One main mechanism of insulin resistance (IR), a key feature of type 2 diabetes, is the accumulation of saturated fatty acids (FAs) in the muscles of obese patients with type 2 diabetes. Understanding the mechanism that underlies lipid-induced IR is an important challenge. Saturated FAs are metabolized into lipid derivatives called ceramides, and their accumulation plays a central role in the development of muscle IR. Ceramides are produced in the endoplasmic reticulum (ER) and transported to the Golgi apparatus through a transporter called CERT, where they are converted into various sphingolipid species. We show that CERT protein expression is reduced in all IR models studied because of a caspase-dependent cleavage. Inhibiting CERT activity in vitro potentiates the deleterious action of lipotoxicity on insulin signaling, whereas overexpression of CERT in vitro or in vivo decreases muscle ceramide content and improves insulin signaling. In addition, inhibition of caspase activity prevents ceramide-induced insulin signaling defects in C2C12 muscle cells. Altogether, these results demonstrate the importance of physiological ER-to-Golgi ceramide traffic to preserve muscle cell insulin signaling and identify CERT as a major actor in this process.
A worldwide obesity and diabetes epidemic has been spreading in humans for four decades. It is concomitant with alterations of carbohydrate/lipid metabolism, particularly with dyslipidemia, which have major consequences in the form of cardiovascular disease and insulin resistance (IR). IR is a metabolic condition in which cells fail to respond to normal levels of insulin and a key actor of type 2 diabetes (T2D). Numerous studies performed in animals and humans have demonstrated a strong relationship between IR and increased intramyocellular lipid content. Ceramide has been described in many studies as the lipid species involved in muscle IR (1), although other studies did not find such a relationship and rather privileged diacylglycerols as responsible for muscle IR (2).
According to various studies, skeletal muscle accounts for 30–70% of insulin-stimulated glucose disposal in the postprandial state and is thus a primary target for ceramide anti-insulin action (3,4). In the context of visceral obesity, ceramides primarily are produced de novo from saturated fatty acid (FA) (palmitate) (1). This synthesis takes place in the endoplasmic reticulum (ER) and starts with the condensation of L-serine with palmitoyl-CoA to yield ceramides after several reactions.
Pioneering in vitro data have shown an involvement of ceramides in the development of IR through the direct addition of these lipids on muscle and adipocyte cell lines (5–7). Ceramides inhibit insulin-stimulated glucose uptake and glycogen synthesis by blocking insulin signaling at the level of both insulin receptor substrate 1 and Akt (8–11). These results indicate that saturated FAs in cells induce IR through ceramide synthesis.
After ceramides are synthesized de novo in the ER, they are transported to the Golgi apparatus and metabolized into other sphingolipids, such as sphingomyelin (SM) and glucosylceramide (GlcCer). The intracellular transport of ceramides from the ER to the Golgi involves both ATP-independent and -dependent specific carriers (12). Ceramides intended to be metabolized into GlcCers at the cis side of the Golgi are transported through an ATP-independent vesicular carrier. This carrier is not well characterized, except that its activity is phosphatidylinositol-3-kinase dependent (12). Although to be processed into SM, ceramides are mainly transported from the ER to the Golgi through a nonvesicular ATP-dependent transporter called ceramide transporter (CERT) (12). Through CERT, ceramides are extracted from the surface of the ER and transported toward the Golgi where they are metabolized into SM by SM synthase 1.
Transformation of ceramide into SM may be a critical step in preventing negative actions of ceramides in cells. A metabolomic study demonstrated that reduced levels of plasma C16:1-SM species is predictive of T2D (13). Inhibition of SM synthase in muscle cells induces a rise in ceramide content and impairs insulin signaling (14). Obese individuals with glucose intolerance show increased muscle ceramide content and lower muscle SM compared with obese individuals with normal glucose tolerance (15).
These data suggest that the biosynthesis of SM from ceramides could be protective for maintaining insulin sensitivity. Because CERT is involved in the transfer of ceramides to the Golgi apparatus for the synthesis of SM, we tested the hypothesis that modulation of CERT activity affects muscle insulin signaling.
