Bioactives of Artemisia dracunculus L. Mitigate the Role of Ceramides in Attenuating Insulin Signaling in Rat Skeletal Muscle Cells

PMI 5011 was produced from plants grown hydroponically under uniform and controlled conditions, thereby standardizing phytochemical content. The growing, quality control, biochemical characterization, and preparation of PMI 5011 have been reported (11–14). For this study, PMI 5011 was evaluated at 10 μg/mL, a dose determined to be the lowest most effective level from previous studies (12–14).

Ceramide standards and reagents were purchased from Avanti Polar Lipids (Alabaster, AL). All organic solvents were of high-performance liquid chromatography grade, American Chemical Society certified. Myriocin and free fatty acids (FFAs) were obtained from Sigma-Aldrich. The plasmid pMko.1 puro PTEN short hairpin RNA was obtained from Addgene.

L6 myoblasts were obtained from the American Type Culture Collection and maintained at 37°C, 95% air, and 5% CO₂ in low glucose Dulbecco’s modified Eagle’s medium supplemented with 10% CBS serum and antibiotics. For individual experiments, myoblasts were subcultured onto 6 or 12 well plates, grown to 80–90% confluence, and differentiated into fused myotubes for 5 days by switching to media with 2% horse serum. All cells used were within five passages.

Cultures were exposed to FFAs conjugated with 1% BSA and constituted to a final concentration 200 μM or ceramide C2 in DMSO at 20 or 40 ng/mL. For insulin signaling studies, myotubes were serum-starved in Dulbecco’s modified Eagle’s media containing 0.2% BSA and treated with FFAs or ceramide C2 for 16 h. Insulin was added at a final concentration of 100 nmol/L for 5 min before subsequent analysis.

Glycogen content in basal and insulin stimulated states was assessed on cells grown in 12 well plates with use of glycogen hydrolysis and glucose determination according to the method of Gómez-Lechón et al. (15) with modifications shown in Wang et al. (13). Results were normalized by protein concentration measured by Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, CA), and glycogen content was presented as micrograms glucose equivalent per well.

Cells grown in 6 well plates were collected with 200 μL ice-cold deionized water after 24 h incubation with FFAs with or without PMI 5011. After sonication, 2 μL was removed for protein concentration measurement. An aliquot equivalent to 300 μg protein was extracted according to the method of Haynes et al. (16) with modifications. In brief, 4 μL of a 10 μg/mL ceramide C17 solution were added as the extraction standard. Lipids were extracted with 500 μL of methanol/chloroform (1:2). The suspension was vortexed and placed on a shaker at 4°C for 1 h. Lysates were spun (10 min, 10,000 rpm). The lower chloroformic layer was transferred to a new tube and placed on ice. The upper methanolic and the middle proteinous layers were sonicated and subjected to a second extraction. The lower chloroformic layer was added to the first one and stored at −80°C until analysis. Before analysis the samples were dried under nitrogen gas at room temperature and reconstituted with acetonitrile.

Liquid chromatography-electrospray ionization tandem-mass spectrometry was used to measure intracellular levels of ceramides C16, C18, C18:1, C20, C22, C24, and C24:1. Ceramide C2 was included as the internal standard. Standards were prepared by dissolving ceramides in a 99.8/0.2 (volume for volume) mixture of ethanol/formic acid to a concentration of 5 mg/mL. Further dilutions were made using acetonitrile to create nine working solutions of 200, 100, 50, 25, 12.5, 6.25, 3.125, 1.56, 0.78, and 0 ng/mL. Standard curves were constructed using linear regression of the peak areas of the analyte versus the corresponding concentrations.

