OUP user menu

★ Editor's choice ★

Atheroprotective role of high-density lipoprotein (HDL)-associated sphingosine-1-phosphate (S1P)

Francesco Potì, Manuela Simoni, Jerzy-Roch Nofer
DOI: http://dx.doi.org/10.1093/cvr/cvu136 395-404 First published online: 1 June 2014


Numerous epidemiological studies documented an inverse relationship between plasma high-density lipoprotein (HDL) cholesterol levels and the extent of atherosclerotic disease. However, clinical interventions targeting HDL cholesterol failed to show clinical benefits with respect to cardiovascular risk reduction, suggesting that HDL components distinct from cholesterol may account for anti-atherogenic effects attributed to this lipoprotein. Sphingosine-1-phosphate (S1P)—a lysosphingolipid exerting its biological activity via binding to specific G protein-coupled receptors and regulating a wide array of biological responses in a variety of different organs and tissues including the cardiovascular system—has been identified as an integral constituent of HDL particles. In the present review, we discuss current evidence from epidemiological studies, experimental approaches in vitro, and animal models of atherosclerosis, suggesting that S1P contributes to atheroprotective effects exerted by HDL particles.

  • High-density lipoproteins
  • Sphingosine-1-phosphate
  • Atherosclerosis
  • Inflammation

1. Introduction

Clinical and epidemiological investigations convincingly documented an inverse relationship between high-density lipoprotein cholesterol (HDL-C) and cardiovascular disease (CVD).1,2 Seminal studies, such as Framingham Heart Study3 and Prospective Cardiovascular Munster (PROCAM),4 showed that low HDL-C levels are an independent risk factor for CVD and estimated that an increase of 1 mg/dL (0.026 mmol/L) in HDL-C is associated with a 2–3% of risk reduction.24 Animal experiments also provided evidence that the development of atherosclerotic lesions may be retarded or even reversed by transgenic overexpression or exogenous administration of apolipoprotein (apo) A-I, the constitutive and most abundant protein of HDL.5 Moreover, the infusion of reconstituted HDL particles was shown to reduce coronary plaque volume in both animal models and humans.5

To date, arduous efforts have been made to establish therapies increasing HDL-C, assuming that its higher levels could translate into a reduced CV risk. However, there is accumulating evidence that the plain elevation of HDL-C do not protect against CVD.6,7 For instance, AIM-HIGH and HPS2-THRIVE trials, in which niacin has been applied to target HDL-C levels, failed to show clinical benefits with respect to cardiovascular risk reduction.8,9 Likewise, ILLUMINATE and dal-OUTCOME trials, in which patients with acute coronary syndrome were randomly allocated to placebo or cholesterol ester transfer protein (CETP) inhibitors torcetrapib or dalcetrapib, respectively, revealed a lack of efficacy in reducing cardiovascular events despite the impressive HDL-C increase by 72 and 31–40%.10,11 Though both niacin and CETP inhibitor trials were performed in patients receiving statins, thus raising the possibility that potential beneficial effects of HDL-C elevation were pre-empted by lowering plasma low-density lipoprotein cholesterol (LDL-C), recent results of other experimental approaches also undermine the contention that high HDL-C affords protection against atherosclerosis. For instance, a large Mendelian randomization study (20 913 cases and 95 407 controls)12 showed that common single-nucleotide polymorphisms (SNPs) that exclusively associate with higher levels of HDL-C did not affect the risk of myocardial infarction. In contrast, reduced risk has been observed in carriers of SNPs associating with lower levels of LDL.

Collectively, the results of clinical trials and Mendelian randomization studies suggest that cholesterol content may be a poor metrics for anti-atherogenic effects exerted by HDL.6,7 As a matter of fact, cholesterol represents only one constituent of HDL particles, which in addition are composed of several molecules of apoA-I, a shell of phosphatidylcholine, above 80 quantitatively minor proteins, hundreds of diverse lipids species, and even microRNAs.1316 Theoretically, each of these components may directly or indirectly contribute to the anti-atherogenic potential of HDL. During the last two decades, a number of studies1719 demonstrated that HDL serves as a carrier for sphingosine-1-phosphate (S1P), a lysosphingolipid exerting its biological activity mainly by binding to five specific G protein-coupled receptors, named S1P1–5.20,21 S1P was shown to regulate a plethora of biological responses in a variety of different organs, tissues, and cell types, including the cardiovascular system.21 It has been demonstrated that S1P is a potent regulator of T-cell trafficking22 and S1P signalling pathways may control host response to infections23,24 and play a prominent role in cancer progression.25 Excellent reviews have extensively covered these topics and the interested reader is referred therein for more in-depth details.21,2628 In the present review, we summarize evidence suggesting that HDL-associated S1P is relevant to atheroprotective effects exerted by this lipoprotein and highlighting anti-atherogenic activities mediated by S1P in vitro and in vivo.

