Mechanisms of Cell Death in Hypoxia/Reoxygenation Injury

Pothana Saikumar*,1, Zheng Dong1, Joel M Weinberg2 and MA Venkatachalam1

1Department of Pathology, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, Texas 78284, USA; 2Department of Medicine, University of Michigan, Ann Arbor, Michigan 48109, USA

Investigation of death pathways during cell injury in vivo Cells may die during ischemia itself, in which case, caused by ischemia and reperfusion is of clinical early membrane damage leads to activation of multiple importance, but technically dicult. Heterogeneity of degradative systems in an uncontrolled fashion. This cell types, dierences between organ systems, diversity of form of death is necrotic; dead cells are found in the death paradigms and exacerbation of tissue damage cores of ischemic regions, but may also be diusely caused by in¯ammation are only some of the variables distributed if the ischemic period is relieved by that need to be taken into account. With respect to the reperfusion of blood. Of much clinical and scienti®c identi®cation of necrosis and apoptosis in aected interest is the additional loss of parenchymal cells that organs, technical issues related to preparation artifacts, occurs during reperfusion following ischemia. Sublethal occurrence of internucleosomal DNA cleavage in necrotic ischemic damage may cause some cells to undergo as well as apoptotic cells and other overlaps in death `reperfusion injury' when reoxygenated by blood pathways bear consideration. In that caspase independent re¯ow. Because availability of oxygen and nutrients as well as caspase dependent processes cause cell death during reperfusion allows ischemic cells to restore ATP and that caspase inhibitors can act as anti-in¯ammatory pools, additional losses of cells during reperfusion agents, evaluation of ischemic death mechanisms in could be apoptotic, since this form of cell death parenchymal cells needs to be performed with caution. requires energy (Eguchi et al., 1997; Leist et al., When the eects of in¯ammation are removed by 1997). There have been reports of reperfusion injury appropriate in vitro studies using puri®ed or cultured by apoptosis, and several mechanisms have been cells, several mitochondrial factors that lead to cell death proposed (Piper et al., 1998; MacLellan and Schnei- can be studied. Substantial evidence exists for the der, 1997; Chopp and Li, 1996). However, there is participation of electron transport defects, mitochondrial controversy whether necrosis and apoptosis are permeability transitions (MPT) and release of cyto- uniquely dierent with respect to pathogenetic chrome c from mitochondria, eected by pro-apoptotic mechanisms, and if they can be reliably distinguished. proteins such as Bax. The anti-apoptotic protein Bcl-2 Also, there is uncertainty with respect to death of exerts an overriding protective role in this type of injury parenchymal cells caused by ischemia and reperfusion by preserving mitochondrial structure and function. In per se, relative to cell death caused secondarily by contrast, caspase inhibitors cannot oer long-term in¯ammation in the infarct. In fact, if the identity of protection to ischemically injured parenchymal cells dead cells is not established by morphology or cellular regardless of how eectively they can inhibit apoptotic markers, it is dicult to know if the aected cells are events, if the cells have suered permanent mitochondrial parenchymal, endothelial or in¯ammatory. This damage impairing respiration. diculty is compounded in studies of the central nervous system, which exhibits intimate intermingling Keywords: apoptosis; necrosis; Bax; Bcl-2; hypoxia; of glial cells and neurons. Under these conditions it is reoxygenation dicult to assess if pharmacological agents, including caspase inhibitors, ameliorate ischemic injury by decreasing in¯ammation or by inhibiting damage mechanisms speci®c to parenchymal cells. Such cell Introduction speci®c injury pathways are more easily identi®ed by subjecting puri®ed cells or cell cultures to hypoxia and Ischemia of tissues during arterial occlusion, shock and substrate deprivation in order to simulate ischemia. organ transplantation is a common and important cause of cell death. Aected tissues exhibit profound losses of adenosine triphosphate (ATP), which trigger Factors that may determine expression of cell death as multifarious degenerative processes. Declines of ATP apoptosis or necrosis depend not only on hypoxia and attendant cessation of oxidative phosphorylation, but also on failure of Apoptosis and necrosis were originally described to be anaerobic glycolysis. For this reason, ischemic cell two distinct forms of cell death, that can be clearly damage is more severe and proceeds faster than distinguished (Wyllie, 1994). However, these two hypoxic injury under conditions which permit anaero- types of cellular demise can occur simultaneously in bic glycolysis. tissues or cultured cells exposed to ischemia-hypoxia and reoxygenation (Kato et al., 1997; Kajstura et al., 1998; Li et al., 1998a; Lieberthal and Levine, 1996; Noda et al., 1998; Weigele et al., 1998). The intensity and duration of insults may decide the outcome *Correspondence: P Saikumar (Bonfoco et al., 1995; Leist and Nicotera, 1997). Apoptosis in reoxygenation injury PSaikumaret al 3342 Thus, in certain cases, triggering events can be permeability, in a `ladder' pattern, indistinguishable common for both types of cell death. However, a from that seen in apoptosis (Dong et al., 1997; see downstream controller may be required to direct cells Figure 1). The serine protease inhibitor TPCK, but not towards the organized execution of apoptosis. If the the caspase inhibitors Z-VAD-fmk and Z-DEVD-fmk apoptotic program is aborted before this control point, inhibited DNA ladders (Figure 1a). Moreover, DNA and the initiating stimulus is