Very-Long-Chain Unsaturated Sphingolipids Mediate Oleate-Induced Rat β-Cell Proliferation

The pancreatic β-cell compensates for an increase in insulin demand, such as in obesity-associated insulin resistance, by expanding its functional mass (1). In rodents and possibly humans, the increase in β-cell mass is at least in part due to replication of existing β-cells (2). Whether insulin resistance is the cause of compensatory β-cell mass expansion has been questioned (3) partly on the basis of observations that insulin hypersecretion can be observed before insulin resistance in human obesity (4). In mice, we (5) and others (6) have shown that short-term high-fat diet leads to enhanced β-cell proliferation before an apparent increase in insulin resistance, suggesting a direct effect of nutrients to promote β-cell replication.

Previously, we showed that 72-h infusions of glucose and lipids in rats stimulate a compensatory increase in functional β-cell mass characterized by β-cell proliferation and insulin hypersecretion, despite progressive β-cell dysfunction (7,8). These findings were reproduced ex vivo by showing that a mixture of 65% monounsaturated fatty acids (MUFAs), 20% polyunsaturated fatty acids (PUFAs), and 15% saturated fatty acids (SFAs) promotes β-cell proliferation in the presence of elevated glucose concentrations in isolated rat and human islets (9). Yet, the differential effects of individual fatty acid (FA) species, and the intracellular FA metabolites mediating these effects, remain to be identified.

In the presence of elevated glucose concentrations, intracellular FA oxidation is inhibited, and FA metabolism is preferentially shifted toward complex lipid synthesis, including acylglycerols and sphingolipids (10). This generates various intracellular signaling molecules that affect β-cell function and survival (11). Among these, sphingolipids, including ceramides, have emerged as important players in metabolic diseases (12). Ceramide accumulation is generally considered as detrimental and mediates glucolipotoxicity-induced apoptosis and β-cell dysfunction (13–16). However, recent findings have suggested that this view may be oversimplistic, as ceramides of different acyl chain lengths differentially affect glucose metabolism: Whereas an increase in C16–C22 (long-chain [LC]) ceramides is correlated with insulin resistance, a rise in C>22 (very-long-chain [VLC]) ceramides improves insulin signal transduction and protects mice from high-fat diet–induced obesity and glucose intolerance (17–19). Likewise, LC and VLC sphingolipids differentially affect cancer cell division and apoptosis (20,21). However, the differential effects of individual ceramide species and their metabolites on islet function has not been investigated.

Ceramides are produced through the de novo synthesis pathway in the endoplasmic reticulum following acylation of sphinganine by ceramide synthase (CerS) (Fig. 1A). As such, ceramides combine a sphingoid base and an amide-linked FA residue with a carbon chain length of C16–C24. Ceramides can be modified to produce sphingomyelins or glucosylceramides in the Golgi apparatus or metabolized to sphingosine, which in turn is phosphorylated by sphingosine kinase (SphK) to generate sphingosine-1-phosphate (S1P). S1P promotes insulin secretion (22) and inhibits apoptosis in β-cell lines (23), and S1P administration promotes β-cell proliferation and inhibits apoptosis in diabetic mouse models (24). Furthermore, loss of SphK or S1P recycling enzymes leads to reduced β-cell mass and proliferation (25) and accelerates diabetes onset (26), respectively. Although these studies suggest a role of SphK/S1P in promoting β-cell function and mass, whether they contribute to β-cell proliferation in response to FAs is currently unknown.

The current study was designed to 1) identify the specific FA species that stimulates β-cell proliferation, 2) assess the potential contribution of SphK/S1P to FA-induced β-cell proliferation, and 3) determine the impact of sphingolipid acyl chain length and saturation on FA-induced β-cell proliferation.

All procedures were approved by the Institutional Committee for the Protection of Animals, Centre de Recherche du Centre Hospitalier de l’Université de Montréal (CRCHUM). Two-month-old male Wistar rats (Charles River Laboratories, Saint-Constant, QC, Canada) were housed under controlled temperature on a 12-h light/dark cycle with free access to water and standard laboratory chow.

