* This work was supported by National Institutes of Health Grant R01-DK58784 (to S. A. S.), the American Diabetes Association (to S. A. S.), Deutsche Forschungsgemeinschaft Grant SFB 645 (to K. S.), and the Ben and Iris Margolis Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Recent studies indicate that insulin resistance and type 2 diabetes result from the accumulation of lipids in tissues not suited for fat storage, such as skeletal muscle and the liver. To elucidate the mechanisms linking exogenous fats to the inhibition of insulin action, we evaluated the effects of free fatty acids (FFAs) on insulin signal transduction in cultured C2C12 myotubes. As we described previously (Chavez, J. A., and Summers, S. A. (2003) Arch. Biochem. Biophys. 419, 101–109), long-chain saturated FFAs inhibited insulin stimulation of Akt/protein kinase B, a central regulator of glucose uptake and anabolic metabolism. Moreover, these FFAs stimulated the de novo synthesis of ceramide and sphingosine, two sphingolipids shown previously to inhibit insulin action. To determine the contribution of either sphingolipid in FFA-dependent inhibition of insulin action, we generated C2C12 myotubes that constitutively overexpress acid ceramidase (AC), an enzyme that catalyzes the lysosomal conversion of ceramide to sphingosine. AC overexpression negated the inhibitory effects of saturated FFAs on insulin signaling while blocking their stimulation of ceramide accumulation. By contrast, AC overexpression stimulated the accrual of sphingosine. These results support a role for aberrant accumulation of ceramide, but not sphingosine, in the inhibition of muscle insulin sensitivity by exogenous FFAs.
The peptide hormone insulin stimulates the uptake and storage of glucose into skeletal muscle while simultaneously repressing glucose efflux from the liver. Insulin resistance occurs when a normal dose of the hormone is incapable of eliciting these anabolic responses, and the condition is a major contributor to the pathogenesis of Type 2 Diabetes Mellitus (
), but the relationship between increased fat stores and the development of insulin resistance in tissues other than adipose is unclear. Researchers hypothesize that increased lipid oversupply to tissues not suited for fat storage accounts for the decreased insulin sensitivity, perhaps by promoting the accumulation of fat-derived metabolites that inhibit insulin signaling and action (
). The studies described herein investigate the role of ceramide, sphingosine, and diacylglycerol as potential mediators of the insulin resistance induced by saturated fats.
Insulin initiates its pleiotropic actions through its heterotetrameric receptor with intrinsic tyrosine kinase activity. The activated receptor phosphorylates a family of “insulin receptor substrates” (IRS
). One of these docking proteins, the lipid kinase phosphatidylinositol 3-kinase, catalyzes the phosphorylation of specific phosphoinositides to generate phosphatidylinositol-3,4-bisphosphate and phosphatidylinositol-3,4,5-trisphosphate. These phosphoinositides trigger a signaling cascade leading to the phosphorylation and activation of the serine/threonine kinase Akt/PKB, which is a central regulator of glucose uptake and anabolic metabolism (
) proposed the existence of a glucose-fatty acid cycle by which glucose and lipids could serve as competitive substrates for oxidation in muscle. More recent studies suggest that FFAs inhibit at least two independent steps in insulin signaling (
). Specifically, using both in situ and in vivo assays to evaluate the effects of exogenous lipids on insulin signaling, researchers have shown that exogenous fats can activate pathways leading to the inhibition of IRS-1 or Akt/PKB. Interestingly, infusing a lipid mixture enriched in unsaturated fatty acids has been shown to block insulin signaling to IRS-1 and PI3-kinase, but to not affect Akt2/PKBβ (
). By contrast, findings in cultured myotubes indicate that long-chain saturated fatty acids inhibit insulin stimulation of Akt/PKB, including the Akt2/PKBβ isoform, but not upstream signaling events (
). As shown herein, overexpressing AC prevented the accrual of ceramide, but not DAG or sphingosine, induced by long-chain saturated FFAs. Moreover, cells expressing AC were resistant to the inhibitory effects of saturated FFAs toward Akt/PKB. These findings identify ceramide as a necessary metabolite linking saturated FFAs to the inhibition of insulin signaling.
