Impaired xCT-mediated cystine uptake drives serine and proline metabolic reprogramming and mitochondrial fission in skeletal muscle cells

Abstract

Muscle satellite cell (MuSC) proliferation is tightly regulated by redox homeostasis and nutrient availability, which are often disrupted in muscular pathologies. Beyond its role in maintaining cellular redox homeostasis, this study identified a key metabolic role for cystine/glutamate antiporter xCT in proliferating MuSCs. We investigated the impact of impaired xCT-mediated cystine import in Slc7a11^sut/sut MuSCs isolated from mice that harbor a mutation in the SLC7A11 gene, which encodes xCT. We used complementary approaches to study how disrupted cystine import affects glutathione (GSH) redox, cellular bioenergetics, mitochondrial dynamics, and metabolism. Oxygen consumption rates of Slc7a11^sut/sut MuSCs were lower, indicative of compromised mitochondrial oxidative capacity. This was accompanied by a fragmented mitochondrial network associated with OPA1 cleavage and redox-sensitive DRP1 oligomerization. Metabolomic profiling revealed a distinct metabolic signature in Slc7a11^sut/sut MuSCs, manifested by major differences in BCAAs, pyrimidines, cysteine, methionine, and GSH. Despite lower overall bioenergetic flux, stable-isotope tracing analyses (SITA) showed that xCT deficiency increased glucose uptake, channeling glucose-derived carbons into de novo serine biosynthesis to fuel cysteine production via the transsulfuration pathway, partially compensating for disrupted GSH redox. Furthermore, xCT deficiency triggered upregulated pyrroline-5-carboxylate synthase (P5CS)-mediated proline reductive biosynthesis. By directing glutamate into proline synthesis, MuSCs apparently downregulate oxidative phosphorylation (OXPHOS) and regulate intracellular glutamate levels in response to impaired cystine/glutamate antiporter function. Our findings highlight the roles of xCT in regulating redox balance and metabolic reprogramming in proliferating MuSCs, providing insights that may inform therapeutic strategies for muscular and redox-related pathologies.

Graphical abstract

Highlights

• •Dysfunctional xCT impairs cellular redox and mitochondrial function in MuSCs.
• •xCT-deficient MuSCs undergo global metabolic reprogramming of amino acid metabolism.
• •xCT deficiency induces cysteine biosynthesis via the transsulfuration pathway.
• •xCT deficiency directs excess glutamate toward proline biosynthesis.

1. Introduction

Cysteine, a thiol-containing amino acid, is imported into cells via alanine-serine-cysteine transporters (ASCTs) and excitatory amino acid transporters (EAATs) [1]. In the extracellular space, cysteine predominantly exists in its oxidized form, cystine, which is imported into cells exclusively via system Xc− (xCT), in exchange for intracellular glutamate. xCT is a plasma membrane cystine/glutamate antiporter and comprises a heavy chain (4F2hc/SLC3A2), essential for membrane insertion, and a light chain (xCT/SLC7A11), which determines substrate specificity and transport activity [[2], [3], [4]]. As cysteine is the rate-controlling precursor for the biosynthesis of glutathione (GSH), xCT-mediated cellular uptake of cystine has well-established roles in lowering the levels of reactive oxygen species (ROS), maintaining intracellular redox homeostasis, and increasing cellular resistance to apoptosis [5,6].

Skeletal muscle possesses remarkable regenerative capacity due to a designated population of adult stem cells known as muscle satellite cells (MuSCs) [7,8]. In response to injury or exercise, quiescent MuSCs are activated and enter a proliferative phase to generate a pool of myoblasts dedicated to repairing muscle tissue while others commit to self-renewal ensuring the maintenance of the quiescent MuSC pool [9]. Failure to maintain the quiescent MuSC pool is linked to aging, muscular dystrophies, and metabolic diseases [10,11]. The initiation of MuSC proliferation requires substantial metabolic reprogramming involving a shift from mitochondrial oxidative phosphorylation (OXPHOS) to glycolysis usually ascribed to the high demands for rapid ATP production [12,13]. This metabolic shift during MuSCs proliferation also involves increased amino acid uptake to support biosynthetic requirements for rapid cell growth [14,15]. Mammalian cells have high rates of cyst(e)ine uptake and glutamate excretion during proliferation, implicating an important role for xCT in the control of these processes [14].

We previously demonstrated in Slc7a11^sut/sut mice that the absence of xCT was associated with greater MuSC activation in vivo following cardiotoxin muscle damage and greater commitment to muscle differentiation [16]. An important question emerging from this work was, how does impaired cystine import and glutamate efflux impact metabolic pathways, redox and mitochondrial dynamics in MuSCs? Here, we leveraged metabolomic profiling, stable isotope tracer analysis (SITA), metabolite transport and oxidation analyses, cell imaging and bioinformatic approaches in muscle cells in vitro. Findings show that xCT deficiency or its chemical inhibition in muscle cells promotes metabolic reprogramming, shifting glycolytic intermediates toward serine, cysteine, and proline biosynthesis in an attempt to compensate for disrupted cysteine and GSH metabolism and to restore GSH redox at the expense of oxidative metabolism. These metabolic perturbations are accompanied by impaired mitochondrial oxidative capacity, DRP1-mediated mitochondrial fragmentation, and elevated oxidative stress related to insufficient cyst(e)ine availability for GSH biosynthesis.

2. Experimental Procedures

2.1. Animals

All mouse experiments were conducted according to the principles and guidelines set by the Canadian Council on Animal Care and were approved by the Animal Care Committee at the University of Ottawa. Experiments were conducted using primary muscle cells isolated from male and female C3H/HeSnJ wild-type (WT) mice and background-matched Slc7a11^sut/sut (xCT^−/−) mice. No sex dimorphism was observed in any outcome measured. Mice were housed under standard conditions, maintaining a controlled temperature of 22–23 °C, humidity levels of 30–60 %, and a 12 h light/dark cycle (lights on from 07:00 to 19:00). Mice were given free access to water and fed a standard diet containing 18 % protein and 6 % fat (2018 Teklad Global Diet).

2.2. Mouse primary muscle cell isolation and culture

Skeletal muscles of the hindlimb were rapidly dissected and cleaned from fat and connective tissue. Muscle tissues were washed with PBS supplemented with 1 % antibiotic-antimycotic. Tissues were then treated with an enzymatic cocktail containing 1 mg/ml Dispase II and 1 mg/ml Collagenase B (Sigma-Aldrich) and chopped with a sterile razor blade. The homogenate was incubated at 37 °C for 30 min and vortexed every 5 min during the incubation. The homogenate was then centrifuged at 500 g for 5 min and pellets were resuspended and transferred into Matrigel-coated plates containing DMEM (25 mM glucose) supplemented with 20 % fetal bovine serum, 10 % horse serum, 2.5 ng/ml β-FGF (Sigma-Aldrich), 1x non-essential amino acids (11140050, Gibco), and 1 % antibiotic-antimycotic. Fibroblast populations were eliminated by the differential adhesion method, and primary muscle cell enrichment was achieved [17]. Primary MuSCs were cultured for 48 h. During the initial 24 h period, the cell culture medium was supplemented with 50 μM of 2-mercaptoethanol (2 ME), which was essential for early Slc7a11^sut/sut MuSCs survival as previously reported [16]. 2 ME was removed for the remaining 24 h, except for the Slc7a11^sut/sut + 2 ME condition, which was retained as a rescue treatment.

2.3. Immunostaining of primary muscle cells

To conduct mitochondrial morphological analyses, cells were cultured in 8-well glass slide plates (Millipore) coated with Matrigel (Corning). Cells from different conditions were rapidly washed with warm PBS and then fixed with 4 % PFA (Sigma-Aldrich) for 15 min. After fixation, cells were washed twice with PBS and incubated for 2 h at room temperature (RT) in PBS containing 1 % BSA, 0.2 % Triton-X (Sigma-Aldrich), and anti-TOMM20 (1:500, 11802, Protein Tech). Cells were washed twice with PBS and incubated for 1 h at RT in PBS containing 1 % BSA and a secondary antibody (Alexa Fluor 488 goat anti-rabbit IgG (H + L), 1:500, A-11008, Thermo Fisher Scientific). After 2 washes, cells were incubated for 5 min at RT with 4',6-diamidino-2-phenylindole (DAPI, 1:1000, Sigma-Aldrich). The polypropylene wells were lifted off the slide before mounting the cells in ProLong Gold antifade reagent (Invitrogen). Cells were imaged using a confocal microscope Zeiss LSM880 equipped with AiryScan FAST technology 63X/1.4 oil objective.

Mitochondrial morphological analyses were performed using Mitochondria Analyzer, a three-dimensional mitochondrial analysis pipeline in ImageJ/Fiji, as previously described [18]. This tool integrates adaptive thresholding and 2D/3D quantitative shape analysis to allow unbiased assessments of mitochondrial network characteristics.

