Getting to the heart of the sphingolipid riddle

Abstract

Obesity, Type 2 Diabetes, and Metabolic Syndrome induce dyslipidemia resulting in inundation of peripheral organs with fatty acids. These not only serve as substrates for energy production, but also contribute to aberrant production of bioactive lipids. Moreover, lipid metabolism is affected in many cardiac disorders including heart failure, ischemia reperfusion injury, and others. While lipids serve crucial homeostatic roles, perturbing biosynthesis of lipid mediators leads to aberrant cell signaling, which contributes to maladaptive cardiovascular programs. Bioactive sphingolipids, in particular, have been implicated in pathophysiology in the heart and vasculature by a variety of studies in cells, animal models, and humans. Because of the burgeoning interest in sphingolipid-driven biology in the cardiovascular system, it is necessary to discuss the experimental considerations for studying sphingolipid metabolism and signaling, emphasizing the caveats to some widely available experimental tools and approaches. Additionally, there is a growing appreciation for the diversity of ceramide structures generated via specific enzymes and bearing disparate cellular functions. While targeting these individual species and enzymes constitutes a major advance, studies show that sphingolipid synthesis readily adapts to compensate for experimental targeting of any individual pathway, thereby convoluting data interpretation. Furthermore, though some molecular mechanisms of sphingolipid action are known, signaling pathways impacted by sphingolipids remain incompletely understood. In this review, we discuss these issues and highlight recent studies as well as future directions that may extend our understanding of the metabolism and signaling actions of these enigmatic lipids in the cardiovascular context.

Introduction

Diabetes and the metabolic syndrome affect a number of organs and tissues including pancreas, kidney, adipose tissue, liver, and the cardiovascular system. The pathological processes that occur in some of these organs are thought to arise, at least in part, from dyslipidemia. In type 2 diabetes, unrestrained lipolysis in adipocytes increases circulating free fatty acids; additionally, elevated low-density lipoprotein contributes to fatty acid availability via provision of triacylglycerols, from which fatty acids can be released in the presence of endothelial lipoprotein lipase¹. These fatty acids are taken up and then undergo metabolism through several pathways in the cardiomyocyte, each of which serves a distinct biological function. The bulk of these fuel mitochondrial ATP synthesis, and when fatty acid supply exceeds demand, cardiomyocytes can sequester fatty acids in lipid droplets². Other metabolic pathways utilize fatty acids for synthesis of membrane glycerophospholipids and basal pools of signaling lipids. Under normal conditions, partitioning of fatty acid metabolism among these pathways maintains cell homeostasis. However, it is now appreciated that overabundant fatty acid supply perturbs pools of signaling lipids leading to deleterious effects; among these, sphingolipids may link hyperlipidemia to many of its pathological sequelae³. Altered sphingolipid profiles and levels of enzymes of sphingolipid synthesis have been reported across a range of cardiac pathologies including myocardial infarction, ischemia/repurfusion, heart failure, pressure overload, and arrhythmias. Therefore, potential roles for sphingolipids in cardiac pathology constitutes a burgeoning interest in cardiovascular research. Here we review sphingolipid metabolism, pointing out its characteristics which should be taken into account when experimentally addressing biological functions of sphingolipids. We summarize recent litureature addressing functions of sphingolipids in the cardiovascular system, and we discuss potential molecular mechanisms of action. Finally, we suggest areas where further study is needed in this emerging research area.

Sphingolipid Metabolism

In 1884, the German scientist Johann Thudichum described a lipid fraction extracted from bovine and human brain⁴, which behaved distinctly from other lipids. Due to their enigmatic chemical properties, he named these lipids “sphingolipids” after the riddle of the sphinx. Their molecular structures and signaling functions, however, were only described in the following century. Sphingolipids are synthesized by all eukaryotic cells and arise initially from condensation of acyl-CoA to an amino acid by the rate-limiting enzyme serine palmitoyltransferase, constituting the first committed step in de novo sphingolipid synthesis. Sphingolipids are defined by the resulting aliphatic amino alcohol structure (Fig. 1), the sphingoid base, which serves as the backbone of further sphingolipid synthesis. N-acylation of this base through the Ceramide Synthase enzymes produces a family of ceramides, which serve several signaling functions. Ceramides also take on numerous headgroups including phosphate, phosphocholine, sugars, sialic acid, and others. These modifications generate ceramide-1-phosphate and the complex sphingolipids including sphingomyelin, glycosphingolipids, gangliosides, cerebrosides, and globosides. The complex lipids, in particular sphingomyelin, can also be catabolized to release ceramide, and ceramide can be further metabolized to its constituent sphingoid base and fatty acid. Phosphorylation of sphingoid bases occurs via sphingosine kinase, and the product, sphingosine-1-phosphate (S1P) participates in autocrine, endocrine, and paracrine signaling through a family of G protein-coupled receptors. Notably, the only exit from sphingolipid metabolic pathways occurs through the sphingosine-1-phosphate lyase. This enzyme catabolizes sphingosine-1-phosphate to ethanolamine phosphate and a fatty aldehyde, disrupting the sphingoid base. Otherwise, numerous forward/backward enzymatic interconversions occur to maintain sphingolipid profiles, or, alternatively, to respond to extracellular stimuli.

