Rat Brain Cortex Mitochondria Release Group II Secretory Phospholipase A2 under Reduced Membrane Potential*

Activation of brain mitochondrial phospholipase(s) A2 (PLA2) might contribute to cell damage and be involved in neurodegeneration. Despite the potential importance of the phenomenon, the number, identities, and properties of these enzymes are still unknown. Here, we demonstrate that isolated mitochondria from rat brain cortex, incubated in the absence of respiratory substrates, release a Ca2+-dependent PLA2 having biochemical properties characteristic to secreted PLA2 (sPLA2) and immunoreacting with the antibody raised against recombinant type IIA sPLA2 (sPLA2-IIA). Under identical conditions, no release of fumarase in the extramitochondrial medium was observed. The release of sPLA2 from mitochondria decreases when mitochondria are incubated in the presence of respiratory substrates such as ADP, malate, and pyruvate, which causes an increase of transmembrane potential determined by cytofluorimetric analysis using DiOC6(3) as a probe. The treatment of mitochondria with the uncoupler carbonyl cyanide 3-chlorophenylhydrazone slightly enhances sPLA2 release. The increase of sPLA2 specific activity after removal of mitochondrial outer membrane indicates that the enzyme is associated with mitoplasts. The mitochondrial localization of the enzyme has been confirmed by electron microscopy in U-251 astrocytoma cells and by confocal laser microscopy in the same cells and in PC-12 cells, where the structurally similar isoform type V-sPLA2 has mainly nuclear localization. In addition to sPLA2, mitochondria contain another phospholipase A2 that is Ca2+-independent and sensitive to bromoenol lactone, associated with the outer mitochondrial membrane. We hypothesize that, under reduced respiratory rate, brain mitochondria release sPLA2-IIA that might contribute to cell damage.

Phospholipase A 2 (PLA 2 ) 1 catalyzes the hydrolysis of the ester bond of fatty acids at the sn-2 position of membrane glycerophospholipids. Because this position is rich in arachidonic acid, great attention has been devoted to the relationship between activation of PLA 2 and the production of eicosanoids (1). However, it is well established that the various isoforms of PLA 2 are involved in many other physiological or pathological processes, such as the remodeling of membrane phospholipids, the removal of oxidized fatty acids, intra-and extracellular signaling, inflammation, and tissue repair (2,3).
Because one of the first biochemical events observed after the onset of brain ischemia is the release of free fatty acids from membrane phospholipids, PLA 2 has been ascribed a relevant role in the consequent brain damage (4 -6). A large body of evidence supports the notion that this class of enzyme is involved in chronic neurodegenerative disease (7).
It is now well established that various isoforms of phospholipase A 2 are present in brain tissue (8), and it is conceivable that each of them should have specific roles depending on their cellular and subcellular localization, the metabolic state of the cell, and response to extracellular stimuli. Three main classes of PLA 2 have been found in brain tissue: cytosolic Ca 2ϩ -dependent (cPLA 2 ), cytosolic Ca 2ϩ -independent (iPLA 2 ), and secretory (sPLA 2 ) (8,9) but their roles in various physiological or pathological events are still largely unclear. Particular attention has been devoted to type IV-cPLA 2 because this enzyme shows a great specificity for arachidonic acid, and it is very likely to be involved in the production of eicosanoids after receptor-mediated activation of the enzyme (10). Evidence has been also reported suggesting its involvement in neurodegeneration (7). In fact, cPLA 2 seems to be present in reactive glial cells, particularly in regions of neuronal loss, and it is not detectable in those brain regions in which neurons do not degenerate or are protected from death (11). However, a predominant expression of cPLA 2 mRNA in neurons of rat brain has also been reported (12), suggesting its involvement in neurotransmission or other neuronal functions through the generation of such lipid mediators as arachidonic acid, eicosanoids, and platelet-activating factor (13)(14)(15).
Two Ca 2ϩ -independent PLA 2 s have been isolated from bovine brain cytosol showing different specificities for 1,2-diacylsn-glycero-3-phosphoethanolamine and ethanolamine plasmalogens (16). More recently, an 80-kDa type VI iPLA 2 was purified from rat brain (9). This enzyme shows a head-group preference for phosphatidylcholine, and its specific activity is 20 -50-fold higher than that of type IV-cPLA 2 depending on brain areas. Secretory PLA 2 types IIA and V are also present in all areas of rat brain, and typeIIA-sPLA 2 is the predominant enzyme of this class and is associated with the particulate fraction, whereas type V sPLA 2 is mainly found in the soluble fraction (9). Secretory PLA 2 s have a low molecular mass (13)(14)(15)(16)(17)(18), show a low specificity for fatty acids, and require millimolar Ca 2ϩ for full activity. They are widely distributed in animal tissues and extracellular fluids and exert many different functions (17,18). sPLA 2 -IIA, one of the several groups of this class, was initially purified from synovial fluid (19) and platelets (20). Enzymes belonging to this group are also present in the venom of various snake species, and they have a potent neurotoxic effect (21,22). Because brain tissue possesses specific presynaptic receptors for sPLA 2 -IIA from snake venoms, the involvement of an endogenous sPLA 2 in neurotransmission seems very likely (22,23). In fact, exogenous sPLA 2 modulates AMPA receptor function (24) and an enzyme, immunochemically identical to sPLA 2 -IIA, has been detected in synaptic vesicles and secreted upon depolarization or by neurotransmitter stimulation (25).