severe, cell death may still ladders and caspase activation could be dissociated result, by necrosis (Leist et al., 1997). Consistent with during fas mediated apoptosis of Jurkat cells (Figure this view, caspase inhibition was sucient to prevent 1b). In another study, increases of Ca2+ in chemically proteolysis and DNA fragmentation in Bax or drug hypoxic cells led to nuclear condensation, hydrolysis of induced apoptosis, but did not prevent plasma lamins A and C and DNA cleavage into large 50 ± membrane damage and cell death (Xiang et al., 1996; 150 kb fragments indistinguishable from those seen in Lemaire et al., 1998). The inability of caspase the early stages of apoptosis; with the onset of plasma inhibitors to confer long term viability under some membrane damage, typical DNA ladders also formed circumstances of injury could be related to the (Dong et al., 1998). These considerations underscore involvement of mitochondrial factors in transducing the need for caution when the relative contributions of apoptotic signals (McCarthy et al., 1997; Saikumar et necrotic and apoptotic programs in ischemic or al., 1998). hypoxic cell death are evaluated. Apoptotic processes may involve at least two dierent mechanisms initiated by distinct caspases. When death receptors are ligated, they recruit and proteolytically Cell death in ischemia-hypoxia followed by activate the initiator caspase-8 (Ashkenazi and Dixit, reoxygenation 1998) that in turn triggers eector caspases. This cascade can be ampli®ed by caspase-8 mediated mitochondrial The apoptotic process of cell death is divided into three damage and release of apoptogenic factors (Luo et al., phases: the initiator or induction phase, which depends 1998; Li et al., 1998b). Because the proximate trigger on speci®c death-inducing signals, the eector phase, in involves the initiator caspase, caspase inhibitors abort which an execution mechanism is activated, and the the entire death cascade and provide complete protec- degradation phase, during which cells show morpholo- tion. However, other apoptotic stimuli lead to the gical and biochemical features of apoptosis. What is the activation of caspase-9 and require a proximate trigger nature of the initiating mechanism that could lead to that involves increased mitochondrial permeability apoptosis during reoxygenation injury? Several possibi- mediated by the mitochondrial permeability transition lities need to be considered. First, it has been reported and/or release of cytochrome c (Zhou et al., 1997; Li et that expression of Fas ligand during hypoxia/reoxygena- al., 1997). Whereas anti-apoptotic members of the Bcl-2 tion may trigger Fas dependent death pathways family can abort the death cascade by preventing (Matsuyama et al., 1995; Nogae et al., 1998; Vogt et mitochondrial damage (Yang et al., 1997; Kluck et al., al., 1998). In some models it has been shown that 1997), caspase inhibitors have the power only to prevent inhibition of caspase-1 (interleukin-1b-convertase; ICE) structural damage caused by downstream execution conferred neuronal protection (Hara et al., 1997). ICE events. Because they cannot prevent irreparable mito- activates pro-IL-1b to generate mature IL-1b, and this chondrial damage, cell death is assured (Xiang et al., proin¯ammatory cytokine is also a mediator of ischemic 1996; McCarthy et al., 1997; Miller et al., 1997; Saikumar injury. These results are further supported by the et al., 1998), and may be expressed as necrosis. observation that mutant mice de®cient in the ICE gene are less severely injured by brain focal ischemia as compared with wild-type controls (Schielke et al.,1998). Biochemical overlaps between apoptosis and necrosis How IL-1b induces ischemic injury is still unclear. Second, reoxygenation injury has been attributed to Identi®cation of apoptotic and necrotic death processes the toxic eects of oxygen free radicals generated requires more rigorous methodology than has been during reperfusion of organs with oxygenated blood generally used (McCarthy and Evan, 1998). Artifacts (McCord, 1985). Ample evidence exists for the over- inherent in techniques that detect breaks in double production of radicals in reperfused organs (Chan, stranded DNA as well as overlaps between apoptotic 1994; Flitter, 1993; Gonzalez-Flecha and Boveris, 1995; and necrotic death programs account for these Johnson and Weinberg, 1993), but following an initial problems (McCarthy and Evan, 1998). Internucleoso- increase, cellular concentrations are buered sharply mal DNA cleavage and false positive TUNEL staining down by endogenous antioxidants (Flitter, 1993; of nuclei are readily induced in tissues even by short Gonzalez-Flecha and Boveris, 1995). It is also well delays in tissue preparation (McCarthy and Evan, known that invading in¯ammatory cells add an 1998). These methodological considerations are rele- element of oxidant damage to reperfusion injury that vant to ischemically injured tissues; the necrotic mode may take place by other mechanisms (Engler, 1989; of cell death is abundantly expressed during ischemic Granger, 1988; Thiagarajan et al., 1997). Nonetheless, injury, and the nature of the pathology makes tissue it is not known whether the production of oxygen preparation prone to artifacts. Recent studies on ATP radicals in parenchymal cells of reoxygenated organs is depletion induced by chemical hypoxia show that there the cause or result of injury. are overlapping patterns of nuclear breakdown and A third possibility for which substantial support DNA cleavage between apoptotic and necrotic death exists, is that irreparable mitochondrial damage under- programs. Internucleosomal cleavage of DNA in lies cell death by hypoxia/reoxygenation. Since multiples of 180 bp occurs during necrosis soon after mammalian cells are obligately dependent on mito- the development of increased plasma membrane chondria