Elevation of serum sphingosine-1-phosphate attenuates impaired cardiac function in experimental sepsis

Serum levels of the lipid mediator sphingosine-1-phosphate (S1P) are reduced in septic patients and are inversely associated with disease severity. We show that serum S1P is reduced in human sepsis and in murine models of sepsis. We then investigated whether pharmacological or genetic approaches that alter serum S1P may attenuate cardiac dysfunction and whether S1P signaling might serve as a novel theragnostic tool in sepsis. Mice were challenged with lipopolysaccharide and peptidoglycan (LPS/PepG). LPS/PepG resulted in an impaired systolic contractility and reduced serum S1P. Administration of the immunomodulator FTY720 increased serum S1P, improved impaired systolic contractility and activated the phosphoinositide 3-kinase (PI3K)-pathway in the heart. Cardioprotective effects of FTY720 were abolished following administration of a S1P receptor 2 (S1P2) antagonist or a PI3K inhibitor. Sphingosine kinase-2 deficient mice had higher endogenous S1P levels and the LPS/PepG-induced impaired systolic contractility was attenuated in comparison with wild-type mice. Cardioprotective effects of FTY720 were confirmed in polymicrobial sepsis. We show here for the first time that the impaired left ventricular systolic contractility in experimental sepsis is attenuated by FTY720. Mechanistically, our results indicate that activation of S1P2 by increased serum S1P and the subsequent activation of the PI3K-Akt survival pathway significantly contributes to the observed cardioprotective effect of FTY720.

Most notably, it has been reported recently that S1P serum levels are decreased in patients with sepsis and are inversely associated with disease severity 17 . However, it remains unclear if this observation is associative or whether there is a true cause-effect relationship between S1P levels and cardiac (dys)function. Thus, S1P (and/or its signaling pathway) may serve as a biomarker or even as a theragnostic target in patients with sepsis.
We show that serum S1P is reduced in human and experimental murine sepsis. We then aimed to elucidate the role of S1P and its therapeutic potential in ameliorating sepsis-induced cardiomyopathy in a reverse-translational approach. To address our objectives, we employed two different murine models of experimental sepsis complementing each other by recapitulating some individual features of the human disease. Mice received either cell wall fragments from Gram-negative (lipopolysaccharide, LPS) and Gram-positive (peptidoglycan, PepG) bacteria to induce severe multiple organ dysfunction (MOD) including impaired systolic contractility of the heart (for mechanistic studies) or they underwent cecal-ligation and puncture (CLP) to induce septic cardiomyopathy (for proof of principle studies). Specifically, we investigated the effects and underlying mechanisms of pharmacological (administration of the immunomodulator FTY720) or genetic (sphingosine kinase-2 deficient (SPHK-2 −/− ) mice) approaches to alter S1P receptor signaling on the impaired systolic contractility and S1P levels in experimental sepsis.

Methods
Additional details relating to materials and methodology are provided in the online data supplement.
Use of human subjects-ethic statement. The methods using human subjects were carried out in accordance with the approved guidelines. Specifically, following approval of the local ethics committee of Jena University Hospital, Germany (application no 2160-11/07, 2712-12/09), written informed consent for blood sampling, sample analysis and data collection was obtained from all patients or legal surrogates before enrolment.
Human pilot study. Human samples were collected from a large cohort of septic patients admitted to the multidisciplinary intensive care unit of Jena University Hospital 18 . Samples were snap frozen and stored at −80 °C in the certified biobank of the Center for Sepsis Control & Care at Jena University Hospital. All samples were prospectively collected within 24 h or 72 h (± 6 h) after the patients fulfilled criteria of severe sepsis/septic shock according to ACCP/SCCM matching the newly introduced criteria for sepsis/septic shock (in accordance to the sepsis-3) 1 . To enable best possible homogeneity, septic patients (n = 19) were selected based on fulfilling anastomosis insufficiency after major abdominal surgery as a well-defined focus of this study. We included 28-d survivors (n = 9) and non-survivors (n = 10). Further information on patients characteristics are displayed in Supplementary Table S1.
Control patients (n = 11) underwent minimal invasive direct coronary bypass surgery (n = 5) or cardiac surgery with cardio-pulmonary bypass (n = 6). Serum samples of each patient were collected prior to operation (after cannulation) and early postoperative (suture), reflecting an age-matched cohort with signs of systemic inflammation.