Rats underwent catheterization of the jugular vein for systemic infusion of either saline (SAL) (0.9% w/v NaCl) or 70% w/v glucose plus 20% w/v ClinOleic (GLU + CLI) (an olive/soybean triacylglycerol emulsion composed of 65% MUFAs, 20% PUFAs, and 15% SFAs) as previously described (9). For in vivo studies, SKI II was dissolved in polyethylene glycol 400 (95%)/DMSO (5%) and administered by oral gavage once a day to GLU + CLI–infused rats at a final dose of 50 mg/kg. Control GLU + CLI–infused rats received SAL by oral gavage once a day. Measurement of plasma SKI II levels and immunostaining of pancreatic sections are described in the Supplementary Materials.

Rat and mouse islets were isolated by collagenase digestion and Dextran density gradient centrifugation as previously described (27). Batches of 200 islets were cultured in RPMI medium with 10% v/v FBS or in minimum essential medium (MEM) with 10% v/v dialyzed FBS for 48 or 72 h in the presence of glucose (2.8 or 16.7 mmol/L), an FA mixture mimicking the composition of CLI (MIX) (65% oleate, 20% linoleate, and 15% palmitate), or FA alone, as indicated in the figure legends. FAs were complexed with BSA at a 5:1, 2.5:1, or 1.25:1 ratio (0.5, 0.25, 0.125 mmol/L FA, respectively) as indicated. To assess β-cell proliferation, 5-ethynyl-2′-deoxyuridine (EdU) (10 μmol/L) was added throughout the culture period. Media were replaced daily. At the end of the treatment, islets were processed for β-cell proliferation and apoptosis or sphingolipidomic analyses.

For adenoviral transduction, isolated islets were infected at a multiplicity of infection of 1, 10, or 100 plaque-forming units (pfu)/cell as previously described (28). Sixteen hours later, the medium was replaced with complete medium and cultured for an additional 24 h. Infected islets were cultured in the presence of glucose (2.8 or 16.7 mmol/L) with or without oleate (0.5 mmol/L) for 48 h with EdU (10 μmol/L) before measuring β-cell proliferation of green fluorescent protein–positive (GFP⁺) and GFP⁻ cells. To assess knockdown efficiency, mRNA expression was measured by quantitative RT-PCR. β-Cell proliferation and apoptosis, insulin secretion, quantitative RT-PCR, and sphingolipidomic analyses are described in the Supplementary Material.

Batches of 200 islets were cultured in serine-deficient MEM with 10% v/v dialyzed FBS for 2 h and then switched to MEM containing 0.4 mmol/L L-serine-¹³C₃,¹⁵N in the presence of 16.7 mmol/L glucose and palmitate or oleate (0.5 mmol/L) as indicated. After the 16-h exposure, islets were washed with PBS and stored at −80°C. Sphingolipidomic analyses are described in the Supplementary Material.

Data are expressed as mean ± SEM. Statistical analyses were performed using Student t test or one-way or two-way ANOVA with Tukey, Sidak, or Dunnett post hoc test adjustment for multiple comparisons, as appropriate, using GraphPad Prism 9 (GraphPad Software, San Diego, CA). P < 0.05 was considered significant.

All data generated or analyzed during this study are included in the published article (and its online supplementary files). No applicable resources were generated or analyzed during the current study.

Given the reported effects of S1P on β-cell proliferation (24), we first assessed the effect of nutrient excess on S1P concentrations in islets using our rat model of GLU + CLI infusion (7,9). Wistar rats were infused with SAL or GLU + CLI for 72 h. GLU + CLI infusion increased blood glucose in the target range (∼15 mmol/L) (Fig. 1B) and, consistent with previous studies (9), led to an increase in β-cell proliferation as assessed by the percentage of Ki67⁺/insulin-positive (Ins⁺) cells (Fig. 1C and D). S1P concentrations were increased in islet extracts of GLU + CLI–infused rats compared with the SAL-infused group (Fig. 1E), suggesting that GLU + CLI infusions augment sphingolipid metabolism in islets in vivo.

