Sphingosine 1-Phosphate (S1P) Regulates Glucose-stimulated Insulin Secretion in Pancreatic Beta Cells*

Background: Sphingolipids play an important role in glucose homeostasis. Results: High glucose induces SphK activity, leading to increases in S1P levels and stimulation of insulin secretion. Conclusion: SphK activity and S1P levels are critical for glucose-stimulated insulin secretion. Significance: The presented data uncover a new role for SphK and S1P in regulation of insulin secretion by glucose. Recent studies suggest that sphingolipid metabolism is altered during type 2 diabetes. Increased levels of the sphingolipid ceramide are associated with insulin resistance. However, a role for sphingolipids in pancreatic beta cell function, or insulin production, and release remains to be established. Our studies in MIN6 cells and mouse pancreatic islets demonstrate that glucose stimulates an intracellular rise in the sphingolipid, sphingosine 1-phosphate (S1P), whereas the levels of ceramide and sphingomyelin remain unchanged. The increase in S1P levels by glucose is due to activation of sphingosine kinase 2 (SphK2). Interestingly, rises in S1P correlate with increased glucose-stimulated insulin secretion (GSIS). Decreasing S1P levels by treatment of MIN6 cells or primary islets with the sphingosine kinase inhibitor reduces GSIS. Moreover, knockdown of SphK2 alone results in decreased GSIS, whereas knockdown of the S1P phosphatase, Sgpp1, leads to a rise in GSIS. Treatment of mice with the sphingosine kinase inhibitor impairs glucose disposal due to decreased plasma insulin levels. Altogether, our data suggest that glucose activates SphK2 in pancreatic beta cells leading to a rise in S1P levels, which is important for GSIS.

The onset and progression of type II diabetes have been linked to alterations in sphingolipid metabolism (1)(2)(3)(4). Studies show that ceramide levels are significantly increased in the blood, muscle, liver, and adipose tissue of diabetic rodents and humans (1,(5)(6)(7). Increased levels of ceramide correlate with increased insulin resistance, suggesting that ceramide plays an important role in promoting insulin resistance (1,2,4,7). Consistent with this idea, inhibition of ceramide synthesis enhances insulin sensitivity (8 -10). Recent data indicate that ceramide may also inhibit insulin gene transcription in pancreatic beta cells under diabetic conditions (11,12). In addition, ceramide, which is known to induce apoptosis in many cell types, has been shown to mimic cytokine-induced beta cell death (13,14).
The sphingolipid, sphingosine 1-phosphate (S1P), 2 which opposes the actions of ceramide, also has an effect on beta cell function. S1P protects beta cells from cytokine-induced cell death (14). S1P may also have a role in insulin secretion. Exogenously added S1P promotes insulin secretion in HIT-T15 cells and mouse pancreatic islets (15). S1P acts as a signaling molecule by binding both extracellular and intracellular targets (16,17). Extracellular targets include the five known S1P receptors, S1P [1][2][3][4][5] , four of which (S1P [1][2][3][4] ) are expressed in the rat pancreatic INS1 cell line and mouse pancreatic islets (18). S1P receptors are G protein-coupled receptors; therefore, upon S1P binding, many downstream signaling pathways are activated, including PI3K/Akt, Ras/Erk, PLC, Rho, Rac, and adenylate cyclase (16). The intracellular targets of S1P are largely unknown. However, a recent study discovered that intracellular S1P inhibits HDAC1/HDAC2 activity and thereby regulates histone acetylation and gene transcription (19). S1P is synthesized by phosphorylation of sphingosine via sphingosine kinases (SphKs). There are two sphingosine kinase isoforms, SphK1 and SphK2 (20 -22). SphK1 has been shown to have a role in insulin signaling and in maintaining glucose homeostasis. Specifically, overexpression of SphK1 in KK/Ay diabetic mice leads to decreases in blood glucose levels (23). In addition, high glucose conditions stimulate SphK1 activity in endothelial cells (24). In support of the idea that S1P may be involved in promoting beta cell viability, SphK activity is increased in the presence of cytokines in INS-1 cells and mouse pancreatic islets (14). Furthermore, of the two isoforms, SphK2 appears to be the dominant active isoform in pancreatic beta cells (25). There are two isoforms of sphingosine-1-phosphate phosphatases, Sgpp1 and Sgpp2, that dephosphorylate S1P to sphingosine (20). To date, neither isoform has been directly implicated in contributing to diabetes.