2.4. Cystine and glucose uptake in primary muscle cells

Primary MuSCs were plated at 10,000 cells per well in Matrigel-coated 96-well white/clear bottom plates. Cystine uptake was measured using a cystine uptake assay kit (UP05, Dojindo), while glucose uptake was measured using a cell-based assay kit (600470, Cayman) following manufacturers' instructions. Briefly, for cystine uptake, cells were deprived of cystine for 30 min, then incubated with the cystine analog, selenocystine at 37 °C for 30 min. Cells were then incubated with fluorescein O, O′-diacrylate, and tris(2-carboxyethyl) phosphine for 30 min. For glucose uptake, cells were deprived of serum for 3 h before the assay. During the last hour, cells were starved of glucose for 30 min, then incubated with 200 μg/ml of 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino)-2-deoxyglucose (2-NBDG) for 30 min. Cystine and glucose uptake rates were determined by measuring fluorescence intensity at 490/535 nm and 485/535 nm, respectively, using a BioTek Synergy H1 Multi-Mode Plate Reader (BioTek Instruments).

2.5. GSH and GSSG measurements

Primary MuSCs for each indicated condition were grown in 100 mm Petri dishes, harvested with trypsin, and then washed with ice-cold PBS. A 1:1 homogenization buffer [125 mM sucrose, 1.5 mM EDTA, 5 mM Tris, 0.5 % trifluoroacetic acid (TFA), and 0.5 % meta-phosphoric acid (MPA) in 50 % mobile phase (10 % HPLC grade methanol, 0.09 % TFA – 0.2 μm filtered)] was used to homogenize the cells for 20 min on ice. Cell lysates were centrifuged at 14,000 g for 20 min at 4 °C, and supernatants were used for measurements. To quantify GSH and GSSG levels, an HPLC 1100 Series system (Agilent) equipped with a Pursuit C18 column (150 × 4.6 mm, 5 μm; Agilent) was employed with a 1 ml/min flow rate using a UV–visible wavelength detector at 215 nm (Agilent), as previously described [19]. Data were analyzed using the OpenLab CDS 2.8 software and values were normalized to cellular protein levels using a bicinchoninic acid (BCA) assay (Pierce BCA Protein Assay, 23225, Thermo Fisher Scientific).

2.6. Cellular bioenergetics

A Seahorse XFe96 Analyzer (Agilent) was used to assess oxygen consumption rates (OCR) and extracellular acidification rates (ECAR) in primary MuSCs plated at 10,000 cells/well. Mitochondrial stress tests assessed cellular resting respiration before and following four consecutive injections: oligomycin (2 μg/mL), FCCP (2.4 μM), combined antimycin A (5.5 μM)/rotenone (7.7 μM), and monensin (20 μM). This allowed determinations of resting, leak-dependent, and maximal rates of oxygen consumption. These rates were corrected for non-mitochondrial OCR, measured as antimycin A/rotenone-independent respiration. ATP-linked OCR was calculated as the difference between resting and leak respiration, and reserve capacity was determined by subtracting resting OCR from maximal OCR. The injection of monensin at the end of the injection cycle enabled the measurement of maximal ECAR, a proxy measure of cell glycolytic capacity [20]. Beyond resting levels of glycolytic rates, glycolytic reserve was determined by subtracting resting rates from the maximal rates in the presence of monensin.

For the glycolysis stress test, cells were washed and incubated in a glucose-free medium. After measuring resting ECAR, cells were treated with consecutive injections of glucose (10 nM), oligomycin (2 μg/mL), and 2-deoxyglucose (2DG, 50 mM). Following glucose and oligomycin injections, glycolysis and glycolytic capacity were determined, respectively. The glycolytic reserve was measured as the difference between glycolysis and glycolytic capacity.

2.7. Citrate synthase activity

Maximal citrate synthase activity as a measure of mitochondrial content was determined in protein samples in the presence of 5,5′-dithiobis(2-nitrobenzoate) DTNB as previously described [16]. The change in the rate of absorbance at 412 nm and pathlength was measured using a BioTek Synergy Mx Microplate Reader (BioTek Instruments). The extinction coefficient of 13.6 mM⁻¹cm⁻¹ was used. Values are expressed per μg cellular protein.

2.8. Mitochondrial to nuclear DNA quantification

DNA was isolated from WT and Slc7a11^sut/sut MuSCs as previously described [21]. Briefly, MuSCs were homogenized in DNA lysis buffer (5 mM EDTA, 0.2 % SDS, 100 mM Tris, 200 mM NaCl, pH 8.0) supplemented with proteinase K solution (Sigma). DNA extraction was performed using phenol/chloroform/isoamyl alcohol (25:24:1; PCIAA) and isopropanol (−20 °C). DNA concentration and purity were verified using the NanoDrop™ 2000 UV–Vis spectrophotometer (ThermoFisher). Mitochondrial DNA (mtDNA) to nuclear DNA (nDNA) ratios were determined through qPCR against mitochondrial DNA gene mt-ND1 (FWD: 5′-CTAGCAGAAACAAACCGGGC-3′, REV: 5′-CCGGCTGCGTATTCTACGTT-3′) and nuclear DNA gene n-HK2 (FWD: 5′-GCCAGCCTCTCCTGATTTTAGTGT-3′, REV: 5′-GGGAACACAAAAGACCTCTTCTGG-3′). Ratios were calculated using the 2^−ΔΔCT method [22].

2.9. Cellular protein levels

To measure cellular protein content, RIPA buffer (Millipore) supplemented with a protease inhibitor cocktail (P8340, Sigma-Aldrich) and phosphatase inhibitor cocktail (78420, Thermo Fisher Scientific) was used during cell homogenization. A BCA assay was used to measure protein concentrations, as per the manufacturer's protocol. Protein samples were kept at −80 °C for later use.

2.10. Western blot analyses

Protein samples were prepared in 1 x Laemmli buffer containing 100 mM DTT. Samples were separated on SDS-PAGE and then transferred onto PVDF (Bio-Rad) or nitrocellulose membranes. For glutathionylation, MFN1/2 oligomerization, and DRP1 oligomerization immunoblots, samples were prepared in 1 x Laemmli without DTT. Membranes were blocked with 5 % BSA in Tris-buffered saline containing 0.1 % Tween-20 (TBST) for 1 h at RT. Primary antibody incubations were overnight at 4 °C. The following antibodies were purchased from Abcam: xCT (1:2000, ab175186), GPX1 (1:2000, ab22604), GPX4 (1:2000, ab16800), Grx2 (1:1000, ab191292), NRF2 (1:1000, ab31163), Total OXPHOS (1:1000, ab110413), and MFN1/2 (1:5000, ab57602). Antibodies from Santa Cruz were: GCL-c (1:1000, sc-390811), and GSS (1:1000, sc-365863). Antibodies from Protein Tech were: GCL-m (1:1000, 14241-1), OPA1 (1:2000, 27733-1), BCAT2 (1:2000, 16417-1), PHGDH (1:2000, 14719-1), PSAT1 (1:5000, 10501-1), PSPH (1:2000, 14513-1), CBS (1:2000, 14787-1), CTH (1:2000, 12217-1), GS (1:2000, 11037-2), GLS1 (1:2000, 12855-1), GLS2 (1:2000, 20171-1), GLUD1 (1:5000, 14299-1), P5CS (1:2000, 17719-1), PYCR1 (1:1000, 13108-1), PYCR2 (1:1000, 55060-1), ALDH4A1 (P5CDH) (1:1000, 11604-1), and PRODH (1:2000, 22980-1).

Additional antibodies included: Glutathione (1:1000, 101-A, Virogen), DRP1 (1:2000, 611113, BD Biosciences), and ATF4 (1:1000, 11815-S, Cell Signalling). Primary antibodies used against loading controls were: GAPDH (1:10000, 60004-1-Ig, Proteintech), and Vinculin (1:5000, ab129002, Abcam). A ChemiDoc™ MP Imaging System (Bio-Rad) was used to visualize protein bands and ImageJ software was employed to conduct protein band densitometry. The abundance of all target proteins is presented as normalized to the indicated loading control.

2.11. Mitochondrial H₂O₂ emission

Mitochondrial H₂O₂ release was measured in cells using the fluorescent probe Amplex Red (ex/em: 563/587) using a Hitachi F2500 spectrophotometer as previously described [23]. Briefly, 1.5 million cells were incubated in 600 μL of buffer Z (in mM: 110 K-MES, 35 KCl, 1 EGTA, 5 K₂HPO₄, 3 MgCl_2·6H₂O, and 0.5 mg/mL BSA, pH 7.3 at 4 °C) in a 1 cm quartz cuvette with magnetic stirring at 37 °C supplemented with 1.2 U/mL horseradish peroxidase, and 20 μM Amplex Red. Cells were permeabilized with 2 μg/μL digitonin. After baseline readings, the following were added sequentially: 2.5–5 mM malate-glutamate, 5 mM succinate, 10 mM ADP, and 8 μM antimycin-A. Values are reported as arbitrary fluorescence units.

