Impaired energy metabolism of senescent muscle satellite cells is associated with oxidative modifications of glycolytic enzymes

Alteration of protein homeostasis during replicative senescence of human satellite cells

Muscle derived satellite cells isolated from a 5-day-old infant were cultivated until they reached replicative senescence at about 48 cumulative population doublings (CPD). Muscle derived satellite cells, also referred as myoblasts, were considered as young until 30 CPD and senescent at the end of their replicative life span when they ceased to respond to mitogenic stimuli and no population doublings were observed during a 4 week period. Senescent myoblasts exhibited typical morphological changes characteristic of senescent cells as they became flattened and enlarged when compared to young cells (Figure 1A). To further validate our model of cellular aging, we investigated the expression of the biomarker of senescence p16 (INK4a) protein (Figure 1B) and found it significantly increased during replicative senescence (Figure 1C). The accumulated p16 will bind the cyclin-dependent kinase 4 (Cdk4), thus inhibiting its activity and blocking the cell-cycle progression [26].

Proteasome inhibition appears to be significantly associated with the onset of in vitro cellular senescence in multiple cell types such as BJ [27] and WI38 fibroblasts [28]. However, data on senescent human myoblasts have not yet been reported. As depicted in Figure 2, a significant decrease in chymotrypsin-like (Figure 2A), trypsin-like (Figure 2B) and caspase-like (Figure 2C) proteasome peptidase activities was observed in human myoblasts undergoing replicative senescence (starting after 40 CPD). A decreased expression of the catalytic beta subunits in senescent WI 38 human fibroblasts has also been reported [28], explaining the observed decreased activities. However, in human myoblasts, the protein level of the three beta catalytic subunits (ß1, ß2 and ß5) remained constant during their replicative span (Figure 2D and 2E). These results suggest the presence of internal inhibitors, such as highly oxidized and cross-linked proteins that could explain the decreased proteasome activity observed in senescent myoblasts. Previous studies have also shown an inactivation of the proteasome as a consequence of the oxidation of some of its subunits [29–31].

Therefore, to gain further insights into the mechanisms that may be implicated in the impairment of proteasome activity and to further demonstrate the occurrence of deleterious protein modifications, the accumulation of proteins modified by carbonylation (Figure 2F), glycation (Figure 2G) and conjugation with the lipid peroxidation product HNE (Figure 2H), were monitored. Both carbonylated, glycated and HNE-modified proteins were found to be increased during replicative senescence of human myoblasts. Interestingly, the threshold of accumulation starts after 40 CPD, which is coincident with the kinetics of proteasome inhibition.

Senescence-associated protein modifications target only a subset of the total proteome involved in key cellular pathways

Although an increased level of damaged proteins has been shown to accumulate during aging and cellular senescence as well as in certain pathological conditions, in most cases the specific proteins targeted by these modifications have not yet been identified. The identification of these proteins may give insights into the mechanisms by which these damaged proteins could affect cellular functions. For this purpose, total protein extracts from young (30 CPD) and senescent myoblasts were resolved by two-dimensional electrophoresis (2D-PAGE). For each sample, two gel series were prepared in parallel, one for Coomassie colloidal blue staining to visualize the total protein profile and equal loading control, and the second for immunodetection of carbonylated, glycated or HNE-modified proteins (Figure 3A, 3B and 3C). Interestingly, the pattern of the modified proteins was not superimposable with the pattern obtained for the total protein staining, suggesting that certain proteins represent preferential targets for these detrimental modifications.

After analyses, during replicative senescence 22 protein spots were shown to be increasingly carbonylated (Figure 3A), 24 were increasingly glycated (Figure 3B), and 8 were increasingly modified by HNE (Figure 3C). Each spot was excised from the gel and analyzed by MALDI-TOF/TOF-MS for protein identification. All spots were successfully identified (Tables S1, S2 and S3). Proteins were identified as targets of one, two or the three modifications (Figure 3D). The identified proteins were then analyzed and grouped by metabolic pathways and cellular functions (Figure 3E). Major biological functions include carbohydrate metabolism, cellular morphology, migration and proliferation, as well as protein quality control, protein degradation and free radical scavenging. We identified 6 enzymes involved in glycolysis to be increasingly modified in the senescent myoblasts (Figure 4A), these were aldolase (ALDOA), triosephosphate isomerase (TPIS), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), phosphoglycerate mutase (PGAM), enolase (ENOA) and pyruvate kinase (PKM1/M2). To further investigate whether these modifications result in a modification of the glycolytic pathway, the steady-state levels of related metabolites were measured and glucose oxidation flux analyses were performed.

