**Inflammatory Conditions Disrupt Constitutive Endothelial Cell Barrier Stabilization by Alleviating Autonomous Secretion of Sphingosine 1-Phosphate**

#### **Jefri Jeya Paul 1,2,3, Cynthia Weigel 1,2,4,5, Tina Müller 1,2, Regine Heller 2,3,6, Sarah Spiegel <sup>5</sup> and Markus H. Gräler 1,2,3,\***


Received: 6 March 2020; Accepted: 7 April 2020; Published: 10 April 2020

**Abstract:** The breakdown of the endothelial cell (EC) barrier contributes significantly to sepsis mortality. Sphingosine 1-phosphate (S1P) is one of the most effective EC barrier-stabilizing signaling molecules. Stabilization is mainly transduced via the S1P receptor type 1 (S1PR1). Here, we demonstrate that S1P was autonomously produced by ECs. S1P secretion was significantly higher in primary human umbilical vein endothelial cells (HUVEC) compared to the endothelial cell line EA.hy926. Constitutive barrier stability of HUVEC, but not EA.hy926, was significantly compromised by the S1PR1 antagonist W146 and by the anti-S1P antibody Sphingomab. HUVEC and EA.hy926 differed in the expression of the S1P-transporter Spns2, which allowed HUVEC, but not EA.hy926, to secrete S1P into the extracellular space. Spns2 deficient mice showed increased serum albumin leakage in bronchoalveolar lavage fluid (BALF). Lung ECs isolated from Spns2 deficient mice revealed increased leakage of fluorescein isothiocyanate (FITC) labeled dextran and decreased resistance in electric cell-substrate impedance sensing (ECIS) measurements. Spns2 was down-regulated in HUVEC after stimulation with pro-inflammatory cytokines and lipopolysaccharides (LPS), which contributed to destabilization of the EC barrier. Our work suggests a new mechanism for barrier integrity maintenance. Secretion of S1P by EC via Spns2 contributed to constitutive EC barrier maintenance, which was disrupted under inflammatory conditions via the down-regulation of the S1P-transporter Spns2.

**Keywords:** S1P receptor; inflammation; S1P transporter; spinster homolog 2; barrier dysfunction

#### **1. Introduction**

Endothelial cell (EC) barriers are important intercellular structures that regulate the movement of fluids and dissolved substances into tissues. Maintaining barrier function is particularly important at sites where fluids need to be efficiently separated from tissues such as the vasculature, lymph vessels, gut, brain, and lung [1]. Several different junctional complexes are involved in barrier formation, including tight and adherens junctions, gap junctions, and desmosomes [2]. Adherens junctions are formed by cadherins and nectins and provide a mechanical linker similar to zippers, while tight

junctions are formed by claudins, occludin, and junctional adhesion molecules in the transmembrane regions and perform most of the sealing functions to prevent passage of fluids and molecules [3,4]. The bioactive sphingolipid signaling molecule sphingosine 1-phosphate (S1P) and its G protein-coupled receptor S1PR1 are critical mediators of adherens junction assembly [5]. The deletion of S1PR1 in ECs or the deletion of the two known S1P-producing sphingosine kinases (SphK1 and SphK2) in hematopoietic cells and ECs of adult mice result in severe disruption of the EC barrier [6–8]. Despite this apparent phenotype, the exact mechanism of barrier maintenance by S1P is still unknown. One of the most puzzling questions is of how high S1P concentrations in circulation that are sufficient to induce activation-induced internalization and desensitization of S1PR1 are able to constitutively maintain the vascular EC barrier. Two models were proposed as potential explanations [8]: (1) the static model postulating that there is constantly sufficient S1PR1 expression on the luminal cell surface of ECs even at high S1P concentrations, due to efficient receptor recycling and (2) the dynamic model suggesting that S1PR1 is only expressed on the tissue-facing side of vascular ECs which are activated by S1P leaking through the EC barrier and subsequently induce adherens junction assembly and EC barrier stabilization. In either case, reduced S1P leakage and decreased barrier stability occur until the amount of S1P leaking through the EC barrier increases again and starts a new cycle of EC barrier formation. The validity of either of these models has not yet been verified.

The collapse of the EC barrier is a life-threatening condition and a major severity factor in sepsis [9]. The S1P concentration in circulation decreases significantly during systemic inflammation [10–13]. Whether or not this observed decrease of S1P has something to do with the vascular EC barrier collapse is not known. Previous data indicate that even low amounts of S1P in plasma are sufficient to maintain S1P and S1PR1 mediated lymphocyte circulation [14].

Here, we show that, in vitro, ECs can autonomously produce and secrete S1P, rendering their ability to maintain EC barrier formation largely independent from exogenously added S1P. The S1P transporter Spinster homolog 2 (Spns2) plays a crucial role in the proper release of S1P into the extracellular space. Our work has uncovered an important function of Spns2 in ECs to regulate barrier stability. Spns2 deficient mice demonstrated significantly reduced EC barrier formation presumably due to the lack of S1P exportation from ECs. Furthermore Spns2, but not S1PR1, was down-regulated in ECs stimulated with lipopolysaccharides (LPS) and pro-inflammatory cytokines. Inflammation-induced EC barrier breakdown due to down-regulation of Spns2 resulted in decreased S1P release. Thus, decreased exportation of S1P from ECs due to reduced expression of Spns2 may contribute to EC barrier dysfunction during inflammation. This mechanism may be particularly important in sepsis, where inflammation-induced collapse of the EC barrier significantly contributes to increased morbidity and mortality. The observed stable expression of S1PR1 and the most likely local autocrine and paracrine activity of Spns2-released S1P points to local approaches for S1P supplementation in tissues rather than systemic alteration of S1P in circulating plasma as a potential medical intervention.

#### **2. Materials and Methods**

#### *2.1. Cell Culture*

Human umbilical vein EA.hy926 cells (ATCC CRL-2922) were grown in M199 medium containing 10% fetal bovine serum (FBS; Biochrom, Berlin, Germany), 1% penicillin/streptomycin (100 U/mL, Biochrom), 0.2% glutamine (200 mM, Lonza, Basel, Switzerland), 0.2% heparin (12.5 mg/mL, Carl Roth, Karlsruhe, Germany), and 0.6% ascorbic acid (20 mM, Sigma-Aldrich, Steinheim, Germany). HUVEC were freshly isolated from human umbilical cords and grown in M199 containing 17.5% FBS, 1% penicillin/streptomycin (100 U/mL), 0.34% glutamine (200 mM), 0.2% heparin (12.5 mg/mL), 0.5% ascorbic acid (20 mM) and endothelial mitogen (5 mg/mL, Alfa Aesar, Karlsruhe, Germany). FBS was heat inactivated at 56 ◦C. Rat hepatoma HTC4 cells expressing human S1PR1 together with human Gαi subunit of trimeric G proteins [15] were grown in minimal essential medium (MEM) with Earle's salts (MEM Earle's medium) containing 10% FBS, 2% 100× non-essential amino acids (Biochrom),

1% 100 mM sodium pyruvate (Biochrom) and 1% penicillin/streptomycin (100 U/mL). Cells were incubated at 37 ◦C and 5% CO2 in a humidified incubator (Panasonic, Hamburg, Germany).

#### *2.2. Isolation of Primary Lung Endothelial Cells*

Mice deficient for the S1P-transporter Spinster homolog 2 (Spns2) and their wild-type (wt) littermates were obtained from the NIH Knockout Mouse Project [16]. All described animal procedures were done with dead mice in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC), Animal Welfare Assurance Number AD10000996, effective date March 26, 2017, renewed February 21, 2020 at the Virginia Commonwealth University School of Medicine, Richmond, VA, USA. Mice were killed, and after perfusion through the left ventricle with PBS supplemented with 0.1% heparin (10 U/mL), lungs were dissected from the thoracic cavity and cut into single lobes. The bronchial area was removed and the remaining tissue was digested in 15 mL conical tubes containing 5 mL of PBS with 0.5 mg/mL Liberase TL (Sigma-Aldrich) and 0.02 mg/mL DNAse 1 (Thermo Scientific, Braunschweig, Germany) for 45 min at 37 ◦C with shaking. The tissue was subsequently minced, further incubated for 30 min and passed through a 70 μm cell strainer (BD Biosciences, Heidelberg, Germany). Single cell suspensions were centrifuged at 300 rcf for 10 min at 4 ◦C. Supernatants were aspirated and cell pellets re-suspended in 90 μL of PBS containing 2 mM EDTA, 0.5% BSA, and 10 μL of CD31 MicroBeads per 107 cells (Miltenyi Biotec, Bergisch-Gladbach, Germany). The suspensions were incubated for 15 min at 4 ◦C and re-suspended in 500 μL PBS with 2 mM EDTA and 0.5% BSA. The labeled cells were separated with LS columns and the QuadroMACS separation system according to the manufacturer's instructions (Miltenyi Biotec). The isolated ECs were cultured in endothelial growth medium (5 × 105 cells/mL, Cell Biologics M1168) on poly-l-lysine-coated six-well plates.

#### *2.3. cDNA Synthesis and Quantitative Polymerase Chain Reaction (qPCR)*

Total RNA was extracted using the Quick-RNA Miniprep Plus kit (Zymo Research, Freiburg, Germany). cDNA was synthesized with the RevertAid first strand cDNA synthesis kit (Thermo Scientific) according to the manufacturer's instructions and diluted in nuclease-free water to a final concentration of 5 ng/μL. qPCR was performed with 8 μL cDNA, 4.4 μL of 25 mM MgCl2 (VWR), 2 μL of 10× reaction buffer (VWR) with 15 mM MgCl2, 0.2 μL of 10 mM dNTPs (Thermo Scientific), 0.8 μL of 5 μM 6-carboxy-X-rhodamine (ROX, Eurofins, Ebersberg, Germany), 0.048 μL of 5 U/μL of Taq-Polymerase (VWR), 3.752 μL of nuclease free water (Thermo Scientific), and 0.8 μL of TaqMan primer/probe-mix (2.5 μM each, Eurofins) for each reaction. The reaction was performed with the Mastercycler Realplex (Eppendorf, Hamburg, Germany) using the following program: initial activation at 94 ◦C for 10 min followed by 40 amplification cycles of denaturation at 94 ◦C for 10 sec and annealing and extension at 60 ◦C for 1 min. The internal reference dye ROX was used to normalize the fluorescent reporter signal. The expressions of genes of interest were normalized to hypoxanthine-guanine phosphoribosyltransferase (*HPRT*). Relative gene expression was calculated by the ΔCt and 2−ΔΔCt methods [17]. Primers used for the reaction are listed in Table 1.

