Cell Death in the Pathogenesis of Heart Disease: Mechanisms and Significance

Abstract

Cell death was once viewed as unregulated. It is now clear that at least a portion of cell death is a regulated cell suicide process. This type of death can exhibit multiple morphologies. One of these, apoptosis, has long been recognized to be actively mediated, and many of its underlying mechanisms have been elucidated. Moreover, necrosis, the traditional example of unregulated cell death, is also regulated in some instances. Autophagy is usually a survival mechanism but can occur in association with cell death. Little is known, however, about how autophagic cells die. Apoptosis, necrosis, and autophagy occur in cardiac myocytes during myocardial infarction, ischemia/reperfusion, and heart failure. Pharmacological and genetic inhibition of apoptosis and necrosis lessens infarct size and improves cardiac function in these disorders. The roles of autophagy in ischemia/reperfusion and heart failure are unresolved. A better understanding of these processes and their interrelationships may allow for the development of novel therapies for the major heart syndromes.

INTRODUCTION

Cell death has traditionally been viewed as an unregulated process. Although cell death is a prominent feature of multiple diseases, this assumption of passivity did not encourage attempts to delineate its pathogenic role. The one type of cell death that had been suspected of being programmed was the deletion of specific cells during development, although the mechanism was obscure. In the mid-1980s, Horvitz and colleagues (1) carried out a series of genetic analyses in the developing nematode Caenorhabditis elegans that triggered a paradigm shift in our view of cell death. These studies identified a small network of genes that regulate programmed cell death during development in this organism (Figure 1a) and, in so doing, provided definitive evidence that some instances of cell death are actively carried out by the cell it-self. During an intense period of investigation from approximately 1985 to 2000, multiple investigators defined biochemical and molecular mechanisms in mammalian cells that mediate a specific type of cell death called apoptosis. The findings showed that the basic schemas for developmental programmed cell death in the worm have been conserved over 600 million years of metazoan evolution to mediate apoptosis in mammalian cells (Figure 1b). In the past decade, attention has also become focused on nonapoptotic forms of cell death, particularly necrosis and autophagic cell death. Although our understanding of these areas is rudimentary, it is clear that at least a subset of nonapoptotic cell death is also actively mediated. This collective work presents a new model for cell death as an active and regulated process in which noxious stimuli originating outside or inside the cell are the triggers to die but the death machinery is part of the cell itself.

Figure 1.

Evolutionary conservation of apoptosis pathways. (a) Programmed cell death during Caenorhabditis elegans development. Precisely 131 cells (out of the 1090 somatic cells in the adult hermaphrodite) die at specific times during nematode development. Mutagenesis studies have revealed genes that regulate these deaths (162). These genes are termed cell death abnormal, or ced, and include ced-3 and ced-4, which promote death, and ced-9, which inhibits death. Loss-of-function mutations of ced-9 result in widespread death, which can be rescued by loss-of-function mutations of either ced-4 or ced-3. Thus, ced-9 is upstream of and inhibits ced-4 and ced-3. The relationship between ced-4 and ced-3 was elucidated in studies in which ced-4 killing was shown to require ced-3 (163). (b) Mammalian apoptosis. Although apoptosis in worms and mammals differs in some respects, the genetic blueprint has been conserved over 600 million years of evolution from worms to mammals. The ortholog of Ced-3 is the caspase family of cysteine proteases. Ced-4 is represented by a single protein, Apaf-1, which functions as an adaptor in the apoptosome. Orthologs of Ced-9 are the antiapoptotic branch of the Bcl-2 family. The mammalian Bcl-2 family also contains two subfamilies of proapoptotic proteins, one of which, the BH3-only branch, is present in C. elegans. The worm BH3-only protein Egl-1 (Egg laying defective-1) inhibits Ced-9 to promote apoptosis. The equivalent step in mammals is the inhibition of Bcl-2 by various BH3-only proteins. Mammalian BH3-only proteins may also induce apoptosis through additional mechanisms (35, 36; see text).

How do these important advances in biology relate to disease? A wiring diagram for various forms of cell death, albeit quite rudimentary in places, has allowed investigators to manipulate these processes genetically and pharmacologically in order to define the significance of cell death in the pathogenesis of disease. Such approaches have revealed that cell death plays important causal roles in multiple diseases including cancer, stroke, myocardial infarction, heart failure, diabetes, sepsis, and some neurodegenerative diseases. Specifically, genetic manipulations in the mouse have shown that regulated forms of cardiac myocyte death are important in the pathogenesis of myocardial infarction and heart failure (2).

This article critically reviews the role of cell death in heart disease. In particular, we discuss (a) the observations that causally link cell death with pathogenesis, (b) current understanding of the contributions of apoptotic and nonapoptotic death, and (c) the relevance of cell death in a larger context and therapeutic implications.

CELL DEATH

What Is Cell Death?

Just as debated for organismal death, there is considerable controversy over what constitutes cell death (3). Parameters that indicate unequivocal cell death include loss of plasma membrane integrity, cellular fragmentation, and phagocytosis by neighboring cells. Parameters thought to indicate serious injury that usually, but not invariably, leads to cell death include (a) outer mitochondrial membrane permeabilization with release of apoptogens and (b) inner mitochondrial membrane permeabilization and loss of electrical potential. As discussed below, not all these parameters are applicable to every form of cell death. The deeper issue, however, is that, despite some understanding of the pathways that mediate cell death, we have little understanding as to what cell death is. The most incontrovertible feature of death is its irreversibility, and the molecular parameters that impart irreversibility remain to be defined.

Types of Cell Death

Cell death can be classified in many ways. Probably the most useful differentiation involves those instances that are actively regulated versus those that are passive or accidental. This review deals with molecular mechanisms that mediate several forms of actively regulated cell death in the heart. In making these distinctions between active and passive forms of cell death, however, one should keep in mind that actively mediated programs may play a role even in situations of cell death traditionally considered passive (e.g., acute massive trauma). Below, we discuss three forms of cell death: apoptosis, necrosis, and autophagic cell death. Although these are currently defined morphologically, recent progress has provided molecular insights.

Apoptosis

Apoptotic cells exhibit cytoplasmic shrinkage [“shrinkage necrosis” (4)], plasma membrane blebbing, nuclear condensation, and later fragmentation of both cytoplasm and nucleus into membrane-enclosed apoptotic bodies. These bodies undergo phagocytosis by macrophages or neighboring cells, thereby avoiding an inflammatory response.

Apoptosis is an actively regulated form of cell death that is mediated by two pathways (Figure 2). The extrinsic pathway utilizes cell surface receptors, whereas the intrinsic pathway involves the mitochondria and endoplasmic reticulum (ER), and each of these pathways leads to caspase activation. In addition, connections between the pathways amplify signals, increasing the efficiency of killing. Apoptosis plays important roles in development and in postnatal life, when it is critical for tissue homeostasis and surveillance for damaged or transformed cells. Inadequate, excessive, or inappropriate apoptosis can contribute to the pathogenesis of disease.

Figure 2.

