Effect of Inorganic Phosphate Concentration on the Nature of Inner Mitochondrial Membrane Alterations Mediated by Ca 21 Ions

Addition of high concentrations (>1 mM) of inorganic phosphate (Pi) or arsenate to Ca-loaded mitochondria was followed by increased rates of H2O2 production, membrane lipid peroxidation, and swelling. Mitochondrial swelling was only partially prevented either by butylhydroxytoluene, an inhibitor of lipid peroxidation, or cyclosporin A, an inhibitor of the mitochondrial permeability transition pore. This swelling was totally prevented by the simultaneous presence of these compounds. At lower Pi concentrations (1 mM), mitochondrial swelling is reversible and prevented by cyclosporin A, but not by butylhydroxytoluene. In any case (low or high phosphate concentration) exogenous catalase prevented mitochondrial swelling, suggesting that reactive oxygen species (ROS) participate in these mechanisms. Altogether, the data suggest that, at low Pi concentrations, membrane permeabilization is reversible and mediated by opening of the mitochondrial permeability transition pore, whereas at high Pi concentrations, membrane permeabilization is irreversible because lipid peroxidation also takes place. Under these conditions, lipid peroxidation is strongly inhibited by sorbate, a putative quencher of triplet carbonyl species. This suggests that high Pi or arsenate concentrations stimulate propagation of the peroxidative reactions initiated by mitochondrial-generated ROS because these anions are able to catalyze C-aldehyde tautomerization producing enols, which can be oxidized by hemeproteins to yield the lower C-aldehyde in the triplet state. This proposition was also supported by experiments using a model system consisting of phosphatidylcholine/dicethylphosphate liposomes and the triplet acetone-generating system isobutanal/horseradish peroxidase, where phosphate and Ca cooperate to increase the yield of thiobarbituric acid-reactive substances.

The decrease in ATP levels that occurs under pathological conditions such as prolonged anoxia or ischemia/reperfusion results in inorganic phosphate (P i ) accumulation and alterations in intracellular Ca 2ϩ homeostasis (1)(2)(3)(4)(5)(6)(7). It is argued that high P i concentrations mimic the metabolic conditions prevalent during ischemia and that, at concentrations higher than 10 mM, P i inhibits mitochondrial oxidative phosphorylation (8). Indeed, it is well known (see Refs. 9 and 10, and references therein) that accumulation of Ca 2ϩ and P i by mitochondria results in increased permeability of the inner mitochondrial membrane. This is proposed to be a key step in the pathogenesis of cell injury that occurs during ischemia and reperfusion (8 -11). Different mechanisms such as lipid peroxidation, phospholipid hydrolysis by phospholipase A 2 , or opening of the mitochondrial permeability transition pore (MTP) 1 have been proposed to take place under these conditions (9 -14). In this regard, P i concentration, which is reported to increase up to 20 mM in cells exposed to prolonged anoxia (6), has a profound effect in the rate, extent, and nature of mitochondrial membrane alterations caused by Ca 2ϩ ions (8 -11, 14 -16).
Data from our laboratory (17)(18)(19)(20) provided evidence that in the absence of added P i , the membrane alterations caused by Ca 2ϩ seem to be mediated mainly by the attack of mitochondrial generated reactive oxygen radicals (ROS) to membrane proteins, resulting in opening of the MTP, a mechanism sensitive to cyclosporin A (9,10). This process is stimulated by prooxidants such as t-butyl hydroperoxide or diamide (17,19,21), which exhaust mitochondrial reducing power such as NAD(P)H and GSH (9,22). Under these conditions, the rate of lipid peroxidation is low and the membrane can be quickly resealed by the addition of EGTA and disulfide reductants (17,21). In contrast, lipid peroxidation is the main form of irreversible mitochondrial membrane permeabilization caused by Ca 2ϩ in the presence of other oxidants such as Fe(II)citrate or Fe(II)ATP (23,24). Interestingly, other data (18,20) from our laboratory indicate that, while the reduced state of coenzyme Q stimulates the protein oxidation that occurs in the presence of Ca 2ϩ and t-butyl hydroperoxide, it protects against damage that occurs in the presence of Ca 2ϩ and Fe(II)citrate. This supports the notion that mitochondrial membrane alterations caused by Ca 2ϩ in situations of oxidative stress may occur via protein thiol oxidation and/or lipid peroxidation.
