Caspase-independent cell death.pdf

NATURE MEDICINE VOLUME 11 | NUMBER 7 | JULY 2005 725

Caspase-independent cell deathGuido Kroemer1 & Seamus J Martin2

Caspase activation has been frequently viewed as synonymous with apoptotic cell death; however, caspases can also contribute to processes that do not culminate in cell demise. Moreover, inhibition of caspases can have cytoprotective effects. In a number of different models, caspase inhibition does not maintain cellular viability and instead shifts the morphology of death from apoptosis to nonapoptotic pathways. Here, we explore the contribution of caspases to cell death, either as upstream signals or as downstream effectors contributing to apoptotic morphology, as well as alternative strategies for cell death inhibition. Such alternative strategies may either target catabolic hydrolases or be aimed at preventing mitochondrial membrane permeabilization and its upstream triggers.

Cell death is central to a cornucopia of disease states, including acute and chronic degenerative processes. In recent years, the molecular mechanisms of at least one type of cell death, apoptosis, have been deciphered with unprecedented celerity. In nematodes and insects, most developmental cell death involves a unique class of cysteine proteases called caspases. In Caenorhabditis elegans and in Drosophila melanogas-ter, genetic inactivation or inhibition of caspases prevents cell death, leading to the persistence of many cells that otherwise would have died during development1,2. Nonetheless, even in C. elegans, some cells suc-cumb to developmental cell death in a caspase-independent fashion3. Although caspases are evolutionarily conserved (and frequently acti-vated during cell death)4,5, it seems that the equations ‘programmed cell death = apoptosis’ and ‘apoptosis = caspase activation’6 do not apply to mammals for several reasons.

First, mammalian cells can die through different, biochemically and morphologically distinct pathways; apoptosis represents only one particular mode of programmed cell death. Second, although certain caspases operate as signal transducers and catabolic enzymes in apop-tosis, a subset of caspases (caspases 1, 4 and 5 in humans) are involved in inflammatory cytokine production, and much evidence suggests that these enzymes are incapable of propagating cell death signals. So, cas-pase activation does not necessarily lead to apoptosis7. Third, caspase inhibition often does not prevent cell death, causing a shift to caspase-independent self-destruction processes. The primary reason for this, as we shall discuss, is that many stimuli that promote caspase-dependent apoptosis also provoke mitochondrial damage that cannot be reversed by blocking caspase activity. This implies that the inhibition of caspases has a limited therapeutic potential when the avoidance of unwarranted cell loss is the goal. In many cases, cytoprotection can only be achieved by targeting caspase-unrelated processes and molecules.

Here, we will critically evaluate the contribution of caspases and caspase-independent processes to cell death. Moreover, we will discuss the possibility of targeting caspase-independent cell death mechanisms for cytoprotection.

Caspase-mediated cell deathThe human genome encodes 12 to 13 distinct caspases (caspase 12 is a pseudogene in whites and is functional in a subset of individuals of African descent). Some of them function in cytokine processing and inflammation, and at least seven (caspases 2, 3, 6, 7, 8, 9 and 10) contrib-ute to cell death. Caspases that participate in apoptosis can be divided into the initiator caspase group, which includes caspases 2, 8, 9 and 10, and the effector caspase group, to which caspases 3, 6 and 7 belong.

Caspases are synthesized as proenzymes that have very low intrinsic activity and require activation, either by proteolytic maturation or by interaction with an allosteric activator. Initiator caspase activation can be achieved by a series of polyprotein complexes such as the inflammasome8, the piddosome9, the death-inducing signaling complex (DISC)10 and the apoptosome11. The inflammasome promotes activation of caspase 1, an enzyme required for the secretion of mature interleukin (IL)-1β8 and has little, if any, impact on apoptosis. The inflammasome also promotes activation of caspases 4 and 5, although their role in the inflammatory response remains poorly defined. The piddosome catalyzes the proteolytic maturation of procaspase 2 in response to nuclear DNA damage9. The DISC is generated upon ligation of death receptors such as CD95-Fas and stimulates the proteolytic maturation of procaspase 8 (ref. 10). The DISC represents the caspase-activating complex that is triggered by the ‘extrinsic’ apoptotic pathway. The apoptosome, which is triggered by the ‘intrinsic’ apoptotic pathway, is formed upon mitochondrial release of cytochrome c, leading to the Apaf-1–mediated activation of procaspase 9.

Caspases can autoactivate (e.g., caspase 3 can cleave and activate procaspase 3) or activate other caspases (e.g., caspases 8 and 9 catalyze the proteolytic activation of caspase 3)12. So, in the absence of efficient caspase inhibitors and in the absence of an efficient subcellular com-partmentalization, caspase activation tends to occur in a rapid, chain reaction–like fashion in which all of the apoptosis-associated caspases eventually experience proteolytic maturation, regardless of the initial trigger12.

1Centre National de la Recherche Scientifique, UMR8125, Institut Gustave Roussy, 39 rue Camille-Desmoulins, F-94805 Villejuif, France. 2Molecular Cell Biology Laboratory, Department of Genetics, The Smurfit Institute, Trinity College, Dublin 2, Ireland. Correspondence should be addressed to G.K. ([email protected]).