As CLI is composed of MUFAs, PUFAs, and SFAs, we measured the effect of each type of FA on β-cell proliferation in isolated rat islets. Islets were exposed to an FA mixture, mimicking the composition of CLI (MIX), or to palmitate (SFA, C16:0), oleate (MUFA, C18:1), or linoleate (PUFA, C18:2) in the presence of 16.7 mmol/L glucose for 48–72 h. β-Cell proliferation was assessed by immunostaining of islet sections or by flow cytometry. Immunohistochemistry revealed that the FA mixture increased β-cell proliferation as expected (Fig. 2A and B). The increase in β-cell proliferation in response to oleate was comparable to the MIX, but palmitate and linoleate had no effect. Likewise, oleate, but not palmitate, dose-dependently increased β-cell proliferation as assessed by flow cytometry (Fig. 2C–E). Neither oleate nor palmitate affected α-cell proliferation (Supplementary Fig. 1A). A significant increase in β-cell apoptosis was detected in response to palmitate but not oleate (Supplementary Fig. 1B). The other MUFA palmitoleate (C16:1) did not significantly increase β-cell proliferation (Supplementary Fig. 1C). Overall, these results show that among the various FAs tested, only oleate stimulates β-cell proliferation, suggesting that the carbon chain length and the position and degree of saturation are important in this process.

To extend these finding in vivo, using procedures previously described for the infusion of ethyl palmitate in mice (29), we infused ethyl oleate with glucose in rats for 48–72 h. Unfortunately, the rats did not tolerate the infusion well, even at reduced rates, obligating the termination of these studies.

To test whether the effect of oleate on β-cell proliferation in isolated rat islets was applicable to other species, we compared the response of mouse islets. Preliminary experiments indicated that mouse β-cell proliferation was not increased by oleate at 0.5 mmol/L in the presence of 16.7 mmol/L glucose (data not shown), conditions that were effective in rat islets. We therefore performed a dose-response experiment and found that β-cell proliferation was stimulated in the presence of 0.25 mmol/L oleate and 11.1 mmol/L glucose (Fig. 2F). Thus, oleate potentiates glucose-induced β-cell proliferation in both rats and mice, albeit at different concentrations.

Our findings so far suggest that 1) β-cell proliferation in response to elevated glucose levels and an FA mixture is associated with an increase in intraislet S1P levels in vivo and 2) FA stimulation of β-cell proliferation is restricted to oleate. These two observations are seemingly contradictory since palmitate, but not oleate, is the preferential FA substrate for the first step of de novo sphingolipid synthesis. To confirm the implication of this pathway, we first blocked that step with the serine palmitoyl transferase (SPT) inhibitor myriocin. Myriocin dose-dependently decreased oleate-induced β-cell proliferation (Fig. 3A and B). Second, we depleted the culture medium of L-serine, the other substrate of the SPT reaction. Islets cultured in L-serine–deficient medium were unresponsive to oleate, and the addition of L-serine restored the β-cell proliferative response (Fig. 3C).

We then asked whether SphK/S1P contributes to oleate-induced β-cell proliferation. As rat islets express two SphK isoforms, SphK1 and SphK2 (30), we blocked total SphK activity with the inhibitor SKI II. SKI II inhibited oleate-stimulated β-cell proliferation, as assessed by immunohistochemistry (Fig. 3D and E) and flow cytometry (Fig. 3F), but not the response to 16.7 mmol/L glucose alone (Fig. 3F). Treatment with SKI II was not associated with detectable apoptosis (Supplementary Fig. 1D).

