Cytotoxic Phospholipid Oxidation Products

Phospholipid oxidation products accumulate in the necrotic core of atherosclerotic lesions, in apoptotic cells, and circulate in oxidized low density lipoprotein. Phospholipid oxidation generates toxic products, but little is known about which specific products are cytotoxic, their receptors, or the mechanism(s) that induces cell death. We find the most common phospholipid oxidation product of oxidized low density lipoprotein, phosphatidylcholine with esterified sn-2-azelaic acid, induced apoptosis at low micromolar concentrations. The synthetic ether phospholipid hexadecyl azelaoyl phosphatidylcholine (HAzPC) was rapidly internalized, and overexpression of PLA2g7 (PAF acetylhydrolase) that specifically hydrolyzes such oxidized phospholipids suppressed apoptosis. Internalized HAzPC associated with mitochondria, and cytochrome c, and apoptosis-inducing factor escaped from mitochondria to the cytoplasm and nucleus, respectively, in cells exposed to HAzPC. Isolated mitochondria exposed to HAzPC rapidly swelled and released cytochrome c and apoptosis-inducing factor. Other phospholipid oxidation products induced swelling, but HAzPC was the most effective and was twice as effective as its diacyl homolog. Cytoplasmic cytochrome c completes the apoptosome, and activated caspase 9 and 3 were present in cells exposed to HAzPC. Irreversible inhibition of caspase 9 blocked downstream caspase 3 activation and prevented apoptosis. Mitochondrial damage initiated this apoptotic cascade, because overexpression of Bcl-XL, an anti-apoptotic protein localized to mitochondria, blocked cytochrome c escape and apoptosis. Thus, exogenous phospholipid oxidation products target intracellular mitochondria to activate the intrinsic apoptotic cascade.

Vascular cells are exposed to exogenous phospholipid oxidation products in the circulation but particularly so in the concentrated environment of atherosclerotic lesions (1). Advanced lesions contain extracellular deposits of oxidized lipids (2)(3)(4) formed by oxidation of lipoprotein particles trapped in the vas-cular matrix (5). These lesions contain apoptotic cells (6), and the apoptotic process plays a critical role in atherogenesis (7)(8)(9).
A significant portion of the cytotoxic material generated during lipoprotein oxidation is oxidized phospholipid, because phospholipase A 2 treatment decreases the toxicity of oxidized LDL 2 (10). Conversely, inhibition of LDL-associated PAF acetylhydrolase with an irreversible inhibitor increases the toxicity of oxidized LDL (11). Plasma PAF acetylhydrolase hydrolyzes oxidized phospholipids with sn-2 fragments up to nine carbon atoms long, including the common azelaoyl fragment, yet, unlike most phospholipases A 2 cannot hydrolyze intact, long chain fatty acyl residues (12). This implies that oxidatively modified phospholipids are abundant cytotoxic agents of oxidized LDL and that neither unmodified phospholipids, the fragmented fatty acids themselves, nor the lysolipid backbone constitute the toxic lipids of oxidized lipoprotein particles.
The majority of the fatty acyl fragments that remain esterified to the phosphatidylcholine backbone after oxidative fragmentation of LDL phospholipids are 9-carbon-long azelaoyl fragments (13). Azelaic acid, a 9-carbon di-acid, is formed by oxidative fragmentation of the 9,10-double bond (14) of the most abundant unsaturated (oleoyl) and polyunsaturated (linoleoyl and linolenoyl) fatty acyl residues of LDL. Accordingly, these azelaoyl phosphatidylcholines (AzPCs) account for almost two-thirds of the oxidized phospholipid in oxidized LDL (13). Indeed, octadecyl-AzPC has long been identified as the cytotoxic oxidation product of oxidant-stressed erythrocytes (15), but whether this phospholipid is toxic to nucleated cells is unknown.
