Cysteine metabolism dysregulation promotes ferroptosis in osteoarthritis: Insights from multi-species bioinformatics and experimental validation

Abstract

Objective

Osteoarthritis (OA) was traditionally viewed as a mechanically driven disorder. However, emerging evidence implicated metabolic dysregulation in its pathogenesis. This study investigated the novel interaction among cysteine metabolism, ferroptosis, and OA progression, while evaluating the therapeutic efficacy of L-cysteine supplementation.

Methods

Integrated transcriptomic analyses of murine (GSE112641), rat (GSE118559), and human (GSE114007) OA cartilage datasets identified conserved metabolic perturbations. Proteomic and metabolomic profiling of human OA cartilage validated pathway-level dysregulation. Mechanistic validation employed IL-1β-stimulated inflammatory chondrocytes under cysteine-depleted conditions, with parallel assessment of ferroptosis markers, including expression of GPX4, SLC7A11 and TFRC, glutathione metabolism, iron accumulation, and lipid peroxidation. Therapeutic potential was further tested in a surgery-induced OA murine model receiving L-cysteine administration.

Results

Cross-species analysis revealed cysteine/glutathione metabolism as the most consistently dysregulated pathway in OA. Multi-omics profiling demonstrated dysregulation of cysteine/glutathione metabolism and ferroptosis in OA. In vitro cysteine deprivation triggered ferroptosis hallmarks in inflammatory chondrocytes, including viability reduction, malondialdehyde (MDA) increase, and glutathione depletion. L-cysteine supplementation reversed these effects, restoring viability and normalizing redox balance. The in vivo study results further demonstrated that L-cysteine supplementation reduces cartilage degeneration severity by protecting against ferroptosis and lipid peroxidation.

Conclusion

This study establishes cysteine metabolism as a master regulator of ferroptosis in OA pathogenesis. The mechanistic chain from cysteine depletion to glutathione collapse, iron overload, and lipid peroxidation explains chondrocyte loss patterns observed clinically. L-cysteine supplementation emerges as a dual-action therapy, simultaneously addressing oxidative stress and ferroptosis.

Translational potential

This study underscores the translational potential of L-cysteine supplementation as a novel therapy for OA. By targeting the dysregulated cysteine/glutathione metabolism and its role in ferroptosis, L-cysteine presents a dual-action approach to mitigate cartilage degeneration. Future clinical trials should evaluate its efficacy and safety in diverse patient populations, focusing on both clinical outcomes and metabolic markers. This research paves the way for a redefined management strategy for OA as a metabolically driven disorder.

Graphical abstract

1. Introduction

Osteoarthritis (OA), a common degenerative joint disorder, is mainly characterized by progressive cartilage destruction, subchondral bone remodeling, and synovial inflammation, which impacting a global population exceeding 300 million [1,2]. As a leading cause of chronic pain and disability in aging populations, OA imposes substantial socioeconomic burdens. However, its molecular pathogenesis remains elusive [3,4]. Traditional paradigms emphasize biomechanical stress and low-grade inflammation as primary drivers, while emerging evidence positions metabolic dysregulation as a critical pathogenic axis that intersects with these mechanisms [[5], [6], [7]].

Current therapeutic strategies, ranging from NSAIDs and intra-articular corticosteroids to joint replacement, primarily address symptomatic relief rather than disease modification [8,9]. This therapeutic gap resultants from an incomplete understanding of underlining regulation mechanism of OA, particularly the metabolic reprogramming observed in chondrocytes and synovial tissues. Recent advances have revealed that OA progression involves profound alterations in energy metabolism, lipid handling, and amino acid utilization, with cysteine metabolism emerging as a pivotal regulator of redox homeostasis and cellular survival [[10], [11], [12], [13]].

The centrality of cysteine metabolism in OA pathophysiology arises from its dual role in maintaining glutathione (GSH) biosynthesis and modulating ferroptosis, which is an iron-dependent cell death pathway characterized by lethal lipid peroxidation. Mechanistically, cysteine serves as the rate-limiting substrate for GSH synthesis, the primary cellular antioxidant [14]. Its depletion disrupts redox balance, rendering chondrocytes vulnerable to oxidative stress while simultaneously activating ferroptosis cascades through GPX4 inactivation and iron overload [15]. This metabolic vulnerability is exacerbated in OA by chronic inflammation and mitochondrial dysfunction, creating a self-perpetuating cycle of cartilage degradation [14,16].