To substantiate the findings obtained with SKI II, we infected isolated rat islets with adenoviruses expressing shRNA against SphK1 and SphK2 (Adv-shSphK1 and Adv-shSphK2, respectively). To titrate the viral load and exclude nonspecific effects of the infection on β-cell proliferation, we first infected islets for 48 h with increasing titers of the GFP-expressing control adenovirus (Adv-shCTL) and quantified GFP⁺ β-cells by flow cytometry. Following infection with 1, 10, and 100 pfu/cell, ∼10, 25, and 40% of C-peptide–positive cells were GFP⁺, respectively, and these percentages were similar across culture conditions (Supplementary Fig. 2A). Surprisingly, adenoviral infection at 10 and 100 pfu/cell inhibited oleate stimulation of β-cell proliferation without affecting the response to glucose (Supplementary Fig. 2B). Despite this limitation, we chose to use 100 pfu/cell for optimal gene knockdown. Compared with Adv-shCTL, infection with Adv-shSphK1 and Adv-shSphK2 led to a significant reduction in SphK1 and SphK2 RNA levels, respectively, in GFP⁺ cells at 48 h postinfection (Supplementary Fig. 2C and D). Knockdown of SphK2, but not of SphK1, led to a significant decrease in the proliferative response to glucose and oleate (Fig. 3G). Although knockdown efficiency was lower for SphK1 compared with SphK2 (Supplementary Fig. 2C and D), these data are consistent with the reported weaker expression of SphK1 compared with SphK2 in β-cells (30) and support a dominant role of SphK2.

To investigate the role of SphK in nutrient-induced β-cell proliferation in vivo, we administered SKI II (50 mg/kg/d) by oral gavage to 72-h GLU + CLI–infused rats. SKI II treatment did not affect target blood glucose levels (Supplementary Fig. 3A) or glucose infusion rates (mg/kg/min) in GLU + CLI–infused animals (Supplementary Fig. 3B). Whole-body and pancreas weights at the end of the infusion were reduced in GLU + CLI–infused animals compared with the SAL-infused group but not affected by SKI II treatment (Supplementary Fig. 3C and D). Plasma SKI II levels increased over time but were highly variable between animals by the end of the infusion (Supplementary Fig. 3E). Accordingly, the effect of SKI II on GLU + CLI–induced β-cell proliferation, as assessed by immunostaining for Nkx6.1 and Ki67, was also highly variable and, therefore, not significantly different from the control group (Supplementary Fig. 3F). Hence, we plotted plasma SKI II levels at 72 h against β-cell proliferation in individual animals (Fig. 3H) and noted a significant inverse correlation (r = −0.72), indicating that increased plasma SKI II levels are associated with a decrease in GLU + CLI–induced β-cell proliferation. Taken together, these data suggest that de novo sphingolipid synthesis and SphK activity are necessary for the β-cell proliferative response to oleate.

We then asked whether the products of SphK, S1P, and sphinganine-1-phosphate (Sa1P) mediate the effect of oleate on β-cell proliferation. Surprisingly, S1P levels increased in islets following 48 h exposure to palmitate, but not oleate (Fig. 4A). SKI II, at a concentration shown to inhibit β-cell proliferation (10 μmol/L) (Fig. 3), did not significantly diminish S1P or Sa1P levels in the presence of oleate or 16.7 mmol/L glucose alone (Fig. 4B and C). Finally, blockade of S1P lyase, the enzyme that catalyzes the irreversible degradation of S1P, with 2-acetyl-5-tetrahydroxybutyl imidazole (THI) did not affect β-cell proliferation in response to oleate or 16.7 mmol/L glucose alone (Fig. 4D). These results suggest that the sphingolipid metabolites S1P and Sa1P are not involved in oleate-induced β-cell proliferation.

Although palmitate is the preferential FA for the sphingoid base of ceramides and complex sphingolipids, both oleate and palmitate are substrates for the amino-linked FA portion (Fig. 1A). ELOVLs extend FA chain length to C16–C24, depending on the substrate. Hence, oleate and palmitate might affect islet sphingolipid synthesis at the level of the amino-linked FA chain length rather than total levels of sphingolipid classes. To test for this possibility, we performed a comprehensive and untargeted analysis of sphingolipids by liquid chromatography-tandem mass spectrometry (LC-MS/MS) on lipid extracts following islet treatment with palmitate or oleate for 48 h (Fig. 5A). Principal component analysis revealed major differences in steady-state lipid profiles among vehicle-, palmitate-, and oleate-treated islets (Fig. 5B). Whereas palmitate increased the level of saturated LC and VLC ceramides (C16:0–C24:0), oleate only increased the unsaturated VLC ceramides (C24:1) (Fig. 5C). In addition, palmitate increased C22:0 and C24:0 but decreased C24:1 sphingomyelins, while oleate diminished C16:0–C24:0 and increased C24:1 sphingomyelins (Fig. 5D). Similarly, palmitate increased C16:0 and C22:0 but decreased C24:1 glucosylceramides, whereas oleate diminished C22:0 and increased C24:1 glucosylceramides (Fig. 5E). In a complementary analysis performed under similar conditions, we observed that in the presence of high-glucose concentrations, exposure to oleate increased total triacylglycerols, but not diacylglycerols, whereas palmitate was without effect (Supplementary Fig. 4A), in line with previous studies (31). Moreover, the total amount of islet sphingolipids was not affected by either palmitate or oleate treatment (Supplementary Fig. 4B). However, as with the previous analyses, amino-linked FA chain length was altered: Palmitate increased several saturated LC and VLC sphingolipid species, while oleate decreased these saturated sphingolipids and increased unsaturated VLC (C24:1) sphingolipid species (Supplementary Fig. 4C–G).