Apoptosis pathways. Apoptosis is mediated by an extrinsic pathway involving cell surface death receptors and by an intrinsic pathway that utilizes the mitochondria and endoplasmic reticulum. The extrinsic pathway is activated by binding of death ligand to its receptor, which triggers formation of the DISC. Caspase-8 is activated by forced proximity within the DISC and then cleaves and activates downstream procaspases. Caspase-8 can also cleave the BH3-only protein Bid, the carboxyl portion of which translocates to the mitochondria to trigger apoptotic mitochondrial events. The intrinsic pathway is activated by diverse biological, chemical, and physical stimuli. These signals are transduced to the mitochondria and ER (not shown) by proapoptotic Bcl-2 proteins: Bax (a multidomain protein) and BH3-only proteins. These death signals trigger the release of apoptogens from the mitochondria into the cytosol, one of which, cytochrome c, is depicted here. Cytosolic cytochrome c triggers the formation of a second multiprotein complex, the apoptosome, in which procaspase-9 undergoes activation. Caspase-9 then cleaves and activates downstream procaspases. Downstream caspases cleave several hundred cellular proteins to bring about the apoptotic death of the cell. See text for details.

Caspases.

Caspases are a class of cysteine proteases that hydrolyze peptide bonds following aspartic acid residues (5). They are synthesized as largely inactive procaspases, which are arranged in a hierarchy consisting of upstream procaspases-2, -8, -9, and -10 and downstream procaspases-3, -6, and -7. Upstream procaspases, which normally exist as monomers, undergo activation when forced to dimerize (6). Dimerization takes place in the death-inducing signaling complex (DISC) and apoptosome, which are multiprotein complexes (discussed below). Once activated by dimerization, upstream procaspases undergo autocleavage and then proceed to cleave downstream procaspases-3 and -7. These downstream procaspases are already dimerized, and cleavage is the activating event (7). The primary function of downstream caspases is to bring about cellular demise by cleaving hundreds of structural and regulatory proteins. In addition, caspases amplify apoptotic signaling by cutting and activating a variety of proapoptotic mediators (8, 9).

Extrinsic pathway.

The extrinsic pathway is specialized to transduce signals from soluble and cell-bound death ligands [e.g., Fas ligand and tumor necrosis factor (TNF)-α], which bind and activate their cognate cell surface death receptors (10). The binding of Fas ligand triggers a conformational change in Fas that allows its intracellular death domain (DD) to bind to a DD in the adaptor protein FADD (Fas-associated via death domain). This, in turn, allows FADD to recruit procaspase-8 or -10, an interaction mediated by death effector domains (DED) in each of the proteins. This multiprotein complex is termed the DISC (11). Recruitment to the DISC results in the activation of procaspases-8 and -10 through forced proximity (6, 12), following which these activated caspases can cleave and activate downstream procaspases. Although direct activation of downstream procaspases in the extrinsic pathway can be important in some cells (type I cells), in most cells it is inadequate to bring about significant downstream caspase activation (type II cells) (13). The extrinsic pathway usually requires amplification at the mitochondria. This amplification is mediated by cleavage of the BH3 [B cell leukemia/lymphoma-2 (Bcl-2) homology domain 3]-only protein Bid [BH3-interacting domain death agonist] by caspase-8, following with truncated Bid (tBid) translocating to the mitochondria and contributing to mitochondrial apoptosis events (described below; 9). TNF signaling is more complex, as TNF can promote survival as well as death. This increased complexity has required that the simple Fas DISC model be revised for TNF signaling (14), which is considered in the necrosis section below.

Intrinsic pathway.

In contrast to the extrinsic pathway, the more ancient intrinsic pathway is responsible for transducing most apoptotic stimuli, including those due to inadequate nutrients/survival factors, hypoxia, oxidative stress, nutrient stress, proteotoxic stress, DNA damage, and chemical and physical toxins. These stimuli ultimately converge at the mitochondria to trigger the release of apoptogenic proteins (15) and at the ER to stimulate the release of luminal Ca²⁺ (16, 17) and to initiate other events (18, 19).

Death signals are transduced to the mitochondria and ER by two classes of proapoptotic Bcl-2 proteins: Bax (Bcl-2-associated X protein), a multidomain proapoptotic, and BH3-only proteins (21). In healthy cells, a pool of Bax resides in the cytosol in an inactive conformation. In response to apoptotic stimuli, Bax undergoes conformational activation, which includes exposure of its C-terminal transmembrane domain, and translocation to the mitochondria and ER (22, 22a). A variety of Bax-binding proteins regulate the conformational state of Bax, although the precise mechanisms that mediate Bax activation are incompletely understood (23–25). In contrast to the general involvement of Bax, BH3-only proteins transduce death signals in a relatively stimulus-specific manner (21). Moreover, death stimuli regulate the abundance, activity, and localization of these proteins through transcriptional mechanisms, posttranslational modifications including phosphorylation and cleavage, and association with other proteins. For example, deficiency of certain survival factors results in dephosphorylation of Bad, allowing its release from the 14–3-3 protein and translocation to the mitochondria (26).

The translocation of Bax and BH3-only proteins to the mitochondria triggers the release of mitochondrial apoptogens. Although additional aspects of mitochondrial remodeling are involved (15), the key event is the permeabilization of the outer mitochondrial membrane. The primary regulators of outer mitochondrial membrane permeabilization are Bax and a second multidomain proapoptotic Bcl-2 protein called Bak (Bcl-2 homologous antagonist/killer) (27). In contrast to Bax, a pool of Bak is constitutively anchored at the outer mitochondrial membrane, where its activity is regulated by interactions with several proteins (28, 28a). Bax and Bak promote mitochondrial apoptogen release. Although their functions are redundant in some contexts (27), their effects are additive in other systems (29). Apoptogen release is opposed by the antiapoptotic Bcl-2 proteins Bcl-2 and Bcl-x_L (B cell leukemia/lymphoma-x, long isoform) (30, 31). Little is known about the physical or molecular basis for outer mitochondrial membrane permeabilization or how Bax and Bak regulate this event. It is known, however, that Bax and Bak can undergo a complex pattern of homo- and hetero-oligomerization at the mitochondria (32, 33). BH3-only proteins contribute to this oligomerization (34), possibly through direct interactions with Bax and Bak (35), although an alternative model postulates that the BH3-only proteins interact only with Bcl-2 and Bcl-x_L to stimulate the release of Bax and Bak (36). A precise understanding of how these Bax-Bak interactions bring about outer mitochondrial membrane permeabilization remains unknown.

Permeabilization of the outer mitochondrial membrane permits the release of several apoptogens into the cytosol. Most of these proteins are thought to have important physiological functions in healthy cells but exhibit toxic properties when released into the cytosol. For example, cytochrome c is part of the electron transport chain in oxidative phosphorylation. Once cytosolic, however, cytochrome c binds the adaptor protein Apaf-1 (apoptotic protease activating factor-1) along with dATP, which is already present in the cytosol. The binding of cytochrome c triggers a conformational change in Apaf-1, resulting in its oligomerization and recruitment of upstream procaspase-9 into a heptameric structure termed the apoptosome (37, 38). Similar to procaspase-8 in the DISC, procaspase-9 in the apoptosome is activated by forced proximity. It subsequently undergoes autocleavage and then activates downstream procaspases. Other apoptogens are discussed below.

As might be expected, a number of endogenous inhibitors oppose the various activators of apoptosis. To unify the discussion, inhibitors of both extrinsic and intrinsic pathways are considered together. A potential inhibitor of the extrinsic pathway is c-FLIP [FLICE (FADD-Like IL-1β-converting enzyme)-inhibitory protein], which exists as two alternatively spliced isoforms (39). The short isoform is composed of only two DEDs, whereas the long isoform is homologous to procaspase-8 but is enzymatically dead due to mutation of the catalytic cysteine. The effects of FLIP are complex, however. The short isoform inhibits the extrinsic pathway, probably by occupying DEDs on FADD and procaspase-8, thereby precluding DISC assembly. In contrast, low concentrations of the long isoform result in the recruitment and activation of procaspase-8. Only at high concentrations is the long isoform inhibitory.

