Elevating Endogenous Sphingosine-1-Phosphate (S1P) Levels Improves Endothelial Function and Ameliorates Atherosclerosis in Low Density Lipoprotein Receptor-Deficient (LDL-R−/−) Mice

Abstract Background Sphingosine-1-phosphate (S1P) is a bioactive lysosphingolipid and a constituent of high-density lipoprotein (HDL) exerting several atheroprotective effects in vitro. However, the few studies addressing anti-atherogenic effects of S1P in vivo have led to disparate results. We here examined atherosclerosis development in low-density lipoprotein receptor (LDL-R)-deficient (LDL-R−/−) mice with elevated endogenous S1P levels. Methods and Results Sub-lethally irradiated LDL-R−/− mice were transplanted with bone marrow deficient in sphingosine kinase 2 (SphK2), which led to the elevation of S1P concentrations in erythrocytes, plasma and HDL by approximately 1.5- to 2.0-fold in SphK2−/−/LDL-R−/− mice. Afterwards, mice were fed a Western diet for 14 weeks. Elevation of endogenous S1P significantly reduced atherosclerotic lesion formation by approximately half without affecting the plasma lipid profile. Furthermore, the macrophage content of atherosclerotic lesions and lipopolysaccharide-induced monocyte recruitment to the peritoneal cavity were reduced in SphK2−/−/LDL-R−/− mice. Studies using intra-vital microscopy revealed that endogenous S1P lowered leukocyte adhesion to capillary wall and decreased endothelial permeability to fluorescently labelled LDL. Moreover, SphK2−/−/LDL-R−/− mice displayed decreased levels of vascular cell adhesion molecule 1 in atherosclerotic lesions and in plasma. Studies in vitro demonstrated reduced monocyte adhesion and transport across an endothelial layer exposed to increasing S1P concentrations, murine plasma enriched in S1P or plasma obtained from SphK2-deficient animals. In addition, decreased permeability to fluorescence-labelled dextran beads or LDL was observed in S1P-treated endothelial cells. Conclusion We conclude that raising endogenous S1P levels exerts anti-atherogenic effects in LDL-R−/− mice that are mediated by favourable modulation of endothelial function.