To confirm that palmitate and oleate stimulate de novo sphingolipid synthesis, we performed stable isotope labeling of sphingolipids. Islets were cultured in the presence of L-serine-¹³C₃,¹⁵N, and sphingolipids were analyzed following 16 h exposure to palmitate or oleate (Fig. 6A). Palmitate, but not oleate, increased total isotope-labeled dihydroceramides and ceramides (Fig. 6B and C). Importantly, palmitate increased C16:0–C24:0, whereas oleate increased C24:1 isotope-labeled dihydroceramides and ceramides (Fig. 6D and E). Taken together, these data indicate that palmitate and oleate affect amino-linked FA characteristics through de novo sphingolipid synthesis and that oleate exposure leads to increased synthesis of C24:1 species without an increase in total ceramide levels.

As inhibition of SphK with SKI II blocked oleate stimulation of β-cell proliferation without significantly affecting S1P or Sa1P levels, we asked whether this effect could be mediated by changes in other sphingolipid species. We analyzed lipid extracts following islet exposure to oleate in the presence of SKI II. Addition of SKI II significantly increased the total amount of ceramides in oleate-treated islets (Fig. 7A). Although not significant, there was a trend toward an increase in total dihydroceramides, dihydrosphingomyelins, and sphingomyelins in the presence of SKI II. Interestingly, these changes were often associated with an increase in sphingolipids with specific acyl chains. Hence, C24:1 dihydroceramides (Fig. 7B) and C16:0 ceramide and dihydrosphingomyelins (Fig. 7C and D), but not sphingomyelin or glucosylceramide species (Fig. 7E and F), were augmented in response to SKI II exposure. Overall, these data indicate that the inhibition of β-cell proliferation by SKI II is associated with an increase in several C16:0 sphingolipid species.

Our results so far suggest that β-cell proliferation is regulated by the relative balance between individual sphingolipid species: Conditions associated with an increase in C24:1 species, such as in the presence of oleate, are stimulatory, whereas those associated with an increase in C16:0 sphingolipids, such as in the presence of palmitate or SKI II, are neutral or inhibitory. To further ascertain the involvement of C24:1 sphingolipids in β-cell proliferation induced by oleate, we knocked down ELOVL1 or acyl-CoA binding protein (ACBP) in islets exposed to oleate. Knockdown of ACBP is expected to reduce the available pool of LC acyl-CoA for elongation by ELOVL1 and consequently decrease the synthesis of VLC sphingolipids (32,33). Infection of islets with Adv-shELOVL1 or Adv-shACBP significantly diminished their respective mRNA levels (Supplementary Fig. 5A and B). Knockdown of ELOVL1 completely blocked the ability of glucose and oleate to stimulate β-cell proliferation, and knockdown of ACBP significantly decreased the response to oleate, in GFP⁺ (Fig. 8A and B), but not GFP⁻ (Supplementary Fig. 5C and D) cells. Analysis of whole-islet lipid extracts following ELOVL1 knockdown revealed diminished oleate-induced C24:1 and C24:0 dihydroceramide and dihydrosphingomyelin and C24:1 glucosylceramide species (Supplementary Fig. 6). Taken together, these data indicate that the synthesis of VLC sphingolipids via ELOVL1 is necessary for oleate-induced β-cell proliferation.