Key inhibitors of the intrinsic pathway include the antiapoptotic Bcl-2 proteins discussed above, which inhibit mitochondrial apoptogen release. In addition, proteins of the IAP (inhibitor of apoptosis) family [e.g., XIAP (X-linked inhibitor of apoptosis protein)] inhibit already activated downstream caspase-3 and -7 by binding these caspases directly and occluding substrate access (40–42). IAPs also bind procaspase-9 and inhibit its activation in the apoptosome (42a). IAPs also possess E3 ubiquitin ligase activity that can target downstream caspases, as well as themselves, for destruction in the proteasome (43, 44).

The mitochondrial apoptogens Smac/DIABLO (second mitochondria-derived activator of caspase/direct IAP-binding protein with low PI) (45, 45a) and Omi/HtrA2 (Omi/High temperature requirement protein A2) (46, 47) reverse the inhibition of caspases by IAPs. This involves direct binding of Smac/DIABLO and Omi/HtrA2 to the IAPs, thereby releasing the caspases. In addition, Omi/HtrA2 possesses a serine protease activity that cleaves and irreversibly inactivates XIAP (48). The need for Smac/DIABLO and Omi/HtrA2 in the efficient execution of apoptosis underscores that cytochrome c release alone may not be adequate and that it may be necessary to inactivate inhibitory mechanisms as well.

The inhibitors discussed above antagonize either the extrinsic pathway or the intrinsic pathway. In contrast, ARC [Apoptosis repressor with a CARD (Caspase recruitment domain)] is unusual in that it inhibits both the extrinsic and intrinsic pathways through several mechanisms. ARC blocks the extrinsic pathway by interacting directly with Fas and FADD, preventing formation of the DISC (49). ARC inhibits the intrinsic pathway by binding Bax and suppressing its conformational activation and translocation to the mitochondria in response to an apoptotic stimulus (49, 50). ARC also inactivates p53, a transcription factor whose targets include proapoptotic genes. The mechanism involves a direct interaction between ARC and p53 in the nucleus, which inhibits p53 tetramerization (51). Disruption of p53 tetramerization both disables p53 function as a transcription factor and exposes a nuclear export signal in p53 that relocates it to the cytoplasm.

Necrosis

In contrast to apoptosis, which is actively mediated, necrosis has been the traditional example of unregulated cell death. Although a significant proportion of necrotic deaths are passive, evidence has emerged that necrosis can also be regulated. The relative proportion of unregulated versus regulated necrotic death is not currently known, but regulated necrosis has clearly been shown to be an important component of myocardial infarction, heart failure, and stroke. A subset of regulated necrotic cell death, initiated by death receptor activation along with simultaneous caspase inhibition, has been termed necroptosis (52), but we use the term necrosis for all regulated and unregulated forms.

The quintessential features of necrosis are loss of plasma membrane integrity and depletion of cellular ATP, although the causal relationship between these events is not yet clear. As a result of plasma membrane dysfunction, necrotic cells become swollen, in contrast to the shrunken appearance of apoptotic cells. There is also swelling of organelles such as the mitochondria, which remain morphologically intact in apoptosis until late in the process. These events result in a general collapse of intracellular homeostasis. In contrast to the clean-up operation that characterizes apoptosis, the release of cellular contents into the extracellular space engenders an inflammatory response in necrosis. We next consider some of the molecular pathways that mediate regulated necrosis (Figure 3). In contrast to the relatively mature body of knowledge that characterizes apoptosis mechanisms, the investigation of necrosis signaling is in the early stages.

Figure 3.

Necrosis pathways. Information about necrosis signaling is currently limited to two pathways. The first involves death receptors, as exemplified by TNFR1 (tumor necrosis factor-α receptor 1). Depending on context, activation of TNFR1 can promote cell survival or either apoptotic or necrotic cell death. These choices are mediated by multiprotein complexes I and II. The binding of TNF-α to TNFR1 stimulates formation of complex I, which contains TNFR1, TRADD, RIP1, TRAF2, and cIAP1/2. The exact relationships among these proteins has not yet been defined, but one model postulates that TNFR1-TRADD-RIP1 proteins are linked through their respective death domains (DDs). RIP1 and TRAF2 undergo K63 polyubiquitination by cIAP1/2 in conjunction with TRAF2 (not shown). Polyubiquitinated RIP1 and TRAF2 recruit TAK1, which activates NF-κB, thereby stimulating transcription of survival genes. Death effects of TNFR1 signaling are mediated via complex II, which forms following endocytosis of complex I, the dissociation of TNFR1, and the deubiquitination of RIP1 by CYLD and A20 (not shown). TRADD recruits FADD (DD-DD interactions), and FADD recruits procaspase-8 (DED-DED interactions). Unless inhibited, procaspase-8 undergoes activation and cleaves RIP1, rendering RIP1 unable to signal either survival or necrosis. Caspase-8 also activates downstream caspases inducing apoptosis. In contrast, if procaspase-8 is inhibited (genetically or pharmacologically), RIP1 is not cleaved and instead recruits RIP3. RIP1 and RIP3 undergo a complex set of phosphorylation events, and necrosis ensues through unclear mechanisms. One potential mechanism may involve the activation of catabolic pathways and ROS production as shown. A second necrosis pathway involves the mitochondrial permeability transition pore (MPTP) in the inner mitochondrial membrane and its regulation by cyclophilin D (CypD). This pore may be opened by increased Ca²⁺, oxidative stress, decreased ATP, and other stimuli that operate during ischemia/reperfusion and heart failure. Ischemia/reperfusion can lead to increased Ca²⁺ and ROS as depicted. MPTP opening results in profound alterations in mitochondrial structure and function as described in text, which results in decreased ATP. No definitive connections have been delineated between death receptor and mitochondrial necrosis pathways. A possible connection is RIP3-induced ROS generation.

Death receptors/RIP.

Experiments going back more than a decade demonstrate that death receptor activation can result in outcomes other than apoptosis, including survival, proliferation, and necrosis (53). This is illustrated by TNF signaling. Treatment of most cells with TNF alone induces little cell death because survival and death mechanisms are both activated. When survival mechanisms are inhibited, however, TNF induces apoptosis. In contrast, when caspases are inhibited, e.g., using z-VADfmk (N-benzyloxycarbonyl-valine-alanine-aspartic acid-fluoromethylketone), TNF elicits necrotic cell death.

Over the past decade, investigators have elucidated mechanisms that may explain some of these observations. Two distinct complexes are involved (14) (Figure 3). The binding of TNF to TNFR1 (TNF receptor 1) stimulates formation of complex I, which also includes the adaptor TRADD [TNFRSF1A (TNF receptor superfamily 1A)-associated via death domain], the serine/threonine kinase RIP1 (receptor interacting protein 1), TRAF2 (TNF receptor-associated factor 2), and inhibitor of apoptosis proteins cIAP1 and −2, which possess E3-ubiquitin ligase activity. In conjunction with TRAF2, cIAP1/2 stimulate K63 polyubiquitination of RIP1 and TRAF2 (54–56). Polyubiquitinated RIP1 and TRAF2 recruit and activate TAK1 (TGF-β-activated kinase 1) through its binding protein TAB (TAK1-binding protein) 2/3 (54, 55). TAK1, a MAPKKK (mitogen-activated protein kinase kinase kinase) then phosphorylates the IKK [IκB (inhibitor of κB) kinase] complex to activate NF-κB (nuclear factor κ light-chain enhancer of activated B cells) (54), a transcription factor that stimulates the expression of multiple survival genes. In summary, complex I promotes survival.