Introduction
Sphingosine-1-phosphate (S1P) is a membrane-derived sphingolipid that regulates diverse cellular processes including growth, survival, migration, angiogenesis and inflammation. 1,2 S1P is synthesized intra-cellularly via phosphorylation of sphingosine by two distinct sphingosine kinases (SphK), SphK1 and SphK2, and was initially postulated to act as a second messenger. 1,2 Subsequent studies revealed that S1P is actively transported across the plasma membrane to the extra-cellular space, where it interacts with five cognate cellsurface receptors designated S1P [1][2][3][4][5] . 3 Although widely expressed, S1P receptors display tissue-specific distribution patterns with S1P 1 , S1P 2 and S1P 3 being represented in vascular tissues. In endothelial cells, S1P was found to promote adherence junction assembly resulting in the enhancement of endothelial barrier. 4,5 In addition, S1P decreased the expression of endothelial adhesion molecules such as vascular cell adhesion molecule 1 (VCAM1), and thereby abrogated the adhesion of monocytic cells to endothelial monolayer. [6][7][8] In smooth muscle cells, S1P suppressed the production of chemoattractant chemokines. 8,9 Consequently, S1P was postulated to exert anti-inflammatory effects by restricting the recruitment of leukocytes to sites of inflammation. Actually, lowering S1P concentration in plasma or defective expression of S1P receptors in endothelial cells exacerbated inflammatory response in murine models of ischaemia-reperfusion injury, acute lung injury and anaphylaxis, while opposite effects were seen after restoration of plasma S1P levels or administration of S1P receptor agonists. 1,4,10,11 Erythrocytes, platelets and endothelial cells are major sources of S1P in plasma, where it is mainly associated with an apolipoprotein M (apoM)-containing sub-fraction of highdensity lipoprotein (HDL)-a potent plasma-borne antiatherogenic factor. 1,4,12 Growing evidence indicates that S1P accounts for a substantial portion of atheroprotective effects attributed to HDL. For instance, S1P closely correlates with HDL in a concentration range, in which HDL most effectively protects against atherosclerosis, and decreased HDL-bound S1P levels were noted in patients with coronary artery disease and myocardial infarction. [13][14][15] Under in vitro condition, HDLbound S1P was reported to mitigate endothelial apoptosis, stimulate nitric oxide and prostacyclin generation in endothe-lial and smooth muscle cells, inhibit expression of endothelial adhesion molecules and produce persistent enhancement of endothelial barrier and reduction of vascular leak. 16,17 In addition, synthetic S1P mimetics, such as FTY720-a high affinity agonist for S1P 1,3,4,5 , and KRP203-a specific S1P 1 agonist diminished atherosclerotic lesions in low-density lipoprotein (LDL) receptor-deficient (LDL-R À/À ) or apoE-deficient (apoE À/À ) mice fed a high cholesterol diet, improved endothelial function by reducing leukocyte adhesion and suppressed inflammatory activation of macrophages. [18][19][20] More recently, cell-specific deletion of S1P 1 in endothelial cells or macrophages was shown to enhance atherosclerotic development in apoE À/À or LDL-R À/À mice, respectively. 21,22 By contrast, reduced vascular lesion formation was observed in apoE À/À mice deficient in S1P 2 or treated with synthetic S1P 2 antagonist. 23 Moreover, elevation of plasma S1P in apoE À/À mice over-expressing apoM augmented aortic root but not aortic arch atherosclerosis, and this effect was abolished under conditions of uraemia. 24 Thus, the involvement of S1P in the atherosclerosis development appears to critically depend on the S1P receptor involved, the localization of lesions as well as the experimental setting, and anti-atherogenic effects of HDLbound S1P remain controversial.
The objective of this study was to assess the impact of endogenously synthesized S1P on the atherosclerotic lesion development. Our findings demonstrate that LDL-R À/À mice with haematopoietic SphK2 deficiency, which are characterized by elevated S1P levels in plasma, display attenuated development of atherosclerotic lesions, and attribute this beneficial effect to the enhancement of endothelial function.

Materials and Methods
Animals Generation of B6N.129S6-Sphk2 tm1Rlp /J mice deficient in SphK2 (SphK2 À/À ) was described elsewhere. 25 Female LDL-R À/À mice (B6.129S7-Ldlr tm1Her /J) were purchased from Jackson Laboratories, Bar Harbor, Maine, United States. Bone marrow (BM) aplasia was induced in LDL-R À/À mice (6-8 weeks of age) with a single dose of 11 Gy total body irradiation. Single-cell BM suspensions from SphK2 À/À and wild-type (WT) mice were injected intravenously to irradiated recipients (5.0 Â 10 6 cells/ animal). The haematological chimerism of transplanted animals adhesion to capillary wall and decreased endothelial permeability to fluorescently labelled LDL. Moreover, SphK2 À/À /LDL-R À/À mice displayed decreased levels of vascular cell adhesion molecule 1 in atherosclerotic lesions and in plasma. Studies in vitro demonstrated reduced monocyte adhesion and transport across an endothelial layer exposed to increasing S1P concentrations, murine plasma enriched in S1P or plasma obtained from SphK2-deficient animals. In addition, decreased permeability to fluorescence-labelled dextran beads or LDL was observed in S1P-treated endothelial cells. Conclusion We conclude that raising endogenous S1P levels exerts anti-atherogenic effects in LDL-R À/À mice that are mediated by favourable modulation of endothelial function.
Thrombosis and Haemostasis Vol. 118 No. 8/2018 was determined in genomic deoxyribonucleic acid (DNA) from blood leukocytes 4 weeks after transplantation. Thereafter, animals were put on Western diet (0.25% cholesterol, 21% fat; Altromin, Lage, Germany) for 14 weeks (assessment of atherosclerosis) or 4 weeks (assessment of endothelial function), sacrificed and used for further analysis. All animal protocols used in this study conformed to the national law and were approved by the animal protection authority (LANUV).

