Suppression of endothelial cell apoptosis by high density lipoproteins (HDL) and HDL-associated lysosphingolipids.

Apoptotic cell death following injury of vascular endothelium is assumed to play an important role in the pathogenesis of atherosclerosis. In this report, we demonstrate that high density lipoproteins (HDL), a major anti-atherogenic lipoprotein fraction, protect endothelial cells against growth factor deprivation-induced apoptosis. HDL blocked the mitochondrial pathway of apoptosis by inhibiting dissipation of mitochondrial potential (Deltapsi(m)), generation of reactive oxygen species, and release of cytochrome c into the cytoplasm. As a consequence, HDL prevented activation of caspases 9 and 3 and apoptotic alterations of the plasma membrane such as increase of permeability and translocation of phosphatidylserine. Treatment of endothelial cells with HDL induced activation of the protein kinase Akt, an ubiquitous transducer of anti-apoptotic signals, and led to phosphorylation of BAD, a major Akt substrate. Suppression of Akt activity both by wortmannin and LY-294002 or by a dominant negative Akt mutant abolished the anti-apoptotic effect of HDL. Two bioactive lysosphingolipids present in HDL particles, sphingosylphosphorylcholine and lysosulfatide, fully mimicked the survival effect of HDL by blocking the mitochondrial pathway of apoptosis and potently activating Akt. In conclusion, the present study identifies HDL as a carrier of endogenous endothelial survival factors and suggests that inhibition of endothelial apoptosis by HDL-associated lysosphingolipids may represent an important and novel aspect of the anti-atherogenic activity of HDL.

Injury of the vascular endothelium is a critical event in the pathogenesis of atherosclerosis. Endothelial cells in lesionprone regions are characterized by increased turnover rates, and local apoptosis has been implicated in this process (1). Apoptosis of endothelial cells has been observed in regions of atherosclerotic plaques located downstream of blood flow where low shear stress and disturbed hemodynamic flow patterns prevail (2). Several studies indicate the involvement of endothelial cell apoptosis in acute coronary syndromes; apoptotic endothelial cells are pro-coagulant and constitute an integral part of highly thrombogenic "shed apoptotic membrane microparticles", which are found in the blood of patients with acute coronary events (3,4).
Several pro-athrogenic factors in plasma were demonstrated to initiate endothelial cell apoptosis in vitro. These include oxidized and minimally oxidized low density lipoproteins, angiotensin II, homocystein, and hyperglycemia (5)(6)(7)(8). Furthermore, endothelial cell apoptosis can be triggered by reactive oxygen species (ROS), 1 which are abundantly produced by several cell types found in atherosclerotic lesions (9). Much less, however, is known about the ability of endogenous anti-atherogenic factors to prevent endothelial cell apoptosis. Laminar shear stress and nitric oxide have been shown to suppress endothelial apoptosis triggered by various stimuli (7,10,11). Anti-atherogenic hormones such as oestrogens or insulin were also observed to protect endothelial cells against apoptosis (12,13). Although high density lipoproteins (HDL) are regarded as one of the most potent anti-atherogenic factors, little is known about their effect on endothelial cell survival and susceptibility to apoptosis.
In the present study, we demonstrate that HDL potently protect endothelial cells against growth factor deprivation-induced apoptosis via suppression of the mitochondrial apoptotic pathway. We characterize the serine/threonine kinase Akt as a mediator of the anti-apoptotic effect of HDL and identify molecular entities in HDL particles responsible for their protective effects in endothelial cells.