Use of experimental animals-ethic statement. The experimental protocols using animals were carried out in accordance with the approved guidelines and were approved by the Home Office, London, UK (project licenses reference numbers: PPL 70/6525, PPL 70/7348; personal license number: PIL 70/22807). The local ' Animal Use and Care Committee' approved animal experiments in accordance with the derivatives of both, the 'Home Office guidance on the Operation of Animals (Scientific Procedures) Act 1986' , and the 'Guide for the Care and Use of Laboratory Animals' of the National Research Council.
Quantification of organ dysfunction/injury. Cardiac function was assessed in mice subjected to LPS/ PepG or CLP at 18 h or 24 h, respectively, via echocardiography using a Vevo-770 imaging system (Visual Sonics, Toronto, Canada) 20,21 . Then, the experiment was terminated and organ and blood samples were collected for quantification of organ dysfunction and injury.  Immunoblot analysis. Semi-quantitative immunoblot analyses were carried out in mouse heart tissues as described previously 20 . Statistics. Values are presented as box and whisker (min to max) with mean or as mean ± standard error of the mean (SEM) of n observations, where n represents the number of patients/animals studied. Due to the relatively low n-numbers, data (assumed to be not normally distributed) was assessed by Kruskal-Wallis test and Dunn´s test (corrected for multiple comparisons) unless otherwise stated. A P-value of less than 0.05 was considered to be statistically significant.

Results
Additional results are provided in the online Supplementary. S1P serum concentrations are decreased in patients with sepsis and in mice challenged with LPS/PepG. S1P serum levels are significantly lower in septic patients compared to control patients (P = 0.0006, control T1 vs. sepsis D1). This effect was independent of the chosen time point (day 1 or day 3 after sepsis diagnosis) of blood sampling (Fig. 1a). A subanalysis revealed lower S1P levels in 28-day non-survivors compared to survivors, however, this effect was not significant (Fig. 1b).

Cardiac dysfunction following LPS/PepG co-administration is attenuated by FTY720 treatment.
The immunomodulator FTY720 is a structural analogue of S1P and acts in its phosphorylated isoform as an unselective agonist on S1P 1 and S1P 3-5 and a selective functional antagonist on S1P 1 24 . There is evidence that FTY720, enhances serum S1P levels by inhibiting S1P lyase activity 25 . Thus, we investigated the effects of FTY720 on both serum S1P levels and cardiac function in mice challenged with LPS/PepG. There were no significant differences in percentage EF, fractional shortening (FS) and fractional area change (FAC) in sham animals treated with FTY720 26 or vehicle ( Fig. 3a-d). When compared to the sham animals, mice subjected to LPS/PepG demonstrated a significant reduction of percentage EF (P = 0.0001), FS (P < 0.0001) and FAC (P < 0.0001), indicating impaired systolic contractility in vivo ( Fig. 3a-d). Delayed intravenous administration of FTY720 (0.1 mg/kg) 1 h after LPS/PepG challenge significantly attenuated this impaired systolic contractility, indicated by significantly higher values for EF (P = 0.0332), FS (P = 0.0378) and FAC (P < 0.0482) ( Fig. 3a-d).
Effect of LPS/PepG co-administration and treatment with FTY720 on Akt, eNOS and ERK1/2 phosphorylation in murine heart tissue. Activation of S1P receptors results in activation (phosphorylation) of Akt, eNOS and ERK1/2 30 , while activation of Akt and eNOS improves cardiac function in mice with sepsis 31 and activation of ERK1/2 mediates cardioprotection 32 . Thus, we investigated the effects of FTY720 on the degree of phosphorylation of Akt on Ser 473 (Fig. 5a), eNOS on Ser 1177 (Fig. 5b) and ERK1/2 on Thr 202 /Tyr 204 and Thr 185 /Tyr 187 (Fig. 5c), respectively. FTY720 did not affect the degree of phosphorylation of any of the above proteins in sham-operated animals. In contrast, FTY720 significantly increased phosphorylation of Akt on Ser 473 (P = 0.0208) (Fig. 5a), eNOS on Ser 1177 (P = 0.00273) (Fig. 5b) and ERK1/2 on Thr 202 /Tyr 204 and Thr 185 /Tyr 187 (P = 0.00273) (Fig. 5c) in mice challenged with LPS/PepG. Role of S1P 1 and S1P 2 on the FTY720 mediated effects on cardiac function in LPS/PepG challenged mice. Activation of the S1P receptors S1P 1 , S1P 2 or S1P 3 results in activation of the PI3K/Akt/eNOS pathway (Fig. 6a). The beneficial effects of FTY720 on cardiac dysfunction in LPS/PepG challenged mice was abolished in mice that received either the selective phosphatidylinositol 3 (PI3)-kinase inhibitor LY294002 33 or the selective S1P 2 antagonist JTE013 34 15 min prior to FTY720 treatment (Fig. 6b,c). Of note, one mouse of the JTE013 treated group died. Administration of the selective S1P 1 agonist SEW2871 35 (1 mg/kg i.v.) (Fig. 6b,c) or 10 mg/kg (data not shown) at 1 h after LPS/PepG challenge did not significantly attenuate percentage EF, FS and FAC in LPS/ PepG challenged mice. These results suggest that activation of S1P 2 by S1P plays an important role in mediating the cardioprotective effects of FTY720 via down-stream PI3K signaling.

