Sphingosine 1-phosphate may be a major component of plasma lipoproteins responsible for the cytoprotective actions in human umbilical vein endothelial cells.

Sphingosine 1-phosphate (S1P), a novel lipid mediator, is concentrated in the fraction of lipoproteins that include high density lipoprotein (HDL) and low density lipoprotein (LDL) in human plasma. Here, we show that oxidation of LDL resulted in a marked reduction in the S1P level in association with a marked accumulation of lysophosphatidylcholine (LPC). We therefore investigated the role of the lipoprotein-associated lipids especially S1P in the lipoprotein-induced cytoprotective or cytotoxic actions in human umbilical vein endothelial cells. The viability of the cells gradually decreased in the absence of serum or growth factors in the culture medium. The addition of oxidized LDL (ox-LDL) accelerated the decrease in the cell viability. LPC and 7-ketocholesterol mimicked ox-LDL actions. On the other hand, HDL and LDL almost completely reversed the serum deprivation- or ox-LDL-induced cytotoxicity. Exogenous S1P mimicked cytoprotective actions. Moreover, the S1P-rich fraction and chromatographically purified S1P from HDL exerted cytoprotective actions, but the rest of the fractions did not. The cytoprotective actions of HDL and S1P were associated with extracellular signal-regulated kinase (ERK) activation and were almost completely inhibited by pertussis toxin and PD98059, an ERK kinase inhibitor. The HDL-induced action was specifically desensitized in the S1P-pretreated cells. Taken together, these results indicate that the lipoprotein-associated S1P and the lipid receptor-mediated signal pathways may be responsible for the lipoprotein-induced cytoprotective actions. Furthermore, the decrease in the S1P content, in addition to the accumulation of cytotoxic substances such as LPC, may be important for the acquisition of the cytotoxic property to ox-LDL.

Plasma lipoproteins are responsible for lipid transport to cells and control of cholesterol synthesis. Low-density lipoprotein (LDL) 1 provides cholesterol to cells through LDL receptors, and this lipoprotein is thought to play an important role in atherosclerosis after undergoing oxidative modifications (1)(2)(3)(4). Thus, ox-LDL is present in atherosclerotic lesions and exerts a variety of biological actions, including cytotoxicity on the cells of the artery wall, potentially involved in atherogenesis (1)(2)(3)(4). Recent studies show that LPC mimics some of ox-LDL-induced actions (5)(6)(7)(8). On the other hand, HDL levels have been shown to be inversely correlated with the risk of cardiovascular disease (1)(2)(3)(4). Several mechanisms have been proposed for the anti-atherogenic functions of HDL. These include the promotion of the efflux of cholesterol from atherosclerotic plaques, inhibition of the oxidative modification of LDL, and inhibition of the expression of adhesion molecules such as vascular cell adhesion molecule-1 (VCAM-1) (1)(2)(3)(4). HDL has also been shown to protect endothelial cells from serum deprivation-and ox-LDL-induced cytotoxicity (1-4, 9, 10), but the mechanisms by which HDL exerts cytoprotective action are not fully understood. S1P, one of the sphingolipid metabolites, has been shown to participate in a variety of cellular responses including proliferation, differentiation, adhesion, motility, and apoptosis (11)(12)(13)(14)(15)(16). These cellular responses elicited by S1P were first thought to be mediated through an intracellular target(s), but extracellular mechanisms through G-protein-coupled S1P receptors have also been suggested. Supporting the latter extracellular mechanisms, several isoforms of S1P receptors have been identified (11)(12)(13)(14)(15)(16). These S1P receptor subtypes are expressed and functioning in a variety of cells including endothelial cells. In vascular endothelial cells, S1P has been shown to regulate a wide range of cellular activities involved in angiogenesis, wound healing, apoptosis, and atherosclerosis (17)(18)(19)(20). Thus, S1P induces cell migration, expression of several cell adhesion molecules, DNA synthesis, and cell survival (17)(18)(19)(20).
Sachinidis et al. (21) were the first to show that S1P-like lipids are associated with plasma lipoproteins (21). Recently, we specified one of the S1P-like lipids as S1P (22). We also succeeded in quantifying the S1P content in plasma components: this lipid was concentrated per unit amount of protein in lipoprotein fractions with the rank order of HDLϾLDL ϭ VLDLϾlipoprotein-deficient plasma (albumin fraction) (22). These results raise the possibility that S1P mediates some of * This work was supported in part by a research grant-in-aid for scientific research from the Japan Society for the Promotion of Science. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
the lipoprotein-induced actions in endothelial cells. In the present study, we show that S1P may mediate the lipoproteininduced cytoprotective actions through S1P receptors and their intracellular signaling pathways. We also found that oxidation of LDL markedly reduced its S1P content in association with a marked increase in cytotoxic LPC content. Thus, plasma lipoprotein-associated S1P may be an important factor for determining whether these actions are cytoprotective or cytotoxic.
