Increased FAT/CD36 Cycling and Lipid Accumulation in Myotubes Derived from Obese Type 2 Diabetic Patients

Background Permanent fatty acid translocase (FAT/)CD36 relocation has previously been shown to be related to abnormal lipid accumulation in the skeletal muscle of type 2 diabetic patients, however mechanisms responsible for the regulation of FAT/CD36 expression and localization are not well characterized in human skeletal muscle. Methodology/Principal Findings Primary muscle cells derived from obese type 2 diabetic patients (OBT2D) and from healthy subjects (Control) were used to examine the regulation of FAT/CD36. We showed that compared to Control myotubes, FAT/CD36 was continuously cycling between intracellular compartments and the cell surface in OBT2D myotubes, independently of lipid raft association, leading to increased cell surface FAT/CD36 localization and lipid accumulation. Moreover, we showed that FAT/CD36 cycling and lipid accumulation were specific to myotubes and were not observed in reserve cells. However, in Control myotubes, the induction of FAT/CD36 membrane translocation by the activation of (AMP)-activated protein kinase (AMPK) pathway did not increase lipid accumulation. This result can be explained by the fact that pharmacological activation of AMPK leads to increased mitochondrial beta-oxidation in Control cells. Conclusion/Significance Lipid accumulation in myotubes derived from obese type 2 diabetic patients arises from abnormal FAT/CD36 cycling while lipid accumulation in Control cells results from an equilibrium between lipid uptake and oxidation. As such, inhibiting FAT/CD36 cycling in the skeletal muscle of obese type 2 diabetic patients should be sufficient to diminish lipid accumulation.


Introduction
Intramyocellular lipid accumulation in skeletal muscle is abnormally high in type 2 diabetes and contributes to the etiology of the pathology [1]. Long chain fatty acid (such as palmitate) uptake is achieved by a concert of co-existing mechanisms. These lipids can passively diffuse, but certain membrane proteins can also accelerate the transport. Membrane fatty acid transporters can modulate the balance between fatty acid uptake and subsequent storage and/or oxidation in muscle tissue. The principal muscle fatty acid transporter FAT/CD36 (fatty acid translocase) is involved in regulating the uptake of long-chain fatty acids into skeletal muscle [2,3]. Abnormal increased fatty acid transport [4] and membrane FAT/CD36 relocation [4,5] independent of mitochondrial dysfunction result in an excessive accumulation of intramyocellular lipid in skeletal muscle tissue of type 2 diabetic patients. To date, little is known about FAT/CD36 regulation in human skeletal muscle. In this tissue, the mechanisms known to induce FAT/CD36 translocation from intracellular storage compartments towards plasma membrane are insulin, (AMP)activated protein kinase (AMPK) signaling pathways and muscle contraction [6,7].
Mature muscles possess a population of satellite cells located on their surface. In vitro satellite cells differentiation is characterized by withdrawal of myoblasts from the cell cycle, induction of musclespecific gene expression, and cell fusion into multinucleated myotubes. After differentiation in vitro, satellite cells are composed of two populations of cells: mature myotubes and quiescent undifferentiated myoblasts (reserve cells) [8]. In adipocytes, FAT/ CD36 expression appeared to be closely linked to preadipocyte differentiation [9], however, the regulation of FAT/CD36 expression and localization during the differentiation of primary human satellite cells has not been previously studied. These primary human satellite cells, when derived from type 2 diabetic patients, have been shown to display the majority of the defects previously described for type 2 diabetic muscle in vivo including defective insulin signaling pathway [10,11,12], metabolic inflexibility [13,14,15] and abnormal cell surface FAT/CD36 localization [5].
In this context, cell culture of human primary satellite cells offers an excellent and dynamic model to study the link between lipid accumulation and the regulation of FAT/CD36 localization. The aim of this study was to compare FAT/CD36 regulation between primary human satellite cells isolated from healthy subjects and from obese type 2 diabetic patients [5] in order to uncover new mechanisms of FAT/CD36 regulation associated with obesity and/or type 2 diabetes.