#### *2.4. Agarose Gel Electrophoresis*

Gel electrophoresis was performed with 1.2% agarose gels in TBE buffer according to standard protocols for 30 min at 150 V. Gels were stained with ethidium bromide and visualized with a UV trans-illuminator.


**Table 1.** Sequences of primers and probes used for quantitative polymerase chain reaction (qPCR) analyses.

Abbreviations: FAM, 6-carboxyfluorescein; TAM, tetramethylrhodamine.

#### *2.5. Flow Cytometry*

Cells were suspended in 200 μL of 5% FBS in PBS and transferred to a 96-well round bottom plate. The plate was centrifuged at 265 rcf for 5 min. After centrifugation, 300 μL of the primary antibody specific for the human S1PR1 receptor produced in mouse (20 μg/mL, custom-made by Abmart, Berkeley Heights, NJ, USA) was added and incubated for 1 h at 4 ◦C. After incubation, cells were centrifuged and washed with 5% FBS in PBS, followed by 30 min incubation with 50 μL of anti-mouse Biotin-SP (1:200 dilution, Jackson Immuno Research, West Grove, PA, USA, 115–065–075). Subsequently, cells were washed and incubated with 50 μL of 40 ng/mL streptavidin-PE (Biolegend, San Diego, CA, USA, 405203) for 30 min. Cells were washed again and resuspended in 200 μL of 5% FBS and 1 μg/mL propidium iodide (BD Biosciences) in PBS. Resuspended cells were analyzed with the Accuri C6 Plus (BD Biosciences). After doublet exclusion and life-dead discrimination by propidium iodid, the mean floerescence intensity was analyzed.

#### *2.6. Western Blot and Calcium Measurement*

Cells were washed and lysed with buffer containing radioimmunoprecipitation assay buffer (RIPA, Merck, Darmstadt, Germany), 0.5 M EDTA, and protease and phosphatase inhibitor cocktail (Thermo Scientific). Samples were adjusted to 10 μg protein with the BCA protein assay kit (Thermo Scientific) and blotted to polyvinylidenfluoride (PVDF) membranes (GE Healthcare Life Sciences, Freiburg, Germany) according to standard protocols. Detection was performed with antibodies against GAPDH (1:2000 dilution, Santa Cruz Biotechnology, Dallas, TX, USA SC-166574) and VE-cadherin (1:1000 dilution, BD Biosciences 610252), anti-mouse HRP secondary antibody (1:1000 dilution, Carl Roth 47591) and SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) with the C-Digit Blot Scanner (LI-COR Biosciences, Bad Homburg, Germany).

Calcium measurements were performed as described [18].

#### *2.7. Electric Cell-Substrate Impedance Sensing (ECIS)*

ECIS arrays (Ibidi, Gräfelfing, Germany 8W10E PET) were coated with 200 μL of 0.2% gelatin (Sigma-Aldrich) for 20 min at room temperature. Gelatin was replaced with 400 μL of medium containing 1 × 105 cells. Measurements were carried out at 6 kHz with 10 s interval. Once a monolayer was attained, 200 μL of the medium was replaced by 200 μL of starvation medium (2% FBS containing growth medium). The monolayer was stimulated 6 h later with various stimulants. Results were analyzed using ECIS software v1.2.214.0 (Ibidi) by normalizing the data points after treatment to the data point before treatment. The basal barrier function was set to 1.

#### *2.8. Fluorescence Microscopy*

Cells were fixed with 100% ice-cold methanol for 15 min at −20 ◦C. Coverslips were washed five times by dipping them in ice-cold HBSS containing Ca2+/Mg2<sup>+</sup>. Slides were blocked in 50 μL blocking buffer (5% goat serum in HBSS -including Ca<sup>2</sup>+/Mg2<sup>+</sup> and 0.1% saponin) for 1 h at room temperature followed by overnight incubation with 40 μL of mouse primary VE-cadherin antibody (BD Biosciences 610252) diluted 1:100 in 1% BSA in HBSS+ Ca2+/Mg2<sup>+</sup> +0.1% saponin at 4 ◦C in a dark wet chamber. Slides were washed three times in washing buffer (0.1% saponin in HBSS containing Ca<sup>2</sup>+/Mg2+) and incubated in 40 μL of 10 μg/mL goat anti-mouse Cy3 secondary antibody (Life Technologies, Darmstadt, Germany A10521) diluted in 1% BSA in HBSS+ Ca2+/Mg2<sup>+</sup> +0.1% saponin at room temperature in a dark wet chamber. Slides were washed and stained with 40 μL of 300 nM DAPI (VWR) for 20 min at room temperature. Then, the slides were washed three times in ice-cold PBS and mounted using CC Mount (Sigma-Aldrich).

#### *2.9. Measurement of S1P and Sphingosine*

Lipid extraction and quantification of S1P and sphingosine by liquid chromatography coupled to triple-quadrupole mass spectrometry (LC-MS/MS) was done as described [19].

#### *2.10. FITC-Dextran Leakage Assay*

Sixty thousand primary lung ECs in 100 μL medium were seeded in the upper chamber of Transwell inserts with 0.4 μm pores (Sarstedt, Nürnbrecht, Germany) in 24-well plates and grown until confluency. The lower chamber was filled with 600 μL medium. Afterwards, the media of the upper chamber was replaced with 200 μL of media containing 2 mg/mL 70 kDa FITC-dextran (Sigma-Aldrich). The amount of FITC-dextran was measured in the medium of the lower chamber after 24 h incubation at 37 ◦C and 5% CO2 with the Infinite 200 plate reader (Tecan, Stadt Crailsheim, Germany) at 485 nm excitation and 530 nm emission wave lengths. The exact amount of FITC-dextran was determined using a standard curve of FITC-dextran diluted in cell culture media.

#### *2.11. Albumin Measurement*

The detection of mouse albumin in bronchoalveolar lavage fluid (BALF) was carried out with the Mouse Albumin ELISA Quantitation Set (Bethyl Laboratories, Montgomery, TX, USA) according to the manufacturer´s instructions. BALF was diluted 1:20.000 in dilution buffer (50 mM Tris base, 0.14 M NaCl, 1% BSA, 0.05% Tween 20), and 100 μL were added to anti-mouse albumin antibody pre-coated 96-well plates.

#### *2.12. Reagents*

Stimuli used in this study were S1P (Sigma-Aldrich), FTY720 (Cayman Chemicals, Ann Arbor, MI, USA), FTY720-phosphate (Cayman Chemicals), W146 (Tocris, Wiesbaden-Nordenstadt, Germany), LPS (Sigma-Aldrich), TNFα (ImmunoTools, Friesoythe, Germany), and IL1β (Thermo Scientific). The anti-S1P antibody Sphingomab (LT1002) and the corresponding isotype control antibody LT1013 were kindly provided by Roger Sabbadini (LPath Inc. and San Diego State University, San Diego, CA, USA).

#### *2.13. Statistics*

Statistical analysis was performed using GraphPad Prism® Software Version 5.00 (San Diego, CA, USA). Data are presented as mean ± SEM. Unpaired two-tailed t-tests were used to compare two groups. The significance threshold was set to \* *p* < 0.05, \*\* *p* < 0.01, and \*\*\* *p* < 0.001.

#### **3. Results**

#### *3.1. EC Barrier Stabilizing Function of S1P and S1PR1*

To investigate the role of S1P in EC barrier function, the human endothelial cell line EA.hy926 and primary HUVEC were used. EA.hy926 represents a somatic cell hybrid of HUVEC and the lung epithelial carcinoma cell line A549. Quantitative PCR demonstrated that both, HUVEC and EA.hy926 expressed mainly *S1PR1* followed by *S1PR3*, although HUVEC expressed both receptors stronger than EA.hy926 (Figure 1A). Higher expression of S1PR1 in HUVEC was confirmed by flow cytometry (FACS) using a highly specific antibody against human S1PR1 (Figure 1B). Specific staining was demonstrated by the incubation of cells with 1 μM FTY720 overnight, which led to S1PR1 internalization and consequently low cell surface staining as expected (Figure 1B). In line with these expression data, HUVEC responses were greater than those of EA.hy926 after stimulation with 100 nM S1P in intracellular calcium flux measurements (Figure 1C). However, to our surprise, EA.hy926 responses were greater in ECIS measurements after stimulation with 1 μM S1P compared to HUVEC, indicating a higher barrier stabilization in EA.hy926 (Figure 1D). Basal resistance of the EC monolayer, however, was lower in EA.hy926 compared to HUVEC (Figure 1E).

**Figure 1.** S1PR expression and signaling in EC. (**A**) qPCR analysis of S1PR expression in EA.hy926 and HUVEC. Dara are means ± SEM, *n* = 3. (**B**) Flow Cytometric analysis cell surface expression of S1PR1 on EC before and after treatment with 1 μM FTY720 overnight. means ± SEM, *n* = 3. (**C**) Intracellular calcium responses in EA.hy926 and HUVEC upon stimulation with 100 nM S1P. Data were normalized to the response of 10 μM ATP. Means ± SEM, *n* = 3. (**D**) Resistance following treatment with 1 μM S1P, normalized resistance values were taken at the time of the established maximum resistance after S1P treatment divided by resistance of carrier-treated control cells at the same time and are means ± SEM, *n* = 3, \*\* *p* <0.01, determined by two-sided Student's t-test. Line plots represent one experiment out of three with black arrows indicating the addition of S1P or vehicle at the corresponding time. (**E**) Difference in initial non-stimulated resistance of EA.hy926 and HUVEC in ECIS measurements 60 h after seeding, means ± SEM, *n* = 3, \* *p* < 0.05, determined by a two-sided Student's t-test. Line plot represents one experiment out of three.