Cardiac dysfunction following polymicrobial sepsis is attenuated by FTY720 treatment.
Having shown that FTY720 prevents the impairment in cardiac function caused by LPS/PepG, we wished to confirm these findings in a model of polymicrobial sepsis (CLP with fluid resuscitation and antibiotics) mimicking the clinical syndrome (Fig. 7). We obtained no significant differences in percentage EF, FS and FAC in sham animals treated with FTY720 or vehicle. When compared to sham animals, mice that underwent CLP demonstrated a significant reduction in percentage EF (P = 0.0002), FS (P = 0.0002) and FAC (P = 0.0002), indicating impaired systolic contractility (Fig. 7). Delayed intravenous administration of FTY720 (0.1 mg/kg) 1 h after CLP significantly attenuated this impaired systolic contractility, indicated by a significant increase of percentage EF (P = 0.0117), FS (P = 0.0205) and FAC (0.0031) (Fig. 7).

Figure 5. Effect of LPS/PepG co-administration and treatment with FTY720 on Akt, eNOS and ERK1/2 phosphorylation in heart tissue of C57BL/6J wild type mice. At 1 h after administration of LPS/PepG or vehicle (sham)
, mice were treated with FTY720 (0.1 mg/kg) or vehicle (10% DMSO). Signaling events in heart tissue were assessed at 18 h. Each immunoblot is from a single experiment and is representative of three separate experiments. All values were corrected for the corresponding β -actin band. Densitometric analysis of the bands is expressed as relative optical density (OD) of (a) phosphorylated Akt (pSer 473 ) corrected for the corresponding total Akt content (Σ Akt) and normalized using the related sham band; (b) phosphorylated eNOS (pSer 1177 ) corrected for the corresponding total eNOS content (Σ eNOS) and normalized using the related sham band and; (c) phosphorylated ERK1 (pThr 202 /pTyr 204 ) and ERK2 (pThr 185 /pTyr 187 ) corrected for the corresponding total ERK1 or ERK2 content (Σ ERK1 or Σ ERK2) and normalized using the related sham band. Data are expressed as mean ± SEM for n number of observations. * P < 0.05 vs. LPS/PepG + vehicle (Kruskall-Wallis test with Dunn´s multiple comparisons test).

Discussion
This study reports that serum S1P levels are significantly reduced in patients with sepsis and in animals challenged with LPS/PepG (reverse translation), which might contribute to septic cardiomyopathy. The lowest levels of S1P were found in patients that died in the acute phase of sepsis, while 28-day survivors appeared to have higher levels of serum S1P. Thus, our findings confirm and extend recently published data by Winkler et al. 17 . These authors investigated serum S1P levels in a larger cohort of patients with sepsis and found that serum S1P was decreased in sepsis and S1P levels negatively correlate with disease severity assessed by sofa score and 28-day mortality, however, their clinical study did not address the role of S1P in the pathophysiology of sepsis.
In mice, low levels of serum S1P were associated with a significant impairment in cardiac systolic contractility, and both pharmacological (FTY720) and genetic approaches (sphingosine kinase 2 deficiency) that increase serum S1P levels attenuated the cardiac dysfunction caused by LPS/PepG. These findings support our hypothesis that approaches, which enhance S1P serum levels, may reduce cardiac dysfunction and improve outcome in patients/animals with sepsis.
What, then, is the mechanism(s) by which high serum levels of S1P preserve cardiac function in sepsis? The S1P receptors S1P 1 , S1P 2 and S1P 3 activate the Akt survival pathway 30 (Fig. 6a). When phosphorylated by it's upstream regulator PI3K, Akt modulates inflammation, cell survival and growth 36 . Most notably, activation (resulting in phosphorylation of Ser 473 ) of Akt attenuates the cardiac dysfunction caused by sepsis in mice 20,21,31 .
Here, we show that the cardioprotective effects of FTY720 in sepsis are associated with increases in a) the serum levels of S1P and b) the phosphorylation of Ser 473 on Akt resulting in the activation of the Akt survival pathway in the heart.