In this study, we investigated the role of S1P in pancreatic beta cell function and discovered that exposure to high glucose increases S1P levels by activation of sphingosine kinase isoform 2. Furthermore, we demonstrate that the increase in S1P level is important for glucose-stimulated insulin secretion.

EXPERIMENTAL PROCEDURES
Animals-C57BL6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were housed at the University of Kentucky Animal Facility. All experiments carried out with animals were performed according to protocols approved by the University of Kentucky Institutional Animal Care and Use Committee (IACUC).
Cell Culture-Mouse insulinoma-6 (MIN6) cell line was maintained as described previously (26). Experiments were performed using MIN6 cells grown to passages 22-30. Insulin Enzyme-linked Immunosorbent Assay (ELISA)-ELISA was carried out according to the manufacturer's specifications (Mercodia Inc.). All insulin secretion assays were run in duplicates, and results were expressed as -fold differences (1 mM glucose control ϭ 1-fold) in secreted insulin. Insulin levels were normalized to total protein.
Glucose Tolerance Test-Mice were i.p. injected with DMSO or sphingosine kinase inhibitor (SKI) (65 mg/kg) followed by a 4-h starvation period. After starvation, 2 g of glucose/kg of body weight was i.p. injected. Blood glucose (mg/dl) was measured every 15 min up to 120 min. Blood samples were collected to measure insulin levels using the insulin ELISA.
Insulin Secretion Assay-Cells were washed with 1ϫ PBS and preincubated on DMEM with 1 mM glucose for 16 h. Following this incubation, cells were incubated with freshly made KRBB containing 1 mM glucose for 2 h. For experiments performed with SKI (Calbiochem), cells were incubated with 1 mM glucose in KRBB for 1.5 h followed by a 30-min incubation on KRBB with 1 mM glucose with or without 10 M SKI or DMSO. Following preincubations, cells were incubated in KRBB with 1 or 25 mM glucose for 1 h, at the end of which supernatant was collected and insulin was quantified using ELISA. Insulin secretion in purified islets was assayed by pretreatment of 50 purified islets with KRBB buffer containing 3 mM glucose with or without 40 M SKI or DMSO for 1h. After the pretreatment, islets were transferred into KRBB buffer containing 3 or 20 mM glucose with or without SKI (40uM) for 1 h, and the amount of secreted insulin was quantified using the insulin ELISA assay.
In Vitro Sphingosine Kinase Assay-The sphingosine kinase assay is a modification of a previously described assay (27) and was carried out using the assay buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM MgCl 2 , 1 mM DTT, and 3 M sodium orthovanadate) with 1 mM ATP, 20 M C 17 sphingosine, 10 M C 18 S1P (internal standard), and cell lysate at 37°C for the indicated times. The reaction was stopped by adding 0.4 ml of 0.1 M HCl. Lipids were extracted and quantified using LC-MS/MS.
Lipid Extraction and Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS)-Sphingolipids were extracted and quantified as described previously (28). Extracted lipids were quantified using the Applied Biosystems 4000 Q-Trap mass spectrometer. The obtained data were normalized to lipid recovery, using the C 17 S1P control, and to total starting material, using lipid PO 4 .
Pancreatic Islet Isolation-Pancreatic islets were isolated from C57BL6 mice. Pancreata were removed aseptically via bile duct injection of Hanks' balanced salt solution (Invitrogen) containing 0.5 mg/ml collagenase type V (Sigma) and incubated at 37°C. Collagenase digestion was stopped by adding chilled Hanks' balanced salt solution with 10% fetal calf serum. Islets were purified using a Histopaque gradient (Histopaque-1077, Sigma). The purified islets were incubated in RPMI 1640 with 10% FBS. Islets were handpicked into KRBB and transferred to low (3 mM) or high (20 mM) glucose in KRBB for the indicated time points.
siRNA Knockdown-siRNA knockdown experiments were performed as described previously (29,30). siRNAs were transfected into MIN6 cells using the Amaxa cell line Nucleofector kit V (Lonza) according to the manufacturer's specifications. About 4 g of siRNA per 1.0 ϫ 10 Ϫ6 cells was used to achieve optimal knockdown. The sequence of the siRNAs is listed in supplemental Table 1.