Reagents—Fetal bovine serum was from Atlas Biologicals (Fort Collins, CO), and silica gel 60 thin layer chromatography (TLC) plates were from Merck. The following additional reagents were obtained from Sigma: palmitate, stearate, arachidate, lignocerate, Dulbecco's modified Eagle's medium (DMEM), fatty acid-free bovine serum albumin (BSA), and N-oleoylethanolamine (NOE). Antibodies utilized included the following: rabbit polyclonal anti-phospho-Akt/PKB (Ser-473), anti-phospho-GSK3β (serine 9), and anti-insulin receptor substrate-1 antibodies from Cell Signaling (Beverly, MA); a rabbit polyclonal anti-Akt/PKB and horseradish peroxidase-conjugated anti-rabbit and anti-mouse antibodies from Santa Cruz Biotechnology (Santa Cruz, CA); and a mouse anti-acid ceramidase monoclonal antibody from BD Biosciences.
Cell Culture—C2C12 myoblasts were maintained at 37 °C in DMEM containing 10% fetal bovine serum. For differentiation into myotubes, the myoblasts were grown to confluence and the medium replaced with DMEM containing 10% horse serum. Myotubes were used for experiments 2–3 days following differentiation.
FFA Treatment—Free fatty acids were administered to cells by conjugating them with FFA-free BSA. Briefly, FFAs were dissolved in ethanol (75 mm) and diluted 1:25 in 1% fetal bovine serum-DMEM containing 2% (w/v) BSA to obtain a FFA concentration of 3 mm. After brief sonication and a 10-min incubation at 55 °C, samples were diluted to the desired concentration in 1% fetal bovine serum-DMEM-2% BSA, cooled at room temperature, filter sterilized, and administrated to myotubes for 14 h. Two hours prior to performing the experiments, myotubes were placed in serum-free DMEM containing 2% BSA in either the presence or absence of FFAs.
Immunoblot Analysis—Cell lysates were resolved by SDS-PAGE, transferred to nitrocellulose, and immunoblotted using methods described previously (
). Detection was performed using the Enhanced Chemiluminescence Plus kit from Amersham Biosciences according to the manufacturer's instructions. For some experiments, 4–20% LongLife precast gels from Gradipore Ltd. (Frenchs Forest, Australia) were used for protein separation.
Ceramide Assay—Total lipids from myotubes were extracted using the method described previously (
) as a template, the forward primer 5′-CTC GAG GCC GCC ACC ATG GAG CAG AAG CTG ATT TCC GAG GAG GAC CTG CCG GGC CGG AGT TGC GTC G-3′ and the reverse primer 5′-GGA TCC TCA CCA ACC TAT ACA AGG GTC-3′. The resulting product included a XhoI restriction site, the start codon of the coding region followed by a c-Myc tag, the 1218-base pair coding region, and a BamHI restriction site. The PCR fragment was ligated into pGEM-T easy vector from Promega (Madison, WI). Nucleotides 1–1218 of Myc-AC were then isolated by digesting pGEM-T-Myc-AC with NotI, and the fragment containing the PCR product was shuttled into the pLNCX1 retroviral vector using NotI, thus obtaining the full-length Myc-tagged construct (pLNCX1-Myc-AC). The constructs were restriction digested to confirm proper orientation of insert and subsequently sequenced to verify that no errors were introduced by PCR. Recombinant retrovirus was generated using methods described previously (
). Briefly, the pLNCX1-AC constructs were co-transfected with plasmids encoding the gag/pol and vsv genes into 293T cells. Supernatant containing the recombinant retroviruses was collected and used to infect C2C12 myoblasts. Because pLNCX1 contains a neomycin (G418) resistance gene, transfected populations were selected and maintained in medium supplemented with G418 (600 μg/ml active concentration).
Sphingosine Assay—Four 10-cm plates containing C2C12 myotubes were incubated in serum-free DMEM containing 2% BSA in either the presence or absence of 0.75 mm palmitate for 12 h. Cells were washed twice in ice-cold phosphate-buffered saline prior to quantifying sphingosine levels using the methods of Edsall and Spiegel (
). Under conditions where palmitate inhibits Akt/PKB, it induces the accumulation of both ceramide and sphingosine (Fig. 1), two lipid metabolites shown previously to inhibit insulin-stimulated glucose transport (
). Moreover, both sphingosine and a short-chain ceramide analog, C2-ceramide, markedly inhibited insulin stimulation of Akt/PKB (Fig. 2). In the studies below we investigated which of these sphingolipids mediate the inhibitory effects of saturated FFAs on insulin signaling.