Published online 6 July 2005; doi:10.1038/nm1263

R E V I E W©

2005

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e

726 VOLUME 11 | NUMBER 7 | JULY 2005 NATURE MEDICINE

Although massive caspase activation is frequently found in apoptosis, the circuitry of the system is often complicated (Fig. 1). For example, activation of caspases 2 or 8 is often insufficient to stimulate the direct activation of apoptotic effector caspases (such as caspases 3, 6 and 7), and this reaction then requires a mitochondrial amplification step with cytochrome c–dependent apoptosome formation. Thus, caspase 8 cleaves and activates proteins from the Bcl-2 family, such as Bid, which catalyzes mitochondrial outer membrane permeabilization (MOMP) and cytochrome c release13. It has also been suggested that caspase 2 can directly interact with mitochondrial membranes, inducing MOMP through a mechanism that does not rely on its catalytic activity14.

When MOMP has occurred and cytochrome c has been released, activation of caspases 3 and 9 can ensue. But it should be noted that a physiological or pathological context in which endogenous levels of caspase 2 are sufficient to provoke direct effects on mitochondria remains to be determined.

Apoptosis is morphologically defined by a few hallmarks, namely cellular and nuclear shrinkage (pyknosis), chromatin condensation and nuclear fragmentation (karyorhexis) with formation of apoptotic bodies15. Moreover, for historical reasons, nuclear DNA degradation has been viewed as a central and near-defining feature of apoptosis. So, methods that assess DNA fragmentation, such as horizontal aga-rose gel electrophoresis (which visualizes the ‘ladder’ of oligonucleosomal DNA fragmen-tation) and the TUNEL technique (which stains nuclei containing fragmented DNA), have been widely used to measure apoptosis. One of the prominent enzymes required for this kind of apoptotic DNA fragmentation is caspase-activated DNAse, and pharmacologi-cal inhibition of caspases (and hence caspase-activated DNAse activation) largely suppresses DNA ladders and TUNEL positivity. Moreover, caspase inhibitors prevent or reduce pyknosis and the formation of apoptotic bodies. This is possible because prominent caspase substrates include proteins that participate in cellular blebbing (gelsolin, ROCK-1, PAK2), determine cellular shape (cytokeratin-18, vimentin, Gas2 and plectin, fodrin, focal adhesion kinase, Cas or paxillin) or influence nuclear architecture16, providing a molecular link between caspase activation and apoptotic morphology that is interrupted when caspases are inhibited.

On the basis of the fact that caspase inhibi-tors reduce the frequency of cells with mor-phological and biochemical signs of apoptosis, it has been widely inferred that they exert true cytoprotective effects. So, when human cells exposed to an apoptogenic stimulus (such as chemotherapy) are treated with polycaspase inhibitors such as N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethylketone (Z-VAD-fmk), suppression of caspase activity results in what seems like inhibition of apoptosis if chroma-tin condensation and DNA fragmentation

are used as identifying criteria. When other signs of cell death (such as mitochondrial dysfunction, phosphatidylserine exposure, plasma membrane permeabilization or loss of clonogenic survival) are consid-ered, caspase inhibition frequently does not confer cytoprotection17–19. Similarly, caspase inhibition is often insufficient to maintain cells alive in vivo.

Advocates of the obligatory involvement of caspases in cell death might advance two reasons to explain the relatively poor cytoprotec-tive potential of caspase inhibitors. First, the ideal inhibitor with opti-mal pharmacokinetic and pharmacodynamic properties might still be lacking. So, although polycaspase inhibitors such as Z-VAD-fmk are

Figure 1 Major pathways to caspase-dependent and caspase-independent cell death. Two major pathways to caspase-dependent apoptosis have been identified: the extrinsic pathway, which involves stimulation of members of the TNF receptor–nerve growth factor receptor superfamily such as TNF receptor (TNFR), CD95 (Fas)–APO-1 receptor or TRAIL receptor (‘death receptors’), and the intrinsic pathway, in which MOMP occurs and results in assembly of a caspase-activating complex between caspase 9 and Apaf-1 (‘the apoptosome’). Death receptor stimulation typically results in the recruitment and activation of caspase 8 by the adaptor proteins FADD and TRADD to form a DISC, which can propagate death signals in two ways: by proteolysis of the BH3-only protein Bid, which provokes translocation of the latter to mitochondria and MOMP, and by direct proteolysis of downstream effector caspases, which results in their activation. In the intrinsic pathway, a range of BH3-only proteins act as sentinels for cell stress, damage or infection and can be mobilized to initiate MOMP through post-translational modification. The BH3-only proteins probably initiate MOMP by triggering oligomerization of Bax and/or Bak in the mitochondrial outer membrane, forming channels that permit the escape of multiple proteins from the mitochondrial intermembrane space. In the context of DNA damage, stabilization of the p53 tumor suppressor protein can result in transcriptional activation of the BH3-only proteins Puma and Noxa, which can promote MOMP through the Bax-Bak channel. An alternative pathway to p53-dependent apoptosis has been proposed to occur through transcriptional upregulation of PIDD protein. PIDD can promote assembly of a complex between itself, RAIDD and caspase 2 (‘the piddosome’). It remains unclear how assembly of the piddosome can promote cell death, but this may involve caspase 2–dependent MOMP. Some of the mitochondrial proteins released as a result of MOMP (AIF, HtrA2/Omi, endonuclease G) can promote caspase-independent death through mechanisms that are relatively poorly defined. Caspase-independent death can also result from stimuli that cause lysosomal membrane permeabilization (LMP) with the consequent release of cathepsin proteases.

R E V I E W©

2005

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e

NATURE MEDICINE VOLUME 11 | NUMBER 7 | JULY 2005 727

effective against most caspases, whether such inhibitors can effectively neutralize all caspases that are activated during apoptosis has been the subject of considerable debate20,21. Second, an intrinsic difficulty of the approach might be the need to obtain very high fractional inhibition to observe cytoprotection21. If a small percentage of uninhibited caspases were sufficient to trigger lethal effector mechanisms (such as caspase-activated DNAse), then it would be necessary to completely inhibit caspase activity to protect cells, and this may be an unrealistic goal. But, as a third possibility, caspases may simply not be a useful therapeutic target, for reasons that we describe in the following sections.