2. S1P distribution in plasma and lipoproteins

S1P is generated by phosphorylation of sphingosine, the direct precursor coming from the ceramide/sphingomyelin metabolic pathway. Sphingosine kinase 1 or 2 (SphK 1 or 2) is responsible for S1P synthesis, while S1P phosphatases and S1P lyase account for S1P dephosphorylation to sphingosine or irreversible degradation to phosphoethanolamine and hexadecanal, respectively.21,27,29 S1P is present at high nanomolar concentrations (200–900 nmol/L) in plasma and even higher in serum,30 while lower levels are encountered in tissues.31,32 Significantly higher plasma S1P levels were found in women than in men in the same age and in premenopausal than in postmenopausal women.33 It has been estimated that S1P concentrations in lymph correspond to ∼25% of plasma levels,34 whereas in the interstitial fluid S1P occurs in the low nanomolar range,31,32 thus generating a relevant biological gradient. Different S1P sources are believed to account for the high S1P concentrations in blood. Platelets were initially proposed to represent the major repository of S1P due to the low activity of S1P lyase.35 The release of stored S1P during platelet activation explains why S1P concentrations are higher in serum than in plasma. However, nuclear factor-erythroid 2-deficient mice, which have no circulating platelets, were found to possess normal S1P levels in plasma.34 In addition, no correlation between platelet-related parameters in blood and plasma S1P concentrations could be noted.36 It seems, therefore, that platelets do not significantly contribute to plasma S1P content in steady state, albeit they might increase local S1P concentrations during thrombus formation. Erythrocytes were postulated as the source of plasma S1P alternative to platelets, as they were found to efficiently incorporate, store, and release S1P to plasma, but not to other plasma-free media, and the transfer of wild-type erythrocytes into SphK1/2-conditional double knock-out mice, which are characterized by undetectable S1P in blood, restored the normal plasma S1P levels.34,37 Moreover, in contrast to platelets, close correlation between erythrocyte count in blood and plasma S1P levels has been reported.36,37 While experiments using animal models with selective SphK deficiency in haematopoietic cells additionally supported the role of erythrocytes in regulating plasma S1P,38 Venkataraman et al.39 demonstrated that other important sources of S1P exist and suggested that vascular endothelium may significantly contribute to plasma S1P levels.

While S1P is present in blood in relatively high concentrations, in vitro assays indicate that the potency of S1P to trigger S1P receptors is ∼10–100 nmol/L.40 Consequently, if plasma S1P were fully active, S1P receptors on blood and endothelial cells would be expected to reach saturation. As a matter of fact, Murata et al.30 observed a significant rightward shift in the dose–response curves by stimulating cells with charcoal-treated, lipid-depleted plasma, which contains only a minimum amount of S1P. Assuming that under physiological conditions blood and endothelial cells may come into full contact with plasma, they calculated that only 2% of plasma S1P could be active. Approximately 55 and 35% of plasma S1P partition into HDL and albumin, respectively.30 Only ∼10% of plasma S1P resides in other lipoproteins, chiefly LDL. Early in vitro studies demonstrated that HDL-associated S1P was as potent as albumin-bound S1P to stimulate short-term responses in cells, such as activation of mitogen-activated protein kinase (MAPK) and endothelial nitric oxide synthase (eNOS), while significantly more potent than albumin-bound S1P in stimulating long-term functions, such as cell survival.18,40,41 Recently, Wilkerson et al.42 showed that HDL-associated S1P sustains endothelial cell barrier longer than S1P present in albumin. In addition, these authors found that HDL-S1P but not albumin-S1P reduces the degradation of S1P1 protein and promotes the receptor recycling on the cell surface, providing clear evidence that the type of S1P carrier is a key determinant in this process.42 It is noteworthy that HDL-associated S1P possesses four-fold longer half-life than S1P bound to albumin.40 Collectively, these observations suggest that HDL may represent a stable reservoir of fully active S1P in blood, modulating both short- and long-term anti-atherogenic actions. On the contrary, because of its high concentration in plasma, albumin may act both as a reservoir and a molecular trap for S1P, effectively preventing the over-stimulation of S1P receptors (see Figure 1 for schematic representation).

Figure 1

Schematic representation of S1P transport, distribution, and action. S1P is synthesized mainly in platelets, erythrocytes, and endothelial cells and transported to HDL and/or albumin by Spns2 or ATP-binding cassette transporters (ABC) A1, C1, or G2. S1P bound to apoM in HDL particles interacts with S1P receptors (S1P1, S1P2, and S1P3) on cells in the vasculature to activate intracellular signalling pathways encompassing trimeric G-proteins (Gi, Gq, and G12/13) and other signalling molecules adenylate cyclase—AC; phosphoinositid-3-kinase; phosphatidylinositol-specific phospholipase C (PLC); small G proteins Rac1 and Rho; adapted from ref.27. Docking of HDL particles to SR-BI mediated by apoA-I may facilitate S1P-dependent signalling.