Isolation of Lipoproteins, Apoproteins, and HDL-Lipid and HDL-Protein
Fractions-HDL (d ϭ 1.125-1.210 g/ml) and LDL (d ϭ 1.019 -1.250 g/ml) were isolated from human plasma as described by Havel et al. (14). LDL was oxidized with CuSO 4 as described previously (15). Lipid-free apoA-I and apoA-II were isolated by reversed phase HPLC (16). To obtain lipid and protein fractions, 0.1 ml of native HDL (10 g/L) was diluted with 0.9 ml water, adjusted to pH 3.0 with sulfuric acid, and subsequently mixed with 1.0 ml acetonitrile. After addition of 0.5 g NaCl, the sample was centrifuged (5 min, 800 ϫ g). The upper organic phase arbitrarily called HDL lipid fraction was dried, dissolved in ethanol, and rapidly injected into phosphate-buffered saline while vortexing. The lower aqueous phase containing apoproteins was arbitrarily called HDL protein fraction, collected, and lyophilized. Both fractions were used for cell stimulation in amounts indexed to original HDL concentration.
HPLC Characterization of Lipoprotein-associated Lysosphingolipids-To characterize sphingosylphosphorylcholine (SPC) and lysosulfatide (LSF) contents in HDL and LDL, HPLC analysis of the lipid fractions obtained from each lipoprotein was performed on a Kontron (Neufahrn, Germany) liquid chromatograph with two Model 422 pumps and a Model 440 diode array detector. 0.03 ml of the HDL sample was introduced onto a column using a Rheodyne 7125 loop injector (Cotati, CA). The separation was carried out on a 5 m Nucleosil 100-Si column (25 cm ϫ 4 mm ID) (Macherey-Nagel, Dü ren, Germany), with a flow rate of 1.0 ml/min at an ambient temperature. Mobile phase component A was acetonitrile and component B was acetonitrile/water (80:20, v/v). The following gradient was used: 0 min 100% A; 45 min 10% A; 47 min 10% A; 48 min 100% A; 55 min end of run. Phospholipids were detected by UV absorption at 193 nm.
Fluorescence-activated Cell Sorter Analysis of Apoptotic Cells-Floating endothelial cells were collected by centrifugation. Adherent cells were collected by trypsinization. For determination of the annexin V binding, combined cell populations (5 ϫ 10 5 /ml) were resuspended in 140 mmol/liter NaCl, 10 mmol/liter Hepes, and 2.5 mmol/L CaCl 2 . Annexin V-FITC (Bender Med-Systems Diagnostics, Vienna, Austria) was added for 30 min at room temperature according to the supplier's instruction. For determination of the cell membrane permeability, endothelial cells (5 ϫ 10 5 /ml) were incubated for 30 min in phosphatebuffered saline with YO-PRO-1 at a final concentration of 0.1 mol/ liter. Flow cytometric measurements of annexin V binding and YO-PRO-1 staining were performed on a Coulter Epics Elite flow cytometer equipped with a 15-milliwatt argon ion laser (Coulter Corporation, Hialeah, FL) (excitation wavelength: 488 nm; fluorescence emission: 525 nm; data rate: ϳ300 cells/s).
Fluorimetric Determination of ROS and Mitochondrial Potential (⌬⌿m)-Determination of ROS production and ⌬⌿ m was performed as described by Quillet-Mary et al. (17). Briefly, production of ROS was detected with the fluorescent probe C2938. After labeling with 0.5 mol/liter C2936 for 1 h floating and adherent cells were collected and washed twice, and fluorescence was determined using a PerkinElmer Life Sciences LS70 fluorimeter (excitation wavelength: 510 nm, emission wavelength: 534 nm). To evaluate ⌬⌿ m cells were labeled with 40 nM DiOC6 and washed twice, and the fluorescence of the suspension was determined (excitation wavelength: 488 nm, emission wavelength: 500 nm).
Subcellular Fractionation and Cytochrome c Release-Cells were washed with cold phosphate-buffered saline, scraped, and centrifuged, and the cell pellet was resuspended in 20 mM Hepes-KOH, pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol, 250 mM sucrose, 100 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin and 1 g/ml aprotinin. Subcellular fractionation was performed as described (18). Briefly, cells were homogenized using a glass Dounce (10 strokes) and were centrifuged at 15,000 ϫ g for 15 min at 4°C. The particulate represented the mitochondria-containing nuclear-heavy membrane fraction. The supernatant was centrifuged for 15 min at 20,000 ϫ g at 4°C. This second supernatant represented the cytosol (including the light membrane fraction).