Cell Culture-HUVECs with passages of 3 were purchased from Whittaker Bioproducts (Walkersville, MD). The cells (passage number between 5 and 12) were cultured in RPMI 1640 medium supplemented with 15% (v/v) FBS (Sigma) and several growth factors as previously described (19). Where indicated, PTX (100 ng/ml) or its vehicle (final 2 mM urea) was added to the culture medium 24 h before experiments, unless otherwise stated. CHO cells, which express Edg-1 or Edg-3, were cultured as previously described (15,22,23).
Cell Survival Assay-HUVECs were cultured for 24 h with test agents in fresh RPMI 1640 medium containing 0.1% BSA unless otherwise specified. In the experiments with PD98059 (10 M) or SB203580 (1 M), the cells were pretreated with these inhibitors for 1 h and then cultured for another 24 h with test agents in the presence of these inhibitors. The cells were then washed twice with PBS and harvested with trypsin. The viable cells were determined by trypan blue (0.2%) exclusion assay. The results were expressed as percentages of the value obtained with 15% FBS in the control cells.
Measurement of ERK1/2 Activity-HUVECs were incubated for 4 h in fresh RPMI 1640 medium containing 0.1% BSA unless otherwise noted. Where indicated, the cells were treated without or with PD98059 (10 M) or SB203580 (1 M) for the last 1 h during this incubation period. The cells were then washed once and preincubated for 20 min with or without these inhibitors at 37°C in a HEPES-buffered medium (15) and finally incubated for 5 min with test agents in the same medium. For the desensitization experiments with S1P (Fig. 5), the cells were incubated for another 5 h with or without S1P (1 M) after the 4 h-culture with RPMI 1640 medium containing 0.1% BSA. After the S1P pretreatment procedure, the cells were preincubated for 20 min in HEPES-buffered medium and incubated for 5 min with the test agents as described above. The incubation was terminated by washing twice with ice-cold PBS and adding 0.5 ml of a lysis buffer as previously described (24). The kinase activity was determined with an assay kit (Amersham Pharmacia Biotech) that measures the incorporation of [␥-32 P]ATP into a synthetic peptide (KRELVEPLTPAGEAPNQALLR) as a specific substrate. The enzyme activity was expressed as percentages of the basal activity without test agents in control cells. The same lysate was also analyzed by Western blotting with an ERK-specific antibody to detect the change in gel mobility reflecting phosphorylation of the enzyme as described previously (24).
Extraction of Active Components of HDL-HDL (about 4 mg in 2 ml) was extensively mixed with chloroform (3 ml), methanol (2 ml), water (0.5 ml), and 10 N NaOH (0.1 ml), and the phases were separated. The upper alkaline phase was collected. To the lower phase, 4 ml of synthetic upper phase mixture was added, and the phases were separated again. The lower phase containing the majority of phospholipids and neutral lipids evaporated to dryness (fraction a). The pooled upper alkaline phase containing S1P (about 8 ml), chloroform (4 ml) and HCl (0.2 ml) were mixed extensively and the phases were separated. The lower chloroform phase was collected. This extraction procedure was repeated another four times more by adding chloroform (4 ml) to the upper aqueous solution and phase separation. The pooled chloroform phase (about 20 ml) containing S1P was evaporated to dryness (fraction b). S1P recovery was about 90% as determined by including [ 3 H]S1P as an internal standard in the lipid purification procedure. The upper aqueous phase containing water-soluble substances was also dried by evaporation (fraction c). Fraction b was further processed by a silica gel high-performance thin layer chromatography (HPTLC) (Merck) using a solvent system consisting of 1-butanol/acetic acid/water (3:1:1). The silica gel with the resolved lipids (about 1-cm length each) was scraped off to obtain lipids covering the entire area of migration. The lipids were then eluted with chloroform/methanol/HCl (100:100:1) and dried by evaporation. All fractions thus separated were dissolved in PBS containing 0.4% BSA (2 ml) and were used at a final concentration of 10%.
