Although necrotic PCD involves the loss of mitochondrial membrane potential (D Ym), this cell death program can be initiated by the organelles other than mitochondria (e.g. lysosomes, endoplasmatic reticulum, or nucleus) and proteases other than caspases (e.g., calpains or cathepsins). Indeed, it seems that initial, organelle-specific death responses finally lead to mitochondrial alteration. Mitochondria might function as a central integrator of the programmed necrotic pathway, thereby streamlining lysosome-, endoplasmatic reticulum-, or nucleus-elicited responses into a common pathway. A good example of such integration is programmed necrosis induced by extensive DNA-damage (Moubarak et al. 2007). This form of cell death is regulated by mitochondrial (AIF), cytoplasmic (Bax and calpains), and nuclear (PARP-1) effectors. Indeed, this recent work from our laboratory provides novel insights into the molecular mechanisms of programmed necrosis, since we have demonstrated the sequential activation of PARP-1, calpains, Bax, and AIF in a linear cell death pathway. Single ablation of PARP-1, the proapoptotic Bcl-2 member Bax, but not Bak, as well as inactivation of calpains or AIF prevented programmed necrosis.
In the following sections we will describe these key programmed necrosis effectors in further detail.
AIF was the first identified protein involved in caspase-independent cell death (Susin et al. 1999). AIF is expressed as a precursor of 67 kDa, which is addressed and compartmentalized into mitochondria by two mitochondrial localization sequences (MLS) located within the N-terminal prodomain of the protein. Once in mitochondria, this prodomain is removed, giving rise to a mature form of ~62 kDa (Susin et al. 1999; Otera et al. 2005). This form comprises three structural domains: FAD-binding domain, NADH-binding domain, and C-terminal domain (Lorenzo and Susin 2004, 2007). Under physiological conditions, AIF remains confined to the internal mitochondrial membrane, with its N-terminal region oriented towards the matrix (Otera et al. 2005). AIF is here a mitochondrial FAD-dependent oxi-doreductase that plays a role in oxidative phosphorylation (Mate et al. 2002; Vahsen et al. 2004; Miramar et al. 2001). However, after a cellular insult, AIF is cleaved by calpains and/or cathepsins to yield tAIF (truncated AIF) a 57 kDa form of AIF (Otera et al. 2005; Polster et al. 2005; Yuste et al. 2005a). tAIF then translocates from mitochondria to cytosol and further to the nucleus, where it interacts with DNA and causes caspase-independent chromatin condensation and participates in large-scale (~50 kb) DNA fragmentation (Susin et al. 1999; Ye et al. 2002). The mitochondrial release, nuclear translocation, and DNA fragmentation associated with AIF have now been extensively demonstrated in several cell death systems and cell types (~900 references in Medline to date).
Three additional AIF isoforms have been characterized (1) AIF-exB, generated by an alternative use of exon 2b (rather than exon 2) (Loeffler et al. 2001); (2) AIFsh (AIF short) derived from an alternative transcriptional start site located at intron 9 of AIF (Delettre et al. 2006a); and (3) AIFsh2 (AIF short 2), produced by the alternative use of the newly discovered exon 9b (Delettre et al. 2006b). While AIF-exB is similar to AIF in terms of regulation and subcellular distribution (Loeffler et al. 2001), AIFsh and AIFsh2 are respectively restricted to cytoplasm and mitochondria, at least under physiological conditions (Delettre et al. 2006a, b). Upon the induction of cell death, like AIF, AIFsh translocates to the nucleus leading to large-scale (~50 kb) DNA-fragmentation (Delettre et al. 2006a). In contrast, AIFsh2, which presents oxidoreductase but no proapoptotic activity, translocates from mitochondria to cytosol after an apoptotic insult but cannot be further translocated to the nucleus (Delettre et al. 2006b).
As indicated above, AIF release from mitochondria occurs in a number of cell types, independently of the nature of the death-triggering signal (Kim et al. 2003a; Murahashi et al. 2003; Zhu et al. 2003). The requirement of nuclear translocation for AIF-mediated induction of caspase-independent, apoptosis-like PCD has been explicitly demonstrated by microinjection experiments of AIF-directed antibodies, which potently inhibited AIF targeting to the nucleus as well as cell death (Susin et al. 1999; Yu et al. 2002; Cregan et al. 2004). Following nuclear translocation, the integrity of AIF's C-terminal domain, but not of the oxi-doreductase-related (NADH) and FAD-binding N-terminal domains, is required to trigger PCD (Miramar et al. 2001; Loeffler et al. 2001). This was confirmed by a study showing that AIFsh can trigger cell death despite not having NADH and FAD-binding domains (Delettre et al. 2006a).