for long term viability, it follows that Apoptosis in reoxygenation injury PSaikumaret al 3343 irreversible mitochondrial damage should have lethal release of apoptogenic factors from mitochondria consequences. This may happen in several ways. induced by pro-apoptotic members of the Bcl-2 family Electron conduction defects have been demonstrated of proteins. to occur in ischemic and hypoxic cells (Piper et al., 1994), and if the defects are not corrected during reoxygenation, a lethal outcome is certain since any Are oxygen radicals involved in apoptosis associated with capacity for anaerobic glycolysis that the aected cells reoxygenation? may have will eventually be compromised by mitochondrial dysfunction. Other factors, discussed Oxygen free radicals have been implicated in the below, are mitochondrially generated oxygen free pathogenesis of reperfusion injury (McCord, 1985). radicals, inner membrane permeability defects termed However, it is probable that the free radicals are mitochondrial permeability transitions (MPT) and the derived from in®ltrating leukocytes in aected tissues

Figure 1 Eect of protease inhibitors on internucleosomal DNA cleavage during Apoptosis (a) and necrosis (b). (a) After 20 min preincubation without or with caspase inhibitors z-VAD-fmk (VAD; 10 mM); z-DEVD-fmk (DEVD; 14 mM), or with serine protease inhibitors 3,4-dichloroisocoumarin (DCI; 100 mM), phenylmethane sulphonyl ¯uoride (PMSF; 1 mM), tosyl phenylalanyl chloromethyl ketone (TPCK; 100 mM) or with the cysteine protease inhibitor iodoacetic acid (IAA; 1 mM) 100 ng/ml Fas antibody was added to Jurkat cells for 5 h to induce apoptosis. Caspase activity was blocked in cells treated with -SH reactive agent IAA (not shown), but these cells developed early membrane damage, and typical DNA ladders neverthless. (b) Necrosis was induced in MDCK cells by incubating with 0.2 mg/ml saponin. Cell lysis was indicated as % of lactate dehydrogenase released (free LDH) and shown at the top of each lane. DNA fragments were analysed by gel electrophoresis. Identical patterns of DNA cleavage were seen in necrosis induced by ATP depletion (adopted from Dong et al., 1997) Apoptosis in reoxygenation injury PSaikumaret al 3344 rather than parenchymal cells per se (Cotran et al., peroxide (Figure 2b). However, with or without the 1998). To examine whether oxidant stress is involved in drugs, DCF ¯uorescence during reoxygenation was triggering apoptosis during reoxygenation, we em- less, not more, than in control cells that had not been ployed a model of reoxygenation injury in cultured subjected to hypoxia (Figure 2b). This is consistent rat kidney proximal tubule cells (RPTC; Saikumar et with the demonstration that reoxygenation of hypoxic al., 1998). RPTC were subjected to hypoxia for 5 h kidney proximal tubules results in decreased rather without glucose, and were then returned to either than increased oxidant stress (Zager, 1993). normoxic complete growth medium or oxygen-free complete growth medium that had been equilibrated

with 95% N2 and 5% CO2. Incubation of hypoxic cells Loss of mitochondrial cytochrome C during

in O2 -free medium led to 42% cell death after 2 h ischemia-hypoxia

(Figure 2a, 5H/2R: N2), which was not dierent from values observed in cells returned to normoxic medium Loss of cytochrome c may be one reason for respiratory (44%; Figure 2a, 5H/2R: N/A), excluding oxygen or dysfunction in post-ischemic-hypoxic mitochondria free radicals derived therefrom as death triggers. These (Piper et al., 1994; Borutaite et al., 1996; Naro et al., results agree with earlier reports that apoptosis can be 1993). In support, electron transport was restored by the induced without the involvement of oxygen free addition of exogenous cytochrome c to mitochondria radicals (Shimizu et al., 1995; Jacobson and Ra, isolated from ischemic hearts (Borutaite et al., 1996; 1995) and were further substantiated by the failure of Mur®tt et al., 1978; Toleikis et al., 1980). More recently, N-acetyl-cysteine (NAC; 1 ± 10 mM), a chemical anti- studies of hypoxia/reoxygenation injury in cultured oxidant (Sundaresan et al., 1995) and Mn(III) proximal tubule cells of rat kidney (Saikumar et al., tetrakis(4-benzoic acid) porphyrin chloride (MnTBAP; 1998) have documented evidence for mitochondrial 10 ± 100 mM), a cell permeant superoxide dismutase damage during hypoxia by a process that led to complete mimetic (Faulkner et al., 1994), to aect cell death loss by diusion of cytochrome c into the cytosol. during reoxygenation (Figure 2a; shown for 1 mM NAC, 100 mM MnTBAP). Production of oxygen free radicals was measured as the oxidation of dihydro- Mechanisms of mitochondrial cytochrome C release dichloro¯uorescein (H2-DCF) to generate the fluor- escent compound DCF . The relative DCF ¯uorescence Mitochondrial permeability transition (MPT) with NAC was less, indicating that reactive oxygen species were being scavenged, while MnTBAP in- Evidence for the occurrence of the MPT in cells injured creased the signal, consistent with accelerated dismuta- by ischemia or hypoxia has been obtained from studies tion of superoxide and production of hydrogen on organs subjected to ischemia/reperfusion, parench-

