Apoptosis and Activation of the Sphingomyelin-Ceramide Pathway Induced by Oxidized Low Density Lipoproteins Are Not Causally Related in ECV-304 Endothelial Cells*

Oxidized low density lipoproteins (oxLDL) are thought to play a central role in the development of atherosclerosis. Toxic concentrations of mildly oxidized LDL elicit massive apoptosis of endothelial cells (Escargueil-Blanc, I., Meilhac, O., Pieraggi, M. T., Arnal J. F., Salvayre, R., Nègre-Salvayre, A. (1997)Arterioscler. Thromb. Vasc.Biol. 17, 331–339). Since the lipid mediator ceramide emerged as a potent inducer of apoptosis, we aimed at investigating the occurrence of ceramide formation and its potential role in oxLDL-induced apoptosis. In ECV-304 endothelial cells, toxic concentrations of oxLDL triggered an early activation of the sphingomyelin-ceramide pathway, as shown by both sphingomyelin hydrolysis and ceramide formation.N-Tosyl-l-phenylalanyl chloromethyl ketone (TPCK) and dichloroisocoumarin (DCIC), two serine-protease inhibitors (serpins), blocked the oxLDL-induced ceramide generation but, unexpectedly, did not inhibit the oxLDL-induced apoptosis. Conversely, treatment of endothelial cells by bacterial sphingomyelinase, under conditions effectively generating ceramide, did not induce apoptosis. In contrast, short-chain permeant C2- and C6-ceramides elicited apoptosis of ECV-304. However, the mechanisms of apoptosis triggered by C2-ceramide and by oxLDL were (at least in part) different, because C2-ceramide-induced apoptosis was calcium-independent, whereas oxLDL-induced apoptosis was calcium-dependent. In conclusion, it is suggested that oxLDL-induced apoptosis is calcium-dependent but independent of the activation of the sphingomyelin-ceramide pathway and that the toxic effect of short chain permeant ceramides is calcium-independent and does not mimic the effect of natural ceramides induced by oxLDL.

In ECV-304 endothelial cells, toxic concentrations of oxLDL triggered an early activation of the sphingomyelin-ceramide pathway, as shown by both sphingomyelin hydrolysis and ceramide formation. N-Tosyl-L-phenylalanyl chloromethyl ketone (TPCK) and dichloroisocoumarin (DCIC), two serine-protease inhibitors (serpins), blocked the oxLDL-induced ceramide generation but, unexpectedly, did not inhibit the oxLDL-induced apoptosis. Conversely, treatment of endothelial cells by bacterial sphingomyelinase, under conditions effectively generating ceramide, did not induce apoptosis. In contrast, short-chain permeant C 2 -and C 6 -ceramides elicited apoptosis of ECV-304. However, the mechanisms of apoptosis triggered by C 2 -ceramide and by oxLDL were (at least in part) different, because C 2 -ceramide-induced apoptosis was calcium-independent, whereas oxLDL-induced apoptosis was calcium-dependent.
In conclusion, it is suggested that oxLDL-induced apoptosis is calcium-dependent but independent of the activation of the sphingomyelin-ceramide pathway and that the toxic effect of short chain permeant ceramides is calcium-independent and does not mimic the effect of natural ceramides induced by oxLDL.
Low density lipoproteins (LDL) 1 play a physiological role in delivering cholesterol to peripheral cells (1). After undergoing oxidative modifications, LDL are involved in the genesis of atherosclerosis (2)(3)(4), one of the most prevalent causes of morbidity and mortality in Western countries. LDL oxidation is mediated in vitro by various types of cultured vascular cells (2)(3)(4) and is thought to occur in vivo in the subendothelial area, as suggested by the presence of oxidized LDL in atherosclerotic areas (5). Oxidized LDL (oxLDL) exhibit a wide variety of biological properties, such as formation of foam cells and fatty streaks, induction of gene expression, alterations of coagulation pathways, and arterial vasomotor properties (reviewed in Ref. 4). In addition, oxidized LDL induce a dramatic cytotoxic effect in cultured cells (6 -8). The morphological changes of cultured endothelial cells treated by oxidized LDL (9, 10) exhibit some similarities with those of the endothelial cover of atherosclerotic areas (11). We have recently reported that ox-LDL elicit apoptosis of cultured endothelial cells through a calcium-dependent process (12).
