Ceramide Content Is Increased in Skeletal Muscle From Obese Insulin-Resistant Humans

Subjects were placed into one of two study groups based on BMI. Individuals with a BMI <27 kg/m² were categorized as lean control subjects (n = 10). This group did not have a family history (first-degree relative) of type 2 diabetes. Subjects with a BMI of 27–33 kg/m² were categorized as obese nondiabetic (n = 10). Two of the obese subjects had a single first-degree relative with type 2 diabetes. All subjects had a normal 75-g oral glucose tolerance test and were judged to be in good health based on physical examination, medical history, routine blood work, urinalysis, and electrocardiogram. All subjects were free of medications and were not participating in any form of regular exercise. Subjects were instructed to maintain their normal diet and abstain from exercise for 3 days before the study. The purpose, nature, and potential risks of the study were explained to all subjects, and written consent was obtained before participation. All studies were reviewed and approved by the Institutional Review Boards at the University of Texas Health Science Center as San Antonio and Georgia Tech University. All studies were performed at the Audie L. Murphy Memorial Veterans Hospital, General Clinical Research Center, San Antonio, Texas, in the morning after a 10- to 12-h overnight fast.

A hyperinsulinemic-euglycemic clamp with biopsies of the vastus lateralis muscle was performed as described (25). An antecubital vein was cannulated for the infusion of insulin and 20% glucose. A hand vein was cannulated in a retrograde fashion, and the hand was placed in a heated box (55°C) for sampling of arterialized blood. Blood was drawn every 10 min for the last 30 min of the rest period for the determination of basal plasma insulin and plasma glucose concentrations. After 60 min bed rest, a percutaneous muscle biopsy was obtained with a Bergstrom cannula from the vastus lateralis muscle under local anesthesia. Muscle specimens (75–200 mg) were immediately blotted free of blood, frozen in liquid nitrogen, and stored under liquid nitrogen until processing. Subjects rested for 60 min after the basal biopsy. At time t = 0, a constant infusion of insulin [40 mU/(m² · min)] was started. The plasma glucose concentration was measured every 5 min by a glucose analyzer (Beckman Instruments, Fullerton, CA), and an infusion of 20% glucose was adjusted to maintain euglycemia. Blood was obtained every 10 min during the insulin clamp for measurement of plasma insulin concentrations. A second percutaneous muscle biopsy was obtained from the other leg at 30, 45, or 60 min after insulin infusion began. Insulin infusion was terminated after the last biopsy. All subjects had at least one muscle biopsy basally. Some of the subjects received multiple insulin infusions and a second biopsy; if so, the infusions were conducted on separate days. The rate of infusion of glucose required to maintain euglycemia was taken as a measure of insulin-stimulated glucose uptake. This measure underestimates the true rate of glucose uptake to the extent that endogenous glucose production is incompletely suppressed by insulin.

Frozen muscle specimens were homogenized as described (26,27), and protein concentration was determined by the Lowry method (28). Akt activation was assessed using immunoblot analysis of 100 μg total muscle protein by determining the phosphorylation of Ser⁴⁷³ using a specific antiphospho-Akt antibody. All blots were stripped and reprobed for protein expression, and protein phosphorylation was expressed relative to protein levels.

Muscle sphingolipid content was determined as described (29). Tissues were homogenized in the smallest volume possible of PBS such that the amount of tissue per unit volume is the same for all samples. Approximately 2 × 2 mg of protein per sample is transferred to a 13 × 100-mm screw-cap glass test tube for sphingolipid extraction via established methods (25).

A 20-μl aliquot of the reconstituted sample was added to a final volume of 1 ml of a methanol:acetic acid (99:1) solution containing 5 mmol/l ammonium acetate. This solution was infused into the ion source at ∼5–10 μl/min, and individual sphingolipids were identified via precursor ion scans (25). Optimal ionization and collision conditions for individual molecular species in each class of sphingolipids were determined.