Because mediators such as tumor necrosis factor-␣ and interleukin-1␤ induce sPLA 2 mRNA and enzymatic activity in immortalized astrocyte cell lines (26), the involvement of this group of enzymes in inflammation has been also suggested.
The overall message emerging from these reports indicate that, in general, PLA 2 (s) are involved in physiological mechanisms but they also participate in the development of neurodegeneration after short-term ischemia or as the consequence of long-term neurodegenerative diseases. There is a large body of evidence that, in both cases, mitochondrial functions undergo substantial alterations that reduce their capability of producing ATP and address neurons toward necrotic or apoptotic cell death (27)(28)(29)(30)(31).
The presence of PLA 2 activity in brain mitochondria has been known since the early 1970s (32)(33)(34), but the identities and the functions of the(se) enzyme(s) are still unclear. More recently, it has been reported that gerbil brain mitochondria contains a PLA 2 with an estimated molecular mass of 14 kDa whose activity increases after ischemia and reperfusion (35). This observation and the detection of sPLA 2 -IIA in rat brain particulate fraction suggests that brain mitochondria might contain a sPLA 2 , similar to that isolated from liver mitochondria (36), and it prompted us to verify whether the enzyme could exit from the organelle under impairment of mitochondrial functions, similarly to other mitochondrial proteins such as cytochrome c and caspase-9 (37,38). In this study, we have demonstrated that mitochondria purified from rat brain cortex and mitochondria of neuron-like PC-12 cells and U-251 astrocytoma cells contain an enzyme having biochemical and immunological properties characteristic of sPLA 2 -IIA. By contrast, in the same cell types, sPLA 2 -V is not present in mitochondria, but it has mainly cytosolic and nuclear localization. The enzyme present in isolated rat brain cortex mitochondria is very probably associated with mitoplasts, and it is released as the consequence of a reduction in membrane potential. Rat brain cortex mitochondria contain also a Ca 2ϩ -independent PLA 2 that seems to be localized on the outer membrane.  (39 -41), was prepared as reported previously (42). In brief, bacteria (E. coli; American Type Culture Collection) were grown overnight in triethanolamine medium and resuspended in fresh medium containing [9 -10-3 H]oleic acid (10.3 Ci/mmol) complexed with 0.02% bovine serum albumin (fatty acid-free). After incubation (3 h at 37°C), bacteria suspension was autoclaved (20 min at 120°C and 1 kg/cm 2 ). After washings, labeled E. coli was sedimented by centrifugation at 15,000 ϫ g for 30 min. The specific radioactivity of this substrate (0.7-3 nCi/nmol of phosphatidylethanolamine) was calculated by measuring the radioactivity associated with phosphatidylethanolamine, which represents Ϸ70% of total phospholipids and contained more than 90% of the incorporated labeled oleic acid.

Materials-[
Preparation of Rat Brain Cortex Mitochondria-Mitochondria prepared from rat brain cortex (2-month-old CD rats; Charles River Laboratories) were purified on sucrose gradient as reported previously (43) and then resuspended in an isotonic medium containing 2 mM HEPES, pH 7.4, and 0.32 M sucrose (S/H buffer). The purity of brain mitochondria was checked by marker assay (44). The specific activity of cytochrome c oxidase (mitochondrial marker enzyme) was 5.7 times higher in mitochondria than in homogenate, whereas NADPH-cytochrome c reductase (microsomal marker enzyme), Na ϩ ,K ϩ -ATPase (plasma membrane marker), and arylsulfatase A (lysosomal marker) were less than 5% with respect to the homogenate (45). The integrity of the outer mitochondrial membrane was in the range 90 -93%, as calculated by the latency in cytochrome c oxidase assay (45). In some experiments, mitochondria (2-3 mg of protein/ml of 0.32 M sucrose and 2 mM HEPES, pH 7.4) were incubated with 50 g of trypsin or Pronase at 30°C, a concentration known to be ineffective on mitochondrial functional parameters (46). After 30 min, aliquots were taken and mitochondria recovered by centrifugation at 10,000 ϫ g for 10 min. Protein content was determined by the method of Bradford (47) using bovine serum albumin as standard.
