Human muscle biopsies were obtained from 10 ALS patients referred from the Paris Motor Neuron Disease Center (Pitié-Salpêtrière Hospital, Paris, France). All patients met the El Escorial World Federation of Neurology criteria for the diagnosis of definite, probable or possible ALS (Brooks et al., 2000). Muscle specimens were taken from the anconeus muscle for three patients and the deltoid muscle for seven patients. Six subjects with histochemically normal deltoid muscle were used as controls. These subjects had undergone muscle biopsy as a diagnostic procedure and, after careful clinical and histological evaluation, had been considered free of any neuromuscular disorder. The study was approved by the local ethics committee, and all patients provided written informed consent before muscle biopsy, according to institutional guidelines. Biopsy specimens were immediately frozen in liquid nitrogen, and stored at -80°C until use.

The care and treatment of animals followed the French national authority guidelines for the detention, use and the ethical treatment of laboratory animals. All experimental procedures that include minimizing the number of animals used and their suffering were approved by the policies of the French Agriculture and Forestry Ministry. All the experiments using mice were performed in a blind systematic fashion so as to minimize bias.

Transgenic male B6/SJL mice with the G93A human SOD1 mutation [B6/SJL-Tg (SOD1-G93A) 1Gur/J] (ALS mice) were purchased from The Jackson Laboratory (Bar Harbor, ME, United States). ALS male were crossed with wild-type B6/SJL females (Janvier, Le Genest-Saint-Isle, France), and only males were used for this study.

Amyotrophic lateral sclerosis males were trained from 70 days of age (P70), 30 min a day, 5 days a week, to a running-based or swimming-based exercise as previously described (Grondard et al., 2008; Deforges et al., 2009) and until 115 days of age (P115). 18 ALS mice (Run ALS) were submitted to a running-based training on a speed-regulated treadmill (max 13 m.min^-1). Eighteen ALS mice (Swim ALS) were submitted to a swimming-based training in an adjustable-flow swimming pool (max 5 L.min^-1). These groups were compared to 29 Sedentary ALS and 20 Sedentary Control mice.

For the autophagy analysis, specific groups of ALS males at P70 were subjected to the same swimming-based exercise for 3 days, and treated for autophagy inhibition as previously reported (Kang et al., 2014). Accordingly, each mice was intraperitoneally injected 1 h before exercise with 10 μl.g^-1 of either chloroquine (CQ) at 5 μg.μl^-1, or 3-methyladenine (3-MA) at 1.5 μg.μl^-1, or a placebo NaCl 0.9% solution.

Glucose tolerance tests were performed on 4 or 12 h fasted mice at P70 and P115, with the same results. Mice were intraperitoneally injected with D-glucose (Sigma–Aldrich, St. Louis, MO, United States) at a dose of 2 g.kg^-1. Glucose concentration was then measured every 15 min using a Glucotrend R analyser (Boehringer Mannheim GmbH, Germany) from whole blood samples taken from the tail.

Blood lactate samples were collected from right ventricle on anesthetized mice, with 1% pentobarbital solution (6 μl.g^-1) diluted in 0.9% saline buffer, using a 1 ml syringe mounted with a 22-gauge needle and coated with heparin (5000 UI.ml^-1, PanPharma Luitré, France), 2 h after training. Blood samples were centrifuged (1000 g, 10min, +4°C) and serum was frozen at -80°C. The serum lactate level was measured using Lactate assay kit (Biovision research, Mountain view, CA, United States) following manufacturer’s instructions. For each mouse, 1 μl of serum were incubated at +22°C in 50 μl of incubation buffer and 50 μl of reaction mix in a 96-well plate, and read on spectrophotometer at 570 nm wavelength after 30 min of incubation. Lactate levels were given nmol.μl^-1.

Sample preparation was made using the classic protocol of tissue extraction (Beckonert et al., 2007). The combined extraction of polar and lipophilic metabolites from tissues was obtained using the methanol/chloroform/water mixture.