Despite growing recognition of the metabolic dimension of OA, the specific pathways and molecular mechanisms by which metabolic disturbances contribute to cartilage degeneration remain poorly defined. In particular, the interplay between cysteine metabolism, oxidative stress, and regulated cell death pathways in chondrocytes is not fully understood. While individual studies have implicated disrupted redox homeostasis and ferroptosis in OA, a systematic and cross-species exploration of cysteine metabolism and its downstream effects is lacking. Furthermore, it remains unclear whether targeting metabolic vulnerabilities, such as cysteine depletion, can provide therapeutic benefit beyond symptomatic relief. Addressing these unresolved questions is essential for advancing the conceptual framework of OA and identifying novel avenues for disease-modifying interventions.

This study employs an integrative translational approach to address these gaps. Through multi-tiered omics investigation, including cross-species bioinformatics analysis of cartilage transcriptomes (murine: GSE112641, rat: GSE118559, human: GSE114007), human cartilage proteomics and metabolomics, combined with in vitro and in vivo functional validation, we systematically investigate how cysteine metabolic rewiring drives ferroptosis in OA. We further evaluate L-cysteine supplementation as a novel disease-modifying strategy, demonstrating its capacity to restore redox balance, inhibit iron overload, and preserve cartilage integrity. Our findings reposition OA as a metabolic disorder with actionable therapeutic targets, bridging the gap between mechanistic understanding and clinical translation.

2. Materials and methods

2.1. Ethical approval and patient inclusion

This study was approved by the Ethics Committee of the authors' institution (Approval No. 2024-LP-032). Patients with osteoarthritis (n = 3) were recruited from the Orthopedics Department and were eligible for inclusion if they met the following criteria: 1) aged between 60 and 80 years. 2) clinically diagnosed with moderate to severe osteoarthritis as per the criteria of American College of Rheumatology. 3) scheduled to undergo total knee arthroplasty. Written informed consent was obtained from all participants. The demographic details of human patients were summarized in Supplementary Table S1.

2.2. Sample collection and processing

Cartilage tissues were harvested from the femoral condyle of osteoarthritis patients undergoing total knee arthroplasty in the Orthopedics Department, following the institutional regulations and procedures for sample collection. For each patient, paired cartilage samples were collected from non-weight-bearing intercondylar groove region (serving as control) and central weight-bearing region of the medial femoral condyle (representing the OA group) of the same joint. Immediately after resection, tissues were rinsed with ice-cold PBS, dissected into 5 × 5 × 2 mm³ full-thickness fragments under sterile conditions, and snap-frozen in liquid nitrogen within 10 min. All samples designed for Proteomics and Metabolomics analyses were stored at −80°C until shipment to BGI Genomics (BGI-Shenzhen, China) on dry ice for further processing. For histological study, cartilage samples are fixed in a solution like 4% paraformaldehyde and reserved for further section and histological staining.