Preparation of Mitoplasts-Mitoplasts were prepared as described previously (48). In brief, purified mitochondria were treated with digitonin (0.4 mg/mg of protein) for 30 min at 0°C and then loaded on sucrose gradient for their separation from intact mitochondria.
Cytofluorimetric Analysis of Mitochondrial Membrane Potential (⌬⌿ m )-Cytofluorimetric analysis of mitochondrial membrane potential was done as reported previously (48). In brief, purified mitochondria (0.3 mg of protein) were incubated for 10 min at room temperature in the presence of 1 M DiOC 6 (3), in different respiratory conditions. Suspensions were immediately subjected to cytofluorimetric analysis, using a FACScan flow cytometer (BD Biosciences) equipped with a focused argon laser. For complete depletion of ⌬⌿ m (positive control), the mitochondria uncoupler CCCP (100 M) was used. Data were analyzed and stored with the use of a data management system (LYSYS software). DiOC 6 (3) green fluorescence was plotted on a logarithmic scale versus the frequency of events. The mean value of the integral of fluorescence was also evaluated.
Determination of Phospholipase A 2 Activity Released from Mitochondria-Purified mitochondria were resuspended in S/H buffer (resting state) or, when indicated, in the same buffer containing respiratory substrates or the uncoupler CCCP. The pH of the solutions of ADP, pyruvate, and malate were adjusted to pH 7.4 with KOH before additions. In any case, Ca 2ϩ was not added and its concentration in S/H buffer, measured with Fura-2 (acid) (42), was 1 M. Mitochondria were incubated at 37°C and then pelleted at 10,000 ϫ g for 10 min. Control samples (T 0 ) were prepared in identical conditions, but they were not incubated. A convenient aliquot of the supernatant was taken for PLA 2 assay and incubated at 37°C with [ 3 H]oleate-labeled E. coli (ϳ30 nCi per sample) in a total volume of 300 l containing 125 mM Tris-HCl, pH 7.4, 2 mM CaCl 2 and 1 mg/ml BSA (fatty acid-free). The reaction was stopped by the addition of 150 l of 2 N HCl and 150 l of BSA (20 mg/ml). After a period of 20 min at 4°C, labeled bacteria were pelleted at 10,000 ϫ g for 5 min. The radioactivity of the supernatant, containing free [ 3 H]oleic acid, was determined (49). The activity of PLA 2 was calculated by subtracting blank sample radioactivity from that of samples with extramitochondrial medium. Determination of Fumarase Activity-Fumarase activity was determined spectrophotometrically at 240 nm. The supernatant (4 -10 g of protein), recovered after the incubation of mitochondria in S/H buffer, was incubated with 0.5 mM malate for 15 min at 25°C in 0.35 ml of phosphate buffer (100 mM, pH 7.4) (50).
Western Blot Analysis-Western blot analysis of proteins, released from rat brain cortex mitochondria, was performed using monoclonal antibody against rat sPLA 2 -IIA (generous gift from Dr. H. van den Bosch and Dr. J. Aarsmann, Utrecht) essentially as indicated by van der Helm et al. (51). In brief, mitochondria purified from 10 rat brain cortices were incubated in S/H buffer for 60 min at 37°C, and the incubation medium was subjected to dialyze against 50 mM sodium acetate, pH 4.5, in the presence of 0.2 M NaCl at 4°C overnight (52). Samples were concentrated and then subjected to Western blot analysis. Blot was analyzed using Duoscan T1200 (AGFA, Seifert, Germany) scanning densitometer and Quantity One 4.4 software (Bio-Rad Laboratories, Hercules, CA).
Confocal Immunofluorescence Microscopy-PC-12 and U-251 cells, after washings with phosphate-buffered saline (PBS), were placed on cover slips and loaded with 500 nM Mito-Tracker Red CM-XRos at 37°C for 15 min. Cells were fixed and permeabilized with a mixture of methanol-acetone (1:1 by vol) at Ϫ20°C for 10 min. When indicated, PC-12 cells were fixed with methanol at Ϫ20°C for 7 min, rinsed in PBS, and treated with 0.1% (v/v) Triton X-100 in PBS for 10 min at room temperature. Nonspecific binding of antibodies or antisera was blocked by a preliminary incubation of fixed cells with 1% (w/v) BSA in PBS (blocking buffer). This was followed by the incubation at room temperature for 1 h and 30 min with monoclonal antibody against rat sPLA 2 -IIA diluted 1:10 in blocking buffer or with polyclonal antibody against sPLA 2 -V (1:100 in blocking buffer). The cells were washed for 20 min in PBS and incubated with rabbit Ig conjugated to Alexa 488 diluted 1:100 in blocking buffer at room temperature for 1 h. After washing with PBS for 20 min, cells were loaded with 10 M TO-PRO-3, a monomeric cyanine stain for double-stranded DNA giving a blue fluorescence, for 7 min at room temperature. Confocal analysis was performed with a confocal microscope (Bio-Rad MRC 1024) using an Ar/Kr laser. The fluorescences of Mito-Tracker and Alexa 488 were detected with excitation wavelengths of 568 and 488, respectively, whereas that of TO-PRO-3 was at 647 nm. Images were elaborated with the colocalization module of Imaris software (Bitplane, Zurich, Switzerland) on a SGI Octane work station (SGI, Mountain View, CA).