The identity of specific cytotoxic lipids is not well defined, but their cellular targets are even less apparent. Some phospholipid oxidation products activate the PAF receptor displayed on plasma membranes of inflammatory cells (16) to promote inflammation, whereas others interact with the scavenger receptor CD36 (17). A third class of phospholipid oxidation products interact with a phosphatidylinositol-linked membrane protein (18) that may be distinct from CD14 (19).
The mechanism of cell death following exposure to oxidized phospholipids is also incompletely characterized. Cells may die * This work was supported by Grant HL44513 from the National Institutes of by necrosis or by a regulated pathway of apoptosis. Mitochondria play a critical role in the intrinsic pathway to apoptotic death where release of cytochrome c from the mitochondrial inter-membrane space to the cytoplasm allows it to associate with the structural protein apaf-1 and pro-caspase 9 to form the apoptosome (20). This apoptotic machine proteolytically activates caspase 9, which then cleaves and activates the executioner caspase 3 (20). Other proteins released subsequent to the loss of mitochondrial compartmentalization include apoptosisinducing factor (AIF), which is then free to move to the nucleus to participate in the DNA fragmentation characteristic of apoptotic cell death (21). Oxidant-induced cell death is suppressed by the mitochondria-targeted Bcl-2 family member Bcl-X L (22) that prevents mitochondrial permeabilization and release of pro-apoptotic proteins (23). Mitochondrial dysfunction subsequent to oxidative stress plays a primary role in atherogenesis (24).
Here we used synthetic hexadecyl-AzPC (HAzPC), to understand how this abundant extracellular oxidized phospholipid initiates cell death. We find this lipid is toxic in a receptorindependent way: it is rapidly internalized, damages mitochondria, and initiates the intrinsic apoptotic caspase cascade.

Mitochondria
Minced liver from an adult Sprague-Dawley rat in a protocol approved by Cleveland Clinic IACUC was homogenized (0.1 g/ml) in EB buffer (200 mM D-mannitol, 70 mM sucrose, 20 mM Hepes, pH 7.4, 0.5 mg/ml defatted bovine serum albumin, and 1 mM EGTA). Homogenates were cleared twice (1000 ϫ g, 5 min) before recovery of the mitochondria (9500 ϫ g, 10 min). This pellet was washed once with an equal volume of bovine serum albumin-free media and then resuspended (35 mg/ml) in medium without bovine serum albumin. HL60 cells (3 ϫ 10 8 ) stably transfected with Bcl-X L or its empty vector were washed 2ϫ in phosphate-buffered saline and once in EB media before the cells were suspended in 10 ml of EB medium containing bovine serum albumin (2 mg/ml), and then mechanically homogenized and mitochondria purified as above.
Caspase 9 Activation-Cells were treated (4 h) with HAzPC or buffer and then FAM-LEHD-fmk for 1 h before washing and single color flow cytometry.
MitoTracker Red-HUVEC medium was changed to a protein-free medium, and the cells were treated or not with 5 M HAzPC for 30 min at 37°C before 25 nM MitoTracker Red was added to the medium. After 30 min, the cells were washed thrice and then fixed and permeabilized with BD Cytofix/Cytoperm (BD Biosciences, San Jose, CA) for 30 min at 4°C. The cells were then washed with phosphate-buffered saline thrice and mounted with medium containing DAPI.

HAzPC Internalization
HL60 cells were treated in suspension with 5 M HAzPC for the stated times and recovered by centrifugation. The cells were washed once with RPMI containing 0.5% human serum albumin and once with RPMI. [ 2 H]PAF was added as an internal standard, and the lipids were extracted (25) and analyzed by liquid chromatography/electrospray ionization/tandem mass spectrometry (Quattro Ultima, Micromass, Wythenshawe, UK). Lipids in methanol were separated with a Prodigy ODS C18 high-performance liquid chromatography column (150 ϫ 2 mm, 5 m, Phenomenex, Torrance, CA). The solvent gradient started at 85% methanol (0.2 ml/min), then a linear gradient to 100% methanol over 5 min was applied and held for 20 min. The mobile phase was linearly changed back to 85% MeOH over 0.5 min and held for 4.5 min. Mobile phase solvents contained 0.2% formic acid. Source temperature was 120°C, desolvation temperature was 250°C, N 2 flow was 735 liters/h, and the cone N 2 flow was 71 liters/h. Argon was used for collisioninduced dissociation. The multiplier was set at an absolute value of 500 V, total ion current was obtained over m/z 200 -1000 using a cone energy of 50 V in the positive ion mode, and 5.0 kV was applied to the electrospray capillary. Multiple reaction monitoring used 22 eV of collision energy, and the transition used to identify HAzPC were the molecular cation [MϩH ϩ ] of m/z 652 and the daughter ion of m/z 184.