Complex I can transition to complex II. Changes include endocytosis of complex I, dissociation of TNFR1, deubiquitination of RIP1 by CYLD (cylindromatosis) and A20, and recruitment of FADD and procaspase-8 (14, 57, 58). Unless inhibited, procaspase-8 undergoes activation to stimulate apoptosis. Caspase-8 also cleaves RIP1, which renders RIP1 defective to signal either necrosis (59) (see below) or survival through NF-κB (60). Moreover, the C-terminal cleavage fragment of RIP1 may further drive procaspase-8 activation and enforce apoptosis (60). In contrast, if caspase-8 is inhibited pharmacologically or genetically, RIP1 is not degraded and is able to recruit RIP3 into a complex where both undergo phosphorylation (61, 62). One mechanism by which RIP3 may stimulate necrosis is through activation of metabolic pathways that lead to marked increases in reactive oxygen species (63). Thus, the RIP1-RIP3 complex appears important in TNF/z-VADfmk-induced necrotic death. The kinase activities of both RIP1 and RIP3 are critical for necrosis (53, 61). Identification of downstream targets will further define this pathway (64).

The general applicability of the RIP1-RIP3 pathway in mediating necrosis and its physiological significance awaits further investigation. Other pathways parallel to RIP1-RIP3 may also be important. To underscore this point, additional mediators have emerged from siRNA screens. One somewhat surprising candidate is Bmf (Bcl-2 modifying factor), a classic BH3-only protein with a well-defined apoptotic role (65) in some contexts that appears to also mediate necrosis in other situations (58).

MPTP/cyclophilin D.

Another important necrosis mechanism involves mitochondrial permeability transition (MPT) (66) (Figure 3). This refers to the opening of the mitochondrial permeability transition pore (MPTP) in the inner mitochondrial membrane, a nonselective channel allowing the passage of molecules of less than 1.5 kD. Opening of MPTP may be triggered by elevated matrix Ca²⁺ concentration, oxidative stress, elevated phosphate concentration, and adenine nucleotide depletion. Pore opening results in loss of the electrical potential difference (Δψ_m) due to the proton gradient that normally exists across the inner mitochondrial membrane. This results in ATP depletion. In addition, MPTP opening allows the equilibration of solutes across the inner membrane and the influx of water down its osmotic gradient into the mitochondrial matrix, which causes marked mitochondrial swelling, a hallmark of necrosis. In addition, the redundant inner mitochondrial membrane can swell without breaking and rupture the nonredundant outer mitochondrial membrane. This process can release mitochondrial apoptogens into the cytosol, although it is unknown whether these add significantly to necrotic death. Importantly, this outer mitochondrial membrane rupture differs from the outer mitochondrial membrane permeabilization that mediates apoptogen release in apoptosis.

The molecular composition of the MPTP is unresolved (66). It is clear, however, that cyclophilin D regulates pore opening. Cyclophilin D is a peptidyl-prolyl cis-trans isomerase, and this enzymatic activity is required for necrosis, although its critical substrates are not yet known. Transgenic overexpression of cyclophilin D results in spontaneous mitochondrial swelling and cell death. In contrast, mitochondria lacking cyclophilin D are remarkably resistant to Ca²⁺-induced swelling while still releasing cytochrome c in response to Bax and tBid. Primary cells lacking cyclophilin D are resistant to death induced by Ca²⁺ and oxidative stress but remain sensitive to killing by Bax. These data indicate that the mode of cell death mediated by cyclophilin D is distinct from apoptosis and bears the hallmarks of necrosis (67–69).

Ca²⁺ and proteases.

As noted above, apoptosis signaling is highly conserved between C. elegans and mammals. Accordingly, analyses in the worm have been carried out to identify genes that mediate necrosis. Gain-of-function mutations in MEC-4 (Mechanosensory 4), a channel reported to conduct Ca²⁺ and Na⁺, induce neuronal death with features of necrosis (70–72). This can be suppressed by knockdown or mutation of orthologs of the ER Ca²⁺ release channels ryanodine receptor (UNC-68) or IP3 receptor (ITR-1), or calnexin (CNX-1) or calreticulin (CRT-1). Furthermore, the reversal of MEC-4-induced necrosis by CRT-1 mutation can be restored by the drug thapsigargin, which increases cytosolic Ca²⁺ by inhibiting Ca²⁺ reuptake into the ER. These studies in C. elegans confirm the active nature of some instances of necrosis. In addition, they implicate cytoplasmic Ca²⁺ in necrotic death, an observation consistent with its role in MPTP opening.

Ca²⁺ can also activate calpains, which are noncaspase cysteine proteases. Cathepsins are lysosomal proteases that may be liberated during cellular stress (73). In C. elegans, loss-of-function mutations in orthologs of calpains (CLP-1 and TRA-3) and cathepsins (ASP-3 and ASP-4) suppress necrosis induced by Na⁺ and/or Ca²⁺ channels, suggesting that these proteases are involved in necrotic death in the worm (74). In light of these data, it is reasonable to consider the role of these proteins in mammalian cell death. Calpain is involved in some cell death signaling events, particularly those involving apoptosis (75). For example, calpain can cleave and activate the proapoptotic proteins Bid (76) and Bax (77). In addition, calpain-induced cleavage of Atg5 (autophagy related 5 homolog), a critical mediator of autophagy (discussed below), triggers apoptosis through translocation of an Atg5 fragment to the mitochondria, where it binds Bcl-x_L (78). Conversely, calpain can inhibit apoptosis by cleaving both upstream and downstream caspases (79, 80). The role of calpain in signaling aside, a major question is whether this or other proteases play general roles in the execution of necrosis similar to those of caspases in apoptosis. Only scant evidence, however, currently supports the notion of calpain or another protease functioning as a global mediator of protein degradation during necrosis. This also raises the fundamental question of whether an ordered proteolytic program plays a major role in the terminal phase of necrosis.

Syntheses of necrosis mechanisms.

Two areas of necrosis signaling, RIP kinases and cyclophilin D, have been described. Both appear important, although a connection between them has not yet been established. Moreover, these pathways provide at least partial explanations for some key features of the necrotic phenotype. For example, depletion of cellular ATP results at least in part from MPT. However, the precise mechanisms for some important features of necrosis, such as loss of plasma membrane integrity, remain to be elucidated. Another mechanism that appears relevant in at least certain instances of necrosis (e.g., DNA damage from alkylating agents) is activation of PARP [Poly (ADP-ribose) polymerase] (81). PARP polymerizes ADP ribose onto histones to facilitate DNA repair and, in the process, consumes NAD⁺, thereby depleting ATP. Some progress has been made in understanding a third core phenotype, the inflammatory reaction elicited by necrotic death. HMGB1 (High-mobility group box 1), a chromatin-binding protein, is released into the extracellular space following failure of the plasma membrane and can elicit an inflammatory response (82). Interestingly, HMGB1 is also capable of inhibiting phosphatidylserine-mediated clearance of apoptotic cells (83). Thus, HMGB1 may provide a switch from an apoptotic to a necrotic postdeath phenotype.