Electron Microscopy-U-251 astrocytoma cells were rapidly frozen in a Cryopress device (Med Vac Inc., St. Louis, MO) by impact onto a liquid helium-cooled copper block. Frozen samples were rapidly transferred into a Balzers FSU 010 cryosubstitution unit and treated with: osmium tetroxide in 2% methanol for 8 h at Ϫ90°C, for 8 h at Ϫ60°C, and another 8 h at Ϫ30°C. Osmium tetroxide was then removed by two washings with methanol at Ϫ30°C. After cryosubstitution, samples were embedded in Lowcryl K4M as described by Carlemalm et al. (53).
Cryosections of U-251 cells were treated with sodium metaperiodate for 1 h at room temperature and then incubated at room temperature for 10 min in PBS containing 3% BSA and 20 mM NaN 3 (buffer A). Next, samples were incubated in buffer A containing 1% normal goat serum and 100 mM glycine for 4 h at room temperature. This step was followed by incubation for 3 h at room temperature and then overnight at 4°C with a monoclonal anti-sPLA 2 -IIA antibody (1:20 in buffer A ϩ 1% normal goat serum). Sections were washed in buffer A containing 1% Tween 20 (five times, 1 min each) and incubated for 3 h at room temperature with goat anti-mouse IgG antibody labeled with 5-nm gold particles (1:80). Sections were washed in PBS, postfixed for 3 min in 2% glutaraldehyde in PBS, and counterstained with uranyl acetate. Sections were finally examined by electron microscopy (Philips TEM 208).
Statistical Analysis-Statistical significance of the data was evaluated by analysis of variance followed by Student's t test or Scheffé's multiple comparison test. Differences were considered statistically significant when p Ͻ 0.05.

Release of Phospholipase A 2 from Rat Brain Cortex Mitochondria: Time
Dependence-A time-dependent and highly significant increase of PLA 2 activity was observed in the extramitochondrial medium after incubation of purified mitochondria in S/H buffer, pH 7.4 ( Fig. 1). After 60 min, PLA 2 -specific activity in the medium was nearly 6 times that of nonincubated samples (T 0 ). A further increase was observed after 90 min but a greater variability of the data was noticed.
The homogenization of brain cortex in the presence of 100 g/ml heparin reduced PLA 2 activity of the homogenate to 40% of that measured in control samples (homogenate without heparin). Purified mitochondria were then prepared from both homogenates, and the release of sPLA 2 was determined as described above. No differences were observed in the specific activity of the enzyme released from mitochondria prepared in the presence or in the absence of heparin, even if the direct addition of this compound to the assay system reduced PLA 2 activity by 70% (data not shown). This indicates that the mitochondrial enzyme was not accessible to heparin during the homogenization of the tissue and during the preparation of mitochondria. Therefore, it had to be associated with the inner mitochondrial compartments in the tissue before the homogenization.
To exclude that the observed phenomenon could be due to an unspecific mitochondrial leakage, the same medium was assayed for fumarase and protein content and no significant changes were observed with time up to 90 min. Thus, it is conceivable to conclude that PLA 2 is preferentially released from rat brain cortex mitochondria with respect to other mitochondrial proteins.
Properties of the Released Mitochondrial Phospholipase A 2 -To provide a characterization of the released PLA 2 , we have assayed the effect of Ca 2ϩ concentration on its activity. As shown (Fig. 2), the rate of substrate hydrolysis significantly increased with 10 M Ca 2ϩ and was maximal at millimolar concentrations. A relatively low activity (15.5 Ϯ 1.0 nmol/mg of protein/h) was measured in the presence of EGTA (10 M). Mitochondria were pelleted at 10,000 ϫ g for 10 min, and supernatant was assayed for PLA 2 (OE), fumarase (q), or total protein content (f). Results are expressed as a percentage of values (mean Ϯ S.E.) with respect to medium of nonincubated mitochondria and are from three independent experiments in duplicate. **, p Ͻ 0.01 and ***, p Ͻ 0.001 versus nonincubated samples A significant reduction of enzyme activity (Ϫ80%) was observed in the extramitochondrial medium, with DTT (5 mM), which is an inhibitor of sPLA 2 (54). BEL, a specific inhibitor of Ca 2ϩ -independent PLA 2 (55), and arachidonoyltrifluoromethyl ketone, known to inhibit type IV cytosolic Ca 2ϩ -dependent PLA 2 (56), were both ineffective (Fig. 3).