LC was performed using a Waters Acquity ultraperformance liquid chromatography (UPLC) with a binary pump and Accquity autosampler set at 10°C. Extracted lipids (10 μL) were injected on a Pursuit 3 Diphenyl reversed-phased column (50 × 2.0 mm inner diameter, 3 micron particle column; Varian, Walnut Creek, CA), and flow rate of 0.3 mL/min was applied. The column was at room temperature. The mobile phases were as follows: phase A: 0.1% formic acid and 25 mmol/L ammonium acetate in water; and phase B: 100% acetonitrile. Gradient was 0.10 min B = 30%, 1.10 min B = 95%, 3.00 min B = 95%, and 5.00 min B = 30%. Total run time was 5.00 min. The detector was a Waters Aquity TQD triple quadruple MS/MS with ion source ESI operated in the positive mode. Source and desolvation temperatures were 110°C and 300°C, respectively. The desolvation gas (nitrogen) flow was 600 L/h. No cone gas was used. Collision gas (Argon) flow rate was 0.15 mL/min, and the capillary voltage was 2.47 kV. All ceramides gave the protonated molecular ion [M+H]⁺ and the same ion after losing a water molecule [M+H-H₂O]⁺as the major ions and the common product ion 264.2 because the loss of fatty acid and two hydroxyl groups (Supplementary Table 1). According to the retention times of standards, the individual long-chain FFAs were quantified and normalized by protein concentration.

Akt-1, Akt-2, and protein tyrosine phosphatase 1B (PTP1B) protein content and phosphorylation were determined by Western blotting as described previously (13,14). Specific antibodies used were anti–Akt-1 and -2, Akt-1 phos (Ser473), Akt-2 phos (Ser474) (Millipore, Temecula, CA), anti-PTP1B and anti-PTP1B phos (Ser50) (ECM Biosciences, Versailles, KY), and β-actin (Santa Cruz Biotechnology, Santa Cruz, CA). In all cases, Western blot results were verified in three separate experiments. To assess the effect of PTP1B expression levels on AKT-1 and -2 phosphorylation, interference RNA (RNAi) was used to downregulate PTP1B expression in FFA and C2-treated cells. Cells in the process of differentiation after 1 day (90–95% confluence) were transfected with 2.5 μg DNA and Lipofectamine 2000 at a ratio of 1:2.5 in 6-well plates in serum-free media. After 6 h, media were switched to that with 2% horse serum for 4 days before treating the cells for 16 h with FFA, ceramide C2 with or without PMI 5011 in serum-free media.

For gene expression, the same treatments above were performed; however, cells were treated for 24 h with 2% horse serum. Total RNA was purified using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. RNA total content and quality was determined spectrophotometrically by absorbance measurements at 260 and 280 nm using the NanoDrop system (NanoDrop Technologies). The cDNAs were synthesized with 2.5 μg of DNA for each sample by reverse transcription (Qiagen) following manufacturer’s protocol. Synthesized cDNAs were diluted fourfold. Five microliters of each diluted sample were used for PCR reactions of 20 μL final volume. Other components making up this final volume for PCR reactions were 5 μL of 6 μM gene-specific commercial primers (synthesized by IDT, Coralville, IA) and 10 μL SYBR Green PCR master mix (Qiagen). PCR amplification was carried out by denaturing at 95°C for 15 s, annealing and extension at 60°C for 1 min for 40 cycles. The RT-PCR was analyzed by the Sequence Detection System SDS (Rotor Gene 6000). The amounts of the PTP1B, AKT1, and AKT2 mRNA were normalized to the amount of β-actin mRNA, as an internal control.

PTP1B activity was assayed as described previously (14). In brief, the colorimetric p-nitrophenyl phosphate (PNPP) hydrolysis assay based on the ability of phosphatases to catalyze the hydrolysis of p-nitrophenyl phosphate to p-nitrophenol, a chromogenic product, was used. The intensity of the color reaction was measured at 410 nm on a Bio-Rad microplate spectrophotometer. Results were normalized by protein concentration. Experiments were also conducted with and without the presence of a selective inhibitor of PTP1B [3-(3,5-dibromo-4-hydroxy-benzoyl)-2-ethyl-benzofuran-6-sulfonicacid-(4-(thiazol-2-ylsulfamyl)-phenyl)-amide; Calbiochem, Gibbstown, NJ] used at a concentration of 20 μM.