Transfection of HUVEC with Dominant Negative Akt (dnAkt) Constructs-HUVEC were transfected with 11 g dnAkt (c-myc-tagged dominant negative Akt1 (K179M) in pUSEamp, Upstate Biotechnology) and a control vector, respectively, using lipid pF7 (Invitrogen) in Optimem (Life Technologies, Inc.) as described earlier (19,20). After 24 h of recovery by culture in complete growth medium, cells transfected with control vector or dnAkt were exposed to growth factor (GF) deprivation in the presence of 1.0 g/liter HDL for 24 h. Floating apoptotic and adherent survived cells were collected and fixed in 3.7% (w/v) paraformaldehyde. Separate analysis of c-myc-positive cells (apoptotic and surviving) in both transfected cell populations (vector control and dnAkt) was performed by fluorescence-activated cell sorter analysis using a primary anti-c-myc-antibody (Upstate Biotechnology) and a secondary fluorescein isothiocyanate-labeled anti-mouse antibody. Apoptosis was calculated as percent c-myc-positive apoptotic cells from all c-mycpositive cells in vector control and dnAkt-transfected cell populations.

HDL Inhibit Apoptosis of Human Endothelial Cells-HU-
VEC undergo apoptosis when deprived of GF (19,20). Some of the early molecular events in apoptosis include alterations of the plasma membrane properties such as translocation of phosphatidylserine from the endofacial to the exofacial leaflet and increased permeability. To test the effect of HDL on endothelial cell apoptosis in our system, we exposed HUVEC to GF deprivation in the presence or absence of HDL and quantified phosphatidylserine-positive cells using annexin V staining as a measure of apoptosis. GF deprivation led to a significant increase of annexin V-positive cells over time, which was consistently inhibited by 40 -50% in the presence of 1.0 g/liter HDL at any time point measured (Fig. 1A). The time-dependent in- crease of apoptosis-associated membrane permeability measured with the dye YO-PRO-1 was also substantially inhibited by 1.0 g/liter HDL (Fig. 1B). The blocking effect of HDL on endothelial apoptosis was concentration-dependent, with a maximal suppression at 0.5-1.0 g/liter HDL (Fig. 1C). After undergoing apoptosis, HUVEC detach from the substratum and appear in the supernatant as apoptotic "floaters" (20). HDL suppressed the number of these apoptotic floaters in the conditioned medium in a concentration-dependent manner as well, with an ϳ50 -60% inhibition of apoptosis at 1.0 g/liter HDL (Fig. 1D).
HDL Suppress the Mitochondrial Pathway of Apoptosis-The execution of the apoptotic program requires the activation of the caspase family of cystein proteases, which disassemble cells through the selective proteolysis of key protein substrates. One of the principal mechanisms of caspase activation requires the release of cytochrome c from mitochondria, which together with Apaf-1 and dATP/ATP form the "apoptosome", the complex that recruits and activates the "initiator" caspase-9. In turn, caspase-9 recruits, cleaves, and activates "effector" caspases such as caspase-3 (21). Therefore, we examined the effect of HDL on the activation of caspase-9 and caspase-3 by measuring cleavage of their fluorogenic substrates, Ac-LEHD-AFC and Ac-DEVD-AMC, respectively. HDL dramatically suppressed the activity of both caspase-9 and caspase-3 measured 8 h after withdrawal of GF in a concentration-dependent manner with an ϳ75% inhibition at a concentration of 1.0 g/liter ( Fig. 2A). As a biochemical marker for caspase activation we also determined the effect of HDL on the proteolytic processing of pro-caspases-3 and -9, which is required for the full caspase activity. 1.0 g/liter HDL almost completely abolished the generation of the active 23 kDa caspase-3 subunit and markedly reduced the generation of the active 37 kDa caspase-9 subunit after GF deprivation (Fig. 2B).