An Exogenous Oxidized Phospholipid Caused Apoptotic Cell
Death-We treated the promyelocytic cell line HL60 with increasing concentrations of synthetic HAzPC for 6 h and then tested for an early marker of apoptotic cell death, exposure of phosphatidylserine on the exterior leaflet of the plasma membrane. HAzPC caused a concentration-dependent increase in annexin V staining of exposed phosphatidylserine, where 2.5 M HAzPC increased by 5-fold the number of cells stained by annexin V (Fig. 1A). Doubling the amount of exogenous HAzPC to 5 M produced a further increase in the number of cells displaying phosphatidylserine on their surface. HAzPC also increased the number of cells unable to exclude the nuclear dye propidium iodide, but at this early time few cells had progressed to become doubly positive (not shown). The liver cell line HepG2 (Fig. 1B) and human umbilical endothelial cells (not shown) also lost plasma membrane phosphatidylserine asymmetry when exposed to increasing concentrations of HAzPC, so the effect of HAzPC was not specific to HL60 cells.
We determined whether HAzPC caused a lethal increase in intracellular free Ca 2ϩ using the fluorescent dye FURA2. We found that, although the cells retained the dye, and therefore remained intact, there was only a small transient increase in Ca 2ϩ (not shown), suggesting the cells did not die by lysis or necrosis.
Exogenous HAzPC Is Internalized and Associates with Mitochondria as an Intact Phospholipid-We added HAzPC to the medium of HL60 cells, recovered the cells at varied times by FIGURE 1. The oxidized phospholipid HAzPC induces apoptosis. A, HL60 cells were incubated with the stated concentrations of synthetic HAzPC for 6 h, or exposed to 1 milliwatt/cm 2 UVB (not shown) for 5 min as a positive control for apoptosis. The cells were stained for surface phosphatidylserine with fluorescent Alexa Fluor 488 -annexin V as described under "Experimental Procedures." Flow cytometry for the intensity of annexin V fluorescence on the x-axis versus cell number was normalized to the maximal number of cells in a channel to overlay the plots. This experiment is representative of three independent experiments. B, HepG2 cells were exposed to HAzPC at the stated concentrations before the level of Annexin V staining determined as in panel A. centrifugation, and washed them once with albumin. HL60 cells rapidly internalized the added HAzPC, with maximal accumulation by ϳ30 min (Fig. 2A). Apoptosis can be initiated through mitochondrial damage, so we determined whether mitochondria were exposed to the internalized oxidized phospholipid. We treated HL60 cells with HAzPC, washed them once with albumin and once with buffer, and then homogenized the cells and fractionated the homogenate by differential centrifugation. We found that the mitochondrial fraction was enriched with HAzPC compared with other cellular membranes and cytosol (Fig. 2B). At least a portion of the internalized HAzPC trafficked to mitochondria as the intact molecule, because the mass spectrometer monitored the molecular ion.