Figure 2 Oxidant stress is not the death trigger during reoxygenation. (a) Cell death was measured as percent fragmentation of total DNA. Rat kidney proximal tubule cells (RPTC) in culture, after labeling with 3H-thymidine, were subjected to 5 h of hypoxia followed by 2 h of incubation in full growth medium (5H/2R) under anaerobic conditions (N2) or aerobic conditions without any agents (N/A), or with 100 mM Carbobenzoxy-Val-Ala-Asp-¯uoromethyl ketone (VAD), or 1 mM N-acetylcysteine, an anti-oxidant, (NAC), or 100 mM Mn(III)tetrakis(4-benzoic acid) porphyrin chloride (Calbiochem-Novabiochem Co., CA, USA), a superoxide dismutase mimetic (MnTBAP). Control cells (CON) received no treatment. For anaerobic incubation, the culture medium was pre- equilibrated with 95% N2/5% CO2 for 48 h. To ensure lack of O2 , EC Oxyrase (Oxyrase Inc., OH, USA), an oxygen scavenging enzyme preparation, was present for the last 2 h of pre-equilibration. DNA fragmentation was analysed as described (Dong et al., 1997). (b) Reactive oxygen species within cells were measured by the oxidization of 2',7'-dichlorodihydro¯uorescein diacetate (H2- DCF) to generate the ¯uorescent product, dichloro¯uorescein (DCF; Sundaresan et al., 1995). Cells were loaded with H2-DCF (10 mM) for 30 min under normoxic conditions or during the last period of hypoxic incubation. Control cell: cells that received no treatment. Reoxygenated cell: cells made hypoxic for 5 h and reoxygenated. Both groups of cells were transferred to normoxic Krebs ± Ringer buer (KRB) containing 17.5 mM glucose in the absence (N/A) or presence of 1 mM N-acetylcysteine (NAC) or 100 mM MnTBAP (MnTBAP). DCF ¯uorescence (Ex 480 nm/Em 520 nm) during 60 min of KRB incubation was recorded. Similar results were obtained with NAC at 1 or 10 mM and with MnTBAP at 10 or 100 mM (not shown). Values are means+SE (n=4). Data shown here are our unpublished results Apoptosis in reoxygenation injury PSaikumaret al 3345 ymal cells isolated after hypoxia/reoxygenation, and demonstrated for the ®rst time that Bax translocation mitochondria isolated from ischemic tissues (Lemasters can take place under ATP depleted conditions, et al., 1998; Piper et al., 1994). Although the suggesting that the hypoxic model will provide new development and pathological signi®cance of the clues in identifying the mechanisms by which Bax is MPT has been rigorously documented by in vitro normally retained in the cytosol despite having studies, extrapolation of these results to pathogenetic mitochondria targeting signals in its carboxy terminus. events during ischemic injury in vivo, or even hypoxic Our studies have shown that under conditions of injury of whole isolated cells in vitro, has been fraught hypoxia, Bax induced cytochrome c release is unlikely with diculties. This has been largely due to the to be due to the induction of MPT. The mechanism by simultaneous occurrence of several destructive pro- which MPT are thought to induce cytochrome c release cesses during cell damage by ATP depletion and the involves loss of the normal inner membrane barrier to lack of interventions which have sucient power to small solutes leading to matrix swelling and rupture of discriminate between events that are causal, and those outer membranes. This should result in irreversible which evolve secondarily as eects. Current hypotheses damage to mitochondria. However, our results (Figure regarding the role of MPT in ischemic hypoxic damage 3) show that regeneration of ATP by glycolysis during center around ATP depletion related increases of reoxygenation is associated with recovery of mitochon- ionized calcium, which induce hyperpermeability of drial potential in the same cell populations that had the inner mitochondrial membrane to solutes in a non- leaked cytochrome c during hypoxia. By demonstrating selective manner. This `mitochondrial permeability the ability of mitochondrial inner membranes to retain transition' or MPT behaves like a membrane pore gradients of protons, these studies show that MPT, if which allows diusion of solutes 51 500 daltons in they had occurred at all during hypoxia, must have size. Its formation is thought to be mediated by been reversible. A more likely explanation for the improper interactions between proteins in a multi- release of cytochrome c from the mitochondrial inter- protein complex centered on the adenine nucleotide membrane space is the formation of pores in translocase, a mitochondrial cyclophilin, adenylate mitochondrial outer membranes, mediated by the kinase and porin, located in `focal contacts' between insertion of Bax, leaving the inner membranes intact. inner and outer mitochondrial membranes (Frade and Studies of Bax induced cell death in other models Michaelidis, 1997). Although it can occur as a support this hypothesis (Eskes et al., 1998; Jurgensme- temporary event, the MPT can rapidly become ier et al., 1998). irreversible, with resulting loss of mitochondrial homeostasis, and cause high amplitude mitochondrial Role of caspases in ischemia-hypoxia mediated injury swelling. Since the inner membrane has a larger surface area than the outer membrane, mitochondrial swelling It has been reported earlier that caspases are involved can cause rupture of the outer membrane, releasing in hypoxic or ischemic cell death (Shimizu et al., 1996; intermembrane proteins into the cytosol (Petit et al., Cheng et al., 1998; Yaoita et al., 1998; Kaushal et al., 1998; Reed et al., 1998). There have been several 1998). We have reported that DEVDase activity reports to the eect that cyclosporin A, a drug that (mediated by caspase-3, a distal eector caspase) is binds mitochondrial cyclophilin, prevents the MPT and increased dramatically during hypoxia or ATP ameliorates reoxygenation injury (Halestrap et al., depletion in a time dependent manner, concurrently 1997). with Bax translocation and cytochrome c release (Saikumar et al., 1998). However, Caspase-3 activation is not required for Bax translocation and cyt.c Role of Bax release, since neither process is prevented by caspase Members of the Bcl-2 family of proteins associate with inhibition (Saikumar et al., 1998). The immediate mitochondrial membranes and regulate mitochondrial proximate cause of caspase-3 activation during pathophysiology. Bcl-2, the prototype anti-apoptotic hypoxia or ATP depletion is unclear. Three mutually protein, inhibits MPT in some models of apoptosis and non-exclusive possibilities could account for the thus prevents the release of apoptogenic factors such as activation of