Statistical Analysis-The Student's paired t test was used to determine statistical significance. All experiments analyzed for significance were performed at a minimum of three times. A p value Ͻ 0.05 was considered statistically significant. Error bars represent Ϯ S.D.

S1P Levels Are Increased under High Glucose Conditions in
Pancreatic Beta Cells-Sphingolipid levels are known to be altered in obesity and during type 2 diabetes, indicating that these lipids may be involved in regulating glucose homeostasis. Because the production and secretion of insulin in pancreatic beta cells are regulated by changes in glucose levels, we measured S1P levels in the mouse insulinoma cell line MIN6 after incubation with 1 or 25 mM glucose. Exposure of MIN6 cells to high glucose (25 mM) for 1 h resulted in about 2-fold increase in S1P levels when compared with 1 mM glucose incubated cells (Fig. 1A), whereas there were no significant changes in the levels of other sphingolipids, such as sphingomyelin and ceramide (supplemental Fig. S1). We observed a similar increase in S1P levels in purified mouse pancreatic islets by treatment with 20 mM versus 3 mM glucose (Fig. 1B). These data suggest that S1P levels are regulated by changes in glucose levels in pancreatic beta cells.
High Glucose Stimulates SphK Activity in MIN6 Cells-To determine the mechanism(s) by which glucose mediates changes in S1P levels, we analyzed SphK expression, localization, and activity. Incubation of MIN6 cells with 1 or 25 mM glucose did not alter SphK1 and SphK2 expression, protein stability, or localization (supplemental Figs. S2 and S3). However, we found that SphK activity was increased over 4-fold in MIN6 cells incubated with high glucose (Fig. 2). Previous data indicate that phorbol 12-myristate 13-acetate (PMA) stimulates SphK activity in other cell types (32). Incubation of MIN6 cells with 1 mM glucose in the presence of 100 nM PMA stimulated SphK activity by 8-fold when compared with 1 mM glucose alone (Fig.  2). SphK activity in cell lysates was measured by quantification of newly formed S1P (C 17 S1P). The SphK activity assay was not linear, which is likely due to the low level of constitutive degradation of S1P by S1P phosphatases. In summary, exposure of MIN6 cells to high glucose stimulates SphK activity in pancreatic beta cells. S1P Levels Are Important for GSIS-Because SphK activity is induced by high glucose and S1P levels are significantly increased in the presence of high glucose, we tested whether changes in S1P levels are important for glucose-stimulated insulin secretion (GSIS). First, we blocked the production of S1P using SKI (Fig. 3C). MIN6 cells treated with 10 M SKI displayed a significant drop in insulin secretion on 25 mM glucose and completely lacked GSIS when compared with untreated cells (Fig. 3A). Treatment of MIN6 cells with SKI did not affect total insulin levels (supplemental Fig. S4). Furthermore, treatment of isolated islets with 40 M SKI for 1 h significantly diminished GSIS (Fig. 3B). These data suggest that the rise in S1P levels in response to high glucose is important for GSIS. To further substantiate the role of S1P in GSIS in pancreatic beta cells, we either blocked S1P production on high glucose by knockdown of SphKs or increased S1P levels on low glucose by knockdown of Sgpp1. Knockdown of both SphK1 and SphK2 by transfection of MIN6 cells with the corresponding siRNA oligonucleotides completely abolished GSIS (Fig.  4A). Surprisingly, knockdown of SphK1 alone did not have a significant effect on insulin secretion (Fig. 4B), although knockdown of SphK1 reduced overall S1P levels by 30% (Fig. 4D). However, knockdown of SphK2 in MIN6 cells using four different siRNA oligonucleotides reproducibly resulted in complete lack of GSIS when compared with cells transfected with control siRNA (Fig. 4C). Knockdown of SphK2 resulted in a 75% decrease of S1P levels (Fig. 4D). Thus, knockdown of SphK2 completely abolished GSIS. The efficiency of SphK1 and SphK2 knockdown in MIN6 cells was confirmed as shown in supplemental Fig. S5. Knockdown of SphK2 did not affect cell viability when compared with cells transfected with control siRNA (supplemental Fig. S6). In support of an important role for Sphk2 in beta cells, glucose induction of SphK activity was reduced by 6-fold in MIN6 cells transfected with SphK2 siRNAs when compared with control cells (Fig. 5). These data suggest that the increase in S1P levels by high glucose is likely due to activation of SphK2 in the presence of glucose.