Western blot analysis of proteins released upon incubation of mitochondria revealed that monoclonal antibodies against rat liver sPLA 2 recognized a protein having a molecular mass Ϸ 14 kDa, similar to that of human synovial sPLA 2 , which did not immunoreact with the antibody (Fig. 4). On the basis of this result, the protein will be referred as type IIA-sPLA 2 even if it cannot be completely excluded that the antibody might recognize another isoform of the same group.
From densitometric analysis and from total amount of proteins present in whole mitochondria and in the medium 60 min after incubation in the absence of respiratory substrates, it was possible to estimate that nearly 50% of the total mitochondrial enzyme immunoreacting with sPLA 2 -IIA antibody was released.
Effect of Respiratory Substrates on Membrane Potential and on the Release of sPLA 2 from Mitochondria-The respiratory activity of purified mitochondria resuspended in S/H buffer, pH 7.4, was evaluated by flow cytometric analysis in different metabolic states. Single-parameter fluorescence histograms of mitochondria stained with the membrane-potential-sensitive DiOC 6 (3) are reported in Fig. 5. The addition of respiratory substrates caused a shift toward higher fluorescence intensity with respect to that of purified mitochondria resuspended in S/H buffer only (resting state), whereas the treatment with the uncoupler CCCP caused the opposite effect.
Because the increase of fluorescence correlates with an increase of membrane potential and consequently with the metabolic state of mitochondria, we hypothesized that the reduction of respiratory rate could trigger the release of sPLA 2 from mitochondria.
Therefore, PLA 2 release in different metabolic conditions was tested. As shown in Fig. 6, the addition of respiratory substrates caused nearly 60% reduction of the release of mitochondrial sPLA 2 , whereas a slight increase was observed in the presence of CCCP (100 M) that caused a collapse of ⌬⌿ m (Fig.  5). Under identical conditions, the addition of CCCP to samples containing ADP-malate-pyruvate partially reversed the effect of respiratory substrates on the release of sPLA 2 from mitochondria. These observations indicate that a reduction of mem-brane potential in mitochondria isolated from rat brain cortex induces the release of an enzyme that as biochemical and immunological properties identical to those of sPLA 2 -IIA.

Incubation of [ 3 H]Oleic
Acid-labeled E. coli with Mitochondria from Rat Brain Cortex-We have also assayed the liberation of labeled oleic acid from E. coli substrate by incubating rat brain cortex mitochondria, without respiratory substrates, in S/H buffer with 2 mM Ca 2ϩ or in the presence of 1 mM EGTA at different pH values (Fig. 7A). In the presence of Ca 2ϩ , the maximal rate of the liberation of labeled oleic acid was at pH 7.4, whereas it was at pH 8.0 when Ca 2ϩ was substituted by EGTA.
The preincubation of mitochondria with 5 mM DTT reduced the hydrolysis of the substrate in the presence of 2 mM Ca 2ϩ , pH 7.4, which was not affected by BEL, known to inhibit Ca 2ϩ -independent PLA 2 (Fig. 7B). On the other hand, when PLA 2 activity was assayed at pH 8.0 without Ca 2ϩ (Fig. 7C), the treatment with BEL caused a significant decrease of the liberation of labeled oleic acid and was not affected by DTT. These observations indicate that the direct incubation of purified brain mitochondria with E. coli-labeled substrate allows the detection of two distinct PLA 2 s, one Ca 2ϩ -dependent and the other Ca 2ϩ -independent, confirming previous reported data with rat liver mitochondria (57).
To verify the localization of the two enzymes, purified mitochondria were treated with proteases or digitonin and then assayed for PLA 2 activity. Mitochondria were treated with trypsin or Pronase (25 g/mg mitochondrial protein) and then assayed for Ca 2ϩ -independent or -dependent activities. The choice of protease concentration was based on the observation that it did not affect mitochondrial membrane potential, as shown previously (46). The treatment with proteases caused a decrease of more than 50% in the activity of the Ca 2ϩ -independent enzyme, whereas it did not affect that of the Ca 2ϩ -dependent one (Fig. 8A). The partial hydrolysis of the Ca 2ϩindependent enzyme indicates that it is not fully accessible to proteases.
Under identical conditions, recombinant sPLA 2 -IIA (a gift from Prof. H. van den Bosch), sPLA 2 released from rat brain mitochondria, and sPLA 2 II-A from C. atrox venom were inactivated by incubation with proteases (Fig. 8B), it can be concluded that Ca 2ϩ -dependent enzyme is not accessible to proteolysis in intact mitochondria.
Similar experiments were performed with digitonin-treated mitochondria that allowed the removal of the outer membrane and the isolation of the mitoplasts (48). As shown in Fig. 9A, the specific activity of the Ca 2ϩ -dependent PLA 2 increased in mitoplasts with respect to intact mitochondria, and it was significantly reduced by the preincubation of mitoplasts with DTT (Fig. 9B).