All data are presented as means ± SE. Difference between means was determined by ANOVA with subsequent Fisher’s protected least significant difference tests or Student’s t test. P < 0.05 was considered significant. All results are an average of three experiments.

Insulin stimulation resulted in a threefold increase in glycogen content in L6 myotubes when compared with basal conditions (Fig. 1). Incubation with the saturated FFA (i.e., palmitic acid) or ceramide C2 resulted in significantly reduced glycogen content compared with control. Specifically, palmitic acid resulted in a 32% decrease in glycogen content, whereas exposure to C2 resulted in 38.2% and 40.5% decrease at concentrations of 20 ng/mL and 40 ng/mL, respectively (Fig. 1). Incubation with PMI 5011 in the presence of the FFA or C2 resulted in a significant increased glycogen content when compared with incubation of FFA or C2 alone (P < 0.05; Fig. 1).

Double extraction with methanol/chloroform (1:2) yielded 90.2% recovery for C17, the extraction standard giving assurance of the efficiency of the extraction process. Efficacy of PMI 5011 to reduce ceramide content was compared with that of the fungal toxin myriocin (1 μg/mL), a known inhibitor of serine palmitoyltransferase (SPT), the enzyme that catalyzes the first and rate determining step in the de novo ceramide synthesis process (Fig. 2). Treatment with palmitic acid resulted in a more than twofold increase in all ceramide species. PMI 5011 did not alter total ceramide content when incubated with palmitic acid compared with palmitic acid alone (P = ns). However, myriocin incubation with palmitic acid resulted in ceramide levels that were not different from the control. Ceramide C16 was the most prevalent species formed followed by C24:1. All ceramide species remained elevated in the presence of PMI 5011 and palmitic acid, suggesting that PMI 5011 had no effect of ceramide formation or transformation into metabolites.

L6 cells incubated with saturated fatty acids (i.e., palmitic [16:0] and stearic [18:0]) resulted in a significant increase in total ceramide content as compared with a nonsignificant increase in ceramide content observed when cells were exposed to the unsaturated fatty acids (palmitoleic [16:1] and oleic acid [C18:1]) (Figs. 3 and 4). Stearic acid (18:0) resulted in a significant increase in ceramide C18 species approximately sevenfold compared with control conditions. Although both stearic (18:0) and oleic (C18:1) acid resulted in a threefold increase in the C18:1 ceramide species, the total level of C18:1 was very low (Fig. 3).

The saturated fatty acids palmitic (C16:0) and stearic (C18:0) resulted in significantly reduced glycogen content after insulin stimulation (Fig. 4A) and increased total ceramide content (Fig. 4B). By contrast, the monosaturated fatty acids, i.e., palmitoleic acid (C16:1) and oleic acid (C18:1), had minimal effect on insulin-stimulated glycogen content and ceramide accumulation (Fig. 4A and B).

Incubation of muscle cells with ceramide C2 or palmitic acid significantly blunted insulin-stimulated phosphorylation of both Akt-1 and Akt-2 compared with control conditions (see lanes 4 and 8 vs. lane 2, Fig. 5), and significantly increased PTP1B levels (see lanes 3 and 4 and 7 and 8 vs. lanes 1 and 2). PMI 5011 restored insulin stimulated Akt-1 and -2 phosphorylation in the presence of ceramide or FFA as compared with the ceramide or FFA alone (see lanes 6 and 10 vs. lane 4 and 8, respectively). PMI 5011 reduced PTP1B protein abundance in the presence of ceramide or FFA as compared with the ceramide or FFA alone (see lanes 5 and 6 and lanes 9 and 10 vs. lanes 3 and 4 and 7 and 8, Fig. 5). Besides reducing PTP1B protein abundance, 5011 also resulted in increased phosphorylation of PTP1B at Ser50 when incubated with C2 or FFA (lanes 5 and 6 and lanes 9 and 10, respectively) when compared with incubation of myotubes with C2 or FFA alone (lanes 3 and 4 and lanes 7 and 8, respectively).