caspase-3. First, released cyt.c, in cytochrome c and apoptosis inducing factor (AIF) conjunction with the cytosolic factors apaf-1 and from mitochondria into the cytosol (Susin et al., 1996; caspase-9, may stimulate the conversion of pro- Kluck et al., 1997). Overexpression of Bax, a pro- caspase-3 to active protease in presence of dATP/ apoptotic protein of the Bcl-2 family, results in loss of ATP (Thornberry and Lazebnik, 1998; Li et al., 1997; mitochondrial potential (Dc), induction of MPT and Zhou et al., 1997; Pan et al., 1998). Whether release of cytochrome c into the cytosol (Pastorino et sucient dATP is still available during ATP deple- al., 1998). However, it remained unclear whether MPT tion to trigger this mechanism is not known. Second, and loss of Dc were the cause or result of apoptotic it has been reported that a novel apoptosis inducing events, since these mitochondrial end-points might factor (AIF) is released from mitochondria by agents have been measured in cells that were moribund or that induce mitochondrial permeability transition post-mortem. Recently, we have shown that Bax is (MPT; Kroemer et al., 1997; Susin et al., 1996). AIF translocated from cytosol to mitochondria during ATP is reported to be released as a result of permeability depletion caused by hypoxia or inhibition of oxidative transition (PT) pore opening (Susin et al., 1996) and phosphorylation by an uncoupler, causing the release may directly activate caspase-3. However, the nature of cytochrome c from mitochondria into the cytosol of AIF is unknown, and its role in our model cannot (Saikumar et al., 1998). Although the mechanisms by be speculated upon. Third, increased DEVDase which Bax promotes cytochrome c release are still activity could be due to the release of mitochondria- unclear, these studies of cultured proximal tubule cells associated caspase-3 during hypoxia. Recently, a Apoptosis in reoxygenation injury PSaikumaret al 3346

Figure 3 Hypoxic cells that leaked cytochrome c can mount potential across inner mitochondrial membrane after reoxygenation. After 5 h of hypoxia, RPTC were loaded with rhodamine 123 for 5 min. under normoxic conditions and photographed during ¯uorescence microscopy (b). Immediately after the rhodamine 123 ¯uorescence was recorded, cells were ®xed and immunostained for cytochrome c using CY3-conjugated secondary antibodies (Saikumar et al., 1998). Fluorescence microscopy of immunostained cells (a) show that the majority of cells released cytochrome c into their cytosol, although almost all cells were able to mount mitochondrial potential (b). Identical Fields showing the same cell population in which mitochondria that leaked cytochrome c into cytosol were able to upload rhodamine 123 are shown in the insets. These results clearly indicate that the Bax translocation we observed during hypoxia (Saikumar et al., 1998) did not damage the mitochondrial inner membrane even after releasing cytochrome c. Data shown are our unpublished results

caspase-3 precursor was identi®ed to be present both ATP availability during reoxygenation determines death in mitochondria and the cytosol (Mancini et al., 1998). outcome Although the precursor was lost from the mitochondria following apoptotic stimulation, it was not Cells injured by ischemia/reperfusion can display determined whether the precursor was activated in necrotic as well as apoptotic morphology (Kato et mitochondria and then released into the cytosol, or al., 1997; Kajstura et al., 1998; Li et al., 1998a; Wiegele whether it was activated after release. Bcl-2 regulation et al., 1998; Lieberthal and Levine, 1996). Apoptosis is is common to all these possible pathways of caspase-3 an energy dependent process (Wyllie, 1994; Vaux and activation. While the mechanism of caspase-3 activa- Strasser, 1996). Impaired electron transport due to loss tion during hypoxia is likely to involve the formation of mitochondrial cyt.c would prevent recovery of of ternary complexes between cytochrome c, the cellular ATP pools if anaerobic glycolysis is not protein Apaf-1 and the initiator caspase-9 (Li et al., feasible, and thus lead to death by caspase-indepen- 1997; Zhou et al., 1997; Pan et al., 1998), the almost dent processes due to prolonged ATP depletion. On the certain participation of these transducing events in other hand, apoptosis will result if energy levels hypoxic injury remains to be shown. increase by virtue of glycolysis. Studies in cultured Although caspase-3 was activated during hypoxia, kidney cells showed that incubation of previously morphological alterations resembling apoptosis did not hypoxic cells in glucose free medium led to early occur suggesting that ATP is required to eect plasma membrane damage and cell death with necrotic apoptosis. The events that require ATP probably morphology (Saikumar et al., 1998). In sharp contrast, involve translocation of active caspase-3 to the cellular cells reoxygenated in presence of glucose showed targets such as nuclei and other cytoplasmic structures. characteristic apoptotic morphology. Cell death by Mitochondrial translocation of Bid, another death necrosis during reoxygenation in glucose free medium agonist of the Bcl-2 family, requires caspase-8 was not suppressed by the caspase inhibitor VAD activation (Luo et al., 1998; Li et al., 1998b). Thus, (100 mM) suggesting a caspase-independent process. Bid induced mitochondrial damage is contingent upon Importantly, overexpression of Bcl-2 completely an initiator caspase trigger situated proximally in blocked cell death regardless of whether the death death-receptor mediated apoptotic pathways. mechanisms were mediated by caspases or not Based on our results (Saikumar et al., 1998) and (Saikumar et al., 1998). It was also shown that Bcl-2 other information, we propose a model for Bax preserved mitochondrial integrity during hypoxia and induced release of cytochrome c from mitochondria promoted ATP generation even in the absence of (Figure 4). This model takes into account a mitochon- glucose, utilizing respiratory substrates during reox- drial membrane protein or protein complex to form ygenation. Based on the studies available to date, we protein-transporting pores by Bax. In this concept, Bcl- have proposed a general scheme for the evolution of 2 inhibits Bax pore formation by competing with a cell death in the hypoxia and reoxygenation model of protein or protein complex. injury (Figure 5). Apoptosis in reoxygenation injury PSaikumaret al 3347