Muscle mass ranged between 15 and 40 mg. The 13 polar sample extracts were transferred in 5 mm nuclear magnetic resonance (NMR) tubes to obtain a final volume of 580 μL/samples.

Mouse tissues were collected 2 h after training in liquid nitrogen and RNA was extracted using TRizol reagent (Invitrogen, Life Technologies, Saint-Aubin, France). 1 μg of human or mouse RNAs was reverse transcribed with oligodT using reverse transcriptase Improm II (Promega France, Charbonnières, France). Quantitative real time PCR was performed with standard protocols using SYRB Green ROX detector in ABIPrism 7000 (ABgene, Courtaboeuf, France). Specific primers were used at 300 nM (Table 1). The relative amounts of cDNA in each sample were determined on the basis of the threshold cycle (Ct) for each PCR product and normalized to either RPL13 or Rps26 Ct for human or mouse samples, respectively. These housekeeping genes has been determined as best internal controls using Bestkeeper (Pfaffl et al., 2004) and Normfinder (Andersen et al., 2004) algorithms on 10 control and 10 ALS tibialis and over 10 different housekeeping genes (data not shown). The analysis was done relative to respective control samples and given by 2^-ΔΔC_T.

Muscles from P115 mice were frozen in cold isopentane (-80°C) and transversally cut in 10 μm sections using cryostat (CM 3050S, Leica, Le Pecq France). After an incubation in a blocking solution [Tris Buffer Saline (TBS), 0.1% tween, 0.1% triton and 4% donkey serum] for 30 min at room temperature, sections were incubated with primary polyclonal rabbit anti-GLUT4 antibody (1:50; GT41-A Alpha Diagnostic, San Antonio, TX, United States), overnight at +4°C. Sections were then washed and incubated with secondary polyclonal goat anti-rabbit Cy3-conjugated antibody (1:400; 111-165-144, Jackson laboratories, Baltimore, MD, United States) for 2 h at room temperature. Sections were washed and mounted in VECTASHIELD^® mounting medium (Vector Laboratories, Burlingame, CA, United States). The staining specificity was checked in control incubations performed in the absence of the primary antibody.

Epifluorescence was detected with a CMOS camera (ORCA Flash 2.8, Hamamatsu Photonics France, Massy, France) and histological images were recorded with a RGB camera (ICC1 Zeiss, Carl Zeiss SAS, Le Pecq, France) mounted on Zeiss AxioObserver (Z1) using the ZEN 2012 software (Carl Zeiss SAS) with 100 (10× Zeiss Plan NeoFluar NA 0.3) and 200 (20× Zeiss EC-Plan-Apo NA 0.8) magnifications.

All images were analyzed using Image J v1.47 software (National Institutes of Health, Bethesda, MD, United States). Identical brightness, contrast, and color balance adjustments were applied to all groups. GLUT4 staining was counted in positive-fibers and characterized as full membrane-stained or not.

Mouse muscles were dissected and frozen in liquid nitrogen and homogenized in 50 μl/5 mg of RIPA buffer [50 mM Tris/pH 8.3, 150 mM NaCl, 0.1% SDS, 1% NP40, 10 mM NAF, 1X Protease inhibitor (Roche, BASEL, Switzerland), 1% Phosphatase inhibitor (Sigma–Aldrich, St. Louis, MO, United States)]. Protein concentration was determined in the supernatant after centrifugation (17000 g, 15 min, +4°C) by Bradford protein assay (Bio-Rad, Hercules, CA, United States). Protein samples (100 μg for LC3B, 40 μg for the other proteins) were submitted to 12.5% SDS PAGE electrophoresis (1.5 M Tris pH 8.3, 12.5% Acrylamide, 0.07% Bis, 0.1% SDS, 0.05% APS, 0.06% TEMED). The separated proteins were transferred on PVDF membranes (Bio-Rad) and incubated overnight at +4°C with either rabbit anti-PDHE1α P-Ser293 (1:1000; AP1062; Merck Millipore, Darmstadt, Germany), rabbit anti-PDHE1α (1:1000; ab110334, Abcam, Cambridge, United Kingdom), rabbit anti-LC3B (1:1000; NB100-2331, Novus Biological), mouse anti-GAPDH (1:10.000; MAB374, Millipore) or mouse anti-α-tubulin (1:20.000; T6074, Sigma) diluted in blocking solution. Membranes were washed and incubated in horseradish peroxidase-conjugated secondary antibody goat anti-rabbit or goat anti-mouse (1:10.000 Jackson ImmunoResearch). Revelation was performed with Amersham^TM ECL^TM Western Blotting Analysis System (GE Healthcare, Bio-Science, Upsala, Sweden). Images were done using ImageQuant LS4000 (GE Healthcare Bio-Science, Upsala, Sweden) and quantification was performed using Image J v1.47 software.