2.3. Bioinformatics analysis

Three independent OA cartilage transcriptomic datasets (GSE112641 [17], GSE118559 [18], GSE114007 [19]) were retrieved from the Gene Expression Omnibus (GEO) database. GSE112641 (Illumina HiSeq 2000) contained 5 OA and 5 healthy control knee cartilage from sham non-injury C57BL/6J mice, GSE118559 (Illumina NextSeq 500) included 3 OA and 2 control knee joint cartilage from healthy control group of rats, while GSE114007 (Illumina HiSeq 2000) comprised 20 OA and 18 normal human knee cartilage tissues from donors without history of joint disease or trauma. Differential gene expression analysis was performed using OmicsBean web server (http://www.omicsbean.cn/). Conserved metabolic pathways were identified through KEGG enrichment analysis (OmicsBean platform) and cross-species Venn diagram visualization. Proteomics data processing was conducted using the OmicsBean web server (http://www.omicsbean.cn/), while metabolomics data were processed through the MetaboAnalyst web tool (https://www.metaboanalyst.ca/). Throughout the multi-omics statistical analyses, a selection threshold was set at absolute value of log2FC greater than 0.26, corresponding to a FC greater than 1.2 or less than 0.833, with a P-value threshold of less than 0.05 to identify statistically significant differences. The rationale for choosing the above FC threshold is twofold. First, articular cartilage has low metabolic turnover, and OA-related pathophysiological changes are often subtle; chondrocytes comprise only 2–5% of tissue volume, so many OA-associated genes exhibit modest fold changes in bulk tissue analyses. A stringent cutoff such as logFC >1 would miss functionally important but quantitatively modest changes, as exemplified by the transporter SLC7A11. Second, differential expression was only the initial filtering step: genes with logFC >0.26 but unadjusted p > 0.05 were excluded. Subsequent pathway enrichment, protein-protein interaction network analysis, and cross-dataset comparison served as additional filters to remove spurious signals. Only genes that appeared in at least two of the three datasets or were consistently enriched in ferroptosis-related pathways were retained for further analysis.

2.4. In vitro inflammatory chondrocyte model

Primary chondrocytes were isolated from knee articular cartilage of 4-week-old male C57BL/6 mice (n = 4) following the previous protocol [19]. Briefly, the murine primary chondrocytes were collected and digested through enzymatic digestion with 0.2% collagenase II (Sigma–Aldrich) in Dulbecco's Modified Eagle Medium (DMEM; Gibco) for 4 h at 37°C under 5% CO₂ atmosphere. Isolated cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin (Thermo Fisher) until passage 2. P2 chondrocytes were used due to low initial yield (8.2 × 10³±1.5 × 10³ cells/joint). SOX9 expression remained consistently high (Supplementary Fig. S1), validating phenotype retention. To establish an in vitro OA model, P2 chondrocytes were stimulated with 10 ng/mL recombinant murine IL-1β (PeproTech) for 24 h. To model L-cystine/cysteine metabolism dysfunction in osteoarthritis, we induced extracellular cystine deficiency using customized L-cystine-free DMEM (Thermo Fisher). This approach leverages the physiological hierarchy where extracellular L-cystine (oxidized dimer) serves as the primary precursor for intracellular L-cysteine (reduced monomer) via reduction after SLC7A11-mediated transport. By depleting extracellular L-cystine, we recapitulate the intracellular L-cysteine deficiency observed in OA, since L-cystine and L-cysteine function in the same integrated pathway. Although our model specifically targets L-cystine deficiency, it effectively induces comprehensive L-cyst(e)ine metabolism dysfunction. We note that commercial DMEM formulations are exclusively available as cystine-free (not cysteine-free), as L-cysteine is unstable in cell culture media and typically auto-oxidizes to L-cystine. Rescue experiments supplemented 200 μM L-cystine (Sigma) to bypass the transport defect and directly restore intracellular cysteine pools.

2.5. In vivo OA model

Osteoarthritis was induced in 8-week-old male C57BL/6 mice (n = 6 per group) through anterior cruciate ligament transection combined with destabilization of the medial meniscus (ACLT + DMM) surgery following the previously reported procedures under isoflurane anesthesia [20]. The animals were randomized into four experimental groups: sham-operated controls receiving phosphate-buffered saline (PBS), the OA model group receiving PBS (OA), and the two treatment groups with low-dosage and high-dosage treatment, respectively (OA+15 mg/kg L-cysteine or OA+30 mg/kg L-cysteine administered by oral gavage starting from week 2 after surgery, every other day for another 4 weeks). All the animals were randomized into each group and concentration in accordance with the random number table. All the animals were kept in the same environment, and minimized the potential confounder. At the conclusion of the experiment, knee joints were harvested for histological and immunohistochemical analyses. All animal experiments were approved by the Ethics Committee of Chinese University of Hong Kong and performed in accordance with the Code of Ethics of the World Medical Association.

2.6. Cell viability assays

Cell viability was measured in chondrocytes with the MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide) assay. MTT solution (2.5 mg/mL) (Invitrogen) was added, followed by dimethyl sulfoxide solution (Sigma Aldrich), and absorbance was read using Benchmark Microplate Reader (Bio-Rad).