As oleate (34,35) and sphingolipid metabolism (36) affect β-cell function, we asked whether oleate-induced β-cell proliferation is associated with changes in glucose-stimulated insulin secretion (GSIS). Hence, we measured GSIS in static incubations in islets exposed to 16.7 mmol/L glucose with or without 0.5 mmol/L oleate for 48 h following knockdown of ELOVL1 (Supplementary Fig. 7). In control islets, oleate treatment led to a reduction in GSIS and total insulin content, but not insulin secretion as a percentage of content, compared with glucose alone. Knockdown of ELOVL1 did not affect GSIS or insulin content. These data suggest that oleate negatively affects insulin secretion by reducing insulin content and that this effect is independent of VLC sphingolipid synthesis.

The objectives of this study were to identify the specific FA species stimulating rat β-cell proliferation and to investigate the underlying mechanisms. We showed that among several FAs tested, only oleate potentiates β-cell proliferation in the presence of elevated glucose concentrations. This was unexpectedly associated with increased de novo sphingolipid synthesis, which did not lead to higher S1P and Sa1P levels but to a selective increase in unsaturated VLC (C24:1) sphingolipids. Importantly, blocking VLC (C24:1) sphingolipid synthesis inhibited oleate-induced β-cell proliferation. Overall, these results identify a key role for an FA-CoA-C24:1 sphingolipid axis in the β-cell proliferative response to nutrients (Fig. 8C).

Infusion of GLU + CLI in rats increased β-cell proliferation, consistent with previous studies demonstrating a proproliferative role of a MUFA-enriched emulsion in vivo (7–9,37). However, as our attempts to infuse glucose and oleate in vivo were not successful, whether oleate promotes β-cell proliferation in vivo remains unknown. Nevertheless, in rat islets ex vivo, we demonstrated that a lipid mixture mimicking the FA composition of CLI and, importantly, the MUFA oleate (C18:1) alone, increased β-cell proliferation in the presence of glucose. Oleate also potentiated glucose-induced β-cell proliferation in isolated mouse islets. Although we previously demonstrated that an FA mixture containing predominantly oleate (65%) in the presence of glucose stimulates β-cell proliferation in adult human islets (9), in the current study, we did not test oleate alone. Thus, whether oleate promotes human β-cell proliferation remains to be determined. Although previous studies in INS-1 cells reported an inhibition of glucose-induced β-cell proliferation by palmitate and linoleate (38), neither FA reduced β-cell proliferation in rat islets ex vivo in the current study. In agreement with our findings, a 1:1 mixture of oleate and palmitate was shown to promote rat β-cell proliferation (39). However, palmitate did increase β-cell apoptosis, as shown previously (40). Collectively, these findings highlight the key role of chain length and degree of saturation in FA-induced β-cell proliferation, as reported previously for FA-induced β-cell dysfunction (16,41).

Inhibition of β-cell proliferation in the presence of the SPT inhibitor or following depletion of L-serine demonstrated that de novo sphingolipid synthesis is implicated in the response to oleate. Our observations that pharmacological inhibition or knockdown of SphK inhibited oleate-induced β-cell proliferation were suggestive of a role for the SphK metabolites S1P and Sa1P, as shown in cancer cell lines (42). Surprisingly however, oleate did not increase S1P or Sa1P levels in isolated islets, suggesting that the rise in islet S1P levels following GLU + CLI infusion was likely due to the presence of palmitate in the GLU + CLI emulsion. In addition, although SKI II decreased oleate-induced β-cell proliferation, it did not alter S1P or Sa1P levels in isolated islets. Furthermore, blocking S1P degradation with the S1P lyase inhibitor had no effect on β-cell proliferation. This led us to conclude that oleate-induced β-cell proliferation is not mediated by S1P or Sa1P. Although SphK/S1P exerts a critical role in β-cell survival in response to lipotoxic or cytokine stress in vitro (23,43) and in response to high-fat feeding in vivo (26), the role of S1P in β-cell proliferation is unresolved. Whereas β-cell proliferation increases in streptozotocin-induced diabetic mice following S1P administration (24), inhibiting S1P metabolism in S1P phosphatase 2 knockout mice reduces β-cell proliferation in response to high-fat feeding (25). Furthermore, exogenous S1P compromises insulin-induced β-cell proliferation in MIN6 cells (44).