The removal of the outer membrane proteins alone cannot account for the 4-fold increase of the specific activity, but it may reflect an easier release of the enzyme or a better interaction with the labeled substrate. Hydrolysis of the labeled substrate was also observed upon incubation of mitoplasts in the presence of EGTA but to a lesser extent than that measured with purified mitochondria (Fig. 9A).
These experiments allowed us to detect two distinct PLA 2 s in rat brain mitochondria: one Ca 2ϩ -dependent and associated with mitoplasts, probably the sPLA 2 -IIA that is released under certain conditions, and the other one Ca 2ϩ -independent and sensitive to BEL present mainly in the outer membrane, which might be the cytosolic iPLA 2 previously reported (9) or that found in liver mitochondria (57). In the latter case, we have to suppose that, under our experimental conditions, a change in iPLA 2 localization had to take place, because this enzyme seems to be present in the inner mitochondrial membrane (57), and it might be responsible for the activity that we have found in mitoplasts in the presence of EGTA (Fig. 9A).
Subcellular Localization of sPLA 2 -IIA and sPLA 2 -V in PC-12 and U-251 Cells-The presence of sPLA 2 -IIA in mitochondria was also assessed in neuron-like PC-12 cells that were analyzed by confocal laser microscopy using a monoclonal antibody raised against the liver enzyme (51) (Fig. 10B, green signal) and showed a high degree of colocalization (Fig. 10 C and D) with the red signal of Mito-Tracker, which accumulates in active mitochondria (Fig. 10A) (58). As shown in E the percentage of colocalization in various cell layers reached a maximum of 20%. A similar degree of colocalization was obtained when cells were fixed with methanol and acetone, thus avoiding Triton-X100 treatment.
To assess the specificity of monoclonal antibody against mitochondrial sPLA 2 -IIA and to demonstrate that its colocalization with Mito-Tracker dye cannot be the consequence of an artifact of the experimental procedure, experiments with PC-12 cells were repeated comparing the subcellular localization of sPLA 2 -IIA with that of another secretory PLA 2 , sPLA 2 -V, which is also present in rat brain (9). The sequences of the two enzymes have 41% identity as suggested by the ClustalW data In fact, as shown in Fig. 11B, mitochondria are devoid of sPLA 2 -V because the enzyme does not colocalize with Mito-Tracker (red fluorescence) but it is mainly present in punctate cytoplasmic structures (green fluorescence). Furthermore, a certain degree of colocalization can be observed in a central section of the cells in which sPLA 2 -V (green fluorescence) colocalizes with the nuclear marker TO-PRO-3 (blue fluorescence). Fig. 11, C and D, show confocal immunofluorescence analysis of the subcellular distribution of sPLA 2 -IIA and sPLA 2 -V in U-251 astrocytoma cells. The choice of a glial cell line was suggested by the consideration that astrocytes are quantitatively predominant in brain cortex and that both sPLA 2 are expressed in these cells (62). It is quite evident that, whereas sPLA 2 -IIA has a large colocalization with perinuclear mitochondria (C; yellow signal), sPLA 2 -V colocalize with the nuclear marker (D; green plus blue fluorescence), and it is completely absent in mitochondria. Both isoforms are also present in cytosol.
The mitochondrial localization of sPLA 2 -IIA was confirmed by immunoelectron microscopy in U-251 cells (Fig. 12B). Immunogold particles are mainly localized in the inner compartments of mitochondria; this observation is in agreement with the biochemical evidence reported in Figs. 8 and 9. The enzyme is also present in other cellular compartments and very likely to be present in the endoplasmic reticulum. Control experi-ments, in which the antibody against type IIA-sPLA 2 was omitted, showed no immunogold particles (Fig. 12A).

DISCUSSION
In this study, we have demonstrated that rat brain mitochondria possess two PLA 2 s, at least. One is Ca 2ϩ -independent, is sensitive to BEL, and seems to be associated with the outer membrane, because its activity is partially lost by the treatments with proteases. Because this enzyme has properties similar to those of the iPLA 2 purified from rat brain cytosol and is sensitive to BEL (9, 63), we may suppose that it might bind to the mitochondrial membranes under certain circumstances. We cannot exclude the possibility that the cytosolic enzyme binds to the outer mitochondrial membrane during tissue homogenization and isolation of mitochondria. We have no further indication of the identity of the mitochondrial iPLA 2 , but it is worth mentioning that Woelk and Porcellati (33) also reported a Ca 2ϩ -insensitive PLA 2 using mixed micelles of specifically radiolabeled substrates. More recently, it has been reported that liver mitochondria possess a Ca 2ϩ -independent PLA 2 that has been identified as BEL sensitive-iPLA 2 (57), and it has been proposed that the enzyme might participate in the removal of damaged mitochondria. This enzyme seems to be localized in the inner membrane, and then it should not be accessible to proteases.