#### *3.2. Endogenous Di*ff*erences in S1P Signaling between HUVEC and EA.hy926*

To explore the reason for the different behavior of HUVEC and EA.hy926 in ECIS measurements, both cells were treated with 3 μM of the S1PR1 antagonist W146. While EA.hy926 resistance was not affected by W146 treatment, HUVEC monolayers showed significantly reduced resistance by 60% in ECIS measurements, suggesting involvement of S1PR1 in constitutive basal EC barrier maintenance in HUVEC, but not in EA.hy926 (Figure 2A). A similar observation was recorded in ECIS measurements after treatment with the anti-S1P antibody Sphingomab. Sphingomab (120 μg/mL) reduced the basal resistance of the HUVEC monolayer by 30%, while EA.hy926 did not respond at all (Figure 2B). Determination of S1P in the supernatant of both cell types revealed three fold greater S1P level in HUVEC medium than EA.hy926 medium (Figure 2C). Conditioned HUVEC medium consequently provided a four-fold enhanced calcium signal in S1PR1, overexpressing rat hepatoma HTC4 cells compared to EA.hy926 conditioned medium (Figure 2D). Conditioned medium from HUVEC induced a significant 20% increase of the measured resistance in ECIS experiments when added to EA.hy926, while conditioned medium from EA.hy926 in contrast reduced the corresponding resistance by 20% of a HUVEC monolayer (Figure 2E). HUVEC re-established their barrier integrity within hours, while the observed increased resistance in EA.hy926 after incubation with conditioned medium from HUVEC subsequently decreased further and fell below the value of HUVEC (Figure 2E).

#### *3.3. Reversibility and VE-Cadherin Disturbance of EC Barrier Destabilization by S1PR1 Antagonism and S1P Blocking*

Since HUVEC responded immediately to treatment with the S1PR1 antagonist W146 and the anti-S1P antibody Sphingomab with EC barrier disruption, we next asked if this destabilizing effect was reversible. To this end, HUVECs were applied to ECIS measurements and treated with either 3 μM W146 or 120 μg/mL Sphingomab. After decreased resistance leveled off in ECIS measurements, the medium was replaced without the addition of W146 and Sphingomab. Subsequently, the recorded resistance increased immediately and eventually reached normal levels of untreated control cells (Figure 3A). This experiment demonstrated that S1PR1 had to be constantly and persistently activated to induce the EC barrier stabilizing effect, and S1P had to be present all the time as a stimulus. Barrier stabilization was not induced by a single long-lasting activation of S1PR1, but required continuous stimulation. This was also supported by staining HUVEC monolayers for VE-cadherin expression. HUVEC showed pronounced VE-cadherin staining in the intercellular regions of monolayers, which was severely disrupted after treatment with 3 μM W146 (Figure 3B and Figure S1).

#### *3.4. Role of Spns2 in EC Barrier Maintenance*

Autonomously produced S1P by HUVEC, but not by EA.hy926 obviously contributed to constitutive basal EC barrier maintenance. qPCR data revealed that HUVEC expressed significant amounts of the S1P transporter Spns2 mRNA, while EA.hy926 were negative (Figure 4A). Monolayers of primary lung EC isolated from Spns2 deficient mice showed significantly decreased resistance values in ECIS measurements compared to those isolated from wt mice (Figure 4B). Spns2 deficient mouse lung EC also demonstrated increased leakage of fluorescein isothiocyanate (FITC) labeled dextran compared to wt mouse lung EC in Transwell cell monolayer permeability assays (Figure 4C). These results supported a significant contribution of Spns2-driven export of autonomously produced S1P for EC barrier formation ex vivo. To examine whether Spns2 also contributed to EC barrier stabilization in vivo, serum albumin leakage from circulation into the lung was measured in bronchoalveolar lavage fluid (BALF) of wt and Spns2 deficient mice. In line with ex vivo data, BALF retrieved from Spns2 deficient mice contained significantly more serum albumin compared to wt mice (Figure 4D).

**Figure 2.** Comparison of S1P-signaling in HUVEC and EA.hy926. (**A**) Resistance following treatment with 3 μM of the S1PR1 antagonist W146. Normalized resistance values were taken at the time of the established maximal change of resistance after W146 treatment divided by resistance of carrier-treated control cells at the same time and are means ± SEM, *n* = 3, \*\* *p* < 0.001, determined by two-sided Student's t-test. Line plots represent one experiment out of three with black arrows indicating the addition of W146 or vehicle at the corresponding time. (**B**) Resistance following treatment with 120 μg/mL of the anti-S1P antibody Sphingomab. The difference in resistance is the difference between S1P-antibody treatment and isotype control antibody treatment taken at the time of maximal change of resistance after treatment. Shown are means ± SEM, *n* = 3, \*\*\* *p* < 0.001, determined by a two-sided Student's t-test. Line plot represents one experiment out of three with a black arrow indicating the addition of Sphingomab (S1P Ab) or isotype control antibody (Isotype Ctrl) at the corresponding time. (**C**) LC-MS/MS quantification of extracellular S1P production by EA.hy926 and HUVEC. (**D**) Intracellular calcium response in rat hepatoma HTC4 cells transfected with human S1PR1 and the human Gαi subunit of trimeric G proteins, stimulated with lipid extracts from EA.hy926 and HUVEC as indicated. (**E**) Resistance following medium exchange. The difference in resistance is the difference between exchange of conditioned medium and control medium at the time of maximal change of resistance after medium exchange. Line plot represents one experiment out of three with a black arrow indicating the exchange of medium at the corresponding time, controls represent unconditioned medium. (**C**–**E**) Data are means ± SEM, *n* = 3, \*\* *p* < 0.01, \*\*\* *p* < 0.001, determined by a two-sided Student's *t*-test.

**Figure 3.** Dependence and reversibility of EC barrier stability in HUVEC and EA.hy926. (**A**) Resistance following treatment with 3 μM S1PR1 antagonist W146 or 120 μg/mL of anti-S1P antibody Sphingomab, followed by the removal of the added substances. Line plot represents one experiment out of three with black arrows indicating the addition and removal of W146 or Sphingomab at the corresponding time points. (**B**) Immunofluorescence staining of VE-cadherin in HUVEC after addition of 3 μM S1PR1 antagonist W146, followed by removal of the added substance. Representative images from one out of three individual experiments are shown. Pictures were taken 6 h after addition of W146 and 12 h following removal of W146.

#### *3.5. S1P-Mediated EC Barrier Maintenance Under Inflammatory Conditions*

To test the potential influence of inflammation on EC barrier formation, HUVEC and EA.hy926 monolayers were treated with a mix of the pro-inflammatory cytokines tumor necrosis factor-alpha (TNF-α) and interleukin-1beta (IL-1β) together with lipopolysaccharide (LPS). While HUVEC responded with a severe decrease of resistance in ECIS measurements upon treatment, EA.hy926 only showed a weak response (Figure 5A). S1P measurements demonstrated increased S1P levels in the supernatant of HUVEC, but not EA.hy926, which was markedly reduced after treatment with cytokines and LPS (Figure 5B). Conditioned HUVEC medium consequently reduced the calcium signal by 60% in S1PR1 overexpressing rat hepatoma HTC4 cells after treatment with cytokines and LPS compared to non-treated HUVEC medium (Figure 5C). qPCR analyses revealed decreased expression of Spns2 in HUVEC after cytokine and LPS treatment, while expression of the S1P-degrading enzyme S1P-lyase (SGPL1), the S1P-dephosphorylating enzyme lipid phosphate phosphatase 3 (LPP3), and SphK1 was increased (Figure 5D). SphK2 and S1PR1 expression did not change (data not shown). Compared to EA.hy926, HUVEC showed pronounced VE-cadherin staining in the intercellular regions of monolayers, which was severely disrupted after treatment with cytokines and LPS (Figure 6A). In contrast to EA.hy926, VE-cadherin protein expression was reduced after treatment with cytokines and LPS (Figure 6B). Importantly, S1P was able to substantially rescue barrier disruption of HUVEC monolayer in ECIS measurements by transiently increasing the resistance (Figure 6C).

**Figure 4.** Role of Spns2 for EC barrier stability. (**A**) qPCR analysis of *Spns2* in EA.hy926 and HUVEC, *n* = 3, means ± SEM. Images show representative agarose gel electrophoresis signals of amplified PCR products for *HPRT* and *Spns2*. (**B**) Difference in initial non-stimulated resistance of primary lung ECs isolated from wt and Spns2 deficient mice in ECIS measurements, means ± SEM, *n* = 3, \* *p* <0.05, determined by two-sided Student's *t*-test. Single values represent the resistance values for separate mice taken 90 h after seeding. Line plot represents one experiment out of three. (**C**) FITC-dextran leakage assay with primary lung endothelial cells isolated from wt and Spns2 deficient mice, means ± SEM, *n* = 6, \* *p* < 0.05, determined by two-sided Student's t-test. (**D**) Serum albumin measurement in BALF isolated from wt and Spns2 deficient mice, means ± SEM, *n* = 6, \*\* *p* < 0.01, determined by a two-sided Student's *t*-test.