We next determined whether internalized HAzPC was responsible for the cytotoxicity of the oxidized phospholipid. To do this, we stably expressed PAF acetylhydrolase (PLA2g7) in these cells, because this enzyme hydrolyzes PAF and oxida-FIGURE 2. HAzPC internalized by HL60 cells is cytotoxic. A, time-dependent internalization of HAzPC. HAzPC (5 M) was added to suspended HL60 cells, and the cells recovered by centrifugation at the stated times. The cells were washed once with 0.5% human serum albumin, a [ 2 H]PAF internal standard was added, and the lipids were extracted and purified by reversed phase high-performance liquid chromatography. HAzPC was determined by tandem mass spectrometry as the molecular cation [MϩH ϩ ] of m/z 652 producing a phosphocholine daughter ion of m/z 184. The average of duplicate values is shown, and this experiment is representative of three experiments. B, exogenous HAzPC accumulates in mitochondria. HL60 cells (7 ϫ 10 8 ) were incubated with or without 5 M HAzPC for 20 min at room temperature. The cells were washed once in RPMI containing 0.5% human serum albumin and once in RPMI before the cells were suspended in 10 ml of EB medium (200 mM D-mannitol, 70 mM sucrose, 20 mM Hepes (pH 7.4), 1 mM EGTA, and 100 M Pefabloc) and then mechanically homogenized. Homogenates were centrifuged (2000 ϫ g, 5 min) to remove unbroken cells. The resulting lysates were centrifuged (9500 ϫ g, 10 min), and the mitochondrial pellets were resuspended in the same volume of buffer. Lipids were extracted from these duplicate samples and the content of HAzPC determined by high-performance liquid chromatography-tandem mass spectrometry as described under "Experimental Procedures." A separate experiment showed the mitochondrial fraction contained no endoplasmic reticulum-associated calnexin by Western blotting, and that the mitochondrial recovery, visualized by blotting adenine nucleotide translocase, was incomplete. Therefore, the relative amount of HAzPC recovered in the mitochondrial fraction is an underestimate. C, HL60 cells expressing PAF acetylhydrolase are protected from HAzPC toxicity. HL60 cells stably expressing the oxidized phospholipid phospholipase PLA2g7 (plasma PAF acetylhydrolase) or its vector control were treated with 5 M HAzPC for 6 h and stained with Alexa Fluor 488 -annexin V as in Fig. 1. tively damaged phospholipids, including HAzPC, without attacking unmodified membrane phospholipids. Cells expressing PLA2g7 became almost fully resistant to the toxic effect of HAzPC (Fig. 2C).
Intracellular Mitochondria Are Damaged by Exogenous HAzPC-We considered that the unusual structure of HAzPC potentially distorts membrane bilayer structure and that one form of apoptosis results from altered mitochondrial membrane barrier function that allows pro-apoptotic proteins to escape. To define mitochondrial integrity after exposure to HAzPC, we stained adherent HUVECs with MitoTracker Red. We found extensive accumulation of the cationic dye by mitochondria that produced a bright, localized staining pattern (Fig.  3). In contrast, MitoTracker RED was diffusely present in the cytoplasm of cells exposed to HAzPC, suggesting the mitochondrial transmembrane potential that drives the cationic dye into the matrix may have been dissipated by HAzPC exposure.
AIF is an integral mitochondrial inner membrane protein, but proteolysis of the membrane anchor, coupled with outer membrane damage, allows the protein to escape this confine. A portion of this released AIF moves to the nucleus to participate in the DNA fragmentation characteristic of apoptotic cells (26). We used adherent HepG2 (Fig. 4A) and HUVECs (Fig. 4B) to image AIF to find that AIF was localized in punctate, perinuclear structures in control cells, but that it had a diffuse distribution in cells exposed to HAzPC. Staining of the nuclei with DAPI showed that a portion of the AIF was now associated with the nucleus of HAzPC-treated cells. Structurally compromised mitochondria also allow cytochrome c to escape its association with the inner membrane, and we observed that immunoreactive cytochrome c had leaked from the mitochondrial compartment into the cytoplasm in HL60 (Fig. 4C) and HepG2 (Fig. 4D) cells exposed to HAzPC.