The triacylglycerols (TAG) of the Bligh and Dyer extract (Bligh and Dyer, 1959) were prepared from homogenized muscles in water and separated on a normal phase PVA-Sil column (Polymerised Vinyl Alcohol silica (5 μ) support; L 250 mm X ID 4 mm) (YMC, Kyoto, Japan). The PVA-Sil column was fitted on a Agilent 1200 HPLC equipment coupled to the electrospray ionization (ESI) source of a triple-quadrupole mass spectrometer (API3000, AB Sciex). TAG species were quantified by integration of molecular ions (as ammonium ion adducts) from full scan mass spectra and application of standard curves that relate the responses of known amounts of reference standards to that for a single internal standard as described elsewhere (Hutchins et al., 2008).

All data are presented as mean and standard deviation (SD). For the glucose tolerance studies, a one-way ANOVA for repeated measures with Tukey post hoc analysis was performed on the data. For the other studies, a Kolmogorov–Smirnov normal distribution analysis was performed on all data followed by either a student’s t-test for normally distributed data or a non-parametric Kruskal–Wallis test, to verify significant differences between groups (Systat v 8.0, SPSS Inc., Chicago, IL, United States). All the data presented in this study were considered as statically different when the statistical power exceeds 95% (AnaStats.fr, France). All graphics were done with GraphPad Prism v6.01 and Adobe Illustrator CS6 v16.0.3.

The ALS-induced loss of body mass detected from the asymptomatic phase of the disease in several mouse models of ALS (Dupuis et al., 2004b) is differentially altered by two different physical exercises (Deforges et al., 2009). Indeed, unlike running, the swimming-based training significantly limited the body weight loss of ALS mice (Deforges et al., 2009). Then, we questioned whether this preservation was linked to an intrinsic change in energy metabolism. To address this question, we assessed glucose tolerance in sedentary controls and sedentary and trained ALS mice in the presymptomatic (P70) and late symptomatic (P115) phases of the disease. The results of an intraperitoneal glucose tolerance test revealed that, from P70 until death, sedentary ALS mice displayed a severely impaired glucose tolerance. The blood glucose levels remained over 16 mmol.L^-1 1 h after a 11.1 mmol.kg^-1 glucose injection (2 g.Kg^-1), whereas in wild-type mice whole blood glucose concentration was closed to baseline levels at that time (Figures 1A,B). Since glucose intolerance was recorded for the first time in SOD1(G93A) ALS mice, we decided to verify and confirm these results by quantifying glycogen and glucose concentration levels using nuclear magnetic resonance (NMR) in the skeletal muscle of SOD1(G93A) ALS and control mice. We focused our study on the tibialis muscle, which is a fast-twitch muscle, mainly using glucose as fuel for energetic supply. Importantly, our NMR results revealed a significant decrease in glycogen concentration levels in ALS vs. control tibialis (Figure 1C). The concentration levels of free glucose converged toward a global decrease in the ALS tibialis compared to controls, although we found a high heterogeneity among samples, which is normally expected with a highly metabolized molecule (Figure 1D). All these data concurred with a glucose intolerance in SOD1(G93A) ALS mice and further substantiated a defect in glucose uptake.