Detection of LPC in Lipoproteins-The total lipids were extracted from lipoproteins as described for fraction a in the previous section except that 1 N HCl was used instead of 10 N NaOH. LPC was then separated by an HPTLC using a solvent system consisting of chloroform, methanol, 20% NH 4 OH (60:35:8). The bands were stained with primulin and visualized under UV light. The LPC fraction was scraped, and the lipid content was quantified with the malachite green method (25).
Lipoprotein Electrophoresis-Agarose gel film, TITAN GEL LIPO KIT J3045 (Helena Laboratories, Japan), was used. After electrophoresis, the film was dried and thereafter stained with Fat Red 7B. Other experimental conditions are described in the previous study (26).
Quantitative Measurement of S1P-S1P in plasma lipoproteins was selectively extracted, and its content was measured by a radioreceptor binding assay using Edg-1-expressing CHO cells as described previously (22,23).

Measurement of Inositol Phosphate Production in S1P
Receptor-expressing CHO Cells-This was performed as described previously (22). In brief, vector-or Edg-3-transfected CHO cells, which had been labeled with [ 3 H]inositol, were harvested from 10-cm dishes with trypsin, washed by sedimentation (250 ϫ g for 5 min) and resuspended in HEPES-buffered medium. The cells were then incubated to measure the production of 3 H-labeled IP 2 and IP 3 . To normalize the effects of lipoproteins in vector-transfected or Edg-3-transfected cells, data were first normalized to 10 5 dpm of the total radioactivity incorporated into the cellular inositol lipids in each experiment, and then the results were expressed as percentages of the maximal activity obtained at 1 M S1P in Edg-3-transfected cells.
Data Presentation-All experiments were performed in duplicate or triplicate. The results of multiple observations were presented as means Ϯ S.E. of at least three separate experiments unless otherwise stated. Statistical significance was assessed by Student's t test. Fig. 1A, we measured the S1P content in the lipoprotein particles of human plasma by a radioreceptor binding assay, which we recently established (23). For this quantitative measurement, S1P was extracted from lipoproteins. Consistent with the previous result (22), S1P contents in LDL and HDL reached ϳ100 -200 pmol/mg protein, respectively, which are 20 -40 times higher than the S1P content in the lipoprotein-deficient plasma (22). Because oxidation of LDL is thought to be a major risk factor for the development of atherosclerosis, we examined the effect of oxidation on the S1P content in LDL. The CuSO 4 treatment of LDL induced degradation of Apo B (Fig. 1B). The copper treatment also induced a marked accumulation of LPC at an expense of reduction of phosphatidylcholine (Fig. 1C) (5,6). Under these conditions, the S1P content was reduced to about 25% of the initial value (Fig. 1A).

Oxidation of LDL Resulted in a Decrease in the Lipoproteinassociated S1P Content-In
To examine whether the change in the S1P content is reflected in the functional activity, we measured S1P receptormediated phospholipase C-stimulating activity by the intact lipoprotein samples without the extraction procedure of S1P. In the vector-transfected CHO cells, inositol phosphate production in response to lipoproteins regardless of the lipoprotein species was very small (Fig. 1D, upper panel). On the other hand, in the S1P receptor Edg-3-overexpressing CHO cells, HDL and LDL markedly stimulated inositol phosphate production reflecting activation of phospholipase C, whereas ox-LDL exerted only a small effect on the activity (Fig. 1D, lower panel). It is reasonable to assume that the increase in the activity induced by the receptor transfection may be mediated by the S1P receptor. Thus, the change in the S1P content in lipoprotein particles seems to reflect their ability to stimulate the S1P receptor.