Incubation of myotubes with FFAs and ceramide C2 resulted in a significant increase in PTP1B activity compared with that of the control cells (Fig. 6). PMI 5011 significantly reduced PTP1B activity in both FFA-treated and C2-treated cells to levels comparable with those by the PTP1B inhibitor (Fig. 6).

To assess effect of PTP1B expression levels on AKT phosphorylation, we used the use RNAi to downregulate PTP1B expression in FFA-, C2-, and 5011-treated cells. Results showed that when compared with L6 myotubes incubated with FFA or ceramide C2 alone, RNAi lowered PTP1B levels in L6 myotubes incubated with FFA or C2 (see lanes 9 and 10 vs. lanes 5 and 6, Fig. 7A and B, respectively), and this resulted in increases in insulin stimulated AKT1 and AKT2 phosphorylation (see lane 10 vs. lane 6, Fig. 7A and B). Addition of PMI 5011 to cells treated with FFA or C2 appeared to have greater effects on Akt-1 and Akt-2 phosphorylation as compared with transfected cells treated with FFA or C2 (see lanes 8 vs. 10, Fig. 7A and B). It is noteworthy that PMI 5011 was noted again to have effects to enhance PTP1B phosphorylation when either treated with FFA or C2 alone or in transfected cells treated with FFA or C2 (see lanes 8 and 12 vs. lanes 6 and 10, Fig. 7A and B, respectively).

Gene expression revealed that both FFA and ceramide C2 significantly increased PTP1B mRNA expression (Fig. 8A and B). Use of RNAi or PMI 5011 separately as intervention reduced PTP1B expression to levels not different from those of control. AKT1 and AKT2 expression remained unchanged after intervention (Fig. 8A and B).

Our data demonstrate that pretreatment of myotubes with saturated FFAs mirrors the effect of clinical metabolic dysregulation because this intervention attenuated insulin signaling and resulted in blunted insulin action. Specifically, we noted a reduction in both Akt-1 and Akt-2 phosphorylation and reduced insulin-stimulated glycogen content with exposure to saturated FFAs. In agreement with other reports (2,3), we demonstrated a consistent increase in lipid metabolites, i.e., ceramide content, with FFA incubation and that ceramide C2, a cell-permeable analog of the long-chain ceramides, closely replicated the effects of the saturated FFAs. We report that the mechanism by which PMI 5011 improves insulin action was not secondary to an effect on the formation or accumulation of ceramides. However, novel information was obtained suggesting that the botanical reverses the effect of intramuscular-formed ceramides on Akt phosphorylation and hence modulated their role in insulin signaling in L6 cells.