Consistent with a role for S1P in GSIS, knockdown of Sgpp1 did not have any effect on insulin secretion on high glucose, but increased insulin secretion by about 3-fold on low glucose (Fig.  6A). Sgpp1 dephosphorylates S1P to sphingosine, thereby reducing S1P levels in the cells. Knockdown of Sgpp1 resulted in an about 50% increase of S1P levels (Fig. 6B) that led to an increase in insulin secretion (Fig. 6A). The efficiency of Sggp1 knockdown using four different siRNA oligonucleotides was confirmed as illustrated in supplemental Fig. S7.
Intraperitoneal Injection of SKI into Mice Impairs Glucose Disposal-Because SKI inhibits GSIS in MIN6 cells, we tested whether SKI treatment of mice affected insulin secretion and thereby glucose disposal during the intraperitoneal glucose tolerance test. For this purpose, 10-week-old C57BL6 mice were i.p. injected with DMSO (control mice) or with 65 mg/kg of body weight of SKI and fasted for 4 h. Following the fasting period, the mice were i.p. injected with 2 g/kg of body weight of glucose and blood glucose, and insulin levels were determined (Fig. 7). Within 30 min after intraperitoneal injection of glucose, the SKI-treated mice had significantly higher blood glucose levels than control mice (Fig. 7, A and B).
The glucose intolerance observed in SKI-treated mice correlated with decreased serum insulin levels when compared with   APRIL 13, 2012 • VOLUME 287 • NUMBER 16 control mice (Fig. 7, C and D). Plasma insulin levels in SKItreated mice were significantly lower than in DMSO-treated control mice (Fig. 7C). The intraperitoneal glucose tolerance test experiments suggest that treatment of mice with the sphingosine kinase inhibitor SKI inhibits insulin secretion and thereby results in impaired glucose disposal.

DISCUSSION
In this study, we have presented evidence that sphingosine kinase activity is regulated by glucose in pancreatic beta cells. Exposure of beta cells to high glucose induces SphK activity, leading to a rise in S1P levels. The increase in S1P levels during high glucose conditions is important for GSIS. Inhibition of S1P production by treatment of cells with the sphingosine kinase inhibitor SKI or by knockdown of SphK2 using siRNAs caused a complete lack of GSIS in pancreatic beta cells. On the other hand, knockdown of Sgpp1, which normally dephosphorylates S1P, led to increased insulin secretion even on low glucose. We conclude from these data that S1P has a novel role in pancreatic beta cells and that changes in S1P levels are important for GSIS.
Although it has previously been reported that exogenously administered S1P to HIT-T15 beta cells stimulates insulin secretion, it was not clear from this previous study whether the addition of exogenous S1P mimicked GSIS (15). Extracellular S1P can function as a signaling molecule by binding to five G protein-coupled S1P receptors, but intracellular S1P can also activate distinct signal pathways (33,34). We believe that the regulation of GSIS by S1P in pancreatic beta cells involves the activation of intracellular targets of S1P because the rise in S1P coincides with GSIS. Furthermore, we could not detect any extracellular S1P in the medium of MIN6 cells using LC-MS/MS (data not shown).
Regulation of SphK activity and S1P levels by glucose has also been reported previously in endothelial cells (24), suggesting that S1P may play a key role in maintaining glucose homeostasis. Consistent with our data, a recent study also demonstrates that SphK activity is regulated by high glucose in rat pancreatic islets (35).