Interestingly, it is currently unknown whether the mechanisms of mitochon-drial AIF release might be different and conditioned by the organelle at which the cell death program has been initiated (i.e. nucleus in the case of DNA damage-induced death vs. mitochondria in the case of Ca2+ overload). This is of a particular interest because both of these pathological alterations have been involved in triggering programmed necrosis (see sections "Increase in intracellular Ca2+ concentration" and "DNA damage"). It is important to keep in mind that different organelles (e.g. nucleus and mitochondria) might be engaged in an orderly manner to mediate induction of the cell death program. For example, the nucleus might be the site of cell death process initiation in the paradigm of genotoxic DNA damage, whereas it can be involved in a secondary fashion when oxidative DNA damage is triggered by reactive oxygen species (ROS) generated after mitochondrial impairment (Fig. 3.1).
The Bcl-2 family includes proteins with antagonistic functions that either cause programmed cell death or inhibit it. Their structures in solution are similar to those of bacterial channel-forming toxins. Indeed, just like these toxins, the Bcl-2 proteins cycle between soluble and membrane-associated forms, and form channels in mitochondrial membranes. Bax is a member of the Bcl-2 family that, together with Bak, belongs to the multidomain BH1-BH3 (Bcl-2 homology domains 1-3) pro-apoptotic Bcl-2 proteins (Kuwana et al. 2002, 2005). Both Bax and Bak act by sequestering antiapoptotic members of the Bcl-2 family, such as Bcl-2 and Bcl-xL (proteins containing BH1-BH4 domains). Other proapoptotic
Fig. 3.1 (continued) compatible with the reported Bax-dependency (Moubarak et al. 2007). It has to be noted that the endonucleases recruited by tAIF (circle 9 and square 12) are probably distinct. In addition, there might exist a cross-talk between these two distinct pathways leading to different AIF-mediated PCD outcomes. Indeed, intra-cellular Ca2+ over-load might lead to increased ROS production subsequent to mitochondrial metabolism impairment. Among other deleterious effects, ROS might mediate oxidative damage of DNA, which can in turn trigger PARP-1 activation. This will result in the switch from the Ca2+-induced pathway depicted in the lower part of the figure to the DNA-damage-induced pathway depicted in the upper part of the figure. An interesting possibility is that these two pathways might operate in parallel in neurons exposed to some acute injuries, and thus explain the existence of "hybrid" death phenotypes seen, for example, in neurons exposed to ischemia that display both necrotic features and TUNEL-positivity (Unal-Cevik et al. 2004)
Fig. 3.1 Unifying model for AIF-mediated PCDs. The proposed model is based on the initial trigger of PCD: the increase in Ca2+ intracellular level for apoptosis-like PCD (lowerpart of the scheme) and DNA damage for programmed necrosis (upper part of the scheme). Initial rise in Ca2+ level (circle 1) in neurons is mainly due to the activation of the NMDA type of glutamate receptors (NMDAR) (Cheung et al. 2005). The increase in Ca2+ level may lead to the opening of the high conductance permeability transition (PT) pores in the mitochondrial inner membrane (IMM), as well as to subsequent mitochondrial membrane depolarization associated with mitochondrial permeability transition (MPT), oxidative phosphorylation uncoupling, and mitochondrial swelling (circle 2) (Kim et al. 2003b). In parallel, the increase in intracellular Ca2+ level leads to calpain activation (circle 3); activated calpain might enter the mitochondrial inter-membrane space via putative PT pore (circle 4) and subsequently cleave AIF from the inner mitochondrial membrane yielding truncated AIF (tAIF) (circle 5). The latter is then released from mitochondria (circle 6) and is translocated to the nucleus (circle 7), where it participates in chromatin condensation (circle 8) and DNA fragmentation (circle 9) (Susin et al. 1999; Susin et al. 2000; Susin et al. 1996; Moubarak et al. 2007; Susin et al. 1997; Yuste et al. 2005b). This hypothetical model is compatible with reported Bax-independence of NMDA-induced apoptosis-like PCD (Cheung et al. 2005). When the initial lesion concerns DNA damage (square 1), PARP-1 activation and increased synthesis of PAR polymers occur (square 2). PARs can be released from the nucleus (Andrabi et al. 2006; Yu et al. 2006) and can trigger the loss of mitochondrial membrane potential by a yet unknown mechanism. The latter might be associated with the release of Ca2+ from mitochondrial Ca2+ stores (square 4) and subsequent calpain activation (square 5) (Heeres and Hergenrother 2007). Activated calpain might first activate (square 6) and then trigger Bax translocation to the mitochondria (square 7). Bax-formed pores on the outer mitochondrial membrane allow activated calpain to enter the inter-membrane space (square 8) and reach AIF, which is imbedded at the inner mitochondrial membrane. AIF is then cleaved by calpain (square 9) and tAIF is released from mitochondria (square 10). After translocation of tAIF to the nucleus (square 11), tAIF might associate with an unknown endonuclease (square 12) to cleave DNA in TUNEL-positive fragments (Moubarak et al. 2007). This hypothetical model for programmed necrosis induction is members are characterized by the presence of only one BH domain (BH3-only). Among them, two groups can be distinguished based on the mechanism of action. BH3-only proteins Bim and tBid (truncated Bid) cooperate with Bax and Bak to induce PCD (Youle and Strasser 2008), whereas BH3-only proteins such as Bad, Puma, and Noxa act as apoptosis derepressors. These derepressors act through a competitive binding of Bcl-2 and Bcl-xL which, by rendering Bax and Bak free, favors the mitochondrial-dependent PCD program (Kuwana et al. 2005; Certo et al. 2006).
Bax and Bak are key players in triggering apoptosis. However, genetic deletion of either Bak (Lindsten et al. 2000) or Bax (Knudson et al. 1995) displays relatively low phenotypic impact, although Bax deficiency results (among other alterations) in a mild neuronal overgrowth (Knudson et al. 1995). By contrast, double Bax/Bak deletion leads to severe developmental defects pointing to a functional redundancy of these two Bcl-2 multidomain proteins (Lindsten et al. 2000). Importantly, the cells lacking both Bax and Bak are resistant to death-inducing signals that might trigger programmed necrosis (Wei et al. 2001).
Bax activation is causally related to the initiation of MOMP. In the absence of apoptogenic stimuli, Bax is an inactive cytoplasmic protein that is constitutively expressed at a relatively constant level. Its activation results mainly from its interaction with other members of Bcl-2 family. It has been shown (at least in cell-free membrane permeabilization assays) that Bax acts in synergy with the BH3-only proteins tBid or Bim to trigger apoptosis, probably through a direct (physical) interaction (Kuwana et al. 2005; Certo et al. 2006; Willis and Adams 2005). However, neither of these two BH3-only proteins appears to be required for Bax activation, as suggested by the data from double Bim/Bid knockout mice (Willis et al. 2007). Alternatively, Bax could be activated by a large panel of apoptotic regulators, such as JNK, through Bad phosphorylation (Donovan et al. 2002), or calpains, by calpain-dependent proteolytic cleavage (Wood et al. 1998; Gao and Dou 2000).
Since Bax is mostly a cytoplasmic, monomelic protein (in contrast to Bcl-2 which is mainly embedded in the outer mitochondrial membrane, OMM), the translocation from cytoplasm to mitochondria is an essential step for its death-inducing activity. Translocation is followed by Bax insertion into OMM via its C terminus, resulting in a conformational change revealing a hidden epitope in the N terminus. This translocation-related conformational alteration coincides with the increased oligomerization capacity of Bax, which is in turn directly linked to its membrane pore-forming activity. It has been shown that in liposomes, Bax dimers form a pore of approximately 11 A whereas tetramers give rise to 22-A pores (Korsmeyer et al. 2000). Thus, Bax alone is capable of forming pores of increasing size in mitochondrial membranes. Consequently, translocation of more Bax from cytoplasm to mitochondria increases the probability of multimeric Bax complex formation, giving rise to pores of increasing size. It is then reasonable to assume that the increase in the size of Bax-formed pores is time-dependent, which is compatible with the release of proteins with increasing mass from the intermembrane mitochondrial space (Munoz-Pinedo et al. 2006).