Caspase activation in nonlethal signal transductionA subclass of ‘inflammatory’ caspases (in particular, caspases 1, 4 and 5 in humans, and caspases 1 and 11 in mice) are thought to partici-pate in the maturation of cytokines (in particular, IL-1β and IL-18), without any relationship to apoptosis. Notably, in a subset of individu-als, caspase 12 may serve to antagonize or limit inflammatory caspase activation or effector function, thereby predisposing these individuals to an increased risk of bacterial sepsis or chronic infection22. This may explain why most humans do not express a catalytically active form of caspase 12. In addition, a series of caspases are required for activation and differentiation processes that are not directly linked to cell death. Although it can be argued that caspase-dependent processes such as lens cell enucleation, proplatelet maturation (which implies budding off of parts of the cytoplasm) or erythroblast differentiation (with loss of cytoplasmic organelles and the nucleus) may constitute a sort of par-tial cell death, there are a number of other caspase-dependent processes that are clearly not related to cellular demise: T-cell activation (which depends on caspase 8 activation mediated by the adaptor protein FADD and death receptors, and is deficient in individuals with inactivating caspase 8 mutations)23, maintenance of embryonic endothelial cells (which depends on caspase 8, at least in mice)24, monocyte differentia-tion into macrophages (which depends on caspase 8)24,25, osteogenic differentiation of bone marrow stromal stem cells (which depends on caspase 3 in mice)26, myoblast differentiation and fusion into myotubes (which depends on caspase 3)27, and syncytial fusion of trophoblast cells in the placenta (which depends on caspase 8)28. In neurons, cas-pases may contribute to axon guidance and synaptic plasticity29. So, caspases can act like ordinary signal-transducing molecules. An inter-esting dilemma is how cells avoid massive caspase activation (which would lead to cell death) in these settings. At present, little information is available regarding whether caspase activation is limited spatially (by confining caspase activation to restricted subcellular compartments) or chemically (by caspase inhibitors), or both. It is also possible that crucial caspase substrates are protected from caspase cleavage by post-translational modifications, such as dephosphorylation of lamin30 or phosphorylation of Bid31.

In view of the multiple physiological roles of caspases, it is possible that their inhibitors may exert unexpected side effects, especially when they block caspase-dependent processes that are unrelated to cell death. Thus, Z-DEVD-fmk causes accelerated bone loss in ovariectomized mice, presumably by preventing the osteogenic differentiation of bone marrow stromal stem cells26. Side effects of this sort need to be taken into account if caspase inhibitors are to be used for the treatment of prolonged disease states such as neurodegenerative diseases.

In some cases, caspase inhibitors could precipitate cell death. Sublethal, partial caspase 3 activation has been observed in a pre-conditioning model of transient middle cerebral artery occlusion, a manipulation in which a short-lasting arterial occlusion renders mice resistant to a permanent occlusion performed 24 h later. This caspase 3 activation is essential for the upregulation of heat-shock protein 70

and neuroprotection in preconditioning32, indicating that caspases can also be cytoprotective in some instances. Caspase-mediated cleavage of poly(ADP-ribose) polymerase has also been suggested to participate in ischemic preconditioning33. Moreover, partial cleavage of RasGAP by caspase 3 may be required for activation of an Akt-dependent cell survival pathway under stress conditions34. Similarly, the inactivation of phospholipase A2 by caspase 3 might have cytoprotective effects35. Caspase 2 activation, if confined to the nucleus, may be instrumental for the stimulation of DNA repair9.

In summary, caspase inhibition could also aggravate cell loss, at least on theoretical grounds. Accordingly, caspase inhibitors have been shown to sensitize cells to apoptosis inducers in vitro36,37, and sys-temic injections of Z-VAD-fmk can sensitize mice to the lethal effect of recombinant human tumor necrosis factor (TNF)-α35. Whether this paradoxical effect results from an intrinsic toxic effect of Z-VAD-fmk or truly involves caspase inhibition has not been elucidated.

Dual involvement of caspases in cell deathDepending on the initial cell death–inducing stimulus, there are two different ways in which caspases can contribute to cell death. First, cas-pases can be activated at a late stage of the process, after MOMP. In this case, caspase activation constitutes a sign rather than a mechanism of cell death, and the decisive event has occurred upstream or at the level of MOMP, which frequently marks the ‘point of no return’ of the lethal process38,39. Although there is no doubt that massive caspase activation is sufficient for the induction of cell death, inhibitor experiments indi-cate that, in mammalian cells, caspase inhibition is often not sufficient to avoid cell death. When inhibition of caspases does not inhibit cell death, this often leads to a shift in the morphology of cell death40, from the aspect of classical apoptosis to the occurrence of ‘apoptosis-like’ cell death41,42 (a cell death routine that resembles apoptosis, yet lacks full-blown chromatin condensation, karyorhexis, oligonucleosomal DNA fragmentation and/or formation of apoptotic bodies), autophagic cell death43,44 (cell death with accumulation of autophagic vacuoles) or even necrosis36,45 (death with pronounced swelling of cytoplasmic organelles; Fig. 2). Indeed, caspases can act on specific substrates to modify their function in ways that ensure that the cell does not die from necrosis46.