We next recorded the effects of each training paradigm on glucose tolerance in SOD1(G93A) ALS mice. Very interestingly, the glucose tolerance in SOD1(G93A) ALS mice was significantly improved following specific training. The swimming-based trained in ALS mice completely restored the glucose tolerance at P115 (Figure 1E) while the running-based slightly improved it after 45 days of training (Figure 1F).

We next evaluated the effect of the swimming-induced increase in glucose tolerance on its subsequent use as an energetic substrate in SOD1(G93A) ALS mouse tissues. To this end, we measured the levels of circulating lactate in sedentary control and in sedentary and trained SOD1(G93A) ALS mice at P115 in resting conditions (Figure 1G). We found as expected (Ferraiuolo et al., 2011) a significant decrease of circulating lactate levels in sedentary SOD1(G93A) ALS mice, in comparison to control mice. Interestingly, unlike running, the swimming-based training induced a significant increase in the levels of circulating lactate in SOD1(G93A) ALS mice, suggesting a specific-exercise-induced activation of glycolysis in skeletal muscles.

All these results converge on the idea that glucose resistance in SOD1(G93A) ALS mice, also found in ALS patients, impacts muscle metabolism, as suggested by the abnormal decrease in circulating lactate. Interestingly, only the swimming-based training managed to limit this metabolic defect.

We next investigated the molecular mechanisms underlying, in the one hand, glucose intolerance in SOD1(G93A) ALS mice and, on the other, specific exercise-induced potential restoration of glucose uptake by ALS mouse tissues, including skeletal muscles. Among glucose transporters, GLUT4 is the most abundant in skeletal muscle and its level of expression is a major determinant of glucose uptake by skeletal muscle fibers (Richter and Hargreaves, 2013). We analyzed GLUT4 expression in skeletal muscles of SOD1(G93A) ALS mice, particularly the fast-twitch tibialis and the slow-twitch soleus, affected earlier and later by the disease, respectively (Hegedus et al., 2008). We found a significant decrease of Glut4 mRNA expression in the two muscles in ALS mice, in the late symptomatic phase of the disease (P115) (Figures 2A,B). Interestingly, physical exercise enhanced Glut4 expression in ALS muscles, but with exercise and muscle specificities. Indeed, only the swimming-based training was able to significantly increased Glut4 expression and specifically in the ALS tibialis (Figures 2A,B).

Since muscle glucose uptake is not only dependent on GLUT4 expression levels, but also of its abundance at the myofiber sarcolemma level, we analyzed the sub-cellular distribution of GLUT4 in the tibialis and soleus of SOD1(G93A) mice. We found that the GLUT4 subcellular distribution was largely intracellular, throughout the cytosol, and mostly absent at the sarcolemma level in ALS tibialis muscles, compared to controls (Figures 2C,D). By contrast, no difference in GLUT4 subcellular distribution was observed in the ALS soleus compared to controls (Figure 2E). Since exercise is thought to induce GLUT4 mobilization to the sarcolemma (Douen et al., 1990; Lauritzen et al., 2010), we next investigated whether, depending of its type, exercise training could restore GLUT4 subcellular distribution in the ALS muscles. In contrast to the running-based training, the swimming-based training induced a sub-cellular redistribution of the transporter in ALS tibialis as evidenced by both the increase in GLUT4 staining at the cell periphery and the decrease in GLUT4 staining throughout the cytoplasm of the muscle cells (Figures 2C,D). In the ALS soleus, both exercise types had no effect on GLUT4 subcellular distribution (Figure 2E).

Since these data substantiated a defect in glucose transport in myofibers in SOD1(G93A) ALS mice, we evaluated GLUT4 expression in human samples from ALS patients by RT-qPCR. Consistent with ALS mouse data, we found a significant decrease in GLUT4 mRNA expression in ALS human muscles compared to controls (Figure 2F).