Knockdown of SphK1 in MIN6 cells had no effect on GSIS. This is interesting considering the fact that knockdown of SphK1 reduced S1P levels by 30% without having an effect on GSIS. It is possible that the overall increase in S1P levels is not important for GSIS, but increases in S1P levels at specific locations or compartments, such as the granules, may be required for GSIS. Knockdown of SphK2 led to a 75% decrease in S1P levels and abolished GSIS in MIN6 cells. This suggests that the observed increase in SphK activity mediated by high glucose is due to activation of SphK2 and that the rise in S1P levels by glucose is likely to be mediated by activation of SphK2. Indeed, knockdown of SphK2 drastically reduced the glucose-induced SphK activity in MIN6 cells (Fig. 5), confirming that the rise in S1P levels on high glucose is mainly due to SphK2 activity. Consistent with this idea, a previous study reported that SphK2 is the major active isoform in pancreatic beta cell lines and rat islets (25).
Although the mechanism(s) by which glucose activates SphK2 in pancreatic beta cells remains to be established, it is most likely that glucose stimulates SphK activity by a posttranslational modification. It has been previously shown that PMA induces the phosphorylation and activation of both SphK1 and SphK2 by stimulating the activation of protein kinase C (PKC) and extracellular signal-related kinases 1/2 (ERK1/2) (36 -38). These same signaling pathways are also important for glucose-induced beta cell function. Our data show that PMA induces SphK activity on low glucose conditions and that this increase in activity is comparable with that seen on high glucose.
Glucose tolerance experiments in mice indicate that intraperitoneal injection of SKI impairs glucose disposal in mice due to inhibition of insulin secretion. These data suggest that S1P has an important role in GSIS in rodents. Although deletion of SphK1 and SphK2 in mice causes lethality, SphK1 or SphK2 single knock-out mice are viable (39,40). Neither the SphK1 nor the SphK2 knock-out mice have been reported to be diabetic. Our data presented here suggest that SphK2 activity is essential for GSIS in the MIN6 cell line. Thus, it is expected that  mice deleted for SphK2 should have defects in insulin secretion. One explanation for why SphK2 knock-out mice are not diabetic could be that SphK1 compensates for the lack of SphK2. Alternatively, other compensation mechanisms could be active in SphK2 knock-out mice, such as decreased S1P phosphatase activity. However, detailed studies on insulin secretion and glucose disposal in SphK2 knock-out mice are lacking.
Although S1P levels increased during incubation with high glucose within the pancreatic beta cells, the levels of other sphingolipids, including sphingomyelin and ceramide, were not changed. A large body of evidence indicates that chronic hyperglycemia and hyperlipidemia, caused by nutrient excess and diabetes, results in dramatic increases in ceramide levels (1,(5)(6)(7)9). Ceramide and S1P are known to have opposing actions.  Mice treated with sphingosine kinase inhibitor SKI have impaired glucose tolerance and decreased plasma insulin levels. 10-week-old C57BL6 mice were injected with DMSO (Control) or 65 mg/kg of body weight of SKI and starved for 4 h. After starvation, the mice were i.p. injected with 2 g/kg of body weight with glucose (intraperitoneal glucose tolerance test), and blood glucose (A and B, Control, n ϭ 6; SKI-treated mice, n ϭ 5) and plasma insulin (C and D, n ϭ 4) levels were monitored every 15 min and quantified using insulin ELISA. *, p Ͻ 0.05. AUC, area under the curve.
Typically, when ceramide levels are increased, S1P levels are decreased and vice versa. Therefore, it is likely that under normal conditions, S1P has an important function in pancreatic beta cells and regulates GSIS, but under diabetic conditions, increases in ceramide levels may oppose the actions of S1P, contributing to defects in insulin secretion. Because ceramide appears to negatively affect glucose-stimulated insulin gene transcription, it is possible that it may have the same affect on insulin secretion during diabetes (11). Thus, it would be interesting to determine whether S1P levels are altered within pancreatic islets of diabetic animals, and if so, whether activation of SphK2 within diabetic islets restores impaired GSIS.
In summary, consistent with previous data that indicate an important role for sphingolipids in glucose homeostasis, we demonstrate in this study that changes in S1P levels in pancreatic beta cells impact GSIS. Furthermore, we show that changes in S1P levels are mediated by regulation of SphK2 activity by glucose. Thus, altered S1P levels may interfere with normal glucose homeostasis and contribute to metabolic disorders such as diabetes. Detailed studies on S1P function in pancreatic beta cells may contribute to the development of novel strategies for the treatment of diabetes.