We previously demonstrated that NOE, an inhibitor of AC, recapitulated and augmented the effects of exogenous palmitate on insulin signaling (
). This finding suggested that ceramide, and not sphingosine, was the likely metabolite linking FFAs to the antagonism of insulin signaling. Reasoning that enhanced AC expression might promote the fast removal of intracellular ceramide, thus protecting cells from the inhibitory effects of FFAs, we used recombinant retrovirus to generate stable C2C12 myotubes overexpressing human AC. We infected C2C12 myotubes with either an empty pLNCX1 retrovirus or one encoding AC. The pLNCX1 virus carries the Neo resistance gene, and selecting the cells in medium containing G418 produced stable lines. Using a monoclonal, anti-human AC antibody we detected significant expression of AC in C2C12s infected with the recombinant virus compared with myotubes infected with the empty retroviral vector (Fig. 3A). We next determined the effect of AC overexpression on ceramide, DAG, and sphingosine levels. To quantify ceramide and DAG, we used the ceramide/DAG kinase assay. Briefly, DAG kinase can phosphorylate both DAG and ceramide to produce phosphatidic acid and ceramide 1-phosphate, respectively, which can be resolved from each other by TLC (
). When the reaction is allowed to proceed in the presence of [32P]ATP, the phosphorylated products can be detected using a Storm phosphorimager. As expected, palmitate stimulated the production of ceramide in cells transfected with an empty vector (Fig. 3B). However, AC overexpression largely attenuated the increase in ceramide accumulation induced by palmitate (Fig. 3B). By contrast, AC overexpression did not have any affect on the accumulation of DAG (Fig. 3C).
To quantify sphingosine levels, we used a previously described assay that relies on sphingosine kinase to label sphingosine present in lipid extracts with a radiolabeled phosphate from [γ-32P]ATP (
). The resulting product, sphingosine 1-phosphate, can then be resolved by thin layer chromatography and detected using a phosphorimager. In vector-only transfectants, palmitate induced a significant 2.5-fold increase in sphingosine levels (Fig. 4). However, in AC-overexpressing cells, palmitate induced a more than 7-fold increase in sphingosine (Fig. 4). These data clearly indicate that the overexpressed AC was active in vivo and effectively metabolized ceramide to sphingosine, thus reducing the accumulation of long-chain ceramides that inhibit insulin signaling.
To determine whether ceramide or sphingosine intermediated the effect of palmitate on insulin signaling, we assessed the ability of FFA to regulate Akt/PKB phosphorylation in both cell lines. In vector cells, palmitate inhibited insulin-stimulated phosphorylation of Akt/PKB and glycogen synthase kinase-3 beta (GSK-3β), one of the physiological targets of Akt/PKB linking it to the regulation of glycogen synthesis (Fig. 5). However, in AC-expressing cells, palmitate was without effect (Fig. 5). As we reported previously (
), palmitate did not inhibit upstream signaling events, including the activation of phosphatidylinositol 3-kinase (Fig. 6). Interestingly, however, AC overexpression did slightly increase insulin-stimulated PI3-kinase, though this effect is unlikely to fully explain the protection from palmitate provided by AC overexpression.
To confirm that the protective effects of AC were the result of its enzymatic activity, we tested whether the aforementioned AC inhibitor (i.e. NOE) retained its ability to block Akt/PKB activation. As predicted, the inclusion of NOE blocked insulin stimulation of Akt/PKB, which is consistent with its observed ability to induce ceramide accumulation (
). NOE retained its ability to block insulin signaling in the AC line (Fig. 7), where it blocks AC enzymatic activity. Moreover, though palmitate was not able to inhibit insulin stimulation of Akt/PKB in the AC lines, it retained its ability to augment the inhibitory effects of NOE. These findings indicate that the enzymatic activity of AC protects cells from the antagonistic effects of palmitate.
We previously found that other long-chain saturated FFAs, including stearate (18:0), arachidate (20:0), and lignocerate (24:0), also inhibited insulin stimulation of Akt/PKB while inducing ceramide synthesis (
). To determine whether the ceramide produced by these FFAs was an obligate intermediate in their antagonistic effects, we tested whether AC overexpression negated the effects of other FFAs on insulin signaling. As shown in Fig. 8, AC overexpression completely blocked the inhibitory effects of these other FFAs on insulin stimulation of Akt/PKB.
The findings above suggest that ceramide, and not sphingosine, was the primary intermediate linking saturated FFAs to the inhibition of insulin signaling. We thus speculated that sphingosine may be inhibiting insulin action by promoting the accumulation of ceramide. Indeed, prior studies reveal that feeding of sphingosine to cells increases intracellular ceramide levels (
). As shown in Fig. 9, exogenous sphingosine increased cellular ceramide levels within 10 min of its addition, and the effects on ceramide accumulation mirrored its inhibitory effects on Akt/PKB phosphorylation.
