A unique pathway of cardiac myocyte death caused by hypoxia–acidosis

High blood pressure and elevated cholesterol are risk factors for coronary artery disease, promoting the formation of fatty streaks and lesions in the coronary vasculature that are the precursors of plaque. Progressively expanding atherosclerotic plaque ultimately restricts blood flow through coronary arteries and provides a substrate for occlusive thrombus formation. Reduced blood flow produces hypoxia in the tissues downstream of the lesion. In the most extreme cases of coronary artery disease, a complete occlusion occurs, eliminating blood flow and presenting an acute threat to the viability of the effected tissue (acute coronary event or heart attack). When the myocardial tissue is reperfused by removal of the plaque (thrombolysis),downstream cells are subjected to further damage involving oxidative stress,and a region of permanent injury containing dead and dying cells, known as an infarct, develops (Jennings et al.,1995). Blood supply is interrupted inside the developing infarct and in the bordering tissue, and hypoxia may persist in these regions for days or weeks, exacerbating the injury (Narula et al., 1998; Vanoverschelde et al., 1997; Zuurbier et al.,1999). Hypoxia causes a switch from oxidative to glycolytic energy generation, with increased glucose consumption, lactic acid production and lower intracellular pH (Dennis et al.,1991; Neely and Grotyohann,1984; Webster et al.,2000). Increased blood lactate levels reflect this metabolic shift and are diagnostic of infarction in ischemic heart disease. The extent of tissue loss to infarction is determined by the severity and duration of the ischemic period and is known to involve both necrotic and apoptotic cell death pathways (Anversa and Kajstura,1998; Gottlieb et al.,1994; Kajstura et al.,1996). Oxidative stress caused by reperfusion may account for up to 60% of the tissue damage during early infarction(Fliss and Gattinger, 1996; Horwitz et al., 1994; Yoshida et al., 1996). Multiple additional factors contribute to cell death as the infarcted area expands and more border cells die. These include collateral damage from necrosis and infiltrating macrophages(Williams et al., 1994),additional necrotic and apoptotic death resulting from energy (glucose and ATP) depletion (Buja et al.,1993; Hochachka et al.,1996; Majno and Joris,1995; Reimer and Ideker,1987; Mani and Kitsis,2003), and changes associated with hypoxia. The pathways of ischemic injury are the subjects of intensive study because descriptions of these pathways may lead to new methods of reducing tissue loss and minimizing the damage caused by infarction and heart attack.

Numerous reports have documented the extensive cell death and tissue loss that accompanies myocardial ischemia(Anversa and Kajstura, 1998; Narula et al., 1996; Chen et al., 1997; Fiss and Gattinger, 1996). The contribution of hypoxia to cell death is controversial and different groups have reported markedly different results(Tanaka et al., 1994; Long et al., 1997; Matsui et al., 1999; Kang et al., 2000; Regula et al., 2002; Kubasiak et al., 2002). The discrepancies are almost certainly due to differences in the models,especially related to glucose levels and pH. Our results indicate that it is the secondary consequences of hypoxia rather than hypoxia per se that causes cell death (Webster and Bishopric,1992; Webster et al.,1999). Although hypoxia is an obligatory consequence of ischemia,it rarely exists alone. Hibernating myocardium may represent a condition of chronic hypoxia (Heusch and Schulz,1996) but in most other conditions of moderate to severe ischemia,hypoxia is accompanied by or precedes energy depletion, acidosis and/or reoxygenation. Each of these latter conditions involves cell and tissue loss,and the pathways of cell death in each case are unique. Reperfusion damage has been the subject of numerous articles and reviews (Gottlieb et al., 2003; Kumar and Jugdutt, 2003; Valen, 2003) and will not be addressed in detail here. The damage caused by reperfusion is widely believed to be initiated by surges of oxygen free radicals that initiate stress responses and culminate in cell death. Our studies showed that the exposure of cardiac myocytes to cycles of hypoxia–reoxygenation results in ∼30%cell death in each cycle (Webster et al.,1999; Dougherty et al.,2002). The response of cardiac myocytes to ischemia–reperfusion (I/R) is complex, and the survival/death pathways have not been fully described. During early ischemia, ATP levels may be maintained by increased glycolysis but at the expense of the limited reserves of glucose and glycogen. Glucose utilization increases by >10-fold in ischemic/hypoxic cardiac myocytes with corresponding lactate accumulation(Webster et al., 1993, 1994, 1999). Acidosis is further exacerbated when ATP levels begin to decline(Allen et al., 1989; Allen and Orchard, 1987; Neely and Grotyohann, 1984). If ATP is depleted during ischemia, necrosis will occur because of the passive loss of transmembrane ion gradients, followed by cell swelling and loss of membrane integrity (Buja et al.,1993; Hochachka et al.,1996; Majno and Joris,1995; Reimer and Ideker,1987). In the latter condition, apoptosis as well as necrosis probably occurs, and both may be `programmed' at least in the early stages(Bishopric et al., 2001). Programmed death is an active, energy-consuming process requiring ATP; as ATP levels fall at late time points, the programmed pathways may fail and be replaced by a more classical necrotic death(Kajstura et al., 1996; Ohno et al., 1998; Buja and Entman, 1998).