Toxic cell injury induces a complex sequence of events leading to cell death (13). Two types of cell death, termed necrosis and apoptosis, have been discriminated on the basis of morphological studies (14). Necrosis is characterized by cellular swelling, organelle alterations, rupture of plasma membrane, and finally cell lysis and leakage of the cellular components (14). Apoptosis, or programmed cell death, is characterized by DNA fragmentation, alterations of nucleus morphology (chromatin condensation and nucleus fragmentation), organelle relocalization, and cell fragmentation without leakage of cytosolic macromolecules (14 -16).
In addition to the above mentioned proteolytic pathways, various other intracellular signals may trigger apoptosis (17). The lipid second messenger ceramide has recently emerged as a prominent mediator implicated in apoptosis (22) and other cellular responses (23,24), such as cell proliferation and cell cycle arrest (23)(24)(25). Ceramide is generated through activation of the sphingomyelin-ceramide pathway, i.e. sphingomyelin (SM) hydrolysis mediated by signal-regulated sphingomyelinases (SMases) (23)(24)(25), or through ceramide biosynthesis (26). Several targets of ceramide have been identified, namely protein kinase C, a proline-directed protein kinase, a serine/ threonine protein phosphatase, and mitogen-activated protein and SAP kinases (23)(24)(25). Despite a large literature on the subject, the precise molecular mechanism of action of ceramide in apoptosis remains largely hypothetical.
OxLDL are able to induce apoptosis of lymphoid and endothelial cells (12,27) and to activate the SM-ceramide pathway in vascular smooth muscle cells (28). Because ceramide may be involved in apoptotic signaling, we hypothesized that oxLDL may trigger activation of the SM-ceramide pathway that, in turn, may be involved in oxLDL-induced apoptosis of cultured endothelial cells.
The reported data show that, in cultured endothelial cells, oxLDL induce an early activation of SM hydrolysis and ceramide formation, but, quite unexpectedly, our results suggest that the toxic effect of oxLDL is independent of the endogenously generated ceramide.
Cell Culture-The human umbilical vein endothelial cell line ECV-304 (CRL-1998) was obtained from the ATCC (Rockville, MD) and grown under the previously used conditions (12). All passages were made using a splitting ratio 1/4. Under the standard conditions, endothelial cells (0.4 ϫ 10 6 cells/ml) were seeded in 6-multiwell plates or in flasks (Nunc) when required and grown in their respective medium with Glutamax TM supplemented with 10% heat-inactivated fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were incubated in a humidified incubator (5% CO 2 , 37°C). 24 h before LDL incorporation, this medium was replaced by a serum-free medium.
LDL Isolation and Oxidation-LDL from human pooled heat-inactivated (1 h at 56°C) sera were isolated by ultracentrifugation according to Havel et al. (30), dialyzed against phosphate-buffered saline (PBS) containing 0.1 mmol/liter EDTA, sterilized on 0.2-m Millipore membrane, and stored at 4°C under nitrogen (up to 3 weeks). ApoB was determined by immunonephelometry.
Mildly oxidized LDL were obtained by (UV ϩ copper/EDTA)-mediated oxidation under mild conditions: LDL solution (2 mg of apoB/ml, containing 2 mol/liter CuSO 4 ) was irradiated for 2 h, as a thin film (5 mm) in an open beaker placed 10 cm under the UV-C source (HNS 30W OFR Osram UV-C tube, max 254 nm, 0.5 milliwatt/cm 2 determined using a Scientech thermopile Model 360001), under the standard conditions previously utilized (12,27). At the end of the irradiation, aliquots were taken up for analyses and oxidized LDL (200 g of apoB/ml under standard conditions or at the indicated concentration) were immediately incorporated in the culture medium.
The level of LDL oxidation was evaluated by monitoring the formation of lipid hydroperoxides, using the FOX-2 procedure of Wolff (31), and thiobarbituric acid-reactive substances (TBARS) (32). The relative electrophoretic mobility was evaluated on Hydragel (Sebia, Paris, France), and the level of trinitrobenzenesulfonic acid-reactive amino groups was determined according to Steinbrecher (33).