The other enzyme is very likely to be sPLA 2 -IIA, based on its biochemical and immunological properties. In this case, it is unlikely that an enzyme having identical properties but different subcellular localization might account for the activity detected in the mitochondrial fraction. In fact, the treatment of purified mitochondria with proteases did not affect the Ca 2ϩdependent enzyme (Fig. 8), whereas the removal of the outer membrane by digitonin treatment greatly increased its specific activity (Fig. 9). This excludes the possibility that sPLA 2 -IIA, present in other cellular compartments at the time of tissue homogenization, might have contaminated mitochondrial preparations by interacting with components of the outer membrane. On the other hand, it should be also excluded that, during the isolation of mitochondria from brain tissue, sPLA 2 -IIA might have been unspecifically transported into mitochondria by crossing the outer membrane and then becoming associated with mitoplasts. In fact, the transport of proteins into mitochondria from other cell compartments is a process requiring their recognition at the outer membrane surface and a complex machinery of translocases that are present in the outer and inner membranes, have specific binding sites, and are strictly regulated (64). These considerations are consistent with the mitochondrial localization of sPLA 2 -IIA in the cells of the nervous tissue, similarly to previously reported data for liver mitochondria (36). The mitochondrial localization of sPLA 2 -IIA is also supported by the demonstration of the presence of the enzyme, for the first time, to the best of our knowledge, in the mitochondria of intact PC-12 cells and U-251 astrocytoma cells, as shown by immunofluorescence colocalization with a specific mitochondrial dye (Figs. 10 and 11). Further and conclusive evidence for the mitochondrial localization of the enzyme has been provided by immunoelectron microscopy of ultracryofixed U-251 astrocytoma cells (Fig. 12).
The results of our immunofluorescence experiments and immunoelectron microscopy have shown that sPLA 2 -IIA is also associated with cellular components other then mitochondria. It is worth mentioning that previous immunocytochemical studies have localized a type II sPLA 2 in granule-like organelles of PC-12 cells, and the release of the enzyme in the extracellular medium after stimulation of these cells with epinephrine or KCl has also been reported (25). These observations further support the hypothesis that sPLA 2 may play both intracellular and extracellular roles (3), depending on cell types and/or conditions. How proteins such as sPLA 2 s, synthesized with an N-terminal signal sequence typical for secretion, could be targeted to intracellular organelles is still unknown. However, it has been proposed that the secreted sPLA 2 -IIA, lacking the N terminus signal sequence, could undergo internalization through the association with heparan sulfate chain of glycosylphosphatidylinositol (65). The internalized enzyme might then be targeted to intracellular compartments.
The most relevant finding of our study is that isolated mitochondria from rat brain cortex release PLA 2 when incubated in a isotonic medium at pH 7.4 containing Ϸ1 M Ca 2ϩ and no respiratory substrates. The addition of respiratory substrates increases membrane potential and reduces the release of the enzyme in the extramitochondrial medium. The released PLA 2 has properties identical to those reported for type IIA secreted enzyme isolated from other cells or fluids (20, 66, 67) and immunoreacts with monoclonal antibody raised against rat liver mitochondrial sPLA 2 (68).
A further indication supporting the identification of the enzyme released from mitochondria as a type II secretory PLA 2 derives from the observation that this antibody does not recognize sPLA 2 -V, as reported previously (51). We cannot exclude the possibility that a type II-sPLA 2 other than IIA might also be present in neural cells and cross-react with the antibody raised against recombinant sPLA 2 -IIA (51) used in this study. In fact, other type II-sPLA 2 s (IIC, IID, IIE, and IIF) can be expressed in various mammalian cells (65). For instance, it has been reported that, although type IIA sPLA 2 mRNA is present in all rat brain regions and peripheral tissues, type IIC mRNA seems to be specifically expressed in brain (69). However, it seems unlikely that the mitochondrial enzyme might be identified as sPLA 2 -IIC, because the enzyme released from mitochondria is inhibited by heparin (data not shown), whereas type IIC activity is not affected by this proteoglycan (65). Type IID-sPLA 2 has been cloned based on a tBLASTn search, but Northern analysis has shown that it is not expressed in rat brain (70). Tissue distribution of type IIF isoform, determined by reverse transcriptase-PCR on human adult cDNA panels and Southern blot, has shown its presence in some human organs but not in brain (71) and, in adult mice, the expression seems to be limited to testis (72). Type IIE-sPLA 2 has been also cloned and is expressed in human tissue including brain (73). However, contrary to type IIA, its expression in rat tissues has not been reported so far. In conclusion, it is reasonable to think, on the light of current knowledge, that brain mitochondrial sPLA 2 might be identified as type IIA, which has been isolated from rat and human tissues (36,66).