There is very good evidence that the activation of Akt results in the phosphorylation of eNOS (on Ser 1177 ) and, hence, activation of eNOS 37 . Indeed, activation of S1P 1 , S1P 2 and S1P 3 activate the Akt/eNOS pathway. Activation of eNOS inhibits neutrophil adhesion, maintains microvascular patency 38 and reduces the cardiac dysfunction in sepsis 20,21,31 . We have discovered that prevention of the cardiac dysfunction in sepsis afforded by FTY720 is associated with an increase in phosphorylation (on Ser 1177 ) of eNOS. Thus, activation by FTY720/S1P of the Akt/ eNOS survival pathway may contribute to the beneficial effects of FTY720/S1P in mice with sepsis. Activation of the S1P receptors S1P 1 , S1P 2 and S1P 3 also activates the ERK1/2 pathway 30 (Fig. 6a). Activation of ERK1/2 promotes cell survival and proliferation 30 . Here we demonstrate that the cardioprotective effects afforded by FTY720 are associated with an increase in the phosphorylation (on Thr 202 /Tyr 204 and Thr 185 /Tyr 187 ) of ERK1/2 resulting in the activation of these kinases. One could argue that the cardioprotection afforded by FTY720 is independent of a rise in serum S1P levels.
To address this important question we have used a molecular approach to maintain elevated S1P levels in sepsis: Mice with a functional deletion of sphingosine kinase 2 have been reported to have higher endogenous S1P serum levels [27][28][29] . We show here that SPHK-2 −/− mice in comparison to wild type mice have a) elevated S1P serum levels b) exhibit no fall in S1P serum levels when challenged with LPS/PepG, and c) have less impairment in cardiac function when challenged with LPS/PepG. These findings further support the view that higher serum levels of S1P are associated with preservation in cardiac function in sepsis.
The activation of Akt/eNOS by the S1P receptors S1P 1 , S1P 2 and S1P 3 is dependent on the prior activation of PI3K (Fig. 6a). We report here that inhibition of PI3K (with LY294002) prevents the cardioprotective effects of FTY720 in sepsis (Fig. 6b). Thus, activation of PI3K (presumably secondary to the activation of the S1P receptor(s) by S1P) plays an essential role in the cardioprotective effects of FTY720/S1P.
We designed further experiments to gain a better understanding of the specific S1P receptor(s) that mediate the cardioprotective effects of FTY720. SEW2871 is a selective and specific agonist of S1P 1 35 . SEW2871 did not significantly attenuate the cardiac dysfunction caused by LPS/PepG and, hence, did not mimic the effects of FTY720 in sepsis (Fig. 6b). Thus, it is less likely that the observed cardioprotective effects of FTY720 are secondary to the activation by S1P (or FTY720 itself) of the S1P 1 . In contrast, the S1P 2 antagonist JTE 013 34 abolished the cardioprotective effects of FTY720 indicating that these effects are (at least in part) secondary to the activation of S1P 2 by S1P/FTY720 (Fig. 6b). It is possible that the activation of S1P 3 also contributes to the cardioprotective effects of FTY720, as a) this receptor results in the activation of PI3K, and b) prevention of the activation of PI3K with LY294002 abolished the cardioprotective effects of FTY720. Thus, we provide evidence to suggest that the preserved cardiac function afforded by FTY720 in sepsis is at least in part secondary to the activation (by S1P) of S1P 2 (and possibly S1P 3 ). Interestingly, S1P activates Akt in cardiomyocytes and reduces the injury caused by myocardial ischemia-reperfusion, and both effects are lost in S1P 2 and S1P 3 double knock out mice 39 . Mechanistically, S1P 2 and S1P 3 are coupled to Ras homolog gene family member A (RhoA) activation 40,41 . RhoA Figure 8. Effect of cecal ligation and puncture (CLP) and treatment with FTY720 on Akt, eNOS and ERK1/2 phosphorylation in murine heart tissue. At 1 h after CLP or sham operation, mice were treated with either FTY720 (0.1 mg/kg) or vehicle. Signaling events in heart tissue were assessed at 24 h. Each immunoblot is from a single experiment and is representative of three separate experiments. All values were corrected for the corresponding β -actin band. Densitometric analysis of the bands is expressed as relative optical density (OD) of (a) phosphorylated Akt (pSer 473 ) corrected for the corresponding total Akt content (Σ Akt) and normalized using the related sham band; (b) phosphorylated eNOS (pSer 1177 ) corrected for the corresponding total eNOS content (Σ eNOS) and normalized using the related sham band and; (c) phosphorylated ERK1 (pThr 202 /pTyr 204 ) and ERK2 (pThr 185 /pTyr 187 ) corrected for the corresponding total ERK1 or ERK2 content (Σ ERK1 or Σ ERK2) and normalized using the related sham band. Data are expressed as mean ± SEM for n number of observations. * P < 0.05 vs. CLP + vehicle (Kruskall-Wallis test with Dunn´s multiple comparisons test). activation protects cardiomyocytes against ischemia/reperfusion injury 42,43 . In addition, cardiomyocyte apoptosis is reduced in vitro by activation of the RhoA/Rho-associated protein kinase/Focal adhesion kinase/PI3K/Akt signaling pathway 44 . Furthermore, S1P can also mediate cardiomyocyte survival via activation of the RhoA signaling pathway involving RhoA, phospholipase C-ε and protein kinase D1 42 . Importantly, it should be noted that FTY720 lacks affinity for S1P 2 45 , again supporting our view that the cardioprotective effects of FTY720 reported here are secondary to an increase in serum S1P levels after FTY720 treatment.