Plasma membrane FAT/CD36 and lipid accumulation are increased in OBT2D cells
We have previously shown that lipid accumulation was dependent upon the obesity of the donor subject and partially due to cell surface FAT/CD36 localization [5]. We confirmed this result by showing that palmitate-induced lipid accumulation was significantly increased in myotubes derived from obese type 2 diabetic patients (OBT2D) compared to myotubes derived from non-obese healthy subjects (Control) (p,0.05; Fig. 1 A). We also confirmed cell surface localization of FAT/CD36 in OBT2D cells, by immunofluorescence staining of FAT/CD36 in living cells (no permeabilization) using two different antibodies raised against FAT/CD36, a polyclonal antibody (H300) (Fig. 1 B, left panel) and a monoclonal conjugated to alexa488 antibody (CD36-alexa488) ( Fig. 1 B, right panel). Moreover, cell surface localization of FAT/CD36 was specifically observed in multinucleated troponin T (TT) positive OBT2D myotubes with both antibodies (left and right panels), with no such staining in Control cells. In order to show that cell surface staining of FAT/CD36 was quantitatively associated with plasma membrane relocation, we performed fractionation experiments and prepared plasma membrane (Mb) and post-plasma membrane (P) fractions ( Fig. 1  C). Since caveolin 3 is expressed exclusively in muscle cell types [16] and is expressed in plasma membrane microdomains, it was used as a marker of plasma membrane fractions. We monitored total FAT/CD36 and caveolin 3 expression in Control derived from four different healthy subjects (1 to 4) and in OBT2D derived from three obese type 2 diabetic patients (1 to 3) cells before fractionation ( Fig. 1 C, top panels). After fractionation ( Fig. 1 C, bottom panels), FAT/CD36 in Control (1 to 4) differentiated cells was only detected in caveolin 3 negative P fractions. In contrast, in OBT2D (1 to 3) differentiated cells, FAT/CD36 was found in both P and Mb fractions.

FAT/CD36 membrane localization in OBT2D myotubes is involved in lipid accumulation
During skeletal muscle differentiation, myoblasts exit the cell cycle and differentiate, giving rise to a heterogeneous population of cells. A major subpopulation consists of myotubes, quiescent multinucleated cells expressing muscle-specific differentiation proteins such as troponin T. A minor subpopulation of myoblasts remains quiescent and undifferentiated [8]. Positive oil red O staining was only observed in multinucleated Control and OBT2D myotubes but not in reserve cells (Fig. 2 A, arrows show reserve cells). As shown in figure 2 B (overlay of FAT/CD36, troponin T and DAPI staining) cell surface FAT/CD36 appeared as a punctuate staining only detected at the cell surface of troponin T positive cells and not in reserve cells (arrows). To determine if the specific localization of FAT/CD36 in OBT2D myotubes was due to increased expression of FAT/CD36 during differentiation, we monitored total FAT/CD36 expression by Western blot in two Control (1 and 2) and two OBT2D (1 and 2) cell cultures (0, 2, 4, 6 and 8 days of differentiation) (Fig. 2 C). a-tubulin expression was used as a loading control. OBT2D satellite cells have a normal induction of troponin T and caveolin 3 expression during differentiation. Total FAT/CD36 expression was already detected in proliferative cells and was not increased during differentiation in Control and OBT2D cells (Fig. 2 C, right panel).
Plasma membrane FAT/CD36 and lipid accumulation were observed in differentiated OBT2D cells. To determine whether a positive correlation between membrane FAT/CD36 and lipid accumulation could be observed in OBT2D cells, cell surface FAT/CD36 and oil red O co-staining was performed during differentiation. Positive oil red O and membrane FAT/CD36 costaining was only detected in differentiated cells, after 4 and 8 days of differentiation (Fig. 2 D). We next determine whether FAT/ CD36 membrane localization and lipid accumulation were linked in OBT2D myotubes. Palmitate uptake occurs through passive diffusion and protein-mediated transport. Phloretin is a general transport inhibitor [17]. We tested the possible influence of phloretin ( Fig. 2 E) prior to palmitate treatment. Phloretin inhibited palmitate-induced lipid accumulation to a greater extend in OBT2D cells than in control cells (278.467.3% versus 249.469.5%, p,0.05). We next tested the possible influence of sulfo-N-succinimidyloleate (SSO), which specifically binds to plasma membrane FAT/CD36, resulting in an arrest of the transport function of this protein [18] (Fig. 2 F). Incubation of Control and OBT2D differentiated cells with SSO inhibited palmitate-induced lipid accumulation. Again, the inhibition was significantly more pronounced in OBT2D cells than in Control cells (246.663.7% versus 223.062.2%, p,0.05).