Second, caspases may link signal transduction (through the DISC) or a DNA damage sensor system (through the piddosome) to the cell death cascade. In such systems, inhibition of specific initiator caspases (e.g., caspase-8 in the DISC or caspase-2 in the piddosome) may inter-rupt proapoptotic signaling at an upstream level and hence inhibit cell death10,47. As a caveat, however, it should be mentioned that CD95-mediated or TNF-α receptor–mediated signals may also be transmitted, at least in some cell types, by caspase-independent signals36,37,48. So,

Figure 2 Relationship between caspase dependency and cell death morphotype.

R E V I E W©

2005

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e

728 VOLUME 11 | NUMBER 7 | JULY 2005 NATURE MEDICINE

caspase inhibition may be expected to confer protection against cell death under only a limited set of circumstances. Where a death-initiat-ing stimulus has engaged the mitochondrial pathway and has provoked MOMP (with the consequent release of several mitochondrial pro-teins), to attempt to rescue the cell by blocking caspase activity is akin to shutting the stable door after the proverbial horse has bolted.

On the basis of this argument, the most probable reason for the fail-ure of caspase inhibitors to provide clonogenic rescue is their incapacity to prevent MOMP17–19,39,45, which occurs upstream of the point of caspase activation in many pathways that culminate in cell death (Fig. 1). Whereas there is no doubt that caspase inhibitors can prevent the death of animals injected with CD95-specific antibodies, such inhibi-tors are relatively inefficient against cell death induced by signals that activate the intrinsic pathway. As an example, neonatal brain ischemia and subsequent reperfusion causes the death of cortical neurons that cannot be inhibited by intrathecal injection of Z-VAD-fmk, although Z-VAD-fmk does inhibit the degradation of caspase substrates in vivo49. Similarly, the intraocular injection of Z-VAD-fmk does not prevent the apoptotic death of photoreceptor cells induced by retinal detachment50. Systemic injections of Z-VAD-fmk also do not affect the contraction of the number of virus-specific CD8+ T cells, although this schedule of Z-VAD-fmk does prevent the fulminant hepatic necrosis induced by Fas (CD95)-specific antibodies51. In neurons, transgenic expression of the protein XIAP under the control of the Thy1 promoter also does not antagonize neuronal loss induced by irradiation, although XIAP does prevent activation of caspases 3 and 9 in these circumstances52. Similar results have been obtained when kainate-induced cell death was assessed in mice expressing the baculovirus p35 caspase inhibitor in neurons53.

Alternative targets for cell-death inhibitionMOMP is a key event during the cell death process, and often defines the point of no return38,39. In some modalities of pathogenic cell death, cyclophilin D–dependent mitochondrial permeability transition deter-

mines lethal MOMP54,55. In addition, a subset of Bcl-2 family proteins, the BH3-only pro-teins, seem to be particularly important trig-gers of MOMP in a variety of pathways13,56. These proteins seem to act as cellular sensors for stress, infection, cellular transformation, cytokine deprivation, attack by cytotoxic T cells or natural killer cells and cellular damage, and can be mobilized, through post-translational modification, to translocate to mitochondria and trigger MOMP13. BH3-only proteins most probably initiate MOMP by inducing oligomerization of Bax and Bak in the mito-chondrial outer membrane13,57. Bax and Bak oligomers form large channels through which many proteins escape from the mitochondrial intermembrane space57. Recent studies indi-cate that BH3-only proteins fall into two sub-families: those that can directly trigger opening of the Bax-Bak channel to precipitate MOMP, and those that sensitize for opening of the latter channel by neutralizing members of the Bcl-2 family that can inhibit channel opening58,59. The direct agonists of Bax-Bak channel open-ing, Bid and Bim, have been suggested to be important participants in several cell death pathways. Indeed, genetic inactivation of Bim

leads to clonogenic survival of cells in several experimental models in which caspase-inhibitors have little effect56. In addition, gene targeting of other members of this family, such as Puma and Noxa, also show cru-cial roles for these proteins in p53-dependent cell death pathways56.

Isolated mitochondria can be used to screen for drugs that prevent MOMP, and some inhibitors of the permeability transition and/or mitochondrial ion channels inhibit cell death in models of stroke or myocardial infarction39,60,61 (Table 1). Because BH3-only members of the Bcl-2 family seem to be important instigators of MOMP in many pathways that culminate in cell death, small-molecule antagonists of these proteins should be effective inhibitors of MOMP. Alternatively, stimulation of a BH3-only protein-targeting ubiquitin ligase might be attempted62. Furthermore, it is possible to target noncaspase pro-teases, such as cathepsins and calpains, that act as upstream MOMP triggers. As an example, cathepsin B can act as a MOMP inducer link-ing lysosomal damage (which causes cathepsin B release) to MOMP63. Cathepsin B knockout cells are particularly resistant to induction of apoptosis by TNF-α, and cathepsin B knockout mice show reduced liver damage in response to TNF-α64 or cholestasis65. The transgenic expression of two cathepsin B inhibitors, namely the serine protease inhibitors 2A and 6, can boost the immune response by preventing death of CD8+ memory T cells66. Moreover, the cathepsin B inhibitor CA-074 can protect neurons from focal cerebral ischemia67. Calpains, a family of 14 Ca2+-activated neutral cysteine proteases, have also been suggested to be involved in the regulation of cell death upstream of MOMP, for instance in endoplasmic reticulum damage, which causes activation of µ-calpain. Indeed, recent experiments on mice in which the endogenous calpain inhibitor calpastatin has been overexpressed or knocked out underscore the importance of calpains as an activator of lethal MOMP in neuronal cell death68.