While HDL seems to effectively extract S1P from erythrocyte membranes,43 the mechanisms underlying S1P accumulation in this lipoprotein remain obscure (Figure 1). Spinster homologue 2 (Spns2) is a specific transporter involved in S1P mobilization from cells.44 Although Spns2 primarily affects S1P levels in the lymphatic compartment, it nevertheless contributes to the S1P export from endothelial cells and the concentration of HDL-bound S1P in the plasma is lower in Spns2-deficient animals.4548 However, the evidence suggesting a direct contribution of Spns2 to S1P cargo carried by HDL particles has not been provided to date. Recent findings indicate that ABC transporters may play a pivotal role in mediating the export of S1P from cells. It has been shown that glibenclamide, a non-specific inhibitor of ABC transporters, suppresses S1P release from platelets49 and erythrocytes.50 Other studies further demonstrated that ABCC1 is involved in the export of S1P from mast cells, endothelial cells, adipocytes, Langerhans cells, and, together with ABCG2, in oestradiol-induced S1P release from MCF-7 breast cancer cells.5155 The latter observation may explain higher plasma S1P levels encountered in women (see above). Similar to Spns2, the evidence indicating a direct interaction between ABCC1 and HDL particles is currently missing. However, HDL is well known to interact with other members of the ABC transporter family such as ABCA1 and ABCG1 and thereby to promote the efflux of cholesterol and phospholipids from cells. Few studies suggested that ABCA1 may be responsible for S1P enrichment and the generation of HDL-like lipoproteins in the central nervous system.56,57 This mechanism, however, has not been as yet demonstrated to account for the accumulation of S1P in plasma lipoproteins.

Calculating the S1P/HDL molar ratio in plasma indicates that on average every 10th HDL particle serves as a carrier to S1P cargo. This observation raises an interesting possibility that S1P may be specifically associated with a distinct subpopulation of HDL particles. Actually, apolipoprotein M (apoM)—a member of the lipocalin protein superfamily—has been identified as a S1P-binding protein in plasma HDL58 ( Figure 1). ApoM concentration in plasma is ∼0.9 μmol/L, with >95% bound to HDL, mainly to the HDL3 subclass, and the remaining to LDL and very low density lipoproteins.59,60 Crystallographic analysis and ligand-binding studies performed on wild-type or mutant apoM isoforms showed that S1P binds to apoM with an IC50 of 0.9 μmol/L, which is concordant with physiological S1P concentrations in plasma.61 A seminal study revealed that S1P is absent in HDL obtained from apoM knock-out mice, while transgenic mice overexpressing human apoM show a significant increase of S1P in HDL particles.58 Moreover, in endothelial cells, only apoM-containing HDL induced typical responses of S1P–S1P1 axis activation, such as S1P1 receptor internalization, activation of kinases MAP and protein kinase B (PKB or Akt), and migration. In addition, apoM-deficient mice display impaired endothelial barrier function resulting in vascular leakage in the lungs.58 The same study further demonstrated that, also in human, plasma S1P content of HDL is restricted to the apoM-containing particles. However, no significant correlation has been found between total S1P and apoM in human plasma.62 Few important issues may, at least partially, help to explain this apparent discrepancy.63 First, it is likely that the albumin-bound S1P contributes more to the variation in total plasma S1P than in S1P bound to apoM-containing HDL. Secondly, apoM is not completely saturated with S1P and may bind other lipophilic compounds, for example oxidized phospholipids64 or retinol,65 thus attenuating the possible correlation between plasma apoM and S1P. Very recently, hepatocyte-specific apoM transgenic (apoM-Tg) mice have been generated.66 ApoM-Tg mice presented with large S1P-enriched HDL particles in plasma, although total HDL-C levels were similar in transgenic and wild-type animals. Importantly, primary hepatocytes isolated from apoM-Tg mice displayed increased S1P synthesis and secretion, suggesting that the export of S1P from hepatocytes is largely apoM-dependent.66