Crude sphingolipid extracts were loaded onto a high-performance liquid chromatography instrument directly coupled with the ion source of a tandem mass spectrometer (MS/MS). Free sphingoid bases and complex sphingolipids were separated by either reverse- or normal-phase chromatography, respectively (25). The former used shorter columns (2.1 × 50 mm) and higher flow rates (1 ml/min), whereas the latter used an amino column that was both shorter (2.1 × 50 mm) and had higher flow rates (1.5 ml/min.). Quantitation was determined by comparing the peak areas generated by the internal standards with those of the endogenous species.

Plasma insulin concentrations were determined by radioimmunoassay (Diagnostics Products, Los Angeles, CA). Body composition was determined by bioimpedance (RJL Bio-106 Spectrum Body Composition Analyzer; RJL Systems, Detroit, MI).

Anti-phospho-Akt antibody was purchased from Upstate Biotechnology (Waltham, MA). Polyclonal anti-Akt was purchased from Cell Signaling Technology (Beverly, MA). Anti-rabbit horseradish peroxidase conjugate secondary antibody and the enhanced chemiluminescence reagent system were purchased from Amersham Pharmacia Biotech (Piscataway, NJ). All other chemicals and reagents were purchased from Sigma-Aldrich (St. Louis, MO) or Bio-Rad (Hercules, CA).

A two-way ANOVA was used to determine differences in Akt phosphorylation and sphingolipid content between the two groups, with a posteriori comparisons (StatView; SAS Institute, Cary, NC). Correlation analysis was performed using Pearson’s product moment correlation.

The groups were well matched with respect to age, sex, and ethnicity. BMI was used to place individuals into either the lean control or obese nondiabetic group. Body weight, BMI, percent lean body mass, fasting plasma glucose, and fasting plasma insulin levels were significantly elevated in the obese nondiabetic group when compared with the lean control subjects (Table 1). All subjects had normal glucose tolerance. Fasting plasma insulin concentrations were significantly greater in the obese subjects (Table 1), as were area under the curve for insulin concentrations during the oral glucose tolerance test (P < 0.001).

Plasma insulin and FFA concentrations during the euglycemic clamp are given in Table 2. Plasma insulin concentrations were significantly greater in the obese subjects basally and at every time point during the insulin infusion. Likewise, fasting plasma FFA concentrations were increased in the obese subjects. Moreover, plasma FFA concentrations were less inhibited by insulin infusion in the obese group. The glucose infusion rates required to maintain euglycemia during the insulin infusions are shown in Fig. 1. Because the insulin concentrations achieved during the insulin infusions were greater in the obese subjects (presumably due to decreased insulin clearance because the insulin infusion rates were identical in the two groups), glucose infusion rates were also expressed relative to the insulin concentration. The glucose infusion rate, expressed per kilogram of fat-free mass, was lower in obese nondiabetic subjects at all time points (P < 0.01).

Sphingolipid content of skeletal muscle was determined using high-performance liquid chromatography electrospray ionization tandem mass spectrometry. The results of the analysis for total ceramide content are shown in Fig. 2. The results at 45 and 60 min did not differ in either group; therefore, data from these time points were pooled for statistical comparisons. The results for the individual ceramide species are also given in Table 3. Under basal conditions, total muscle ceramides were increased by nearly twofold in the obese subjects (P < 0.05), and ANOVA showed the obese subjects to have greater ceramide content overall (P < 0.05). Although total ceramides decreased somewhat in the obese subjects during the insulin infusion, this effect was not statistically significant. Of the total ceramides present, C24, C24:1, and C18 ceramides were the most abundant. A small amount of C16 dihydroceramide was also detected and did not differ between the two groups. No individual species was overrepresented in the obese subjects; rather, the increase was uniform. Insulin infusion did not affect the composition of ceramides (Table 3). In addition, neither obesity nor insulin infusion affected the content of other sphingolipids assayed (Table 4).