Taken together, these results suggested an alteration of GLUT4 expression, the major glucose transporter in the skeletal muscle, specifically in the fast-twitch muscles of SOD1(G93A) ALS mice. More importantly, the GLUT4 expression was restored by the swimming-based training, suggesting a link between GLUT4 expression defects and the glucose tolerance status in sedentary and trained ALS mice.

In order to determine if the potential defects in glucose uptake, suggested by (1) a glucose intolerance, (2) a decrease in lactate output and (3) defects in muscular GLUT4 expression, would impact glycolysis in ALS muscles, we analyzed the expression pattern of the glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Indeed, GAPDH has long been recognized as a pivotal enzyme for the production of ATP and pyruvate from glucose through glycolysis. We found that, in both tibialis and soleus muscles of SOD1(G93A) ALS mice, the levels of Gapdh mRNA were significantly decreased in the late symptomatic phase of the disease (P115), with a more pronounced decrease in tibialis than in soleus when compared to the controls (Figures 3A,B). In contrast with soleus, we recorded a significant decrease in GAPDH protein in tibialis muscle consistent with mRNA expression level (Figures 3C,D and Supplementary Figures S1A,B).

We next investigated whether exercise training could improve GAPDH expression in ALS muscles. In the ALS tibialis, both exercise paradigms resulted in a significant increase in GAPDH mRNA and protein levels with the strongest efficiency observed with the swimming-based training (Figures 3A,C). In the ALS soleus, no physical training effect could be recorded on GAPDH mRNA and protein levels (Figures 3B,D).

Importantly, we confirmed these data in human muscle samples, where we found a significant decrease in GAPDH mRNA expression in ALS patients (Figure 3E).

Taken together, these data suggested that glycolysis is altered in ALS skeletal muscles and that high-intensity exercise could efficiently re-balanced the enzymatic equipment in ALS muscle, leading to the improvement in anaerobic capacity of fast-twitch muscles.

Pyruvate, a glycolysis by-product, can be metabolized in two different ways in muscles, i.e., the anaerobic pathway that will result in the production of lactate, or the aerobic pathway that will result in the production of acetate. Consequently, we investigated the regulation of the PDH complex, which is pivotal in determining the partitioning of anaerobic compared with aerobic use of pyruvate produced by the glycolysis. The PDH E1-alpha subunit is considered as the on/off switch of the PDH complex, since its phosphorylation by the PDH kinase 4 (PDK4) in the muscle leads to PDH complex inactivation (Wu et al., 1999). Importantly, the alteration of PDK4 expression has been recently highlighted in ALS skeletal muscle (Palamiuc et al., 2015). Therefore, we analyzed the level of PDH E1-alpha phosphorylation in tibialis and soleus of SOD1(G93A) ALS mice. The level of PDH E1-alpha phosphorylation was significantly higher in the ALS tibialis compared to controls (Figure 4A and Supplementary Figure S2A), while no significant change was found in the soleus (Figure 4B and Supplementary Figure S2B). We next evaluated the level of Pdk4 mRNA expression by RT-qPCR. As previously found in the SOD1(G86R) ALS mouse model (Palamiuc et al., 2015), and consistently with higher PDH phosphorylation levels, ALS tibialis displayed significantly higher Pdk4 expression levels compared to control muscles (Figure 4C), whereas no significant change was found in the soleus (Figure 4D). Interestingly, the effects of physical exercise in ALS tibialis altered the PDH phosphorylation status in an exercise-type dependent manner. Indeed, the running-based training induced a significant decrease in PDH E1-alpha phosphorylation and Pdk4 mRNA expression levels in ALS tibialis when compared to sedentary ALS mice (Figures 4A,C). By contrast, the swimming-based training had no effect on PDH activation pattern, nor on Pdk4 mRNA expression levels (Figures 4A,C), strongly suggesting that all the swimming-induced changes in ALS tibialis unexpectedly occurred independently of PDH modulation. In the soleus, only the running-based training induced a significant decrease of Pdk4 mRNA expression, without a significant change in PDH E1-alpha phosphorylation pattern (Figures 4B,D).