Figure 4 Model for Bax induced mitochondrial outer membrane permeability changes. Based on our results (Saikumar et al., 1998), we propose that Bax, a mitochondrially targeted protein, is retained in the cytosol by yet to be identi®ed tethering chaperones. During hypoxia, Bax is translocated to mitochondria. This results in cytochrome c release into the cytosol. Released cytochrome c can activate cytosolic caspases through Apaf-1, a CED-4 homolog. This model also postulates an outer membrane protein or protein complex to be involved in Bax mediated pore formation. Overexpressed Bcl-2 may compete for the associated protein or protein complex to inhibit Bax pore formation. This model also includes the possibility of an activated caspase molecule being released from Bax translocated mitochondria

Figure 5 Evolution of cell death in the hypoxia and reoxygenation model of injury. Mitochondria release apoptogenic factors both by Bax mediated and Bax-independent pathways during ischemia-hypoxia. Bax independent mechanisms may involve membrane permeability transition (MPT) pore formation, which leads to mitochondrial swelling and rupture of outer membrane. Bax- mediated pathways only aect outer membrane and release cytochrome c. If glycolysis prevails during reoxygenation of hypoxia damaged cells, glycolytic ATP drives cells to apoptotic form of death. In the absence of glycolytic ATP and lack of ATP production by defective mitochondria, cells devoid of ATP undergo necrosis Apoptosis in reoxygenation injury PSaikumaret al 3348 Conclusions cyclophillin binding drug, cyclosporine A. However, using the hypoxia/reoxygenation model in cultured It is increasingly clear that mitochondria constitute an renal cells, we have shown that mitochondrial inner important component of the cell death machinery. membrane retained its integrity as indicated by its Since mammalian cells are obligately dependent on ability to restrict proton permeability and maintain mitochondrial function for long-term viability, irrever- potential gradients during reoxygenation. On the other sible mitochondrial damage during ischemia or hypoxia hand, depletion of ATP, associated with hypoxia, is certain to assure a lethal outcome. Much progress resulted in translocation of Bax from cytosol to has been made in understanding how the ATP mitochondria, and concurrent release of cytochrome c depletion which characterizes hypoxic insults leads to and activation of caspase-3. While the transducing mitochondrial damage and how the damaged orga- mechanism responsible for caspase-3 activation re- nelles may transduce downstream death pathways. mains to be identi®ed, the results strongly suggest Relevant abnormalities include mitochondrial electron that Bax translocation is a determining pathogenetic transport defects, generation of oxygen free radicals, event, and that Bax permeabilizes the mitochondrial mitochondrial permeability transitions (MPT) and outer membranes. It seems likely that translocated Bax release of apoptogenic factors induced by death does not require MPT in order to eect cytochrome c proteins of the Bcl-2 family. That irreversible blocks release. Other results using the same injury model of electron transport have lethal signi®cance is showed that Bcl-2 does not antagonize Bax transloca- intuitive, and such defects have been shown to occur tion during hypoxia, but prevents mitochondrial release in mitochondria isolated from ischemic organs. How of cytochrome c and aborts all downstream death these defects may lead to further steps in structured events. Thus, Bcl-2 not only inhibits apoptosis, but, by death pathways is not known. Increased generation of maintaining the integrity of the electron transport oxygen free radicals by damaged mitochondria has chain and promoting oxidative phosphorylation, also been considered to be an important pathogenetic factor can prevent necrosis that may otherwise aect cells in reoxygenation injury. However, we have been unable with glycolytic disability. An important clinical to detect such increases during reoxygenation following consideration that follows from these results is that hypoxia of cultured renal cells, and could show caspase inhibitors cannot oer long-term protection to unabated activation of apoptotic cascades even under ischemically injured parenchymal cells regardless of oxygen free conditions, provided that recovering cells how eectively they can inhibit apoptotic events, if the could synthesize ATP by anaerobic glycolysis. Devel- cells have also suered mitochondrial damage sucient opment of irreversible MPT, a structurally disruptive to permanently impair respiration. Because of emerging event, is in itself a plausible predictor of cell death recognition of the bewildering diversity of cellular during reoxygenation injury, without need to invoke death paradigms, the relative importance of mitochon- release of apoptogenic factors as its direct consequence. drial electron transport defects, MPT, cytochrome c There is precedent for its occurrence as a terminal leakage and other non-mitochondrial factors during event during reoxygenation injury, and for its ischemic injury in vivo needs to be determined prevention by pharmacological intervention using the individually for dierent cell types and organs.