Determination of Cytosolic Calcium Concentration [Ca 2ϩ ] i -[Ca
] i was determined by employing the permeant calcium probes fura-2/AM, under the previously used conditions (12,27). Briefly, cells were incubated for 30 min at 37°C in RPMI medium buffered with 20 mmol/liter HEPES and containing 0.5% bovine serum albumin and fura-2/AM (2 mol/liter). After dilution and incubation in RPMI for 45 min, cells were washed twice in PBS, and their fluorescence was recorded at the dual excitation wavelength of 340 and 380 nm and emission at 510 nm.
[Ca 2ϩ ] i was calculated according to the ratio method (34).

Determination of [ 3 H]Thymidine
Incorporation-To determine [ 3 H]thymidine incorporation, cells were incubated with the indicated mitogenic agent (native LDL, oxLDL, sphingomyelinase, short-chain ceramides) and [ 3 H]thymidine (0.5 Ci/ml for the indicated time). Then cells were washed three times with PBS, harvested, precipitated by 10% trichloroacetic acid, and centrifuged at 10,000 ϫ g for 10 min. The precipitate was dissolved in 1 M NaOH and 1% SDS overnight and mixed with Aquasafe-300 (BAI, Zinsser) for liquid scintillation counting (Packard Tri-Carb 4530). [ 3 H]Thymidine incorporation was expressed as the percent of that measured in control cells grown in serum-free medium.
Determination of the Cytotoxicity and Evaluation of Necrosis and Apoptosis-The whole cytotoxic effect was evaluated by using the MTT test (35). Necrosis, i.e. loss of the plasma membrane integrity, was evaluated by the leakage of cellular lactate dehydrogenase into the culture medium (Roche MA kit 10 TM ) and by the trypan blue exclusion test, as previously utilized (27).
Apoptotic cells were counted microscopically after fixation of cells (3% paraformaldehyde for 15 min and washing in PBS) and staining of the nucleus by 4Ј,6-diamidino-2-phenylindole, a fluorescent intercalating DNA probe, under the previously used conditions (12,27) or after staining by May-Grü nwald-Giemsa (MGG). Alternatively, morphological changes of the nucleus were detected after staining the nucleus of living cells with 5 mol/liter SYTO-11 (a fluorescent permeant intercalating DNA probe) and immediately examined by fluorescence microscopy (Leica Model Diaplan). As both fluorescent staining of the nucleus and MGG gave similar results, only one set of data (MGG) was reported.
It may be noted that cellular debris were not counted (thus probably excluding late steps of the apoptotic and necrotic processes) DNA fragmentation was also visualized in situ on fixed cells by the TUNEL (terminal transferase-mediated dUTP-biotin nick end labeling) procedure of Gavrielli et al. (36), using the terminal transferase (TdT) kit of Boehringer Mannheim (Meylan, France). Briefly, cells grown on glass coverslides were fixed in 3% buffered paraformaldehyde, and endogenous peroxidases were inactivated by 2% H 2 O 2 . After rinsing, the slides were incubated with 150 l of terminal deoxynucleotidyltransferase (0.3 unit/l) and biotinylated dUTP in terminal deoxynucleotidyltransferase buffer (150 mmol/liter potassium cacodylate, 25 mmol/liter Tris-HCl, pH 6.6, 0.25 mg/ml bovine serum albumin) and 2 mmol/liter CoCl 2 for 1 h at 37°C. Then, after rinsing four times, the slides were covered by 150 l of ExtrAvidin peroxidase (Sigma) diluted 1/15 in water and incubated for 30 min at 37°C, washed twice, and stained with 1 mg/ml diaminobenzidine for 5 min at 37°C. The positive control was treated by DNase I (1 g/ml from Sigma, for 10 min), before being processed through TUNEL procedure.
Determination of Chromatin Fragments-DNA fragmentation assays were essentially derived from the procedure of McConkey et al. (37) under the previously described conditions (27). Cells were allowed to lyse for 15 min in 1 ml of lysis buffer (5 g/liter Triton X-100 and 20 mmol/liter EDTA, 5 mmol/liter Tris, pH 8.0), then ultracentrifuged for 20 min at 27,000 ϫ g to separate the chromatin pellet from cleavage products. The pellet (resuspended in 1 ml of 10 mmol/liter Tris-HCl pH 8.0 buffer containing 1 mmol/liter EDTA) and the supernatant were assayed for DNA by the fluorometric method of Kapuscinski et al. (38).