Correlation analysis was performed to determine the relationship between total ceramide content and the glucose infusion rate required to maintain euglycemia during the insulin infusions as well as basal plasma FFA concentrations. When data from all subjects were pooled, there was no significant correlation of ceramide with either variable. However, exclusion of one outlying subject in the obese group revealed a significant correlation between ceramide and the fasting plasma FFA concentration (r = 0.51, P < 0.05, Fig. 3).

The effect of insulin infusion on Akt Ser⁴⁷³ phosphorylation is given in Table 5 and illustrated in Fig. 4. Muscle biopsies of the vastus lateralis were performed basally and after 30, 45, or 60 min of insulin infusion, and Akt Ser⁴⁷³ phosphorylation and Akt protein content were quantified by immunoblot analysis. Figure 4A shows a representative comparison of a lean and obese subject who received biopsies basally and after 60 min of insulin infusion. Average data for the blots are shown in Fig. 4B. Because the plasma insulin concentrations during the insulin infusions were ∼40% higher in the obese subjects, Akt phosphorylation was expressed relative to the protein content and plasma insulin concentration (Fig. 4B). Akt Ser⁴⁷³ phosphorylation was significantly reduced in the obese subjects after either 30 or 45–60 min of insulin infusion (Fig. 4B). Because each study had a basal time point, values from all basal biopsies were pooled. As for ceramides, data from the 45- and 60-min time points did not differ from each other and so they were pooled for analysis. Table 5 shows Akt Ser⁴⁷³ phosphorylation data expressed relative to Akt protein content, without regard to insulin concentration during the clamp. Results of these analyses showed that, under basal conditions, Akt Ser⁴⁷³ phosphorylation was slightly, but not significantly, decreased in the obese subjects. After 30 min of insulin infusion, however, Akt phosphorylation was stimulated in both groups of subjects (Table 5). In lean insulin-sensitive subjects, Akt phosphorylation remained elevated for 45–60 min of insulin infusion. In contrast, Akt phosphorylation after 45–60 min of insulin infusion in the obese subjects decreased significantly (P < 0.05) compared with the 30-min value for that group and was also significantly decreased compared with the lean control group (Table 5, P < 0.05). There were no differences in Akt protein expression between the two groups (Table 5). Correlation analysis between total ceramide content under basal conditions and insulin stimulation of Akt Ser⁴⁷³ phosphorylation revealed a modest negative correlation (r = −0.45, P = 0.09).

Insulin resistance characterizes skeletal muscle in obesity and type 2 diabetes. Increasing evidence supports the notion that in insulin-resistant muscle, abnormalities in lipid use lead to increased myocellular lipid content (30). The increase in lipid content is believed to result in decreases in insulin receptor signaling and insulin resistance (13,31). Although the increase in lipid content is manifested as an increase in triacylglycerol (2–6), it is likely that it is not triacylglycerol itself but some other less abundant lipid species that are the proximal cause of insulin resistance. One candidate for such a lipid species is ceramide. Ceramides are a family of sphingolipids that differ in the fatty acyl moiety and are known to induce insulin resistance and apoptosis in cultured cells (7,17–20). As an intermediate in the sphingomyelin pathway, ceramide can be generated by hydrolysis of sphingomyelin or can be produced by de novo synthesis (32). In cultured C2C12 cells, a cell-permeable ceramide analog decreased insulin-stimulated glucose uptake, glycogen synthesis, and Akt serine phosphorylation, but not insulin receptor and IRS-1 tyrosine phosphorylation and IRS-1-associated PI 3-kinase activity (7). Because these abnormalities mirrored those produced by incubation of the cells with palmitate, de novo synthesis of ceramide was invoked as the mechanism of palmitate-induced insulin resistance (7). Recent studies have confirmed the role of ceramide in inhibition of Akt and downstream signaling, but not IRS-1 function, in 3T3-L1 adipocytes (33) or L6 muscle cells (34). However, the results from another study indicate that ceramides may also inhibit insulin receptor signaling upstream of Akt, at the level of IRS-1 (19). Moreover, infusion of a triglyceride emulsion into rodents inhibited IRS-1-associated PI 3-kinase activity, possibly via Ser³⁰⁷ phosphorylation of IRS-1 (13), and this may also occur in human muscle (14). These latter abnormalities were unaccompanied by increased muscle ceramide concentrations; rather, increases in fatty acyl CoA and diacylglycerol were invoked as potential explanations (13). Whether ceramide plays a role in insulin resistance in human muscle is unknown.