**Figure 5.** Influence of LPS and cytokines on EC barrier stabilization. (**A**) Resistance following treatment with a mix of LPS and cytokines (50 ng/mL IL1β, 50 ng/mL TNFα, 1 μg/mL LPS). Normalized resistance values are resistance of cells 12 h after LPS/cytokine treatment divided by resistance of carrier-treated control cells at the same time and are means ± SEM, *n* = 3, \*\* *p* <0.01, determined by a two-sided Student's *t*-test. Line plots represent one experiment out of three with black arrows indicating the addition of LPS and cytokines or vehicle at the corresponding time. (**B**) LC-MS/MS quantification of extracellular S1P production by EA.hy926 and HUVEC after stimulation with LPS and cytokines or vehicle control. (**C**) Intracellular calcium response in rat hepatoma HTC4 cells transfected with human S1PR1 and the human Gαi subunit of trimeric G proteins, stimulated with lipid extracts from HUVEC without or with LPS and cytokines as indicated. (**D**) qPCR analysis of *Spns2*, *SGPL1*, *LPP3*, and *SphK1* expression in cytokine mix treated HUVEC. (**B**–**D**) Data are means ± SEM, *n* = 3, \* *p* < 0.05, \*\* *p* < 0.01, determined by two-sided Student's *t*-test.

**Figure 6.** Role of S1P and S1PRs in EC barrier destabilization induced by LPS and cytokines. (**A**) Immunofluorescence staining of VE-cadherin and Western blot analysis of VE-cadherin and GAPDH in EA.hy926 and HUVEC after stimulation with LPS and cytokines or vehicle. Representative images from one out of three individual experiments are shown, size bar = 50 μm. (**B**) Western blot quantification (top) and representative Western blot (bottom) of VE-cadherin expression in EA.hy926 and HUVEC after stimulation with LPS and cytokines or vehicle, means ± SEM, *n* = 3, \* *p* < 0.05, determined by two-sided Student's t-test. (**C**) Resistance following treatment with a mix of LPS and cytokines (50 ng/mL IL1β, 50 ng/mL TNFα, 1 μg/mL LPS), and re-stimulated with 1 μM of the S1PR1 agonist CYM5442, 1 μM of the S1PR1,3,4,5 agonist FTY720-phosphate and its non-phosphorylated precursor FTY720, and 1 μM S1P. Line plots represent one experiment out of three with black arrows indicating the addition of stimuli at the corresponding time points. The dark grey line represents an unstimulated control, the light grey line represents a control stimulated with LPS and cytokines without second stimulation. Bar graph represents means ± SEM, *n* = 3, \*\* *p* < 0.01, \*\*\* *p* < 0.001, determined by two-sided Student's *t*-test. Normalized resistance values were taken before (controls) and after treatment with S1P, FTY720-P, FTY720, and CYM5442 at the time of the established maximal change of resistance of cells divided by resistance of cells before LPS/cytokine treatment.

#### **4. Discussion**

S1P is present in the circulation at high nM up to μM concentrations. Recent studies suggested that S1P in blood is particularly important for EC barrier maintenance [8]. Data presented in this study, however, demonstrate that autonomously produced S1P by EC is important for basal constitutive maintenance of barrier function. This is also consistent with previous studies that focused on the role of S1P in the circulation, but used inducible knockout strategies that would also delete the only S1P-producing enzymes SphK1 and SphK2 in EC [20]. Deletion of both SphK1 and SphK2 in EC prevented the autonomous production and release of S1P in EC. Furthermore, the important role of autonomously produced S1P in EC was previously demonstrated by the treatment of HUVEC monolayer with the S1PR1 antagonist W146 prior to addition of platelets in transendothelial electrical resistance (TEER) measurements, which also induced a significant reduction of resistance [20]. While the role of S1PR1 in EC barrier formation and maintenance was investigated many times, the source of S1P required for S1PR1 mediated EC barrier stabilization is still under debate. Our data support the autonomous contribution of EC to produce their own barrier-stabilizing S1P.

S1P is produced intracellularly and needs to be transported out of the cell to act as an extracellular ligand for S1PRs. The main S1P transporter expressed in EC is Spns2 [21,22]. Our data indicate that Spns2-deficient EC suffer from a compromised barrier function due to the defective export of S1P and low extracellular S1P concentrations, which resembled the decreased circulating levels of S1P in mice lacking Spns2. Isolated primary lung EC established very quickly a stable resistance base line as a measure for barrier integrity, lacking the common slow increase in resistance in the first couple of hours of this assay that we observed with EA.hy926 and HUVEC. One reason for this unusual behavior could be the lack of cell growth and division, and seeding of these cells in quantities sufficient to rapidly establish a confluent cell layer to compensate for their observed growth arrest. EC barrier destabilization in Spns2 deficient mice was not observed in a previous study using Evans Blue dye to investigate EC barrier leakage [16]. A possible reason could be different experimental approaches. In this study, we found that endogenous serum albumin was significantly increased in BALF of Spns2-deleted mice compared to wt mice, which reflects the steady-state of barrier stability formed over a long period of time. In contrast, Evans Blue leakage is a single event monitored over a very brief period of 90 min, which may not be sufficient to detect differences in basal EC barrier disturbances. Our ex vivo data was consistent with our in vivo data obtained in three different experimental setups with primary lung EC that all confirmed a disturbed EC barrier function after depletion of Spns2.

Deficiency of S1P in circulation contributes to detrimental effects during inflammation [8,20]. Although the EC barrier stabilizing function of S1P is well accepted, a contributing role of S1P production, transportation, and signaling in inflammation-induced EC barrier breakdown is a novel observation of this study. Particularly, the demonstration that cytokine and LPS-induced down-regulation of Spns2 in EC contributed significantly to lower basal extracellular S1P levels and consequent EC barrier disruption, has not been reported before. Thus far, many studies, including ours, observed an up-regulation of SphK1 during infection, and it therefore could be considered to be an inflammatory kinase [23,24]. The down-regulation of Spns2 in EC might be an effective response to support the infiltration of leukocytes at sites of local infection. In the event of a systemic infection, however, reduced expression of Spns2 in endothelial cells likely contributes to a global collapse of the vascular EC barrier leading to septic shock. Treatment of ECs with inflammatory stimuli did not compromise the expression of S1PR1 on EC, which opens the possibility of using S1PR1 agonists for EC barrier stabilization. Previous studies investigated the influence of S1PR1 agonists on EC barrier protection in various different disease models with mixed results. While some studies showed beneficial effects [7,25], others reported they were ineffective [26]. Thus far, even successful approaches with S1PR1 agonists were not very effective. These results are consistent with our data, which showed incomplete rescue of the EC barrier under inflammatory conditions, probably due to concomitant cytokine and LPS-induced down-regulation of VE-cadherin, which is an essential player in EC barrier formation as well [3].

The contribution of autonomously produced and secreted S1P by EC also implicates S1PR1 and EC barrier function. Since levels of S1P released by EC with 4–15 nM are much lower than the typical concentrations measured in plasma with 200–1000 nM, it is unlikely that activation occurs at the plasma-facing side of EC. EC barrier stabilizing stimulation supposedly occurs at the tissue-facing side of EC, and strategies to increase S1P in tissues such as inhibition of the S1P-metabolizing enzyme SGPL1 may be more promising for future medical interventions than gross application of S1PR1 agonists [27]. SGPL1 inhibitors have already been tested in rheumatoid arthritis [28]. Additional studies are required to evaluate the full potential of targeting S1P signaling and metabolism for EC barrier stabilization.

#### **5. Conclusions**

We confirmed that S1P is a major EC barrier-stabilizing factor, predominantly via S1PR1 stimulation. The constitutive production of S1P by EC and its release in the local environment by Spns2 rather than systemic S1P-levels in plasma were important for basal EC barrier stabilization. Inflammatory stimuli resulted in EC barrier disruption due to the down-regulation of the S1P-transporter Spns2, while the expression of S1PR1 was not altered. Based on our data, inflammation-induced EC barrier disruption may be prevented by local application of S1P or S1PR1 agonists, e.g., via permanent inhalation in the lung or by inhibition of cellular S1P-degradation in tissues with S1P-lyase inhibitors to compensate for the reduced release of S1P from EC.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/9/4/928/s1, Figure S1: Dependence and reversibility of VE-cadherin assembly in HUVEC. Immunofluorescence staining of VE-cadherin in HUVEC after addition of 3 μM S1PR1 antagonist W146, followed by removal of the added substance. Representative images from three individual experiments are shown. Pictures were taken 6 h after addition of W146 and 12 h following removal of W146.

**Author Contributions:** Conceptualization, M.H.G.; Methodology, J.J.P., C.W., T.M., and R.H.; Formal analysis, J.J.P.; Investigation, J.J.P., C.W., and T.M.; Resources, R.H., S.S., and M.H.G.; Writing—Original Draft, M.H.G.; Writing—Review and Editing, J.J.P., C.W., T.M., S.S., R.H., and M.H.G.; Visualization, J.J.P. and M.H.G.; Supervision, M.H.G.; Project Administration, M.H.G.; Funding Acquisition, S.S., R.H., and M.H.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Center for Sepsis Control and Care (CSCC), the Jena School for Microbial Communication (JSMC), and the National Institutes of Health, Grant R01GM043880 (to S.S.). We also acknowledge support by the German Research Foundation and the Open Access Publication Fund of the Thueringer Universitaets- und Landesbibliothek Jena Projekt-Nr. 433052568.