2.7. Real time-PCR

All procedure was performed as described previously [21]. Total RNA from chondrocytes or cartilage tissues were extracted using Tissue Total RNA Mini Kit (Favorgen) and reverse transcribed by High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). RT-PCR was performed through QuantStudio 7 Flex Real-Time PCR System with a 384-well block module (Applied Biosystems) using Power SYBR® Green PCR Master Mix (Applied Biosystems). Quantitative real-time PCR (qRT-PCR) was performed to measure the expression of ferroptosis-related genes, including Slc7a11, Slc3a2, and Gpx4, using SYBR Green Master Mix on a StepOne™ Real-Time PCR System (Fisher Scientific). GAPDH was used as an internal control for normalization. Relative expression was calculated using the 2^−ΔΔCt method. Results are presented as mean ± 95% confidence interval. Primer sequences are listed in Supplementary Table S2.

2.8. Western blot analysis

All procedures were performed as described previously [21]. Cells and cartilage tissues were homogenized in RIPA lysis buffer (Millipore) with 1% protease inhibitor and phosphatase inhibitor (Thermo Fisher Scientific). Protein was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to 0.45 μM PVDF membrane. The primary antibodies used in this study include: rabbit anti-GPX4 (1:1,000, CST59735, Cell Signaling Technology), rabbit anti-GLC7A11 (1:500, MA5-44963, Thermo Fisher), rabbit anti SLC3A2 (1:500, 15193-1-AP, Proteintech), and rabbit anti-GAPDH (1: 5,000, CST2118, Cell Signaling Technology). Secondary antibodies include horseradish peroxidase (HRP)-conjugated Goat anti-rabbit IgG (Cell Signaling Technology) and anti-murine IgG (Invitrogen) (dilution 1:5000). The target protein level was normalized with the protein level of the internal loading control GAPDH and expressed as a fold of control.

2.9. Biochemical analysis

For analysis of GSH, H2O2 and MDA level in chondrocytes, the cultured chondrocytes were harvested by washing with PBS and lysing them with the provided lysis buffer from each kit, respectively. For cartilage analysis, the cartilage tissue was dissected, liquid nitrogen frozen and homogenized in cold buffer or lysis solution provided in the kit with a pestle and mortar. The homogenate was clarified by centrifugation to remove debris. The GSH levels were assessed through GSSG/GSH detection kit (Beyotime). The H₂O₂ levels were detected using the hydrogen peroxide assay kit (Beyotime). The extent of lipid peroxidation was quantified by measuring malondialdehyde (MDA) levels using the Lipid Peroxidation MDA Assay Kit (Beyotime). The intracellular iron concentration was determined using the Iron Assay Kit (ScienCell).

2.10. Histological analysis

Human and mice cartilage tissues were fixed in 4% paraformaldehyde under 4°C for 48 h, followed by decalcification in 10% EDTA for six and three weeks, respectively, and subsequently embedded in paraffin. Sections of 7-μm thickness were stained with safranin O/fast green and toluidine blue to evaluate cartilage morphology and proteoglycan content. Human cartilage degradation was evaluated by MANKIN score analysis (Supplementary Fig. S2). Mice cartilage degradation was assessed using the Osteoarthritis Research Society International (OARSI) scoring system following the published protocol [22].

2.11. Immunohistochemistry staining

Immunohistochemistry (IHC) staining was conducted following standard protocols. Primary antibodies were applied, including rabbit anti-GPX4 (1:100, Ab125066, Abcam, USA), rabbit anti-SLC7A11 (1:100, ab307602, Abcam, USA), and rabbit anti-TFRC (1:100, Ab214039, Abcam, USA), and incubated overnight at 4°C. After washing with phosphate-buffered saline (PBS), the sections were incubated with a secondary antibody, goat anti-rabbit IgG H&L (ab97047, Abcam, Cambridge, UK). Signal visualization was achieved using the horseradish peroxidase-streptavidin system (Dako, USA), followed by counterstaining with hematoxylin. Images of selected areas were captured using a Leica DMIRB Inverted Microscope (Leica, Wetzlar, Germany).