With regard to stimuli that engage the intrinsic apoptotic pathway, successful inhibition of MOMP would be expected to preclude the release and associated destructive effects of caspase-independent and caspase-dependent cell death effectors. The alternative prospect would

Table 1 Pharmacological targets for cytoprotection

Target Observation Reference

MOMP Inhibition of MOMP by expression of antiapoptotic 38, 39, 61members of the Bcl-2 family has a broad spectrum ofcytoprotective action, including in the heart and in the brain.

Pharmacological inhibition of the permeability transition 39, 54, 55, 60pore or of mitochondrion-specific ion channels has alsoneuroprotective and cardioprotective effects.

Cathepsin B Cathepsin B knockout mice show reduced liver 64, 65damage in response to TNF-α or cholestasis, and thiseffect can be mimicked by injecting a cathepsininhibitor into wild-type mice.

Calpains A variety of calpain inhibitors protect against ischemic 81and preservation-reperfusion liver injury, focal cerebralischemic injury and traumatic spinal cord injury in rodents.

Mice expressing calpastatin in neurons are resistant 53, 68against kainate-evoked excitotoxicity.

HtrA2/Omi The HtrA2/Omi-specific inhibitor ucf-101 inhibits the 78nephrotoxic, proapoptotic effect of cisplatin in vivo.

Ucf-101 reduces infarct size after myocardial ischemia. 79

AIF Microinjection of AIF-specific antibodies can prevent 82, 83caspase-independent neuronal death induced by DNAdamage or excitotoxins.

A hypomorphic mutation of AIF confers protection 74against kainate-mediated excitotoxicity in vivo.

Expression of heat-shock protein 70, which sequesters 84AIF, protects against neonatal brain damage.

R E V I E W©

2005

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e

NATURE MEDICINE VOLUME 11 | NUMBER 7 | JULY 2005 729

be the development of antagonists that block the activities of multiple caspase-independent effectors that are released upon the induction of MOMP, since MOMP results in the unleashing of a battery of destruc-tive molecules within the cell. The proteins that are released from the mitochondrial intermembrane space exert pleiotropic effects, ranging from caspase activation to chromatin condensation, DNA strand break-age and generation of reactive oxygen species (ROS)39,69. Under nor-mal circumstances, caspase activation through release of cytochrome c and assembly of the apoptosome is sufficient to ensure rapid and efficient disposal of the cell. But in contexts where caspase activation is prevented—either through inhibition of caspase activity, genetic inac-tivation of a key caspase (such as caspase 9), loss of Apaf-1 through a similar mechanism, or some other reason—the other mitochondrial constituents that are released through MOMP come to the fore.

This applies to apoptosis-inducing factor (AIF), a protein that translocates from mitochondria to the nucleus, where it interacts with DNA, stimulating the formation of a ‘degradosome’ that in mammals and yeast contains cyclophilin A as an obligate cofactor for AIF-medi-ated chromatin condensation70,71. Another mitochondrial factor that is released during many forms of cell death and translocates to the nuclear compartment is endonuclease G, an enzyme that participates in the degradosome of C. elegans but not in humans72. Pharmacological inhibitors of AIF and endonuclease G are currently not available. But genetic experiments suggest that AIF may have an important role in caspase-independent cell death in mice73, particularly in neurons74. The protein HtrA2/Omi is a mitochondrial serine protease that can mediate caspase-independent apoptosis. One of the crucial substrates of HtrA2/Omi is the antiapoptotic protein HAX-1 (bearing Bcl-2–homology BH1 and BH2 domains), which resides in mitochondria, presumably in the outer membrane, and may be cleaved by HtrA2/Omi when it is still confined in mitochondria75. Other potential substrates of HtrA2/Omi are Ped/Pea-15 (an inhibitor of the DISC and of stress kinase)76 and the inhibitor of apoptosis proteins (IAPs), which function as endog-enous caspase inhibitors. Of note, presenilin-1 functions as an endog-enous HtrA2/Omi activator, and HtrA2/Omi can mediate the death of cells expressing a presenilin-1 mutation similar to that occurring in ∼60% of cases of familial early-onset alzheimer disease77. Inhibitors of HtrA2/Omi are being generated, and one such inhibitor can exert some cytoprotective effects in vivo78,79 (Table 1). In addition to lethal pro-teins, mitochondria can generate and release highly toxic ROS, which contribute to cell death. So, ROS scavengers can have cytoprotective effects in vitro or in vivo80.

ConclusionsThe studies of the past decade have highlighted the importance of caspases for diverse cellular processes, but have also led to the conclu-sion that targeting of caspases for therapeutic purposes will probably be restricted to situations in which apoptotic caspase activation has been achieved through death receptor signaling or when inhibition of inflammatory caspases is the therapeutic goal. For the purpose of this discussion, we can envisage three relationships between death and cas-pase activation: cell death through caspase activation, cell death with caspase activation, and cell death without caspase activation (the latter two cases being independent of caspases). It seems that, in contrast to earlier expectations, cell death with caspase activation is frequently not prevented by caspase inhibitors, leading to a phenotypic shift in the morphological manifestation of cell death (Fig. 2).

One of the major problems associated with attempts to therapeuti-cally manipulate caspase-independent cell death mechanisms is that many details concerning how cells die in the absence of caspase par-ticipation are still lacking. From the preceding discussion it is evident

that caspase-independent death is still largely defined in indirect terms. That is to say, essentially any cell death that takes place during which caspase activity has been genetically or pharmacologically neutralized can be attributed to a caspase-independent mechanism. But this does not prove that these cell deaths share an underlying cause. Indeed, it seems more probable that caspase-independent cell death represents a panoply of cell deaths that share little in common, except their inability to be rescued by pharmacological or genetic inactivation of caspases. But there is little doubt that cells frequently die in a caspase-indepen-dent manner when all routes to caspase activation have been closed off. So, where does this leave us in terms of therapeutic intervention in these situations?