Figure 1. Basic pathways of sphingolipid metabolism.

After its synthesis, dihydrosphingosine is N-acylated by one of 6 ceramide synthase enzymes, each of which have distinct acyl-CoA preference. This produces an array of ceramides bearing different N-acyl chains; in mammals, C14-C26:1 are detected, though C16:0 and C24:1 tend to be the most abundant. The complex sphingolipids are the most abundant cell sphingolipids; ceramides are second most abundant. Production of sphingosine-1-phosphate is tightly regulated, and thus, its abundance is extremely low except under conditions that stimulate its production. Note that almost each step is reversible; much data show that perturbing one enzyme leads to perturbations and/or compensations in other parts of the pathway. Therefore, biology arising from overexpression or inhibition of one enzyme may not always be directly attributable to that enzyme (see text). For the purpose of simplicity, some species of low abundance, and steps not thought to be regulated are not shown. Molecular structures courtesy of avantipolarlipids.com.

Recent advances in sphingolipid research and especially the advent of high throughput lipidomics have led to new appreciation of the tremendous diversity of sphingolipid molecules including ceramides. “Ceramide”, which was historically treated as a single substance, occurs as a wide variety of chemically distinct species, now referred to as “ceramides,” which are synthesized by a family of 6 genetically distinct enzymes (Ceramide Synthases 1–6). This is important not only when considering ceramides per se, but also ceramide derivatives, for example, sphingomyelins and glycosphingolipids, which also demonstrate variety in the acyl chain lengths of their ceramide moieties. These variations among the sphingolipids seem to determine cell function via effects on subcellular compartmentation, alterations to membrane structure and function, and sphingolipid-protein interactions.

In addition to de novo synthesis and complex sphingolipid catabolism, Hannun and colleagues identified a ‘recycling’ or ‘salvage’ pathway by which existing ceramides are de-acylated and then re-acylated with different Acyl-CoAs⁵. This process reconfigures the chain length distribution of intracellular ceramides. This pathway has been exploited experimentally, as, because of difficulty in adding the hydrophobic ceramides to cells, synthetic ‘short chain’ (C2-C6) ceramides, which have improved solubility, have been widely used to test effects of ceramides in cells. While cell effects of these short chain ceramides may bear little in vivo relevance, their de-acylation and subsequent re-acylation through the salvage pathway have been demonstrated to generate natural ceramide species intracellularly.

The complex sphingolipids including sphingomyelin, glycosphingolipids, and their derivatives serve critical structural functions in membranes. These abundant lipids also function as a ceramide ‘sink’, which serves as a substrate source for catabolic metabolism. Both sphingomyelin and ceramide undergo catabolism to ceramide and sphingosine, respectively, the latter of which undergoes phosphorylation to yield sphingosine-1-phosphate. These reactions are stimulated in response to acute stressors such as ionizing radiation and chemotherapeutic agents⁶. In contrast, chronic oversupply of free fatty acids stimulates anabolic (de novo) synthesis of bioactive mediators⁷. Depending on severity, the increased reactive oxygen species that occur upon excessive mitochondrial fatty acid oxidation may also stimulate ceramide synthesis through catabolic pathways (e.g. stimulation of neutral sphingomyelinase). While the catabolic pathways play essential roles in acute stress responses, all sphingolipids initially arise from de novo synthesis; therefore, blocking de novo synthesis, over time, as occurs in genetic models of serine palmitoyltransferase deficiency or long in vivo or in vitro time courses of myriocin treatment, could reduce complex lipids thus perturbing catabolic metabolism in addition to de novo synthesis. Therefore these two routes of sphingolipid synthesis should not be thought of as completely independent.