HDL and LDL Protect HUVECs from Cytotoxicity Induced by Serum-deprivation and ox-LDL-When HUVECs were cultured without serum or growth factors, the cells gradually lost their viability and were detached from the dishes. At 24 h after serum deprivation, only 50% of the cells had survived (Fig. 2A). Under these conditions, both HDL and LDL at 100 g/ml almost completely reversed the serum deprivation-induced cytotoxicity ( Fig. 2A). On the contrary, ox-LDL accelerated the cytotoxicity ( Fig. 2A). As shown in Fig. 2B, the oxidative lipids, including 7-ketocholesterol, 25-hydroxycholesterol, and LPC, which were accumulated during LDL oxidation (5,6,27,28) mimicked the ox-LDL-induced action. The cytotoxicity induced by these agents including ox-LDL was reversed by HDL (Fig.  2B). LDL was also effective for inhibiting the ox-LDL-induced action (Fig. 2C). S1P has been shown to protect HUVECs from cytotoxicity or apoptosis induced by serum deprivation (18 -20). We confirmed this observation (Fig. 2D). Furthermore, we found that S1P also inhibited the ox-LDL-induced cytotoxicity (Fig. 2D). Thus, S1P mimicked cytoprotective action of HDL or LDL. LPA has also been shown to regulate the variety of functions of HUVECs (29), but this lipid was ineffective for cytoprotection of HU-VECs at concentrations less than 10 M and exerted a rather cytotoxic effect at higher concentrations (data not shown). We also examined the effects of other lipids including plateletactivating factor, phosphatidic acid, phosphatidylserine, phosphatidylinositol, and phosphatidylethanolamine, but we could not detect any significant cytoprotective effect at concentrations lower than 10 M (data not shown).
HDL and S1P-induced Cytoprotective Actions May Be Mediated by G i /G o Protein-regulated ERK Pathways-We next examined the signaling pathways involved in the S1P and HDLinduced cytoprotective actions. For this, we used PTX, an inhibitor for G i /G o protein functions; PD98059, an inhibitor for ERK kinase (MEK); and SB203580, an inhibitor for p38 MAP kinase. Any drug treatment minimally affected the viability of the cells in the presence of serum (Fig. 3A). Among these agents, a prior treatment of the cells with PTX or PD98059, but not SB203580, almost completely inhibited S1P-or HDL-induced cytoprotective actions against the cytotoxicity induced by serum deprivation and ox-LDL (Fig. 3B). When LDL was used instead of HDL, we observed a similar cytoprotective action that was sensitive to both PTX and PD98059 (data not shown).
These results suggest involvement of G i /G o proteins and ERK in the S1P-and HDL-induced actions. Actually, S1P and HDL induced the phosphorylation of the ERK 1/2 as evidenced by the gel mobility-shift (Fig. 4A) and activated the enzyme as evidenced by the phosphorylation of the ERK-specific substrate peptide (Fig. 4, B and C). As expected, the activation of ERK was completely suppressed by the treatment of the cells with PTX and PD98059 (Fig. 4, A and D). These results indicate that HDL and S1P-induced cytoprotective actions may be mediated by ERK signaling pathways that are regulated by G i /G o protein-coupled receptors. S1P May Be a Major Component Mediating the HDL-induced Cytoprotective Actions-Thus, we could not discriminate the action mode of HDL from that of S1P. This suggests that HDL-induced cytoprotective actions may be mediated by S1P. To demonstrate this possibility, we performed desensitization experiments as shown in Fig. 5. When the cells were treated with S1P, the ERK activity peaked at around 5 min and then gradually decreased to the initial level at around 5 h (data not shown). After the S1P pretreatment, the cells no longer responded to the second applied S1P, but ATP-induced ERK activation was little affected by the S1P pretreatment (Fig. 5). Thus, the cells were undergoing homologous desensitization when the cells were pretreated with S1P. Under these conditions, HDL-induced ERK activation was also completely lost (Fig. 5). Thus, S1P seems to mediate HDL-induced ERK activation and hence the cytoprotective action of the lipoprotein.
The participation of S1P in the HDL action was further confirmed in Fig. 6. In this experiment, components of HDL were separated into three fractions: fraction a, lipid fractions containing the majority of lipids including fatty acids, neutral lipids and phospholipids; fraction b, lipids soluble under an alkaline aqueous solution such as S1P and LPA; fraction c, substances soluble in an aqueous solution. The cytoprotective activity (Fig. 6A) and ERK-activating activity (Fig. 6B) of HDL were recovered in the S1P-rich fraction b but not in fraction a or fraction c. The lipid components of fraction b were further separated by an HPTLC (Fig. 6E), in which S1P was mostly recovered in the fraction 4. The S1P-containing fraction 4 clearly induced the cytoprotective action (Fig. 6C) and ERK activation (Fig. 6D).