2.12. Metabolomic stable isotope tracer analysis (SITA) and LC-MS

Stable isotope tracing was performed as previously described [24]. Briefly, cells were seeded in 60 mm dishes to achieve ∼75 % confluency for 24 h. DMEM was then replaced with equivalent media without 2 ME, FBS, and HS, supplemented with 20 % dialyzed FBS for 24 h. Then, an equivalent labelled medium with 25 mM [U–¹³C]-glucose (CLM-1396-1, Cambridge Isotope Laboratories Inc) was added for the indicated time points. Cells were washed three times with ice-cold 150 mM ammonium formate solution, quenched in 230 μL ice-cold LC/MS grade 1:1 methanol:water solution, and vortexed for 10 s, before adding 220 μL acetonitrile. The collected cells were then homogenized using a bead mill homogenizer at 4 °C for two rounds of 60 s at 30 Hz (Fisherbrand Bead Mill 24 Homogenizer). Homogenates were incubated with a 2:1 dichloromethane:water solution on ice for 10 min then centrifuged at 1500 g for 10 min at 1 °C. Water-soluble metabolites were collected from the upper phase, dried using a refrigerated CentriVap Vacuum Concentrator at −4 °C (LabConco Corporation), and stored at −80 °C before LC-MS analyses. Control samples that were not incubated with [U–¹³C]-glucose were included in all tracer experiments.

Samples were randomized and re-suspended with 75 % acetonitrile, cleared by centrifugation, and run in negative ESI on a 6545B Q-TOF mass spectrometer (Agilent) equipped with a 1290 Infinity II ultra-high-performance LC (Agilent) using hydrophilic interaction chromatography (HILIC-Z). Continuous internal mass calibration was executed using signals from purine [12,000 full width at half maximum (FWHM) resolution] and hexakis (1H, 1H, 3H-tetrafluoropropoxy) phosphazine (24,000 FWHM resolution). HILIC separation was obtained using the Poroshell 120 HILIC-Z column (2.1; 100 mm, 2.7 mm; Agilent) and the corresponding guard column. The chromatographic conditions and mass spectrometry acquisition parameters are described elsewhere [25]. The binary solvent system consisted of 10 mM ammonium acetate (pH 9) in water (solvent A) and 100 mM ammonium acetate in 85 % acetonitrile (solvent B), both having 0.1 % medronic acid. The gradient for separation started at 96 % B for 1.5 min, then decreased from 96 % to 65 % B for 6.5 min followed by a 2 min hold at 65 % B, and 7 min of re-equilibration to 96 % B. This took the total run time to 17 min at a 0.25 mL/min flow rate. The injection volume was 10 μL, and the column temperature was maintained at 35 °C.

Mass spectrometry detection was performed in negative ESI full scan mode with a mass range of 50–1000 m/z. The mass spectrometer source conditions consisted of a capillary voltage of 3000 V. Drying and sheath gas temperatures were set to 200 and 300 °C and flow rates to 10 and 12 L/min, respectively. Nebulizer pressure was set to 40 psi and the fragmentor voltage was 175 V. Data were acquired in centroid mode at the rate of 3 spectra per second in the extended dynamic range mode (2 GHz). A target list of metabolites was created using “MassHunter Pathways to PCDL” software (Agilent). Metabolite retention times were provided from an in-house database.

2.13. Metabolomic profiling

Samples were collected and run using LC-MS as mentioned in the SITA protocol but without tracer. All values were normalized to the protein content of parallel plates. Data were analyzed using Python (Python Software Foundation. Python Language Reference, version 3.8.18. Available at www.python.org) and R (version 4.2.2.) [26]. All figures were produced using the matplotlib [27] and ggplot2 [28] libraries. Unless stated otherwise, statistical tests were performed through their respective Scipy method [29]. All data and code for these analyses are available in our GitHub repository: https://github.com/lkenn012/xCT_metabolomics.

2.14. Analysis of metabolomic profiling data

Hierarchical clustering and several statistical methods were applied after pre-processing the data from sample groups. Three different feature selection methods were employed for differential metabolite determination including Welch's t-test; orthogonal partial least squares discriminant analysis (OPLS-DA); and significance analysis of microarrays (SAM). Missing values were imputed according to a limit of detection imputation defined as ⅕ of the minimum value per metabolite feature. Following this, abundance values were log₁₀-transformed and z-score normalized. Z-score normalized values were used for all statistical analyses except for significance testing by Welch's t-test, where the raw values including missing values were used.

For OPLS-DA, the “ropls” R package (version 1.30.0) [30] was used with 7-fold cross-validation for computing predictive performance (Q²) and 20 permutation tests for computing significance. Important metabolites for sample group separation were determined based on the peak predictive performance of OPLS-DA models using increasingly stringent VIP score thresholds (using the formulation defined as Vip_4,pred by Galindo-Prieto, Eriksson, and Trygg [31]) of metabolite features, as suggested by Anderson and Bro [32]. For SAM tests, significantly different metabolites between WT and Slc7a11^sut/sut samples were carried out using the “samr” R package (version 3.0) with 100 permutation tests (nperms = 100) [33].

Enrichment plots for metabolite clusters were generated using the Metaboanalyst web server [34] via over-representation analysis, using all identified metabolites in the dataset as background (excluding non-specific metabolites such as total hexoses). Agglomerative hierarchical clustering linkages were determined according to the Ward algorithm applied to Euclidean distances (maximum cluster distance of 4) for Slc7a11^sut/sut and WT unlabelled metabolomics data. Robinson-Fould metrics were computed using the ETE3 Python library [35] (ete3 version 3.1.3); these metrics estimate the conservation between two clustering dendrograms based on the proportion of shared linkages between metabolites.

2.15. Statistics

Unless otherwise mentioned, all data are shown as a mean ± standard error of the mean (SEM). Statistical analyses were conducted using Prism (GraphPad, La Jolla, CA). A two-tailed Student's t-test was used to determine statistical significance between WT vs. Slc7a11^sut/sut. The statistical significance of primary MuSCs experiments with 3 different conditions (WT, Slc7a11^sut/sut, and Slc7a11^sut/sut + 2 ME) was determined using one-way ANOVA with Tukey post hoc tests. P-values <0.05 were considered statistically significant.

3. Results

3.1. xCT controls GSH levels and redox in proliferating muscle cells

Our first objective was to determine how xCT-mediated cystine import influences cellular GSH metabolism in proliferating MuSCs. To this end, we utilized primary muscle cells isolated from Slc7a11^sut/sut mice and WT controls. Slc7a11^sut/sut mice have a recessive mutation in the Slc7a11 gene, resulting in the truncation of the xCT protein at the C terminus [36,37]. Slc7a11^sut/sut MuSCs were cultured in media supplemented with β-mercaptoethanol (2 ME) to support cell viability during proliferation [16]. 2 ME was removed 24 h before experimental analyses to allow analyses of the impact of xCT deficiency on metabolism, given that 2 ME reduces extracellular cystine to cysteine, which can be transported into cells via Na (+)-dependent amino acid transporters, thereby masking the impact of xCT on cellular functions. We first measured cystine uptake levels to validate our cell model and approach. As expected, cystine uptake was lower (52 %) in Slc7a11^sut/sut MuSCs than in WT (Fig. 1A). A similar decrease of cystine uptake was observed in WT MuSCs treated with the xCT inhibitor, erastin (10 μM) (Fig. 1A). In addition to impaired cystine transport, Slc7a11^sut/sut MuSCs exhibited decreased xCT protein levels (Fig. 1B), confirming that xCT mutation compromises the protein transport activity and expression levels. In line with impaired cystine uptake, HPLC analyses revealed lower intracellular levels of GSH, GSH:GSSG, and total GSH, but higher GSSG in Slc7a11^sut/sut MuSCs compared to WT MuSCs (Fig. 1C). However, treating Slc7a11^sut/sut MuSCs with 2 ME effectively increased GSH, GSH:GSSG, and total GSH to levels higher than those in WT MuSCs (Fig. 1C). Furthermore, immunoblotting analysis revealed that Slc7a11^sut/sut MuSCs have a trend for decreased protein glutathionylation levels compared to WT (Fig. 1D). These findings indicate that the xCT mutation impairs cystine influx and intracellular glutathione redox.