Lipid Analysis and Lipoprotein Isolation and Fractionation
Plasma total cholesterol (TC) and triglycerides (TG) were determined enzymatically (Siemens, Eschborn, Germany). LDL and HDL were isolated from plasma by a discontinuous gradient centrifugation. LDL was labelled with DyLight 594 fluorescent dye (DyL, Thermo Fischer, Schwerte, Germany) as recommended by the manufacturer. Plasma lipoproteins were fractionated using Smart chromatographic system (Pharmacia, Uppsala, Sweden) as described previously. 19 S1P concentrations were determined using liquid-chromatography tandem mass spectrometry as published previously. 26

Assessment of Leukocyte Adhesion In Vivo
Leukocyte adhesion in vivo was studied in anaesthetized LDL-R À/À mice transplanted with SphK2 À/À or WT BM. In vivo leukocyte staining was performed by intravenous injection of 1.0 µg/g rhodamine 6G (Sigma, Deisenhofen, Germany) and leukocyte adhesion was assessed by intra-vital microscopy of the mesenterial venules using inverted fluorescence microscope (Eclipse 300, Nikon, Düsseldorf, Germany) as described previously. 27

Assessment of Vascular Permeability and Monocyte Recruitment
To assess vascular permeability, LDL-R À/À mice transplanted with SphK2 À/À or WT BM were intravenously administered with Evans blue, fluorescein isothiocyanate (FITC)-dextran (500.0 kDa), or DyL-labelled LDL (DyL-LDL) 15 minutes prior to injection intraperitoneally of lipopolysaccharide (LPS, 25 µg/ animal), sacrificed after 3 hours and dye concentrations were measured in peritoneal lavage fluid by photometry or fluorescence spectrometry. For the assessment of monocyte recruitment to peritoneal cavity, LPS-injected animals were sacrificed after 18 hours and monocytes in the lavage fluid were stained with anti-F4/80 and anti-CD11b antibodies and counted using a FACSCalibur flow cytometer (BD Bioscience, San Jose, California, United States). Vascular permeability in situ was assessed in the ileal mesentery superfused with bradykinin (1.0 µg/mL). The DyL-LDL extravasation was monitored for 15 minutes by intra-vital microscopy and quantified using the image analysis software ImageJ (National Institute of Mental Health, Bethesda, Maryland, United States).

Determination of Cytokine and Adhesion Molecule Concentrations
Concentrations of cytokines and soluble adhesion molecules in plasma and cell supernatants were quantified by commercially available enzyme-linked immunosorbent assays (R&D Systems, Wiesbaden, Germany).

Endothelial Adhesion and Permeability Assays
For the assessment of monocyte adhesion, U937 monocytes labelled with calcein-acetoxymethyl were added to confluent bEnd.5 murine endothelial cells on cover slip in a Dulbecco-modified Eagles medium supplemented with glutamine (2.0%, v/v), sodium pyruvate (1.0%, v/v), non-essential amino acids (1.0%, v/v) and endothelial cell growth supplement (Promocell, Heidelberg, Germany) containing epidermal growth factor, basic fibroblast growth factor and foetal calf serum (2.0%, v/v). S1P (1.0 µmol/L or 2.0 µmol/L in some experiments) was added directly to the cell culture media. The number of adherent cells was counted under fluorescence microscope Leica DM-IRE (Leica Mikrosysteme, Wetzlar, Germany). For endothelial permeability testing, bEnd.5 cells were seeded onto collagen-coated TransWell culture inserts (Corning Life Sciences, Lowell, Massachusetts, United States) to form monolayers. FITCdextran, DyL-LDL or calcein-labelled U937 monocytes were placed in the apical insert compartment for 0.5 or 4 hours. Samples were taken from basolateral chambers for fluorescence measurement.

Real-Time Quantitative Reverse Transcription Polymerase Chain Reaction
Total ribonucleic acid (RNA) was isolated from bEnd.5 murine endothelial cells using RNAeasy Plus Purification Kit (Qiagen, Hilden, Germany) and complementary DNA was synthesized by reverse transcription. Fully automated reverse transcription polymerase chain reaction (RT-PCR) set-up was done on a Genesis 150 workstation (TECAN, Creilsheim, Germany) and PCR products were detected using the ABI7900ht sequence detection system (Applied Biosystems, Darmstadt, Germany). Relative gene expression was calculated by applying the 2 -ΔΔCt method.