3. Plasma S1P in CVD

To date, only few investigations systematically analysed the relationship between plasma S1P levels and CVD. An early study, which involved 308 patients undergoing coronary angiography for various indications, showed that subjects with coronary artery disease (CAD) had higher levels of S1P in serum.67 In addition, S1P levels better predicted obstructive disease than classic cardiovascular risk factors and closely correlated with its severity. Unfortunately, neither the distribution of S1P between HDL and albumin, nor the relationship between HDL-associated S1P and CVD, has been specifically addressed therein. Some years later, Sattler et al.68 found that plasma levels of HDL-bound S1P were lower in individuals with myocardial infarction and stable CAD, whereas those of non-HDL-bound S1P were higher, when compared with healthy controls. In addition, the study reported that the levels of HDL-bound plasma S1P were inversely correlated with the severity of symptoms, whereas those of non-HDL-bound S1P increased proportionally with the severity of CAD.68 Another recent study, involving 204 subjects selected from the Copenhagen City Heart Study (CCHS) cohort, found a significant inverse correlation between HDL-associated S1P levels in serum and the occurrence of CAD.69 Interestingly, this inverse relationship was independent of HDL-C levels upon subject stratification for high (females: ≥73.5 mg/dL; males: ≥61.9 mg/dL) and low (females: ≤38.7 mg/dL; males: ≤34.1 mg/dL) serum HDL-C. These authors also found higher HDL-bound S1P in individuals without CAD than in CAD-affected patients, suggesting that the pathological status may impact the partitioning of S1P to HDL by shifting the lysosphingolipid towards other lipoproteins or albumin.69 A small study by Knapp et al.70 showed that patients with acute myocardial infarction had significantly lower levels of total plasma S1P than controls. In line with previous results, the same group recently found that patients with acute ST-segment elevation myocardial infarction had lower plasma S1P levels than healthy controls.71 This effect, already present upon admission, was maintained for at least 30 days after the infarction. Interestingly, 2 years post-infarction, plasma S1P level recovered only partially.71 A recent study investigated the role of plasma sphingolipids following transient cardiac ischaemia occurring during routine percutaneous coronary intervention (PCI).72 Total S1P, as well as sphingosine and sphinganine, were assessed in blood samples collected from either the coronary sinus or femoral vein from 31 patients at baseline and at different time points after PCI. A significant increase in S1P levels over the baseline with the maximum peak 5 min after PCI has been observed, which strongly correlated with troponin T levels, a biomarker of myocardial damage.72

Other studies demonstrated that plasma S1P levels may be altered in pathologies affecting vascular homeostasis and contributing to the development of atherosclerosis, such as Type 2 diabetes mellitus (T2DM) and obesity. For instance, Kowalski et al.73 found higher total S1P levels in high-fat diet-induced and genetically obese (ob/ob) mice compared with lean controls, and this result was paralleled also in obese humans. In these subjects, S1P levels positively correlated with total body fat percentage, body mass index, waist circumference, fasting insulin, HbA1c, and also LDL-C. Tong et al.74 recently showed that patients with T2DM may have higher HDL-bound S1P than healthy individuals, and suggested that this may represent a physiological compensatory mechanism to counteract vascular complications of the disease, by inducing cyclooxygenase 2 (COX2) expression and prostacyclin (PGI2) release. Taken together, these results suggest that the partitioning of S1P into HDL and non-HDL plasma pool may be relevant to the pathogenesis of CAD, with different or even opposite effects of S1P on the cardiovascular system depending on whether it is HDL-bound or not. In general, HDL-associated S1P has been shown to exert beneficial effects, whereas albumin-bound S1P has been shown to variably affect inflammation sites by contributing either to pro- or to anti-inflammatory reactions.

4. S1P-mediated anti-atherogenic effects: in vitro findings

The results of epidemiological studies demonstrating the inverse relationship between HDL-C levels in plasma and the cardiovascular risk triggered a wave research attempting to dissect molecular mechanisms underlying this relationship. In the last two decades, it has become increasingly clear that HDL particles interact with almost every cell type involved in the pathogenesis of atherosclerosis, and that the alterations of cell functions arising as a consequence of these interactions may directly account for the atheroprotective action of this lipoprotein. Mechanistically, docking of HDL particles on the cell surface and the subsequent interaction of one or more HDL particle components with cell surface receptors are believed to provide a molecular framework, by which HDL induces diverse cellular responses (Figure 1). Out of several HDL-binding partners [HDL-binding protein (HBP/vigilin), HB1, HB2, heat shock protein 60 (HSP60), glycosylphosphatidylinositol-anchored HDL-binding protein 1 (GPI-HBP1), and cubilin], scavenger receptor Type BI (SR-BI) has been firmly established to act as a full HDL receptor transducing signal across the cell membrane upon interaction with the lipoprotein holoparticle (for recent review, see ref.75). There is growing body of evidence suggesting that the binding of HDL to SR-BI may provide a spacial proximity required for the effective interaction of the S1P cargo of lipoprotein and the membrane-localized S1P receptors. For instance, some potentially anti-atherogenic effects exerted by HDL in isolated endothelial cells or murine aortas were attenuated in the presence of antibodies against SR-BI or in SR-BI deficiency.76,77 In addition, adhesion molecule expression elicited in endothelial cells by low-concentrated S1P disappeared in the presence of physiological concentrations of HDL in a manner sensitive to SR-BI.78 Conversely, elevation of SR-BI expression in endothelial cells potentiated the effects of HDL attributable to its S1P content.79 However, the obligatory involvement of trimeric G-proteins—canonical signalling partners of heptahelical receptors—in several cellular effects of HDL18,76,78,8084 argues for S1P receptors rather than SR-BI as principal transducers of HDL-triggered intracellular signalling. In the recent years, a number of intracellular signalling events induced by HDL-associated S1P, located downstream S1P receptors, and producing modulatory effects relevant to atheroprotective action unfolded by this lipoprotein have been identified. In the following we summarize potentially anti-atherogenic cellular effects of HDL particles in endothelial cells, smooth muscle cells, cardiomyocytes, and macrophages, which could have been clearly attributed to their S1P cargo (Table 1).

