Associations of plasma sphingolipid profiles with insulin response during oral glucose testing in Icelandic horses

Abstract Background Sphingolipids modulate insulin sensitivity in mammals. Increased synthesis of ceramides is linked to decreased insulin sensitivity of tissues. Conversely, activation of the insulin signaling pathway can downregulate ceramide synthesis. Elucidating the association between sphingolipid metabolism and insulin response during oral glucose testing may help explain the pathophysiology of insulin dysregulation in horses. Hypotheses Horses with insulin dysregulation will have a plasma sphingolipid profile characterized by increased ceramide concentrations. The plasma sphingolipid profile will have decreased ceramide concentrations after acute activation of the insulin signaling pathway by oral glucose testing. Animals Twelve Icelandic horses. Methods Horses were subjected to an oral glucose test (0.5 g/kg body weight glucose), with plasma insulin concentrations measured at 0, 30, 60, 120, 180, and 240 minutes postglucose administration. Plasma samples were collected at 0 and 120 minutes for sphingolipid profiling using a liquid chromatography‐mass spectrometry‐based metabolomics analysis. Eighty‐three species of sphingolipids were detected, including 3‐ketosphinganines, dihydroceramides, ceramides, dihydrosphingomyelins, sphingomyelins, galatosylceramides, glucosylceramides, lactosylceramides, and ceramide‐1‐phosphates. Results Glucose administration did not significantly alter plasma sphingolipid profiles. C22:0‐ceramide, C24:1‐ceramide, C23:0‐ceramide, C16:1‐sphingomyelin, C22:0‐dihydroceramide, and C24:0‐ceramide were positively correlated with the insulin response (area under the curve). Conclusion and Clinical Importance Positive correlation between the insulin response and sphingolipid concentrations implies upregulated sphingolipid metabolism in insulin dysregulated horses. A high plasma ceramide concentration can indicate insulin dysregulation in horses.


| INTRODUCTION
Increasing attention has been directed to the pathophysiology of sphingolipid metabolism because of their cell signaling potential for modulating insulin sensitivity, apoptosis, 1 and inflammation. 2 Specifically, ceramides were identified as biomolecules inhibiting insulin action by dephosphorylating protein kinase B (Akt), a key element of the intracellular insulin signaling pathway. 3,4 Although the molecular actions of sphingolipids have not yet been confirmed in horses, a higher ceramide concentration is likely to decrease peripheral insulin sensitivity, leading to hyperinsulinemia instead of a normal insulin response. Thus, increased risk of insulin dysregulation (ID) is plausible in horses that have increased plasma ceramide concentrations, considering the action of ceramides on insulin signaling in mammalian cells. 3 In horses, the most common clinically relevant sequela of ID is laminitis, 4 which has been linked to ID and hyperinsulinemia in previous studies. [5][6][7][8] Furthermore, laminitis and laminar separation were shown to be associated with altered tissue sphingolipid concentration. 9,10 These findings suggest that sphingolipids, especially ceramides, could play a pathophysiological role in ID, impacting insulin sensitivity and metabolic health in horses.
Bidirectional crosstalk occurs between the sphingolipid metabolic pathway (illustrated in Figure 1) and the insulin signaling pathway. 2,11 Constitutive activation of phosphatidylinositol 3-kinase (PI3K), a key regulator of the insulin signaling pathway, can inhibit the sphingomyelinase (SM) pathway and suppress ceramide biosynthesis. 12 Furthermore, the mechanistic target of rapamycin (mTOR) signaling pathway, which overlaps with the insulin signaling network, can inhibit de novo sphingolipid synthesis by orosomucoid protein phosphorylation. 13 Conversely, ceramides can attenuate insulin signaling by dephosphorylation of Akt at Ser473, 14 and ceramide-1-phosphate (C1P) can activate PI3K and Akt by rapid phosphorylation. 15,16 Thus, based on the crosstalk between sphingolipid metabolism and insulin signaling, glucose administration during oral glucose testing (OGT) may perturb sphingolipid metabolism and alter the plasma sphingolipid profile.
The OGT is commonly used to classify insulin response in horses. 17,18 The increased plasma glucose concentration during OGT might impair insulin response in ID if insulin resistance in peripheral tissues leads to excessive insulin secretion and hyperinsulinemia. If this insulin resistance is caused by increased ceramide concentrations, it could explain the link between alteration in sphingolipid metabolism and ID. However, the impact of an acute OGT challenge on sphingolipid metabolism in horses has been characterized only for a limited number of sphingolipids, mainly SM. [19][20][21] The correlation between sphingolipids and insulin response has not yet been well established. We hypothesized that the OGT would alter plasma sphingolipid profiles, specifically that the acute activation of the insulin signaling pathway by an OGT would downregulate ceramide biosynthesis. Furthermore, we hypothesized that horses affected by ID during OGT would have a sphingolipid profile characterized by an increased ceramide concentration. We aimed to provide a comprehensive characterization of the plasma sphingolipid profile of healthy and hyperinsulinemic horses during OGT. Our objectives were to assess the shortterm impact of OGT on sphingolipid metabolism by comparing the plasma sphingolipid profiles at 0 and 120 minutes, relative to glucose administration, and to evaluate the correlation between sphingolipids and the insulin response during OGT.