Experimental Considerations

Because of these metabolic complexities, there are some considerations to data interpretation when single enzymes are targeted experimentally, either genetically or pharmacologically. Many experimental strategies have targeted enzymatic steps of sphingolipid synthesis common to large numbers of sphingolipid species, for example, myriocin treatment or genetic manipulation of serine palmitoyltransferase. While this approach enables conclusion that sphingolipid metabolism in general is or is not involved in a process of interest, it precludes the pinpointing of a specific lipid or metabolic pathway. In addtition to this concern, it should be taken into account that a distinguishing characteristic of sphingolipid metabolism is that for nearly every enzyme mediating sphingolipid biosynthesis, there exists another enzyme or enzymes that catalyze the reverse step (Fig. 1). This interconnected nature of the pathway enables a rapid reconfiguration of the cell sphingolipid landscape in response to cell stress, however, this also presents experimental pitfalls. For example, in many systems, expression of Ceramide Synthases readily adapts when function of a single CerS is lost, potentially in a compensatory effort⁸. These compensations can potentially convolute data interpretation, for example, biology initially attributed to loss of Ceramide Synthase 2 arose in fact from compensatory upregulation of Ceramide Synthase 6⁹. Additionally, to generate some mediators, others may be consumed; therefore, overexpression of ceramidase not only may stimulate sphingosine-1-phosphate production but may also reduce ceramide; this makes it difficult to determine the mechanism underlying experimental outcomes. The converse is also true, i.e. blocking production of one metabolite may increase precursors, which precludes whether absence of the blocked metabolite or accumulation of a precursor mediates an experimental outcome. Furthermore, as indicated above, addition of exogenous sphingolipids, most notably sphingoid bases or short-chain ceramides, may increase various lipids and therefore lipidomics profiling should be performed to determine whether the target lipid pools were effectively perturbed in isolation. This extends beyond sphingolipid pathways, per se, as there are numerous points of interaction between sphingolipid and other lipid metabolic pathways (e.g., phosphatidylcholine, from which the headgroup of sphingomyelin is derived).

A large body of work addressing metabolic disease in cardiovascular systems has relied on treating cells in culture with fatty acids, most often palmitate, in an effort to mimic dyslipidemia. Effects of sphingolipids are then teased out by ablation of enzymes of sphingolipid metabolism including serine palmittoyltransferase (pharmacologically and/or genetically). This approach has yielded many insights but is not without caveats. In many cases, experimental outcomes are attributed to effects on de novo sphingolipid synthesis. However, in addition to generally stimulating sphingoid base synthesis, fatty acyl-CoA excess can increase N-acylation of the base, which produces ceramides. Importantly, palmitate would preferentially increases ceramides with N-palmitoyl moieties (i.e. C16-ceramide); because of the increased appreciation for distinct activities amongst the various ceramides, specific increase in C16-ceramide may have effects distinct to those in vivo, when multiple fatty acids are available for incorporation into sphingolipids. Furthermore, these ceramides are subject to modifications generating complex lipids with C16 ceramide moieties; this artificial increase in a single species may complicate data interpretation and relevance to in vivo situations. In addition, our group initially reported that palmitate and other fatty acids differentially regulate enzymes of sphingolipid biosynthesis at the transcriptional level, with particular distinctions between saturated fatty acids, which were shown to induce both dihydroceramide desaturase and sphingosine kinase, and monounsaturated fatty acids, which were shown to prevent these effects^{7, 10}. Additionally, fatty acids indirectly alter sphingolipid metabolism via generation of acute oxidative and/or ER stress (processes which likely occur more via catabolic production of signaling lipids rather than their de novo synthesis). In sum, while the increased understanding of nuances of sphingolipid synthesis and complexity of sphingolipid species complicates data analysis and interpretation, it also presents rich opportunities for further understanding of this complicate metabolic system.

Sphingolipid Functions in the cardiovascular system

Numerous literature supports roles for sphingolipid signaling in the cardiovascular system, including cell culture, rodent models, and human studies (Table 1). In the cardiomyocyte, ceramide synthesis has been demonstrated to influence a variety of pathological processes including insulin resistance ¹¹, contractile dysfunction¹², apoptosis¹³, autophagy ¹⁴, and hypertrophy ^{14, 15}. The first major in vivo study addressing sphingolipids in cardiac lipotoxicity was performed by Ira Goldberg and colleagues. Using both genetic (SPTLC1 haploinsufficciency) and pharmacological (myriocin) approaches, they demonstrated that blocking sphingolipid biosynthesis improved systolic function in mice with dilated cardiomyopathy resulting from perturbed lipid metabolism (transgenic cardiomyocyte overexpression of lipoprotein lipase)¹⁵. A key consideration when interpreting studies where the first step of synthesis is blocked, is that, while these approaches undoubtedly reduced ceramides, production of dihydroceramides, sphingosine-1-phosphate, sphingomyelins and other complex sphingolipids would likely also be altered. Therefore, experimental outcomes in these and similar studies (including our own, described below) may not be unequivocally attributed to ceramide, and, in actuality, the specific sphingolipids mediating these processes remain unknown. As an example, some recent studies outside the cardiac context have revealed functions for both glycosphingolipids and, to a greater extent sphingosine-1-phosphate^{16, 17}, which had been previously attributed to ceramide.

TABLE 1.