It has been recognized for some time that proton pumps and pH regulation may play a role in apoptosis signaling(Anversa and Kajstura, 1998; Gottlieb et al., 1996; Karwatowska-Prokopczuk et al.,1998; Long et al.,1998). Ischemic cardiac myocytes generate excess H⁺through increased anaerobic metabolism, net hydrolysis of ATP and CO₂ retention (Dennis et al.,1991). These protons are extruded from the myoplasm to the interstitial space by the combined action of three major ion-specific membrane transporters, including the Na⁺/H⁺ exchanger, the Na⁺/HCO₃^– cotransporter and the vacuolar proton-ATPase(Karwatowska-Prokopczuk et al.,1998; Lagadic-Gossmann et al.,1992; Lazdunski et al.,1985). Increased activity of the Na⁺/H⁺exchanger can cause Ca²⁺ overload because the elevated intracellular Na⁺ is subsequently exchanged for Ca²⁺via the Na⁺/Ca²⁺ exchanger(Pierce and Czubryt, 1995). Inhibition of Na⁺/H⁺ exchange has been shown to protect against ischemic injury, possibly by preventing this increase in Ca²⁺ (Bond et al.,1993; Shimada et al.,1996). Conversely, inhibition of the vacuolar ATPase promotes apoptosis, in part by shifting the proton load towards the Na⁺/H⁺ transporter, thereby increasing Ca²⁺uptake, and in part by reducing the myocyte capacity to control intracellular pH (Gottlieb et al., 1996; Karwatowska-Prokopczuk et al.,1998; Long et al.,1998). Acidosis has been shown to correlate with apoptosis in a number of other systems where it may promote the activation of caspases(Gottlieb et al., 1995; Li and Eastman, 1995; Perez-Sala et al., 1998).

The Bcl-2 gene family encodes a group of proteins with the ability to promote or repress programmed cell death in response to a wide variety of stimuli (Boise et al., 1995; Hockenbery, 1995; Korsmeyer, 1995). Additional genes in this family include those encoding Bcl-Xl, Mcl-1, A1, Bcl-W and CED-9, which are anti-apoptotic, and Bak, Bax, Bcl-xs, Diva and Mtd/Bok, which are pro-apoptotic. These proteins are usually associated with the cell membranes, particularly the mitochondria, endoplasmic reticulum (ER) and nuclear envelope, where they are anchored by a C-terminal domain. Individual family members may remain in the cytosol or be loosely membrane bound and translocate after a death signal is received(Adams and Cory, 2001; Regula et al., 2002; Vande Velde et al., 2000). The physical location and activity of each Bcl-2 family protein is determined partly by its binding to other Bcl-2-related proteins in the cell cytosol. This in turn is determined by the relative concentrations of each protein, and the balance of pro- and anti-apoptotic members is an important feature of the regulation. Bcl-2 has been attributed antioxidant and proton translocating properties (Hockenbery et al.,1993; Shimizu et al.,1998), and one major function of the family as a whole is to determine the on/off state of the mitochondrial permeability transition pore(MPTP; reviewed in Green and Reed,1998; Earshaw et al.,1999; Crompton,2000). In some animal ischemia models of coronary occlusion,myocardial levels of Bcl-2 proteins have been shown to decline while pro-apoptotic Bax increases (Kajstura et al., 1996; Hockenbery,1995). Over-expression of Bcl-2 reduces apoptosis in some models of neuronal ischemia (Kane et al.,1993; Lawrence et al.,1996).