Quantitation of SM and Ceramide-Cells were metabolically labeled to equilibrium with [methyl-3 H]choline (0.5 Ci/ml) or [ 3 H]palmitic acid (0.5 Ci/ml) in RPMI medium containing 1% fetal calf serum. After a 48-h incubation, cells (about 10 7 /assay) were washed once with PBS and chased for 2 h in RPMI. Then the medium was replaced by fresh medium containing or not containing oxLDL (200 g of apoB/ml). At the indicated times, cells were washed with ice-cold PBS, harvested using a rubber policeman, and sedimented by centrifugation (500 ϫ g for 5 min). Cell pellets were immediately frozen at Ϫ20°C. Cell pellets were suspended in 0.6 ml of distilled water and homogenized by sonication (2 ϫ 10 s, using a MSE probe sonicator). An aliquot was saved for protein determination. Lipids from 0.5 ml of the cell lysate were extracted by 2.5 ml of chloroform/methanol, and [choline-3 H]SM and [palmitoyl-3 H]ceramide were quantified as described previously (28).
Protein concentrations were determined using the procedure of Smith et al. (39). The statistical significance was estimated by Student's t test.

RESULTS
Oxidized LDL Induce Both DNA Synthesis and Apoptosis of ECV-304 Endothelial Cells-Mildly oxidized LDL (oxLDL), obtained by UV ϩ copper/EDTA oxidation, contained moderate levels of lipid peroxidation derivatives (4.2 Ϯ 0.7 nmol of TBARS/mg of apoB and 57 Ϯ 9 nmol of lipid hydroperoxide/mg of apoB) and exhibited only minor alterations of the relative electrophoretic mobility (1.2 Ϯ 0.1 in comparison to the control, i.e. native LDL) and of 2,4,6-trinitrobenzenesulfonic acid-reactive amino groups (92 Ϯ 4% of the control). OxLDL are taken up through the apoB/E receptor pathway (40), in contrast to extensively modified oxLDL (4).
In ECV-304 starved for 24 h in serum-free medium, addition of oxLDL (200 g of apoB/ml) induced an early stimulation (peaking at 8 -10 h) followed by a marked decrease of DNA synthesis, as assessed by monitoring [ 3 H]thymidine incorporation (Fig. 1A). After 10 -12 h of pulse with oxLDL, a relatively rapid decay of MTT index (Fig. 1B) was associated with cellular changes characteristic of apoptosis, as assessed by morphological methods and chromatin fragmentation (Fig. 1, C-E). It may be noted that cell counts of morphologically apoptotic cells stained by MGG or by TUNEL methods gave similar values (in order to avoid redundant data only one set of data is reported). The apoptotic process was also confirmed by DNA laddering (data not shown). During this period, we observed no significant increase of trypan blue staining, thus excluding the occurrence of primary necrosis (data not shown). As reported in Fig.  1, under the conditions used, native (i.e. unoxidized) LDL did not induce any significant mitogenic or toxic effect. It is noteworthy that: 1) toxic concentrations of oxLDL induce massive apoptosis in various endothelial cell lines (either immortalized or nonimmortalized) (12,41); 2) cell-mediated oxLDL were also able to induce apoptosis (data not shown). We used UV ϩ copper oxLDL, because cell-mediated oxLDL preparations may contain growth factors or other mediators secreted by cells, potentially able to trigger intracellular signaling and interfere with the effects of oxLDL.
The mitogenic and apoptotic effects of oxLDL were dose-dependent (Fig. 1, F-H). Under the used conditions, the mitogenic effect of oxLDL began at low concentrations (lower than 100 g of apoB/ml), whereas the apoptotic effect began at a higher concentration (higher than 100 g/ml).