Increased translocation of FAT/CD36 in Control cells after AMPK activation or insulin stimulation
To test whether translocation of FAT/CD36 in Control myotubes could be increased, cells were acutely treated with AMPK activators (AICAR and metformin (MET)) or insulin (INS), two stimuli known to induce the translocation of FAT/ CD36 to the plasma membrane of muscle tissue [7] (Fig. 3 A). Acute INS, AICAR or MET treatments induced FAT/CD36 cell surface localization in Control myotubes. We also overexpressed a constitutively activated form of AMPKalpha 2 subunit [19] in Control cells. This constitutively activated form of AMPK alpha2 was well expressed and phosphorylated (Fig. 3 B). FAT/CD36 translocation (in red) in Control cells was observed after AMPK alpha2 overexpression (Fig. 3 C) whereas FAT/CD36 was not localized to the plasma membrane after infection with the empty adenovirus (GFP).

AMPK activation by AICAR induced increased betaoxidation and no change in lipid accumulation in Control cells
To determine whether increased translocation of FAT/CD36 in Control myotubes was associated with increased lipid accumulation, we tested the effect of acute AICAR stimulation alone or in combination with specific FAT/CD36 inhibition (SSO) on lipid accumulation. SSO treatment did not significantly reduce lipid accumulation in Control cells and AICAR treatment did not induce increased lipid accumulation (Fig. 4 A). We also tested the effect of expressing the constitutively activated form of AMPK alpha 2 on lipid accumulation (Fig. 4 B) and found that Control myotubes overexpressing AMPK alpha 2 did not show increased positive oil red O cells. Lipid accumulation in myotubes is the result of the balance between FA uptake, FA esterification and FA oxidation. Since AMPK activation is known to induce increased  mitochondrial beta oxidation [20] we measured beta oxidation before and after AICAR stimulation in Control cells. As shown on figure 4 C, beta-oxidation was significantly increased by acute AICAR treatment (p,0.05).

FAT/CD36 is not associated with lipid rafts in OBT2D myotubes
In order to gain insight into mechanisms of FAT/CD36 localization in OBT2D myotubes, we examined intracellular staining of FAT/CD36 (after fixation and permeabilization) in Control and OBT2D muscle cells. Intracellular FAT/CD36 appeared similar between Control and OBT2D cells when using the two antibodies directed against FAT/CD36, H300 (Fig. 5 A, left panel) and CD36-alexa488 (Fig. 5 A, right panel). Moreover, intracellular FAT/CD36 was observed in both differentiated cells (troponin T positive) and reserve cells (arrows). In adipocytes, FAT/CD36 has been found in lipid rafts and caveolae [21], suggesting that lipid rafts regulate the expression and function of FAT/CD36 at the level of the plasma membrane. Lipid rafts are composed of DRMs (detergent resistant-membranes). Isolation of DRMs was performed in order to compare lipid rafts between Control and OBT2D differentiated cells. Caveolin 3 expression was used to monitor DRMs purification and the molecular chaperone binding protein (BIP) as an endoplasmic reticulum marker excluded from DRMs. As shown on Figure 5 B, caveolin 3 expression was variable between cells (Control 1, Control 2, OBT2D 1 and OBT2D 2) but was mainly found in fractions 3 and/or 4 (identified as DRMs) in Control and in OBT2D differentiated cells. Extraction with Triton X-100 yielded good separation of lipid rafts from BIP, as it was not detected in DRM fractions. The association of FAT/CD36 with lipid rafts was then analyzed. Surprisingly, FAT/CD36 expression was only detected  in fractions 9 and 10 in both Control and OBT2D cells (Fig. 5 C) but not in DRMs fractions showing that FAT/CD36 is not associated with lipid rafts in primary human myotubes.

