEP-1 (the C. elegans ortholog of p53) does not influence the cell cycle; and (ii) loss-of-function mutations in ced-4 affects the cell cycle also in a genetic background where CED-3 cannot be activated and hence caspase-dependent apoptosis does not occur.¹⁰⁸

Since mammalian AIF possesses a bona fide yeast ortholog¹⁴² and since AIF depletion causes a respiratory defect in mouse and human cells,⁸¹ we made the prediction that the knockout of yeast aif1 gene would also lead to deficient oxidative phosphorylation. Indeed, aif1⁻ yeast cells exhibited normal growth in rich media, yet proliferated less efficiently than isogenic controls in non-fermentable energy sources such as lactate or glycerol.⁸¹ This is remarkable because, in mammalian cells, AIF depletion causes mostly a defect in respiratory complex I, indicating a major phylogenetic conservation of the contribution of AIF to optimal mitochondrial function. As an aside, it should be noted that the knockout of the D. melanogaster homolog of aif1 also causes a major respiratory defect,⁷³ underscoring the importance of AIF in normal mitochondrial function.

EndoG has also been investigated for its putative cell death-unrelated roles in non-mammals organisms. Interestingly, it has been demonstrated that EndoG is required for the survival of tetraploid S. cerevisiae cells, an observation that was confirmed in tetraploid human colon cancer cell clones.⁸⁷

Finally, metacaspases (which may represent a phylogenetic ancestor of mammalian caspases) are not only implicated in various cell death scenarios,¹⁴³ but have been recently shown to modulate cell cycle progression and stress responses in fungal and protozoan models.^{144, 145}

Concluding Remarks

The results outlined above suggest that cell death regulators have vital (as opposed to the lethal) functions that are phylogenetically conserved. Although formal proof for this concept is lacking, it appears plausible that the cell death-unrelated function of such proteins is actually the most ancestral one and that the lethal function has been acquired later during evolution. In particular, it seems that apoptosis effectors exhibit vital functions that are prominently involved in the adaptation to stress such as redox stress (AIF), metabolic stress (BH3-only proteins/EGL-1), DNA damage (Apaf-1/CED-4, EndoG) or thermotolerance (Omi/HtrA2). This applies also to caspases, which play a role in inflammation and immunity, the host responses to pathogen invasion-induced stress.¹⁴⁶ Based on this concept, it is tempting to speculate that proteins involved in stress adaptation of individual cells might become potential death effectors later during evolution.

As a caveat to this speculation, however, it should be noted that the contribution to lethal processes of ‘housekeeping’ genes (which cannot be genetically manipulated because their depletion leads to cell death) is difficult to be investigated. Thus, it remains possible that – for methodological reasons – we are grossly underestimating the proteins and processes involved in the regulation and execution of cell death. The adventure that started in the 1960s with the examination of insect intersegmental muscle cells undergoing PCD¹ is only in its infancy. There is no death without life.