Fig. 1.

xCT controls GSH levels and redox in proliferating muscle cells

(A) Rate of cystine uptake measured using the fluorescent probe fluorescein O, O′-diacrylate in WT, Slc7a11^sut/sut, and WT + erastin MuSCs. (B) Immunoblot of WT and Slc7a11^sut/sut MuSCs against xCT. (C) Reduced glutathione (GSH), oxidized glutathione (GSSG), GSH:GSSG ratio, and total glutathione (GSH + (2 x GSSG)) measured by HPLC in WT, Slc7a11^sut/sut, and Slc7a11^sut/sut + 2 ME. (D–E) Immunoblots of WT and Slc7a11^sut/sut MuSCs against (D) post-translational glutathionylation, and (E) GSH synthesis enzymes, glutamate-cysteine ligase (GCL) catalytic (GCL-c) and modifier (GCL-m) subunits, and GSH synthetase (GSS). (F) Mitochondrial H₂O₂ emissions in digitonin permeabilized WT and Slc7a11^sut/sut MuSCs. (G–H) Immunoblots of WT and Slc7a11^sut/sut MuSCs against (G) GPX1, and (H) GPX4. Comparisons between groups were determined using a one-way ANOVA with post hoc Tukey HSD test, n = 5–6 (A,C); two-tailed Student's t-test, n = 5–6 (B, D-H). Results are presented as mean ± SEM, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

To evaluate the impact of xCT deficiency on mechanisms of GSH biosynthesis we measured the protein levels of key enzymes involved in this process. Immunoblotting analyses demonstrated similar levels of glutamate-cysteine ligase catalytic subunit (GCL-c) in both groups, but lower levels of the regulatory glutamate-cysteine ligase modifier subunit (GCL-m) in Slc7a11^sut/sut MuSCs compared to WT (Fig. 1E) [38]. Interestingly, protein levels of GSH synthetase (GSS), which catalyzes the last step of GSH biosynthesis by adding glycine, were higher in Slc7a11^sut/sut MuSCs (Fig. 1E), in a failed attempt to compensate for decreased cyst(e)ine availability for GSH biosynthesis. Altogether, these findings are consistent with the idea that the decrease in intracellular GSH in Slc7a11^sut/sut MuSCs results from impaired cystine import via xCT.

To assess if the perturbed cellular GSH redox influenced the susceptibility of Slc7a11^sut/sut MuSCs to oxidative stress, we quantified H₂O₂ production. H₂O₂ production was higher in Slc7a11^sut/sut MuSCs than in WT across different mitochondrial respiratory states (Fig. 1F). Despite increased H₂O₂ production, there were similar protein levels of GSH peroxidase 1 (GPX1), GSH peroxidase 4 (GPX4), and NRF2 between WT and Slc7a11^sut/sut MuSCs (Fig. 1G and H, Supplementary Figure 1A). These findings suggest that the increase in H₂O₂ levels was not attributable to lower protein levels of GSH-dependent antioxidant enzymes; instead, it may result from limited GSH availability.

3.2. Impaired cellular bioenergetics accompanied by fragmented mitochondrial structure in Slc7a11^sut/sut MuSCs

Given that cellular GSH redox is known to be important in maintaining mitochondrial energy homeostasis, we performed Seahorse XF analyses to assess mitochondrial respiration and glycolytic flux in WT and Slc7a11^sut/sut MuSCs [39,40]. During the mitochondrial stress test, MuSCs were provided with a standard substrate mixture containing glucose (25 mM), glutamine (4 mM), and pyruvate (1 mM) to support major metabolic pathways. Slc7a11^sut/sut MuSCs displayed overall lower oxidative capacity with a 42 % decrease in resting, a 33 % decrease in leak, and a 50 % decrease in maximal oxygen consumption rates (OCR) compared to WT MuSCs (Fig. 2A and B). Quantification of extracellular acidification rates (ECAR), a proxy measure of glycolysis, revealed that Slc7a11^sut/sut MuSCs display a trend for a decrease in resting and maximal ECAR rates compared to WT MuSCs (Fig. 2C and D). When forcing the cells to use glycolysis by inhibiting OXPHOS, the Slc7a11^sut/sut MuSCs exhibited an impaired ability to upregulate glycolysis (Supplementary Fig. 2A and B). However, when using the Mookerjee method [41] to evaluate absolute ATP levels from glycolysis and OXPHOS under different metabolic states, it was clear that Slc7a11^sut/sut MuSCs were more dependent on glycolytic than OXPHOS for ATP production. Specifically, there was a higher proportion of ATP derived from glycolysis in Slc7a11^sut/sut MuSCs during resting and maximal respiration (Fig. 2E and F). In line with these findings, glucose uptake was higher in Slc7a11^sut/sut than in WT MuSCs, further confirming the greater dependence on glycolysis in xCT-deficient cells (Fig. 2G).

Fig. 2.

Impaired cellular bioenergetics accompanied by fragmented mitochondrial structure in Slc7a11^sut/sut MuSCs

(A-B) Oxygen consumption rates (OCR) and (C–D) extracellular acidification rates (ECAR) measured in primary WT and Slc7a11^sut/sut MuSCs. The contributions of OXPHOS and glycolysis to ATP production were calculated following the Mookerjee et al., 2017 method for (E) Basal and (F) Maximal bioenergetic capacities in WT and Slc7a11^sut/sut MuSCs. (G) Glucose uptake quantified using 2-NBDG in WT and Slc7a11^sut/sut primary MuSCs. (H) Citrate synthase enzyme activity in WT and Slc7a11^sut/sut MuSCs. (I) mtDNA:nDNA ratio measured in WT and Slc7a11^sut/sut MuSCs. (J) Immunofluorescence staining of TOMM20 (green) and DAPI (blue), scale bar = 10 μm. Thresholded TOMM20 signals are shown in white. (K–L) Quantitative morphometric analyses of TOMM20-staining examining (K) total branch length per mitochondria (μm), and (L) number of branches per mitochondria. Each data point represents a region of interest (ROI) containing 2 to 3 mitochondria. 2 ROI were imaged and analyzed for each sample, n = 4. The statistical significance of the differences between groups was determined using a two-tailed Student's t-test, n = 4 (A-F, J-L); n = 6 (G–I). Results are presented as mean ± SEM, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

To determine whether the diminished OXPHOS capacity observed in Slc7a11^sut/sut MuSCs resulted from decreased mitochondrial content, we measured citrate synthase activity, mtDNA:nDNA, and the protein levels of key OXPHOS proteins and TOMM20. Findings showed that citrate synthase activity and protein levels of OXPHOS complexes I, II, and IV were lower in Slc7a11^sut/sut compared to WT MuSCs (Fig. 2H, Supplementary Fig. 2C). Consistently, mtDNA:nDNA in Slc7a11^sut/sut MuSCs demonstrated a significant trend for a decrease (p = 0.052) (Fig. 2I), further indicating a decreased mitochondrial content in Slc7a11^sut/sut compared to WT MuSCs. In contrast, TOMM20 protein levels were comparable between WT and Slc7a11^sut/sut MuSCs (data not shown).

GSH redox and oxidative stress are critical in regulating mitochondrial dynamics [42]. Specifically, high GSSG levels promote mitochondrial hyperfusion [42,43], and oxidative stress is associated with increased fission [44]. Given our observations of elevated GSSG levels and H₂O₂ emissions in Slc7a11^sut/sut MuSCs, we were curious as to whether the xCT deficiency would cause hyperfusion, increased fission, or no change in the mitochondrial network. Quantitative analysis of TOMM20-stained cells was performed using a 2D/3D mitochondrial morphology analysis pipeline established by Chaudhry et al. [18]. Results revealed that the mitochondrial network was more fragmented in Slc7a11^sut/sut MuSCs compared to WT MuSCs, as indicated by the lower total branch length per mitochondrion and lower number of branches per mitochondrion (Fig. 2J, K, and L). The fragmented mitochondrial reticulum observed in Slc7a11^sut/sut MuSCs was not attributed to differences in protein levels of the large GTPase protein that mediates mitochondrial fission, dynamin-related protein-1 (DRP1) (Supplementary Figure 2D). Furthermore, no differences were detected in the phosphorylation levels of DRP1 at serine residues Ser⁶¹⁶ and Ser⁶³⁷ (Supplementary Figure 2E), which are responsible for fission induction and inhibition, respectively. Additionally, we measured the protein levels of other GTPase proteins involved in mitochondrial fusion. While mitofusin 1 and 2 (MFN1/2) protein levels were comparable between WT and Slc7a11^sut/sut MuSCs, the long isoform of optic atrophy 1 (L-OPA1) showed decreased protein levels in xCT-mutant cells (Supplementary Fig. 2F and G). Furthermore, Slc7a11^sut/sut MuSCs showed a trend toward increased levels of the short isoform of OPA1 (S-OPA1) (Supplementary Figure 2G), consistent with the established role of S-OPA1 accumulation in response to mitochondrial stress conditions [45]. As GSH redox also plays a role in determining MFN1/2 oligomerization which drives mitochondrial fusion by regulating disulfide modifications in the mitochondrial membrane [43,46], we then performed immunoblots under non-reducing conditions to preserve disulfide bonds and found that Slc7a11^sut/sut MuSCs had less MFN1/2 oligomers (between 160 and 250 kDa, Supplementary Figure 2H) compared to WT MuSCs, but displayed higher DRP1 oligomerization than WT MuSCs (Supplementary Figure 2I). In summary, xCT deficiency results in a fragmented mitochondrial network in MuSCs associated with OPA1 cleavage, lower MFN1/2 oligomers, and higher DRP1 oligomerization, which may contribute to impaired mitochondrial oxidative capacity.