Ceramides with different compositions are produced by de novo synthesis via condensation of palmitoyl-CoA and Ser (4) or activation of sphingomyelinases under physiological stress (7). In the presence of excess FFAs, long-chain fatty acyl-CoAs derived from the FFAs are diverted away from carnitine palmitoyltransferase 1, the mitochondrial enzyme that catalyzes the first step in β-oxidation, and are instead preferentially partitioned toward the synthesis of signaling intermediates diacylglycerol, triglycerides, and ceramides. We confirmed this pathway by preincubation of cells with different FFAs and observed distinct effects on insulin sensitivity and accumulation of both total ceramides generated and specific ceramide species. The rate-limiting reaction in ceramide synthesis is the SPT catalyzed condensation of Ser with palmitoyl-CoA. Myriocin, an inhibitor of SPT, inhibited accumulation of ceramides demonstrating the involvement of the de novo pathway. Palmitate provides the precursor palmitoyl-CoA and/or induces SPT expression (2,4). Unsaturated palmitoleic (16:1) and oleate (18:1) did not result in a significant increase in total ceramides and had no effect on the ceramide profile because SPT is highly specific to saturated FFAs with 16 ± 1 carbon atoms (4). The incorporation of the second fatty acyl-CoA during the conversion of sphinganine into dihydroceramide uses both unsaturated and saturated FFAs. Oleate caused an increase in the corresponding ceramide C18:1 almost fivefold, confirming its involvement in this step. However, total levels of C18:1 were too low compared with those of C16, C18, and C24:1. Our observations demonstrated that the saturated FFAs palmitic (16:0) and stearic (18:0) reduced glycogen content, and this was associated with increased ceramide levels. In contrast, palmitoleic (16:1) and oleic (18:1), which caused minimal ceramide increase, resulted in no net effect on glycogen content, confirming the importance of ceramides as a metabolite linking lipid oversupply to the antagonism of insulin signaling. The observation is consistent with dietary studies indicating that saturated fats markedly decrease insulin responsiveness in peripheral tissues, whereas unsaturated fats may have weaker or in some cases insulin-sensitizing effects (2). Our findings correlate with those of Ghosh et al. (5) who showed that the proinflammatory cytokine tumor necrosis factor-α (TNF-α) activates both de novo and hydrolysis pathways of ceramide generation. In their study, L6 cells assayed for ceramide content also showed that C16 and C24:1 ceramides were the predominant species detected after TNF-α treatment.

Clinical interventions, i.e., thiazolidinediones and exercise, which reduce cellular lipids, have been reported to increase insulin action (4,8,9,17). Stearoyl-CoA desaturase, a microsomal enzyme, catalyzes the synthesis of monounsaturated fatty acids from saturated fatty acyl-CoAs and plays a key role in the synthesis of triglycerides, cholesteryl esters, and membrane phospholipids. Dobrzyn et al. (18) reported that depletion of stearoyl-CoA desaturase lowers ceramide levels and improves insulin sensitivity. Because PMI 5011 has been shown to significantly increase glucose uptake and partially restore glycogen accumulation (13,14), we hypothesized that the insulin-sensitizing effects of PMI 5011 may be secondary to reducing the accumulation of ceramides observed in the presence of increasing FFA concentrations. However, all seven ceramide species remained elevated in the presence of PMI 5011. We therefore concluded that the mechanism by which PMI 5011 restores insulin signaling is not through preventing accumulation of ceramides. This study did not, however, rule out the fact that 5011 may have some effects on downstream metabolism of the ceramides into forms that are less deleterious to insulin signaling, while still maintaining high overall ceramide levels. Specifically, ceramides can be metabolized to less toxic forms by the addition of a phosphorylcholine group to form sphingomyelin. They can also be deacylated (by ceramidases), phosphorylated (by ceramide kinase), or glycosylated (by glucosylceramide synthase) resulting in molecules that may have a less deleterious effect on insulin signaling (4,17). Overexpression of acid ceramidase has been shown to prevent palmitate-induced insulin resistance in C2C12 muscle cells (19). Thus the study did not rule out whether PMI 5011 promotes any of these metabolism processes.

Consistent with the reports of Chavez and Summers (2), Schmitz-Peiffer et al. (19), and Bourbon et al. (20), ceramide C2 inhibited insulin-stimulated phosphorylation of PKB/Akt. Ghosh et al. (5) also showed that ceramide C6 mimicked the effects of TNF-α in attenuating insulin signaling by reducing AKT phosphorylation at Ser473. However, our study showed that PMI 5011 significantly enhanced insulin sensitivity by restoring PKB/Akt phosphorylation even in the presence of elevated FFA or ceramide levels. Akt activation is dependent on its phosphorylation at two key residues: Thr308 and Ser473/474. In particular, phosphorylation at Ser473 and Ser474 is necessary for the full activation of Akt kinase activity. Reduced insulin-stimulated Akt is evident in insulin-resistant obese individuals. Akt/PKB promotes the uptake of lipids and amino acids in various tissues and regulates a variety of intracellular enzymes that promote the incorporation of nutrients into glycogen, protein, and triacylglycerol including glycogen synthase kinase 3b (GSK3b), an upstream inhibitor of glycogen synthesis. Thus the reduction of glycogen content observed with pretreatment with the saturated fatty acids and ceramide C2 may stem from the inhibition of Akt/PKB phosphorylation and hence attenuation of its role in the insulin signaling pathway.