Autophagy

In contrast to necrosis and apoptosis, autophagy is primarily a survival mechanism. Autophagy is an intracellular recycling process in which organelles, proteins, and lipids are catabolized by lysosomal degradation (84). Autophagy provides cells with amino acids, free fatty acids, and energy in times of nutrient deprivation. In addition, it serves to regulate protein and organelle abundance and quality. There are three types of autophagy (85). Macroautophagy, the focus of this section, is discussed below. Chaperone-mediated autophagy, in which Hsc70 binds a target protein and escorts it to the lysosome, provides a means for the selective degradation of proteins (86). Microautophagy is thought to provide a constitutive quality-control function.

Macroautophagy (herein referred to as autophagy) is mediated by the formation of a double-membrane vesicle, known as the autophagosome, surrounding the material to be degraded. Autophagosome formation is initiated by vesicle nucleation involving Beclin-1, UVRAG (UV radiation resistance–associated gene), Vps34 (Vacuolar protein sorting 34), and IP3R, among others, and is followed by vesicle elongation mediated by the Atg12 and Atg8 conjugation pathways (85). The autophagosome fuses with the lysosome to form an autophagolysosome, resulting in lysosomal degradation of the cargo.

Two major pathways—mTOR (Mammalian target of rapamycin) and Beclin-1—regulate autophagy. Under fed conditions, mTOR inhibits autophagy by phosphorylating and inactivating Atg proteins (87, 88). Under starvation conditions, however, reduced class I PI3K-Akt signaling decreases mTOR activity, resulting in activation of autophagy. Autophagy is positively regulated by the BH3-only protein Beclin-1, which binds the class III PI3K Vps34, thereby facilitating autophagosome formation (89, 90). Antiapoptotic Bcl-2 proteins, such as Bcl-2 and Bcl-x_L, can bind the BH3 domain of Beclin-1 (91), and this interaction inhibits autophagy induced by starvation in the hearts of intact mice (92). Interestingly, the interaction between Bcl-2 and Beclin-1 does not affect the antiapoptotic properties of Bcl-2 (93).

Although the role of autophagy in promoting survival is well established, there has been considerable debate regarding whether autophagy is an independent form of cell death. Because autophagic morphology can be associated with cell death, the questions are whether the two processes are causally connected and, if they are, whether autophagy can be responsible for cell death. The most clear evidence that autophagy can cause cell death is provided by a genetic analysis of Drosophila salivary gland degradation during development, showing that both autophagy and apoptosis contribute to this process (94). Although these data indicate that autophagy can kill cells, they raise questions as to the molecular nature of autophagic cell death. Is it a unique form of cell death, or does it feed into any of the existing forms? A second critical question is what converts autophagy from a survival to a death process. The assumption that death is simply the result of too much autophagy has not been definitively tested. It remains possible that yet-to-be-identified factors provide a switch from survival to death.

CELL DEATH IN MYOCARDIAL INFARCTION

Sudden and sustained thrombotic occlusion of a coronary artery results in acute ST-segment elevation myocardial infarction (STEMI). The optimal treatment is prompt mechanical reperfusion. These human events are modeled in animals by surgical permanent coronary occlusion and prolonged coronary occlusion followed by reperfusion [the latter termed ischemia/reperfusion (I/R)]. The ischemic phase deprives the myocardium of oxygen, nutrients, and survival factors and results in the accumulation of waste products. The reintroduction of oxygenated blood into ischemic myocardium during the reperfusion phase results in oxidative stress, increased cytosolic and mitochondrial Ca²⁺, rapid resolution of intracellular acidosis (which paradoxically can result in MPTP opening), and inflammation, a syndrome termed myocardial reperfusion injury (95). Despite its potentially deleterious components, the net effect of reperfusion is to limit infarct size (96).

Both permanent coronary occlusion and I/R stimulate large amounts of myocardial cell death within the ischemic zone. This occurs in both myocytes and nonmyocytes (97, 98). In fact, nonmyocytes may be even more susceptible because the low levels of Apaf-1 in myocytes render these cells particularly sensitive to XIAP-mediated inhibition of apoptosis (99, 99a). This is consistent with the observation that cytochrome c can be found in the cytoplasm of some living myocytes (100, 101).

Cardiac myocyte death during permanent coronary occlusion and I/R occurs by apoptosis (102, 103), by necrosis (67, 68, 103, 104), and in association with autophagy (105–107). It can be problematic to compare the incidence, prevalence, and timing of the various types of cell death because of different sensitivities and windows of detection for death markers. It has been reported that cardiac myocyte apoptosis becomes maximal 4.5 h following permanent coronary occlusion, whereas necrosis peaks at 24 h (103). Reperfusion appears to accelerate the timing of apoptosis compared with permanent occlusion (108). In contrast to the modestly elevated, but chronic, levels of cell death during heart failure (discussed below), myocardial infarction is characterized by a large burst of cardiac myocyte death that is usually complete within 24 h. This section discusses the mechanisms and significance of cell death in the infarct zone. The heart failure section deals with the role of cell death in the remote myocardium during postinfarct remodeling.

Apoptosis in Myocardial Infarction

We focus here on in vivo studies that have established a causal connection between cardiac myocyte apoptosis and myocardial infarction. Much of this work has involved manipulations of the central apoptosis pathways. Both the extrinsic and intrinsic apoptosis pathways have been shown to be critical in the pathogenesis of myocardial infarction. Mice that lack Fas (lpr mice) exhibit marked reductions in infarct size following I/R compared with controls (109). Isolated buffered-perfused hearts from these mice exhibited a similar effect (110), demonstrating that deficiency of Fas, specifically in heart cells, is responsible. Interestingly, these hearts secreted several death ligands (including Fas ligand) into the perfusate early in reperfusion, suggesting that this pathway is activated by a paracrine loop.

On the basis of the pathogenic effects of Fas signaling, one might expect the TNF pathway to also mediate infarction. Surprisingly, deletion of TNFR1 or TNFR2 does not affect infarct size. In contrast, deletion of both results in significantly larger infarcts following permanent coronary occlusion (111). Increases in apoptosis, but not necrosis, were observed. Had loss of TNFR1 alone exacerbated infarct size, one might have postulated that abrogation of the RIP1-mediated NF-κB pathway was sufficient to abrogate survival. But exacerbation of infarct size required deletion of both TNF receptors, suggesting that TNFR2 also plays a protective role.

The intrinsic apoptosis pathway plays a central role in myocardial infarction. Indeed, cardiac-specific overexpression of antiapoptotic Bcl-2 substantially reduces infarct size, cardiac myocyte apoptosis, and cardiac dysfunction following I/R (112, 113). Similarly, germline deletion of the multidomain proapoptotic Bax reduces infarct size and lessens dysfunction following I/R in isolated, perfused hearts and following permanent coronary occlusion in intact mice (29, 114). In contrast to the functional overlap of Bax and Bak in some contexts (27), these data suggest that Bax and Bak are largely nonredundant in this setting.

Cleavage of Bid by caspase-8 connects the extrinsic and intrinsic pathways. Bid can also undergo activation by calpain cleavage, and both of these mechanisms operate during myocardial I/R (76, 115). Mice lacking Bid demonstrate reductions in infarct size following I/R (116). Targeted deletion of PUMA (p53-upregulated modulator of apoptosis), another BH3-only protein that is p53-responsive, also reduces infarct size following I/R in isolated, perfused hearts (117).