Skeletal muscle from insulin-resistant subjects displays a wide variety of abnormalities in insulin receptor signaling, including decreased tyrosine phosphorylation of the insulin receptor and IRS-1, diminished IRS-1-associated PI 3-kinase activity, and impaired phosphorylation of Akt on Thr³⁰⁸ and Ser⁴⁷³ (26,27). Moreover, infusion of a triglyceride emulsion induces insulin resistance (27) and decreases IRS-1-associated PI 3-kinase activity during an insulin infusion (31). Although various lipid species have been proposed to be relevant to insulin resistance (13,15,16), little data are available in human muscle. The present study was undertaken, in part, to determine whether ceramide levels are increased in skeletal muscle from obese insulin-resistant subjects. We used a highly sensitive high-performance liquid chromatography electrospray ionization tandem mass spectrometry technique to analyze the content of individual ceramide species and other sphingolipids (29). The major new finding of this study is that ceramide concentrations are increased nearly twofold in muscle from the obese subjects. Furthermore, the results show that ceramide concentrations were increased in the obese subjects regardless of the nature of the fatty acyl moiety. This twofold increase was similar to the increase in ceramide produced by incubation of C2C12 cells with palmitate, a maneuver that resulted in insulin resistance (decreased insulin-stimulated glycogen synthesis) and decreased insulin-stimulated Akt phosphorylation in these cells (7). The increase in ceramide content of the obese muscle was specific because there were no differences in muscle content of sphingomyelin or the sphingoid bases. Because of the lack of increase in dihydroceramide, sphingosine, and sphinganine, the present data provide no evidence that the increase in ceramide is due to de novo synthesis. However, the increase in plasma FFA concentration observed in this study and its correlation with the muscle ceramide content suggest that increased substrate for ceramide synthesis is available. On the other hand, the sphingomyelin content of obese muscle was somewhat higher than that in lean control subjects, and these data suggest that the likely source of ceramide elevation is via turnover of sphingomyelin. Such a situation would be consistent with an increased response of muscle to inflammatory cytokines such as TNF-α, which is known to be increased in obesity. However, statistical analysis taking into account the variability in the measured amounts does not make this conclusion unequivocal. Additional studies will be required to determine whether the increased ceramide content in skeletal muscle of obese subjects is due to de novo synthesis or sphingomyelinase activity.