**Acknowledgments:** We thank Elke Teuscher for the preparation of HUVEC, Roger Sabbadini (LPath Inc. and San Diego State University, San Diego, CA, USA) for kindly providing the anti-S1P antibody Sphingomab and the corresponding isotype control antibody LT1013, and Mareike Lipinski for technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Opposing Roles of S1P3 Receptors in Myocardial Function**

#### **Dina Wafa 1,\*, Nóra Koch 1, Janka Kovács 1, Margit Kerék 1, Richard L. Proia 2, Gábor J. Tigyi 1,3, Zoltán Benyó <sup>1</sup> and Zsuzsanna Miklós 1,\***


Received: 15 May 2020; Accepted: 22 July 2020; Published: 24 July 2020

**Abstract:** Sphingosine-1-phosphate (S1P) is a lysophospholipid mediator with diverse biological function mediated by S1P1–5 receptors. Whereas S1P was shown to protect the heart against ischemia/reperfusion (I/R) injury, other studies highlighted its vasoconstrictor effects. We aimed to separate the beneficial and potentially deleterious cardiac effects of S1P during I/R and identify the signaling pathways involved. Wild type (WT), S1P2-KO and S1P3-KO Langendorff-perfused murine hearts were exposed to intravascular S1P, I/R, or both. S1P induced a 45% decrease of coronary flow (CF) in WT-hearts. The presence of S1P-chaperon albumin did not modify this effect. CF reduction diminished in S1P3-KO but not in S1P2-KO hearts, indicating that in our model S1P3 mediates coronary vasoconstriction. In I/R experiments, S1P3 deficiency had no influence on postischemic CF but diminished functional recovery and increased infarct size, indicating a cardioprotective effect of S1P3. Preischemic S1P exposure resulted in a substantial reduction of postischemic CF and cardiac performance and increased the infarcted area. Although S1P3 deficiency increased postischemic CF, this failed to improve cardiac performance. These results indicate a dual role of S1P3 involving a direct protective action on the myocardium and a cardiosuppressive effect due to coronary vasoconstriction. In acute coronary syndrome when S1P may be released abundantly, intravascular and myocardial S1P production might have competing influences on myocardial function via activation of S1P3 receptors.

**Keywords:** sphingosine-1-phosphate; ischemia/reperfusion; cardioprotection; vasoconstriction; coronary flow; myocardial function; myocardial infarct; albumin

#### **1. Introduction**

Ischemic heart disease, including acute coronary syndrome (ACS), is a major cause of death worldwide [1]. ACS is the sudden loss of adequate blood perfusion to the heart, most commonly initiated by the rupture of an atherosclerotic plaque and consequent activation of blood coagulation. This process results in thrombotic occlusion of the coronary artery causing cardiac tissue damage [2]. Urgent reestablishment of blood perfusion to the affected area is crucial to minimizing ischemic tissue injury. Besides the therapeutic time window, the success of reperfusion depends on several other factors such as vascular response to pathophysiological events happening prior to and during thrombus formation. Platelet activation might be relevant in this context as it releases numerous vasoactive mediators which might have an impact on the dynamics and severity of ischemic injury. Sphingosine-1-phosphate (S1P) is one of these many mediators [3–8].

S1P is a sphingolipid mediator which is produced by a wide variety of cells [9]. In vivo, albumin and APO-M in HDL are the most recognized carriers of S1P in blood plasma, which have also been reported to modulate the actions of S1P [10,11]. S1P actions include regulation of diverse physiological and pathophysiological processes such as inflammation, autoimmunity, and neurodegeneration [12,13]. In the cardiovascular (CV) system, activated platelets synthesize and release S1P in large amounts [3–8], and S1P has been reported to play a role in regulating vascular tone [14,15], atherogenesis, cardiac remodeling, and cardioprotection [16–18]. To date, five different G protein-coupled receptors belonging to the endothelial differentiation gene (EDG) family have been identified as specific S1P receptors (S1P1–5) [19,20]. From these, S1P1, S1P2 and S1P3 receptors are expressed abundantly in the CV system and have been reported to mediate CV actions of S1P [16].

S1P has been attributed with cardioprotective effects against ischemia-reperfusion (I/R) injury by several research groups [18,21–27]. The key enzymes in S1P synthesis, sphingosine-kinase 1 and 2 (SphK1 and 2), have been implicated in the ischemia-induced increased release of S1P from cardiomyocytes as well as in mediating the beneficial effects of ischemic pre- and post-conditioning [18,23,24]. Combined deletion of S1P2 and S1P3 receptors increased the infarcted area and enhanced apoptotic cell death after I/R, suggesting that activation of these receptors is cardioprotective [25].

Preischemic S1P treatment has also been reported to decrease infarct size in ex vivo experimental models [24,26]. It has already been shown by Theilmeier and colleagues that HDL and S1P directly protect the heart against I/R injury via the S1P3 receptor in vivo [27]. However, in ACS when S1P is released in substantial amounts from platelets and endothelial cells in blood plasma, it might bind to carriers other than HDL. Several studies have highlighted the vasoconstrictive effects of S1P in various vessel beds from multiple species including the coronaries: S1P had a constrictive effect on isolated porcine pulmonary artery rings [28], in canine, rat, murine, and leporine basilar and middle cerebral arteries [29,30], in rat portal veins [31] and in canine coronaries [32]. S1P administration to the coronary perfusate has been shown to diminish coronary flow (CF) in Langendorff-perfused rat hearts [33]. This effect was attenuated by pharmacological inhibition of S1P3 receptors in the same experiment [33]. Another study conducted on coronary smooth muscle cells raised the potential involvement of S1P2 receptors in the vasoconstrictor response elicited by S1P [34]. Beside its actions on CF, S1P exerts other short-term cardiac effects including generation of arrhythmias and negative inotropy [35–38].

The cardiac effects of S1P reported in I/R injury are controversial. Activation of S1P receptors seems to be cardioprotective, whereas the acute effects of S1P to reduce CF and cardiac contractility are expected to interfere with successful post-ischemic recovery. Moreover, S1P2 and S1P3 receptors have been shown to be involved in both mechanisms. In ACS, when S1P is released in large amounts from activated platelets, its favorable and potentially deleterious effects might clash with one another. In the present study, we aim to delineate how these opposing S1P actions actually affect postischemic cardiac injury after a non-fatal ischemic insult.

For this purpose, we conducted ex vivo experiments in isolated murine hearts mounted on the Langendorff-system. First, we mimicked ACS-related massive S1P release into the blood in order to characterize its coronary effects and consequences on heart function. In these experiments we used albumin as an S1P chaperone and also in subsequent experiments Krebs buffer without added chaperone protein as the vehicle. Second, using S1P receptor gene knockout (KO) mouse models, we aimed to identify receptors involved in these cardiac effects. Third, to understand the role of S1P3 receptor in cardioprotection, we applied a non-fatal I/R protocol in S1P3 deficient hearts. Finally, the complete sequence of ACS was modeled with an initial exposure of the coronaries to S1P as it occurs during plaque rupture and platelet activation, followed by 20 min of complete ischemia, during which the myocardial S1P-producing machinery can be activated, and concluding with 120 min reperfusion representing successful reopening of the coronary artery in a clinical setting. With this approach, we were able to separate the consequences of intravascular and myocardial S1P-related effects during ACS and also to evaluate their combined effects.

#### **2. Materials and Methods**

#### *2.1. Animals*

All experiments reported here were performed in hearts of 130–150-day-old male mice. Animals were bred and housed in the animal facility at Semmelweis University, kept in a 12/12-h light/dark cycle and with free access to water and food. C57BL/6 (WT) mice were bred from breeding pairs obtained from Charles River Laboratories (Isaszeg, Hungary). To answer our specific questions, S1P2-KO and S1P3-KO animals along with wild-type littermates on C57BL/6 genetic background were tested [39]. All procedures were carried out according to the guidelines of the Hungarian Law of Animal Protection (28/1998) and were approved by the Government Office of Pest County (Permission number: PEI/001/820-2/2015).

#### *2.2. Isolated Perfused Heart Experiments*

General anesthesia was induced by intraperitoneal injection of 40 mg/kg pentobarbital, followed by thoracotomy and isolation of the heart. The isolated heart was mounted in a Langendorff apparatus (Experimetria Ltd., Budapest) and perfused at constant 80 mmHg pressure with modified Krebs-Henseleit buffer (118 mM NaCl, 4.3 mM KCl, 25 mM NaHCO3, 1.2 mM MgSO4, 1.2 mM KH2PO4, 0.5 mM NaEDTA, 2.0 mM CaCl2, 11 mM glucose, 5 mM pyruvate (pH 7.4) - all purchased from Sigma-Aldrich, Budapest, Hungary) [40–43]. The solution was continuously gassed with 95% O2 and 5% CO2 at 37 ◦C. During the experiment, the heart was surrounded by a thermally-regulated chamber filled with Krebs-Henseleit buffer.

CF was continuously monitored with a transit-time flow meter placed into the inflow line (Transonic 2PXN flow probe, Transonic Systems Inc., Ithaca, NY, USA). In order to measure left ventricular pressure (LVP), a fluid-filled balloon catheter connected to a pressure transducer was inserted into the ventricle to maintain diastolic pressure at 8 mmHg.

Devices were connected to a computer and data were recorded and analyzed by the Haemosys software (Experimetria Ltd., Budapest, Hungary). Left ventricular developed pressure (LVDevP) was calculated as the difference between peak systolic and minimum diastolic (LVDiastP) pressures. The positive and negative maximum values of the first derivative of the LVP (+dLVP/dtmax, −dLVP/dtmax) were determined as indices of left ventricular contractile and lusitropic performance, respectively.

#### *2.3. Experimental Protocol*

After cannulation of the isolated heart, a 30-min stabilization period was allowed. Subsequently, baseline data were recorded and S1P (D-erythro-sphingosine-1-phosphate, Avanti Inc., 10−<sup>6</sup> M) or vehicle was infused to the perfusion line for 5 min. One mg S1P was dissolved in 263 μL 0.3 N NaOH. This solution (10−<sup>2</sup> M) was further diluted with Krebs solution to give the required concentration in the perfusate.

First, to characterize S1P effects on CF we performed dose-response experiments with S1P in the concentration range of 10<sup>−</sup>9–10−<sup>5</sup> M. S1P was administered in the presence of S1P chaperon human serum albumin (HSA) (Sigma Aldrich, Budapest, Hungary, Cat.No: A3782, lyophilized powder, fatty acid free, globulin free, >99%) or diluted directly in chaperone-free Krebs buffer. The molar S1P to HSA ratio was 1:2 at every concentration [44]. In these experiments we applied S1P in cumulative doses, the next S1P dose was added to the previous dose after the response had reached its maximum. The biological effects of the applied vehicles were tested in separate experiments, and they proved to be without an effect.