Summary of selected studies

Model System	Treatment	Lipid Changes (if measured)	Experimental Outcome	Ref.
Cell Models
C2C12 myotubes	palmitate	↑long and very long chain ceramides ↑S1P	Palmitate induced Sphingosine Kinase 1	7
C2C12 myotubes	Palmitate, stearate	↑ceramide (total) and diacylglycerol	Inhibition of ceramide, but not diacylglycerol, led to attenuation of insulin-stimulated Akt/PkB phosphorylation	48
C2C12 mouse myotubes, primary mouse skeletal myoblasts	Palmitate, oleate	Palmitate ↑dihydroceramides, ceramides, and DAG Oleate prevented these changes	Palmitate induced the dihydroceramide desaturase DES1 and led to insulin resistance; this was prevented by oleate. Oleate protection from insulin resistance was overcome by overexpression of DES1.	10
H9c2 cardiomyoblast cell line; cultured primary neonatal rat cardiomyocytes	lactosylceramide	N/A	Lactosylceramide treatment induced reactive oxygen species, signaling through p44, protein kinase C, and other canonical pathways, and led to cardiomyocyte hypertrophy.	36
H9c2 cardiomyoblast cell line; cultured primary neonatal rat cardiomyocytes	K6PC-5, sphingosine kinase activator; Various sphingosine kinase inhibitors Short-chain ceramide	K6PC-5 ↑S1P	Sphingosine Kinase 1 activation or overexpression protected from mitochondrial dysfunction and cell death due to oxygen/glucose deprivation/reoxygenation (OGD/reoxygenation). Cell death was exacerbated by inhibition of sphingosine kinase 1. OGD/reoxygenation increased total ceramide. C6 ceramide recapitulated OGD/reoxygenation.	32
H9c2 cardiomyoblast cell line; cultured primary mouse cardiomyocytes	relaxin	↑S1P ↓ceramide, sphingomyelin	Relaxin-induced extracellular matrix remodeling was blocked by inhibiting sphingosine kinase. Inhibiting sphingolipid signaling reduced connective tissue growth factor.	59
Isolated Rat Ventricular Myocytes	C6-ceramide	N/A	Phosphorylation of PKCε, depressed contractility Phosphorylation of troponin I and myosin binding protein-C	12
Sprague–Dawley rat Cardiomyocytes	C2-ceramide	N/A	C2-ceramide increased lactic dehydrogenase and cytoplasmic calcium, and caused cell death.	46
Neonatal rat cardiomyocytes; H9c2 cardiomyoblast cell line	Sphingosyl-phosphorylcholine (lyso-sphingomyelin)	N/A	Led to adaptive autophagy via PTEN/AKT1/mTOR Protected from ischemic apoptosis	47
Human myoblasts	C2-ceramide Palmitate; Overexpression of perilipin	Palmitate increased ceramides and diacylglycerols; Perilipin increased lipid droplets	C2 ceramide decreased insulin signaling; palmitate decreased insulin signaling in a sphingolipid synthesis dependent manner; increasing lipid droplets also protected from insulin resistance	56
Sprague–Dawley rat myocytes	C2-ceramide	N/ A	Augmented myocyte peak shortening and relengthening; enhanced calcium release	57
cultured rat neonatal cardiomyocytes	C2-ceramide	N/A	C2-ceramide caused apoptosis via activation of Caspases 3 and 8.	58
cultured rat neonatal cardiomyocytes	C2-ceramide	N/A	increased mitochondrial fission and fragmentation decreased mitochondrial volume increased mitochondrial fission	60
H9c2 cardiomyoblast cell line; cultured primary mouse cardiomyocytes	insulin	↑Ceramide	Increased Sptlc2 and Des1 Loss of mitochondrial respiration and insulin resistance were ceramide dependent	51
Rodent Models
Glycosylphosphatidyl-inositol-anchored human lipoprotein lipase (LpLGPI) transgenic expressing Mice modelling cardiomyopathy	Myriocin, Haplosufficient SPTLC1 heterozygote	↑Ceramides, sphingomyelin in LpLGPI mice, reduced by myriocin or crossing with SPTLC1 −/− mouse.	Blocking sphingolipid synthesis normalized insulin sensitivity and markers of heart failure in LpLGPI transgenic mice	15
Cardiomyocyte overexpression of Diacylglycerol Acyltransferase 1 (DGAT1)	Cross with long-chain acyl-CoA synthetase	↑triglycerols ↓diacylglycerol and ceramide	Cardiac defects in the long-chain acyl-CoA synthetase were ameliorated by DGAT1 overexpression.	60
Cardiomyocyte-specific SPTLC2 deletion		↓C18- and very long chain ceramides; increased saturation in acyl chains of other lipids	Mice exhibited constitutive decreases in fractional shortening and ventricular wall thickness. Markers of heart failure and ER stress were induced.	18