To elucidate the impact of HDL on the molecular pathways upstream of caspase activation, we tested the effect of HDL on the dissipation of the ⌬ m and the generation of ROS, which are associated with the release of cytochrome c into the cytoplasm. After GF deprivation of HUVEC for 8 h a marked decrease of ⌬ m from 620 Ϯ 162 to 78 Ϯ 17 (n ϭ 3) was observed. This ⌬ m collapse was reverted in the presence of increasing concentrations of HDL (Fig. 2C). Because ⌬ m dissipation leads to production of ROS at the ubiquinone site of the respiratory chain, we also investigated ROS generation during endothelial apoptosis. Withdrawal of GFs for 8 h increased ROS levels from 18.2 Ϯ 4.1 (n ϭ 5) to 62.3 Ϯ 8.5 (n ϭ 5), which was inhibited by HDL in a concentration-dependent manner (Fig. 2D). These protective effects of HDL on the mitochondrium were associated with the inhibition of mitochondrial cytochrome c release into the cytoplasm, the initial step for the activation of caspases (Fig. 2E); while cytoplasmic cytochrome c levels gradually increased over time after GF deprivation in the absence of HDL, addition of 1.0 g/liter HDL to the culture medium substantially reduced cytochrome c levels released into the cytoplasm. As expected, the release of cytochrome c into the cytoplasm was accompanied by a decrease in the cytochrome c content in mitochondria, which occurred between 2 and 8 h after GF withdrawal. No such decrease was observed in endothelial cells preincubated with 1.0 g/L HDL at this time. To investigate the specificity of cytochrome c release during apoptosis we examined the effect of GF deprivation on VDAC, a membrane channel expressed in the mitochondrium as described previously (18). As shown in Fig. 2E, no VDAC could be detected in the cytoplasmic fraction of endothelial cells either in the presence or absence of HDL, thus excluding contamination of the cytosolic fraction with mitochondria. Akt Is Essential for HDL-mediated Cell Survival-The protein kinase Akt was identified as an important survival factor that suppresses the mitochondrial pathway of apoptosis in a number of experimental systems (22). To assess the involve-FIG. 2. HDL protect HUVEC against mitochondrial apoptosis. HUVEC were exposed to GF deprivation in the absence or presence of HDL as in Fig. 1. A, caspase-3 and -9 activities in lysates from HDL-treated and control cells were measured by the amount of fluorescence generated from the cleavage of Ac-DEVD-AMC or Ac-LEHD-AFC, respectively, and were expressed as fold increase over the basal level in GF-cultured cells. B, lysates from HDL-treated and control cells were subjected to Western blot analysis with monoclonal antibodies to caspase-3 and caspase-9. C, ⌬ m was determined by fluorescence spectroscopy in HUVEC stained with DiOC6 directly after treatment with HDL. D, ROS production was measured by fluorescence spectroscopy in HUVEC labeled with C2938. Data points represent the average Ϯ S.D. of at least three independent experiments with different HUVEC populations. E, cytoplasmic and particulate fractions of HUVEC treated for the indicated times with 1.0 g/liter HDL were subjected to Western blot analysis with monoclonal antibodies against cytochrome c or VDAC. Blots are representative for one experiment of two to four. ment of Akt in the HDL-mediated inhibition of endothelial apoptosis, we used wortmannin and LY-294002, two selective and structurally unrelated inhibitors of phosphatidylinositol 3-kinase (PI3-K), an upstream activator of Akt. Both wortmannin and LY-294002 reverted the inhibitory effect of HDL on caspase-3 activation and annexin-V binding during GF deprivation-induced apoptosis in a concentration-dependent manner (Fig. 3A). 100 nM wortmannin and 10 mol/liter LY-294002 completely abrogated the anti-apoptotic effects of HDL.