BNIP3 is a member of the so-called BH3-only subfamily of Bcl-2 family proteins that antagonize the activity of pro-survival proteins and promote apoptosis (Ray et al., 2000; Vande Velde et al., 2000). These proteins do not possess the same protein binding domains (BH1 and BH2)as the other Bcl-2 family members but instead bind through a common BH3 domain. Related members of this group include Bik, Blk, Hrk, BimL, Bad, Bid and Nix. BNIP3 was originally identified as a Bcl-XL or E1B 19K-binding protein (Zhang et al., 2003). The BH3 domain of BNIP3 may not be required for the death-promoting functions,but the C-terminal transmembrane domain is required, indicating membrane targeting as essential for pro-apoptosis function. BNIP3 is expressed below detectable levels in most organs including the heart under normal(non-ischemic) conditions (Bruick,2000; Vande Velde et al.,2000). Overexpression of BNIP3 protein by transfection of the cDNA into some cultured cell lines results in membrane translocation and initiation of cell death (Vande Velde et al.,2000). The death pathway has characteristics of both apoptosis and necrosis, including DNA fragmentation and MPTP opening, but early loss of plasma membrane integrity and no caspase activation. Expression of the BNIP3-encoding gene is induced by hypoxia through a 5′ promoter hypoxia-inducible factor-1 (HIF-1)-binding site(Bruick, 2000).

In the next sections, we will review the evidence, including new data, for a unique pathway of death caused by hypoxia with acidosis. These data are a compilation of results reported by our groups over the past decade and include, in particular, the following references: Webster and Bishopric, 1992;Webster et al., 1999, 2000; Dougherty et al., 2002, 2004; Kubasiak et al., 2002.

As discussed above, the role of hypoxia in ischemia-mediated cell death is controversial. We previously reported that rapidly contracting cardiac myocytes remained fully viable and contractile during culture under severe hypoxia for up to 5 days (Webster and Bishopric, 1992). Glycolysis was induced by 10-fold within 1 h,and there was no evidence of major cell loss; ATP remained high and the only significant change was a reduced level of contractility that correlated with lower intracellular cAMP in the hypoxic cultures. These results contrast with other reports of significant cardiac myocyte cell loss by apoptosis during 24–72 h of exposure to an equivalent degree of hypoxia(Tanaka et al., 1994; Long et al., 1997; Matsui et al., 1999; Regula et al., 2002). To resolve this apparent controversy, we subjected cardiac myocytes to severe hypoxia for a week under conditions where the glucose and extracellular pH([pH]o) were constantly monitored and maintained within the physiological range, and we measured both caspase activity and cell death at intervals. As indicated in Fig. 1A, there was no significant change in the number of apoptotic nuclei at either 3 days or 7 days compared with aerobic cultures. Under the same conditions of hypoxia,endothelial cells died within 3 days (data not shown). Caspase activities are shown in Fig. 1B; there were no significant changes in the activities of caspases 3, 8 or 9 at any time during this time course (staurosporine-treated plates incubated in parallel generated a maximal 2.7-fold activation of caspase 3 and 1.8-fold activation of caspase 9; not shown). These results confirm our previous reports that hypoxia alone does not activate programmed cell death in neonatal cardiac myocytes. This conclusion cannot necessarily be extrapolated to other models or to the intact heart. To remain viable under hypoxia, the cells must be able to maintain glycolysis at a level that is sufficient to sustain ATP, a condition that requires a continuous supply of glucose. In addition, the cell must be able to clear excess acid produced by anaerobic glycolysis. If these conditions can not be fully accommodated, the cell will die. Todor et al.(2002) recently reported that cardiac myocytes from failing hearts, but not those from normal hearts, were susceptible to hypoxia-mediated death. It is possible that the failing myocytes were compromised in their ability to activate adaptations to hypoxia. It should also be noted that oxygen tension may be an important determinant in the cellular response including the mechanism of oxygen sensing. Budinger et al. (1998) described a model of myocardial ischemia where myocytes were subjected to 3% O₂. There was a small decrease in contractility under hypoxia coincident with a similar small loss of ATP but no loss of cell viability. The authors concluded that oxygen sensing, and the positive adaptation, was probably initiated by reactive oxygen species (ROS) generated in the mitochondria under these conditions because of partial inhibition of cytochrome oxidase. Under some conditions, mitochondrial ROS may play a driving role in cell death during hypoxia. Yermolaieva et al.(2004) described a stroke model where death of PC12 cells subjected to a brief period of hypoxia followed by reoxygenation correlated with ROS. In our studies, oxygen tension is maintained at less than 0.5%, mitochondrial functions are severely impaired but we did not detect any increase of ROS as measured by the lucigenin reagent(N-methyl-acridinium nitrate; Dougherty et al., 2004).