Isolated Mitochondria Swell When Exposed to HAzPC-The inner mitochondrial membrane supports a chemiosmotic gradient and mitochondria swell when this permeability barrier is breached. Rat liver mitochondria maximally increased their volume when treated with a high concentration of Ca 2ϩ and the poreforming peptide alamethicin (Fig.  5A). Isolated mitochondria swelled to 70% of this maximal change when treated with HAzPC. Depolarization of mitochondria alone was insufficient to cause swelling, because the protonophore carbonylcyanide 3-chlorophenylhydrazone had little effect on mitochondrial volume. The homologous phospholipid palmitoylglutaroyl (a 5-carbon di-acid) PC, formed from oxidative fragmentation of palmitoylarachidonoyl-PC, caused mitochondria to increase their volume to 50% of that of the positive control. Other short chain phospholipids, and lysophosphatidic acid, were less effective in this regard and were about as effective as the underivitized lysophosphatidylcholine in changing mitochondrial volume. We found (Fig. 5B) that the acyl form of HAzPC, palmitoyl azelaoyl-PC, induced swelling, but was about half as effective as the ether phospholipid. A minor portion of this swelling was blocked by cyclosporin A, a small molecule that binds to cyclophilin D of the mitochondrial permeability transition pore and keeps in it a closed position (Fig. 5C).
We examined the effect of HAzPC on the physical structure of mitochondria by electron microscopy. The photomicrographs we obtained (Fig. 5D) show this normally compact and electron dense organelle was affected by lysophosphatidylcholine and to a greater extent by HAzPC. Mitochondria treated with lysophosphatidylcholine were enlarged and less opaque, whereas those treated with HAzPC were greatly enlarged with transparent domains. The positive control Ca 2ϩ and alamethicin showed the complete disruption of the organelle.
Cytochrome c and AIF Escape from Isolated Mitochondria Exposed to HAzPC-The two pro-apoptotic proteins cytochrome c and AIF lost their exclusive mitochondrial localization in cells exposed to HAzPC, and a similar loss of mitochondrial retention occurred when isolated mitochondria were exposed to HAzPC (Fig. 6). We found by immunoblotting that HAzPC induced a concentration-dependent loss of cytochrome c from isolated mitochondrial to their supernatant and that cyclosporin A partially reduced the loss of cytochrome c from the mitochondrial pellet at lower HAzPC concentrations. Mitochondria exposed to HAzPC also lost AIF to their supernatant and again cyclosporin A partially suppressed this release.
HL60 Cells Exposed to HAzPC Contain Activated Caspase 9 and Caspase 3-Cytoplasmic cytochrome c is free to associate with cytoplasmic apaf-1 and pro-caspase 9 to form an active pro-caspase 9 cleavage complex, the apoptosome. We tested whether the cytochrome c released to the cytoplasm of HL60 cells exposed to HAzPC promoted the formation of active caspase 9, an upstream initiator caspase of the intrinsic apoptotic pathway. We found by flow cytometry (Figs. 7, A and B) that few control cells contained active caspase 9 but that a significant portion of HAzPC-treated cells contained this active caspase and hydrolyzed its fluorogenic z-LEHD-fmk substrate to a greater extent than our positive control, UVB irradiation. Caspase 3 cleavage is catalyzed by activated caspase 9, which then allows caspase 3 to function as a downstream executioner caspase. We probed HL60 cells, treated or not with a range of HAzPC concentrations, for proteolytically activated caspase 3 by immunoblotting, and found a concentration-dependent increase in 17-and 19-kDa caspase 3 fragments (Fig. 7C). The specific caspase 9 inhibitor z-LEHD-fmk blocked formation of the 17-and 19-kDa caspase 3 fragments (not shown), and it prevented the formation of active caspase 3 in HAzPC-treated HL60 cells (Fig. 7D). Caspase 3 activation in response to HAzPC therefore depends on active caspase 9.