C57bl mice	streptozotocin	↑Ceramide Synthase2 ↑SptlC1 ↑ Dihydroceramide desaturase 1 mitochondrial increase in lactosylceramide; no increase in mitochondrial ceramide	Lactosylceramide suppressed both mitochondrial calcium retention capacity and state 3 respiration.	16
ApoE−/−mice	High fat high cholesterol feeding with or without an inhibitor of glucosylceramide and lactosylceramide synthase enzymes (D-PDMP)	HFHC diet↑ glucosylceramides	HFHC in the ApoE−/− mice developd multiple maladaptive cardiac pathologies, each of which was ameliorated by pharmacological inhibition of glucosyl-/lactosylceramide synthase. Mechansitically, this occurred at least in part by decreased phosphorylation of MAP kinase.	36
C57BL mice	High saturated fat feeding With or without myriocin	↑medium, long, and very long chain ceramides	Mice on high saturated fat diet developed cardiac hypertrophy and impaired systolic and diastolic dysfunction. Cardiomyocytes were hypertrophied and exhibited markers of increased autophagy. All pathology was blocked by myriocin. Cell studies supported a role for Ceramide Synthase 5 and/or medium chain ceramides in cardiomyocyte autophagy and hypertrophy.	14
C57BL/6 mice	High saturated fat feeding	↑atypical sphingolipids with a 16-carbon sphingoid base	Serine palmitoyltransferase subunit 3 mediated production of atypical sphingolipids and led to cardiomyocyte apoptosis.	22
S1P3-deficient mice	Ischemia/reperfusion	↑S1P bound to high density lipoproteins	Ischemia/reperfusion injury increased S1P borne by HDL. Mice lacking sphingosine-1-phosphate receptor 3 had exacerbated injury.	30
C57BL/6 mice	hypoxic model of pulmonary arterial hypertension with and without PF-543, an inhibitor of Sphingosine Kinase 1, or RB-005, a ceramide synthase inhibitor		Inhibition of Sphingosine Kinase 1, but not Ceramide Synthases, prevented right ventricular hypertrophy and cardiomyocyte death, reduced p53, and increased Nrf-2 (nuclear factor (erythroid-derived)-like 2).	31
C57BL mice	Ischemia/reperfusion with or without inhibiting ceramide synthesis	Ischemia/reperfusion increased ceramides	Inhibiting ceramide synthesis post-ischemia decreased infarct size and reduced expression of inflammatory mediators.	13
Nogo-A/B–deficient mice	Chronic pressure overload with or without myriocin or a sphingosine-1-phosphate receptor 1 agonist	↑sphingosine-1-phosphate	Mice lacking Nogo-A/B showed increased sphingosine-1-phosphate and were protected from pathological remodeling. Wild type mice could recapitulate this when treated with an agonist of sphingosine-1-phosphate receptor 1.	34
Male Wistar rat	Atrial pacing	↑sphingoid bases, sphingosine-1-phosphate ↓ceramides	Ventricular tachycardia or right ventricular pacing elevated sphingoid bases and decreased ceramides, suggesting that pacing alters ceramide catabolism.	25
Human Studies
Coronary Artery Disease Patients, Corogene study	Increased ceramides and changed ratios between long vs. very long chain ceramides were predictive of death in patients with coronary artery disease or acute coronary syndromes.			38
Chronic Heart Failure Patients	Ceramide increased and correlated positively with with severity of chronic heart failure and were an independent mortality risk factor.			39
Patients, FINRISK 2002	Circulating ceramides predicted cardiovascular outcomes.			41
pulmonary arterial hypertension patients	Defects in fatty acid metabolism occurred in pulmonary arterial hypertension pateints and were linked with altered ceramide levels in the right ventricle.			42
Type 2 Diabetic Patients	Dietary medium chain fatty acids led to production of medium chain ceramides and improved cardiac function in type 2 diabetic patients.			43
Severe heart failure, left ventricular assist patients	Myocardial ceramides of various chain lengths were increased in patients receiving left ventricular assist devices. Some changes in expression for proteins and genes involved in sphingolipid metabolism were also observed.			44