Increased FAT/CD36 cycling in OBT2D myotubes
In order to monitor the dynamic of FAT/CD36 cycling, we performed immunofluorescence staining of FAT/CD36 in living cells by incubating CD36-alexa488 antibody for different periods of time (15 min, 1 h and 16 h). Incubation of living Control cells with CD36-alexa 488 antibody only showed important cell surface FAT/CD36 staining after 16 h (Fig. 6 A, top panels). Conversely, cell surface FAT/CD36 was already detected in OBT2D myotubes after 15 min of incubation and was independent of incubation time (Fig. 6 A, bottom panels). We incubated CD36-alexa488 at a lower temperature (room temperature, 22uC) to reduce the rate of FAT/CD36 cycling in OBT2D myotubes. As shown on Figure 6 B, cell surface FAT/CD36 was no longer detected in OBT2D myotubes after 15 min of incubation with the antibody at 22u C. To test the hypothesis that factors released by OBT2D myotubes in the medium were responsible for FAT/ CD36 cycling, we tested supernatant of OBT2D differentiated cells on Control differentiated cells (Fig. 6 C). Interestingly, supernatants from two different OBT2D differentiated cells (OBT2D1 and OBT2D2) increased cell surface FAT/CD36 in two different Control differentiated cells (Control 1 and Control 2).

Primary human muscle cells represent a good model for studying molecular mechanisms involved in type 2 diabetes
Primary human satellite cells have been shown to display the majority of the defects previously described for type 2 diabetic muscle in vivo including defective insulin signaling pathway [10,11], reduced insulin-stimulated glycogen synthase and glucose transport activities [22,23], and reduced lipid oxidation [24,25]. We have previously shown an abnormal cell surface FAT/CD36 localization [5] in cells derived from obese patients and an abnormal metabolic flexibility in response to high palmitate concentrations in myotubes derived from obese type 2 diabetic patients [15]. In the present study we have obtained new interesting results, which confirm that this cellular model can be used to study mechanisms related to type 2 diabetes in human skeletal muscle. By performing fractionation experiments, we were able to show biochemically that using CD36-alexa 488 antibody in living cells is a simple and powerful tool to follow cell surface localization of FAT/CD36. Moreover, this tool can be used to test molecules involved in FAT/CD36 translocation since we have shown that signalling pathways (insulin and AMPK) involved in FAT/CD36 translocation in the skeletal muscle [6,7] were also able to induce endogenous FAT/CD36 translocation in Control cells. A permanent relocation of FAT/CD36 was observed in the skeletal muscle tissue of obese and T2D patients [4,5]. The term permanent relocation was used since the increase in membraneassociated FAT/CD36 [4,26] was accompanied by a decrease in intracellular (vesicle-bound) FAT/CD36 and not by an increase in total FAT/CD36 expression. In our study, we have shown that total FAT/CD36 expression was not increased in OBT2D cells. Cell surface FAT/CD36 was dependent upon duration of antibody incubation. Control cells presented a slow cycling rate of FAT/CD36 and undetectable FAT/CD36 protein expression in plasma membrane fractions under basal conditions compared to OBT2D myotubes. Moreover decreasing the temperature at which the experiment was performed decreased cell surface FAT/ CD36 in OBT2D myotubes consistent with a continuous cycling of FAT/CD36. As such, the increase in sarcolemmal FAT/CD36 observed in the skeletal muscle of type 2 diabetic patients may result from a continuous cycling of FAT/CD36 suggesting that defects in both, endocytosis and exocytosis may be expected. FAT/CD36 cycling was specific to myotubes and was not observed in reserve cells despite the presence of intracellular FAT/CD36. As such, we may suppose that FAT/CD36 cycling is controlled by a system specific to differentiated cells. Moreover, cell surface FAT/CD36 was observed in Control cells by incubating these cells with supernatants from OBT2D differentiated cells. This result implies that the appearance of cell surface FAT/CD36 is dependent upon a secreted factor or a combination of secreted factors or that a secreted factor inhibiting FAT/CD36 cycling is missing in differentiated OBT2D cell supernatant. Recent papers [27,28] have shown by using proteomic approaches, that many proteins were secreted from human and mouse skeletal muscle cells and that these secreted proteins were modified in response to strength training or to TNF-alpha stimulation. Moreover, skeletal muscle secretome is dynamic during differentiation [29]. As such proteomic experiments should be performed in differentiated Control and OBT2D myotubes in order to identify secreted proteins which could be responsible for increased FAT/CD36 cycling during type 2 diabetes.