Studies have also assessed the roles of mediators of the postmitochondrial intrinsic pathway in myocardial infarction. Cardiac-specific transgenic overexpression of cIAP2 reduces infarct size following I/R of isolated, perfused hearts (118). In addition to inhibiting activated downstream caspases, cIAP2 also functions at TNFR1 complex I to K63 ubiquitinate RIP1 and, thereby, activate the NF-κB survival program (54–56). It would be informative to delineate the extent to which each of these mechanisms mediates the cardioprotective function of cIAP2.

The mitochondrial apoptogen Omi/HtrA2 promotes apoptosis by binding IAP proteins, thereby displacing already activated downstream caspases (46, 47). In addition, Omi/HtrA2 possesses a serine protease activity that cleaves IAPs to irreversibly inactivate them (48). UCF-101 is a small-molecule inhibitor of the serine protease activity. In two independent rodent studies, UCF-101 significantly reduced infarct size following I/R (119, 120). This reduction was accompanied by attenuation of IAP loss as well as decreases in downstream caspase activity, apoptosis, and cardiac dysfunction. Given potential redundancy from Smac/DIABLO, it is interesting that UCF-101 was so effective This may be related to the irreversibility of its serine protease activity. As in the case of cIAP2 above, UCF-101 may also exert indirect effects at TNFR1 complex I through its modulation of IAP abundance. These premitochondrial effects may also be important for the potency of cardioprotection afforded by UCF-101.

Treatment of rodents with a variety of polycaspase inhibitors reduced infarct size by 21–52% following I/R (121–124). Where examined, this reduction was accompanied by decreases in apoptosis and cardiac dysfunction. The mechanisms by which caspase inhibitors reduce infarct size have not been adequately explored. In some experimental systems, caspase inhibitors poorly rescue cell death, which is postulated to be due to mitochondrial dysfunction (e.g., from cytochrome c loss) induced by upstream apoptosis signals. This finding suggests that caspase inhibitors in the present experiments are also antagonizing upstream caspases.

In addition to cleaving ubiquitously expressed substrates that effect the death program, caspase-3 also cuts α-actin and troponin T in cardiac myocytes (125). Moreover, α-actin cleavage is associated with decreased contractile function in skinned fibers. It is not known, however, if cardiac myocytes with cleaved contractile proteins eventually die. If so, contractile protein cleavage would merely be a step in the cardiac myocyte death program. In contrast, if some cardiac myocytes remained viable, cleavage of contractile proteins could represent a mechanism for cardiac dysfunction independent of cell death. Although most current evidence suggests that caspase inhibition improves cardiac function by inhibiting cell death, the possibility that viable cells with cleaved contractile proteins exist in diseased hearts merits further investigation.

ARC is an apoptosis inhibitor that is highly enriched in cardiac and skeletal myocytes and in some neurons (126, 127). Although most apoptosis inhibitors antagonize either the extrinsic pathway or the intrinsic pathway, ARC blocks both through inhibition of DISC assembly, inhibition of Bax activation/mitochondrial translocation, and relocation of p53 to the cytoplasm (49–51). ARC blocks hypoxia-induced cytochrome c release and apoptosis (128). Transgenic overexpression of ARC decreases infarct size following I/R (129). The effect of ARC knockout on infarct size following I/R has been controversial, with one group showing an increase and the other no change (130; S.Y. Ji, S. Jha & R. Kitsis, unpublished data). It is possible that ARC gene deletion does not affect infarct size because ARC protein is normally rapidly degraded during I/R via the ubiquitin-proteasome pathway (131).

Adenoviral transduction of constitutively active (myristoylated) Akt into the myocardium markedly reduces cardiac myocyte apoptosis, infarct size, and cardiac dysfunction following I/R (132, 133). The potent cardioprotection provided by Akt presumably reflects its ability to phosphorylate and inactivate multiple apoptotic mediators, including Bad, Bax, Foxo, and many others. The targets that are most critical for cardioprotection in vivo, however, remain to be determined.

Necrosis in Myocardial Infarction

The most obvious connection between prolonged myocardial ischemia with or without reperfusion and necrotic cell death is MPT. At present, a precise connection with RIP1-RIP3 signaling is not known. Events during both ischemia and reperfusion can contribute to MPTP opening. Ischemia results in hypoxia, anaerobic metabolism, and intracellular acidosis. In response to acidosis, H⁺ is pumped out of the cell by the Na⁺/H⁺ exchanger, which leads to a rise in intracellular Na⁺. This excess Na⁺ is handled by the Na⁺/Ca²⁺ exchanger, which then leads to an increase in intracellular Ca²⁺. Additional elevations in intracellular Ca²⁺ result from Ca²⁺-induced Ca²⁺ release from the ER/SR (sarcoplasmic reticulum) (134) and reperfusion. Reperfusion also stimulates oxidative stress and results in rapid normalization of intracellular acidosis. Each of these events contributes to opening of the MPTP.

Cyclophilin D is an important regulator of MPTP. Cells lacking cyclophilin D are resistant to oxidative stress– and calcium-induced cell death, but sensitive to apoptotic stimuli. Mice null for cyclophilin D exhibit reduced infarct size following I/R (67, 68). These studies confirm that necrosis is a major form of cardiac myocyte death in I/R and demonstrate the importance of cyclophilin D in this process.

Although a connection between I/R and RIP1-RIP3 signaling remains obscure, necrostatin-1, a small-molecule inhibitor of the kinase activity of RIP1 (135), reduces infarct size following I/R (136, 137). Interestingly, necrostatin-1 does not reduce infarct size further in mice lacking cyclophilin D. This finding suggests a connection between RIP1 signaling and cyclophilin D. Although the molecular nature of such a connection is not known, one possibility is the generation of reactive oxygen species from activation of metabolic pathways by RIP3 during necrosis (Figure 3) (63).

Autophagy in Myocardial Infarction

Autophagy is induced during both permanent coronary occlusion and I/R, but the underlying pathways and functional consequences appear to differ (105). In permanent coronary occlusion, there is activation of AMPK (5′ adenosine monophosphate-activated protein kinase), a negative regulator of mTOR, which itself is a potent inhibitor of autophagy. Accordingly, autophagy is induced. Transgenic overexpression of dominant-negative AMPK inhibits autophagy and worsens infarct size (106). These data suggest that autophagy plays a protective role during prolonged ischemia, consistent with its function in helping cells survive nutrient deprivation, although AMPK also affects metabolism and apoptosis. In contrast, during reperfusion, autophagy is activated by a different pathway involving increases in Beclin-1 abundance. Augmentation of autophagy during reperfusion, however, is cytotoxic, as evidenced by decreased infarct size in beclin-1^+/− mice (105). These data suggest that autophagy plays a detrimental role during reperfusion. In contrast, another group has reported that autophagic flux is impaired during I/R and that genetic augmentation of autophagy protects cultured HL-1 cardiac myocytes in this setting (107). Accordingly, further investigation is required to resolve these differences.

CELL DEATH IN HEART FAILURE

Heart failure is a syndrome in which the heart is unable to meet the metabolic needs of the body. This situation refers to hearts with significant systolic dysfunction and pathological left ventricular remodeling consisting of chamber enlargement and wall thinning. Heart failure often results from prior myocardial infarctions. During the postinfarct period, the noninfarcted myocardium undergoes pathological remodeling in response to humoral and hemodynamic factors. Other important etiologies include cardiomyopathies, hypertension, valvular heart disease, and toxins. Heart failure is a multifactorial syndrome with abnormalities of cell signaling, Ca²⁺ handling and excitation-contraction coupling, contractile proteins and cytoskeleton, energetics, and cell death (138).