3.3. Metabolomic profiling analysis reveals distinct phenotypes including differences in branched chain amino acids, cysteine, methionine, and proline in Slc7a11^sut/sut MuSCs

To elucidate the specific metabolic pathways impacted by impaired xCT function, we analyzed the global metabolite profiles in WT and Slc7a11^sut/sut MuSCs under steady-state conditions using ion-pairing (LC-MS). We applied three different feature selection methods to the 107 detected metabolites to characterize the metabolic phenotypes (Welch's t-test p < 0.05; OPLS-DA; and SAM), which consistently identified 27 metabolites that differed in abundance between Slc7a11^sut/sut and WT MuSCs (Fig. 3A and B, Table 1, and Supplementary Fig. 3A, B and C). Slc7a11^sut/sut MuSCs had lower levels of GSH, 4,5-dihydroorotate, uracil, dihydrouracil, and orotic acid, but higher levels of several amino acids compared to WT MuSCs, including proline, methionine, glycine, glutamine, and branched-chain amino acids (BCAAs, valine, leucine, and isoleucine) (Fig. 3A and Table 1). Hierarchical clustering showed well-distinguished metabolic clusters between Slc7a11^sut/sut and WT MuSCs, with a normalized Robinson-Foulds (RF) distance of 0.914 (where RF = 1 no common clustering in dendrograms, see Experimental Procedures) [47]. The 27 identified metabolites differing in abundance were dispersed across several clusters in the WT samples (clusters 1, 2, and 3), but largely showed related behavior and grouped in a single cluster (cluster 3) in Slc7a11^sut/sut MuSCs. While clusters 6 and 7 in WT samples are largely consistent with Slc7a11^sut/sut clusters 5 and 6, differences in the clustering of many other metabolites further indicate an overall metabolic reprogramming (Supplementary Figure 3D).

Fig. 3.

Metabolomic profiling analysis reveals distinct phenotypes including differences in branched chain amino acids, cysteine, methionine, and proline in Slc7a11^sut/sut MuSCs

(A) Volcano plot of metabolite profiles. (B) Metabolites differing in abundance between WT and Slc7a11^sut/sut MuSCs were determined by Welch's t-test, OPLS-DA, and SAM. (C) KEGG metabolite set enrichment analysis (top 10, p-value <0.30) of metabolic pathways altered in Slc7a11^sut/sut MuSCs. (D) Abundance of branched-chain amino acid metabolites valine, leucine, and isoleucine (relative to WT) (E). Immunoblot of branched-chain amino acid transaminase 2 (BCAT2) in WT and Slc7a11^sut/sut MuSCs. (F–G) Abundance of serine, methionine, GSH (reduced), glutamine, ornithine, citrulline, and proline (relative to WT). Comparisons between groups were determined using a two-tailed Student's t-test, n = 6 (D–G). Results are presented as mean ± SEM, ∗p < 0.05, ∗∗p < 0.01.

Table 1.

List of significantly different metabolites between WT and Slc7a11^sut/sut MuSCs.

Metabolite	WT	Slc7a11sut/sut	log2 FC	t-test p-value	SAM	OPLS-DA
Citrulline	14922	33139	1.05198	0.00020	0.02486	1.60270
Tryptophan	55109	94749	0.72614	0.00102	0.02486	1.71256
Ornithine	13565	25924	0.83596	0.00134	0.02486	1.39675
Phenylalanine	30191	48760	0.52096	0.00139	0.02486	1.72568
Leucine	19273	31582	0.52991	0.00151	0.02486	1.73038
Proline	349393	769361	0.88983	0.00173	0.02486	1.66877
Isoleucine	16476	26944	0.52684	0.00210	0.02486	1.65313
Tyrosine	17564	27135	0.54251	0.00263	0.02486	1.64029
Ethyl pyruvate	53655	85704	0.79305	0.00293	0.02486	1.58940
Glycine	17762	30089	0.74111	0.00390	0.02486	1.61603
4,5-dihydroorotate	86398	37895	−1.33959	0.00425	0.02486	1.63022
Uridine	30710	50472	0.87892	0.00444	0.02486	1.54333
Valine	8513	14383	0.63449	0.00444	0.02486	1.60971
Threonine	144251	240015	0.68138	0.00487	0.02486	1.54312
Methionine	158338	263288	0.80749	0.00632	0.02486	1.55373
Histidine	41111	71955	0.72868	0.00691	0.02486	1.56655
Histamine	12046	19978	0.77063	0.00919	0.02486	1.56147
Glutamine	330381	511514	0.55201	0.00945	0.03072	1.47808
Dihydrouracil	31715	12634	−1.28365	0.01022	0.02486	1.68672
Asparagine	48528	68521	0.38894	0.01325	0.03947	1.46708
Lysine	7731	13708	0.63061	0.01340	0.03397	1.17329
Orotic acid	35192	20464	−0.71437	0.01510	0.03408	1.27957
Serine	111068	158695	0.41710	0.02090	0.04093	1.36111
Uracil	65576	37279	−0.80083	0.02378	0.03072	1.43802
Malonic acid	14312	20449	0.47252	0.03366	0.04492	1.28729
2-Acetamido-2-Deoxy-d-Glucopyranose	5053	7244	0.61275	0.04414	0.04255	1.25646
Glutathione (reduced)	169737	96518	−0.67949	0.04648	0.03743	1.26352

Metabolite set enrichment analyses of the 27 differing metabolites further highlighted major differences between Slc7a11^sut/sut and WT MuSCs in the metabolism of BCAAs, pyrimidines, cysteine, methionine, and GSH (Fig. 3C). BCAAs are catabolized by branched-chain amino acid transaminases (BCATs), which exist in two isoforms: the cytosolic and redox-sensitive BCAT1, and mitochondrial BCAT2, which is the predominant isoform in skeletal muscle [48,49]. The increased abundance of valine, leucine, and isoleucine (Fig. 3D) could be related to decreased BCAA catabolism, as protein levels of branched-chain amino acid transaminase 2 (BCAT2) was lower in Slc7a11^sut/sut MuSCs (Fig. 3E). As anticipated, metabolites involved in endogenous cysteine biosynthesis, such as serine and methionine, were more abundant in Slc7a11^sut/sut MuSCs (Fig. 3F). Importantly, key metabolites related to proline biosynthesis and urea cycle (glutamine, ornithine, and proline) were also among the most elevated metabolites in Slc7a11^sut/sut MuSCs compared to WT (Fig. 3G).

3.4. xCT deficiency upregulates glucose uptake and promotes cellular de novo serine synthesis

Given the evidence of greater reliance on glycolysis for ATP production in Slc7a11^sut/sut MuSCs, we next performed stable isotope tracing analysis (SITA) of [U–¹³C]-glucose to elucidate alterations in flux through specific metabolic pathways. After testing various incubation times following standard practices, a 12-h incubation with ¹³C-labelled glucose proved optimal for most metabolic pathways of interest (Supplementary Figure 4A) [50]. The integration of ¹³C into the glycolytic metabolites 3-phosphoglycerate (3 PG) and 2-phosphoglycerate (2 PG) was similar between Slc7a11^sut/sut and WT MuSCs (Fig. 4A, B and C). Glucose-derived pyruvate is the major end-product of glycolysis, which can subsequently be converted to acetyl-CoA by pyruvate dehydrogenase (PDH) to enter the TCA cycle or can contribute to TCA cycle anaplerosis via pyruvate carboxylase (PC) production of oxaloacetate. The Slc7a11^sut/sut MuSCs displayed elevated amounts of pyruvate m+3 and lactate m+3 compared to WT (Fig. 4D and E) indicating that pyruvate accumulates instead of entering the TCA cycle. Indeed, the calculated citrate m+2/pyruvate m+3 and citrate m+3/pyruvate m+3 ratios, proxy measures for PDH- and PC-dependent labelling respectively [51], were lower in the Slc7a11^sut/sut MuSCs (Fig. 4F and G). Moreover, Slc7a11^sut/sut MuSCs had a lower carbon labelling to TCA cycle metabolites (Fig. 4H and Supplementary Fig. 4B), further supporting the conclusion that there is an impaired utilization of glucose-derived carbons in the TCA cycle by Slc7a11^sut/sut MuSCs.

Fig. 4.

xCT deficiency upregulates glucose uptake and promotes cellular de novo serine synthesis