Research Design and Methods
Insulin, palmitate, and BSA were obtained from Sigma-Aldrich (Saint-Quentin-Fallavier, France). Gedunin was from Tocris Bioscience (Bristol, U.K.). Broad caspase inhibitor (Q-VD-OPh) was from Merck Chemicals (Nottingham, U.K.). Antibodies against Akt, Akt serine (Ser)-473, Akt threonine (Thr)-308, protein kinase (PKD) Ser-916, GSK3α/β Ser-21/9, ERK Thr-202/tyrosine-204, GAPDH, and cleaved caspase-3 and -9 were from Cell Signaling Technologies (Danvers, MA). The antibody against CERT was from Bethyl Laboratories (Montgomery, TX), and the one directed against β-actin was from Sigma-Aldrich. Secondary horseradish peroxidase antibodies were from Jackson ImmunoResearch (West Grove, PA) and the chemiluminescent substrate from Thermo Fisher Scientific (Waltham, MA). [3H]2-deoxy-d-glucose (26.2 Ci/mmol) and d-erythro-[3-3H]sphingosine (18.6 Ci/mmol) were from PerkinElmer (Boston, MA). High-performance thin-layer chromatography (HPTLC) silica gel plates were from Merck (Darmstadt, Germany). Lipid internal standards (d18:1/12:0 ceramide, d18:1/12:0-SM, and d17:1 sphingosine-1-phosphate) were obtained from Avanti Polar Lipids (Coger SAS, Paris, France). Liquid chromatography-tandem mass spectrometry (LC-MS/MS)–quality-grade solvents were purchased from Fisher Scientific France (Illkirch-Graffenstaden, France).
Culture and Transfection of C2C12 Muscle Cells
Both N-(3-hydroxy-1-hydroxymethyl-3-phenylpropyl)-dodecanamide (HPA12) and gedunin were reconstituted in DMSO (0.4% final concentration). Control cells were incubated with the same quantity of DMSO. Small interfering RNA (siRNA) (25 nmol/L) directed against CERT (Santa Cruz Biotechnology, Dallas, TX) or the same concentration of a nonspecific siRNA were transfected for 96 h into C2C12 myotubes using the transfection reagent DharmaFECT (Dharmacon, Cambridge, U.K.). C2C12 myoblasts were seeded into 12-well plates and transfected for 48 h with a pEGFP N1/hCERT (a gift from T. Levade, Toulouse, France) or pCMV-GFP plasmids (1 µg/well) using the TransfeX transfection reagent (ATCC, Molsheim, France).
C2C12 Glucose Transport
Glucose transport was measured by incubating C2C12 myotubes with 10 μmol/L [3H]2-deoxy-d-glucose (1 µCi/mL [26.2 Ci/mmol]) for 10 min as previously described (17).
C2C12 myotubes were treated either with BSA (1.5%) or with palmitate (0.75 mmol/L) complexed with BSA in the presence or absence of the caspase inhibitor Q-VD-OPh (10 µmol/L), and the activity of caspase-3/7 was measured 24 h later using the Apo-ONE Homogeneous Caspase-3/7 Assay (Promega, Madison, WI).
C2C12 cells were lysed and CERT immunoprecipitated from 200 µg lysates using a CERT antibody. Immunocomplexes were captured by incubation with protein A agarose beads and solubilized in Laemmli buffer before SDS-PAGE and immunoblotting.
C2C12 myotubes were treated with 0.1 or 0.75 mmol/L palmitate in the presence or absence of 10 μmol/L HPA12 (18) for 16 h at 37°C, and then the cells were pulsed for 2 h with [C3-3H]sphingosine (0.3 μCi/mL) at 10°C (19). Stock solutions of [3H]sphingosine in absolute ethanol were prepared and added to conditioned medium. The final concentration of ethanol never exceeded 0.1% volume for volume. At the end of the pulse time, total lipids were extracted and processed as previously described (19). The methanolyzed organic phase was analyzed by HPTLC using chloroform:methanol:water 55:20:3 by volume as the solvent system. Digital autoradiography of HPTLC plates was performed with BetaIMAGER 2000 (Biospace, Nesles la Vallée, France), and radioactivity associated with individual lipids was determined using software provided with the instrument. The 3H-labeled sphingolipids were recognized and identified as previously described (19).
Male C57BL/6 mice (5 weeks old; Charles River Laboratories, Saint-Germain-Nuelle, France) were adapted to their environment for 1 week before the study. The mice were housed with a 12-h light/12-h dark cycle in a temperature-controlled environment and had free access to water and a regular diet (65% carbohydrate, 11% fat, 24% protein) or a high-fat diet (HFD) (EF D12492: 21% carbohydrate, 60% fat, 19% protein, gross energy, 24.0 MJ/kg; ssniff Spezialdiäten GmbH, Soest, Germany) for 12 weeks. All procedures were approved by the Regional Ethics Committee for Animal Experiments No. 5 of Ile-de-France (agreement no. 02852.03).