To investigate the impact of pH on hypoxic myocytes, we exposed parallel cultures of myocytes to hypoxia at neutral pH or hypoxia with coincident acidosis and measured the levels of apoptosis by DNA ladders and Hoechst staining (for experimental details, see Webster et al., 1999). For hypoxia–neutral exposure, the cultures were maintained continuously under hypoxia, and the medium was replaced twice daily with fresh hypoxic medium to prevent the build-up of waste metabolites. For hypoxia–acidosis, the cultures were maintained in parallel but the medium was not replaced. The cultures were monitored for ATP, glucose and[pH]o. Results from these experiments are shown in Fig. 2A,B. In agreement with the results shown in Fig. 1,there was no evidence of DNA fragmentation under the first set of conditions(Fig. 2A), but cultures subjected to hypoxia without daily media change showed significant DNA laddering after 48 h and extensive laddering after 72 h(Fig. 2B). In cultures with medium replacement, ATP levels dropped to ∼75% of aerobic control levels after 24 h and remained stable thereafter; [pH]o did not change, and glucose levels remained high (Fig. 2C). Under these conditions, the myocytes continued to contract for the duration of the experiment, as previously reported(Webster and Bishopric, 1992). When the medium was not replaced, intracellular ATP levels were sustained for up to 48 h, and the dramatic loss at 72 h was coincident with cell death. Glucose declined but was not depleted, and [pH]o declined steadily to a final value of ∼6.0 (Fig. 2C). Under these conditions, contractions ceased before 48 h (data not shown).

In Fig. 2D–F, the effects of coincident hypoxia and acidosis on myocyte structure and integrity are shown by staining with Hoechst 33342 and anti-α-MHC antibody (see Dougherty et al., 2002; Webster et al., 1999). In cultures grown aerobically, abundant myofilaments with clear cross-striations are evident with the α-MHC stain(Fig. 2D). The white arrows indicate smooth, oval-shaped nuclei with sparse evidence of condensation or internal fragmentation. In this field, 29 nuclei were scored normal and one was condensed. At 72 h, thehypoxic–acidotic cultures(Fig. 2E) still stained strongly with α-MHC antibody but there was clear deterioration of the myofilaments, and cross-striations were no longer clearly visible. In the field shown, 22 nuclei were scored condensed (examples are indicated by the arrows), and 14 were normal. By contrast, hypoxic–neutral cells still exhibited myofilaments with intact cross-striations after 72 h(Fig. 2F, see arrow at far right), and most of the nuclei were normal. Control aerobic cultures contained 5–7% apoptotic cells; this increased to 44% after 48 h of hypoxia with metabolite build-up and to 60% after 72 h of hypoxia.

These results demonstrate that cardiac myocyte death under hypoxia occurs only if the media is not frequently changed, indicating either that proapoptotic factors accumulate in the media during hypoxia or vital components are depleted. To test these possibilities, medium from 48 h hypoxic cultures that were just beginning to show signs of DNA laddering was transferred to fresh cardiac myocytes and the cells were incubated for an additional 24–48 hunder either hypoxic or aerobic conditions. Apoptosis was again monitored by DNA fragmentation. Control plates received medium from hypoxic cells that underwent daily medium replacement. Results are shown in Fig. 3. Significant apoptosis was apparent in both aerobic and hypoxic cultures 24 h after exposure to the spent medium, but significantly more DNA fragmentation appeared in the sample from the 24 h hypoxic plate correlating with the lower [pH]o. These results implicate a death factor(s) associated with deteriorating pH control.