Charcoal Treatment Attenuated Not Only Cytoprotective Actions of HDL and LDL but Also Cytotoxic Action of ox-LDL-Finally, we examined the effects of charcoal treatment, which would eliminate low molecular weight substances such as S1P and LPC from lipoprotein action. Charcoal treatment reduced the S1P content to 10 -20% of initial value in either LDL or ox-LDL (Fig. 7A) without any significant change in the Apo composition (Fig. 7B). This treatment also markedly removed LPC from the lipoprotein particles (Fig. 7C). Under these conditions, not only cytoprotective action of LDL but also cytotoxic action of ox-LDL was reversed (Fig. 7D). In the case of HDL, however, charcoal treatment only partially (50%) removed S1P from the lipoprotein particles (Fig. 7A) probably because of its tight binding to the lipoprotein (22). Thus, the charcoal treatment exerted a small but significant inhibitory effect on the cytoprotective action of HDL (Fig. 7D).

DISCUSSION
HDL has been shown to exhibit a wide range of anti-atherogenic functions, including cytoprotective action against cytotoxicity or apoptosis induced by several cytokines, Fas, growth factor-deprivation, and ox-LDL (1-4, 9, 10). Consistent with previous studies (9, 10), HDL protected HUVECs from serum deprivation-and ox-LDL-induced cytotoxic action. We also found that LDL exerted the cytoprotective action to an extent comparable with HDL. Considering the characteristics of LDL as a risk factor for atherogenesis, one might wonder if this observation was anomalous. In previous studies, ox-LDL has been repeatedly shown to be cytotoxic, but, to our knowledge, there is no report showing the cytotoxic action of the native LDL. Thus, we postulate that native LDL itself possesses potentially cytoprotective function, although this lipoprotein might acquire cytotoxic character during its oxidation.
The present studies indicate that S1P and its receptor-mediated signaling pathways are important for HDL-and LDLinduced cytoprotective action. First, the S1P-rich fraction and HPTLC-purified S1P from HDL exerted the cytoprotective action, but the rest of the fraction did not (Fig. 6). Second, the removal of S1P by charcoal treatment of HDL and LDL inhibited the cytoprotective action of these lipoproteins, although the effect was small in the case of HDL because of the insufficient removal of S1P (Fig. 7). Third, S1P-and HDL-induced cytoprotective actions were associated with the activation of ERK, and these responses were suppressed by PTX, an inhibitor of G i /G o -protein function or PD98059, an inhibitor of ERK kinase (Figs. 3 and 4). These results suggest that G i /G o -proteinregulated ERK activation may play an important role in the cytoprotective actions of S1P and HDL. The role of Ca 2ϩ signaling and/or ERK pathway in the S1P-induced cell survival has recently been proposed by other groups (18,20). Fourth, S1P or HDL-induced, but not ATP-induced, ERK activation was specifically desensitized by a prior stimulation of the cells with S1P, suggesting an involvement of S1P receptors in the HDL action (Fig. 5). In relation to this, it has been reported that TNF-␣ increases the intracellular S1P level by activation of sphingosine kinase and thereby induces anti-apoptotic action in HUVECs (17). This suggests that an accumulation of intracellular S1P may also exhibit cytoprotective action. However, the same authors also reported that HDL decreased rather than increased the intracellular level of S1P by inhibiting sphingosine kinase (30). Thus, it would be a minor mechanism, if not negligible, that HDL-associated S1P would be incorporated into the cells and thereby induce cytoprotective action. Although we did not specify the subtype of the S1P receptor involved in the HDL actions in the present study, both Edg-1 and Edg-3 may be responsible for the cytoprotective action (18 -20, 31). The present results are quite consistent with the recent study by Sachinidis et al. (21), in which it was suggested that S1P-like lipids in lipoproteins may mediate the activation of ERK and stimulation of DNA synthesis in vascular SMCs.
In the previous study (9), Apo A as well as HDL exhibited cytoprotective action against ox-LDL-induced cytotoxicity in endothelial cell lines, although HDL was more effective than Apo A. This suggests that not only the lipid component, probably S1P as shown here, but also Apo A may possess the potential cytoprotective activity against cytotoxicity of ox-LDL. However, in that study, the endothelial cell lines seem to be stable for serum deprivation and ox-LDL; the cells survived for at least 48 h even without serum, and more than 24 h was required for the induction of significant cytotoxic effect by ox-LDL. This was somehow different from our system using HUVECs; about 50% of the cells lost their viability during 24-h culture without serum or growth factors even in the absence of ox-LDL. Similar susceptibility to serum deprivation of HU-VECs has been observed by other groups (10,18,20). Thus, Apo A might participate in the cytoprotective action of HDL against predominantly late or chronic phase of cytotoxicity. Alternatively, the cytoprotective mechanisms might differ with different sources of endothelial cells.