The aim of this study was to investigate the mechanisms underlying the stimulation of Ca 2ϩ -mediated mitochondrial inner membrane permeabilization by different P i concentrations. Phosphate is the earliest "inducing agent" described (see Refs. 9 and 10, and references therein), and the molecular mechanism by which it mediates membrane permeabilization promoted by Ca 2ϩ remains unclear. Phosphate accumulation, which accompanies intracellular Ca 2ϩ increase during pathological situations, such as prolonged ischemia and reperfusion (see Ref. 8 and references therein), attests to the relevance of this study. Our hypothesis is that high P i concentrations and Ca 2ϩ act synergistically increasing the extent of irreversible membrane permeabilization, due mainly to aldehyde ␣-peroxidation, which yields triplet carbonyls able to amplify the reaction chain of the peroxidative process (25). This was approached by examining the effect of Ca 2ϩ and phosphate ions on TBARS production by mitochondria and by phosphatidylcholine/dicethylphosphate (PC/DCP) liposomes challenged with isobutanal (IBAL) (1) as a phosphate-dependent source of the enol (2), which produces triplet acetone (3) in the presence of horseradish peroxidase (HPR) (see Scheme I). Mitochondrial swelling and H 2 O 2 production were also monitored in Ca 2ϩ plus phosphate-challenged mitochondria. Inhibition of TBARS production and mitochondrial swelling by sorbate, a quencher of triplet carbonyls (26), supports the proposition that P i stimulates lipid peroxidation via aldehyde tautomerization producing enols, very reactive substrates for peroxidases.

MATERIALS AND METHODS
Isolation of Rat Liver Mitochondria-Mitochondria were isolated by conventional differential centrifugation from the livers of adult Wistar strain rats fasted overnight. The homogenate was prepared in 250 mM sucrose, 1.0 mM EGTA, and 5.0 mM Hepes buffer, pH 7.2. The mitochondrial suspension was washed twice in the same medium containing 0.1 mM EGTA, and the final pellet was diluted in 250 mM sucrose to a protein concentration of 80 -100 mg/ml. These mitochondria contain 8 -10 nmol/mg endogenous calcium as determined by atomic absorption spectroscopy.
Liposome Preparation-PC/DCP liposomes were prepared in deionized water by sonication in a Cole-Parmer ultrasonic homogenizer with a nominal output of 100 watts. After 20 min of sonication (with 1-min interval after each 2-min sonication), the solution was centrifuged at 10,000 ϫ g during 20 min to eliminate titanium from the preparation (27).
Standard Incubation Procedure-The experiments were carried out at 30°C in a reaction medium containing 125 mM sucrose, 65 mM KCl, 10 mM Hepes buffer, pH 7.2, 5.0 M rotenone, 2 mM succinate, and 10 M Ca 2ϩ as determined by atomic absorption spectroscopy. Other additions are indicated in the figure legends. The results shown are representative of a series of at least three experiments reproducible within 10%.
Oxygen Uptake Measurements-Oxygen consumption was measured using a Clark-type electrode (Yellow Springs Instruments Co.) in a glass chamber equipped with magnetic stirring.
Determination of Mitochondrial Swelling-Mitochondrial swelling was estimated from the decrease in the absorbance at 520 nm measured in a SLM Aminco DW2000 spectrophotometer.
Determination of H 2 O 2 Production-H 2 O 2 production was determined by the horseradish peroxidase method, as described in Ref. 28.