In contrast to this deleterious sphingolipid-dependent biology, Park and Goldberg went on to demonstrate that mice happloinsufficient for SPTLC2 exhibited constitutive decreased fractional shortening, thinning of the cardiac wall, and induction of fibrosis and heart failure markers¹⁸. These studies indicated that, while excessive sphingolipids may mediate the cardiac pathophysiology, there exists a basal requirement for sphingolipids. Sphingolipids are present in membranes of every eukaryotic cell type and are constituents in mitochondrial membranes as well as key players in intracellular transport and stress responses. Therefore, it is unsurprising that in a mouse lacking sphingolipid synthesis machinery from birth, some defects might occur. Therefore, sphingolipids should not be thought of as inherently toxic, but rather, that altering their homeostatic levels is maladaptive.

Based on our observations that saturated fatty acids more severely perturbed sphingolipid profiles in the lipotoxic context, our group developed a mouse model of cardiomyopathy by high saturated fat feeding for 18 weeks¹⁹. These mice showed left ventricular hypertrophy, concentric remodeling, and decreased systolic and diastolic function, similar to diabetic cardiomyopathy in humans. At the cellular level, these mice also showed maladaptive autophagy, which has been suggested to mediate cardiomyocyte hypertrophy²⁰. All cellular and in vivo cardiac pathophysiology was prevented by blocking sphingolipid synthesis using myriocin¹⁹. However, lipidomics analysis of these hearts enabled the identification of a specific ceramide species, N-myristoyl sphingosine (C14-ceramide), that correlated with cardiac dysfunction. Studies in primary cardiomyocytes revealed a specific ceramide synthase, CerS5, which produces C14-ceramide, that mediated cardiomyocyte autophagy and hypertrophy in cultured cells. Further work in this vein revealed a specific role for CerS2, and not CerS5, in loss of mitochondrial function in cardiomyocyte cell lines treated with free fatty acids²¹. Specifically, we demonstrated a specific role for VLC-ceramides derived from CerS2 in driving oxidative stress, mitochondrial dysfunction, insulin resistance, and mitophagy preceding apoptosis in a cardiomyocyte model of lipotoxicity²¹. Increased oxidative stress and mitochondrial dysfunction was observed with palmitate oversupply and CerS2 overexpression, but blocked by inhibition of ceramide de novo synthesis. Loss of CerS2 protected cells from these maladaptations, and only VLC mitochondria-targeted ceramide analogs stimulated mitochondrial loss by mitophagy and cell death. These CerS2-depedendent effects were not recapitulated by manipulating levels of CerS5 ¹⁴. These data support a model in which ceramides have distinct signaling functions in the cardiomyocyte dependent on N-acyl chain length. However, other explanations may be subcellular localization or non-catalytic functions of CerS enzymes; further work will shed light on these scenarios.

In addition to the increased appreciation for N-acyl chain length variation in ceramide pools, recent work has identified atypical sphingoid bases, which contain 16 rather than 18 carbons. These bases are derived from inclusion of Sptlc3 in the serine palmitoyltransferase complex, which is typically comprised of Sptlc1 and 2. Our group showed that alterations in the Sptlc3-derived bases also occurred in diabetic cardiomyopathy, and these non-canonical sphingolipids could promote apoptosis, but not autophagy, in cardiomyocytes²². Together these studies suggest that multiple sphingolipid metabolic pathways contribute to cardiac dysfunction. However, despite progress in our understanding of the molecular mechanisms and the specific lipid species involved in cardiomyocyte lipotoxicity, further study is required to pinpoint the exact sphingolipids and mechanisms involved.

Roles for sphingolipids in pathology outside the lipotoxic context have been reported. Cardiac sphingolipids were increased in a rodent genetic model of heart failure and isoproterenol-induced myocadial infarction ^{23, 24}, and a model of tachycardia reported a decrease in ceramide and changes in other sphingolipids²⁵. Reforgiato et al. recently demonstrated how infarct size after ischemia/reperfusion (I/R) injury was positively impacted by inhibition of de novo ceramide synthesis. Mouse hearts intraventricularly injected with myriocin showed a significant decrease in infarct size, local ceramide, and other inflammatory mediators ²⁶. Moreover, alcohol feeding in rats decreased cardiac ceramides, which could be a possible factor in contributing to cardioprotection conferred by alcohol consumption ²⁷. In an inflammation model of advanced glycation end product (AGE) stimulated mitochondrial damage, cardiomyocytes were protected from decreased mitochondrial respiration when ceramide synthesis was inhibited. Intriguingly, mice overexpressing the AGE receptor in the lung also had increased ceramides and CerS isoforms in the heart ²⁸. Therefore, sphingolipid changes are not limited to the lipotoxic situation in metabolic disease, but are more broadly applicable across pathological etiologies.