General Procedures
Data are presented as means AE standard deviation for at least three separate experiments or as results representative of at least three repetitions. Comparisons between the means of two or multiple groups were performed with two-tailed Student's t-test or one-way analysis of variance for independent samples, respectively. Pairwise comparisons were performed with Student-Newman-Keuls post hoc test. p-Values less than 0.05 were considered significant. Detailed methods can be found online.

Haematopoietic SphK2 Deficiency Elevates Plasma S1P Level
SphK2-deficient mice display increased S1P concentrations in plasma. 30 To elevate plasma S1P levels in atherosclerosisprone animals, sub-lethally irradiated LDL-R À/À mice were reconstituted with either SphK2 À/À or WT BM. Analysis of SphK2 À/À transplanted animals revealed that over 90% of blood cells were derived from SphK2 À/À BM (►Fig. 1A). Body weight at sacrifice, plasma levels of TC and TG as well as plasma lipid profiles were comparable between SphK2 À/À and WT transplanted mice (►Fig. 1B and C). However, haematopoietic SphK2 À/À deficiency led to an approximately 1.5-to 2-fold increase in S1P concentration in plasma and significantly raised the S1P content in erythrocytes (►Fig. 1D). In addition, SphK2 deficiency elevated the amount of S1P associated with HDL but not LDL or very low density lipoprotein (VLDL) particles (►Fig. 1D).

Haematopoietic SphK2 À/À Chimeras Show Reduced Atherosclerosis
To determine the effects of haematopoietic SphK2 À/À deficiency on atherosclerosis development, Oil Red O-stained lesions in the aortic root and thoracic aorta of Western diet-fed LDL-R À/À chimeras were analysed. Morphometric quantification of atherosclerosis at the aortic root revealed that both absolute lesion size and the plaque-to-lumen ratio were significantly decreased in the SphK2 À/À transplanted mice as compared with WT transplanted mice (►Fig. 2A). In addition, SphK2 À/À chimeras displayed significantly smaller necrotic cores within aortic lesions. Analysis of en face prepared thoracic aortas demonstrated a remarkable reduction of Oil Red O-positive lesions around branch ostia (►Fig. 2B). Immunohistochemical analysis of lesion composition in the aortic root yielded a significant reduction of MOMA-2-positive macrophage content in SphK2 À/ À chimeras as compared with WT transplanted controls (►Fig. 2C). By contrast, SphK2 À/À chimerism did not affect the collagen content in the atherosclerotic lesions (not shown).

Haematopoietic SphK2 Deficiency Reduces Leukocyte Adhesion
As leukocyte-endothelial interactions represent a pivotal target of anti-atherogenic action of HDL-bound S1P, the influence of haematopoietic SphK2 deficiency on leukocyte was directly assessed in vivo in post-capillary venules. To this aim, leukocytes were intra-vitally labelled with fluorescence dye (rhodamine) in SphK2 À/À and WT transplanted mice and cells adhering to the vascular wall were observed using intravital microscopy. We found that SphK2 À/À chimerism substantially reduced leukocyte-endothelial interaction: the number of leukocytes rolling on vascular endothelium was significantly reduced in SphK2 À/À transplanted LDL-R À/À mice as compared with controls and permanent adhesion showed a trend towards reduction (►Fig. 3A). Concomitantly, circulating levels of soluble isoforms of the endothelial adhesion molecule VCAM1 were significantly lower in the plasma of SphK2 À/À transplanted LDL-R À/À mice (►Fig. 3B). In addition, atherosclerotic lesions of SphK2 À/À chimeras exhibited less expression of VCAM1 (►Fig. 3C).
To investigate in vitro correlates of anti-adhesive effects, which are exerted by elevated S1P in mice with haematopoietic SphK2 deficiency, bEnd.5 murine endothelial cells were preincubated with medium containing S1P in concentrations approximating those seen in WT and SphK2 À/À transplanted mice (1.0 and 2.0 µmol/L, respectively). After 24 hours, cells were exposed to 100 U/mL of tumor necrosis factor (TNF)α and the extent of U937 monocyte adhesion to the endothelium was examined using an in vitro cytoadherence assay, while the expression of endothelial adhesion molecules was determined by qPCR. As shown in ►Fig. 3D and E, the adhesion of U937 monocytes to endothelial cells and the expression of VCAM1 were both reduced in a concentration-dependent fashion after pre-incubation with S1P. In addition, augmented down-regulation of monocyte adhesion and VCAM1 expression was observed after pre-incubation of endothelial cells with WT mouse plasma enriched with S1P (1.0 µmol/L), or with plasma obtained from SphK2-deficient mice (►Fig. 3D and E), indicating that the increment in S1P quantity in plasma is associated with augmentation of the inhibitory effects on monocyte adhesion.