The importance of FAT/CD36 cycling in determining lipid accumulation is different between OBT2D and Control myotubes
In adipocytes, FAT/CD36 expression appeared to be closely linked to preadipocyte differentiation [9]. Furthermore, FAT/ CD36-mediated long-chain fatty acid uptake in adipocytes has been shown to require plasma membrane rafts [21]. In the present study, we showed that this is not the case in primary human muscle cells. As such, expression and localization of FAT/CD36 are differentially regulated during primary human muscle cell differentiation and murine preadipocyte differentiation. The fact that the proteins involved in fatty acids transport would differ in adipocytes and muscle tissue is not surprising given their different metabolic roles and physiological functions [30]. The precise localization of cell surface FAT/CD36 in relation to other known membrane proteins in primary human myotubes remains to be determined.
Fractionation experiments showed that only a minor part of FAT/CD36 was present in plasma membrane fractions implying that under basal condition, FAT/CD36 is mainly localized in intracellular compartments in OBT2D myotubes. Palmitateinduced lipid accumulation was observed in OBT2D differentiated cells with a concomitant cell surface FAT/CD36 localization. Moreover, the increase in palmitate-induced lipid accumulation observed in OBT2D myotubes was due to an increase in transporter-mediated lipid entrance (phloretin inhibition) mainly mediated by FAT/CD36 (SSO inhibition). Altogether these results show that cell surface localization of FAT/CD36 in fully differentiated OBT2D cells allows lipid accumulation. However, the more pronounced inhibition of lipid accumulation with phloretin (by 78%) compared to SSO treatment (by 47%) suggests that other lipid transporters may also be involved. This is consistent with the notion that membrane fatty acid binding protein contributes to fatty acid transport with FAT/CD36 [31,32].
Acute AICAR or metformin treatment induces FAT/CD36 translocation in Control myotubes. This result was similar when using an adenovirus overexpressing a constitutively active form of AMPK, suggesting that FAT/CD36 translocation in response to AICAR or metformin treatment was due to AMPK activation. We hypothesized that similarly to OBT2D myotubes, increased FAT/ CD36 translocation in Control myotubes would lead to increased lipid accumulation. However, inducing FAT/CD36 translocation in Control cells had no effect on lipid accumulation since AMPK activation in Control cells was able to significantly increase mitochondrial beta-oxidation. As such, increased translocation of FAT/CD36 in Control cells is not sufficient to induce lipid accumulation. We know from a previous work [15] that cells derived from obese type 2 diabetic patients display an impaired capacity to respond to metabolic stimuli at the level of mitochondrial activity, which could explain the difference in lipid content between OBT2D and Control cells even when control cells showed a similar cell surface FAT/CD36 content than OBT2D cells. Moreover, protein-mediated lipid transport was not the major mechanism responsible for palmitate-induced lipid accumulation in Control myotubes (less than 50% of inhibition by phloretin).