2.12. Detection of reactive L-cysteine levels in primary chondrocytes

Following a 72h treatment period, primary chondrocytes were harvested, and lysates were prepared via mechanical homogenization on ice, followed by centrifugation at 10,000 × g for 10 min at 4°C to collect the supernatant. The intracellular L-cysteine concentration was then determined using a Cysteine Colorimetric Assay Kit (Elabscience, Wuhan, China) by sequentially adding the reagents to the supernatant, incubating at room temperature for 10 min, and measuring the absorbance at 600 nm, and the relative L‐cysteine levels were calculated.

2.13. Statistical analysis

All experiments were performed at least in triplicate and sample size for each experiment is indicated as n. The sample size was calculated using the G∗Power Calculator (http://www.gpower.hhu.de/) based on the prior studies [23]. Statistical parameters can be found in figure legends. Data were statistically analyzed by GraphPad Prism 10 software (GraphPad Inc.) using the Two-tailed student's t-test with Welsh's correction for comparing two independent groups. One-way ANOVA followed by Tukey's multiple comparison tests were used for comparing three or more independent groups. Two-way ANOVA followed by Bonferroni's multiple comparison tests were used to analyze the effects of two independent variables. Data were presented as mean ± 95% confidence interval. P < 0.05 was considered statistically different.

3. Results

3.1. Cross-species analysis identifies metabolic dysregulation in OA

To understand the molecular dynamics of OA, we examined differentially expressed genes (DEGs) from three independent cartilage tissue datasets across species, including murine (GSE112641 [17]), rat (GSE118559 [18]), and human (GSE114007 [19]). The functional enrichment in KEGG pathways uncovered important metabolic changes regularly linked with the development of OA, underscoring the importance of metabolic reprogramming in the nature of OA among different species.

Analysis of KEGG pathways divided DEGs into distinct functional groups, with metabolic pathways being prominently enriched in all datasets. Within the murine dataset (GSE11264), metabolic pathways accounted for approximately 25% of the total enriched pathways, followed by pathways related to ECM-receptor interaction and signal transduction (Fig. 1A, Supplementary Table S3). Similarly, in the rat dataset (GSE118559), metabolic pathways represented a significant proportion of enriched categories, alongside pathways involved in proteoglycan function and the MAPK signaling pathway (Fig. 1B–Supplementary Table S4). In the human dataset (GSE114007), metabolic pathways were also prominent, comprising approximately 17% of all enriched pathways, further underscoring the conserved role of metabolic dysregulation in OA (Fig. 1C–Supplementary Table S5).

Fig. 1.

Multi-dataset analyses highlight metabolic alteration in OA pathogenesis A-C. Pie charts showing the proportion of differentially expressed genes (DEGs) assigned to relevant KEGG pathways in OA vs. non-OA groups for the datasets GSE112641 (A), GSE118559 (B), and GSE114007 (C). D-F. Bar charts of the top 30 enriched KEGG pathways for DEGs in OA vs. non-OA groups for the datasets GSE112641 (D), GSE118559 (E), and GSE114007 (F). G-I. Bar charts of the top 30 enriched KEGG metabolic pathways for DEGs in OA vs. Non-OA groups for the datasets GSE112641 (G), GSE118559 (H), and GSE114007 (I).

To gain further insights, bubble plots of the top 30 enriched KEGG pathways provided additional insights into the biological processes implicated in OA. In the murine dataset (GSE112641), prominent pathways included metabolic pathways, cytoskeletal organization in muscle cells, and pathways associated with neurodegeneration and cancer (Fig. 1D–Supplementary Table S6). The rat dataset (GSE118559) showed a strong enrichment for metabolic processes and cancer-related pathways (Fig. 1E–Supplementary Table S7), while the human dataset (GSE114007) highlighted PI3K-Akt signaling pathway in cancer, and metabolic pathways as the most significantly enriched categories (Fig. 1F–Supplementary Table S8). These results suggest both shared and species-specific mechanisms in OA pathogenesis.