Haematopoietic SphK2 Deficiency Enhances the Endothelial Barrier
Next, we assessed the effect of elevated S1P concentrations in haematopoietic SphK2 deficiency on the function of the endothelial barrier in vivo. To this purpose, alterations in vascular permeability were estimated by intravenous injection of Evans blue dye, which binds to serum proteins, high molecular weight FITC-dextran or DyL-LDL, and quantification of the respective dye after 3 hours in the peritoneal cavity of SphK2 À/À and WT transplanted mice. Intra-peritoneal injection of LPS produced a substantial influx of Evans blue, FITC-dextran and DyL-LDL into the peritoneal cavity and each of these effects was substantially attenuated in SphK2 À/À chimeras as compared with WT transplanted mice (►Fig. 4A). In addition, the effect of haematopoietic SphK2 deficiency on endothelial permeability for DyL-LDL was examined in situ in mesenterial capillaries observed with intra-vital microscopy. As shown in ►Fig. 4B, DyL-LDL was entirely localized in capillaries both in SphK2 À/À and WT transplanted mice, but underwent massive extravasation after exposure to bradykinin, which compromises the endothelial barrier and enhances the vascular permeability.  concentrations in plasmas obtained from SphK2 À/À -or WT-transplanted LDL-R À/À mice (n ¼ 11 per group). (C) Fluorescence photomicrographs showing VCAM1 staining (green fluorescence) in aortic roots obtained from SphK2 À/À or WT chimeras (both n ¼ 11). Blue fluorescencecounterstaining with 4',6-diamidino-2-phenylindole (DAPI). Bar graph shows the quantification of VCAM1 in atherosclerotic lesions expressed as the percentage of intimal area. Ã p < 0.05, ÃÃ p < 0.01 (SphK2 À/À vs. WT). (D) bEnd.5 murine endothelial cells pre-treated for 24 hours with sphingosine-1-phosphate (S1P) (1.0 or 2.0 µmol/L, left bar graph), WT plasma enriched with S1P (1.0 µmol/L, centre bar graph) or plasma obtained from SphK2 À/À mice (right bar graph) were stimulated with tumor necrosis factor (TNF)α (50.0 ng/mL) for 4 hours. The adherence of calcein-loaded U937 monocytes was evaluated under a fluorescence microscope. (E) bEnd.5 cells pre-treated as described above were examined for VCAM1 expression using quantitative polymerase chain reaction (qPCR). Data are representative for three to five independent experiments.
Quantification of DyL-LDL in capillary vessels after bradykinin treatment revealed that significantly more DyL-LDL was retained intra-capillarily in the SphK2 À/À chimeras as compared with WT transplanted mice (►Fig. 4B).
To investigate the influence SphK2 À/À chimerism on monocyte capacity to migrate across the endothelial barrier, the recruitment of monocytes into the peritoneal cavity was additionally examined in SphK2 À/À and WT transplanted mice after intraperitoneal injection of LPS. As shown in ►Fig. 4C, accumulation of monocytes in the peritoneal cavity 18 hours after LPS injection was significantly reduced in haematopoietic SphK2 À/À deficiency.
To assess the effect of S1P on the function of the endothelial barrier in vitro, confluent bEnd.5 murine endothelial cells on TransWell culture inserts were exposed for 24 hours to medium containing increasing S1P concentrations (1.0 or 2.0 µmol/L), WT mouse plasma enriched with S1P (1.0 µmol/L) or plasma obtained from SphK2-deficient or WT mice. Thereafter, endothelial permeability to FITC-dextran or DyL-LDL was determined. As shown in ►Fig. 4D, increasing the exposure of endothelial cells to S1P consistently decreased the transfer of FITC-dextran or DyL-LDL across the endothelial monolayer irrespective of the exposure ambience. Similarly, migration of U937 monocytes across the endothelial monolayer was gradually inhibited by increasing S1P quantities (►Fig. 4D).