References

Ashkenazi A and Dixit VM. (1998). Science, 281, 1305 ± Faulkner KM, Liochev SI and Fridovich I. (1994). J. Biol. 1308. Chem., 269, 23471 ± 23476. Bonfoco E, Krainc D, Ankarcrona M, Nicotera P and Flitter WD. (1993). Br. Med. Bull., 49, 545 ± 555. Lipton SA. (1995). Proc. Natl. Acad. Sci. USA, 92, 7162 ± Frade JM and Michaelidis TM. (1997). Bioessays, 19, 827 ± 7166. 832. Borutaite V, Morkuniene R, Budriunaite A, Krasauskaite D, Gonzalez-Flecha B and Boveris A. (1995). Biochim. Biophys. Ryselis S, Toleikis A and Brown GC. (1996). J. Mol. Cell. Acta, 1243, 361 ± 366. Cardiol., 28, 2195 ± 2201. Granger DN. (1988). Am. J. Physiol., 255, H1269 ± H1275. Chan PH. (1994). Brain Pathol., 4, 59 ± 65. Hara H, Friedlander RM, Gagliardini V, Ayata C, Fink K, Cheng Y, Deshmukh M, D'Costa A, Demaro JA, Gidday Huang Z, Shimizu-Sasamata M, Yuan J and Moskowitz JM, Shah A, Sun Y, Jacquin MF, Johnson EM and MA. (1997). Proc. Natl. Acad. Sci. USA, 94, 2007 ± 2012. Holtzman DM. (1998). J. Clin. Invest., 101, 1992 ± 1999. Halestrap AP, Connern CP, Griths EJ and Kerr PM. Chopp M and Li Y. (1996). Acta Neurochir., 66, 21 ± 26. (1997). Mol. Cell. Biochem., 174, 167 ± 172. Cotran RS, Kumar V and Collins T. (1998). Cell injury and Jacobson MD and Ra MC. (1995). Nature, 374, 814 ± 816. cellular death. In: Robbins Pathologic Basis of Disease (6th Johnson KJ and Weinberg JM. (1993). Curr. Opin. Nephrol. edition) WB Saunders Company: Philadelphia. Hypertens., 2, 625 ± 635. Dong Z, Saikumar P, Weinberg JM and Venkatachalam Jurgensmeier JM, Xie Z, Deveraux Q, Ellerby L, Bredesen D MA. (1997). Am. J. Pathol., 151, 1205 ± 1213. and Reed JC. (1998). Proc. Natl. Acad.Sci. USA, 95, Dong Z, Saikumar P, Griess GA, Weinberg JM and 4997 ± 5002. Venkatachalam MA. (1998). Am.J.Pathol.,152, 231 ± KajsturaJ,LiuY,BaldiniA.,LiB,OlivettiG,LeriAand 240. Anversa P. (1998). Am.J.Cardiol.,82, 30K ± 41K. Eguchi Y, Shimizu S and Tsujimoto Y. (1997). Cancer Res., Kato H, Kanellopoulos GK, Matsuo S, Wu YJ, Jacquin MF, 57, 1835 ± 1840. Hsu CY, Kouchoukos NT and Choi DW. (1997). Exp. Engler RL. (1989). Am. J. Cardiol., 63, 19E ± 23E. Neurol., 148, 464 ± 474. Eskes R, Antonsson B, Osen-Sand A, Montessuit S, Richter Kaushal GP, Singh AB and Shah SV. (1998). Am.J.Physiol., C, Sadoul R, Mazzei G, Nichols A and Martinou JC. 274, F587 ± F595. (1998). J. Cell. Biol., 143, 217 ± 224. Apoptosis in reoxygenation injury PSaikumaret al 3349 Kluck RM, Bossy-Wetzel E, Green DR and. Newmeyer DD. Pastorino JG, Chen ST, Tafani M, Snyder JW and Farber (1997). Science, 275, 1132 ± 1136. JL. (1998). J. Biol. Chem., 273, 7770 ± 7775. Leist M, Single B, Castoldi AF, Kuhnle S and Nicotera P. Petit PX, Goubern M, Diolez P, Susin SA, Zamzami N and (1997). J. Exp. Med., 185, 1481 ± 1486. Kroemer G. (1998). FEBS Lett., 426, 111 ± 116. Leist M. and Nicotera P. (1997). Biochem. Biophys. Res. Piper HM, Noll T and Siegmund B. (1994). Cardiovasc. Res., Commun., 236, 1±9. 