As discussed earlier, a common event in numerous forms of cell death is MOMP. The resulting escape of mitochondrial proteins that are either toxic within the cytosol, that result in a rapid decline in ATP produc-tion, provoke generation of ROS, or all of the above is clearly a major impediment to cell survival. For all the reasons already outlined, it is not surprising that caspase inhibitors are unable to provide clonogenic rescue from stresses that result in MOMP. In situations where MOMP contributes to caspase activation, approaches aimed at blocking release of both caspase-independent and caspase-dependent effectors of cell death will be required. Keeping the genie in the bottle, rather than trying to put him back in, is likely to be a challenging but worthwhile endeavor.

ACKNOWLEDGMENTSG.K. is supported by Ligue Nationale contre le cancer, European Union (Active p53, Impaled, RIGHT, Trans-Death), Canceropole Ile-de-France, and French Ministry of Science. S.J.M. is supported by a Principal Investigator Award from Science Foundation Ireland.

COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.

Published online at http://www.nature.com/naturemedicine/

1. Horvitz, H.R. Worms, life, and death. ChemBioChem 4, 697–711 (2003).2. Hay, B.A., Huh, J.R. & Guo, M. The genetics of cell death: approaches, insights and

opportunities in Drosophila. Nat. Rev. Genet. 5, 911–922 (2004).3. Abraham, M.C. & Shaham, S. Death without caspases, caspases without death. Trends

Cell Biol. 14, 184–193 (2004).4. Boyce, M., Degterev, A. & Yuan, J. Caspases: an ancient cellular sword of Damocles.

Cell Death Differ. 11, 29–37 (2004).5. Golstein, P., Aubry, L. & Levraud, J.P. Cell-death alternative model organisms: why and

which? Nat. Rev. Mol. Cell Biol. 4, 798–807 (2003).6. Martin, S.J. & Green, D.R. Protease activation during apoptosis: death by a thousand

cuts? Cell 82, 349–352 (1995).7. Garrido, C. & Kroemer, G. Life’s smile, death’s grin: vital functions of apoptosis-execut-

ing proteins. Curr. Opin. Cell Biol. 16, 639–646 (2004).8. Martinon, F. & Tschopp, J. Inflammatory caspases: linking an intracellular innate

immune system to autoinflammatory diseases. Cell 117, 561–574 (2004).9. Tinel, A. & Tschopp, J. The PIDDosome, a protein complex implicated in activation of

caspase-2 in response to genotoxic stress. Science 304, 843–846 (2004).10. Krammer, P.H. CD95’s deadly mission in the immune system. Nature 407, 789–795

(2000).11. Jiang, X. & Wang, X. Cytochrome C-mediated apoptosis. Annu. Rev. Biochem. 73,

87–106 (2004).12. Zhivotovsky, B. Caspases: the enzymes of death. Essays Biochem. 39, 25–40 (2003).13. Danial, N.N. & Korsmeyer, S. Cell death: critical control points. Cell 116, 205–219

(2004).14. Enoksson, M. et al. Caspase-2 permeabilizes the outer mitochondrial membrane and

disrupts the binding of cytochrome c to anionic phospholipids. J. Biol. Chem. 279, 49575–49578 (2004).

15. Kroemer, G. et al. Classification of cell death: Recommendations of the Nomenclature Committee on Cell death. Cell Death Differ. (in the press) (2005).

16. Fischer, U., Janicke, R.U. & Schulze-Osthoff, K. Many cuts to ruin: a comprehensive update of caspase substrates. Cell Death Differ. 10, 76–100 (2003).

17. Xiang, J., Chao, D.T. & Korsmeyer, S.J. Bax-induced cell death may not require interleu-kin 1β-converting enzyme-like proteases. Proc. Natl Acad. Sci. USA 93, 14559–14563 (1996).

18. Carter, B.Z. et al. Caspase-independent cell death in AML: caspase-inhibition in vitro with pan-caspase inhibitors or in vivo by XIAP or Survivin does not affect cell survival or prognosis. Blood 102, 4179–4186 (2003).

R E V I E W©

2005

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e

730 VOLUME 11 | NUMBER 7 | JULY 2005 NATURE MEDICINE

19. Kanzawa, T. et al. Arsenic trioxide induces autophagic cell death in malignant glioma cells by upregulation of mitochondrial cell death protein BNIP3. Oncogene 24, 980–991 (2005).

20. Nicholson, D.W. From bench to clinic with apoptosis-based therapeutic agents. Nature 407, 810–816 (2000).

21. Methot, N. et al. Differential efficacy of caspase inhibitors on apoptosis markers during sepsis in rats and implication for fractional inhibition requirements for therapeutics. J. Exp. Med. 199, 199–207 (2004).

22. Saleh, M. et al. Differential modulation of endotoxin responsiveness by human caspase-12 polymorphisms. Nature 429, 75–79 (2004).

23. Chun, H.J. et al. Pleiotropic defects in lymphocyte activation caused by caspase-8 mutations lead to human immunodeficiency. Nature 419, 395–399 (2002).

24. Kang, T.B. et al. Caspase-8 serves both apoptotic and nonapoptotic roles. J. Immunol. 173, 2976–2984 (2004).

25. Sordet, O. et al. Specific involvement of caspases in the differentiation of monocytes into macrophages. Blood 100, 4446–4453 (2002).