FTY720 is currently being used in the therapy of patients with multiple sclerosis 46 . Thus, it is, in principle, possible to evaluate the effects of FTY720 in patients with sepsis. Although systemic administration of TLR/ NOD-ligands, such as LPS or PepG, can produce many of the features of sepsis including cardiac dysfunction 47 , CLP-sepsis is the model of choice when testing the efficacy of new therapeutics in sepsis. We show here that the delayed administration (after onset of CLP) of FTY720 in a clinically relevant, murine model of sepsis (CLP) with fluid resuscitation and antibiotic therapy attenuates the cardiac dysfunction caused by sepsis (Fig. 7). Interestingly, preservation of cardiac function in sepsis by FTY720 was associated with activation of the Akt/eNOS and ERK1/2 pathways (Fig. 8). These findings support the view that FTY720 attenuated the cardiac dysfunction in CLP-sepsis also in an Akt/eNOS and ERK1/2-dependent manner. FTY720 or S1P reduces the microvascular permeability in lung and kidney in mice challenged with LPS 16 and FTY720 reduces plasma extravasation in rats with sepsis 48 . With this in mind, we have also investigated the effects of FTY720 on the renal dysfunction and liver injury associated with LPS/PepG. Although less pronounced, we observed a trend towards protection by FTY720, which failed to reach significance (see Supplementary Table S2). Thus, it seems likely that the acute beneficial effect of S1P elevation is restricted to the cardiovascular system.
What, then, is known about the role of endogenous S1P in patients with sepsis? Under normal conditions, the main portion of blood S1P is bound to HDL (60%) and serum albumin (35%) and only a minor amount is available as free S1P 49 . In sepsis, HDL levels are reduced 50 and the application of HDL reduced mortality in various animal models of sepsis and multiple organ dysfunction [51][52][53][54] . The mechanism(s) underlying the protective effect of HDL are not completely understood, but may be due to direct binding and trapping of endotoxins by HDL 55,56 , the down-regulation of pro-inflammatory adhesion molecules and of chemotactic factors 51 . Additionally, and especially in view of our data here, the protective effects of HDL may be mediated by HDL-associated S1P. Interestingly, the many of the beneficial effects of HDL including the cardioprotective effects may be mediated by S1P (reviewed in 57 ). It was recently discovered that apoM is the direct carrier of S1P in HDL 58 . Remarkably, apoM also decreases in sepsis 59 supporting the view that the apoM-S1P-HDL may play a role in sepsis. Whether the reduction of apoM transcription in the septic liver 59 is causal to the reduced serum S1P in sepsis remains open.
In conclusion, we confirmed patients with severe sepsis have lower serum levels of S1P and report here for the first time that in mice pharmacological (FTY720) and genetic (SPHK2 deficiency) approaches to enhance S1P serum levels reduce the cardiac dysfunction caused by LPS/PepG. In a model of polymicrobial sepsis we provide proof-of-concept for a potential therapeutic application of strategies to increase S1P. Preservation of cardiac function in sepsis by FTY720 is at least in part secondary to the activation (by S1P) of the S1P 2 (and possibly S1P 3 ) resulting in the PI3K-dependent activation of the Akt/eNOS and ERK1/2 pathways, which are known to be cardioprotective in animal models of sepsis. We speculate that FTY720 may be useful to elevate levels of S1P in patients with sepsis, which, in turn, may improve outcome in these patients. As FTY720 is used in patients with multiple sclerosis, it is, in principle, possible to evaluate the effects of this drug in patients with sepsis, although the known immunosuppressive effects of FTY720 need to be considered.