Electrogene Transfer in Mice
Mice were anesthetized with AErrane (Baxter, Deerfield, IL), and their tibialis anterior muscles were injected with 8 units hyaluronidase 2 h before the injection of 15 μg pEGFP N1/hCERT or 15 μg pCMV-GFP plasmids. Six 65 V/cm pulses of 60 ms, with a 100-ms interval, were applied (20). Fourteen days after gene delivery and before sacrifice, mice were injected or not with 0.75 International Units/kg insulin (Actrapid; Novo Nordisk, La Défense, France) for 15 min. Muscles were then collected under microscope. All experiments were conducted in accordance with European guidelines for the care and use of laboratory animals and were approved by the institutional animal care and use committee (agreement no. 00315.01).
SM and ceramide were extracted according to Bielawski et al. (21). Muscles were crushed in an Omni Bead Ruptor 24 homogenizer (Omni International, Kennesaw, GA) with 950 µL saline and ∼20 1.4-mm zirconium oxide beads. An aliquot equivalent to 3 mg muscle (60 µL lysate) was diluted with 1.94 mL saline and finally spiked with an internal standard mix containing 30 and 125 ng d18:1/12:0 ceramide and d18:1/12:0-SM, respectively. Lipids were extracted with 2 mL propanol 2:water:ethyl acetate 30:10:60 for 30 min. After centrifugation (1100g for 5 min), the organic phase was kept and the aqueous phase further extracted. After centrifugation, both organic phases were combined and evaporated to dryness under vacuum. Samples were solubilized with 200 µL methanol and transferred to injection vials, again evaporated to dryness under vacuum, and finally solubilized with 40 µL methanol.
Quantification of Ceramides and SM by LC-MS/MS
Ceramide analysis was carried out on an Agilent 1200 Series 6460 Triple Quadrupole LC-MS/MS system equipped with an electrospray ionization source (Agilent Technologies, Les Ulis, France) as previously described (22). Samples were injected on a Poroshell 120 EC-C8 2.1 × 100 mm, 2.7-µm column (Agilent Technologies) (flow rate 0.3 mL/min, 50°C), and separation was achieved with a linear gradient of formic acid/ammonium formate 0.2%/1 mmol/L final concentration (solvent A) and methanol containing formic acid/ammonium formate 1 mmol/L (solvent B). Acquisition was performed in positive single reaction monitoring mode. Relative quantitation of ceramide-related compounds was performed by calculating the response ratio of the considered ceramide to d18:1/12:0 ceramide used as the internal standard. Two-microliter samples were used for quantitation of SM.
Human Skeletal Muscle Cells
Biopsy samples from lean healthy adult volunteers were obtained in the context of agreed preclinical and clinical experiences (23) through the tissue bank for research (Myobank) of the French Association Against Myopathies in agreement with French bioethical law (Law No. 94-654 of 29 July 1994, amended 22 January 2002). Samples from patients with T2D were obtained from healthy tissue after leg amputation upon informed consent. Ethical approval for the use of human muscle tissue was given by the Ethics Committee of Pitié-Salpêtrière Hospital (CPP-Ile de France VI–Paris, France). Fresh muscle samples were sliced and dissociated in collagenase. Satellite cells were purified, cultured, and differentiated into myotubes as previously described (16).
Preparation of Whole-Cell Lysates
Cells were lysed after experimental manipulation (see the figure legends) in an appropriate volume of lysis buffer and frozen at −80°C until required (24).
Real-time Quantitative RT-PCR
Total RNA was extracted from muscle cells, and real-time quantitative RT-PCR analyses were performed as described previously (16). One microgram RNA was retrotranscribed using SuperScript II (Invitrogen, Carlsbad, CA). Sequences of sense and antisense primers of the gene to be amplified (CERT) were 5′-TCTGCTTATCTCCTGGTCTCCC-3′ and 5′-CGAATCAAGCCAGCCTTGAC-3′, respectively.
Frozen tissues or cells were homogenized after experimental manipulation in an appropriate volume of lysis buffer, and cell lysates were subjected to SDS-PAGE and immunoblotted as previously described (24).