Further analysis of metabolic pathways demonstrated consistent alterations in amino acid metabolism, carbohydrate metabolism, and lipid metabolism across all three datasets (Fig. 1G–I, Supplementary Tables S9–S11). Notably, amino acid metabolism emerged as a consistently altered process, suggesting its critical role in OA. These findings collectively emphasize metabolic reprogramming as a key feature of OA pathogenesis, with amino acid metabolism as a conserved and critical component across species.

3.2. Cysteine dysregulation links to ferroptosis in OA

To explore the most significant metabolic alterations contributing to OA pathogenesis, we focused on shared amino acid metabolic pathways across the three datasets (GSE11264, GSE118559, and GSE114007) and investigated their potential link to ferroptosis, which is a form of regulated cell death associated with metabolic dysregulation, and is highly correlated with OA progression [24]. A Venn diagram of the top 30 enriched KEGG metabolic pathways identified six common pathways across all datasets (Fig. 2A), with glutathione metabolism and cysteine/methionine metabolism being particularly relevant due to their direct biochemical interconnection in maintaining redox homeostasis (Supplementary Table S12). Analysis of genes involved in cysteine and methionine metabolism and glutathione metabolism revealed consistent dysregulation across species. In GSE112641 (murine), key genes such as Got, Mdh, and Ldh in cysteine and methionine metabolism, as well as GPX4 in glutathione metabolism, were significantly down-regulated (Fig. 2B), suggesting the dysregulation of cysteine and glutathione metabolism. Similarly, in GSE118559 (rat), down-regulation was observed in the expression of cysteine-related genes (Kyat3) and glutathione-related genes (Gpx4) (Fig. 2C). In GSE114007 (human), additional down-regulated genes including Cth and Chac1 were observed, which are involved in both cysteine and glutathione metabolism pathways, further supporting their role in OA pathogenesis (Fig. 2D).

Fig. 2.

Cysteine dysregulation and ferroptosis are targets in OA pathogenesis A. Venn diagram showing the common significant amino acid metabolic pathways among the top 30 metabolic pathways identified in datasets GSE112641, GSE118559, and GSE114007. B-D. Representative pathways illustrating altered genes in the cysteine and methionine metabolism pathway and the glutathione metabolism pathway in OA vs. non-OA groups for the datasets of GSE112641 (B), GSE118559 (C), and GSE114007 (D). E. A summarized schematic diagram illustrating the ferroptosis cascade. F-H. Profiles of key ferroptosis-related genes (red box marked) among DEGs in OA vs. non-OA groups for the datasets of GSE112641 (F), GSE118559 (G), and GSE114007 (H).

Given the close relationship among cysteine metabolism, glutathione homeostasis, and ferroptosis [15,25], we examined the expression pattern of ferroptosis cascade related genes in OA. A simplified schematic illustrates the dependence of ferroptosis on cysteine uptake, glutathione synthesis, and lipid peroxidation regulation (Fig. 2E). Ferroptosis is triggered by reduced cysteine availability, impaired glutathione synthesis, or dysregulated antioxidant defenses, leading to lipid peroxide accumulation and cell death [25,26]. We evaluated the expression profiles of key ferroptosis-related genes across the datasets. In GSE112641 (murine), genes such as SLC39A14 (a zinc transporter linked to oxidative stress [27]) and PTGS2 (encoding COX-2, a ferroptosis marker [28,29]) were significantly up-regulated in OA cartilage (Fig. 2F), indicating the putative induction of ferroptosis. In GSE118559 (rat), dysregulation included decreased GPX4 (a critical regulator of lipid peroxidation [26]) and increased iron metabolism regulators (TF, TFRC, and SLC40A1 [30,31]) (Fig. 2G). In GSE114007 (human), the expression level of other ferroptosis-related genes, such as TF, SLC39A8, SLC38A1, and SLC3A2 [32,33], were significantly altered, further implicating the involvement of ferroptosis in OA development (Fig. 2H). Collectively, these findings suggest that dysregulation of cysteine metabolism and glutathione depletion contribute to ferroptosis, which further exuberate OA progression. Impaired cysteine availability reduces antioxidant defenses, promotes iron accumulation, and increases lipid peroxidation, driving oxidative stress and cartilage degeneration in OA.