(A) Stable isotope tracing diagram for [U–¹³C]-Glucose through glycolysis and TCA cycle via pyruvate dehydrogenase (PDH, green) and pyruvate carboxylase (PC, purple). Labelled metabolites are colored orange (higher in Slc7a11^sut/sut MuSCs), blue (lower in Slc7a11^sut/sut MuSCs), and black (similar between Slc7a11^sut/sut and WT MuSCs). (B–E) [U–¹³C]-Glucose labelling (12 h) of labelled glycolytic intermediate metabolites 3-phosphoglycerate, labelled 2-phosphoglycerate, pyruvate m+3, and lactate m+3, in WT and Slc7a11^sut/sut MuSCs. (F–G) Citrate m+2/pyruvate m+3 and citrate m+3/pyruvate m+3 ratios of [U–¹³C]-glucose labelling (12 h) as proxies of PDH and PC, respectively. (H) Relative proportion of labelled TCA cycle intermediates (citrate, α-ketoglutarate, succinate, fumarate, and malate) derived from [U–¹³C]-glucose precursor. (I) [U–¹³C]-glucose labelling through the de novo serine synthesis pathway. Enzymes and labelled metabolites are colored orange (higher in Slc7a11^sut/sut MuSCs), blue (lower in Slc7a11^sut/sut MuSCs), and black (similar between Slc7a11^sut/sut and WT MuSCs). (J) [U–¹³C]-glucose labelling (12 h) of serine m+3 (from glycolysis) expressed as percentage of total respective metabolite levels. (K) Immunoblot of de novo serine synthesis (PHGDH, PSAT1, and PSPH)., (L) [U–¹³C]-glucose labelling through cysteine biosynthesis via the transsulfuration pathway. (M) [U–¹³C]-glucose labelling (12 h) of GSH m+2 (from [U–¹³C]-glucose-derived glycine or glutamate), GSH m+3 (from [U–¹³C]-glucose-derived cysteine), and GSH m+4 (from [U–¹³C]-glucose-derived glycine and glutamate). (N) Immunoblot of de novo cysteine synthesis (CBS and CTH) enzymes. The statistical significance of the differences between groups was determined using a two-tailed Student's t-test, n = 6. Results are presented as mean ± SEM, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Glucose also serves as a precursor for de novo serine synthesis by converting glycolysis-derived 3 PG to 3-phosphohydroxypyruvate in a reaction catalyzed by phosphoglycerate dehydrogenase (PHGDH) (Fig. 4I) [52]. De novo serine synthesis is upregulated during times of amino acid starvation, as serine is a major donor to the carbon pool for one-carbon (1C) metabolism, linking the folate and methionine cycles with the transsulfuration pathway to support amino acid metabolism, redox homeostasis, nucleotide biosynthesis, and methylation reactions [53]. Serine 1C metabolism also plays a key role in GSH biosynthesis via endogenous cysteine biosynthesis [54] (Fig. 4L). Thus, we hypothesized that the Slc7a11^sut/sut MuSCs would have an increased reliance on glycolytic metabolism to support de novo synthesis of serine as a compensatory mechanism to restore GSH redox. Consistent with our hypothesis, Slc7a11^sut/sut MuSCs had higher serine m+3 levels (from glycolysis-derived 3 PG) compared to WT [55], indicating an increased serine biosynthesis from glycolysis (Fig. 4J). In support of this observation, lactate dehydrogenase activity was comparable between groups (data not shown), further confirming that Slc7a11^sut/sut MuSCs upregulate the early steps of glycolysis to support serine biosynthesis but not the latter steps, including lactate production. Moreover, protein levels of key enzymes involved in de novo serine synthesis (PSAT1 and PSPH) were higher in Slc7a11^sut/sut MuSCs compared to WT MuSCs (Fig. 4K). Notably, WT and Slc7a11^sut/sut MuSCs had similar m+2 labelling to glycine, suggesting that glucose-derived serine is preferentially shunted towards the transsulfuration pathway for cysteine biosynthesis (Supplementary Figure 4C). Indeed, the levels of GSH m+2 (derived from glycine m+2 or glutamate m+2 incorporation), and GSH m+4 (derived from the incorporation of both glycine m+2 and glutamate m+2) were lower in Slc7a11^sut/sut MuSCs (Fig. 4M). Conversely, levels of GSH m+3 (derived from cysteine m+3 incorporation) was not different between genotypes (Fig. 4M). Furthermore, while expression of cystathionine β-synthase (CBS), which catalyzes the conversion of serine and homocysteine to cystathionine, was comparable between genotypes, the protein level of cystathionine γ-lyase (CTH), which converts cystathionine to cysteine, was higher in the Slc7a11^sut/sut MuSCs (Fig. 4N), suggesting that transsulfuration pathway activity in Slc7a11^sut/sut MuSCs resulted in GSH biosynthesis. We also measured protein levels of activating transcription factor 4 (ATF4), a key regulator of serine biosynthesis and cellular response to amino acid starvation [56]. In line with upregulated serine biosynthesis and transsulfuration pathway activity, immunoblots showed increased ATF4 in Slc7a11^sut/sut MuSCs (Supplementary Figure 4D). Lastly, we assessed the synthesis of asparagine that can also be induced by upregulated ATF4. Slc7a11^sut/sut MuSCs exhibited higher protein levels of asparagine synthase (ASNS) and elevated abundance of asparagine (Supplementary Fig. 4E and F).

3.5. Increased proline biosynthesis in the absence of functional xCT

Proline was among the most significantly increased metabolites in Slc7a11^sut/sut MuSCs compared to WT. Proline biosynthesis involves the intermediate, pyrroline-5-carboxylate (P5C), which is produced from glutamate by P5C synthase (P5CS) and subsequently converted to proline by P5C reductase (PYCR) [57] (Fig. 5A). While the ¹³C-labelled glucose labelling to TCA cycle-related metabolites, αKG and glutamate, were lower in Slc7a11^sut/sut MuSCs (Fig. 5B and C), increased levels of proline were revealed by the metabolomic profiling and the stable isotope approaches (Fig. 5, Fig. 3G). Furthermore, Slc7a11^sut/sut MuSCs displayed an increased labelled proline:glutamate ratio (Fig. 5E). These results suggest an increased relative efflux of carbons from the TCA cycle toward proline biosynthesis via the glutamate-P5C-proline axis in Slc7a11^sut/sut MuSCs. This is further supported by higher protein levels of P5CS and higher levels of urea cycle intermediates, including ornithine and citrulline (Fig. 5, Fig. 3G). However, the protein levels of enzymes involved in the last step of proline biosynthesis, pyrroline-5-carboxylate reductases 1 and 2 (PYCR1 and PYCR2), were comparable between both genotypes (Fig. 5F). Levels of proline dehydrogenase (PRODH) and pyrroline-5-carboxylate dehydrogenase (P5CDH), which are enzymes responsible for the degradation of proline and P5C, respectively, were comparable between the two groups (Fig. 5G). The proline:glutamate ratio was found to negatively correlate with cystine uptake (Pearson, R² = 0.34, p = 0.06, Supplementary Figure 5A) and GSH abundance (Pearson, R² = 0.48, p = 0.012, Supplementary Figure 5B). These results are consistent with the conclusion that lower cystine uptake and decreased GSH levels in Slc7a11^sut/sut MuSCs promote proline synthesis from glutamate.

Fig. 5.

Increased proline biosynthesis in the absence of functional xCT

(A) Diagram showing glutamate oxidative and reductive metabolism. Enzymes and labelled metabolites are colored orange (higher in Slc7a11^sut/sut MuSCs), blue (lower in Slc7a11^sut/sut MuSCs), and black (similar between Slc7a11^sut/sut and WT MuSCs). (B–D) [U–¹³C]-glucose labelling (12 h) of α-ketoglutarate, glutamate, and proline. (E) Labelled proline to glutamate ratio as proxy of glucose-labelled glutamate to proline flux. (F–H) Immunoblots of (F) proline synthesis (P5CS, PYCR1, and PYCR2), (G) proline catabolism (PRODH and P5CDH), and (H) glutamate synthesis (GS, GLS1, GLS2, and GLUD1) enzymes. The statistical significance of the differences between groups was determined using a two-tailed Student's t-test, n = 6. Results are presented as mean ± SEM, ∗p < 0.05.

Proline and glutamine metabolism are closely interconnected through glutamate and its derivative P5C [58]. Therefore, we immunoblotted for various enzymes that mediate the interconversion of glutamine and glutamate such as glutamine synthase (GS) and glutaminase (GLS) [59]. The levels of GS, which catalyzes glutamine biosynthesis from glutamate, were lower in Slc7a11^sut/sut MuSCs (Fig. 5H). There were similar GLS1 and GLS2 protein levels in WT and Slc7a11^sut/sut MuSCs (Fig. 5H), supporting the conclusion that glutamine degradation into glutamate was comparable between groups. Thus, glutamine accumulation in Slc7a11^sut/sut MuSCs does not appear to be related to impaired glutaminolysis. Glutamate can also be converted to α-ketoglutarate (α-KG) to enter the TCA cycle in a reaction catalyzed by glutamate dehydrogenase (GLUD1) [59]. Despite lower ¹³C-labelled flux of metabolites into the TCA cycle in Slc7a11^sut/sut MuSCs, GLUD1 protein levels were not different between groups (Fig. 5G). Altogether, these findings are consistent with the conclusion that glutamate in xCT-deficient MuSCs primarily serves as a precursor for proline biosynthesis rather than an anaplerotic substrate for the TCA cycle.

4. Discussion

In response to impaired nutrient uptake, cells attempt to compensate and adapt, sometimes effectively, and other times inadequately, leading to cellular dysfunction. Here, we have examined the impact of impairments in the plasma membrane amino acid transporter xCT, which is responsible for cellular cystine uptake and glutamate efflux. We previously showed that xCT deficiency in vivo in mice disrupts muscle GSH redox but enhances their MuSC activation and myogenic differentiation following cardiotoxin-induced muscle injury [16]. Using metabolomic, bioenergetic, imaging, and molecular approaches in vitro, we demonstrate that xCT deficiency or chemical inhibition leads to impaired cystine import, resulting in disrupted GSH redox balance. This perturbed GSH redox is associated with compromised mitochondrial structure and function, characterized by mitochondrial fragmentation and a greater reliance on glycolytic metabolism to produce ATP. Extensive metabolomic analyses then revealed that xCT deficiency induces broad metabolic reprogramming as evidenced by upregulated de novo cysteine and serine synthesis, and elevated proline biosynthesis. Despite the metabolic reprogramming elicited by xCT dysfunction, redox dysfunction persists and is accompanied by impairments in cellular energetics and mitochondrial dynamics.