Apoptosis in Heart Failure

In contrast to myocardial infarction, in which large numbers of cardiac myocytes undergo apoptosis over a short period of time, failing hearts exhibit modestly elevated, but sustained, levels of apoptosis. Rates of apoptosis in the hearts of patients with end-stage dilated cardiomyopathy are 0.08–0.25% compared with 0.001–0.002% in controls (139–141). Although rates of apoptosis are clearly higher in failing compared with control hearts, the low absolute levels of cardiac myocyte apoptosis in heart failure patients raised questions as to whether this level of cell death could play a role in the pathogenesis of heart failure. This issue was addressed using transgenic mice with cardiac-specific expression of a modified allele of caspase-8 that resulted in low levels of apoptosis (142). Transgenic mice with rates of cardiac myocyte apoptosis as low as 0.023% developed a severe, dilated cardiomyopathy over 8 weeks and died within 2–6 months compared with healthy wild-type mice that exhibited apoptotic rates of ~0.002%. These results were not explained by a nonspecific effect of transgene expression, as mice expressing an enzymatically dead caspase-8 transgene were normal. Moreover, the abnormalities of cardiac anatomy and function were largely rescued by administration of a caspase inhibitor. These data demonstrate that modest, but persistent, levels of cardiac myocyte apoptosis can cause lethal heart failure.

Gαq transduces the activation of several receptors (type 1 angiotensin II receptor, α1-adrenergic receptor, endothelin receptor) that play roles in cardiac hypertrophy and failure (143). Cardiac-specific transgenic overexpression of Gαq results in cardiac hypertrophy and failure at baseline (144). In a subset of female Gαq mice, pregnancy precipitates fulminant heart failure accompanied by markedly increased rates of cardiac myocyte apoptosis and organismal death (145). Expression of Nix/Bnip3L [Nip3 (19 kD interacting protein-3)-like protein X/Bcl-2/adenovirus E1B 19 kD interacting protein 3-like], a BH3-like protein, is transcriptionally activated in the hearts of Gαq mice (146). Transgenic overexpression of Nix/Bnip3L results in massive cardiac myocyte apoptosis and death of the animals. Conversely, transgenic overexpression of sNix, a splice variant of Nix/Bnip3L that functions as a dominant negative, decreases cardiac myocyte apoptosis, improves cardiac function, and attenuates the mortality of pregnancy in Gαq transgenic mice (146). Thus, the Gαq-Nix/Bnip3L pathway plays a prominent role in the pathogenesis of heart failure.

Bnip3 (Bcl-2/adenovirus E1B 19 kD interacting protein 3) is another BH3-like protein that is induced during heart failure and hypoxia (147, 148). Bnip3 ablation does not affect infarct size following I/R, possibly because the window of hypoxia is too brief for Bnip3 induction. However, absence of Bnip3 results in decreased apoptosis, pathological remodeling, and cardiac dysfunction in the peri-infarct and remote myocardium following I/R. These results indicate that Bnip3 is a critical mediator of apoptosis and remodeling of the noninfarcted myocardium following myocardial infarction (149).

Loss of endogenous survival mechanisms may also precipitate heart failure. gp130 is a subunit of receptors for the interleukin-6 (IL-6) family of cytokines (150), including cardiotrophin-1, which mediates cardiac hypertrophy (151), LIF (leukemia inhibitory factor), IL-6, and oncostatin M. Cardiac-restricted knockout of gp130 exhibits no baseline phenotype. In contrast, transverse aortic constriction in gp130-null mice precipitates severe cardiomyopathy accompanied by massive apoptosis. Knockouts also exhibit loss of pressure overload–induced STAT3 (signal transducer and activator of transcription 3) phosphorylation but retain induction of the fetal gene expression. These data demonstrate that endogenous gp130 suppresses the development of pressure overload–induced heart failure (152).

Necrosis in Heart Failure

Although necrosis has long been recognized as a part of ischemic injury, it has only recently been appreciated as a mechanism of heart failure. One early study suggested that necrosis was even more frequent than apoptosis in failing human hearts (139). A recent study provided molecular evidence that necrosis mediates heart failure. Transgenic overexpression of the L-type Ca²⁺ channel β2a subunit in cardiac myocytes resulted in intracellular Ca²⁺ overload, myocyte necrosis, and heart failure (153). This phenotype was rescued by deletion of cyclophilin D, but not by overexpression of Bcl-2. These data show that cardiac myocyte necrosis can be a causal component of heart failure. Although the ability to rescue Ca²⁺ overload–induced heart failure provides proof of principle, it will be important to test whether knockout of cyclophilin D can also rescue postinfarct- and hemodynamic overload–induced models of heart failure.

Autophagy in Heart Failure

Failing human hearts exhibit an increased number of autophagosomes, suggestive of increased autophagy (154, 155). Whether increased autophagy serves a protective or a deleterious role in heart failure is not clear. Several studies have addressed this question. Deletion of Atg5 in the hearts of adult mice precipitates left ventricular enlargement and cardiac dysfunction, structural abnormalities of sarcomeres and mitochondria, increased ubiquitinated proteins, and cardiac myocyte apoptosis (156). Deletion of Atg5 at embryonic day (E)8.0 results in no abnormalities at birth, presumably due to compensation. The imposition of hemodynamic overload in these mice, however, precipitates cardiac enlargement and failure. This study concluded that autophagy is required under basal conditions for normal cardiac structure and function and that increased autophagy is a compensatory mechanism during heart failure. A second study used beclin^+/− mice to examine the role of autophagy in heart failure induced by severe pressure overload (157). Deletion of one allele of beclin-1 decreased autophagy, pathological cardiac remodeling, and cardiac dysfunction. Conversely, beclin-1 overexpression increased autophagy and pathological remodeling. This study concluded that autophagy plays a pathological role in hemodynamic overload–induced heart failure. The reason for these differing conclusions is not readily apparent, but there are clear differences between the studies, including different genetic models, potentially different degrees of severity of aortic constriction (probably more severe in the beclin-1^+/− study), and different readouts. beclin-1^+/− mice were also used to assess the role of autophagy in desmin-related cardiomyopathy due to the R120G αB crystallin mutation (158). In this model, the decrease in autophagy due to beclin-1 heterozygosity resulted in increased accumulation of polyubiquitinated proteins, aggregates, fibrosis, cardiac dysfunction and failure and increased mortality. The study concluded that autophagy plays an adaptive function in desmin-related cardiomyopathy. It is not clear why autophagy played a pathological role in hemodynamic overload–induced heart failure and an adaptive role in desmin-related cardiomyopathy, both investigated using beclin-1^+/− mice. There are, of course, differences between the inciting heart failure stimuli. Pressure overload hypertrophy may result in severe myocardial energy imbalance, whereas mutated αB crystallin causes proteotoxicity. Interpretation of these studies is further confounded by the absence of markers of autophagic cell death in contrast to autophagy itself. Further mechanistic studies will be required to dissect if these differences, or others, account for these different roles of autophagy in heart failure.