Data were analyzed with GraphPad Prism 6.07 statistical software by unpaired or paired two-tailed t-test when two groups were compared and by one-way ANOVA followed by Bonferroni multiple comparison test when more than two groups were compared. P < 0.05 was considered significant.
CERT Expression Is Altered in Lipotoxic Conditions in Muscle Cells
Palmitate treatment (0.75 mmol/L) for 16 h induced a 50% decrease in CERT protein expression (Fig. 1A) concomitantly with a 60% increase in total ceramide content (Fig. 1B) and a 35% increase in total SM content (Fig. 1B) in C2C12 myotubes. In gastrocnemius muscle lysates from mice fed an HFD (12 weeks), we also observed a 58% decrease in CERT protein content (Fig. 1C) and a 28% increase in total ceramide content (Fig. 1D) compared with controls. However, no difference in total SM content was observed between groups (Fig. 1D). We then studied human myotubes differentiated from human satellite cells obtained from either insulin-sensitive or T2D donors (10). Figure 1E shows that insulin-stimulated Akt phosphorylation in myotubes derived from muscles from patients with T2D was drastically reduced compared with nondiabetic myotubes. Of note, a concomitant decrease in CERT expression was observed in T2D myotubes compared with control myotubes (Fig. 1E).
Next, we tested whether the decrease in insulin-induced Akt phosphorylation usually observed after 16 h of palmitate exposure (10,11,16) was concomitant with a decreased CERT expression. C2C12 cells were treated with palmitate for up to 16 h, and insulin-induced Akt phosphorylation and CERT expression were assessed in the same time frame. Supplementary Fig. 1A shows that palmitate needed 16 h to induce both a defect in insulin signaling (decrease in Akt phosphorylation in response to insulin) and decreased CERT protein content. CERT mRNA levels were not decreased after 16 h of palmitate incubation in C2C12 myotubes (Supplementary Fig. 1B) and in muscle of mice fed an HFD compared with control mouse muscles (Supplementary Fig. 1C), suggesting that the alteration of CERT observed in lipotoxic conditions was posttranscriptional.
We next tested whether palmitate could act through ceramide production to alter CERT protein expression in muscle cells. C2C12 myotubes were treated with palmitate in the presence of myriocin (inhibitor of the first enzyme of ceramide biosynthesis) for 16 h before assessing CERT expression. Decreased CERT protein content observed after palmitate treatment (Fig. 1F) was concomitant to an increase in ceramide content in cells (Fig. 1G). Of note, both the decreased expression of CERT and the increased ceramide content observed in response to palmitate were completely abrogated in the presence of myriocin (Fig. 1F and G), suggesting that ceramides produced from palmitate are accountable for the observed CERT alteration in muscle cells.
The type of free FA, saturated or unsaturated, is critical for the development of IR. Although saturated FAs induce IR (25,26), unsaturated FAs have no deleterious effect and even protect cells from the negative action of saturated FAs (27–29). To determine whether unsaturated FAs exert a protective effect on the expression of CERT in the presence of palmitate, C2C12 myotubes were incubated with palmitate, oleate, or linoleate. Supplementary Fig. 2A shows that although palmitate altered CERT expression, the other two unsaturated FAs displayed no significant effects. Furthermore, treatment of C2C12 myotubes with both oleate and palmitate together protected cells against the harmful effect of the latter on CERT expression (Supplementary Fig. 2B). Overall, these data demonstrate that lipotoxic conditions negatively regulate CERT content and activity in muscle cells.