Beyond Ceramides

While emerging roles for ceramides in cardiovascular pathophysiology are gaining further support, there are potential roles for additional sphingolipids in cardiac pathology. Among these, sphingosine-1-phosphate (S1P) has garnered attention. A study by Hayek et al., noted that an increase in circulating progenitor cells in patients undergoing cardiac catherization was linked to increased S1P, but not ceramide-1-phosphate ²⁹. S1P has also been linked to protection from ischemia/reperfusion injury³⁰. In type 1 diabetic rats exposed to I/R injury, S1P agonist FTY720 decreased infarct size and inflammation ¹⁷. Similarly, in a mouse model of hypoxic pulmonary hypertension, inhibition of sphingosine kinase 1 (Sphk1) reduced RV hypertrophy and cardiomyocyte cell death³¹. In an in vitro model of ischemia/reperfusion, cells were protected from oxygen/glucose deprivation and reoxygenation by increasing Sphk1, which protected from I/R-induced oxidative stress, increased ceramide, and cell death ³². S1P produced in the vascular endothelium was shown to play a role in protection from pressure overload induced myocardial inflammation and remodeling, and the S1P receptor 1 (S1P1) was identified as a key regulator of blood pressure^{33, 34}. Additionally, an intracellular inhibitor of serine palmitoyltransferase, Nogo-B, was shown to protect from pressure overload injury. Importantly, however, as S1P tends to arise acutely from catabolic pathways, it is likely that sphingolipids generated de novo through serine palmitoyltrasferase may play distinct roles from S1P in these experimental contexts. In sum, most studies suggest that S1P may be beneficial in contrast to excess ceramide, which data suggest drives cardiac pathology^{12, 35}. Moreover, functions for other spingolipids continue to arise, for example in the ApoE −/− mice a model of cardiolipotoxcitiy, inhibition of lactosylceramide, a glycosphingolipid, protected mice from cardiac hypertrophy³⁶. Additionally, lactosylceramide was shown to accumulate and cause dysfunction in the mitochondria of cardiomyocytes in mice stimulated to develop type 1 diabetes by streptozotocin injection¹⁶. Most studies have focused on the ceramides, however, so potential roles of complex sphingolipids remain to be discovered.

Human Relevance

Poor outcomes in patients with cardiovascular disease have been correlated with changes in circulating sphingolipids including ceramides ^37–39. Availability of metabolomics methods to identify and quantify molecules in patient specimens has enabled the correlation of specific ceramides with major adverse cardiac events and death. As such, recent studies of large patient cohorts have consistently linked long and very long chain ceramides with increased cardiovascular risk. Specifically, circulating C16:0, C:18:0, C:20:0, and C24:1 ceramides were strongly associated with poor cardiovascular patient outcomes^40–42. Consistent with this, diabetes patients offered a diet rich in long-chain fatty acids demonstrated decreased cardiac output and stroke volume; however, those administered a diet rich in medium-chain fatty acids showed improved cardiac function, including increased contractility and decreased circulating ceramides and sphingomyelins⁴³. The mechanism by which medium chain fatty acid consumption reduced circulating ceramides is unclear, but may arise from increased mitochondrial metabolism of medium chain fatty acids, which is CPT1-independent. Thus these fatty acids would more readily be taken in to mitochondria rather than incorporated into cellular lipid pools. Together, these studies strongly support that the long and very long chain ceramides correlate with poor cardiac outcomes; however, as the bulk of mammalian ceramides are indeed in this range, further study will be required to pinpoint specific species and enzymes. For example, long and/or very long chain ceramides may arise from CerS1,2,4,5, and/or 6. An additional caveat to the study of circulating ceramides is that the tissue origin of these lipids remains unknown, and whether they serve as a marker of perturbed metabolism in general or play a mechanistic role remains unknown. An additional consideration is that ceramides circulate bound to lipoproteins, which could potentially deliver them to cells expressing lipoprotein receptors. Therefore circulating ceramides could be a cause, an effect, and/or a marker merely correlated to cardiac pathology.

In addition to circulating ceramides, myocardial ceramide content was shown to increase in human heart failure, and to decrease upon unloading by implantation of a left ventricular assist device. Increased ceramides included both long- and very long-chain species⁴⁴. This study by Schulze and colleagues demonstrated upregulation of serine palmitoyltransferase subunits. Overexpression of either of its subunits in cells led to hypoxia and inflammatory signaling. It is important to note, however, that CerS2, which synthesiszes very long chain ceramides, was downregulated in heart failure, whereas CerS5 was upregulated. Therefore the source of the very long chain ceramides remains unknown and may occur via catabolic (i.e. sphingomyelinase) pathways. Whether these lipids play any mechanistic role remains to be determined. Also, as with most studies to date, the approach was to modulate the first step of sphingolipid synthesis. Therefore, further study is necessary to understand the specific lipids involved.