The present study indicates that S1P mediates the HDLinduced cytoprotective actions through ERK-involving pathways, but it should be noted that there was a considerably large difference in their potency between ERK activation (about 3 nM, see hand, in the case of HDL, the difference was small; 10 g/ml for ERK activation (Fig. 4C) versus 30 g/ml for cytoprotective action ( Fig. 2A). This peculiar observation may be explained by the notion that S1P is metabolized very fast especially in the absence of lipoproteins. Under the present assay conditions using HUVECs, we observed that the half-life of HDL-associated S1P was about 2 h at 100 g/ml HDL (which corresponds to ϳ20 nM S1P), whereas the half-life of exogenous S1P was about 30 min at the same concentration in the absence of HDL but the presence of 0.1% BSA (data not shown). For the ERK assay, the activity was measured 5 min after the addition of test agents, whereas it was measured 24 h after for the cytoprotective activity. Thus, it is reasonable to speculate that a higher concentration of S1P is necessary to observe the long term cytoprotective action compared with the short term ERK activation especially in the absence of HDL.
The mechanism by which ox-LDL induces a variety of responses involved in the development of atherosclerosis was recently extensively investigated although it is still not completely defined (1)(2)(3)(4). During oxidation of LDL, several products such as lipid hydroperoxides, oxysterols, and LPC are produced (5,6,27,28). In addition, the production of lipid mediators such as LPA and platelet-activating factor has also been reported (29,32). Among these oxidative lipid products, LPC has been shown to duplicate a variety of ox-LDL-induced actions including monocyte migration and expression of adhesion molecules on endothelial cells (5)(6)(7)(8). As for cytotoxicity, LPC and oxysterols such as 7-ketocholesterol have been shown to mimic the ox-LDL-induced action in vascular endothelial cells (8,28). Thus, these lipids may be components of ox-LDL responsible for the induction of cytotoxicity, although their molecular targets and the mechanisms causing cytotoxicity remain unknown. This conclusion is further supported by the observation that charcoal treatment of ox-LDL reversed its cytotoxic activity in an association with a marked decrease in LPC content without any apparent change in Apo components (Fig. 7).
In vascular smooth muscle cells, LDL-and HDL-associated S1P-like lipids stimulated DNA synthesis (21). Based on these results, the investigators postulated that the S1P-like lipids might behave as atherogenic mediators and might be increased by oxidation of lipoproteins (21). However, in the present study, we demonstrated that oxidation of LDL markedly reduced, but not increased, its S1P content. The reduction of S1P content by copper treatment was blocked by an antioxidant butylated hydroxytoluene, indicating an oxidation-dependent reaction (data not shown). At present, however, the metabolic pathway of S1P degradation and its mechanism remains uncharacterized. This is an important future subject for investigation. In any event, during LDL oxidation, the contents of cytotoxic LPC and cytoprotective S1P changed reciprocally. The decrease in the S1P content may also be involved in the acquisition of cytotoxicity to ox-LDL. Thus, we propose that the balance between the contents of cytotoxic lipids including LPC and cytoprotective S1P may be an important factor that determines whether plasma lipoproteins are cytotoxic or cytoprotective. This balance might also be an important determinant for lipoproteins to be atherogenic or anti-atherogenic. In this proposal, S1P is postulated to be an anti-atherogenic mediator. In the endothelial cells, S1P has been shown to stimulate nitric oxide production, cell migration, and cell proliferation (18 -20, 33). Furthermore, in vascular smooth muscle cells, S1P is a potent inhibitor of cell migration (34). These responses in addition to cytoprotective action seem to favor anti-atherogenic properties. On the other hand, S1P has been shown to induce expression of adhesion molecules such as VCAM-1 and E-se-lectin in endothelial cells (16). These actions suggest rather atherogenic properties of S1P. Thus, further experiments are necessary to conclude whether S1P is atherogenic or antiatherogenic. However, these findings together with the present study suggest that control of the S1P content in plasma lipoproteins and the S1P receptor function in vascular cells may provide potentially useful means for the therapy of cardiovascular disease.
In conclusion, HDL-associated S1P is a major component of the lipoprotein-induced cytoprotective action in HUVECs. This action is probably mediated by ERK pathways that are regulated by S1P receptors such as Edg-1 and Edg-3. Oxidation of LDL resulted in a marked decrease in S1P content in association with a marked increase in LPC content. Such a reciprocal change in the lysophospholipid composition may be important for cytotoxicity to ox-LDL.