The SM-Ceramide Signaling Pathway Is Activated by oxLDL but Is Not Required for oxLDL-induced Apoptosis-As shown in Fig. 2, toxic concentrations of oxLDL (200 g of apoB/ml) activated the SM-ceramide pathway in ECV-304 metabolically labeled to equilibrium with [methyl-3 H]choline or [ 3 H]palmitic acid. The maximal hydrolysis (25-30% of total labeled SM) was observed within 4 -5 h after addition of oxLDL and returned progressively toward the baseline (Fig. 2, A and B). As shown in Fig. 2A, inset, nontoxic concentrations of oxLDL (75 g of apoB/ml) also induced the activation of the sphingomyelinceramide pathway but at a lower rate (maximal hydrolysis peaked at 12-17% of total labeled SM). In cells prelabeled with [9,10-3 H]palmitic acid, oxLDL induced the formation of [ 3 H]ceramide (concomitantly with SM hydrolysis) (Fig. 2C). In contrast, native unoxidized LDL (used at 200 g of apoB/ml, under similar experimental conditions as oxLDL) induced no detect- able sphingomyelin hydrolysis nor [ 3 H]ceramide generation (Fig. 2, A-C).
In order to investigate whether the activation of the SMceramide pathway is causally related to the apoptotic process triggered by UV-oxLDL, two sets of experiments were performed using SM hydrolysis inhibitors and exogenously generated or added ceramides.
Recent reports showed that TPCK and DCIC, two serine protease inhibitors (serpins), are able: (i) to block the activation of the SM-ceramide pathway stimulated during daunorubicininduced apoptosis (21) and (ii) to inhibit apoptosis induced by various effectors or drugs (19 -21). Under the experimental conditions used here, the two serpins TPCK and DCIC (used at nontoxic concentrations, i.e. 10 and 25 M, respectively) effectively blocked SM hydrolysis, ceramide generation (Fig. 2, B and C), and DNA synthesis (Fig. 2D). But, unexpectedly, the two serpins did not prevent oxLDL-induced apoptosis of ECV-304 (Fig. 2E). This strongly suggests that oxLDL-induced apoptosis is independent of the activation of the SM-ceramide pathway.
The biological effects of ceramides generated by the SMceramide pathway activation are often mimicked by shortchain permeant ceramides or by ceramides generated at the plasma membrane by bacterial SMase treatment. Treatment of ECV-304 by bacterial SMase induced around 30% hydrolysis of cellular labeled SM (Fig. 3A, inset), but elicited no significant mitogenic effect (evaluated by [ 3 H]thymidine incorporation) (Fig. 3A) and no significant increase of apoptosis (as assessed by morphological evaluation and chromatin fragmentation) (Fig. 3B). The short-chain permeant C 2 -and C 6 -ceramides induced no significant mitogenic effect but elicited apoptosis of ECV-304 (Fig. 3, A and B). In Fig. 3C, the toxicity of the newly synthesized thio-C 2 -and thio-C 3 -ceramides was compared with that of C 2 -and C 6 -ceramides. All 4 compounds induced a dosedependent toxicity exhibiting the features of apoptosis. The thio analogs were less toxic than C 2 -and C 6 -ceramides, and in each group, the shorter compounds exhibited a higher toxicity. As expected, the saturated analog C 2 -dihydroceramide (10 mol/liter) induced no DNA synthesis nor apoptosis.
These data clearly show that, in ECV-304, natural ceramides generated at the plasma membrane by bacterial SMase are not mitogenic or apoptotic and do not mimic the biological effect of oxLDL. However, short-chain permeant ceramides triggered no mitogenic response but elicited apoptosis (apparently in contrast with the above conclusion that the SM-ceramide pathway is not required for oxLDL-induced apoptosis).
As these data apparently were in contrast with the above conclusion that the SM-ceramide pathway is not required for oxLDL-induced apoptosis, we further compared the mechanism of apoptosis induced by oxLDL with that induced by shortchain permeant ceramides.
OxLDL-induced Apoptosis Is Calcium-dependent, in Contrast to That Induced by C 2 -Ceramide-OxLDL evoked a delayed and sustained calcium peak (Fig. 4A) that is required for oxLDL-induced toxicity (12,27). In contrast, native unoxidized LDL evoked no significant calcium rise nor toxicity (Fig. 4A). EGTA (used at 0.5 mM, i.e. concentration chosen for buffering extracellular calcium without any toxic effect) inhibited the calcium entry and the sustained calcium peak (Fig. 1B). Under these conditions, the DNA synthesis stimulated by oxLDL was not inhibited by EGTA, while apoptosis was blocked (Fig. 4, C  and D). Conversely, 10 M TPCK, which inhibited effectively both the SM hydrolysis and ceramide generation (Fig. 2B), did not block the oxLDL-induced calcium peak (Fig. 4B) or the apoptotic effect of oxLDL (Fig. 2B). In contrast, apoptosis of ECV-304 induced by 10 M C 2 -ceramide was not preceded by any sustained calcium peak (Fig. 4A) and was not blocked by EGTA (Fig. 4D). These data suggest that the oxLDL-induced apoptosis is calcium-dependent, whereas the C 2 -ceramide-induced apoptosis is calcium-independent. DISCUSSION Although the cytotoxic effect of oxLDL to cultured endothelial cells has been reported a long time ago (7,8), the type of cell death induced by oxLDL has been studied only recently. Ox-LDL induce apoptosis of cultured endothelial cells, but the mechanism involved in the oxLDL-induced apoptotic process is only poorly understood (12).