Haematopoietic SphK2 Deficiency Does Not Affect Macrophage Activation
Previous studies attributed anti-atherogenic effects exerted by S1P mimetics to the inhibition of macrophage activation and/or to alterations in T cell distribution and function. 18,19 Therefore, we tested inflammatory responses in peritoneal macrophages isolated from SphK2 À/À chimeras and WT transplanted mice. Both, basal and LPS-stimulated secretion of pro-inflammatory cyto-and chemokines (TNFα, monocyte chemoattractant protein 1 [MCP1], interleukin [IL]-12p70) was comparable between SphK2 À/À and WT macrophages, which also showed similar expression levels of major histocompatibility complex class II (MHC-II) and CD86-the surface markers of the proinflammatory macrophage activation (►Supplementary Fig. S1A, available in the online version) and similar monocyte blood count (►Supplementary Fig. S1C, available in the online version). Moreover, SphK2 À/À chimerism exerted no effect on the lymphocyte count in blood and on CD4 þ and CD8 þ T cell distribution in blood and spleen (►Supplementary Fig. S1B and C, available in the online version). Splenocytes obtained from SphK2 À/À or WT transplanted mice showed similar response to concanavalin A stimulation as assessed by the determination of IL-2 and interferon-γ (IFNγ) concentrations in cell supernatants (►Supplementary Fig. S1B, available in the online version). In addition, comparable levels of macrophageand T cell-secreted pro-inflammatory cyto-and chemokines (TNFα, IFNγ, IL12p70, MCP1, regulated on activation, normal T cell expressed and secreted) were noted in plasmas from SphK2 À/À chimeras and WT transplanted mice (►Supplementary Fig. S1C, available in the online version).