Conclusion
This study is the first to show that increased lipid accumulation in OBT2D myotubes results from increased FAT/CD36 cycling.
We propose that lipid content in Control myotubes may mostly be influenced by catabolic pathways (lipolysis, fatty acid oxidation) than anabolic pathways (uptake, esterification) in contrast to myotubes derived from obese type 2 diabetic patients.

Materials and Methods
Primary human satellite cell culture Skeletal muscle biopsy of the vastus lateralis was performed according to the percutaneous Bergström technique after local anesthesia (xylocaine) [33,34]. The experimental protocol was approved by ''Le Comité de Protection de Personnes (CPP) «Sud Méditerranée IV»» (03/10/GESE, Montpellier, France). Informed and written consent was obtained from all subjects after explanation of the protocol. In the present study, biopsies were taken from four control subjects (control, age: 49.262.7 years, body mass index: 24.061.2 kg/m 2 ) with no familial or personal history of diabetes and from five obese type 2 diabetic patients (OBT2D, age: 53.462.4 years, body mass index: 31.560.6 kg/ m 2 ). Clinical characteristics of the subjects were previously described [5]. The number of cell cultures used in this study is not high, however these cells were isolated from homogeneous populations of male subjects matched by age and physical activity [5]. Type 2 diabetic patients are obese since we have previously shown that cells derived from non obese type 2 diabetic patients did not show increased lipid accumulation compared to body mass index matched healthy subjects [5]. Cell culture of primary human satellite cells was performed as previously described [5,12,15,35], cultures were maintained in a growth medium composed of (DMEM, 10% fetal bovine serum (FBS) and 1% Ultroser G) and when myoblasts reached confluence, medium was changed to differentiation medium (growth medium minus Ultroser G) and the differentiation process occurred until fusion and terminal differentiation into contractile myotubes (8 days). The experiments were performed on passages 2 to 4.
FAT/CD36 has an extracellular loop allowing performing immunofluorescence in living cells by adding the antibody directly to the culture medium (FAT/CD36 staining in living cells can only be detected when the transporter is at the plasma membrane). We have used primary antibodies (SM) already coupled to fluorochrome (alexa 488 or PhytoErythrine) to avoid using a secondary antibody. Briefly, cells were washed with PBS and anti-CD36 Alexa Fluor 488 (SM) antibody was added at a dilution of 1/500 in PBS 2% BSA for 15 min, 1 h or 16 h either at 37uC or at room temperature (22uC) prior to paraformaldehyde fixation. For co-staining with GFP-expressing adenovirus, anti-CD36 PhytoErytrhine (CD36-PE, SM, Tebu-BIO) diluted at 1/500 was used instead of anti-CD36 Alexa fluor 488 (CD36-Alexa 488, SM, Tebu-BIO) (Fig. 3C).
For Figure 6C, medium from two independent differentiated Control (Control 1 and Control 2) cell cultures (day 7) was replaced by 1 ml of supernatant from two independent differentiated OBT2D (OBT2D 1 and OBT2D 2) cell cultures (Day 8) for 16 hours prior to staining on living cells with CD36 alexa 488.

Virus infection
Adenovirus overexpressing a constitutively active truncated (40 kDa) form of AMPKa2 was previously described [19]. Infection of myotubes with either the empty adenovirus (GFP) or the adenovirus overexpressing the constitutively active form of AMPK alpha 2 (alpha 2) was performed on Day 6, 48 hours prior to the end of myotube differentiation.