Because the removal of the outer mitochondrial membrane increases its specific activity, the enzyme seems to be associated with mitoplasts excluding the possibility of a contamination of a PLA 2 from other cellular compartments. Thus, we may conclude that rat brain cortex mitochondria contain sPLA 2 -IIA, which under reduced respiratory activity is released in the extramitochondrial space. The release of the enzyme is not caused by a damage of mitochondrial membranes severe enough to cause the leakage of proteins from matrix, because fumarase is not released under the same conditions.
The presence of a secretory PLA 2 in mitochondria raises per se a number of questions concerning whether or not it may be in the active form in the organelle and induces one to speculate on its functions in vivo. Although sPLA 2 are fully active at millimolar Ca 2ϩ concentration, a consistent activity can be reached at Ca 2ϩ concentrations that can be achieved in intra-  (74) and, very likely, in mitochondria. In fact, it is well established that mitochondria play an important role in Ca 2ϩ homeostasis and that they can accumulate a large amount of this cation when its cytosolic concentration increases (75,76). Thus, we may suppose that oscillations of Ca 2ϩ concentrations modulate mitochondrial sPLA 2 activity and consequently the rate of production of free fatty acids and lysophosphoglycerides, which participate in the regulation of some functions of mitochondrial membrane proteins. It is known that free fatty acids modulate cytochrome c oxidase activity (77) and uncouple oxidative phosphorylation (78). Free fatty acids also act as physiological modulators of mitochondrial permeability because they cause swelling of the organelle without opening cyclosporin-sensitive mitochondrial permeability transition pores (79).
Among the free fatty acids produced by the action of PLA 2 , great attention has been devoted to the effects of polyunsaturated fatty acids on mitochondrial functions because they seem to play important roles in the decision of cellular life or death. In HL-60 cells, polyunsaturated fatty acids inhibit their growth and induce apoptosis by releasing cytochrome c from mitochondria by two distinct mechanisms: membrane depolarization-dependent and -independent (80). It has been reported recently that micromolar arachidonic acid induces permeability transition in isolated rat liver mitochondria and in hepatoma cells (81). This effect was erroneously attributed to cPLA 2 activation because treatment with aristolochic acid did not prevented permeability transition induced by arachidonic acid in isolated mitochondria or in intact cells, but it prevented permeability transition pore opening, cytochrome c release, and cell death induced by tumor necrosis factor-␣ (81). However, aristolochic acid is not an inhibitor of either Ca 2ϩ -dependent or -independent cytosolic PLA 2 (82, 83), but it is considered an inhibitor of sPLA 2 . Thus, those results support the involvement of this class of enzymes in apoptosis.
The other products of phospholipid hydrolysis by PLA 2 , lysophospholipids, also interfere with some mitochondrial functions. For instance, lysophosphatidylcholine modulate the activities of various mitochondrial enzymes, causing their increase or decrease depending on its concentration (84). However, at relatively high concentrations, lysophosphatidylcholine causes the breakdown of mitochondrial membranes. Lysophosphatidylcholine and other compounds, having similar chemical properties (i.e. the lipid mediator platelet-activating factor and the experimental anticancer drug hexadecylphosphocholine), also influence mitochondrial Ca 2ϩ transport and membrane potential. Thus, an impairment of mitochondrial functions caused by the accumulation of PLA 2 products is very likely (85,86).
When ⌬⌿ m decreases as a consequence of reduced respiratory rate, the mitoplast-associated sPLA 2 is released in the extramitochondrial space, and we may suppose that it hydrolyzes membrane phospholipids, thereby causing severe cell damage. Although the molecular mechanisms causing the release of sPLA 2 from mitoplasts toward the extramitochondrial space are still unknown, our findings suggest that the released enzyme might participate to the cascade of events leading to cell death as the consequence of mitochondrial dysfunction. It is possible that the release of sPLA 2 in the extramitochondrial compartment causes additional effects to those induced by the receptor-mediated activation of Ca 2ϩ -dependent cPLA 2 type IV (1). This might be the case for neuronal degeneration and death induced by excessive glutamate release. In vivo and in vitro studies have shown that glutamate excitotoxicity is linked to the activation of NMDA receptors, which induces the increase of cytosolic Ca 2ϩ , phosphorylation, and translocation of cPLA 2 from cytosol to membranes and the liberation of fatty acids from phospholipids (88,89). However, it has been also observed a reduction of ⌬⌿ m and a failure of energy production in mitochondria of neuronal cells by elevated glutamate concentration (87) and release of pro-apoptotic proteins as cytochrome c (37) and caspase-9 (38). Our results show that, in addition to these proteins, sPLA 2 is also released from mitochondria when ⌬⌿ m is reduced, and it might contribute to neuronal cell death.