Influence of the Modulation of CERT Activity/Expression on Muscle Cell Insulin Signaling In Vitro
We next determined whether an artificial reduction of CERT function could potentiate palmitate-induced defects in insulin signaling in myotubes. We used a concentration of palmitate (0.1 mmol/L) that had a minimal effect on total ceramide content (Supplementary Fig. 3) and CERT expression (Fig. 2B). We inhibited the activity of CERT in muscle cells by using the CERT inhibitor HPA12 (18). HPA12 inhibited CERT activity through its interaction with the stAR-related lipid-transfer domain of CERT that usually binds ceramides (30). HPA12 treatment enhanced the ceramide concentration induced by a low concentration of palmitate (Supplementary Fig. 3), suggesting that the inhibition of CERT activity prevents the ceramide produced in the ER to be metabolized into SM in the Golgi apparatus. To evaluate the effect of both palmitate and HPA12 on ceramide utilization for the biosynthesis of SM, we studied ceramide metabolism using [3H]sphingosine as a metabolic precursor because it is rapidly internalized into cells and N-acylated to ceramide and then metabolized to form SM and GlcCer (31). The experiment was performed at 10°C, a nonpermissive temperature for the ER-to-Golgi vesicle flow, allowing us to assess essentially a CERT-dependent transport. At that temperature and after a short time pulse, we found comparable levels of radioactivity incorporated in control and palmitate-treated cells (data not shown), and most of the radioactivity remained associated with ceramides (Fig. 2A). A 0.1 or 0.75 mmol/L palmitate treatment inhibited synthesized [3H]-SM levels by 31% and 70%, respectively (Fig. 2A) but did not affect GlcCer biosynthesis (Fig. 2A). HPA12 treatment mimicked high-concentration levels of palmitate and blocked the conversion of ceramide to SM. The addition of 0.75 mmol/L palmitate did not enhance the negative action of HPA12 (Fig. 2A). Taken together, the data demonstrate that both high palmitate concentration and HPA12 inhibit SM biosynthesis in the Golgi apparatus. We then assessed insulin signaling. At 0.1 mmol/L, palmitate did not affect CERT expression (Fig. 2B) but only partially blocked SM biosynthesis (Fig. 2A) and did not inhibit insulin-induced Akt phosphorylation (Fig. 2B). However, at the same concentration of palmitate but in the presence of HPA12, we observed a complete inhibition of ceramide transport from the ER to the Golgi (Fig. 2A) and an accentuated inhibitory action of the lipid on insulin signaling (Fig. 2B). Of note, at 0.75 mmol/L, the negative effect of palmitate on CERT expression and SM biosynthesis was maximal, and HPA12 did not potentiate any further the deleterious action of the lipid on Akt phosphorylation (Fig. 2B).
Similar results were obtained by using another CERT inhibitor, gedunin, which inhibited CERT-mediated extraction of ceramides from the ER membranes (32). Figure 2C shows that like HPA12, gedunin unmasked the inhibitory action of 0.1 mmol/L palmitate on insulin signaling.
To confirm the importance of the negative effect of a decreased-CERT activity on insulin signaling in muscle cells, we used an siRNA directed against CERT (siCERT). Figure 3A and B show that siCERT decreased CERT mRNA and protein content in C2C12 myotubes. Figure 3B demonstrates that 0.1 mmol/L palmitate induced only a slight reduction in CERT protein content and had no significant effect on Akt phosphorylation. However, in cells transfected with siCERT, CERT protein was strongly reduced (Fig. 3B) along with an increased action of palmitate on insulin-induced Akt phosphorylation, confirming that the absence of CERT expression unmasks the action of a low palmitate concentration on insulin signaling (Fig. 3B).
To demonstrate that a correct ceramide transport from the ER to the Golgi apparatus is essential in preventing the inhibitory effect of ceramides on insulin signaling, we overexpressed CERT in C2C12 myoblasts before treating them with palmitate and insulin. For this experiment, we used myoblasts instead of myotubes because of their higher transfection efficiency. Endogenous CERT expression was identical before and after C2C12 myoblast differentiation into myotubes (data not shown). A 16-h palmitate exposure induced a sevenfold increase in total ceramide content in C2C12 myoblasts (Fig. 4A) (a much higher increase than in myotubes [e.g., Fig. 1B]). CERT overexpression, however, reduced the total ceramide increase in response to palmitate to 4.1-fold (Fig. 4A). Overexpression of an exogenous CERT prevented endogenous CERT downregulation in response to a high palmitate concentration (Fig. 4B). In addition, CERT overexpression also counteracted palmitate’s deleterious action on insulin signaling. Indeed, palmitate induced an inhibition of the insulin-induced phosphorylation of Akt, GSK3, and ERK. CERT re-expression, however, induced a major improvement in insulin signaling (Fig. 4B).
Mechanism of Alteration of CERT Action in Lipotoxic Conditions: A PKD- and Caspase-Dependent Mechanism
CERT function is downregulated by phosphorylation on its Ser-132 residue by PKD. PKD is activated by phosphorylation of its Ser-960 residue in response to various stresses (33), and CERT phosphorylation on Ser-132 by PKD decreases CERT affinity to phosphatidylinositol-4-phosphate in the Golgi apparatus, thus reducing ceramide transfer activity (34). Palmitate induced both phosphorylation of PKD on its Ser-960 residue (Supplementary Fig. 4A) and CERT on its Ser-132 residue (Supplementary Fig. 4B). Treatment of cells with a PKD inhibitor (kb-NB-142-70), however, reduced palmitate-induced CERT phosphorylation (Supplementary Fig. 4B).