The present study shows that in ECV-304: (i) oxLDL trigger activation of the SM-ceramide pathway, transient DNA synthesis, and apoptosis (the oxLDL-induced apoptosis being calcium-dependent); (ii) serpins are able to inhibit the activation of the SM-ceramide pathway and DNA synthesis, but are unable to block the calcium peak and apoptosis triggered by oxLDL; (iii) SM hydrolysis and ceramides generated by exogenous bacterial SMase induce neither DNA synthesis nor apoptosis; (iv) short-chain synthetic analogs of ceramide are able to trigger apoptosis through a calcium-independent mechanism (in contrast to oxLDL-induced apoptosis).
Typical apoptosis and necrosis are discriminated on the basis of morphological features (14). ECV-304 apoptotic cells exhibited biochemical (DNA ladder) and morphological alterations of the nucleus (chromatin margination and condensation and nucleus fragmentation) easily visualized by fluorescent DNA staining and electron microscopy (12). In the early steps of oxLDL-induced apoptosis, ECV-304 cells are still attached to the culture flask and do not exhibit any of the characteristic features of primary necrosis (or oncosis, i.e. loss of plasma membrane integrity and staining by trypan blue, associated to swelling and rounding of the cell and of cellular organelles, without any nucleus condensation). Later on, apoptotic cells lose their adherence to the substratum, and floating cells undergo plasma membrane alterations (characteristic of postapoptotic necrosis) and are progressively disintegrated into small fragments. All these alterations are consistent with the classical description of apoptosis and postapoptotic alterations (14).
Apoptosis is generally triggered through complex sequences of intracellular signaling. Besides the caspase cascade, which is a common signaling pathway triggered by a variety of apoptosis-inducing agents (18,42), the sphingomyelin-ceramide pathway has also been proposed to be a key component in apoptosis (22,23). However, in some experimental model systems, ceramide increase seems to be a consequence rather than a cause of the caspase activation or cell death (43). Ceramide or derivatives, e.g. sphingosine 1-phosphate, may exhibit also an antiapoptotic effect (44). Finally, the effect of ceramide pathway is regulated by the balance between pro-and antiapoptotic systems (45).
We have recently reported that nontoxic concentrations of oxLDL induced SM-ceramide pathway activation and triggered mitogenic signaling in vascular smooth muscle cells, both events (ceramide generation and mitogenic signaling) being inhibited by serpins (28,46). The data reported in the present paper show that toxic concentrations of oxLDL also activate the SM-ceramide pathway in ECV-304, but the effects of serpins suggest that this signaling pathway is not a prerequisite for the apoptotic process.
The activation of the SM-ceramide pathway was investigated during the first 10 h following addition of oxLDL. During the 1st h, SM hydrolysis was not finely investigated because changing the medium may trigger the activation of the SM-ceramide pathway during this early period of time (47). SM metabolism was not studied after 10 h of incubation with oxLDL because, under the experimental conditions used here, apoptosis began to rise at that time, and later increases of ceramide would be a consequence rather than a cause of cell death (42).