As a second approach to identify the death factors, we collected the spent medium from 48 h hypoxic cultures and readjusted the pH to 7.6 before transferring to fresh cardiac myocytes. The effect of neutralized spent medium is shown in Fig. 4. Acidic spent medium again caused extensive apoptosis of fresh myocytes but the neutralized medium caused minimal DNA fragmentation(Fig. 4A). This suggests that acidosis is necessary to induce apoptosis of hypoxic cultures. To confirm this, parallel cardiac myocyte cultures were again exposed to hypoxia; in one set, acid accumulated in the medium and in the other the acid was neutralized every 12 h to maintain [pH]o above 7.0 during hypoxic incubation. These results are shown in Fig. 4B. In the absence of additional buffer, [pH]o dropped to 5.7 at 48 h, and there was extensive apoptosis. In the pH-neutral cultures, there was no visible DNA fragmentation. Hoechst 33342 and anti-α-MHC antibody stains again revealed >60% apoptosis of acidotic cardiac myocytes and <10% when the pH was neutralized.

Results described above indicate an essential contribution of acidosis to promote death under hypoxia when glucose and ATP levels are sustained. In further experiments, we sought to investigate the specific components of this death pathway. As a first step, we used a microarray screen to determine which gene groups responded to hypoxia. Cardiac myocytes were subjected to normoxic incubation or 24 h hypoxia, and mRNA was reverse-transcribed and hybridized to an array of 20 000 specific gene sequences from rat and mouse libraries (these experiments were carried out in collaboration with Incyte, Inc., Freemont, CA,USA). Results from one such set of screens, subjected to analysis using an Excel 98 program, are shown in Fig. 5A. A gene transcript identified as the Bcl-2 family member BNIP3 appeared as the second most strongly induced gene. Hypoxia-inducible marker genes, including metallothionein-1 (MT-1), heme oxygenase (HO), pyruvate dehydrogenase kinase (PDK), glucose transporter(GLUT), tyrosine amino transferase (TAT) and triosephosphate isomerase (TPI),are also indicated. The microarray data were confirmed by northern and western blot analyses of BNIP3 (Fig. 5B,C; for details, see Kubasiak et al., 2002). BNIP3 mRNA levels increased progressively during hypoxia and peaked after 8 h at similar levels in both neutralized and acidic conditions. BNIP3 mRNA was degraded after 48 h of hypoxia in acidic but not in neutral pH media in parallel with cell death. BNIP3 protein accumulated more rapidly under acid pH and peaked at a significantly higher level than the pH-neutral samples. There were no corresponding changes in Bax, Bak or β-actin proteins. These results demonstrate that hypoxia activates BNIP3 transcription, and the protein may be stabilized by acidosis. Fig. 5D shows that BNIP3 transcripts also increased in a Langendorf-perfused intact rat heart preparation subjected to 1 h of no-flow ischemia or 2 h of perfusion with hypoxic medium. Procedures for these experiments are described in detail in Webster et al. (1999) and Kubasiak et al. (2002).