In cancer cell lines, SKI II inhibits not only SphK activity but also dihydroceramide desaturase 1 activity, leading to a loss of S1P and an accumulation of dihydroceramide and dihydrosphingomyelin (45). Although no significant changes in S1P were detected in oleate-treated rat islets exposed to SKI II, an increase in total ceramides, specifically C16:0 ceramides, as well as C24:1 dihydroceramides and C16:0 dihydrosphingomyelins was observed. Given that ceramide levels are at least an order of magnitude higher than dihydroceramides and dihydrosphingomyelins, changes in ceramide levels are likely to be more biologically significant. The decrease in oleate-induced β-cell proliferation following depletion of SphK activity might be due to a change in the relative balance of specific ceramide species, notably the C16:0/C24:1 ratio. In support of this possibility, whereas S1P promotes cell survival and proliferation, ceramides mediate apoptosis and growth inhibition. Antagonistic effects between sphingolipid species have led to the concept of the “sphingolipid rheostat,” highlighting a critical role of the dynamic balance between ceramides and S1P in the control of cell biology (46). Furthermore, N-acyl chain length significantly affects the biological activities of ceramides (47). Whereas CerS2^−/− mice present glucose intolerance associated with an increase in C16 and a decrease in C24 ceramides in the liver (19,48), loss of CerS6 protects from diet-induced obesity and insulin resistance by decreasing C16 ceramides (18,49). In the β-cell, C14:0–C24:0 ceramides are associated with palmitate-mediated apoptosis and inhibition of insulin secretion (13,14). Similarly, in cancer cells, upregulation of CerS4 and 6 leads to an increase in C16:0–C20:0 ceramides and inhibition of cell proliferation (20). Further studies will be required to determine whether saturated LC and VLC ceramide species negatively regulate β-cell proliferation. In addition, whether potential changes in the levels of other sphingolipid species, including galactosylceramides, sulfatides, and gangliosides of different chain length and saturation, following SphK inhibition contribute to the reduction of β-cell proliferation under these conditions remains to be investigated.

In the presence of high-glucose concentrations, oleoyl-CoA is preferentially shunted toward complex lipid synthesis, and palmitoyl-CoA is used in de novo sphingolipid synthesis, whereas both are substrates for acyl chain elongation and ceramide synthesis through the action of ELOVL and CerS, respectively. In agreement with this model, we found that palmitate increased saturated LC and VLC (C16:0–C24:0) dihydroceramides and ceramides as well as dihydrosphingomyelins, sphingomyelins, and glucosylceramides. Using labeled serine, we observed an increase in ceramide flux characterized by a significant rise in total amount of labeled dihydroceramides and ceramides in response to palmitate. These results are consistent with previous studies that described an increase in ceramide flux in palmitate-exposed MIN6 cells (50). In contrast to palmitate, oleate did not change global sphingolipid flux or steady-state levels. Instead, oleate specifically increased the synthesis of unsaturated C24:1 sphingolipids, including dihydroceramides, ceramides, dihydrosphingomyelins, sphingomyelins, and glucosylceramides. Although oleate does not affect levels of ceramide species in MIN6 cells (14), to our knowledge, the effect of oleate on the synthesis of other sphingolipid species in the β-cell had not been previously described.

Knockdown of ELOVL1 or ACBP in rat islets inhibited oleate-induced β-cell proliferation. By elongating SFA and MUFA to generate C24:0 and C24:1 acyl-CoA, ELOVL1 controls the synthesis of VLC sphingolipids (33). ELOVL1 knockdown in HeLa cells and knockout in mice dramatically reduce C24 sphingolipid levels (21,51). Accordingly, we found that depletion of ELOVL1 diminished the concentrations of several C24:1 sphingolipid species in response to oleate. By binding acyl-CoA, modulating expression of genes involved in FA metabolism and interacting with CerS and ELOVL1, ACBP regulates VLC sphingolipid synthesis (32). Accordingly, ACBP knockout mice exhibit a significant decrease in VLC acyl-CoA (32). In view of the effect of oleate on sphingolipid metabolism described above, we conclude that C24:1 sphingolipids contribute to oleate-induced β-cell proliferation. Interestingly, unlike oleate, palmitoleate, an n-7 MUFA not elongated into VLC acyl-CoA by ELOVL (33), did not induce β-cell proliferation. We conclude that oleoyl-CoA is likely among a limited number of FA-CoA species that provide the necessary substrate for ELOVL1 to increase the pool of C24:1 acyl-CoA leading to the synthesis of C24:1 sphingolipids and β-cell proliferation.