CELL DEATH THERAPEUTICS FOR HEART DISEASE

This section considers potential therapies for which some proof of principle exists. A more speculative discussion is provided elsewhere (2). In contemplating antideath therapies for heart disease, it is worth considering several conceptual issues. First, myocardial infarction, rather than heart failure, is likely to be the initial disease target because it requires treatment for only a limited period of time. This is an important consideration for several reasons, including the risk of carcinogenesis, which is increased with extended periods of cell death inhibition. Although the mortality and morbidity of heart failure may eventually justify accepting risks of cancer, myocardial infarction would probably be a more preferable initial target. Second, therapies that preserve mitochondrial structure and function by acting upstream of (or at) the mitochondria may be most likely to inhibit cell death. Third, small-molecule inhibitors are most practical.

Small-molecule polycaspase inhibitors reduce infarct size 21–52% following I/R in rodents and merit further consideration. This includes cell-based studies to further investigate the mechanism, which, as discussed above, is likely to include inhibition of upstream caspases. In addition, trials in large animals (e.g., pigs) are needed to gauge applicability to human disease.

UCF-101, a small-molecule inhibitor of the Omi/HtrA2 serine protease, shows promise based on rodent studies. Further mechanistic studies are needed to determine the extent to which this drug works at the level of the postmitochondrial intrinsic pathway versus TNFR1 complex I. Large-animal studies will be of interest regarding translation to human disease.

Cyclosporin A inhibits necrosis by blocking MPTP. The reduction in infarct size by cyclosporin A in the mouse is similar to that exhibited by cyclophilin D deletion (67). In a small study of patients with STEMI, cyclosporin A, administered immediately prior to angioplasty/stenting, reduced creatine kinase release (days 1–3; P < 0.04) and infarct size by MRI (day 5; P < 0.04) (159). Troponin I release (days 1–3) was not significantly affected. The study included a total of only 58 patients, only 27 of whom were studied by MRI. This pilot study suggests that cyclosporin A may reduce infarct size in human STEMI treated with reperfusion. But the numbers are too small to draw any firm conclusions, and evaluation of cardiac function and long-term follow-up are needed.

Necrostatin-1, which inhibits the kinase activity of RIP1, decreases infarct size 41–67% in mice, although the protocol involved 30 min of ischemia and only 2 h of reperfusion (136, 137). Nevertheless, these data and those from other animal studies (52, 160) suggest that necrostatin-1 shows promise in myocardial infarction and stroke. Again, large-animal studies would be helpful in assessing the compound’s potential efficacy for human patients.

POSSIBLE CONNECTIONS BETWEEN DEATH PROGRAMS

This section speculates as to potential links between apoptosis and necrosis. It should be emphasized that these concepts have been minimally investigated. However, a unified death map is important from a biological perspective. In addition, experimental validation of the concepts put forth below has potential therapeutic implications, as drugs that inhibit one pathway may shift death to another (e.g., caspase inhibitors shifting death receptor–induced apoptosis to necrosis).

MPTP opening results in catastrophic depletion of cellular ATP levels (Figure 4a). In addition, shifts in water down its osmotic gradient result in swelling of the inner mitochondrial membrane that can lead to rupture of the outer mitochondrial membrane with a subsequent release of apoptogens. Although this is not the mechanism of apoptogen release during apoptosis, outer mitochondrial membrane rupture can lead to the activation of downstream caspases (67). Although the mitochondrial consequences of MPTP likely kill the cell, one could view these downstream events as showing that necrosis and “apoptosis” can be linked in series, with necrosis upstream of apoptosis.

Figure 4.

Model of potential relationships between apoptosis and necrosis. Apoptosis and necrosis are postulated to be linked in series as well as to exist in parallel. (a) Necrosis triggering downstream apoptotic events. A necrotic death stimulus, acting through yet-to-be-defined upstream pathways, triggers mitochondrial permeability transition pore (MPTP) opening. This results in redistribution of solutes and water down their respective gradients, inner mitochondrial membrane swelling, and rupture of the outer mitochondrial membrane (OMM), all manifestations of necrosis. Due to OMM rupture, apoptogens are released to the cytoplasm and trigger apoptosome assembly and activation of procaspases-9 and -3 (67). This sequence of events places necrosis upstream of apoptosis signaling. Given the initial MPTP insult, it is unclear if the activation of downstream apoptotic events/caspases contributes significantly to cell death. (b) Apoptosis triggering necrosis. An apoptotic death stimulus triggers Bax/Bak-dependent permeabilization of the OMM (without rupture), resulting in release of apoptogens to the cytosol and leading to procaspase-3 activation. Caspase-3 is then presumed to cross the permeabilized OMM and enter the mitochondrial intermembrane space. Caspase-3 then cleaves NDUFS1 (161), an inner mitochondrial membrane protein that faces the mitochondrial intermembrane space. NDUFS1 is a component of respiratory complex 1, and its cleavage interrupts electron transport, leading to loss of inner mitochondrial membrane potential (Δψm), decreased ATP levels, and increased reactive oxygen species (ROS) levels. Mitochondria then become swollen, suggesting MPTP opening, although the latter was not assessed (161). In addition, the plasma membrane becomes leaky through a yet-to-be-determined mechanism. Mitochondrial swelling, ATP depletion, and plasma membrane failure are all hallmarks of necrosis. Expression of a noncleavable NDUFS1 mutant partially rescues this phenotype. In this scenario, apoptosis leads to necrosis.

It is possible that apoptosis and necrosis also exist in series with the opposite relationship: apoptosis upstream of necrosis (Figure 4b). In this case, a link could be caspase-mediated cleavage of NDUFS1 [NADH dehydrogenase (ubiquinone) Fe-S protein 1], a 75-kD subunit of respiratory complex 1 on the inner mitochondrial membrane (161). In this scenario, much of which has been demonstrated, apoptotic signals stimulate their expected events, including permeabilization of the outer mitochondrial membrane, apoptogen release, and procaspase-3 activation. Caspase-3 then cleaves NDUFS1. Caspase-3 presumably gains access to the mitochondrial intermembrane space through the already permeabilized outer mitochondrial membrane and is believed to gain access to NDUFS1 because NDUFS1 faces the intermembrane space from its position on the inner membrane. Cleavage of NDUSF1 disrupts electron transport, leading to loss of Δψ_m, decreased ATP levels, increased reactive oxygen species, mitochondrial matrix swelling, and plasma membrane permeabilization. Expression of a noncleavable NDUFS1 mutant leads to maintenance of Δψ_m and ATP levels, decreased reactive oxygen species, preservation of mitochondrial morphology, and delay in plasma membrane permeabilization. These events could be interpreted as apoptosis upstream of necrosis. In addition, NDUFS1 need not be the only link between caspase activation (apoptosis) and necrosis.

Thus, apoptosis and necrosis may exist in a variety of relationships: (a) in parallel and (b) in series, with either upstream of the other. It is also possible for some or all of these relationships to exist simultaneously. Should this scenario bear out, it would necessitate a more careful examination and review of the contribution of each type of cell death to disease pathogenesis. Although prior experiments would still be valid in terms of effects on reduction of total cell death, a reexamination may provide a more complete understanding of the mechanisms of cell death. For example, these concepts may impact the interpretation of experiments demonstrating that inhibition of apoptosis reduces infarct size. If apoptosis can beget necrosis, then inhibition of apoptosis may be working through inhibition of necrosis. Further mechanistic and genetic experiments are needed to test these concepts.

CONCLUDING COMMENTS

Apoptosis and necrosis of cardiac myocytes play critical roles in the pathogenesis of myocardial infarction and heart failure. Autophagy is also a feature of these syndromes, but its role in pathogenesis is currently unresolved. In addition, the relationship between autophagy and cell death is not yet defined at the mechanistic level.