The serpins TPCK and DCIC have been used because of their ability to inhibit several forms of apoptosis (19,20). Moreover, serpins have recently been shown to inhibit simultaneously SM hydrolysis, ceramide generation, and apoptosis induced by daunorubicin (21). In the reported experiments, TPCK (used at a nontoxic concentration) effectively blocked SM hydrolysis and ceramide generation, in agreement with the idea that some (yet unknown) signal-transducing serine protease acts upstream of SM hydrolysis (21,46). Surprisingly, however, TPCK and DCIC did not inhibit oxLDL-induced apoptosis. This suggests that, apparently, oxLDL-induced apoptosis of ECV-304 does not require ceramide generation or TPCK-sensitive serine proteases. Accordingly, ceramide generated at the plasma membrane by bacterial sphingomyelinase treatment does not trigger apoptosis. Various hypotheses may be put forward to explain this lack of apoptotic effect of natural ceramides in ECV-304: (i) rapid degradation of ceramide or conversion into non-or antiapoptotic derivatives, such as sphingosine 1-phosphate (44); (ii) compartmentalization of ceramide that makes it unavailable to intracellular targets as suggested by Zhang et al. (48,49); (iii) lack or insensitivity of some ceramide target necessary for triggering apoptosis, this latter hypothesis being unlikely since short chain ceramides trigger apoptosis. More-over, the data of Modur et al. (50) exclude the possibility that endothelial cells may be generally insensitive to ceramides.
The lack of inhibition of oxLDL-induced apoptosis by serpins was somewhat surprising because serine proteases have been implicated, not only upstream from SMase activation (21), but also at various stages of the proteolytic signaling cascade leading to activation of apoptotic events (51,52). However, in other experimental model systems, TPCK-sensitive serine proteases are not required for inducing apoptosis (51), thus supporting the hypothesis that apoptosis can be triggered by various and distinct signaling pathways (51), involving or not involving TPCK-sensitive serine proteases.
The effects of C 2 -and C 6 -ceramide are also apparently puzzling because, in contrast to natural ceramides, the short-chain permeant ceramides are able to trigger apoptosis. Several hypotheses may be proposed in an attempt to explain this discrepancy. The natural ceramides generated by exogenous sphingomyelinase and synthetic ceramides may exhibit differences in intracellular microcompartmentalization, routing, and metabolic fate, which may explain differences of intracellular targets and cellular responses (48,49,53).
In ECV-304, the reported data suggest that oxLDL-induced apoptosis is apparently independent of ceramide and of TPCKsensitive serine proteases but is calcium-dependent as shown by the antiapoptotic effect of EGTA, a chelator of extracellular calcium, and nifedipine, a calcium channel blocker (12). In contrast, C 2 -ceramide-induced apoptosis did not induce any sustained calcium peak or require any calcium influx (since it was not blocked by EGTA). These data strongly suggest that apoptosis triggered by oxLDL and by C 2 -ceramide are mediated through two separate signaling pathways differing in their calcium dependence.
While this paper was in progress, Harada-Shiba et al. (54) reported that oxLDL elicited an early generation of ceramide (peaking at 15 min) and concluded that ceramide generation by acid sphingomyelinase is indispensable for the endothelial apoptosis induced by oxLDL. This conclusion (opposite from ours) is based on the use of (i) desipramine which inhibited both ceramide generation and apoptosis and (ii) C 2 -ceramide, which induces apoptosis. We agree with these data, but (in the light of our results) they could be interpreted as follows. Because acid sphingomyelinase is probably not the sole cellular target of desipramine (55,56), it cannot be excluded that desipramine may inhibit an early step common to both the apoptotic signaling and ceramide generation. In contrast (although serpins are not specific inhibitors of the SM-ceramide pathway), because serpins inhibit ceramide generation but not apoptosis, it may be concluded that ceramide generation is not indispensable to oxLDL-induced apoptosis and rather is a side effect in the oxLDL-induced apoptosis. Moreover our conclusion is supported by experiments with EGTA, which inhibited oxLDLinduced apoptosis, but not C 2 -ceramide-induced apoptosis, thus demonstrating that the apoptotic signaling triggered by C 2 -ceramide and that triggered by oxLDL are two separate pathways.
In conclusion, toxic concentrations of oxLDL elicit apoptosis of ECV-304 and trigger the activation of the sphingomyelinceramide pathway via a serpin-sensitive step. Short-chain permeant C 2 -and C 6 -ceramides are able to induce apoptosis, but through a calcium-independent process, whereas oxLDL-induced apoptosis is calcium-dependent. Moreover, as TPCK blocked selectively ceramide generation but not the apoptotic process, it is suggested that apoptosis and ceramide generation triggered by oxLDL are apparently unrelated events.