Thiobarbituric Reactive Substances (TBARS) Measurements-Quantification of TBARS production by mitochondria was performed according to Buege and Aust (29). Briefly, 0.4-ml samples were taken after 10 min of incubation in the conditions described above and mixed with 0.4 ml of 1% TBA in 50 mM NaOH, 0.2 ml of 20% H 3 PO 4 , and 40 l of 10 N NaOH. The mixture was heated at 80 -90°C for 15 min in the presence of 1 mM butylhydroxytoluene (BHT). After cooling, 1.5 ml of butanol was added to the solution. The mixture was shaken and centrifuged at 3000 rpm during 5 min. The optical density of the supernatant was determined at 535 nm. In these conditions, the molar extinction coefficient used to calculate TBARS concentrations is 1.56 ϫ 10 5 M Ϫ1 cm Ϫ1 .
Chemicals-All reagents were commercial products of the highest available grade of purity.

RESULTS
Using the classical mitochondrial swelling technique (9), we showed that addition of 4 mM P i to a suspension of rat liver mitochondria preincubated (2 min) in reaction medium containing 10 M Ca 2ϩ results in a fast decrease in absorbance of the mitochondrial suspension compatible with extensive mitochondrial swelling (Fig. 1A, line e). The presence of catalase from the beginning of the experiment caused an almost complete inhibition of mitochondrial swelling (line b), suggesting the participation of mitochondrial generated H 2 O 2 or H 2 O 2derived radicals in this process (19). In order to ascertain the nature of the membrane alterations observed under these conditions, experiments were performed in presence of the antioxidant BHT, an inhibitor of lipid peroxidation, and of cyclosporin A, an inhibitor of MTP opening. Fig. 1 (lines c and d) shows that the mitochondrial swelling was partially prevented either by BHT or cyclosporin A, respectively, and totally prevented by the simultaneous presence of BHT and cyclosporin A (line a). This behavior is compatible with a membrane permeabilization process mediated partly by lipid peroxidation and partly by MTP opening. In contrast, data in the literature (30 -32) indicate that mitochondrial swelling induced by Ca 2ϩ in the presence of P i can be totally prevented by cyclosporin A. Indeed, Fig. 1B shows that, under our experimental conditions, at lower P i concentration (1 mM), cyclosporin A caused an almost complete protection (line a) against membrane permeabilization. The dashed line represents a control experiment in the absence of added phosphate. In this regard, Fig. 2A shows that SCHEME I. HRP-catalyzed generation of triplet acetone from isobutanal. 1B could be reversed almost completely by the addition of EGTA plus ADP, when mitochondria were incubated in a sucrose-free medium. This finding is compatible with membrane resealing (MTP closure), followed by extrusion of the osmotic support that penetrated before addition of EGTA and ADP. At variance with the conditions above, mitochondrial swelling induced by 4 mM P i was only partially reversed by EGTA and ADP (Fig. 2B), as would be expected if, under these conditions, the membrane permeabilization were caused by both MTP opening and lipid peroxidation.
In order to confirm the occurrence of lipid peroxidation during the mitochondrial swelling induced by high P i concentrations, the production of thiobarbituric acid-reactive substances (TBARS) was measured. Table I shows that TBARS production by mitochondria incubated in the presence of P i increases with increasing P i concentrations. This production was greatly diminished when catalase, EGTA, BHT, or sorbate were present. Accordingly, arsenate, a phosphate analog, was also able to cause a great stimulation of TBARS production. In contrast, acetate, which like P i or arsenate stimulates Ca 2ϩ accumulation by mitochondria by preventing matrix alkalinization and elimination of membrane potential (33), had a much lower effect on lipid peroxidation.
Regarding the stimulation of lipid peroxidation by P i , Fig. 3A demonstrates that mitochondrial swelling induced by 4 mM P i (line d) can be significantly inhibited by sorbate (line b), a putative quencher of triplet carbonyl species (26). This inhibition is enhanced by the simultaneous presence of cyclosporin A (line a), but not of BHT (line c). This suggests that sorbate inhibits only the component of membrane permeabilization related to lipid peroxidation and not by MTP opening. Indeed, Fig. 3B shows that sorbate was not capable of inhibiting swelling induced by 1 mM P i (line a).