26. Miura, M. et al. A crucial role of caspase-3 in osteogenic differentiation of bone marrow stromal stem cells. J. Clin. Invest. 114, 1704–1713 (2004).

27. Fernando, P., Kelly, J.F., Balazsi, K., Slack, R.S. & Megeney, L.A. Caspase 3 activity is required for skeletal muscle differentiation. Proc. Natl Acad. Sci. USA 99, 11025–11030 (2002).

28. Black, S. et al. Syncytial fusion of human trophoblast depends on caspase 8. Cell Death Differ. 11, 90–98 (2004).

29. McLaughlin, B. The kinder side of killer proteases: caspase activation contributes to neuroprotection and CNS remodeling. Apoptosis 9, 111–121 (2004).

30. Cross, T. et al. PKC-delta is an apoptotic lamin kinase. Oncogene 19, 2331–2337 (2000).

31. Desagher, S. et al. Phosphorylation of bid by casein kinases I and II regulates its cleav-age by caspase 8. Mol. Cell 8, 601–611 (2001).

32. McLaughlin, B. et al. Caspase 3 activation is essential for neuroprotection in precon-ditioning. Proc. Natl Acad. Sci. USA 100, 715–720 (2003).

33. Garnier, P., Ying, W. & Swanson, R.A. Ischemic preconditioning by caspase cleavage of poly(ADP-ribose) polymerase-1. J. Neurosci. 23, 7967–7973 (2003).

34. Yang, J.Y. et al. Partial cleavage of RasGAP by caspases is required for cell survival in mild stress conditions. Mol. Cell. Biol. 24, 10425–10436 (2004).

35. Cauwels, A., Janssen, B., Waeytens, A., Cuvelier, C. & Brouckaert, P. Caspase inhibi-tion causes hyperacute TNF shock via oxidative stress and PLA2. Nat. Immunol. 4, 387–393 (2003).

36. Vercammen, D. et al. Inhibition of caspases increases the sensitivity of L929 cells to necrosis mediated by tumor necrosis factor. J. Exp. Med. 187, 1477–1485 (1998).

37. Foghsgaard, L. et al. Cathepsin B acts as a dominant execution protease in tumor cell apoptosis induced by tumor necrosis factor. J. Cell Biol. 153, 999–1010 (2001).

38. Kroemer, G. & Reed, J.C. Mitochondrial control of cell death. Nat. Med. 6, 513–519 (2000).

39. Green, D.R. & Kroemer, G. The pathophysiology of mitochondrial cell death. Science 305, 626–629 (2004).

40. Chipuk, J.E. & Green, D.R. Do inducers of apoptosis trigger caspase-independent cell death? Nat. Rev. Mol. Cell Biol. 6, 268–275 (2005).

41. Susin, S.A. et al. Two distinct pathways leading to nuclear apoptosis. J. Exp. Med. 192, 571–579 (2000).

42. Jaattela, M. & Tschopp, J. Caspase-independent cell death in T lymphocytes. Nat. Immunol. 4, 416–423 (2003).

43. Yu, L. et al. Regulation of an ATG7-beclin 1 program of autophagic cell death by caspase-8. Science 304, 1500–1502 (2004).

44. Shimizu, S. et al. A role of Bcl-2 family of proteins in non-apoptotic programmed cell death dependent on autophagy genes. Nat. Cell Biol. 6, 1221–1228 (2004).

45. Hirsch, T. et al. The apoptosis-necrosis paradox. Apoptogenic proteases activated after mitochondrial permeability transition determine the mode of cell death. Oncogene 15, 1573–1582 (1997).

46. Glazner, G.W., Chan, S.L., Lu, C. & Mattson, M.P. Caspase-mediated degradation of AMPA receptor subunits: a mechanism for preventing excitotoxic necrosis and ensuring apoptosis. J. Neurosci. 20, 3641–3649 (2000).

47. Lassus, P., Opitz-Araya, X. & Lazebnik, Y. Requirement for caspase-2 in stress-induced apoptosis before mitochondrial permeabilization. Science 297, 1352–1354 (2002).

48. Vanden Berghe, T. et al. Differential signaling to apoptotic and necrotic cell death by Fas-associated death domain protein FADD. J. Biol. Chem. 279, 7925–7933 (2004).

49. Zhu, C. et al. Involvement of apoptosis-inducing factor in neuronal death after hypoxia-ischemia in the neonatal rat brain. J. Neurochem. 86, 306–317 (2003).

50. Hisatomi, T. et al. Relocalization of apoptosis-inducing factor in photoreceptor apopto-sis induced by retinal detachment in vivo. Am. J. Pathol. 158, 1271–1278 (2001).

51. Nussbaum, A.K. & Whitton, J.L. The contraction phase of virus-specific CD8+ T cells is unaffected by a pan-caspase inhibitor. J. Immunol. 173, 6611–6618 (2004).

52. Fukuda, H. et al. Irradiation-induced progenitor cell death in the developing brain is

resistant to erythropoietin treatment and caspase inhibition. Cell Death Differ. 11, 1166–1178 (2004).

53. Higuchi, M. et al. Distinct mechanistic roles of calpain and caspase activation in neurodegeneration as revealed in mice overexpressing their specific inhibitors. J. Biol. Chem. 280, 15229–15237 (2005).

54. Baines, C.P. et al. Loss of cyclophilin D reveals a critical role for mitochondrial perme-ability transition in cell death. Nature 434, 658–662 (2005).

55. Nakagawa, T. et al. Cyclophilin D-dependent mitochondrial permeability transition regulates some necrotic but not apoptotic cell death. Nature 434, 652–658 (2005).