The protein kinase Akt is activated by binding of PI3-Kphosphorylated phospholipids and by direct phosphorylation at threonine 308 and serine 473. To determine whether HDL regulate the activity of Akt in endothelial cells, we analyzed whole-cell lysates of HDL-treated HUVEC for Akt activation using a phosphospecific antibody to serine 473 of Akt. In timecourse experiments, we found Akt to be activated 5-10 min after exposure to HDL (Fig. 3, B and C). This response gradually decreased after prolonged incubation. Densitometric analysis of signals revealed a maximal 3.4 Ϯ 0.3-fold (n ϭ 3) stimulation by HDL at 1,0 g/liter (Fig. 3C). In addition, HDL induced a time-dependent phosphorylation of BAD, a major substrate of Akt, at serine 136 (Fig. 3B). The maximal 4.2 Ϯ 1.3-fold (n ϭ 2) increase of BAD phosphorylation occurred 30 min after exposure to HDL and decreased thereafter (Fig. 3D). To test the involvement of PI3-K in the activation of Akt in response to HDL, we pretreated cells with 100 nmol/liter wortmannin or 10 mol/liter LY-294002. Both inhibitors completely abolished the phosphorylation of Akt (Fig. 3E).
To examine whether Akt is directly responsible for the survival effect of HDL, we transfected HUVEC with a c-myctagged catalytically inactive Akt mutant, which acts in a dominant negative fashion to suppress the activity of endogenous Akt. We then exposed the transfected HUVEC to GF deprivation in the presence of HDL and determined the percentage of apoptotic cells carrying the c-myc-tagged dominant negative Akt compared with vector controls using fluorescein isothiocyanate-labeled anti-c-myc antibodies and flow cytometry. We observed that almost twice as many endothelial cells expressing dominant negative Akt underwent apoptosis in the presence of HDL as control-transfected cells (Fig. 3F). This experiment documents that HDL mediate survival in endothelial cells via activation of Akt.
SPC and LSF Are Agonists Responsible for the Survival Activity of HDL-To determine specificity of the apoptosissuppressing effect of HDL we compared it with native and oxidized LDL. As shown in Fig. 4, native LDL exerted no significant inhibitory effects on annexin V binding or caspase-3 activity at physiological concentrations (0.5 g/liter). Similarly, oxidized LDL had no effect on the above parameters of apoptosis during the time monitored (8 h). At longer incubation times (24 -48 h) oxidized LDL initiated apoptosis as expected and previously published (5, 6; and data not shown). HDL are complex molecules containing several biologically active proteins and lipids. To determine HDL entities responsible for the inhibition of endothelial apoptosis, we tested different components of HDL for their anti-apoptotic potential. We observed no effect of the HDL protein fraction (0, 8 g/liter), purified apoA-I (0, 25 g/liter), or purified apoA-II (0, 25 g/liter) on annexin V binding or caspase-3 activity after induction of apoptosis (Fig.  4A). In contrast, the HDL lipid fraction exerted a potent antiapoptotic effect in our system (Fig. 4A). Recently, we demonstrated that HDL serve as a carrier for bioactive lysosphingolipids, such as SPC and LSF (23). The content of SPC and LSF in HDL used in the present study was 1.24 Ϯ 0.53 g/mg (n ϭ 5) and 4.71 Ϯ 1.12 g/mg (n ϭ 5), respectively. As shown in Fig.  4B, these two substances are abundantly present in HDL and to much lesser extent in LDL as characterized by HPLC. Therefore, we tested the effect of these two substances on endothelial apoptosis. Both SPC and LSF inhibited the increase of annexin V binding and cell permeability to YO-PRO-1 after GF depri- vation in a concentration-dependent manner (Fig. 4C). Moreover, caspase-3 and caspase-9 activities after induction of apoptosis were dose-dependently reduced in the presence of SPC and LSF (Fig. 4D). The maximal inhibitory effect of both compounds on apoptosis and caspase activation was observed in concentrations close to those present in HDL particles in serum. Finally, both SPC and LSF induced a time-dependent Akt phosphorylation in endothelial cells with similar kinetics to those of HDL (Fig. 4E). Hence, SPC and LSF represent HDL entities responsible for the inhibition of HUVEC apoptosis by HDL.