Discussion
Several studies have demonstrated that synthetic S1P mimetics interacting with S1P receptor types present in vasculature exert anti-atherogenic effects in murine models of atherosclerosis. [18][19][20] Less effort, however, has been devoted towards understanding the relevance of endogenous S1P for the development of atherosclerotic lesions. We have previously reported reduced atherosclerosis in LDL-R À/À mice transplanted with BM deficient in S1P degrading enzyme S1P lyase (Sgpl1), which are characterized by increased S1P levels in plasma. 31 While this finding was in principle consistent with atheroprotective effect exerted by endogenous S1P, the elevation of the plasma S1P concentration in Sgpl1 À/À BM recipients was marginal (approximately 1.2-fold vs. control). By contrast, the haematopoietic Sgpl1 deficiency massively increased S1P content in the spleen and lymph nodes (over 100-fold vs. control), which led to disruption of S1P gradients between blood and lymphatic organs, produced severe lymphopaenia and monocytosis, and profoundly affected T cell and macrophage functions as evidenced by altered proliferation, migration and production of pro-and anti-inflammatory cytokines. In addition, Sgpl1 À/À transplanted mice displayed significantly lower TC and TG levels and less pro-atherogenic plasma lipoprotein profile with decreased cholesterol content in LDL and VLDL fractions. The pleiotropic phenotype seen in in Sgpl1 À/À BM recipients made difficult to pinpoint the mechanism underlying the reduced atherosclerotic lesion formation observed in these animals. In this study, we took advantage of SphK2deficient mice, which display a paradoxical increase in S1P concentrations in plasma and erythrocytes, but not in spleen and lymph nodes. 26,30 The latter effect likely arises because of the predominant involvement of SphK2 in the S1P release from red blood cells and its uptake by endothelial cells and lymphatic tissues rather than S1P production. Actually, impaired S1P transfer from erythrocytes to endothelial cells, but not de novo S1P synthesis was previously observed in SphK2 À/À animals. 30 In our hands, the transplantation of SphK2 À/À BM to LDL-R À/À led to a moderate, but significant increase of S1P level both in plasma and erythrocytes, but failed to produce overt monocytosis, lymphopaenia or changes in CD4 þ and CD8 þ T cell distribution in blood and lymphoid organs, which is concordant with previously published data. 30 In addition, plasma lipid levels and lipoprotein profiles were comparable between mice transplanted with WT and SphK2 À/À BM. Notwithstanding, the elevation of plasma S1P levels in SphK2 À/À chimeras was coupled to the significant reduction of both absolute and fractional area of atherosclerotic lesions in the aortic root and markedly attenuated accumulation of lipid-rich material around the ostia of the intercostal arteries, which represent areas of the aorta pre-disposed to atherosclerosis. In addition, lesions seen in SphK2 À/À transplanted LDL-R À/À mice showed reduced accumulation of macrophages, which are considered a culprit in atherosclerotic plaque build-up. Taken together, these data strongly reinforce the notion that endogenous S1P in plasma exerts anti-atherogenic effects in vivo.
Both macrophages and T cells are abundantly present throughout human and murine plaques and their interaction is required for the development of fully fledged inflammation within the arterial wall. 32,33 Previous investigations exploiting synthetic S1P mimetics such as FTY720 or KRP203 suggested that the anti-atherogenic effects exerted by these compounds in murine models of atherosclerosis are at least partly related to lowering blood lymphocyte count and/or to re-programming macrophages and CD4 þ T cells towards less inflammatory phenotypes (M2, Th2). 18,19 However, this study provides little evidence supporting the notion that similar mechanisms operate in case of elevation of endogenous S1P in plasma. Transplantation of SphK2 À/À BM to LDL-R À/À mice had no impact on blood leukocyte count. In addition, macrophages obtained from SphK2 À/À chimeras showed normal expression of surface activation markers (CD86, MHCII) and their response to pro-inflammatory stimulation was comparable with WT cells. This findings agree with the data published by Xiong et al, who failed to detect any differences between SphK2 À/À and WT macrophages with respect to cellular proliferation and apoptosis as well as nuclear factor kappa B (NF-κB) activation and cytokine expression and concluded that SphK2 is not required for inflammatory responses in macrophages. 34 However, the latter authors observed increased autophagy in macrophages deficient in SphK2. As autophagy is impaired in advanced stages of atherosclerosis and its absence promotes vascular lesion formation in part through activation of the inflammasome, 35 it cannot be excluded that the enhancement of autophagy might additionally contribute to atheroprotective effects seen in the haematopoietic SphK2 deficiency.