In line with previous studies (34,35), we found that chronic exposure to oleate leads to a reduction in β-cell function characterized by reduced GSIS and insulin content. The reduction in insulin content is likely due to oleate-stimulated insulin hypersecretion in the absence of a corresponding increase in proinsulin biosynthesis, as shown previously (35). In contrast, blocking VLC sphingolipid synthesis by ELOVL1 knockdown did not affect GSIS or insulin content. These findings were unexpected as sphingolipid metabolism and ELOVL2 positively regulate insulin secretion (36,52) and mitigate β-cell dysfunction (53) under glucolipotoxic conditions. Nevertheless, these data suggest that unlike its stimulation of β-cell proliferation, the mechanisms underlying the negative effect of oleate on β-cell function are not related to the synthesis of VLC sphingolipids. Instead, other lipid species, including the noncanonical phospholipids ether lipids, which were shown to be depleted in oleate-treated MIN6 cells (54), may underlie defective GSIS.

In conclusion, our results show that oleate-induced β-cell proliferation in isolated rat islets requires the synthesis of C24:1 sphingolipids that is regulated by the availability of oleate. Our findings contribute to the understanding of chain length–specific properties of sphingolipids and support a role for an FA-CoA-C24:1 sphingolipid axis in adaptative β-cell mass expansion in response to nutrients.

Acknowledgments. The authors thank Caroline Tremblay from CRCHUM for valuable technical assistance, Grace Fergusson and Mélanie Éthier from the Rodent Metabolic Phenotyping Core of CRCHUM for assistance with in vivo studies and islet isolations, Julien La Montagne and Alexia Grangeon from the Metabolomics Core of CRCHUM for S1P and SKI II measurements, and Dominique Gauchat and Philippe St-Onge from the Flow Cytometry Core of CRCHUM for assistance with measurements of β-cell proliferation.

Funding. This study was supported by the National Institutes of Health (grants R01-DK58096 to V.P.; DK115824, DK116888, and DK116450 to S.A.S.; and DK108833 and DK112826 to W.L.H.), the Canadian Institutes of Health Research (grant MOP 77686 to V.P.), the JDRF (grant 3-SRA-2019-768-A-B to S.A.S.), the American Diabetes Association (to S.A.S.), the American Heart Association (to S.A.S.), and the Margolis Foundation (to S.A.S.). It benefited from the support of the Canadian Foundation for Innovation (CFI) and the Montreal Heart Institute (MHI) Foundation for the MHI Metabolomic Platform (CFI grants 20415 and 36283 to C.D.R., co-principal investigator). A.-L.C. was supported by the Association pour la Recherche sur le Diabète, the Société Française d’Endocrinologie et Diabétologie Pédiatriques, and Société Francophone du Diabète. V.S.M. was supported by a postdoctoral fellowship from CRCHUM. M.R. is an FRQS Junior 1 Scholar.

Duality of Interest. S.A.S. is a consultant, cofounder, and shareholder of Centaurus Therapeutics. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. A.-L.C. and A.V. contributed to the conceptualization, methodology, investigation, formal analysis, and writing of the original draft of the manuscript. T.S.T. contributed to the investigation and formal analysis. I.R.F., Z.E.A., V.S.M., S.A.C., and M.R. contributed to the investigation. J.G. contributed to the conceptualization, validation, review and editing of the manuscript, and supervision. C.D.R., W.L.H., S.A.S., and V.P. contributed to the conceptualization, validation, review and editing of the manuscript, supervision, funding acquisition, and project administration. V.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 79th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 7–11 June 2019.