In order to answer our main questions, two different experimental protocols were tested (Supplementary Materials Figure S1). To understand the effects of S1P under stable baseline conditions, a 5-min S1P (or vehicle) infusion was applied that was followed by a 20-min washout period. Whereas in the I/R-injury protocol, the 5-min S1P (or vehicle) infusion was followed by a 20-min global ischemia that was brought about by complete cessation of perfusion. At the end of the ischemic period, perfusion

was restarted and reperfusion was maintained for 2 h. In these protocols, S1P was delivered in albumin free Krebs solution at a concentration of 10−<sup>6</sup> M. The S1P concentration applied was defined as a dose approximating its ED50 value (1.17 \* 10−<sup>6</sup> M in Krebs solution).

#### *2.4. Measurement of Infarct Size*

In the I/R experiments, hearts were removed from the apparatus after the 2-h reperfusion period and placed into a −20 ◦ C freezer for at least 15 min. The left ventricle of the frozen heart was cut into ~1 mm thick slices (4 to 6 slices per heart). To visualize the infarcted area of the heart, the slices were then incubated in a phosphate buffer containing 1% triphenyltetrazolium (TTC) (Sigma-Aldrich) for 20 min at a temperature of 37 ◦C. The TTC powder was diluted in a two-part phosphate buffer system at a pH of 7.4 and the slices were fixed in 10% formalin for 15 min [45]. Photos of the TTC-stained slices were captured using a stereomicroscope equipped with a high-resolution digital camera (Rasband, W.S.) and analyzed using Image-J software (National Institutes of Health, Bethesda, MD, USA). The area at risk, defined as the total area consisting of the pale plus red parts and the infarcted pale area, was measured and relative infarct size was calculated as a percentage of the area at risk.

#### *2.5. Statistical Analysis*

Results are presented as mean ± standard error of the mean (SEM). In order to compare time series data between 2 experimental groups, we used two-way repeated measurement ANOVA and Dunnett's *post hoc* test for multiple comparisons. Comparison of data acquired from the 4 experimental groups (WT/S1P3-KO vs. vehicle/S1P infusion) was performed by two-way ANOVA and Sidak's *post hoc* test. To compare the maximal effects of S1P infusion, unpaired *t*-test was applied. To determine the total perfusion loss during S1P infusion, area over the curve (AOC) was calculated. In dose-response experiments non-linear regression analysis was used to find the best fit and ED50 values. To compare dose-responses between 2 experimental groups, comparison of Fits analysis of the statistical software was applied. All statistical analyses were performed using GraphPad Prism 7.0 (San Diego, CA, USA) and *p* < 0.05 was considered as statistically significant.

#### **3. Results**

#### *3.1. Dose-Dependent E*ff*ects of Intravascular S1P on CF Administered with or without S1P-Chaperon Albumin*

To characterize the effect of intravascular S1P on CF, we carried out dose-response experiments with and without albumin as a chaperone. When administered without a carrier, S1P elicited a dose-dependent CF reduction in isolated hearts with an ED50 value of 1.17 <sup>×</sup> 10−<sup>6</sup> M. Therefore, the S1P in further experiments was applied at 1 microM—a dose close to its ED50 value. The coronary effect of S1P was similar in the presence of albumin, however the ED50 value slightly shifted to a smaller concentration range (1.85 <sup>×</sup> 10−<sup>7</sup> M) though it was not statistically significant (*p* = 0.12, *F* = 2.46) (Figure 1). The maximal reduction in CF was also indistinguishable between groups regardless whether S1P was applied carrier-free or with albumin as vehicle.

#### *3.2. E*ff*ects of Intravascular S1P Exposition on CF and Heart Function*

To investigate the effects of a robust S1P release on CF and cardiac function, 10−<sup>6</sup> M S1P or its vehicle was administered to the perfusate of isolated WT murine hearts for 5 min. Administration of S1P reduced CF by 44 ± 3% (Figure 2A). This remarkable decrease started at the beginning of the S1P infusion and continued progressively during the 5 min. During the 20-min wash-out period, CF did not return to the baseline level and remained at a significantly lower value (*p* < 0.0001).

**Figure 1.** Dose-dependent effects of S1P on coronary flow of isolated murine hearts infused alone or in the presence of S1P-carrier albumin. In these experiments S1P was applied in a range of 10−<sup>9</sup> to 10−<sup>5</sup> M in cumulative doses without (S1P) or in the presence of its carrier, human serum albumin (S1P + albumin), and its effects on coronary flow were investigated. Albumin was present in a concentration twice that of S1P. ED50 values were 1.17 <sup>×</sup> 10−<sup>6</sup> M (S1P) and 1.85 <sup>×</sup> 10−<sup>7</sup> M (S1P + albumin). Mean <sup>±</sup> SEM; *n* = 9; 8. Non-linear regression analysis and comparison of Fits using GraphPad Prism 7.0.

**Figure 2.** Effects of S1P on coronary flow (CF) (**A**), left ventricular developed pressure (LVDevP) (**B**), +dLVP/dtmax (**C**) and <sup>−</sup>dLVP/dtmax (**D**) of isolated mouse hearts. S1P (10−<sup>6</sup> M) or its vehicle was administered to the perfusate of isolated wild-type (WT) murine hearts for 5 min. The infusion was followed by a 20-min wash-out period. Administration of S1P resulted in a remarkable decrease in CF, which prevailed throughout the infusion and the wash-out period (*p* < 0.0001). CF reduction compromised left ventricular contractile performance as evidenced by a concomitant decrease in LVDevP, +dLVP/dtmax and −dLVP/dtmax (*p* < 0.0001). Mean ± SEM; *n* = 6, 9; #### *p* < 0.0001 vs. baseline (pre-infusion value), \* *p* < 0.05 vs. vehicle; two-way repeated measurement ANOVA and Dunnett's *post hoc* test.

CF reduction induced by S1P coexisted with compromised left ventricular contractile performance, which is indicated by a 54±9% drop in LVDevP (Figure 2B) and by the markedly decreased+dLVP/dtmax, and −dLVP/dtmax values (*p* < 0.0001) (Figure 2C–D). The vehicle did not affect either CF or other measured heart function parameters (Figure 2A–D).

Earlier studies suggested that S1P might affect coronaries via S1P2 and S1P3 receptors [33,34]. Therefore, we aimed to identify which of these receptors mediate(s) the effect of S1P on the CF. For this purpose, we perfused S1P into the isolated hearts of S1P2 (Figure 3) and S1P3 (Figure 4) KO mice following the experimental protocol described above.

**Figure 3.** Effects of S1P on coronary flow (CF) (**A**–**C**) and left ventricular developed pressure (LVDevP) (**D**–**F**) of hearts isolated from wild-type (WT) and S1P2 knock-out (KO) mice. S1P (10−<sup>6</sup> M) was administered to the perfusate of isolated WT and S1P2-KO murine hearts for 5 min. The infusion was followed by a 20-min wash-out period. CF and LVDevP were monitored during the entire experiment (panels A and D). Maximal decrease in CF and LVDevP compared to preinfusion baseline are shown in panels B and E. Values of area over the curve (AOC) during S1P infusion are shown in panels C and F. In S1P2-KO hearts S1P-induced CF and LVDevP reduction was similar to that observed in WT hearts. Mean ± SEM; *n* = 10, 8; #### *p* < 0.0001 vs. baseline (preinfusion value) in both groups, two-way repeated measurement ANOVA followed by Dunnett's *post hoc* test.

The CF-reducing effect of S1P developing in S1P2-deficient mice was similar to that of WT littermates (Figure 3A–C). The drop of the LVDevP was also similar in the two groups (Figure 3D–F), with no statistically significant difference.

In S1P3-KO hearts, the CF-reducing effect of S1P was markedly diminished compared to WT mice (Figure 4A). There was a significant difference in the maximal effects: CF was dropped by 1.95 ± 0.33 mL/min in WT and only by 0.93 ± 0.10 mL/min in S1P3-KO mice (Figure 4B). The AOC used as an index for total perfusion loss during the infusion period showed similar decrease. During the 5-min S1P infusion, the total perfusion loss was 8.56 ± 1.60 mL in WT vs. 3.70 ± 0.57 mL in S1P3-KO mice (Figure 4C).

The decrease in left ventricular contractile performance upon S1P infusion was also attenuated in S1P3-KO mice (Figure 4D): both the maximal drop in LVDevP (Figure 4E) and the area over the LVDevP curve used as a measure of loss of contractile activity (Figure 4F) were significantly reduced compared to WT controls.

**Figure 4.** Effects of S1P on coronary flow (CF) (**A**–**C**) and left ventricular developed pressure (LVDevP) (**D**–**F**) of hearts isolated from wild-type (WT) and S1P3 knock-out (KO) mice. S1P (10−<sup>6</sup> M) was administered to the perfusate of isolated WT and S1P3-KO murine hearts for 5 min. The infusion was followed by a 20-min wash-out period. CF and LVDevP are shown in panels A and D. Maximal decrease in CF and LVDevP compared to preinfusion baseline are shown in panels B and E. Values of area over the curve (AOC) during S1P infusion are shown in panels C and F. In S1P3-KO hearts, the S1P-induced CF and LVDevP reduction was significantly reduced. Mean ± SEM; *n* = 6, 8; #### *p* < 0.0001 vs. baseline (preinfusion value); \* *p* < 0.05, \*\* *p* < 0.01 vs. WT; two-way repeated measurement ANOVA and Dunnett's *post hoc* test (**A**,**D**) and unpaired *t*-test (**B**–**F**).

#### *3.3. Role of Myocardial S1P3 Receptor Activation in I*/*R Injury*

To better understand the apparent contradiction between the widely reported cardioprotective and observed cardiosuppressive effects of S1P, we aimed to separate the myocardial and coronary actions of S1P in a model of I/R injury.