Impairment of glucose oxidation in senescent human myoblasts

Glycolytic intermediates such as glucose-6-phosphate, fructose-6-phosphate, and the isobar containing fructose 1,6-diphosphate, glucose 1,6-diphosphate, and myo-inositol 1,4 or 1,3-diphosphate were all found to be decreased (Figure 4B), suggesting a decreased glucose metabolism via glycolysis in senescent myoblasts. Pyruvate produced through glycolysis is transported into mitochondria to generate acetyl-CoA. The major metabolic fate of acetyl-CoA is to enter the tricarboxylic (TCA) cycle to drive the production of reducing equivalents that in turn fuel the mitochondrial electron transport chain for ATP generation by oxidative phosphorylation. Senescent myoblasts had decreased levels of acetyl-CoA, succinate and malate (Figure 4B), suggesting an altered TCA cycle flux in aged myoblasts. Metabolic failure in senescent myoblasts is also supported by a decreased cellular reducing potential indicated by the increase in the NAD⁺/NADH ratio due to higher NAD⁺ levels whereas NADH remained unchanged (Figure 4B). The oxidative modifications of glycolytic enzymes concomitant with the alteration of metabolites involved in this pathway suggested that myoblasts undergoing replicative senescence present dynamic changes in overall glucose metabolism. Nonetheless, the metabolite profiling analyses measure steady-state levels and are not conclusive in determining the actual flux of glycolytic metabolites. To directly address the hypothesis of impaired glucose metabolism during senescence, we measured [U-¹⁴C] glucose oxidation into CO₂, which requires both glycolysis and TCA cycle. As depicted in Figure 4C, ¹⁴CO₂ production from [U-¹⁴C] glucose was decreased by about 40% in senescent myoblasts when compared to their young counterparts. Thus, senescent myoblasts exhibited a substantial decrease in glucose oxidation likely due to impaired glycolysis and TCA cycle.

Since TCA cycle flux depends on the oxidation of reduced equivalents (NADH, FADH₂) by the mitochondrial respiratory chain, the oxygen consumption rate (OCR) of young and senescent cells was measured by a pharmacological profiling approach using the Seahorse instrument as previously described [32]. Previous studies have shown that the mitochondrial content increases in senescent cells as a function of cell size whereas mitochondrial density remained similar [33,34]. Due to the low number of cells required for the OCR experiments, it was more straightforward to determine cellular size rather than mitochondrial protein content, for which a much higher number of cells is needed. When the OCR were normalized to cellular size, the basal and maximal respiration rates were significantly higher in senescent myoblasts compared to young cells (CPD 30), whereas the spare respiratory capacity was unchanged (Figure 4D). The non-ATP generating respiration (leak) was increased (P<0.01), thus maintaining similar respiratory control in young (1.70 ± 0.2) and senescent cells (1.59 ± 0.1, P = ns). No evidence of uncoupling was observed. Taken together these results indicate that senescent myoblasts have a decreased capacity to oxidize glucose but their mitochondrial respiration capacity is not compromised, suggesting a direct impairment upstream in glycolysis and/or in the TCA cycle as supported by the targeted proteomics and the metabolic analyses.

Increased mobilization of non-carbohydrate substrates for energy production in senescent human myoblasts

Degradation of branched chain amino acids (BCAAs), namely leucine, isoleucine and valine, produces acetyl-CoA, propionyl-CoA, and succinyl-CoA, which can contribute to energy production via the TCA cycle [35]. Increased levels of leucine and valine in senescent myoblasts (Figure 5A), concomitant with the accumulation of their carnitine conjugates (Figure 5B), suggests an increased catabolism of BCAAs during replicative senescence.

Another specific metabolic feature of senescent myoblasts concerns lipid metabolism. Indeed, except for arachidonate (C20:4n6) whose level was increased, several free mono- and polyunsaturated fatty acids with a chain length higher than 18 carbons were found to be decreased, while free medium-chain fatty acids remained unchanged (Figure 6A). This may reflect a decreased synthesis of these specific fatty acids or an increased incorporation into specific cellular lipids. However, increased levels of several monoacyl glycerols and glycerol-3-phosphate (Figure 6B, left panel) as well as of glycerophosphorylcholine, phosphoethanolamine, choline and ethanolamine (Figure 6B, right panel) in senescent myoblasts suggest an increased turnover of storage lipids (triacyglycerols) and of membrane lipids (glycerophospholipids), pointing towards their mobilization for energy production.