Calpains encompass a family of calcium-dependent, nonlysosomal proteases characterized by a cysteine-protease domain that includes a conserved catalytic sequence Cys-His-Arg combined with a calmodulin-like Ca2+-binding site (Sorimachi and Suzuki 2001). The configuration of the cysteine-protease domain determines the formation of an active catalytic pocket, which only occurs when calcium is present. Calpains are involved in a large variety of calcium-regulated processes, such as signal transduction, cell proliferation, and PCD (Goll et al. 2003).
The calpain family comprises a heterogeneous group of cysteine proteases with a large expression pattern. The proteases belonging to this family have been subdivided into three groups depending on their primary structure and the presence or absence of regulatory subunits. These three groups are defined as: typical (also called ubiquitous or conventional), atypical, and other EF-calpains (Goll et al. 2003; Saez et al. 2006). The most important and most studied calpains are the typical |-calpain (calpain I) and m-calpain (calpain II). The terms and m-calpain were initially used to refer to the micromolar or millimolar Ca2+ concentration needed to activate |- and m-calpain, respectively (Goll et al. 2003). Both and m-calpain are heterodimeric enzymes (80 kDa) sharing a common small regulatory subunit protein of 28 kDa encoded by the CAPN4 gene. This regulatory subunit protein is critical for calpain function, since its genetic ablation leads to a complete blockage of calpain activity (Tan et al. 2006). Calpains are activated by binding of Ca2+, followed by an autolytic cleavage at the N-terminal moiety of the protein. These proteases do not recognize any specific amino acid sequence. The only amino acid specificity that has been reported involves small hydrophobic amino acids (e.g. leucine) at the P2 position, and large hydropho-bic amino acids (e.g. phenylalanine) at the P1 position (Cuerrier et al. 2005).
Calpains are involved in the regulation of both apoptosis (type 1 PCD) (Toyota et al. 2003; Cartron et al. 2004; Cao et al. 2003a) and AIF-mediated programmed necrosis (Moubarak et al. 2007). The mechanism of calpain involvement in apop-tosis is relatively well understood. It has been initially associated with the cleavage of the cytoskeleton protein fodrin (Nath et al. 1996) and to proteolytic activation of caspases-3 and -12. In addition to such positive regulation, calpain-mediated degradation of p53 and caspases-7 and -9 inhibits apoptosis (Goll et al. 2003). By contrast, the mechanisms by which calpains control programmed necrosis are less known. It has, however, been demonstrated that calpains control mitochondrial AIF-release subsequent to the extensive DNA damage provoked by high doses of alkylating agents such as N-methyl-N'-nitro-N-nitrosoguanidine (MNNG). Chemical inhibitors of the enzyme activity, calpain knock-down, or CAPN4 genetic ablation, preclude mitochondrial AIF-release and DNA damage-induced programmed necrosis (Moubarak et al. 2007).
The relationship between calpains, AIF, and the Bcl-2 family of proteins deserves particular attention. Indeed, proapoptotic members from the Bcl-2 family, such as Bax (Cao et al. 2003a) and Bid (Mandic et al. 2002), are processed and activated by calpains. Using a model of sympathetic neurons, it has been shown that Bax translocation from cytosol to mitochondria is a critical event for neuronal type
1 PCD (apoptosis). In neurons, calpains cleave inactive Bax (21 kDa) into an 18-kDa proapoptotic protein that redistributes from cytoplasm to mitochondria and promotes PCD (Cao et al. 2003a). In a similar way, cleavage of Bid by calpains has also been implicated in mitochondrial permeabilization, AIF release, and PCD (Chen et al. 2001b). Indeed, it has been recently proposed that |-calpain cleaves Bid into tBid (Polster et al. 2005). Next, tBid appears involved in the permeabilization of OMM and favors calpain access to the mitochondrial intermembrane space (Polster et al. 2005). Once in mitochondria, |-calpain cleaves AIF yielding its proapoptotic form, tAIF. This interaction has been further confirmed in focal cerebral ischemia (Culmsee et al. 2005).
Poly(ADP-ribose) polymerase (PARP) is a generic term for a family of enzymes catalyzing the production of poly(ADP-ribose) or PAR polymers from NAD+. The family consists of five members (PARP-1 to 5) with distinct cellular and subcellular localizations (Moroni 2008). PARP enzymes participate in the control of chromosomal segregation, DNA replication, and gene transcription along the relevant phases of the cell cycle. They are also involved in the regulation of chromatin structure and DNA repair (Schreiber et al. 2006).