| Horses
Twelve Icelandic horses (8 mares, 3 geldings, and 1 stallion) aged 15 to 29 years were enrolled in the study, including healthy (insulin sensitive) horses and horses previously found to be affected by ID based on PO dynamic testing. The horses had a median body weight (BW) of 400 kg (range, 230-451 kg) and a median body condition score of 6 (range, [3][4][5][6][7][8]

| Oral glucose test
The horses were fasted overnight for approximately 12 hours before testing. For collection of blood samples, a jugular vein catheter (Intraflon 12 G, Vygon, Ecouen, France) was aseptically inserted into a jugular vein 1 hour before sampling and glucose administration. The OGT was performed by nasogastric tube administration of 0.5 g/kg BW glucose (Glucose, WDT, Garbsen, Germany) dissolved in 2 L of water, as reported previously. 22,23 Blood samples were collected immediately before glucose administration, and at 30, 60, 120, 180, and 240 minutes after glucose administration.

| Insulin quantification
Plasma insulin concentrations were measured in duplicate using an insulin ELISA designed for horses (Mercodia AB, Equine Insulin ELISA, Uppsala, Sweden) following the manufacturer's instructions. In cases of insulin concentrations exceeding the upper range of quantification, plasma samples were diluted 1:4 using the manufacturer's commercially available diabetes sample buffer (Mercodia, Diabetes Sample Buffer, Mercodia AB). User-calculated intra-assay coefficients of variation were 4.6%, and 1.9% in the low and high concentration ranges, respectively. Inter-assay coefficients of variation were 9.7%, 6.9%, and 5.2% in the low, medium, and high concentration ranges, respectively. To identify and quantify the sphingolipids in the plasma samples, sphingolipid profiling was performed using ultraperformance liquid chromatography-tandem mass spectrometry in multiple reaction monitoring mode, in which the sphingolipids were targeted in positive ion mode and the phosphorylated sphingolipids were targeted in negative ion mode. The concentration of detected sphingolipids was calculated using the measured peak area in the linear regression calibration curves. metaboanalyst.ca), 25 with AUC ins as the feature of interest. One sample t-tests were performed to test that the correlation (R) was distinct from zero. The P-values from the t test, comparing 0 and 120 minutes sphingolipid concentrations, were adjusted by false discovery rate (FDR) correction using the Benjamini-Hochberg procedure. 26 Individual correlation plots with simple linear regression were drawn using GraphPad Prism 9. The 95% confidence intervals of the best-fit line were shown as dotted lines. The performance of the line of best fit was indicated by the coefficient of determination (R 2 ).
The C20:0-GluCer was the dominant species among the GluCer (0 minute: 0.0298 ± 0.0058 nmol/mL; 120 minutes: 0.0309 ± 0.0072 nmol/mL). The C24:1-LacCer was the dominant species F I G U R E 1 An overview of sphingolipid metabolic pathways, according to Merrill 11 and Maceyka and Spiegel. 2 The synthesis of ceramide begins with the condensation reaction of serine and palmitoyl-CoA to form 3-ketosphinganine. Along the de novo synthesis pathway (red), 3-ketosphinganine is reduced to sphinganine. Sphinganine is coupled with fatty acyl-CoA to form dihydroceramide, and a double bond is added on dihydroceramide to form ceramide. Ceramide is the central hub of the sphingolipid metabolism, which can be modified into sphingomyelin through the sphingomyelinase pathway (blue) and can be modified into glycosphingolipids or phosphorated sphingolipids through the salvage pathway (green). Double-headed arrows indicate a reversible reaction among the LacCer (0 minute: 0.0351 ± 0.0062 nmol/mL; 120 minutes: 0.0349 ± 0.0046 mmol/mL). The C16:0-C1P was the dominant species among the C1P and within the salvage pathway (0 minute: 17.1 ± 1.04 nmol/mL; 120 minutes: 17.0 ± 0.963 mmol/ mL). Collectively, the sphingolipid profiles before and after OGT were not statistically different in the de novo synthesis pathway, SM pathway, and salvage pathway.