These results demonstrate a correlation between BNIP3 accumulation and cardiac myocyte death during hypoxia–acidosis. Further experiments were able to confirm a functional relationship using antisense suppression of the BNIP3 gene transcripts. As expected, treatment of myocytes with BNIP3 antisense oligonucleotides almost completely prevented hypoxia–acidosis cell death (not shown; see Kubasiak et al., 2002). However, these results do not explain why cell death requires acidosis because BNIP3 accumulated during hypoxia at both neutral and acidic pH. Therefore, acidosis may be required to activate BNIP3. Because Bcl-2 proteins can regulate the MPTP by integrating into the outer mitochondrial membrane, we hypothesized that activation may involve a pH-dependent translocation from the cytosolic compartment to membrane compartments, particularly mitochondrial membranes. To test this possibility,cells were separated into subcellular fractions after exposure to hypoxia and hypoxia–acidosis. Membranes were prepared and treated with strong alkali to dislodge loosely membrane-associated protein(Kubasiak et al., 2002). Results from western blots of untreated and alkali-solubilized fractions are shown in Fig. 6. BNIP3 levels were initially detected in the alkalinized cytoplasmic fraction at 12 h of hypoxia and increased progressively at 24 h and 48 h. BNIP3 was present exclusively in the mitochondrial fraction of hypoxic samples without alkali treatment and was primarily mitochondrial in hypoxia–acidosis samples. Alkaline treatment caused a significant shift of BNIP3 into the cytoplasmic fraction from hypoxic–neutral but not hypoxic–acidic treatments. In 48 h hypoxia–neutral samples, 73% of BNIP3 was in the cytoplasm compared with <10% of the hypoxia–acidosis sample. These results show that acidosis promotes a stronger alkali-resistant association of BNIP3 with mitochondrial membranes and support our hypothesis that BNIP3 activation involves the pH-mediated membrane integration.

Overexpression of BNIP3 by transient transfection of cell lines has been reported to activate a necrosis-like pathway that includes early loss of plasma membrane integrity (Ray et al.,2000; Vande Velde et al.,2000). We did not find this to be the case for hypoxia–acidosis (Kubasiak et al.,2002). Progressive hypoxia–acidosis with >60% cell death caused <20% loss of membrane integrity even at the late time points as determined by Trypan blue exclusion or lactate dehydrogenase (LDH) release(Fig. 7A). These results favor a more apoptosis-like pathway. However, as shown in Fig. 7B, the death pathway was unaffected by the broad-range caspase inhibitor Boc-D (or ZVAD; not shown) at a concentration that completely blocked apoptosis by staurosporine, implying atypical apoptosis. Carrying these arguments a step further, BNIP3 associates with mitochondrial (and ER) membranes and has been shown to hererodimerize with the protective proteins Bcl-2 and Bcl-XL(Ray et al., 2000). Our experiments confirm that the opening of the MPTP is also a central feature of death by hypoxia–acidosis. Treatment of cultures with the MPTP inhibitors bongrekic acid (BA) and decylubiquinone (DUB) significantly blocked DNA ladders (Fig. 8A), and Fig. 8B shows directly that the MPTP opens during hypoxia–acidosis. Aerobic cultures retained the mitochondrial dye MitoTracker Red whereas 48 h hypoxia–acidotic mitochondria did not. These studies support an atypical pathway of cell death that retains plasma membrane integrity, requires MPTP opening but involves no caspase activation. The principal steps in the pathway are illustrated in Fig. 9.

All of the results from this laboratory support the argument that chronic severe hypoxia is not necessarily a lethal stress for cardiac myocytes. Lethal stress is imposed by secondary effects of hypoxia including energy depletion,metabolite accumulation (acidosis) or reoxygenation. The combination of hypoxia with acidosis signals a greater stress than hypoxia alone because it indicates that the microenvironment is limited in its life-sustaining capacity. Numerous signals from the extracellular space communicate to intracellular targets through ion channels. In the case of pH, excess protons are expelled from the intracellular compartment of cardiac myocytes through the actions of three main transporters, the Na⁺/H⁺exchanger, the Na⁺/HCO₃^– cotransporter and the vacuolar proton-ATPase(Karwatowska-Prokopczuk et al.,1998; Lagadic-Gossmann et al.,1992; Lazdunski et al.,1985). Under conditions of hypoxia or ischemia, each of these exchangers contributes to the maintenance of neutral intracellular pH by extruding protons. The activity of each transporter is regulated at least in part by the myoplasmic membrane pH gradient(Lazdunski et al., 1985). A decrease of [pH]o below intracellular pH ([pH]i) requires that protons are transported against the proton gradient, and as this gradient increases,transporter activities decrease, and the [pH]i falls correspondingly. If the extracellular pH falls low enough, proton extrusion stops; this has been reported to occur at a [pH]o of ∼6.0 for Na⁺/H⁺exchange in chick cardiac myocytes(Lazdunski et al., 1985). Therefore, when extracellular protons are not cleared, decreased [pH]o will be paralleled by a corresponding or even greater drop of [pH]i. Massive cell death in our model of hypoxia–acidosis occurs between pH 6.5 and 6.0 and this probably coincides with the failure of the Na⁺/H⁺exchanger. However, apoptosis is initiated well before there is significant loss of ATP, and the pathway is unlikely to involve significant contributions from the ATP-sensitive channels, either the vacuolar ATPase or the ATP-sensitive K⁺ channel(Forgac, 1989; Gottlieb et al., 1996; Karwatowska-Prokopczuk et al.,1998).