As expected, during insulin infusion, the obese subjects exhibited reduced insulin-stimulated glucose metabolism, as assessed by a decreased rate of glucose infusion required to maintain euglycemia. Because all insulin infusions were ≤60 min in duration, it is possible that a portion of the decrease in the glucose infusion rate in the obese subjects may have been due to impaired suppression of endogenous glucose output. However, that could not have accounted for all of the decrease in the glucose infusion rate, and, in addition, there is ample additional evidence that obese subjects are insulin resistant (26). As a marker of an effect of ceramide on insulin receptor signaling, we assessed insulin stimulation of phosphorylation of Akt Ser⁴⁷³. The results showed that there was a modest reduction in basal Akt phosphorylation and a dramatic impairment of insulin stimulation of Akt Ser⁴⁷³ phosphorylation. This is similar to the findings of Krook et al. (35) in patients with type 2 diabetes and is consistent with the decrease in upstream signaling observed in obese subjects (26,27). Taken together, the pattern of increased ceramide concentrations and decreased Akt phosphorylation is consistent with the idea that elevated ceramides contribute to insulin resistance in skeletal muscle. However, the situation is clearly more complex in vivo in human muscle than in cultured cells. As noted above, the actions of ceramide on insulin signaling have been reported to occur at the level of Akt as well as further upstream. In the present study, the relationship between ceramide levels and Akt activation was examined, and upstream events were not characterized. However, in previous studies, we characterized upstream insulin receptor signaling events in obese subjects and found reduced insulin stimulation of IRS-1 tyrosine phosphorylation and association of PI 3-kinase activity with IRS-1 (26). Although muscle ceramide levels were not measured in those studies, the subjects were similar, and we could expect that the obese subjects who took part in the present study also had reduced IRS-1 function. The relationship between increased ceramide concentration and IRS-1 function in human muscle remains to be investigated.

Several mechanisms have been described to explain the effect of ceramides to inhibit Akt activity. Earlier studies suggested that TNF-α, acting through ceramide, increased the activity of a membrane-bound proline-directed serine/threonine kinase (36), later identified as c-jun NH₂-terminal kinase (37). In rat FAO hepatoma cells, sphingomyelinase, cell-permeable ceramide, and TNF-α all produced dramatic decreases in IRS-1 tyrosine phosphorylation and association of the p85 regulatory subunit of PI 3-kinase with IRS-1 (19). More recently, a cell-permeable ceramide analog was used to show that in a neural-derived (HMN1) cell line and in COS-7 cells, ceramide-induced decreases in Akt activity were not due to activation of an Akt phosphatase or direct inhibition of Akt; on the other hand, ceramide had no effect on signaling through IRS-1 and PI 3-kinase (20). More recent studies using brown adipocytes showed that ceramide mediated insulin resistance induced by TNF-α and activated an okadaic acid-inhibitable protein phosphatase (38). Finally, Hajduch et al. (18) also used a cell-permeable ceramide in L6 muscle cells to show that ceramide induced an abnormality in membrane localization of Akt. Thus, most, if not all, studies report that ceramides, whether added exogenously, produced by de novo synthesis from palmitate, or generated by sphingomyelinase activity, inhibit Akt activity. However, the findings differ regarding whether ceramides inhibit insulin signaling through IRS-1. The reasons for the differences are not clear but are likely to involve differences among cell types.

The present results taken together with previous findings suggest the following model for the etiology of abnormalities in insulin signaling in muscle from obese insulin-resistant subjects (Fig. 5). The increase in ceramide content is sufficient to account for decreased insulin stimulation of Akt activity, which could, in theory, itself lead to a reduction in insulin-stimulated glucose uptake. However, muscle from obese subjects also displays marked abnormalities in insulin receptor and IRS-1 tyrosine phosphorylation and decreased association of PI 3-kinase activity with IRS-1 (26,27). The abnormality in insulin receptor function is unlikely to be due to increased ceramides but may be secondary to increases in fatty acyl CoA or diacylglycerol (13,15,16). The abnormality in IRS-1 function may be due in part to the action of ceramides (19), but it also is likely to be due to the influences of other lipid abnormalities (13) and inherited influences (39). The relative contributions of these abnormalities to decreased insulin-stimulated glucose uptake in skeletal muscle remain to be defined.

These studies were supported in part by National Institutes of Health Grants DK41075 (to L.J.M.), DK24092 (to R.A.D.), and GM46368 (to M.C.S.); the Georgia Research Alliance; and by General Clinical Research Center Grant M01-RR-01346.

The authors thank Sheila Taylor and Cindy Muñoz for expert technical assistance and John Kincaid, Jim King, Norma Diaz, and Tricia Wolfe for excellent nursing assistance.