View this table:
Table 1

Protective cardiovascular effects of HDL-associated S1P—findings in vitro

Cell/tissue(Athero)protective effectS1P receptorExtra-/intracellular signalling pathway
Endothelial cellsProliferationS1P1ERK, Ras
SurvivalS1P1Akt, ERK
MigrationS1P1, S1P3Akt, AMPK, EL, p38MAPK
AngiogenesisS1P1Akt, EL, ERK
VasorelaxationS1P1, S1P3Akt, AMPK, Ca2+, EL, eNOS, ERK
Inhibition of monocyte adhesionS1P1Akt, NF-κB
Barrier enhancementS1P1Akt, ERK, FAK?
TGF-β expression?Akt, ERK, Smad2/3
PTX3 expressionS1P1, S1P3Akt
Smooth muscle cellsProliferationS1P2?ERK, Raf
Inhibition of migrationS1P2?
PGI2 productionS1P2COX-2, ERK, p38MAPK
Inhibition of ROS generationS1P3NADPH oxidase, Rac1
Guanyl cyclase B desensitization?Akt, ERK
CardiomyocytesSurvivalS1P2, S1P3Akt, ERK, Ras, STAT3
Protection against ischaemia/reperfusion injuryS1P2, S1P3Akt, NO
Gap junctional intercellular communication?Cx43, PKC
MacrophagesInhibition of pro-inflammatory signallingS1P1, S1P2Akt, NF-κB, Ras
  • Akt, protein kinase B; AMPK, AMP-activated kinase; COX, cyclooxygenase; Cx, connexin; EL, endothelial lipase; ERK, extracellular signal-regulated kinase; FAK, focal adhesion kinase; eNOS, endothelial nitric oxide synthase; MAPK, mitogen-activated protein kinase; NO, nitric oxide; NF- κB, nuclear factor κB; PGI2, prostacyclin; PKC, protein kinase C; PTX, pentraxin; STAT, transcription factor signal transducer and activator of transcription; TGF, transforming growth factor.

4.1 Endothelial cells

HDL triggers an array of signalling events in endothelial cells and many of them can be attributed to the HDL-associated S1P cargo. In particular, the activation of extracellular signal-regulated kinase (ERK), AMP-activated kinase (AMPK), and PKB or Akt is crucial to the S1P-dependent signalling.18,85 As demonstrated by Kimura et al.,18 HDL-associated S1P modulates endothelial cell proliferation and survival by activating ERK. Using antisense oligonucleotides, siRNA, and pharmacological inhibitors of trimeric G-proteins, these authors further showed that both S1P1 and S1P3 contribute to HDL-mediated induction of endothelial migration, while S1P1 primarily accounts for the stimulation of endothelial survival.80 Two later studies by Miura et al.81 revealed that HDL-associated S1P promotes endothelial cell tube formation via the activation of MAPK cascade, and that enrichment of HDL particles with exogenous S1P additionally stimulates the angiogenic response.86

It has been convincingly demonstrated that several anti-atherogenic responses induced by HDL-associated S1P may be attributed to the activation of Akt. Nofer et al.87 demonstrated that HDL and the associated S1P switched off the pro-apoptotic protein Bad via Akt stimulation, thereby inhibiting the disruption of mitochondrial potential, the release of cytochrome C, the activation of caspases 3 and 9, and the consequent ignition of apoptosis. Moreover, HDL-associated S1P partially accounts for the HDL-induced eNOS stimulation, NO generation, and the ensuing vasorelaxation,76,78,85 and these effects appear to be mediated by S1P1 and/or S1P3. Interestingly, the administration of fenofibrate to mice increased HDL and S1P plasma levels, as well as S1P1 and S1P3 receptor expression, which both culminated in the enhanced activation of Akt and eNOS.88 Similarly, the pharmacological stimulation of endothelial cells with statins, such as simvastatin and pitavastatin, up-regulated S1P1 expression and increased the HDL-induced eNOS activity.79,89 As reported by Kimura et al.,78 S1P1 mediates the inhibitory effects of HDL on the endothelial expression of adhesion molecules, such as vascular cell adhesion molecule-1 (VCAM1) and intercellular adhesion molecule-1 (ICAM1), and this is related to down-regulation of the NF-κB activity. A study by Argraves et al.82 strongly supports the role of S1P–S1P1 axis and Akt activation in sustaining the endothelial barrier integrity. As pointed above, Wilkerson et al.42 demonstrated that HDL-associated S1P is more effective than albumin-bound S1P in mediating long-term barrier effect, and this is related to the inhibitory effect of HDL-S1P on S1P1 degradation.