It is well recognized that cellular levels of cysteine can be derived from GSH degradation, protein catabolism, and methionine metabolism; however, dietary cystine uptake remains the primary source of cellular cysteine [60]. In plasma, approximately 65 % of cysteine is bound to proteins via S-cysteinylation, 30 % circulates as cystine, and only 5 % exists as free reduced cysteine [61]. While reduced cysteine enters cells through non-specific transporters such as ASCTs and EAATs, cystine, the predominant form in circulation, is exclusively imported via the cystine/glutamate antiporter xCT [62,63]. Our findings of decreased intracellular levels of cysteine and GSH in Slc7a11^sut/sut MuSCs further highlight the importance of xCT in regulating cysteine and GSH metabolism. Beyond serving as a rate-limiting precursor for GSH synthesis, cysteine plays a central role in metabolism through its degradation, generating H₂S and other organic intermediates that contribute to carbon metabolism and serve as substrates for the TCA cycle [61]. The importance of H₂S in muscle metabolism was recently highlighted by Sprenger et al. [64]. Additionally, cysteine can be catabolized via cysteine dioxygenase, further influencing cellular metabolic pathways [65]. Cysteine plays another metabolic role through GSH redox, regulating mitochondrial remodeling in response to variable bioenergetic needs [39]. Moreover, mitochondrial cysteine is essential for preserving mitochondrial integrity by supporting the synthesis of ETC proteins and iron-sulfur clusters, as reported in non-small cell lung cancer [66]. Thus, the overall aim of this work was to examine the impact of impaired or inhibited xCT activity on metabolic pathways, cellular bioenergetics, and mitochondrial morphology in MuSCs.

When mitochondrial health is disrupted, cells can initiate a stress response to promote recovery. To buffer cellular metabolism, dysfunctional mitochondria induce a reparative pathway that shifts energy metabolism to glycolysis in an ATF4-dependent manner [40]. We found that Slc7a11^sut/sut MuSCs display lower bioenergetic flux overall, but a higher proportional glucose uptake and a greater reliance on glycolytic metabolism to produce ATP. In addition to greater reliance on glycolysis, higher ATF4 protein levels in Slc7a11^sut/sut MuSCs are consistent with previous findings in mitochondrial myopathy mouse models, where ATF4-mediated response shifts cellular oxidative metabolism to favor glycolysis [67]. These findings suggest that xCT deficiency in MuSCs leads to an adaptive metabolic response, causing increased reliance on glycolysis and other metabolic alterations that are discussed in subsequent sections.

To support cell resilience, mitochondrial morphology can change in response to various cellular stressors, including oxidative stress, nutrient deprivation, and hypoxia [68]. Cellular mitochondrial networks continuously undergo remodeling to regulate metabolism and coordinate complex signaling pathways involved in cell pluripotency, proliferation, differentiation, and senescence [[69], [70], [71], [72]]. For example, mitochondrial hyperfusion can occur as an acute stress response to preserve mitochondrial oxidative capacities and protect against apoptosis [73]. However, when stress conditions are prolonged or exceed the intracellular antioxidant capacities, mitochondrial fission is triggered to eliminate damaged mitochondria and to maintain mitochondrial homeostasis [69,74,75]. Oxidative stress can also alter mitochondrial architecture by oxidizing structural components of the mitochondrial membranes, including cardiolipin, which has been shown to compromise mitochondrial integrity [76]. Mitochondrial redox and more specifically, GSH redox, are key regulators of mitochondrial dynamics, and multiple proteins that interact with fusion/fission proteins, such as CDK5, Parkin, PKA/AKAP1, and ROMO1, contain redox-sensitive residues [39,77]. While we initially hypothesized that the high GSSG observed in Slc7a11^sut/sut MuSCs would promote mitochondrial hyperfusion and MFN1/2 oligomerization, mitochondria in Slc7a11^sut/sut MuSCs were more fragmented with an upregulation of DRP1 oligomerization. These findings align with observations in C2C12 myoblasts, where H₂O₂-induced oxidative stress triggers DRP1-mediated mitochondrial fragmentation, leading to a loss of mitochondrial membrane potential and decreased cellular respiration [78]. Others found that DRP1-mediated fission is tightly regulated by redox signaling, where oxidative stress induces oxidation of the Cys⁶⁴⁴ residue in DRP1, initiating DRP1 GTPase oligomerization and activity [79,80]. Hong et al. showed that DRP1-mediated mitochondrial fission is essential for MuSC activation, supporting the quiescence-to-proliferation transition [81]. Conversely, excessive mitochondrial fission is associated with cellular senescence [78,[82], [83], [84], [85]]. Additionally, Slc7a11^sut/sut MuSCs displayed lower levels of L-OPA1 along with a trend for an increase in S-OPA1 protein levels. S-OPA1 accumulation resulting from L-OPA1 cleavage may serve as a protective mechanism in Slc7a11^sut/sut MuSCs against oxidative stress, as reported in mouse embryonic fibroblasts (MEFs) [45]. Importantly, Liang et al. demonstrated that OPA1 plays a role in mediating mitochondrial ROS generation and suppressing ATF4 in MEFs and human U2OS osteosarcoma cells [86]. Therefore, based on these findings, OPA1 cleavage may be a contributing factor to ATF4 induction observed in Slc7a11^sut/sut MuSCs.

xCT has been extensively studied in the brain due to the role of glutamate as a key excitatory neurotransmitter [87,88]. Glutamate is also a crucial metabolite in muscle as it participates in numerous metabolic pathways [89]. In Slc7a11^sut/sut MuSCs, we observed altered amino acid metabolism, including a pronounced enrichment in BCAA metabolism. These findings are consistent with observations in cancer cells, which adapt their amino acid metabolism to regulate intracellular glutamate levels by suppressing its biosynthesis or enhancing its utilization [90]. BCAAs are oxidized in skeletal muscle by BCATs, converting them into branched-chain α-keto acids through the reversible transfer of an amino group to α-KG, resulting in glutamate production [91]. Our findings revealed that Slc7a11^sut/sut MuSCs suppress glutamate biosynthesis from BCAAs, as evidenced by lower protein levels of BCAT2. Another potential source of glutamate is through GLS, which catalyzes glutamine degradation into glutamate [92]; however, we did not observe any changes in GLS protein in Slc7a11^sut/sut MuSCs. Thus, our findings are consistent with the conclusion that glutamine degradation is not a factor here, though it is important to note that protein levels alone do not reflect the enzymatic activity of those proteins. However, our metabolomics data did reveal an accumulation of glutamine in Slc7a11^sut/sut MuSCs, consistent with lower glutamine catabolism. In line with our findings, Muir et al. showed that extracellular cystine and xCT expression are critical for facilitating glutamine metabolism in cancer cells [90]. It is also important to note that protein levels of GS were lower in Slc7a11^sut/sut MuSCs, which may prevent glutamine accumulation, a phenomenon also observed in C2C12 cells [93].

Beyond its well-established role in supporting cellular redox via the xCT-GSH axis, xCT is increasingly recognized for its crucial role in adapting to the altered metabolic demands of proliferating cancer cells. Specifically, high xCT activity in cancer cell lines can promote glucose dependency and glutamine utilization to support rapid growth and survival under nutrient-stress conditions [[94], [95], [96], [97], [98]]. Therapeutically, erastin-mediated xCT inhibition mitigates tumor cell proliferation by inducing ferroptosis, which is an iron-dependent mechanism of cell death driven by lipid peroxidation [99]. Despite extensive literature highlighting the importance of xCT in cancer cell survival and metabolism, its role in controlling metabolic pathways in muscle cells remains largely unexplored [100]. Interestingly, we observed distinct metabolic clustering between WT and Slc7a11^sut/sut MuSCs, with xCT deficiency driving metabolites to converge to a specialized metabolic cluster. Similar metabolic convergences have been observed in mutant SOD1 ALS mice with hypermetabolism, reflecting decreased metabolic flexibility as an adaptive mechanism to cellular stress [101]. Our findings highlight the critical role of xCT in reshaping amino acid metabolism and driving metabolic reprogramming in MuSCs.