Besides inhibiting Akt phosphorylation, ceramide C2 increased the expression and activity of PTP1B, an abundant PTPase within skeletal muscle and a therapeutic target for type 2 diabetes. Increased PTP1B protein content was consistent with elevated PTP1B phosphatase activities (Fig. 6). Increased activity of phosphatases can be one of the mechanisms inhibiting the insulin signaling pathway (21). PTP1B dephosphorylates the insulin receptor and insulin receptor substrate. A study by Ravichandran et al. (22) showed that PTP1B can function as a substrate for Akt in cells. They demonstrated that phosphorylation of PTP1B by Akt at Ser50 negatively modulates its phosphatase activity, creating a positive feedback mechanism for insulin signaling by impairing its ability to dephosphorylate the insulin receptor. Testing for PTP1B phosphorylation showed that it increased when AKT phosphorylation increased with 5011 (Fig. 5). When Akt is inactivated by reduced phosphorylation by ceramides in this case, its negative modulation of PTP1B is reduced. It is noteworthy that PMI 5011 treatment consistently resulted in much higher levels of PTP1B phosphorylation. Thus, in this study, PMI 5011 was demonstrated to have three specific and possibly interrelated effects on the insulin signaling pathway. First, it restored Akt phosphorylation. Second, it increased phosphorylation of PTP1B, hence negatively modulating PTP1B activity in dephosphorylating the insulin receptor. It was not concluded as to whether this was a direct effect or secondary to the enhanced Akt phosphorylation. Finally, PMI 5011 decreased PTP1B protein expression in FFA and in ceramide C2–pretreated cells. A decrease in PTP1B or its phosphorylation improves insulin action by reducing its capacity to dephosphorylate the insulin receptor thus improving insulin resistance (22). Overall, the net sum of the effects of PMI 5011 was to lower PTP1B activity to a level comparable to that observed of the specific PTP1B inhibitor in both palmitate and C2 pretreated cells.

Our data demonstrate that the mechanism by which PMI 5011 modulates insulin action in metabolically dysregulated states is not secondary to an effect on ceramide formation and accumulation. With PMI 5011 intervention, we observed increased insulin signaling via restoration of insulin-stimulated Akt-1 and Akt-2 phosphorylation despite presence of high ceramide levels in cells. PMI 5011 mitigated the upregulation of PTP1B that is observed in the presence of excess FFA or ceramide accumulation. It is noteworthy that PMI 5011 was noted to consistently enhance PTP1B phosphorylation and, as such, reduce its phosphatase activity against intracellular substrates of insulin signaling, thus providing a novel observation regarding its mechanism of action.

This work was supported by P50AT002776-01 from the National Center for Complementary and Alternative Medicine and the Office of Dietary Supplements, which funds the Botanical Research Center of Pennington Biomedical Research Center and the Biotech Center of Rutgers University.

No potential conflicts of interest relevant to this article were reported.

D.N.O. and W.T.C. designed the studies, reviewed and analyzed the data, and wrote the manuscript. D.N.O., A.H., Y.Y., X.H.Z., and Z.Q.W. conducted research studies and reviewed and edited the manuscript. D.R. prepared the plant material from growth, extraction, and purification of the extract. W.T.C. 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.

Parts of this study were presented in abstract form at the 71st Scientific Sessions of the American Diabetes Association, San Diego, California, 24–28 June 2011.