Norata et al.90,91 found that HDL potently induces the endothelial expression of long pentraxin 3 (PTX3), an acute-phase response protein, and transforming growth factor β (TGFβ), which exerts potent anti-inflammatory functions and modulates immune responses, in an Akt-dependent manner. They attributed these effects to the stimulation of endothelial S1P1 and S1P3 by HDL-associated S1P. Moreover, the increased activation of Smad 2/3, a transcription factor regulated by TGFβ, suggests that the HDL-associated S1P may transactivate TGFβ signalling pathways, thus acting as an autocrine signalling molecule.86 A recent study suggests that endothelial lipase (EL) is crucial for the HDL-stimulated S1P-dependent signalling in endothelial cells.92 It has been demonstrated that the migration, the angiogenic response, and the Akt activation were defective in EL-lacking endothelial cells. Importantly, exogenous S1P could restore these HDL-induced responses. With help of various pharmacological inhibitors, S1P1 could be identified as a receptor mediating the EL-dependent effects of HDL on endothelial cell migration and Akt activation.92

4.2 Smooth muscle cells

HDL strongly promotes smooth muscle cell (SMC) proliferation and S1P cargo of HDL appears to account for this mitogenic effect. Moreover, HDL-associated S1P appears to modulate SMC migration, a hallmark of advanced atherosclerosis. Actually, HDL was found to inhibit SMC migration and this effect was likely mediated by S1P, as it was abrogated by S1P2-specific siRNA and significantly enhanced in S1P2-overexpressing cells.83,93 Furthermore, HDL-associated S1P was found to modulate the activity of several vasoactive factors targeting SMC. For instance, Chrisman et al.94 demonstrated that HDL-bound S1P desensitizes guanyl cyclase B, a C-type natriuretic peptide (CNP) receptor, thereby blocking the CNP-induced accumulation of cGMP in vascular SMC. When applied to vascular SMC, HDL activates S1P2, S1P3, and the downstream p38 MAPK signalling pathway, inducing a strong up-regulation of COX2 and PGI2 production.95 The latter study also showed that simvastatin enhances the HDL- and S1P-induced COX2 expression by increasing cellular amount of S1P3. Finally, Tölle et al.77 reported that HDL-associated S1P suppressed thrombin-induced pro-inflammatory activation of SMC and this effect was mediated by the inhibition of Rac-1-dependent activation of NADPH oxidase and generation of reactive oxygen species. S1P3 was identified to mediate these anti-inflammatory effects of HDL-associated S1P in SMC.

4.3 Cardiomyocytes

A number of studies established that cardiomyocytes express several S1P receptors, and that HDL may exert protective functions on these cells through its S1P content. HDL and S1P effectively protect cardiomyocytes from hypoxia–reoxygenation damage or doxorubicin-induced apoptosis, and these effects were attributed to the induction of protein kinases Src, ERK, Akt, and the transcription factor STAT3.84,96,97 However, the identity of S1P receptors responsible for these protective effects has not been unequivocally established so far, as the evidence equally supports the involvement of S1P1, S1P2, or S1P3. A recent study showed that short-term treatment of cardiomyocytes with HDL or S1P stimulates the PKC-dependent phosphorylation of connexin43, a gap junction protein involved in cardioprotection, which modulates ischaemic preconditioning in ventricular cardiomyocytes.98 Moreover, both HDL and S1P ameliorated gap junctional communication, although in cardiomyocyte conduction velocities were only incrementally affected. Ex vivo studies confirmed the physiological relevance of these cellular effects by demonstrating that both HDL and S1P protect isolated hearts against ischaemia/reperfusion damage and cell death.

4.4 Macrophages

Although the modulation of the S1P signalling in macrophages, which involves S1P receptors Types 1–4 and activation of several trimeric and small G proteins and kinases, is critical in many pathological inflammatory conditions, such as pancreatitis, allergic asthma, and tumour angiogenesis,99 evidence of modulatory effect of HDL-associated S1P on this cells in the context of atherogenesis is still lacking. One study showed that both HDL and S1P prevented the inflammatory activation of murine monocytes challenged with agonists of Toll-like receptor 2 (TLR2),100 but the specific attribution of this effect to the S1P cargo of HDL particles needs further clarification.