First, we investigated the effects of potential S1P3 receptor activation during I/R in the absence of intravascularly administered S1P. WT and S1P3-KO hearts were exposed to an I/R protocol, CF and myocardial function were monitored during reperfusion. CF did not differ significantly between the WT and S1P3-KO mice (Figure 5A1). In contrast, parameters describing myocardial performance showed marked differences. The lack of S1P3 resulted in a far worse postischemic functional recovery as evidenced by the drop of the LVDevP (Figure 5B1), decreased +dLVP/dtmax, and −dLVP/dtmax (Figure 5C1,D1), and elevated LVDiastP (Figure 5E1).

These results indicate that S1P3 receptors play a beneficial role in preventing ischemia-induced myocardial dysfunction, most probably by activation from S1P generated locally by the tissues of the ischemic heart. However, this myocardial S1P release did not induce S1P3-mediated coronary vasoconstriction as observed in the previous experiments with intravascular S1P administration.

Interestingly, lack of S1P2 receptors did not influence any of these functional parameters nor the infarct size in this I/R model (Figure S2.)

**Figure 5.** Postischemic coronary flow (CF) (**A**), left ventricular developed pressure (LVDevP) (**B**), +dLVP/dtmax (**C**), −dLVP/dtmax (**D**) and left ventricular diastolic pressure (LVDiastP) (**E**) in isolated WT and S1P3 knock-out (KO) mouse hearts without (left panels: **A1**–**E1**) or with S1P administration (middle panels: **A2**–**E2**) for 5 min to the perfusate at 10−<sup>6</sup> M before the induction of a 20-min ischemia followed by a 120-min reperfusion period. The right panels (**A3**–**E3**) demonstrate statistical comparison of the parameters captured at the end of the reperfusion period. Mean ± SEM; *n* = 6, 8, 7, 7; \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, with two-way repeated measurement ANOVA and Dunnett's *post hoc* test in the graphs and two-way ANOVA followed by Sidak's *post hoc* test in the table insets.

#### *3.4. E*ff*ects of Preischemic Intravascular S1P Exposure on I*/*R Injury*

Next, we investigated the role of intravascular S1P by administering S1P to the perfusion solution before I/R at a concentration of 10−<sup>6</sup> M for 5 min. Under these conditions, CF returned to a significantly higher value during reperfusion in the S1P3-KO hearts (Figure 5A2) indicating S1P3-mediated coronary vasoconstriction. Postischemic myocardial function failed to return during the reperfusion without any difference between the two groups (Figure 5B2–E2).

In order to determine the effects of S1P infusion on postischemic CF and cardiac performance, results obtained by the two experimental protocols were compared (see the table panels in Figure 4). S1P-exposed WT hearts showed a marked reduction in post-ischemic functional myocardial recovery as compared to WT or S1P3-KO hearts without S1P administration (Figure 5B3–E3). Furthermore, the difference in the functional parameters between WT and S1P3-KO hearts was not statistically significant (B2–E2 table insets in Figure 5.).

Finally, we determined whether alterations in myocardial function were reflected in the irreversible ischemic damage of cardiomyocytes. TTC staining revealed that without S1P administration, the relative infarct size was larger in S1P3-KO (10.72 ± 2.93%) than in WT (1.12 ± 0.37%) hearts (Figure 6A,C). In the S1P-exposed groups, the infarcts were substantially larger, but they did not differ between S1P3-KO and WT hearts (Figure 6B,D).

Comparing the size of the infarcted area in S1P-pretreated (Figure 6B,D) with the untreated groups (Figure 5A,C), we detected a marked increase in the size of the infarcted myocardium as a result of S1P administration (Figure 6E).

**Figure 6.** Relative infarct size (**A**,**B**) and representative sections (**C**,**D**) from hearts subjected to ischemia/reperfusion without (A & C) or with (B & D) 10−<sup>6</sup> M S1P infusion. Mean <sup>±</sup> SEM; *n* = 6, 8, 7, 7; \* *p* < 0.05, \*\* *p* < 0.01, \*\*\*\* *p* < 0.0001, with unpaired *t*-test (**A**,**B**) or two-way ANOVA and Sidak's multiple comparison test (**E**).

#### **4. Discussion**

In the present ex vivo study, three different experimental protocols were tested in order to understand the complexity of S1P-induced alterations of cardiac function in ACS and also to determine the involvement of S1P receptor subtypes in mediating these effects. First, we focused on intravascular S1P release, which occurs at the onset of ACS when plaque rupture initiates platelet activation. We found that S1P caused a 44 ± 3% reduction of heart perfusion and simultaneous suppression of myocardial contractility (54 ± 9% decrease in LVDevP). Both effects were attenuated in hearts of S1P3-KO mice, indicating a major role of S1P3 in signaling. In the second part of our study, we focused on the effects of S1P receptor activation within the heart in response to ischemia. Under these conditions, hearts of S1P3-KO mice exhibited worse postischemic contractile recovery (82 ± 3% vs. 52 ± 3% decrease in LVDevP compared to preischemic baseline value in S1P3-KO and WT mice, respectively) and larger infarct size (11 ± 3% vs. 1 ± 3% in S1P3-KO and WT mice, respectively) than WT hearts, indicating that ischemia-induced myocardium-related S1P actions are cardioprotective via activation of S1P3. Finally, we proposed to model the complex scenario of ACS, when intravascular and myocardial S1P release may occur simultaneously and influence cardiac function. Under these conditions, WT hearts showed limited coronary perfusion without any sign of postischemic functional recovery. In S1P3-KO hearts, coronary reflow was better (Figure 5A2), but this failed to improve cardiac function (Figure 5B2–E2) or to reduce infarct size (Figure 6B) compared to WT. These observations indicate that although S1P3-mediated vasoconstriction contributes to the deleterious no-reflow phenomenon, elimination of this effect in S1P3-KO hearts does not moderate I/R injury because it also abolishes the benefits of S1P3-mediated cardioprotection.

The major sources of S1P in blood plasma are red blood cells, platelets, and endothelial cells [5,9]. Sphingosine kinase is highly active in platelets and synthesizes S1P from sphingosine taken up from plasma and produces it in the outer leaflet of the platelet plasma membrane [4]. Platelets store S1P abundantly and release it upon activation [3,7,8]. In ACS, when blood clotting is activated by the rupture of an atherosclerotic plaque, substantial amount of S1P might be released to the circulation [6]. S1P has been reported to have vasoconstrictor and endothelium-dependent vasodilator actions in different vascular beds [14,46,47]. For instance, S1P was shown to have a constrictor effect in isolated porcine pulmonary artery rings [28], in canine, rat, murine and leporine basilar and middle cerebral arteries [29,30], in rat portal veins [31], and in canine coronaries [32]. Whereas S1P increased NO production in cultured HUVEC [48] and in bovine lung microvascular endothelial cells [49], and HDL, a carrier of S1P was shown to cause endothelium-dependent vasodilation in aortic rings of rats and mice mediated via S1P3 activation [50]. However, despite its potential pathophysiological relevance, only a few of these studies have investigated the effects of S1P on the coronaries, and none of them have attempted to relate it to myocardial function. In our study, we found that S1P causes dose-dependent reduction in CF of Langendorff-perfused murine hearts. This observation is in agreement with earlier reports that also ascribed vasoconstrictor effects to S1P in the coronaries and other vascular beds [33,34,51]. Murakami et al. reported dose-dependent S1P-induced CF reduction in rat hearts in a similar experimental model [33]. When we delivered S1P in the presence of albumin, the coronary effect of S1P remained unchanged (Figure 1.). Although, a slight, nonsignificant shift in the ED50 to lower concentration was observed potentially indicating that albumin may enhance S1P coronary effects by protecting it from degradation by phosphatases in the vessels of isolated hearts.

One of the main aims of our study was to (1) mimic the effect of robust S1P exposure of the coronary arteries that might occur in ACS upon thrombotic platelet activation, and (2) explore its effects on coronary perfusion, and (3) on heart function. For this purpose, we administered S1P to the coronary perfusate of isolated murine hearts at 1 microM, a concentration that might easily occur in a thrombotic coronary artery [5,6,52,53], and was close to the ED50 value, we defined (Figure 1). This produced a remarkable decrease in CF (Figure 2A). This observation is in agreement with earlier studies, which also ascribed vasoconstrictor effects to S1P in coronaries and other vascular beds [33,34,51]. Murakami et al. reported dose-dependent S1P-induced CF reduction in rat hearts in a similar experimental setting [33]. The S1P-induced flow deprivation in our study was associated with a significant decline in cardiac performance, which was evidenced by decreased LVDevP, +dLVP/dtmax and −dLVP/dtmax (Figure 2B–D). This might be primarily attributed to CF reduction. However, direct negative inotropic effect of S1P on cardiomyocytes reported by earlier studies might also play a role [25]).

The cellular actions of S1P are attributed to the presence of five specific G protein-coupled S1P receptors [19,20]. Among these, S1P1, S1P2, and S1P3 receptors are expressed abundantly in the CV system [16]. Detailed description of S1P signaling in coronaries is not available in the literature. However, a few studies provide evidence that S1P2 or S1P3 might play a role in the regulation of heart function. In a recent study, we reported a dominant role of S1P2 in S1P-induced enhancement of vasoconstrictor stimuli in the circulation [54]. Therefore, in the present study we aimed to characterize the role of these two receptors in mediating CF reduction by S1P. Using an S1P3-KO mouse model, we showed that the S1P3 receptor plays a relevant role in mediating S1P-induced CF reduction, because the absence of this receptor diminishes significantly the CF reducing effect of S1P (Figure 3C). This observation confirms the findings of Murakami *et al*., who proposed the role of S1P3 in coronary constriction using the S1P3 receptor antagonist TY-52156 in a similar experimental setting [33]. Levkau et al. found that S1P decreases myocardial perfusion in vivo and this effect was absent in S1P3-KO mice [51]. Other investigators proposed the role of S1P2 receptors because S1P induced constriction in human coronary smooth muscle cells that was attenuated by the S1P2 antagonist, JTE-013 [34]. However, in our experiments S1P2-KO mice did not reproduce these pharmacological observations. We acknowledge that this does not necessarily mean that the S1P2 has no role in regulating coronary vessel tone, because it might be that S1P2 also activates pathways in the heart which cause coronary dilation, and these and the direct vasoconstrictor effects in smooth muscle cells canceled out each other in our experiments. However, this putative mechanism requires further investigation. Moreover, S1P2 activation might also sensitize the smooth muscle to other vasoconstrictor stimuli, as has been shown in the systemic circulation [54].