Caspase Inhibition Suppresses HAzPC-induced Cell Death-The oxidized phospholipid HAzPC activated components of the intrinsic apoptotic cascade, so we next determined whether these activated proteases were required for cell death. We first inhibited caspase function with the general caspase inhibitor z-VAD-fmk, and then with the specific caspase 9 inhibitor z-LEHD-fmk. We found that 46% of apoptotic cells with surface-exposed phosphatidylserine after HAzPC treatment was reduced to ϳ16% by each caspase inhibitor (Fig. 8). We conclude that HAzPC is cytotoxic, because it activates proapoptotic caspases.
Mitochondria Are Primary Targets of HAzPC-induced Apoptosis-HAzPC damages mitochondrial integrity in intact cells and initiates the intrinsic caspase cascade leading to apoptosis, but is this mitochondrial damage responsible for cell death? To investigate this relationship we used Bcl-X L , an antiapoptotic Bcl-2 homolog that localizes to mitochondria where it interferes with pro-apoptotic family members and suppresses mitochondrial permeabilization (27). We stably overexpressed FLAG-tagged Bcl-X L in HL60 cells (Fig. 9A) and isolated mitochondria from these cells. Mitochondria from Bcl-X L -expressing cells were significantly less sensitive to HAzPC than control cells and allowed little cytochrome c to escape after HAzPC exposure (Fig. 9B). Bcl-X L overexpression was similarly efficacious in vivo where it blocked annexin V staining of HL-60 cells exposed to HAzPC (Fig. 9C). HAzPC therefore interacts with mitochondria to initiate an apoptotic cascade.

DISCUSSION
We show that an exogenous oxidized phospholipid is cytotoxic, because it is readily internalized, migrates to mito- C, cyclosporin A minimally affects HAzPC-induced swelling. Isolated mitochondria were pre-treated with 680 nM cyclosporin A at room temperature to interfere with flow through the mitochondrial permeability transition pore before swelling in response to 2.5 M HAzPC was determined as before. D, transmission electron microscopy. Electron micrographs of rat liver mitochondria treated with 5 M lysophosphatidylcholine or HAzPC, or exposed to 5 mM CaCl 2 and 7 g/ml alamethicin for 10 min. The bar is 1 m in length. FIGURE 6. Pro-apoptotic proteins are released from mitochondria exposed to HAzPC. Isolated mitochondria were treated or not with cyclosporin A for 5 min and then with the stated concentration of HAzPC for 10 min. Mitochondria were then separated from soluble material by centrifugation, and the amount of AIF and cytochrome c in the supernatant was determined by Western blotting as defined under "Experimental Procedures." The center lane, labeled "control," was the amount of the two target proteins in detergent-solubilized mitochondria at the start of the experiment.
chondria, and then damages mitochondria in a way that activates the intrinsic caspase cascade. The sequence of events subsequent to internalization of the oxidation product included association with mitochondria, loss of mitochondrial sequestration of pro-apoptotic proteins, formation of the apoptosome with caspase 9 activation, and then caspase 3 activation. Activation of these caspases was required for HAzPC-induced death, because irreversibly inhibiting caspase 9 suppressed cell death. Cellular mitochondria were the targets of the internalized HAzPC, because overexpression of Bcl-X L suppressed cell death. The phospholipid itself was the proapoptotic agent, because hydrolysis by PAF acetylhydrolase suppressed apoptosis.
We tested several cell lines, and circulating human cells (not shown), and found that each was susceptible to HAzPC-induced cell death, although the concentration of HAzPC required to induce death of these cells varied severalfold, perhaps reflecting varied rates of internalization or hydrolysis. However, for each cell type HAzPC at low micromolar concentrations induced phosphatidylserine exposure, cytoplasmic accumulation of cytochrome c, and redistribution of AIF to the cytoplasm and nucleus.