A previous study showed that CERT can be cleaved by caspases during proapoptotic stress in HeLa cells, resulting in a loss of function of CERT and a decrease in SM de novo synthesis in the Golgi apparatus (35). Because palmitate activates both caspase-3 and caspase−9 in C2C12 myotubes (36), we hypothesized that a similar mechanism would occur in our muscle cell model. To test this hypothesis, we treated C2C12 myotubes with palmitate for 16 h in the presence or absence of a broad caspase inhibitor (Q-VD-OPh) (37). Palmitate induced the cleavage of both caspase-3/9 (Fig. 5A) and activity (Fig. 5B) in C2C12 myotubes together with a loss of insulin response (Fig. 5C). Of note, Q-VD-OPh, which blocks caspase activity in the presence of palmitate (Fig. 5B), prevented the alteration of CERT expression in response to palmitate (Fig. 5C) and improved insulin signaling (Fig. 5C).
We next evaluated whether changes in CERT expression have consequences on glucose metabolism downstream of insulin signaling. Insulin induced a 40% increase in glucose transport in C2C12 myotubes (Fig. 6). If at 0.1 mmol/L palmitate had no significant effect on insulin-stimulated glucose transport, in the presence of HPA12, palmitate inhibited the insulin-induced stimulation of glucose transport (Fig. 6). This result confirms the importance of an active ceramide transport to counteract the action of palmitate on insulin signaling. The stimulation of insulin was completely lost when cells were pretreated with high palmitate concentrations (Fig. 6). The caspase inhibitor Q-VD-OPh, however, prevented the inhibition by palmitate of insulin-stimulated glucose transport (Fig. 6).
Influence of CERT Overexpression on Muscle Insulin Sensitivity and Ceramide Content In Vivo
We fed mice an HFD for 10 weeks, and 2 weeks before the end of the diet, we overexpressed a CERT-GFP construct through electrogene transfer (20) in the left-side tibialis anterior muscle of the mice (Fig. 7A). In the right leg, a GFP construct was transferred. Two weeks later, we sacrificed the mice and isolated the tibialis anterior muscles that were overexpressing CERT (GFP-CERT visible under fluorescence microscope) (Fig. 7A). Ceramide species contents (except for C18 and C20) were decreased up to 30% in CERT-overexpressing muscle fibers (Fig. 7B), demonstrating that CERT overexpression counteracts ceramide accumulation in muscles. We also observed a 10% decrease in total ceramide content (although at P < 0.0533). In addition, both caspase-3 and caspase-9 were cleaved in muscle of HFD-fed mice, and CERT overexpression completely abrogated caspase cleavages (Fig. 7C), indicating a decrease in lipotoxicity-induced caspase activation in vivo. Next, we assessed insulin signaling in these muscle fibers. CERT overexpression in the tibialis anterior muscle prevented endogenous CERT degradation observed in response to lipotoxicity and improved significantly the poor in vivo stimulation by insulin of Akt, GSK3, and ERK observed in HFD-fed mice (Fig. 7D).
In the current study, we demonstrate that CERT plays a pivotal role in the control of ceramide content and insulin response in muscle cells. In lipotoxic conditions (i.e., in the presence of saturated FA), a decrease in CERT content induces ceramide accumulation in cells through a defective ceramide transport from the ER to the Golgi apparatus. This leads to an inhibition of SM synthesis and a concomitant loss in insulin response. Of importance, a direct inhibition of CERT expression/activity has an effect on ceramide content and insulin signaling that is similar to lipotoxic conditions. Conversely, an increased CERT expression in vitro and in vivo counteracts the deleterious effects of lipotoxic conditions on muscle insulin signaling. As a reflection of insulin action, we show mainly AKT, GSK3, and ERK phosphorylation. Defects in insulin-mediated phosphorylation of these targets are usually concomitant with a loss of tissue insulin sensitivity (38), although a number of studies have shown a dissociation between Akt phosphorylation and insulin sensitivity (39–41). Thus, it remains to be demonstrated directly that CERT activity/expression modulation translates in vivo into functional changes of muscle glucose metabolism.