56. Strasser, A. The role of BH3-only proteins in the immune system. Nat. Rev. Immunol. 5, 189–200 (2005).

57. Kuwana, T. et al. Bid, Bax, and lipids cooperate to form supramolecular openings in the outer mitochondrial membrane. Cell 111, 331–342 (2002).

58. Kuwana, T. et al. BH3 domains of BH3-only proteins differentially regulate Bax-medi-ated mitochondrial membrane permeabilization both directly and indirectly. Mol. Cell 17, 525–535 (2005).

59. Chen, L. et al. Differential targeting of prosurvival Bcl-2 proteins by their BH3-only ligands allows complementary apoptotic function. Mol. Cell 17, 393–403 (2005).

60. Mattson, M.P. & Kroemer, G. Mitochondria in cell death: novel targets for neuroprotec-tion and cardioprotection. Trends Mol. Med. 9, 196–205 (2003).

61. Stavrovskaya, I.G. et al. Clinically approved heterocyclics act on a mitochondrial target and reduce stroke-induced pathology. J. Exp. Med. 200, 211–222 (2004).

62. Fischer, S.F. et al. Chlamydia inhibit host cell apoptosis by degradation of proapoptotic BH3-only proteins. J. Exp. Med. 200, 905–916 (2004).

63. Muntener, K., Zwicky, R., Csucs, G., Rohrer, J. & Baici, A. Exon skipping of cathepsin B: mitochondrial targeting of a lysosomal peptidase provokes cell death. J. Biol. Chem. 279, 41012–41017 (2004).

64. Guicciardi, M.E., Miyoshi, H., Bronk, S.F. & Gores, G.J. Cathepsin B knockout mice are resistant to tumor necrosis factor-alpha-mediated hepatocyte apoptosis and liver injury: implications for therapeutic applications. Am. J. Pathol. 159, 2045–2054 (2001).

65. Canbay, A. et al. Cathepsin B inactivation attenuates hepatic injury and fibrosis during cholestasis. J. Clin. Invest. 112, 152–159 (2003).

66. Liu, N. et al. Serine protease inhibitor 2A is a protective factor for memory T cell development. Nat. Immunol. 5, 919–926 (2004).

67. Benchoua, A., Braudeau, J., Reis, A., Couriaud, C. & Onteniente, B. Activation of proinflammatory caspases by cathepsin B in focal cerebral ischemia. J. Cereb. Blood Flow Metab. 24, 1272–1279 (2004).

68. Takano, J. et al. Calpain mediates excitotoxic DNA fragmentation via mitochondrial pathways in adult brains: Evidence from calpastatin-mutant mice. J. Biol. Chem. 280, 16175–16184 (2005).

69. Saelens, X. et al. Toxic proteins released from mitochondria in cell death. Oncogene 23, 2861–2874 (2004).

70. Cande, C. et al. AIF and cyclophilin A cooperate in apoptosis-associated chromatinoly-sis. Oncogene 23, 1514–1521 (2004).

71. Wissing, S. et al. An AIF orthologue regulates apoptosis in yeast. J. Cell Biol. 166, 969–974 (2004).

72. Parrish, J.Z. & Xue, D. Functional genomic analysis of apoptotic DNA degradation in C. elegans. Mol. Cell 11, 987–996 (2003).

73. Joza, N. et al. Essential role of the mitochondrial apoptosis inducing factor in pro-grammed cell death. Nature 410, 549–554 (2001).

74. Cheung, E.C. et al. Apoptosis-inducing factor is a key factor in neuronal cell death propagated by BAX-dependent and BAX-independent mechanisms. J. Neurosci. 25, 1324–1334 (2005).

75. Cilenti, L. et al. Regulation of HAX-1 anti-apoptotic protein by Omi/HtrA2 protease during cell death. J. Biol. Chem. 279, 50295–50301 (2004).

76. Trencia, A. et al. Omi/HtrA2 promotes cell death by binding and degrading the anti-apoptotic protein ped/pea-15. J. Biol. Chem. 279, 46566–46572 (2004).

77. Gupta, S. et al. The C-terminal tail of presenilin regulates Omi/HtrA2 protease activity. J. Biol. Chem. 279, 45844–45854 (2004).

78. Cilenti, L. et al. Omi/HtrA2 protease mediates cisplatin-induced cell death in renal cells. Am. J. Physiol. Renal Physiol. 288, F371–F379 (2005).

79. Liu, H.R. et al. Role of Omi/HtrA2 in apoptotic cell death after myocardial ischemia and reperfusion. Circulation 111, 90–96 (2005).

80. Rustin, P. The use of antioxidants in Friedreich’s ataxia treatment. Expert Opin. Investig. Drugs 12, 569–575 (2003).

81. Liu, X., Van Vleet, T. & Schnellmann, R.G. The role of calpain in oncotic cell death. Annu. Rev. Pharmacol. Toxicol. 44, 349–370 (2004).

82. Cregan, S.P. et al. Apoptosis-inducing factor is involved in the regulation of caspase-independent neuronal cell death. J. Cell Biol. 158, 507–517 (2002).

83. Wang, H. et al. Apoptosis-inducing factor substitutes for caspase executioners in NMDA-triggered excitotoxic neuronal death. J. Neurosci. 24, 10963–10973 (2004).

84. Matsumori, Y. et al. Hsp70 overexpression sequesters AIF and reduces neona-tal hypoxic/ischemic brain injury. J. Cereb. Blood Flow Metab. 2 March 2005 (doi: 10.1038/sj.jcbfm.9600080).

R E V I E W©

2005

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

emed

icin

e