DISCUSSION
Low HDL cholesterol levels are one of the most predictive coronary risk factors (24,25). The negative correlation between coronary heart disease and plasma HDL cholesterol has been attributed to the ability of HDL to take up excess cellular cholesterol from the periphery and transport it to the liver, a process termed reverse cholesterol transport (26). In addition, several other activities exerted by HDL such as inhibition of monocyte adhesion to endothelial cells, inhibition of platelet activation, or inhibition of LDL oxidation are potentially antiatherogenic (27). In this study, we demonstrate that in contrast to native and oxidized LDL HDL potently promote endothelial cell survival and protect against apoptosis. HDL suppress the mitochondrial pathway of apoptosis by preserving mitochondrial integrity and thus inhibiting cytochrome c release as well as the subsequent activation of the caspase cascade. As a mediator of the anti-apoptotic effect of HDL upstream of the mitochondrium we identify the protein kinase Akt, an important survival factor in a number of experimental systems. We demonstrate that HDL activate Akt via a PI3-K-dependent pathway similarly to growth factors such as insulin (14), vascular endothelial growth factor (28), and angiopoetin-1 (29) because the inhibitors of this pathway, wortmannin and LY-294002, abolished HDL-mediated Akt activation. Furthermore, both inhibitors as well as overexpression of dominant negative Akt were found to block the anti-apoptotic activity of HDL in endothelial cells. These findings constitute the direct demonstration that the PI3-K/Akt pathway mediates the anti-apoptotic biological activity of HDL. Although cytoprotective effects of HDL against cell death induced by oxLDL (30) or TNF-␣ (31) were reported previously, neither the role of mitochondrial apoptosis nor the involvement of Akt were tested, and extremely long incubation times (Ͼ20 h) were necessary to demonstrate a protective effect, much in contrast to the HDL-mediated protection already visible 3 h after induction of apoptosis in our system. Moreover, the cytoprotective effects of HDL against oxLDL and TNF-␣ have been found to be partially mimicked by free apolipoprotein A-I, whereas none of the HDL apoproteins examined in this study had a protective effect in our system. These differences may be due to the different apoptotic stimuli used or may reflect the complexity of HDL-induced intracellular signaling, which we and others have previously shown to involve several independent pathways (32,33).