This hypothesis was further reinforced by the observation that the level of several acyl-carnitines was markedly increased during replicative senescence (Figure 6C), suggesting an increased fatty acid β-oxidation. Such changes in lipid metabolism may also reflect changes in membrane composition during replicative senescence, which could also impact on lipid signaling as shown by the increased level of arachidonate (Figure 6A), a precursor of various eicosanoids. Remarkably, we also noticed an enhanced sphingomyelin metabolism in senescent myoblasts as shown by the increased levels of palmitoyl and stearoyl sphingomyelin and shingosine, whereas the sphinganine level decreased (Figure 6D). More importantly, sphingomyelin can release ceramide, a signaling molecule implicated in aging [36]. Previous studies have shown that exogenous ceramide inhibited the growth of young dermal fibroblasts and endothelial cells inducing premature senescence by its ability to inhibit DNA synthesis and mitogenesis [37,38]. In addition, ceramide inhibitors increased the expression of myogenic differentiation markers and cell fusion rate in myoblasts, whereas short-chain ceramides inhibited these responses [39]. Indeed, the changes in sphingomyelinase and ceramide appear to specifically distinguish senescence from quiescent growth arrest. Although we were not able to detect ceramide, increased levels of its degradation product sphingosine and a decrease in its precursor sphinganine suggests that ceramide levels are differentially regulated in senescent myoblasts as compared to young myoblasts.

Primary culture of human muscle satellite cells

Satellite cells were isolated as previously described in accordance with the French legislation on ethical rules [49]. Using the explants method, upon autopsy, neonatal satellite cells were isolated from the quadriceps muscle of a 5-day-old female infant that died due to a developmental heart defect without showing signs of neuromuscular disorder. Cells were expanded in growth medium (DMEM 65%(v/v), 199 media 15% (v/v) and fetal bovine serum 20% (v/v)). Cultures were performed in a humidified atmosphere with 5% CO₂ at 37ºC until sub-confluence. Magnetic-activated cell sorting (MACS) was used to purify myogenic cells using Mini MACS (Miltenyi Biotec, Ber-gisch Gladbach, Germany) using anti-neural cell adhesion molecule (NCAM) antibodies, according to manufacturer instructions. Sub-confluent cultures were obtained by seeding 4x10⁵cells/225 cm² culture flasks until they entered senescence. For each passage, cumulative population doublings (CPD) was calculated as: log (N/n)/ln 2, where N is the number of cells counted and n is the number of cells initially plated. In all experiments, myogenicity was greater than 95% as assessed by desmin immunostaining after 24h of differentiation by serum reduction to 2% (v/v).

Proteasome peptidase activities

Cell pellets derived from three independent experiments of cells at CPD: 30, 40, 46 and senescent were homogenized in an extraction buffer containing 50 mM Tris-HCl (pH 7) and 1 mM dithiothreitol. After clearance by centrifugation, protein concentration was determined by the Bradford method (Bio-Rad protein assay). Peptidase activities were assayed using the fluorescent peptide substrates: succinyl-Leu-Leu-Val-Tyr-aminomethylcoumarin (LLVY-AMC) for the chymotrypsin-like activity, Ac-Arg-Leu-Arg-aminomethylcoumarin (RLR-AMC) for the trypsin-like activity and N-benzyloxycarbonyl-Leu-Leu-Glu-ß-naphthylamide (LLE-NA) for the peptidyl-glutamyl peptide hydrolyzing or caspase-like activity, as previously described [46]. Proteasome peptidase activities were determined as the difference between total activity and the remaining activity of the crude lysates in the presence of the proteasome inhibitor MG132 (20 µM). All measurements were performed in three different batches of cells, and results were expressed as mean ± standard deviation (SD). Data were analyzed two-way ANOVA followed by Bonferroni's post hoc test. * P<0.01.

Protein extraction for proteomics analyses

Cellular pellets derived from three independent experiments were homogenized using a lysis buffer (10 mM tris-HCl (pH 7.4), 8 M urea, 2 M thiourea, 4% (w/v) CHAPS) and clarified by centrifugation. Proteins in the supernatant were precipitated using the 2-D Clean-Up kit (GE Healthcare). Protein precipitates were then resuspended in lysis buffer and protein quantification was performed as described above.