As a reparative metabolic response to mitochondrial stress, cells upregulate glycolysis, directing glucose-derived carbons into de novo serine biosynthesis. This process fuels one-carbon metabolism, including the folate and methionine cycles, thereby supporting de novo cysteine biosynthesis via reverse transsulfuration [54]. Our findings showed increased de novo serine biosynthesis in Slc7a11^sut/sut MuSCs, as evidenced by higher m+3 serine labelling and higher protein levels of PSAT1 and PSPH. Additionally, we found that Slc7a11^sut/sut MuSCs actively used the transsulfuration pathway to produce cysteine endogenously, as indicated by higher CTH protein levels and comparable m+3 GSH labelling between groups. These observations align with results in tumor cells cultured in cystine-depleted media, where cysteine deprivation resulted in increased CBS and CTH protein levels in an ATF4-dependent manner to sustain cell proliferation and GSH biosynthesis [102]. It is well known that enhanced flux through the transsulfuration pathway produces hydrogen sulfide (H₂S), which, at low concentrations, can promote mitochondrial biogenesis and enhance mitochondrial bioenergetics in a sulfide quinone reductase (SQR)-dependent manner [103,104]. However, at high concentrations, H₂S can inhibit complex IV, decrease cell proliferation, and shift metabolism towards reductive carboxylation, further linking redox stress to mitochondrial dysfunction [105,106]. Nonetheless, the work of Sprenger et al. recently showed the importance of ergothionine in the control of muscle mitochondrial metabolism and exercise performance through the activation of the H₂S-producing enzyme 3-mercaptopyruvate sulfurtransferase (MPST) [64]. MPST produces pyruvate and H₂S in mitochondria to support mitochondrial respiration [103,107]. Therefore, increased reliance on the transsulfuration pathway for de novo cysteine synthesis may also contribute to impaired mitochondrial respiration in Slc7a11^sut/sut MuSCs. While the Seahorse mitochondrial stress test was performed in standard assay medium containing glucose, glutamine, and pyruvate, future studies can be conducted under a variety of metabolic states to more fully elucidate how xCT dysfunction impacts cellular bioenergetics.

Glutamate, the most abundant intracellular amino acid, can serve as a key precursor for TCA cycle anaplerosis through α-KG. Additionally, glutamate can support proline biosynthesis via its conversion to P5C [108]. Our findings of lower mitochondrial respiration and impaired glycolytic flux into the TCA cycle in Slc7a11^sut/sut MuSCs suggest that glutamate is not being oxidized through the TCA cycle but is instead redirected toward reductive proline biosynthesis. Indeed, we found higher proline abundance and increased protein levels of P5CS in Slc7a11^sut/sut MuSCs. Our findings are consistent with increased glutamate flux toward proline biosynthesis observed in neurons treated with the psychostimulant methamphetamine, where proline biosynthesis serves as a protective mechanism to mitigate glutamate accumulation [109]. However, this is an energetically costly process with glycolysis serving as the primary energy source for this process [110,111]. In this context, proline biosynthesis provides an essential supply of NAD⁺ crucial for sustaining efficient glycolysis, a key hallmark of the reparative metabolic response to mitochondrial stress [112,113]. Thus, proline biosynthesis may offer a bioenergetic advantage for Slc7a11^sut/sut MuSCs, supporting their metabolic dependence on glycolysis to maintain intracellular glutamate homeostasis.

The proline cycle (i.e., the interconversion of proline and P5C) plays a key role in maintaining cellular redox homeostasis. We found that Slc7a11^sut/sut MuSCs favor proline synthesis over degradation by increasing P5CS protein levels, while PRODH protein levels remain comparable between groups. This observation is consistent with the fact that proline catabolism, catalyzed by PRODH, generates ROS by transferring electrons to FAD and converting oxygen into superoxide, while proline biosynthesis is a reductive process that mitigates ROS production and promotes cell survival by preventing apoptosis [114,115]. When cells are exposed to nutrient deprivation or oxidizing agents, P5CS is upregulated and diffusely localized within mitochondria, where it forms oligomerized filament-like structures [114]. Through the proline cycle and the mitochondrially localized P5CS, mitochondria can be hubs for competing metabolic pathways such as OXPHOS and reductive proline biosynthesis [107]. Ryu et al. demonstrated that P5CS, along with mitochondrial fusion/fission, regulate the balance between OXPHOS and reductive proline biosynthesis by creating distinct mitochondrial subpopulations: one enriched in ATP synthase for OXPHOS, and another enriched in P5CS which prioritizes reductive biosynthesis by suppressing TCA cycle activity and glutamate oxidation [116]. Similarly, our findings of enhanced P5CS-mediated proline biosynthesis accompanied by increased DRP1 oligomerization in Slc7a11^sut/sut MuSCs suggest that Slc7a11^sut/sut MuSCs may employ proline reductive biosynthesis to suppress oxidative metabolism and counteract elevated H₂O₂ levels. Altogether, proline metabolism emerges as a crucial adaptive mechanism in xCT-deficient cells, enabling MuSCs to balance between oxidative and reductive pathways.

In conclusion, our research demonstrates that xCT-mediated cystine import regulates GSH redox and promotes de novo serine, cysteine, and proline synthesis in proliferating muscle cells. In many muscular dystrophies and metabolic diseases, MuSC health is compromised due to disrupted redox and metabolic homeostasis [[117], [118], [119], [120], [121]]. Therefore, a deeper understanding of the metabolic role of xCT in muscle cells provides valuable insights into the crosstalk between cellular redox regulation and metabolic homeostasis. Our findings not only enhance our understanding of the fundamental mechanisms governing xCT function but also highlight cyst(e)ine metabolism as a therapeutic target to restore disrupted redox balance and metabolic homeostasis often associated with muscle-related diseases and pathologies.

CRediT authorship contribution statement

Michel N. Kanaan: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Charbel Y. Karam: Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing, Conceptualization, Project administration, Resources, Supervision. Luke S. Kennedy: Data curation, Formal analysis, Methodology, Software, Validation, Visualization. Chantal A. Pileggi: Conceptualization, Project administration, Validation, Visualization, Writing – original draft. Lauren Hamilton: Data curation, Investigation, Methodology. Miroslava Cuperlovic-Culf: Conceptualization, Data curation, Methodology, Supervision, Validation, Visualization. Mary-Ellen Harper: Funding acquisition, Project administration, Resources, Supervision.

Funding

This work was supported by a grant from the Natural Sciences and Engineering Research Council (NSERC) of Canada (RGPIN-2020-04468 to MEH).

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Mary-Ellen Harper reports financial support was provided by University of Ottawa. Mary-Ellen Harper reports financial support was provided by Natural Sciences and Engineering Research Council of Canada. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

Fig. 1(A) Immunoblot of NRF2 in WT and Slc7a11^sut/sut MuSCs. The statistical significance of the differences between groups was determined using a two-tailed Student's t-test, n = 5–6. Results are presented as mean ± SEM.

Fig. 2. (A–B) Extracellular acidification rates (ECAR) were measured in primary WT and Slc7a11^sut/sut MuSCs during a glycolytic stress test. (C–G) Immunoblots of (C) mitochondrial OXPHOS complexes, and (D) DRP1, (E)p-DRP1 at S616 and S637 residues, (F) MFN1/2, and (G) OPA1 short and long isoforms in WT and Slc7a11^sut/sut MuSCs. (H–I) Immunoblots of (H) MFN1/2 oligomerization, and (I) DRP1 oligomerization in primary WT and Slc7a11^sut/sut MuSCs. The statistical significance of the differences between groups was determined using a two-tailed Student's t-test, n = 4–6. Results are presented as mean ± SEM, ∗p < 0.05, ∗∗p < 0.01.

Fig. 3. (A) Variable influence on projection (VIP) scores for metabolite features in OPLS-DA classification of WT and Slc7a11^sut/sut samples, metabolites with VIP scores >0.75 are shown. (B) Q² scores of OPLS-DA models constructed using metabolite feature subsets defined by minimum VIP score thresholds. Vertical lines indicate VIP threshold determined by peak predictive performance (Q²) in subset models before sustained decrease. (C) SAM plot of WT and Slc7a11^sut/sut groups. (D) Hierarchical clustering of WT and Slc7a11^sut/sut metabolomics data. Dendrogram leafs and linkages are colored according to the assigned cluster (Euclidean distance threshold t = 4).

Fig. 4. (A) Kinetic time-course of [U–¹³C]-glucose labeling of relevant metabolites for 5 min, 30 min, 1 h, 2 h, 6 h, 12 h, and 24 h in Slc7a11^sut/sut primary MuSCs, n = 1, expressed as labelled isotopologue percentage or percentage of labelled carbon units within metabolites. (B) [U–¹³C]-glucose labelling (12 h) of TCA cycle intermediates (citrate, succinate, and fumarate) expressed as a percentage of labelled carbon units within metabolites derived from ¹³C precursor in WT and Slc7a11^sut/sut MuSCs. (C) [U–¹³C]-glucose labelling (12 h) of glycine m+2, expressed as labelled percentage of total glycine levels. (D–E) Immunoblot of (D) ATF4 and (E) ASNS in WT and Slc7a11^sut/sut MuSCs. (F) Asparagine abundance in WT and Slc7a11^sut/sut MuSCs. Comparisons between groups were determined using a two-tailed Student's t-test, n = 6 (B–F). Results are presented as mean ± SEM, ∗p < 0.05, ∗∗p < 0.01.

Fig. 5. (A) Correlation analysis between cystine uptake and labelled proline/glutamate ratio. (B) Correlation analysis between GSH abundance and labelled proline/glutamate ratio.

Data availability

Requests should be directed to mharper@uottawa.ca.