In a previous report we have demonstrated that reduced coenzyme Q is the site for electron leakage responsible for Ca 2ϩ -induced ROS production in liver mitochondria (20). In order to ascertain the proposition that mitochondrial damage induced by Ca 2ϩ plus P i is mediated by mitochondrial generated ROS at the level of coenzyme Q, we performed experiments with carbonyl cyanide p-trifluorophenylhydrazone-deenerziged mitochondria treated with antimycin A in the presence or absence of succinate, which maintains coenzyme Q in the reduced or oxidized form, respectively (18). Fig. 4 shows that in the absence of succinate, a condition under which H 2 O 2 production is minimum (20) The nature of the synergism between P i and Ca 2ϩ in the process of lipid peroxidation was assessed by using a model system consisting of PC/DCP liposomes and the well known   triplet acetone-generating system HRP/IBAL (26); DCP confers negative charge to the liposomes. It is well established that triplet acetone can abstract an allylic electron from linolenic and arachidonic acid (see Scheme II, Reaction 2b) (25), initiating their peroxidation. We hypothesized that: (i) whichever the mechanism, the interaction of P i with the membrane lipid phase would be facilitated by neutralization of the membrane surface negative charge by Ca 2ϩ ; and (ii) the phosphate property to catalyze the enolization of aldehyde lipid peroxidation products (25) would favor their further ␣-peroxidation yielding extra triplet carbonyls (Reaction 4), thus increasing the rate of polyunsaturated fatty acid peroxidation (Reaction 2b). Indeed, Fig. 6 shows that P i stimulated lipid peroxidation (as judged by TBARS absorbance increase at 535 nm) in this model system, in a dose-dependent fashion, when a constant Ca 2ϩ concentration (0.5 mM) was present (squares). It should be mentioned that in the absence of Ca 2ϩ the rate of TBARS production (during 10 min) was 50% smaller at a P i concentration of 5 mM (not shown). This supports the hypothesis that the P i effect is potentiated by the electrostatic interaction between Ca 2ϩ and DCP. This is well documented by Fig. 6 (circles) showing that this physical effect of Ca 2ϩ , under constant P i concentration, increases by increasing its concentration from zero to 1 mM. In contrast, when phosphate (5 mM) was replaced by Tris at the same concentration, TBARS production was negligible and Ca 2ϩ had no effect (data not shown). The saturation kinetics displayed by both experiments in Fig. 6 may be rationalized in terms of the liposome concentration being limitant to the peroxidative process. DISCUSSION Previous results (17)(18)(19) from our laboratory show that mitochondrial membrane permeabilization induced by Ca 2ϩ , under conditions of oxidative stress associated with t-butyl hydroperoxide reduction, is mediated by the oxidative attack of reactive oxygen species to membrane protein thiols. This causes thiol cross-linkage and high molecular weight protein aggregate production that may open a membrane pore upon the binding of Ca 2ϩ (34). These alterations of the inner mitochondrial membrane seem to be related to the state designated as mitochondrial membrane permeability transition, characterized by the reversible opening of a membrane protein pore. The current understanding is that MTP is a protein channel (for review, see Refs. 9 and 10) that is opened by low membrane potential (32,35), thiol cross-linking agents (34,36), and various inducing agents in the presence of Ca 2ϩ and closed by EGTA, ADP, Mg 2ϩ (37), sulfydryl reducing agents (21,36), protonation of the mitochondrial matrix (32), and submicromolar concentrations of cyclosporin A (30).