Apart from dysregulation of innate and adaptive immune responses, endothelial dysfunction represents an important component of the pathological process leading to the development of atherosclerotic lesions. There is overwhelming evidence that plasma S1P is critically involved in the regulation of several endothelial functions relevant to the protection against atherosclerosis. 4 and compounds promoting endothelial expression of S1P receptors such as simvastatin were repeatedly demonstrated to down-regulate TNFα-induced surface expression of the endothelial adhesion molecule VCAM1 under in vitro conditions and to suppress ex vivo monocyte adhesion to aortic explants. 6,7,18,36,37 As a consequence, administration of S1P receptor agonists to mice was effective in reducing the recruitment of monocytes into inflamed areas of the heart, kidney, liver and peripheral nerves, and these effects were attributed to the immediate modulation of endothelial functions. [38][39][40][41][42] In addition, S1P has been firmly established as the single most potent plasma-borne factor enhancing endothelial barrier integrity and interfering with S1P signalling was demonstrated to promote vascular leak, which could be ameliorated by S1P mimetic administration. 4,5 Our results substantially expand these previous findings in LDL-R À/À mice and provide several pieces of evidence suggesting that the endothelium represents a pivotal target of antiatherogenic actions of endogenous S1P. First, peri-vital microscopic assessment of leukocyte-vessel wall interaction provided the first demonstration in vivo that raising S1P levels in plasma effectively suppresses leukocyte adhesion to endothelium, with the latter effect being likely dependent on the reduced expression of VCAM1. As a possible consequence of the impaired interaction between endothelial cells and leukocytes, reduced extravasation manifesting as a diminished presence of monocytes/macrophages in peritoneal cavities and atherosclerotic lesions was observed in SphK2 À/À chimeras. Second, the elevation of endogenous S1P in SphK2 À/À chimeras effectively enhanced the endothelial barrier as evidenced by the reduced influx of molecular indicators such as Evans blue or FITC-dextran into the peritoneal cavity. Even more important from the perspective of atherosclerotic plaque pathology and for the first time demonstrated by this study was a sharp reduction of vascular permeability for LDL in SphK2 À/À chimeras, which likely translated into abated lipid deposition around the intercostal artery ostia in the descending thoracic aorta. Third, the beneficial effects exerted by endogenous S1P on endothelial functions in mice could be entirely recapitulated using an in vitro approach, which closely emulated incremental increases in S1P concentrations occurring as a result of SphK2 À/À BM transplantation. Collectively, our findings suggest that endogenous S1P favourably affects several endothelial functions relevant to the pathogenesis of atherosclerosis, which culminates in the reduced penetration of both monocytes and LDL particles into the arterial wall. As both, monocyte and LDL retention in the sub-endothelial space, followed by the ingestion of LDL by intimal macrophages constitute crucial steps in the formation of atherosclerotic lesions, we hypothesize that these effects predominantly account for the anti-atherogenic effects of endogenous S1P seen in haematogenous SphK2 deficiency. S1P is carried in plasma by the apoM-containing subfraction of HDL with the remainder being associated with albumin and LDL. 1,2,12 As elevation of S1P level in plasma after SphK2 À/À BM transplantation likely arose as a consequence of the enhanced release from S1P-enriched erythro-cytes, our results do not allow us to attribute the favourable effects on endothelial function observed in SphK2 À/À chimeras specifically to HDL-bound S1P. However, the increased amount of S1P associated with HDL, but not LDL or VLDL in these animals, is consistent with the notion that these lipoproteins might serve as a preferential acceptor for S1P liberated from erythrocytes, particularly as apoM-containing HDL has been recently identified as a principal mediator of S1P efflux from human red cells. 43 In this context, it is worth noting that HDL-bound S1P has been previously identified to act as a biased agonist for S1P 1 . 21 Actually, HDL-bound S1P was found to most effectively promote the S1P 1 -and β-arrestin-2-dependent signalling in endothelial cells and thereby to prevent TNFα-induced NF-κB activation and adhesion molecule expression. Conversely, apoM-deprived HDL, which contains negligible amounts of S1P, was demonstrated to lose its favourable effects on the endothelial functions, which were severely impaired in apoM-deficient mice. [44][45][46] In our experimental setting, in vitro S1P was added to cell culture medium containing traces of lipoproteins, which might additionally facilitate its beneficial effects on endothelial cells. Taking these into account, it would be tempting to speculate that HDL assumes an important role in mediating the protective effects of S1P on endothelial function and against atherosclerosis, which are observed in LDL-R À/À mice transplanted with SphK2 À/À BM.
In conclusion, our study documents for the first time that raising endogenous S1P levels in LDL-R À/À mice exerts antiatherogenic effects and that vascular endothelium represents the major target of the atheroprotective effects exerted by S1P. Further research efforts will be necessary to delineate the molecular mechanisms responsible for the beneficial action of S1P and to identify the relevant S1P receptors.
What is known about this topic?
• Sphingosine-1-phosphate (S1P) is a constituent of high-density lipoproteins (HDL), but its contribution to the anti-atherogenic potential of these lipoproteins remains controversial. • Exogenous S1P mimetics were found to reduce or to exert no effect on atherosclerosis development in mice depending on animal model and treatment conditions.
What does this paper add?
• Transplantation of sphingosine kinase 2-deficent bone marrow to mice prone to atherosclerosis (LDL-R À/À mice) significantly elevates S1P level both in plasma and HDL. • Elevation of endogenous S1P ameliorates diet-induced atherosclerosis development and reduces macrophage accumulation in the arterial wall. • Attenuated monocyte adhesion and reduced transfer of LDL across endothelial barrier in mice with elevated S1P levels points to vascular endothelium as a principal target of atheroprotective effects of S1P in plasma.

Conflict of Interest
None.