We confirmed these findings in human samples from ALS patients. Indeed, we found an increase in PDK4 mRNA expression ALS when compared to control muscle specimens (Figure 4E), as previously reported (Palamiuc et al., 2015).

Taken together, these data indicated that the mechanisms underlying both the ALS-induced alteration of glucose uptake and glycolysis and the beneficial effects of the swimming-based training on GLUT4 and GAPDH expression were independent of a regulation at the PDH level.

We next investigated whether autophagy, essential for the maintenance of cellular homeostasis and altered in ALS muscles (Dobrowolny et al., 2008; Onesto et al., 2011; Crippa et al., 2013; Xiao et al., 2015), could be involved in the swimming-induced benefits for carbohydrate metabolism in ALS muscles. Among the different systems of autophagy, the Bcl2/Beclin1 (BECN1)-dependent autophagic pathway seems to play an important role in the exercise-induced muscle adaptation of the glucose metabolism in healthy conditions, including GLUT4 relocalization to the sarcolemma and subsequent increase of glucose intake (He et al., 2012). These data prompted us to evaluate the status of this autophagic pathway in tibialis and soleus of SOD1(G93A) ALS mice. We first evaluated the expression levels of Bcl2, Becn1, Map1lc3b (Lc3b), and Sqstm1 (P62) by RT-qPCR. In sedentary ALS tibialis, Becn1, Map1lc3b (Lc3b), and Sqstm1 (P62) mRNA were significantly down-expressed when compared to controls, whereas no significant change was found for Bcl2 expression (Figure 5A). In the soleus of sedentary ALS mice, Bcl2, Becn1, and Map1lc3b (Lc3b) mRNA expressions were down-regulated, with no significant change for Sqstm1 (P62) (Figure 5B). To complete these results at the protein level, we analyzed LC3B levels by western blot in ALS tibialis and soleus. Contrasting to the mRNA expression data, we found an accumulation of the two forms of LC3B, LC3B-I and LC3B-II, the lipidated version of LC3B-I involved in autophagosome formation, in ALS tibialis when compared to controls (Figure 5C and Supplementary Figure S3A). In the ALS soleus, we found an accumulation of LC3B-II expression levels, when compared to controls, while no statistical difference was observed for LC3B-I (Figure 5D and Supplementary Figure S3B).

When we looked at the ALS tibialis after mouse training, we found that, even if both exercise paradigms induced a significant decrease of LC3B-I and LC3B-II proteins compared to sedentary ALS muscles, only the swimming-based training induced a decrease resulting in LC3B protein levels comparable to controls (Figure 5C). This decrease at the protein levels was associated with an increase in Map1lc3b (Lc3b) mRNA expression levels in ALS tibialis (Figure 5A). This increase was also found for Becn1 and Sqstm1 (P62) mRNAs, with no effect for running (Figure 5A). In the ALS soleus, only swimming exercise restored Bcl2 mRNA expression (Figure 5B) while no significant effect was observed at the protein level of LC3B for both exercises (Figure 5D).

Taken together, these data strongly suggested that the swimming-based training only improves the expression of the main autophagic molecules in ALS muscles.

We next investigated whether the swimming-induced enhancement of GLUT4 and GAPDH expression in ALS muscles was dependent on autophagy. We submitted a population of P70 SOD1(G93A) ALS mice to a swimming-based training for 3 consecutive days with or without a blocking of lysosomal degradation by chloroquine (CQ) or the autophagosome formation by 3-methyladenine (3-MA). We observed that, in the early stages of the disease, the expression of Glut4 and Gapdh mRNA was already altered in the tibialis of ALS mice and that a 3-days swimming-based training was sufficient to reactivate Glut4 and Gapdh mRNA expression (Figures 6A,B). Interestingly, autophagy inhibition with CQ or 3-MA impaired the swimming-induced increase of Glut4 and Gapdh mRNA expression in the tibialis of trained ALS mice (Figures 6A,B). Like in late stages of the disease, we observed an alteration of GLUT4 subcellular localization in the tibialis myofibers of sedentary ALS mice at P70 (Figures 6C,D), with a significant decrease of the sarcolemma level localization compared to controls (Figures 6C,D). As expected, a 3-day swimming exercise was sufficient to induce a relocalization of GLUT4 to the cell periphery (Figures 6C,D). This swimming-induced GLUT4 relocalization was totally abolished by autophagy inhibition whatever the mode of inhibition (CQ or 3-MA) (Figures 6C,D).