Several downstream targets of Akt may account for the antiapoptotic effect of HDL. We show that HDL induce phosphorylation of the pro-apoptotic protein BAD, an important substrate of Akt. This phosphorylation is known to promote the association of BAD with 14 -3-3 proteins and its dissociation from BCL-X L , which is then free to suppress mitochondrial apoptosis. Because in this study HDL was found to inhibit mitochondrial apoptosis, it is possible that the anti-apoptotic effect of HDL is associated with the phosphorylation of BAD. In addition, HDL inhibited the activity of caspase-9, which is another negatively regulated direct substrate of Akt (22). Recent studies suggest that in some cell systems Akt controls cell survival via regulation of the transcription factor NF-B (34, HUVEC were subjected to GF depletion for 8 h in the presence of A, vehicle (control), HDL (1.0 g/liter), LDL (0.5 g/liter), oxLDL (0.5 g/liter), HDL-proteins (0.5 g/liter), apoA-I (0.5 g/liter), apoA-II (0.25 g/liter), or HDL-lipids in the amount equivalent to 1.0 g/liter, and C and D, in the presence of increasing concentrations of LSF (left panels) or SPC (right panels). The percentage of apoptotic cells was determined by flow cytometry with annexin V or YO-PRO-1. Caspase-3 and caspase-9 activities were measured as in Fig. 2. Data points represent the average Ϯ S.D. of at least three separate experiments. B, HDL and LDL lipid fractions isolated from 1.0 g/liter lipoproteins were subjected to HPLC on Nucleosil 100-Si column as described under "Experimental Procedures". LSF and SPC peaks are indicated. E, HUVEC cultured as in Fig. 3B were treated for indicated times with 10 mol/liter SPC (upper panel) or 10 mol/ liter LSF (lower panel). Cell lysates (50 g/lane) were subjected to SDS-polyacrylamide gel electrophoresis and Western blot analysis using antibodies directed against phosphoserine 473 of Akt (P-Akt). Results are representative for two to three separate experiments. 35). However, because HDL neither induce the translocation of NF-B to the nucleus nor the binding of NF-B to specific DNA sequences (36), it is unlikely that this pathway contributes to the anti-apoptotic effect of HDL. Finally, Akt stimulation may enhance endothelial NO synthesis and promote endothelial cell survival by inhibiting caspases via nitrosylation (37). In this context, it is of interest that HDL counteract inhibitory effects of oxidized LDL on NO production in endothelial cells (38).
HDL are complex molecules known to induce a multitude of intracellular signals, for which different components of HDL have been made responsible. We previously demonstrated that free apolipoproteins A-I and A-II are responsible for the HDLmediated activation of phosphatidylcholine-specific phospholipase D but not for the activation of phosphatidylinositolspecific phospholipase C (32). Furthermore, neither apoA-I nor apoA-II could substitute for the intact HDL particle in mobilizing intracellular calcium in fibroblasts and endothelial cells (23,39). In contrast, we showed that both phosphatidylinositolspecific phospholipase C activation, Ca 2ϩ mobilization, and cell proliferation in response to HDL can be mimicked by two lysosphingolipids associated with native HDL, namely SPC and LSF (23). The present study extends the unique biological activities of SPC and LSF to the ability to mediate the antiapoptotic effect of HDL. Both substances inhibited activation of caspases and biochemical features of apoptosis in endothelial cells. In addition, SPC and LSF potently stimulated Akt in concentrations similar to those physiologically present in HDL particles in serum. In contrast, neither free apoproteins nor the protein fraction isolated from HDL had any effect. As ϳ93 and 72% of the total SPC and LSF plasma levels, respectively, are present in HDL 2 , these lipoproteins serve as the primary physiological carrier of the two bioactive lysosphingolipids. Consequently, no significant anti-apoptotic effect was observed with other lipoproteins such as LDL.
The intracellular signaling events initiated by SPC and LSF were the focus of several recent studies. In addition to their ability to induce phosphatidylinositol-specific phospholipase C activation and Ca 2ϩ mobilization, SPC has been shown to activate both mitogen-activated protein kinases (Erk-1 and Erk-2) and protein kinase C in several cell lines (40 -42). To our knowledge, SPC-induced activation of Akt and its anti-apoptotic activity has been not reported to date. However, sphingosine-1-phosphate, a compound structurally closely related to SPC was recently demonstrated to promote endothelial cell viability, although no signaling pathway has been identified (43). Both SPC and sphingosine-1-phosphate interact with the same receptor termed endothelial differentiation gene 3 (EDG-3) (44). Due to their lipophilicity, SPC and LSF were suggested to act primarily in an autocrine fashion, but their physiological function has been unknown. The present study identifies SPC and LSF as endogenous plasma-derived antiapoptotic factors, and reveals that the endothelium may constitute an important physiologic target for LSF and SPC delivered by HDL. Therefore, maintenance of endothelial cell integrity and protection against potentially deleterious proatherogenic stimuli may constitute a new biological function of HDL and thus may define a new mechanism that contributes to their atheroprotective function.