Molecular Mechanisms of Sphingolipid Action

Though cell mechanisms of sphingolipid action have not been explored in detail specifically in the cardiovascular system, it is likely that they occur similarly as has been described in many other cell types. The major bioactive sphingolipids, ceramides and sphingosine-1-phosphate, regulate numerous cell functions including apoptosis, autophagy, inflammation, differentiation, senescence, and many others, dependent on context⁴⁵. Studies to address the cellular functions of sphingolipids did not accumulate a critical mass until around 1990, when Hannun, Kolesnick, and others, performing work largely in cancer model systems, described signaling activities of ceramide. These studies revealed roles for ceramide in senescence, differentiation, and proinflammatory signaling⁴⁶. Specific interest in sphingolipids in the context of metabolic disease, however, was stimulated by a study that would become highly influential, in which Unger and colleagues described the attenuation of saturated fatty acid-induced apoptosis of pancreatic β -cells by inhibition of ceramide biosynthesis⁴⁷. Broad interest in roles of sphingolipids in lipotoxic processes emerged from this initial finding, leading to expansion of the field when Summers and colleagues demonstrated that saturated fatty acids increased ceramide, which led to insulin resistance in skeletal muscle⁴⁸.

Despite tremendous progress made in sphingolipid research on metabolism and biochemical analysis over the last 25 years, molecular mechanisms by which ceramides mediate their effects are still very poorly understood. This is largely due to the technical challenges associated with lipid research, and also the difficulty in determining specificity of lipid-protein interactions. Most proteins have hydrophobic areas, and hydrophobic forces, which thermodynamically drive the exclusion of water, promote lipid-protein interactions. Care should be taken when evaluating lipid-protein interaction data, particulalrly when lipids are used at concentrations far beind the biologically relevant levels.

A few molecular mechanisms of sphingolipid action have been identified and become widely accepted. For example, ceramides have been shown to bind and activate protein phosphatases including PP1 and PP2A⁴⁹. Activation of the latter was shown to mediate ceramide-induced insulin resistance via dephosphorylation of Akt/PKB in response to insulin. Sphingolipids play roles in mitochondria, with ceramides of various chain lengths reported to modify the mitochondrial membrane leading to pore formation and regulation of bax or bac-dependent apoptosis⁵⁰, and ceramide or its derivatives have been shown to disrupt mitochondrial metabolism⁵¹, in some cases via specific interaction with respiration complexes¹⁶. Additionally, recent description of C18-ceramide as an adaptor between mitochondria and the autophagolysosome, which led to mitochondrial depletion and subsequent cell death⁵². Additionally, the atypical and recently described deoxysphingolipids were reported to cause mitochondrial fragmentation and inhibit mitochondrial function⁵³. As key players in membrane structure and topology, membrane sphingolipids may regulate numerous functions of membrane-bound proteins⁵⁴ as well as membrane-localized events that are sensitive to alterations in membrane biophysical properties, including signal transduction through membrane signaling platforms, or endocytosis. Thus, while few molecular mechanisms for ceramides in heart have been shown, it seems reasonable that mechanisms established to play roles in heart-relevant processes (e.g. insulin resistance, mitochondrial function) may also apply in the cardiovascular context.

In contrast to ceramide, sphingosine-1-phosphate actions occur via a family of S1P receptors and therefore more mechanistic information is available. In fact, many mechanisms of S1P functions in the cardiovascular system, including maintaining vascular endothelial barrier function and blood pressure, are fairly well understood^{55, 56}. Additional functions of S1P are broad but depending on receptor and context, could mediate heart-relevant processes including inflammation, stem cell and immune cell chemotaxis, fibrogenesis, and others. Furthermore, a couple of intriguing studies provide data supporting receptor-independent functions for S1P as a cofactor of the TNFα signaling protein TRAF2, and as a regulator of histone acetylation^{57, 58}. The latter may be particularly relevant in light of recent studies demonstrating essential roles for chromatin modification in cardiovascular pathophysiology. As with ceramides and other sphingolipids, few of these S1P functions have been demonstrated in cardiovascular contexts; however, these functions in other tissues and systems may provide hints as to furthering our understanding of cardiovascular sphingolipid functions.

Conclusions

Reports continue to paint sphingolipids as dynamic regulators of cardiovascular pathophysiologies. Over the past few years, efforts have been made to pinpoint the function of distinct sphingolipid species and their specific roles in driving cardiomyocyte dysfuction in the context of disease. Therefore, future heart sphingolipid research should both take into consideration the highly interconnected nature of the pathway in terms of compensatory processes and strategies to correctly identify a species of interest. Moreover, in cardiovascular and indeed, all sphingolipid research, investigators should continue to seek to define cellular and molecular mechanisms by which sphingolipids regulate biological functions.

Highlights:

• Sphingolipids contribute to heart pathophysiology in several disease contexts.

• This diverse lipid group comprises thousands of distinct molecules.

• Care must be taken in data interpretation when general inhibitors are used.

• Few molecular mechanisms by which sphingolipids mediate effects are known.

• Mechanisms identified in other models likely function in the cardiovascular context.