After DNA-damage, activation of PARP results in poly(ADP-ribosyl)ation of key DNA-repair proteins at the expense of NAD+ that is cleaved into ADP-ribose and nicotinamide. When DNA-damage is limited, this physiological machinery could repair the injury (Haince et al. 2005; Shall and de Murcia 2000). If DNA breaks are repaired, the damaged cell survives and the cellular NAD+ levels are restored by recycling nicotinamide with two ATP molecules. If DNA repair is not completely achieved, cells undergo apoptosis by a caspase-dependent mechanism. When DNA-damage is extensive, the cell cannot repair the injury. In this case, disproportionate activation of PARP (mainly PARP-1) depletes the cellular pools of NAD+ and ATP, driving the cell to a programmed necrosis PCD pathway (Haince et al. 2005; Shall and de Murcia 2000).
Traditionally, this kind of cell death was attributable to the fact that PARP-1 activation correlated with a rapid cellular depletion of NAD+ and ATP. For example, IL-3-dependent hematopoietic cells represent one of the relatively well-known paradigms of this cell death, where PARP-1 activation results in a rapid loss of cellular ATP and programmed necrosis (Zong et al. 2004). In these cells, which are highly glycolytic, IL-3 deprivation, glucose withdrawal, or glycolytic blockade renders MNNG inefficient (Zong et al. 2004). Moreover, IL-3 deprivation, which blocks the cellular glycolysis, preserves ATP levels after MNNG treatment, even if PARP-1 remains activated. This suggests that the glycolytic control of the metabolic state of the cell regulates DNA-damage mediated death (Zong and Thompson 2006; Zong et al. 2004; Nelson 2004). Albeit necessary, PARP-1 activity is not sufficient to induce programmed necrosis in IL-3-dependent hematopoietic cells. In contrast, our recent results demonstrated that in MEFs, glucose deprivation, glycolytic blockade by methyl-pyruvate, or oxidative phosphorylation inhibition did not preclude either ATP loss or necrosis associated with MNNG (Moubarak et al. 2007). Moreover, PARP-1 genetic ablation blocks both NAD+/ATP loss and alkylating DNA-damage induced PCD. Hence, in MEFs, PARP-1 activity is mandatory. Therefore, it seems that depending on the cell type, there are different mechanisms governing DNA-damage mediated programmed necrosis.
New exciting studies confirm the hypothesis that energy collapse is not the sole mechanism by which PARP-1 contributes to cell death. A new "death link" between PARP-1, mitochondria, and AIF has now been established in an in vitro model of glutamate excitotoxicity via N-methyl-D-aspartate (NMDA)-type receptors (Andrabi et al. 2006; Yu et al. 2006). These recent studies from the Dawson laboratory elegantly demonstrated that the PAR polymers, a major product of PARP-1 activation (Haince et al. 2005), are the death signals that provoke AIF release from mitochondria to cytosol (Andrabi et al. 2006; Yu et al. 2006). The precise mechanism by which PAR polymer induces AIF mitochondrial release is not totally elucidated. The authors propose two possibilities (1) induction of D Ym dissipation due to the charged nature of PAR polymers and (2) binding of PAR polymers to a yet unknown mito-chondrial partner that, in turn, induces AIF release (Yu et al. 2006). Further analysis is necessary to clarify if there exists a relationship between these polymers and other AIF activators such as calpains and/or cathepsins. In this regard, high PAR levels associated with the opening of TRPM2 Ca2+ channels and subsequent lethal rise in intra-cellular Ca2+ (Fonfria et al. 2004) could provoke the calpain activation needed for AIF release. In any case, it becomes clear that a better understanding of the connection between PAR and AIF can guide the development of new therapeutic strategies regulating AIF-mediated PCD pathways in NMDA-mediated excitotoxicity.
The involvement of PARP in glutamate-mediated neuronal death with necrotic morphology has also been reported in animal models of ischemia and stroke [for review, see Moroni (2008)]. In this case, the mechanisms underlying glutamate-triggered neuronal cell death have been related to excitotoxicity via both ionotropic NMDA- (Moroni et al. 2001; Meli et al. 2004) or metabotropic mGluR1 receptors (Meli et al. 2005).
Was this article helpful?