AIF is released from its association with the inner mitochondrial membrane by apoptotic stimuli (26) allowing it to migrate to the nucleus to cause nuclear condensation and cell death (28). Cytosolic cytochrome c is an essential component of the apoptosome (29) that cleaves pro-caspase 9 to its active form. The peptide z-LEHD-fmk is a selective caspase 9 substrate and irreversible inhibitor that prevents cleavage of down stream targets such as caspase 3. Cells treated with HAzPC contained activated caspase 9, identified by fluorescent adduct derived from a fluorogenic irreversible inhibitor, and the cells contained active caspase 3. Activation of the executioner caspase 3 depended on caspase 9 cleavage, because proteolytically activated caspase 3 fragments did not appear in cells containing irreversibly inactivated caspase 9. This inhibition of caspase activity also suppressed HAzPC-induced cell death, so HAzPC initiates death only via the apoptotic intrinsic caspase cascade.
The C 18 homolog of C 16 HAzPC, octadecylazelaoyl-PC, previously has been identified as a lytic agent present in oxidized egg yolk phospholipid when erythrocytes were the target cells (15). However, we find in nucleated cells that contain mitochondria that the primary effect of exogenous HAzPC was not lysis. The cells remained morphologically unremarkable, and their plasma membrane presented a functional permeability barrier that retained cytoplasmic dyes. We did not observe (not shown) a prolonged influx of extracellular Ca 2ϩ , as found in human umbilical vein endothelial cells exposed to lysophosphatidylcholine (30), or non-selective disruption of the plasma FIGURE 7. HAzPC activates caspases of the intrinsic apoptotic cascade. A, intracellular activated caspase 9. HL60 cells were exposed to 5 M HAzPC for 4 h as before and then incubated with the fluorescence substrate and pro-irreversible caspase 9 inhibitor FAM-LEHD-fmk. The cells were cultured at 37°C for an additional hour and washed, and the amount of caspase 9-bound dye was determined by single channel flow cytometry. B, the bar graph presents the average and standard error of three independent experiments. C, active caspase 3 fragments accumulate in cells exposed to HAzPC. HL60 cells were exposed to the stated concentration of HAzPC for 4 h, lysed before the presence of 19-and 17-kDa caspase 3 fragments was assessed by Western blotting. D, HAzPC induces caspase 3 enzymatic activity in a caspase 9-dependent fashion. HL60 cells were treated for 1 h with buffer or the pan caspase inhibitor z-VAD-fmk or z-LEHD-fmk to inhibit caspase 9 and then incubated with buffer or 5 M HAzPC for 4 h before cell lysates were prepared and assayed for caspase 3 activity. Each panel, including panel B, is representative of three independent experiments. membrane as suggested for other short chain phospholipids (31). Instead, we identified a marked change in cellular mitochondrial function in the absence of frank plasma membrane damage.
We conclude that the oxidized phospholipid HAzPC is proapoptotic because of its effect on intracellular mitochondria for several reasons. First, there is a temporal correlation between early mitochondrial dysfunction (3-4 h) and subsequent phosphatidylserine exposure (staring at 6 h). Second, internalized HAzPC was concentrated as an intact phospholipid by mitochondria. Third, Bcl-X L , which physically interacts with the mitochondrial outer membrane to promote cell survival (32), suppressed HAzPC-induced cell death. Mitochondria isolated from cells overexpressing Bcl-X L released little cytochrome c, which, in turn, resulted in fewer cells expressing phosphatidylserine on their surface after being exposed to HAzPC. Because specifically protecting mitochondria with ectopic Bcl-X L blocked HAzPC cytotoxicity, the primary targets of internalized HAzPC for this event are mitochondria. We currently do not know whether mitochondria are particularly sensitive to HAzPC, whether compromising their integrity is more apparent than with other organelles, or whether the high concentration of HAzPC encountered by mitochondria account for the mitotoxicity. It is apparent, however, that this lipid does not act as an indiscriminant detergent.