5. S1P-mediated anti-atherogenic effects: in vivo findings

Whereas investigations in vitro allowed to clearly attributing several potentially anti-atherogenic effects of HDL to its S1P cargo, the evidence pointing to atheroprotective action of this lysosphingolipid in vivo is much more limited and partly ambivalent. The effects of S1P on atherogenesis in vivo have been examined by different approaches, for example by using specific S1P receptor-deficient mouse models or mice, in which plasma S1P levels were increased or decreased through genetic or pharmacological manipulation. An initial study87 demonstrating that HDL-induced NO release and vasodilation were both partially abolished in S1P3 knock-out mice pointed to this receptor as a potential mediator of HDL effects related to its S1P content. This notion has been further strengthened by the findings of Theilmeier et al.101, who observed that intravenous injection of HDL or S1P significantly attenuated the size of infarction in a mouse model of ischaemia–reperfusion injury and that the beneficial effects of both HDL and S1P were completely absent in S1P3-deficient mice. However, in a subsequent study Keul et al.102 demonstrated that S1P3 deficiency did not affect the development of atherosclerosis, while dramatically reducing monocyte/macrophage content in the lesions of S1P3/apoE double-deficient mice. Moreover, upon thioglycollate-induced peritonitis S1P3-deficient macrophages displayed both the reduced monocyte migration to the inflamed site and the decreased secretion of monocyte chemoattractant protein 1. At the same time, however, S1P3 deficiency increased SMC content and neointima formation in the carotid artery ligation mouse model.102 These results suggest that S1P3 may simultaneously exert pro- and anti-atherogenic effects in macrophages and SMCs, respectively, leading to a neutral effect on atherosclerosis development in vivo. Skoura et al.103 focused on the role of S1P–S1P2 axis in atherogenesis. They found that S1P2/apoE double knock-out mice developed significantly less lesions than controls and that the number of macrophage-like foam cells in atherosclerotic plaques was reduced, suggesting that S1P2 receptor may promote macrophage recruitment and retention in the atherosclerotically changed arterial wall. In addition, endotoxin-induced pro-inflammatory cytokines (IL-1β and IL-18) were profoundly reduced in sera obtained from S1P2 knock-out mice.103 These results are generally congruent with the pro-atherogenic and pro-inflammatory role of S1P2 receptor, if expressed in haematopoietic cells. In contrast, activation of S1P2 was found to suppress SMC growth in arteries and to promote the expression of SMC differentiation genes, thus establishing S1P as a negative regulator of neointimal development.104,105 These findings suggest that similar to S1P3, S1P2 may play opposing roles in the development of atherosclerosis depending on the site of expression.

Similar to studies employing animal models with defective S1P receptor signalling, experimental approaches focusing at modulating plasma levels of endogenous S1P led to conflicting results. For instance, Bot et al.106 investigated the impact of haematopoietic S1P lyase (Sgpl1−/−) deficiency on leucocyte subsets relevant to atherosclerosis. They found that Sgpl1−/− bone marrow transplantation into atherosclerosis-prone LDL receptor knock-out (LDLR−/−) mice disrupted S1P gradients and significantly elevated S1P levels in plasma. As a consequence, Sgpl1−/− chimeras displayed lymphopenia and loss of T-cell mitogenic and cytokine response when compared with controls. Chimeras presented also with monocytosis and cytokine expression patterns compatible with classical macrophage activation. Despite these apparently pro-inflammatory phenotypic features, haematopoietic S1P lyase deficiency ultimately resulted in diminished atherogenic response, suggesting that increasing S1P concentration in plasma produces anti-atherogenic effects in LDLR−/− mice. However, due to parallel accumulation of S1P precursors such as sphingosine in Sgpl1−/− chimeras, which additionally compounded the impact of S1P lyase deficiency on the distribution and function of haematopoietic cells, the apparently atheroprotective action of elevated S1P in this model has to be interpreted with caution. Recently, we have undertaken an opposite approach and examined whether a decrement of S1P plasma levels may affect the development of atherosclerosis in LDLR−/− mice fed cholesterol-rich diet. Animals treated with ABC294640, a synthetic SphK inhibitor, presented with ∼30% reduced S1P levels in plasma, but displayed no change in atherosclerotic plaque size and macrophage content.107 While the reasons for the neutral effect of SphK inhibitor on the atherosclerosis development are not entirely clear, they may be related to fact that blocking this enzyme not only decreases plasma S1P concentrations, but also interferes with the intracellular signalling mediated by S1P. Actually, we found increased concentrations of pro-inflammatory cytokines (IL-12p70, RANTES) in plasma and the enhanced activation of dendritic and T cells in ABC294640-treated mice and attributed these effects to the reduction in extracellular S1P levels. In contrast, pro-inflammatory effects mediated by intracellular S1P in endothelial cells were suppressed in LDLR−/− mice after ABC294640 treatment, as evidenced by decreased plasma levels of endothelial activation markers (VCAM1 and ICAM1), as well as down-regulated in vivo leucocyte adhesion to endothelial lining and the CCL19-induced T-cell penetration into peritoneum.107