| Correlation of sphingolipid concentrations with AUC ins
To evaluate correlations between sphingolipids and the insulin response, 0 and 120 minutes concentrations of all sphingolipids were analyzed separately. The 25 sphingolipids most highly correlated with AUC ins , based on 0 minute sphingolipid concentrations, are presented in Figure 6. The C22:0-DHCer showed the strongest correlation with F I G U R E 2 The distribution of sphingolipids in the de novo synthesis pathway in horses (n = 12), before (0 minute) and after (120 minutes) an oral glucose test. The concentration of each sphingolipid was presented as mean ± SD on a logarithmic scale. Cer, ceramide; DHCer, dihydroceramide; Keto, 3-ketosphinganine

| DISCUSSION
Our objectives were to investigate the impact of OGT on sphingolipid metabolism and to identify sphingolipids that are correlated with the insulin response in horses. Our findings showed that the OGT did not acutely alter the sphingolipid profile over a period of 120 minutes. However, a higher insulin response was associated with higher SM, ceramide, and DHCer concentrations, implying that insulin-dysregulated horses likely had upregulated sphingolipid metabolism. Ours is the first study to provide absolute plasma concentrations of the comprehensive sphingolipid metabolome, covering the various sphingolipid pathways, in horses with a characterized insulin response during OGT. Building on previous metabolomics studies showing that a higher insulin response in horses was significantly associated with a higher plasma glycerophospholipid concentration and lower plasma arginine, acylcarnitines, spermidine, trans-4-hydroxyproline F I G U R E 4 The distribution of sphingolipids in the salvage pathway in horses (n = 12), before (0 minute) and after (120 minutes) an oral glucose test. The concentration of each sphingolipid was presented as mean ± SD on a logarithmic scale. C1P, ceramide-1-phosphate; GalCer, galatosylceramide; GluCer, glucosylceramide; LacCer, lactosylceramide

| Limited acute impact of the OGT on sphingolipid metabolism
We observed that the sphingolipid metabolism was not significantly changed after glucose administration by comparing the sphingolipid profiles between 0 and 120 minutes. The acute challenge of OGT may not have a clinically relevant impact on sphingolipid metabolism, a finding that is in agreement with the results of a previous metabolomics study conducted using a PO sugar test in Welsh Ponies 28 that found that palmitoyl SM concentration did not differ statistically between 0 and 75 minutes of the PO sugar test. Furthermore, another study 19 showed that none of the 15 investigated species of sphingolipid were significantly changed after an acute OGT in horses.
These studies collectively confirm the limited impact of an acute sugar load on sphingolipid metabolism in horses. Our findings further extend our understanding that, in addition to SM, other sphingolipid metabolites such as ceramides, glycosylceramides, and phosphorylated sphingolipids also remain unaffected by an acute OGT.
It is possible however that long-term nutritional interventions, such as high carbohydrate diets or repeated sugar administration, could disrupt sphingolipid metabolism. 29,30 In cats, a plasma metabolomics study showed that plasma C18:1-, C24:0-Cer, and C18:1-, C24:0-SM concentrations were significantly decreased in the glucose supplementary diet group after a 21-day feeding regimen. 31 The concentrations of ceramides and SM also were decreased in dogs on a glucose supplementary diet in the same study. In rats, plasma sphinganine-1-phosphate and sphingosine-1-phosphate concentrations, but not ceramide concentrations, were significantly higher in the high-carbohydrate diet rats than in the control rats over a period  However, in combination with the increase in ceramide concentrations, a decrease in plasma concentration of SM could be an indicator of an upregulated SM pathway and perturbed sphingolipid metabolism, suggesting a higher risk of ID.
Dihydroceramides are the precursors of ceramides in the final step of the de novo synthesis pathway (Figure 1). A potential limitation of our study is that our observations were limited to Icelandic horses. Therefore, we cannot conclude that our findings can be generalized to other breeds, although the link between insulin resistance and ceramide synthesis seems to be evolutionarily conserved in mammals. 3 Because our study population was small (n = 12), larger scale studies using more replicates, and preferably using various breeds, are necessary to extrapolate our data to larger populations of horses. Additional mechanistic experiments focusing on the link between cellular sphingolipid metabolism and insulin resistance are warranted to confirm causative relationships.

| CONCLUSION
Our study is the first to present comprehensive plasma sphingolipid profiles in horses in the context of insulin response during OGT. Our findings help elucidate the relationship between sphingolipid metabolism and insulin response in horses. Six species of sphingolipids were evaluated, including ceramides, SM, and DHCer, providing potential for further investigations in biomarker development and inclusion into diagnostic panels for assessment of ID in horses. However, the pathophysiological role of the identified sphingolipids on developing ID in horses is not yet fully explained. In vitro studies using various sphingolipid species are warranted to elucidate the complex mechanistic regulatory relationships between sphingolipid metabolism and insulin response in horses.

ACKNOWLEDGMENTS
No funding was received for this study. The authors acknowledge the UVic-Genome BC Proteomics Center (Victoria, Canada) for performing the sphingolipid profiling with ultraperformance liquid chromatography-tandem mass spectrometry and the staff assistance at the Clinic for Horses, University of Veterinary Medicine Hannover for taking care of the animals.

CONFLICT OF INTEREST DECLARATION
Authors declare no conflict of interest.

OFF-LABEL ANTIMICROBIAL DECLARATION
Authors declare no off-label antimicrobial use of antimicrobials.