Fatty acid preparation and cells treatments
Stock solution (0.6 M) of palmitate was prepared. Palmitate was dissolved in chloroform. Samples were stored at 220uC until required. Fatty acid stocks were diluted to 0.6 mM in DMEM containing 10% FBS (differentiation medium) and added to the differentiated satellite cells. Control satellite cells were incubated in DMEM containing 10% FBS and 0,001% chloroform without palmitate as previously described [5,12,15]. Treatments performed on differentiated satellite cells are detailed in the figure legends and were realized in triplicate for each of the 9 independent cell cultures. The following reagents were purchased from Sigma (Saint Quentin Fallavier, France): L-Glutamine, DMEM, palmitate, phloretin, AMP-mimetic 5-aminoimidazole-4-carboxamide-1-b-D-ribofuranoside (AICAR), metformin and insulin. FBS was purchased from Hyclone (Brebieres, France) and Sulfo-N-succinimidyloleate (SSO) was a kind gift from W. Coumans (Maastricht, The Netherland).

Oil red O staining
Oil red O staining was carried for one hour followed by 2 or 3 washes with distilled water. Lipid droplets were then visualized by light microscopy. In order to quantify lipid accumulation in myotubes, oil red O was extracted using isopropanol. The absorbance value was measured using a spectrophotometer set at 490 nm [41] and blanked to untreated cells. Staining and quantification of lipid accumulation in primary human satellite cells were previously described [5,12,15]. For Figure 2 D, conventional oil red O was modified according to the previously described protocol [42].

Plasma membrane and post-plasma membrane fractions
One million of differentiated satellite cells derived from control subjects (Control) and type 2 diabetic patients (OBT2D) were used to prepare cell lysate, plasma membrane (Mb) and post-plasma membrane fractions (P) as previously described [43].

Isolation of Detergent-Resistant Membranes (DRMs)
DRMs were prepared by detergent extraction of cellular proteins and centrifugation over discontinuous sucrose gradients essentially as described [44]. 4.5610 6 of differentiated satellite cells derived from control subjects (Control) and from obese type 2 diabetic patients (OBT2D) were used to prepare the DRMs. Cells were washed twice with ice-cold PBS and scraped into 667 ml of MBST buffer (25 mM MES, 150 mM NaCl, 1% Triton X-100, protease inhibitors mixture (Roche, Rosny-sous-bois, France) and homogenized with a loose fitting dunce homogenizer (20 strokes). The extract was adjusted to 40% sucrose by the addition of 667 ml of 80% sucrose in MBS lacking triton X-100 and placed in the bottom of an ultracentrifuge tube. A discontinuous sucrose gradient was formed by overlaying this solution with 1.33 ml of 38% sucrose and 1.33 ml of 5% sucrose (both in MBS). The tubes were centrifuged at 40,000 rpm in an SW60 Ti rotor for 16 h at 4uC, and 10 fractions (numbered 1 to 10) of 400 ml were collected manually from the top of the gradient, resuspended in Laemmli buffer and subjected to SDS-PAGE on 10% acrylamide gels followed by Western blot analyses.

Palmitate oxidation
Control cells were cultured in 96-well plates and differentiated. Differentiated satellite cells were exposed or not to cold palmitate (0.6 mM) for the last 16 h of differentiation and then exposed to differentiation medium (DMEM+10% FBS) supplemented with 1% BSA, 50 mM palmitate and 9.5 mM (0,09 mCi) [1-14 C]palmitate. Identical incubations were conducted on parallel plates that contained no cells. Palmitate oxidation rates were determined by measuring production of 14 C-labeled acid-soluble metabolites (ASM), a measure of tricarboxylic acid cycle intermediates and acetyl esters. After incubation for 30 min at 37uC, reactions were terminated by aspiration of the media and addition of 100 ml of HClO 4 at 5% for 15 min at room temperature. The ASM were assayed in supernatants of the acid precipitate. Radioactivity of ASM was determined by liquid scintillation counting by use of 4.5 ml of liquid scintillation cocktail (Optiphase 'Hisafe' 3, Perkin Elmer) in scintillation vials. For protein determination, identical incubations were conducted on parallel plates with the same number of cells.

Statistical analysis
Statistical analyses were performed using Statview 5.0. Data are shown as mean 6 SEM. Statistical analyses were performed using Student's t test for unpaired and paired comparison, or ANOVA with Fisher's PLSD post-hoc test for multiple comparisons. P,0.05 was considered to be significant.