As anticipated from experiments with intact cells, HAzPC had a direct effect on isolated mitochondria and rapidly induced swelling in a concentration-dependent fashion. HAzPC, with an sn-1-ether bond and a 9-carbon di-acidic fragment derived from linoleoyl/linolenoyl oxidation, was more effective in this than its palmitoyl homolog, or the 5-carbon glutaroyl analog derived from arachidonate oxidation. It also was more effective than the shorter arachidonoyl fragmentation products containing 4-carbon butyroyl residue, or PAF that contains the two-carbon acetyl residue. The phospholipase A 2 hydrolytic product of HAzPC, 1-hexadecyl-sn-glycero-3phosphocholine (lyso-PAF), was only mildly effective in altering mitochondrial volume. This means that catabolism of this particular oxidized phospholipid by phospholipase A 2 activity reduces its mitotoxicity. This relationship for the relative effectiveness of precursor and hydrolytic product does not hold for the other less potent oxidized phospholipids. A recent report (33) shows that a calcium-independent phospholipase A 2 localizes to mitochondria and protects cells from oxidative insult and apoptotic death, suggesting phospholipid products such as HAzPC generated in vivo are more toxic than their hydrolyzed components.
Components of oxidized LDL have previously been found to promote mitochondrial dysfunction (34) and caspase 3 activation (35,36), although the relevant mechanisms remain incompletely defined. The oxidized LDL components palmitoylglutaroyl-PC and palmitoyloxovaleroyl-PC (oxidative fragments of palmitoylarachidonoyl-PC (4, 12) that are 4 carbons shorter than HAzPC) also enhance caspase 3 activity, although it requires 50 M of these shorter chained oxidized phospholipids to achieve a modest increase in caspase 3 activity (37), or a 50% decrease in viability (38). This is a larger concentration difference than their effects on mitochondrial swelling, but internalization and intracellular distribution may also differ for individual oxidized phospholipids (39). Reactive non-phospholipid oxidation products, such as acrolein (40,41), 4-hydroxhexenal (42), or 15-deoxy-⌬ 12,14 -prostaglandin J 2 (43), depolarize isolated mitochondria and induce apoptosis. However, these chemically reactive lipids react with many targets and may not accumulate to a level able to affect mitochondrial function, nor are they the cytotoxic agents destroyed by phospholipase A 2 (10) or PAF acetylhydrolase (44 -48) digestion.
Oxidized LDL levels in carotid plaque are some 70-fold greater than circulating levels (49), and accumulation of apoptotic macrophages and smooth muscle cells parallels the progression of atherosclerosis (6). An important component of the toxic soup of atherosclerotic plaque is oxidized phospholipid, because phospholipid hydrolysis reduces oxidized LDL cytotoxicity, at least to endothelial cells (10). Linoleoyl (C 18:2 ) and linolenoyl (C 18:3 ) are the most abundant esterified polyunsaturated fatty acids, and both contain a 9,10-double bond. Preferential bond scission at this site (14) creates phospholipids with a 9-carbon sn-2 fragment that contain an aldehyde, which also is pro-apoptotic (50), or the more oxidized azelaoyl residue. The azelaoyl residue is not a product of fatty acid ␤-oxidation and has long served as a marker of chemical oxidation. Here we show that phospholipids, with this shortened fatty acyl residue, are not only markers of oxidative processes but they also FIGURE 9. Mitochondrial protection by Bcl-X L blocks HAzPC-induced apoptosis. A, Bcl-X L expression. HL60 cells stably overexpressing human Bcl-X L were immunoblotted for total cellular Bcl-X L and actin. B, cytochrome c loss from isolated mitochondria. Mitochondria were isolated from Bcl-X L -overexpressing cells, treated or not with HAzPC, and loss of cytochrome c to the supernatant was determined as in Fig. 6. C, Bcl-X L expression suppresses HAzPC-induced apoptosis. Surface display of phosphatidylserine was quantitated by flow cytometry using Alexa 488 -conjugated annexin V as in Fig. 1. Each panel is representative of three independent experiments. This is an arm of the experiment in Fig. 2B, so the buffer and HAzPC histograms are the same.
directly participate in cell death by targeting mitochondria to initiate the intrinsic apoptotic caspase cascade.