The extreme effect of acidosis on survival of myocytes under hypoxia may explain why other groups have reported that hypoxia alone promotes apoptosis(Tanaka et al., 1994; Long et al., 1997; Matsui et al., 1999; Kang et al., 2000; Regula et al., 2002). BNIP3 is induced under all conditions of hypoxia and may undergo low-level activation even at neutral [pH]o. Although our results show that acid build-up drives newly synthesized BNIP3 into membranes, it is likely that prolonged hypoxia will promote a continuous low-level BNIP3 activation because of the increased effective proton activity. Hypoxic cells produce >10-fold more acid than aerobic cells with a corresponding increase of proton flux. This condition must lead to the establishment of a new ionic equilibrium involving a net increase of intra- and extracellular H⁺(Webster and Bishopric, 1992; Webster et al., 1994). This change may be sufficient to stimulate low-level BNIP3 translocation in some cells and a consequent low-level activation of the death program. Consistent with this model, proton pumps have been shown to play a role in apoptosis signaling (Anversa and Kajstura,1998; Karwatowska-Prokopczuk et al., 1998), and acidosis has been shown to correlate with apoptosis in a number of other systems(Gottlieb et al., 1995; Perez-Sala et al., 1998).

Loss of cardiac myocytes is a central feature of heart disease of both ischemic and non-ischemic origin (reviewed in Anversa and Kajstura, 1998; Kajstura et al., 1998; Narula et al., 1996). It has been described in multiple regions of the myocardium during infarction,hibernation (Heusch and Schulz,1996) and during both ischemia and subsequent reperfusion(Chen et al., 1997; Fliss and Gattinger, 1996). Death pathways involving necrosis, apoptosis and oncosis have been described(reviewed in Kajstura et al.,1996; Ohno et al.,1998). The probability that hypoxia and acidosis coexist in diseased and/or infarcted myocardial tissue is high because of the disrupted vasculature and elevated lactic acid production in ischemic tissue(Chen et al., 1997; Krayenbuehl and Hess, 1992). Direct measurements of myocardial tissue pH indicate that it drops by 1–2 units within 10 min of ischemia(Marzouk et al., 2002). Therefore, BNIP3 may be expected to play a significant role in cell loss during ischemic heart disease. Combined hypoxia and acidosis reflects a greater hemodynamic disruption than acidosis or hypoxia alone, and the dual signal may provide a selective advantage by activating the death pathway only as a last resort. Acidosis may also activate BNIP3 in skeletal muscle to allow myocyte drop-out under conditions of severe ischemia or hypoxia where the probability of irreversible damage is higher.

Further work is required to characterize this pathway of cardiac myocyte death. We need to determine the contribution of BNIP3 to tissue loss in the intact heart during ischemic heart disease and during infarction. To what extent does this pathway contribute to acute and chronic cell loss during coronary artery disease and congestive heart failure? If this is substantial then it will be important to develop methods to selectively block the pathway. Previous work from this and another laboratory(Kubasiak et al., 2002; Regula et al., 2002) indicates that N-terminal deletions of BNIP3 that lack the transmembrane domain are protective. It may be possible to develop cell-permeable peptides or other small organic mimics that can block BNIP activation very specifically. Finally, the requirement of this death pathway for MPTP opening but absence of caspase activation remains a paradox.

This work was supported by National Institutes of Health grants HL44578,HL69812, HL07294, award # BM034 from the Chiles Endowment Biomedical Research Program of the Florida Department of Health (to K.A.W.), HL HL07109 (to N.H.B.) and American Heart Association, Florida Affiliate Fellowships (to R.A.G., D.P.F., J.W.T., H.L. and M.G.S.). Also supported by an endowed chair from the Walter G. Ross Foundation (to K.A.W.).