The connection between obesity and metabolic diseases, such as Type 2 diabetes and Metabolic Syndrome X, is well established with 80–90% of people newly diagnosed with Type 2 diabetes being overweight or obese. A hypothesis gaining credibility is that elevated levels of circulating FFAs contribute to the complications of obesity, particularly insulin resistance, by promoting excessive deposition of fat in tissues not suited for fat storage, such as skeletal muscle (
). Within peripheral tissues, FFAs are quickly converted into acyl coenzyme A that can be converted to different FFA metabolites (i.e. triacylglycerol, DAG, and ceramides), each of which may have potentially deleterious effects on insulin signaling and action (
). The strong correlation between intramyocellular lipid levels and the severity of insulin resistance is consistent with the hypothesis that these inhibitory lipid metabolites are primary mediators of insulin resistance (
Numerous findings suggest a role for sphingolipids in the regulation of insulin signal transduction. When added to cultured cells, both ceramide and sphingosine have been shown to inhibit insulin signaling or action (
). Furthermore, agents that interfere with sphingolipid pathways have been recently demonstrated to modulate the inhibitory effects of saturated FFAs on insulin signaling. In C2C12 myotubes, for example, inhibitors of serine-palmitoyltransferase or ceramide synthase, two enzymes required for de novo sphingolipid synthesis, prevented the antagonistic effects of palmitate on both ceramide accumulation and insulin signaling (
). These observations support the hypothesis that increases in ceramide levels are a potentially important mechanism underlying the development of insulin resistance induced by saturated fats.
Because ceramide accumulation has also been considered a hallmark of apoptosis, cell cycle arrest, and growth suppression, overexpression of ceramidases has been utilized as a tool for determining the role of ceramide in these biological events. For example, overexpression of acid ceramidase protects against apoptosis induced by tumor necrosis factor-α (
). Herein, we aimed to determine whether attenuating intracellular ceramide accumulation counteracts the inhibitory effects of saturated FFAs on insulin signaling. As shown above, AC overexpression negated the antagonistic effects of palmitate and other FFAs on Akt/PKB and GSK3β, two important regulators of glucose storage (Fig. 5). Moreover, overexpressed AC largely prevented the accumulation of ceramide induced by palmitate (Fig. 3B). By contrast, AC overexpression did not affect the induction of DAG (Fig. 3C), and it increased the accrual of sphingosine (Fig. 4). These observations strongly support the hypothesis that ceramide, but not DAG or sphingosine, is indispensable for the inhibition of insulin signaling caused by the long-chain-saturated FFA palmitate.
Interestingly, Northern blot analysis of various murine tissues showed undetectable AC mRNA levels in skeletal muscle compared with other tissues (
). Possibly, because of the lack of AC activity in skeletal muscle, this tissue may be particularly prone to accumulate ceramide when exposed to excess FFAs. An answer for this conjecture may be found using in vivo models of Farber disease, an autosomal recessive disorder caused by lysosomal AC deficiency. Patients suffering from this rare disease accumulate ceramide in the lysosomal compartments of cells in nearly all tissues. However, because of their shortened life spans, there is no evidence as to whether or not patients develop insulin resistance. Recent studies demonstrated that AC–/+ knock-out mice manifest a progressive lipid storage disease associated with elevated ceramide levels and a reduction of AC activity (
). However, no reports with regard to insulin resistance have been provided from phenotypic studies of these mice. Measurements of insulin-stimulated glucose disposal in AC–/+ knock-out mice would be necessary to establish a central role of AC in the development of insulin resistance in vivo.
Conclusion—Ceramide is a bioactive sphingolipid derived from long-chain saturated fats (
). Herein, we evaluated the functional significance of saturated FFA-induced ceramide accumulation in the development of saturated fat-induced muscle insulin resistance by constitutively overexpressing AC in C2C12 myotubes. Overexpression of AC protected cells from the inhibitory effects of palmitate and other long-chain saturated FFAs on insulin signaling. Moreover, as shown in palmitate-treated cells, AC overexpression shifted the equilibrium to lower levels of ceramide while increasing levels of sphingosine. These results confirm the critical role of AC in the regulation of ceramide accumulation and its deleterious effects on insulin signaling. Moreover, this study suggests that preventing the aberrant muscle accumulation of ceramide by promoting its metabolism into sphingosine and sphingosine-derivatives might restore normal insulin sensitivity and glucose metabolism in animal models of insulin resistance.