S1P is frequently implicated in cardioprotection [21,22,55,56]. Indeed, numerous studies have shown that it decreases the infarcted area and apoptotic cell death after I/R injury, and that it plays a role in the mechanism of ischemic pre- and post-conditioning [18,23–27]. However, myocardial function has not yet been evaluated in detail in these previous studies, although the involvement of S1P2 and S1P3 receptors has already been suggested [27,33,34]. This protective effect has been inferred from experiments, through the use of fundamentally different methodological approaches. In most of these studies, inhibition of S1P signaling in ischemia was achieved by using S1P receptor gene-deficient models or the pharmacological inhibition of SphK1 and SphK2 enzymes, which made I/R injury more severe and/or reduced the benefits of ischemic pre- and post-conditioning. These observations suggest that S1P signaling is stimulated in ischemia most likely by locally generated S1P released from the heart tissue.

The other approach introduced intravascular S1P administration into the coronary blood flow before an ischemic insult. Although this experimental setting can be considered as a relevant model for studying S1P effects in ACS, inasmuch as S1P infusion mimics S1P release during thrombus formation, whereas the flow cessation models thrombotic occlusion, only a few investigators have explored S1P effects this way, and they only assessed tissue damage without monitoring postischemic heart function. Nevertheless, these studies consistently reported a decrease in the infarcted area [24,27]. This is surprising considering that S1P has several short-term effects in the heart by reducing CF and causing negative inotropy that might be detrimental to postischemic contractile recovery [35–38].

Our current study was designed to combine these approaches in the context of S1P receptor signaling. Our choice of focus on S1P3 signaling was motivated by the results of our experiments shown in Figures 3 and 4 which highlight that the short-term cardiac effects of S1P are mediated in large part by S1P3. However, S1P2, which is the other receptor proposed to participate in cardioprotection [25], did not have a major role. Moreover, the exposure of S1P2-KO hearts to our I/R protocol produced

similar functional and tissue injury to that observed in control hearts (Figure S2), showing that this receptor has no detectable role in cardioprotection in our experimental setting.

First, we aimed to clarify whether intrinsic activation of S1P3 signaling during ischemia was protective in our experimental setting. Our results showed that, in the absence of S1P3, murine hearts were more susceptible to a 20-min global ischemia. This was indicated by weaker contractile recovery during the 2-h reperfusion period (Figure 5B1–D1), higher postischemic end-diastolic pressure (Figure 5E1) an indicator of more severe myocardial ischemic contracture, and increased infarct size (Figure 6A,C). These observations are in agreement with other studies which also implicated the participation of S1P3 signaling in cardioprotection against I/R injury [25–27]. Notably, the severe functional and morphological injury in S1P3-KO hearts developed despite a relatively maintained CF, which approached the preischemic value and was not worse than that of WT hearts during the reperfusion period (Figure 5A1). The observation that CF during reperfusion was similar in WT and S1P3-KO hearts indicates that vascular S1P3 was not exposed to S1P levels sufficient to induce S1P3-mediated vasoconstriction. This is not surprising if we consider that the perfusion fluid was free of exogenously added S1P.

In our study we also investigated the effects of S1P on I/R injury by the other approach described in the literature, where S1P was administered to the coronary circulation before ischemia. After S1P pretreatment, we applied a non-fatal ischemia protocol and followed up the recovery of cardiac function upon reperfusion. This model allowed for the exploration of S1P actions which may take place in a complex ACS scenario in which simultaneous intravascular and myocardial S1P release may occur. Preischemic S1P infusion can be considered as simulation of platelet-derived S1P release in ACS [3–8], whereas the ischemia protocol as S1P release from the myocardium [23,24]. Furthermore, our experiments investigated S1P effects more broadly than previous studies. In addition to determining infarct size, we also assessed postischemic cardiac function. We found that preischemic S1P exposure exacerbated ischemic injury. After an ischemic insult which is supposed to be non-fatal, the infarcted tissue extended to a large part of the myocardium and restitution of contractile activity was hardly observed in WT hearts. The latter was indicated by extremely low LVDevP, +dLVP/dtmax, and −dLVP/dtmax values (Figure 5) that failed to approach preischemic levels during reperfusion, although CF partly recovered. Comparing ischemic injury of S1P pretreated and non-treated WT hearts, infarct size was significantly larger (Figure 6E), whereas LVDevP, +dLVP/dtmax and dLVP/dtmax (Figure 5B3–D3) values were significantly lower at the end of the 2-h reperfusion period. Interestingly, some researchers observed a decrease in infarct size after preischemic S1P treatment [24,26,27]. This difference might be explained by differences in methodology because the infusion time and concentration of S1P preinfusion were slightly different [21,24,26]. A possible explanation can be that our infusion protocol, where we applied S1P at 10−<sup>6</sup> M, might have caused sustained desensitization of cardiomyocytes to S1P. Several studies have shown in different cell types that S1P causes rapid desensitization of S1P receptors which persists for hours [57,58]. It might well be that in our experimental setting the combination of sustained vasoconstriction, which is potentially detrimental to postischemic recovery, and the loss of S1P3 mediated activation of prosurvival and antiapoptotic protective pathways due to desensitization, results in enhanced I/R injury. Whereas, in experimental settings, where lower doses are used (0.1 micromolar), S1P3 mediated protection is more active and dominates over effects of more moderate perfusion loss.

Because we observed that the coronary effects of S1P are in part mediated by S1P3 (Figure 4), we also explored the effects of preischemic S1P exposure on ischemic damage in S1P3-KO hearts. Although preischemic CF and the function of S1P3-KO hearts were better (data not shown, also cf. Figure 4), their functional recovery was as weak, and infarct size as large, as those of WT hearts. Interestingly, although CF in S1P3-KO hearts returned close to the preischemic value, this relatively better perfusion did not provide any benefit for cardiac performance. All these results indicate, that although the absence of S1P3 receptors might mitigate the detrimental effects of preischemic intravascular S1P exposure by decreasing the CF-reducing effect of S1P and allow for better reflow during reperfusion, the concomitant

loss of S1P3-mediated cardioprotection obliterates this potential benefit. Therefore, the S1P3 receptor seems to mediate two opposing S1P actions in the heart, as schematically shown in Figure 7.

**Figure 7.** Events in acute coronary syndrome related to S1P3-mediated alterations of cardiac function.

#### **5. Conclusions**

In this study, using isolated perfused murine hearts, we designed experimental models to simulate and explore the actions of S1P release in ACS. First, we described the effects of intravascular S1P exposure which occur during thrombus formation; and second, the effects of S1P within the heart in response to myocardial ischemia in a separate experimental paradigm. Finally, we investigated the combined effects of simultaneous intravascular and myocardial S1P effects, which can be a real-life scenario in ACS. Intracoronary administration of S1P caused a substantial decrease in CF and heart function. Using S1P receptor KO mouse models, we established that S1P2 has only a minor role, whereas S1P3 is a key determinant in this effect. In I/R experiments, the postischemic functional recovery was weakened and the ratio of the infarcted area was increased in S1P3-KO hearts, confirming the cardioprotective role of this receptor subtype. Preischemic intravascular S1P administration worsened the recovery of cardiac function and increased infarct size both in WT and S1P3-KO hearts, although coronary hypoperfusion was attenuated by S1P3 deficiency. These findings highlight that S1P has opposing effects in the myocardium: S1P released from the ischemic myocardium seems to be cardioprotective, whereas S1P acting via the coronary circulation is deleterious to the heart. Moreover, both of these effects are in a large part mediated by S1P3 receptor. These results taken together suggest that in clinical situations, when thrombotic coronary occlusion causes cardiac ischemia, the released S1P might compromise postischemic recovery due to its unfavorable coronary effects, which might outweigh the presumed cardioprotective effects of S1P produced by the ischemic myocardium. Clearly, further studies are warranted using in vivo and ex vivo models to obtain a better understanding of the (patho)physiological actions of S1P in ACS.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/9/8/1770/s1, Figure S1: Experimental protocols of isolated heart experiments, Figure S2: Postischemic coronary flow (CF) (A), left ventricular developed pressure (LVDevP)(B), +dLVP/dtmax (C), −dLVP/dtmax (D) and left ventricular diastolic pressure (LVDiastP) (E) in isolated WT and S1P2 knock-out (KO) mouse hearts.

**Author Contributions:** Conceptualization, D.W., G.J.T., Z.B., Z.M.; Validation, Z.M.; Formal analysis, D.W., N.K., J.K., Z.M.; Investigation, D.W., N.K., J.K., M.K., Z.M.; Resources, R.L.P.; Writing—original draft preparation, D.W., Z.B., Z.M.; Writing—review and editing, D.W., R.L.P., G.J.T., Z.B., Z.M.; Visualization, D.W.; Supervision, G.J.T., Z.B., Z.M.; Project administration, Z.M.; Funding acquisition, Z.M., Z.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research was funded by the Hungarian National Research, Development and Innovation Office (K-112964, K-125174 and NVKP\_16-1-2016-0042) as well as by the Higher Education Institutional Excellence Program of the Ministry of Human Capacities in Hungary, within the framework of the Molecular Biology thematic program of the Semmelweis University and supported by the EFOP-3.6.3-VEKOP-16-2017-00009 grant.

**Acknowledgments:** The authors are grateful to Ildikó Murányi for expert technical assistance as well as to Nathan Tipton and Erzsébet Fejes for critically reading the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