We demonstrate the direct involvement of CERT in the transport of ceramides from the ER to the Golgi apparatus and in their synthesis into SM. However, the fact that the GlcCer content did not change with either palmitate or CERT inhibitor confirms that GlcCers are coming from ceramide transported to the Golgi through the vesicular transport, not through CERT (42). We did not, however, observe a decrease in total SM content in cells in the lipotoxic condition (Fig. 1A and B) because the total SM content exceeded the total ceramide content by >20-fold in muscle cells (results not shown).
A targeted decrease in some ceramide species (C16, C22, C24:1, and C24) without a statistically significant change in total ceramide content seems sufficient to modulate the insulin response of the cells in lipotoxic conditions. C16 ceramides have been demonstrated to attenuate the hepatic insulin response (43,44); thus, similar ceramide species also could mediate lipotoxicity in muscle cells. However, and opposite of what was suggested in some studies (40,45), C18 ceramides did not seem to play a role in the inhibition of insulin response in our experimental models. At present, we have no explanation for this discrepancy. Depending on the relative abundance of specific FAs in the HFD, ceramide species likely are affected differently.
The processing of newly synthetized ceramides to give more complex sphingolipid derivatives, such as SM, GlcCer, and complex glycosphingolipids (e.g., gangliosides) (46), occurs in the Golgi apparatus. Therefore, an efficient, rapid, and regulated transport is required because the half-life for spontaneous interbilayer movement of ceramide is in the order of days (47). In mammalian cells, transport of ceramides from the ER to the Golgi occurs through two different mechanisms: vesicular and nonvesicular. In yeast, nonvesicular CERT-dependent transport has been shown to contribute to 50% of transport ceramide from the ER to the Golgi (48). The current data demonstrate that a default in CERT content in response to lipotoxicity is enough to stop the conversion of ceramide to SM in the Golgi and, thus, to increase ceramide concentration in cells and to trigger their negative action on insulin signaling. This suggests that the loss in CERT-dependent transport of ceramides from the ER to the Golgi is not compensated by the vesicular transport. However, a lack of knowledge about the mechanism by which this ceramide vesicular transport is regulated precludes any conclusion about an effect of lipotoxicity on its function.
A key and original result is that in lipotoxic conditions, CERT function is modulated through two mechanisms. First, palmitate inhibits CERT activity through phosphorylation on its Ser-132 residue by PKD in muscle cells. The importance of this mechanism for the control of CERT activity already has been demonstrated in rat islet β-cells, where high-dose palmitate treatment increased PKD-induced phosphorylation of CERT and its dysfunction and deleterious effects on islet β-cells (49). Second, CERT protein content is strongly reduced in lipotoxic conditions, and the decreased protein content is secondary to an activation of caspases. The underlying mechanism is not completely resolved but is likely to be mediated by ceramides. Indeed, decreased CERT expression did not occur when the de novo ceramide biosynthesis pathway was inhibited (Fig. 1F). In addition, when CERT was overexpressed in diabetic mouse muscle, ceramide content was decreased (Fig. 7B) and lipotoxicity-induced caspase cleavage abrogated (Fig. 7C). The caspase-dependent inhibition of CERT activity likely prevailed over the PKD-dependent one. Indeed, inhibition of caspase activity restored insulin signaling on its own (Fig. 5C).
In lipotoxic conditions, GFP-CERT–overexpressing muscle cells displayed higher endogenous CERT concentrations than control muscle cells (Figs. 4B and 7D). This difference in expression could be explained by an enhanced ceramide transport from the ER to the Golgi apparatus in CERT-overexpressing cells, resulting in less ceramide accumulated and a decrease in ceramide-induced caspase activation. Thus, CERT plays a crucial role in the regulation of sphingolipid metabolism in lipotoxic conditions.