28, 1 ± 15. Lemasters JJ, Nieminen AL, Qian T, Trost LC, Elmore SP, Piper HM, Garcia-Dorado D and Ovize M. (1998). Nishimura Y, Crowe RA, Cascio WE, Bradham CA, Cardiovasc. Res., 38, 291 ± 300. Brenner DA and Herman B. (1998). Biochim. Biophys. Reed JC, Jurgensmeier JM and Matsuyama S. (1998). Acta, 1366, 177 ± 196. Biochim. Biophys. Acta, 1366, 127 ± 137. Lemaire C, Andreau K, Souvannavong V and Adam A. SaikumarP,DongZ,PatelY,HallK,HopferU,Weinberg (1998). FEBS Lett., 425, 266 ± 270. JM and Venkatachalam MA. (1998). Oncogene, in press. Li P, Nijhawan D, Budihardjo I, Srinivasula SM, Ahmad M, Schielke GP, Yang GY, Shivers BD and Betz AL. (1998). J. Alnemri ES and Wang X. (1997). Cell, 91, 479 ± 489. Cereb. Blood Flow Metab., 18, 180 ± 185. Li Y, Powers C, Jiang N and Chopp M. (1998a). J. Neurol. ShimizuS,EguchiY,KosakaH,KamiikeW,MatsudaH Sci., 156, 119 ± 132. and Tsujimoto Y. (1995). Nature, 374, 811 ± 813. Li H, Zhu H, Xu CJ and Yuan J. (1998b). Cell, 94, 491 ± 501. Shimizu S, Eguchi Y, Kamiike W, Matsuda H and Tsujimoto Lieberthal W and Levine JS. (1996). Am. J. Physiol., 271, Y. (1996). Oncogene, 12, 2251 ± 2257. F477 ± F488. Sundaresan M, Yu ZX, Ferrans VJ, Irani K and Finkel T. Luo X, Budihardjo I, Zou H, Slaughter C and Wang X. (1995). Science, 270, 296 ± 299. (1998). Cell, 94, 481 ± 490. SusinSA,ZamzamiN,CastedoM,HirschT,MarchettiP, MacLellan WR and Schneider MD. (1997). Circ. Res., 81, Macho A, Daugas E, Geuskens M and Kroemer G. (1996). 137 ± 144. J. Exp. Med., 184, 1331 ± 1341. Mancini M, Nicholson DW, Roy S, Thornberry NA, Thiagarajan RR, Winn RK and Harlan JM. (1997). Peterson EP, Casciola-Rosen LA and Rosen A (1998). J. Thrombosis & Haemostasis, 78, 310 ± 314. Cell Biol., 140, 1485 ± 1495. Thornberry NA and Lazebnik Y. (1998). Science, 281, Matsuyama T, Hata R, Yamamoto Y, Tagaya M, Akita, 1312 ± 1316. Uno H, Wanaka A, Furuyama J and Sugita M. (1995). Toleikis A, Dzeja P and Praskevicius A. (1980). Adv. Brain Res. Mol. Brain Res., 34, 166 ± 172. Myocardiol., 2, 327 ± 337. McCarthy NJ, Whyte MK, Gilbert CS and Evan GI. (1997). Vaux DL and Strasser A. (1996). Proc. Natl. Acad. Sci. USA, J. Cell Biol., 136, 215 ± 227. 93, 2239 ± 2244. McCarthy NJ and Evan GI. (1998). Curr.Top.Dev.Biol.,36, Vogt M, Bauer MK, Ferrari D and Schulze-Ostho K. 259 ± 278. (1998). FEBS Lett., 429, 67 ± 72. McCord JM. (1985). N. Engl. J. Med., 12,159 ± 163. Wiegele G, Brandis M and Zimmerhackl LB. (1998). Miller TM, Moulder KL, Knudson CM, Creedon DJ, Nephrol. Dial. Transplant., 13, 1158 ± 1167. Deshmukh M, Korsmeyer SJ and Johnson EM. (1997). Wyllie AH. (1994). Phil.Trans.R.Soc.Lond.B,345, 237 ± J. Cell. Biol., 139, 205 ± 217. 241. Mur®tt RR, Stiles JW, Powell JR WJ and Sanadi DR. (1978). Xiang J, Chao DT and Korsmeyer SJ. (1996). Proc. Natl. J. Mol. Cell. Cardiol., 10, 109 ± 123. Acad.Sci.USA,93, 14559 ± 14563. Naro F, Fazzini A, Grappone C, Citro G, Dini G, Giotti A, Yang J, Liu X, Bhalla K, Kim CN, Ibrado AM, Cai J, Peng Malatesta F, Franconi F and Brunori M. (1993). TI, Jones DP and Wang X. (1997). Science, 275, 1129 ± Cardioscience, 4, 177 ± 184. 1132. NodaT,IwakiriR,FujimotoK,MatsuoSandAwTY. Yaoita H, Ogawa K, Maehara K and Maruyama Y. (1998). (1998). Am. J. Physiol., 274, G270 ± G276. Circulation, 97, 276 ± 281. Nogae S, Miyazaki M, Kobayashi N, Saito T, Abe K, Saito Zager RA. (1993). Circ. Res., 73, 926 ± 934. H, Nakane PK, Nakanishi Y, and Koji T. (1998). J. Am. Zou H, Henzel WJ, Liu X, Lutschg A and Wang X. (1997). Soc. Nephrol., 9, 620 ± 631. Cell, 90, 405 ± 413. Pan G, O'Rourke K and Dixit VM. (1998). J. Biol. Chem., 273, 5841 ± 5845.