Taken together, these data suggest that the benefits induced by an anaerobic exercise such as swimming for improving glucose uptake in fast-twitch ALS muscles, are linked to autophagy.

We finally asked whether the swimming-induced improvement of glucose metabolism in ALS tibialis could limit the lipid hypermetabolism found in ALS muscles (Dupuis et al., 2004b).

We first looked at the expression profile of the two fatty acid transporters CD36 (Cluster of Differentiation 36) and VLDLR (Very Low Density Lipoprotein Receptor) in the tibialis of controls and sedentary and trained ALS mice. We found that the levels of Vldlr expression were decreased in ALS but no change was observed for Cd36 expression (Figures 7A,B). As expected, the swimming-based training induced a significant increase in Cd36 and Vldlr expression levels, suggesting an improvement in lipid uptake. By contrast, the running-based training had no effect (Vldlr) or even induced a significant decrease (Cd36) in the expression of these fatty acid transporters (Figures 7A,B).

We next looked at the expression levels of several major genes involved in lipogenesis (Liu et al., 2009), including genes coding for the transcription factor Srebp-1c (Sterol Regulatory Element-Binding Protein 1c), Fas (Fatty Acid Synthase), Ucp3 (Uncoupling protein 3), and Acc1 (Acetyl-CoA Carboxylase 1). We found that the expression levels of Srebp-1c were profoundly lowered in ALS tibialis in comparison to controls (Figure 7C). The swimming-based training induced a significant reactivation of Srebp-1c expression in ALS tibialis, whereas no effects could be recorded with the running-based training (Figure 7C). Interestingly, the expression profile of Fas (Figure 7D) and Ucp3 (Figure 7E) paralleled those of Srebp-1c, with a decrease in sedentary ALS tibialis, and a significant increase in swimming-based trained muscles. The running-based training had no effect on the expression of Ucp3 but induced a significant increase in Fas. By contrast, Acc1 expression levels did not display significant alterations in its mRNA expression, in any of the experimental conditions (Figure 7F).

Taken together, these data suggest that lipid metabolism is improved by swimming exercise in the ALS tibialis.

These results, indicating that the expression of several lipogenic genes, i.e., Fas and Ucp3 (Liu et al., 2009) are enhanced by the swimming-based training, prompted us to evaluate the expression profile of diacylglycerol acyltransferase 1 (Dgat1), which increases muscle fat content, associated with PDH inhibition and Ucp3 overexpression (Liu et al., 2009). We found that the levels of Dgat1 mRNA expression were significantly lower in ALS tibialis, when compared to controls (Figure 8A). Importantly, the swimming-based training induced a significant increase in Dgat1 expression levels (Figure 8A) compared to sedentary ALS, suggesting that, in addition to favoring the uptake of fatty acids, the swimming-based training might favor the neo-synthesis of lipids. Compared to sedentary ALS, a significant increase in Dgat1 expression was also recorded in the running tibialis, albeit lower than in swimming (Figure 8A).

Finally, we quantified by HPLC the levels of triacylglyceride (TAG) in the sedentary and trained ALS tibialis and control muscles. As expected, we found a decrease in TAG quantity in the ALS tibialis when compared to controls (Figure 8B). Importantly, the swimming-based training induced a 2-fold increase in TAG levels in the ALS tibialis, reaching levels comparable to controls, whereas the running-based training only increased TAG levels by 1.4-fold (Figure 8B).

Taken together these data suggested that swimming could favor fat deposition in ALS muscle.