Changes in CERT expression observed in the current study are not unprecedented. Indeed, another study has shown that proapoptotic stress (induced by tumor necrosis factor-α) also can result in an inactivation of CERT through its cleavage by caspases in HeLa cells (35) and results in a decreased biosynthesis of SM (35). Of note, another study demonstrated that the inhibition of the de novo ceramide biosynthesis prevents caspase-3 activation in response to palmitate in L6 myotubes (50). As we have observed (Fig. 5), these authors also showed that inhibition of caspase-3 partially improves insulin-stimulated glucose uptake in palmitate-treated L6 myotubes. They did not explain, however, how ceramide-activated caspases could affect insulin signaling, but they did show that it is not through proteolysis of insulin signaling proteins. Recently, a study conducted in yeast also demonstrated that a nonvesicular ceramide transfer out of the ER prevents the buildup of ceramide content (51). Although Saccharomyces cerevisiae does not express a CERT homolog, under certain conditions, the protein Nvj2p can play a similar role by facilitating lipid exchange between the ER and the Golgi apparatus. During ER stress or ceramide overproduction, Nvj2p relocalizes to and increases ER-medial Golgi contacts, facilitating ceramide exit from the ER and preventing toxic ceramide accumulation (51). All these data strengthen the current results in mammalian cells by demonstrating the importance of a functional ceramide transport from the ER to the Golgi to prevent lipotoxicity. Nevertheless, it will be particularly important to completely elucidate CERT regulation in response to an excess of FA.
We previously showed that the mechanisms inhibiting insulin action in the presence of ceramides take place at the plasma membrane (PM) (8,10). A crucial question thus remains: During lipotoxicity, what is the mechanism linking ER-accumulated ceramides to their effects at the PM? One hypothesis would be that ceramides could directly transfer from the ER to the PM at specialized membrane contact sites (MCSs). MCSs are regions of close apposition between two cellular membranes, and the ER can form MCSs with virtually any other organelle within cells. On electron microscopy, the ER also shows contacts with the PM (46). ER-PM MCSs exist in striated muscles between the transverse tubule (T-tubule) (a specialized PM network) and the sarcoplasmic reticulum (52). Of note, most of the insulin-regulated glucose transporter GLUT4 gets translocated into T-tubules in response to the hormone (53). As such, ER-T-tubule MCSs could form important hubs for lipid metabolism and exchange between membranes. To our knowledge, however, no information exists to date on sphingolipid transfer from the ER to the PM by such a mechanism.
In summary, this study shows that ceramide transport from the ER to the Golgi apparatus is altered in lipotoxic conditions and that its artificial recovery prevents ceramide to accumulate and to act negatively on insulin signaling. These results could open avenues to identify new therapeutic targets that allow for the amelioration of insulin sensitivity.
Acknowledgments. The authors thank N. Venteclef (INSERM U1138) for constructive comments and F. Hajduch (Anglais pour vous, Melun, France) for professional editing. The authors also thank J.-T. Vilquin (UPMC UM76, Groupe Hospitalier Pitié-Salpêtrière, Paris, France) and F. Koskas, J. Gaudric, C. Goulfier, C. Jouhannet, and T. Khalife (Service de Chirurgie Vasculaire, AP-HP, Hôpital Pitié-Salpêtrière, Paris, France) for providing the human muscle samples. The authors are grateful for the technical assistance of the staff of the Centre d’Exploration Fonctionnelle (Cordeliers Research Centre and Faculté de médecine–Sorbonne Université, Paris, France) and thank J.-P. Pais de Barros (INSERM UMR866, Université de Bourgogne, Dijon, France) for the assessment of sphingolipids. Finally, the authors are grateful to the tissue bank of the French Association Against Myopathies for control human biopsy samples.
Funding. This work was supported by INSERM, the Société Francophone du Diabète, the Agence Nationale de la Recherche (ANR 11 BSV1 03101-Crisalis), and the Fondation pour la Recherche Médicale (équipe FRM DEQ20140329504). C.L.B. is the recipient of a doctoral fellowship from Sorbonne Université. This study was also supported by Piano di sostegno alla ricerca BIOMETRA–Linea B (grant 15-6-3003005-9) to P.G.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. C.L.B., R.M., J.V., A.S., M.P., P.G., M.C., E.P., A.B.-Z., R.B., X.L.L., D.B., and J.G. participated in data collection and generation and reviewed the manuscript. O.B. collected human muscle samples. D.B. provided the HPA12. P.G., P.F., H.L.S., F.F., and E.H. designed the experiments, participated in data collection and generation, and wrote and edited the manuscript. C.L.B., R.M., J.V., A.S., M.P., P.G., M.C., E.P., A.B.-Z., R.B., X.L.L., O.B., D.B., J.G., P.F., H.L.S., F.F., and E.H. reviewed the results and approved the final version of the manuscript. E.H. 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. Part of this work was presented at the European Association for the Study of Diabetes 50th Annual Meeting, Vienna, Austria, 15–19 September 2014, and the 12th Sphingolipid Club Meeting, Trabia, Italy, 7–10 September 2017.