This work provides evidence that in the presence of P i the increase in membrane permeability caused by Ca 2ϩ can be mediated by alterations of both proteins and lipids. The rela- SCHEME II. Chain length amplification of polyunsaturated fatty acid peroxidation by triplet carbonyl products. Peroxidation of PUFA is initiated by either HO⅐ radicals or triplet carbonyls (A 3 *) to give hydroperoxides (PUFA-OOH) and fragmentation products, which include triplet aldehydes, singlet oxygen and short hydrocarbons (Reaction 1). The chain propagation of PUFA peroxidation can thus be amplified (Reaction 2b) by A 3 *. The typical chemiluminescence which accompanies the process is attributed to radiative deactivation of the triplet products (Reaction 2a) and singlet oxygen (Reaction 3a). Singlet oxygen can also undergo 1,3-cycloaddition to PUFA to yield PUFA-OOH (Reaction 3b). In the presence of phosphate and cytochrome c (cyt Fe 3ϩ ), the peroxidation-generated aldehydes can behave similarly to the isobutanal/HRP system (Scheme I) producing lower aldehydes homologous in the triplet state (Reaction 4), which are able to reinitiate the peroxidation cycle (Reaction 1). tive importance of these processes depends on P i concentration. At high P i concentrations, both lipid peroxidation and MTP opening occur. This is illustrated in Fig. 1A by the partial protection conferred either by cyclosporin A or BHT and the total protection obtained when both compounds were present simultaneously. However, at lower P i concentrations, opening of the MTP appears to be the main alteration, since cyclosporin A alone provides an almost complete mitochondrial protection (Fig. 1B).
The role of P i seems to be mediated by a stimulation of mitochondrial ROS production at the level of reduced coenzyme Q, since maintenance of this coenzyme in an oxidized state prevents H 2 O 2 production and mitochondrial swelling (see Fig.  4). Indeed, Kowaltowski et al. 2 provided evidence that, as in the case of t-butyl hydroperoxide plus Ca 2ϩ (19), P i plus Ca 2ϩ did not cause membrane permeabilization when mitochondria were incubated in the absence of O 2 . The mechanism by which P i increases H 2 O 2 production is still unknown but requires the presence of matrix Ca 2ϩ . The experiments carried out with liposomes (Fig. 6) support the notion that Ca 2ϩ binding to inner membrane surface (cardiolipins) decreases negative surface charge density (38) easing the access of P i to the membrane lipid phase where it may stimulate lipid peroxidation (see Scheme II). This may render the respiratory chain more susceptible to electron leakage and ROS production.
With reference to the mechanisms by which P i or arsenate stimulate lipid peroxidation, it is important to recall their property to catalyze tautomerization of aldehydes (C n ) (25), a process that may be followed by cytochrome-catalyzed peroxidation of the enols formed (Ref. 39 and Scheme II, Reaction 4). This leads to increased production of the C n Ϫ 1 aldehyde in the triplet state (triplet carbonyls), which stimulates lipid peroxidation (Scheme II, Reaction 2b). Lipid peroxidation is a chemiluminescent process, due to the production of triplet carbonyls and singlet oxygen (Scheme II, Reactions 2a and 3a) (40 -42).
Triplet carbonyls (Scheme II, Reaction 2b) can initiate the peroxidation of polyunsaturated fatty acids (PUFA) (25) and singlet oxygen (Scheme II, Reaction 3b) directly produces the final product (PUFA-OOH) by 1,3-cycloaddition (42). This would amplify the propagation length of the peroxidative reaction chain. The stimulation by P i of triplet carbonyl formation is strongly supported by the observed inhibitory effect of sorbate, a triplet carbonyl quencher, on both TBARS yield and mitochondrial swelling. Sorbate has been shown to quench triplet acetone formed by the IBAL/HPR system at the micromolar range (26). Regarding the participation of cytochromes in this process, recent results from our laboratory (39) reveal that mitochondria have the ability to promote a cytochrome-mediated P i -stimulated oxidation of diphenylacetaldehyde. This oxidation produces triplet benzophenone and is followed by a BHT-sensitive process of membrane permeabilization.
In addition, the data presented in this paper is relevant regarding the discrimination of factors determinating reversibility versus irreversibility of Ca 2ϩ -induced cell injury under conditions of ischemia and reperfusion. Although membrane permeabilization determined by MTP opening, in studies in vitro, is easily reversed by pore closure induced by EGTA, disulfide reductants, cyclosporin A, adenine nucleotides, etc. (9) under in vivo conditions, in which phosphate concentrations attain high levels, overlapping of the peroxidative process seems to be an important event leading to irreversible or lethal cell injury.