The recent availability of S1P mimetic compounds interacting with selected S1P receptor subtypes opened new avenues of investigations, aiming at identification of S1P signalling pathways specifically involved in the protection against atherosclerosis. In initial studies, Nofer et al.108 and Keul et al.109 demonstrated that FTY720, a synthetic S1P analogue targeting all S1P receptors except for S1P2, reduced atherosclerosis when administered to LDLR−/− or apoE−/− mice on cholesterol-rich Western diet. In these studies, atheroprotective effects of FTY720 were principally attributed to the capacity of this compound to modulate the distribution and the activation of T cells and macrophages at both peripheral and plaque levels and to reduce the overall inflammatory processes, as evidenced by decreased pro-inflammatory cytokine levels in plasma. However, no effect of FTY720 on atherosclerosis development and plasma inflammation markers was observed in apoE−/− mice fed a normal diet, despite significant reduction of T cell number in blood and peripheral lymphoid organs.110 In addition, we reported that FTY720 fails to affect atherosclerosis in moderately hypercholesterolemic LDLR−/− mice in spite of down-regulating macrophage function, improving metabolic control, and persistently altering lymphocyte distribution.111 It seems, therefore, that the atheroprotective effects of FTY720 effectively unfold only in the setting of pre-existing chronic inflammation, which may be boosted by persistent hypercholesterolaemia. Very recently long-term FTY720 treatment was found to enhance left ventricular function and to increase longevity in atherosclerotic mice with heart failure.112 However, these benefits were primarily attributed to systemic immunosuppression and a moderate reduction of inflammation in the heart rather than atheroprotection.

Although extensively supported by in vitro findings (see above), the role of S1P–S1P1 axis in atherogenesis has not been as yet explored in vivo. Gene knock-out manipulation revealed that S1P1 deficiency is embryonic lethal between E12.5 and E14.5 dpc,113 thus hampering the possible exploitation of these model for studying atherogenesis. However, by using the selective S1P1 agonist KRP203, we recently provided the first evidence that S1P1 receptor may play a prominent role in the atheroprotection in vivo.114 In LDLR−/− mice, both early and advanced lesions were significantly reduced by KRP203 treatment when compared with controls. Concomitantly, KRP203 reduced total and activated T cells in peripheral lymphoid organs as well as pro-inflammatory cyto- and chemokines levels in plasma and aorta. Moreover, macrophages isolated from KRP203-treated animals showed lower expression of MHC-II activation marker as well as reduced secretion of pro-inflammatory cytokines in response to TLR4 or TLR3 stimulation. In addition, concentrations of endothelial activation markers (VCAM1 and ICAM1) were reduced in plasma obtained from KRP203-treated animals and both their expression and monocyte adhesion were down-regulated in endothelial cells in the presence of this compound. These results are consistent with the contention that signalling through S1P–S1P1 axis may contribute to the atheroprotective effects of S1P under in vivo conditions.

The potential contribution of various S1P receptor subtypes to the progression or regression of atherosclerosis is schematically illustrated in Figure 2.

Figure 2

Anti- and pro-atherogenic effects of HDL-associated S1P—findings in vivo. Studies in animal models of atherosclerosis provide evidence supporting both anti- and pro-atherogenic effects of S1P depending on the S1P receptor subtype and the targeted cell/tissue. IFN, interferon; IL, interleukin; IL1RA, IL1 receptor antagonist; MCP, monocyte chemoattractant protein; NO, nitric oxide; TLR, Toll-like receptor; TNF, tumour necrosis factor.

6. Conclusion and future perspectives

The research of last two decades has firmly established S1P as an integral constituent of HDL particles. Owing to extraordinary wide spectrum of biological activities exerted by S1P and the promiscuity of its receptors, this lysosphingolipid has become a natural candidate as a mediator of HDL-related atheroprotection. Actually, several potentially anti-atherogenic effects carried out by HDL particles in vitro could be fully or partially attributed to their S1P cargo. However, the results of studies addressing the influence of S1P on the development of atherosclerosis under in vivo conditions have been less conclusive. Both anti- and pro-atherogenic effects have been observed in animal models of atherosclerosis manipulated to enhance or decrease signalling through S1P receptors and even single S1P receptor subtypes were found to promote or retard the development of atherosclerotic lesions depending on the site and level of expression, the choice of experimental model, and pharmacological treatment. In addition, epidemiological studies addressing the interrelationship between the HDL-bound S1P and atherosclerosis are still scarce, conducted in small populations and in a cross-sectional design, while the important issue, whether HDL-bound S1P contributes to the reduction of cardiovascular risk, has not been investigated to date in a prospective fashion. In our view, the current state of knowledge does not yet allow considering S1P or its mimetics as potential therapeutics for CVD. It nevertheless mandates further research effort to improve our understanding, how HDL-bound S1P modulates cardiovascular risk and how signalling pathway triggered by S1P influences various aspects of atherosclerotic plaque development.

Conflict of interest: none declared.


Preparation of this manuscript was supported by a grant IDEAS RBID08777T from the Italian Ministry of Education, Universities and Research to J-R.N. and M.S., and intramural resources of the Center for Laboratory Medicine to J-R.N.


View Abstract