Immunodetection of carbonylated, HNE- and AGE-modified proteins after 1D or 2D gel electrophoresis

Immunoblot experiments for carbonyl detection were performed using the OxyBlot^TM kit, according to the manufacturer instructions. Detection of HNE-modified proteins was achieved after overnight incubation at 4^oC with polyclonal antibodies raised against HNE-modified KLH (ab46544, Abcam). For AGE-modified proteins, membranes were incubated with polyclonal antibodies raised against AGE-modified RNAse [50]. After washing, secondary antibodies IRDye were used, and membranes were scanned and quantified by the ODYSSEY infrared imaging system (LI-COR, Les Ulis, France). Two-dimensional polyacrylamide gel electrophoresis (2D PAGE) separation of proteins was performed as described previously [46]. For each sample, two gels were performed in parallel, one for colloidal blue staining of total proteins and other for electro-blotting onto nitrocellulose membranes. Membranes were then incubated for 2h in the blocking solution and immunodetection was performed as described above. Membranes were revealed using Amersham ECL Plus Western Blotting Detection System. Films were digitized with UMAX UTA-100 scanner (GE Healthcare, Saclay, France) and densitometry analyses were performed using NIH ImageJ or Image Master 2D Platinum 7 software (GE Healthcare).

Proteomics data acquisition and analysis

Spot detection and quantification was carried out using the Image Master 2D Platinum 7 software (GE Healthcare). For each spot identified by Coomassie colloidal blue as well as from the corresponding immunodetection (carbonylated, AGE- and HNE-modified proteins), a percentage volume was obtained in pixels (%Vol). The %Vol corresponds to a normalized value of the spot volume by considering the total volume of all the spots present in the gels or on films. %Vol of carbonylated spots from young (CPD 30) and senescent cells were normalized with %Vol of the corresponding Coomassie-stained spot to obtain a normalized %Vol (N%Vol). The N%Vol of senescent cells was divided by the N%Vol of young cells to obtain the Relative Modification Index ratio (RMI ratio). Spots with an RMI ratio ≥1.3 were considered as increasingly modified in the senescent cells. All measurements were repeated a minimum of three times and results are expressed as mean ± standard deviation (S.D.). Data were tested for normality and statistical significance for the comparison of two groups using student's t-test: *P < 0.01.

In-gel digestion, mass spectrometry protein identification and databases searches

Spots of interest were manually excised from gels and proteins were digested with trypsin as described previously [40]. Peptide mass spectra were acquired in positive reflector mode on a 4800 Plus MALDI TOF/TOF™ Analyzer (Applied Biosystems, Foster City, CA, USA) using 20 kV of acceleration voltage. MS and MS/MS peak lists were combined into search files and used to search SwissProt database using the Mascot search engine (Matrix Science Ltd, London, UK). Mascot protein scores greater than 56 are significant at p<0.05.

Metabolic analyses

Metabolites profiling analysis was performed by Metabolon, Inc. as previously described [51]. Cellular pellets (10⁶ cells) from six different batches from young (CPD30) and senescent myoblasts were accessioned into the Metabolon LIMS system.

Measurement of glucose flux and oxygen consumption rates

During the last 24 h of culture, the growth medium from young and senescent myoblasts was replaced by a medium containing 5 mM [U-¹⁴C]D-glucose (133 µCi/nmol) and insulin (10 nM). After 24h, the medium was then collected to determine ¹⁴CO₂, whereas cells were washed three times with ice-cold PBS and scraped for protein quantification. Oxygen consumption rate (OCR) was determined using Seahorse XF-96 extracellular Flux analyzer (Seahorse Bioscience, Billerica, MA, USA) as previously described [32]. Briefly, young (CPD 30) and senescent cells were seeded in the 96-well XF-96 cell in sixteen replicates at 20.000 cells/well in growth medium. Thirty minutes prior to the run, cells were equilibrated with new culture medium in a 37 °C CO₂ incubator for pH and temperature stabilization. Three mitochondrial inhibitors: oligomycin (1µM), carbonyl cyanide m-chlorophenylhydrazone (CCCP; 1-30 μM) and KCN (1mM) were sequentially injected (protocol, 12 min of equilibration, followed by measurements of 3 min, separated by mixing for 3 min). After all the measurements were completed, cells were dissociated and cell number was measured using a BD accuri cytometer (BD, Le Pont –de – Claix, France) while cell size was assessed using flow cytometry by monitoring light scattering.