To systematically evaluate the cross-dataset concordance of ferroptosis-related transcriptional alterations, we curated a panel of 37 ferroptosis-associated genes from the KEGG pathway database and examined their expression trends across three transcriptomic datasets representing distinct species and anatomical sites (GSE112641-murine, GSE118559-rat, and GSE114007-human). Consistency was defined as concordant directional change in at least two datasets. Although no single gene exhibited uniform dysregulation across all three datasets, pairwise comparisons unveiled evolutionarily conserved transcriptional signatures. This lack of global concordance is likely attributable to interspecies regulatory divergence and technical heterogeneity between platforms. Specifically, between the rat (GSE118559) and human (GSE114007) datasets, nine genes demonstrated congruent expression changes. Among these, CYBB, TF, DPP4, HMGCR, FADS2, NCOA4, and SLC40A1 were consistently up-regulated, whereas SLC38A1 and, notably, GPX4 exhibited sustained down-regulation. The consistent repression of GPX4, which encodes the central lipid peroxide scavenging enzyme in the ferroptotic cascade, provides compelling evidence for the evolutionary conservation of ferroptosis vulnerability in OA cartilage across mammalian species.

Conversely, several genes displayed discordant regulatory patterns across datasets. PTGS2 and SLC39A14 were up-regulated in murine cartilage (GSE112641) but down-regulated in human samples (GSE114007). Similarly, LPCAT3, SAT1, and TFRC were induced in the rat model yet repressed in human OA cartilage. These discrepancies are likely multifactorial, reflecting fundamental species-specific metabolic network wiring, differential anatomical loading between knee and hip joints, variability in disease stage and severity across source cohorts, and technical platform differences between RNA-seq and microarray technologies. Despite this heterogeneity, the robust conservation of the core GPX4-centric pathway underscores its pathogenic relevance and supports its prioritisation as a therapeutic target. A comprehensive gene-by-gene comparison, including logFC values, adjusted p-values, directionality, and consistency scores, is provided in Supplementary Table S13.

Collectively, these findings implicate cysteine metabolic dysregulation and consequent glutathione depletion as upstream drivers of ferroptosis, which in turn accelerates OA progression. Diminished cysteine availability compromises antioxidant defences, promotes iron accumulation, and exacerbates lipid peroxidation, collectively perpetuating oxidative stress and chondrocyte degeneration in osteoarthritic cartilage.

3.3. Proteomic and metabolomic validation supports the cysteine metabolism-ferroptosis axis in OA

To validate transcriptomic findings and further elucidate the role of the cysteine metabolism-ferroptosis axis in OA, we performed integrated analyses of proteomics and metabolomics data of severe damaged and relatively preserved cartilage tissues from human OA patients. In the proteomic analysis, we identified 922 differentially expressed proteins in OA group, with 869 proteins being downregulated and 53 being upregulated (Fig. 3A, Supplementary Fig. S3A). KEGG pathway enrichment analysis of differentially expressed proteins (DEPs) revealed significant enrichment of metabolic and ferroptosis-related pathways. Among the top 30 enriched pathways, cysteine and methionine metabolism, glutathione metabolism, and ferroptosis were prominently represented, reinforcing their potential roles in OA pathogenesis (Fig. 3B and C).

Fig. 3.

Proteomics and metabolomics validation of cysteine metabolism-ferroptosis axis in OA progression A. Volcano plot of differential proteome expression in cartilage tissues of healthy human or OA patients. A screened threshold of |log2FC| > 0.26 (FC > 1.2 or FC < 0.833) and P value < 0.05 was performed to select significant differences. B. Bar charts of top 30 KEGG pathway enrichment of differentially expressed proteins (DEPs) in OA vs control human cartilage tissues. C. Bar charts of top 30 metabolic pathway enrichment of DEPs in OA vs. control human cartilage tissues. D. Volcano plot of differential metabolites expression in cartilage tissues of healthy human or OA patients. A screened threshold of |log2FC| > 0.26 (FC > 1.2 or FC < 0.833) and P value < 0.05 was performed to select significant differences. E. KEGG pathway enrichment network of differential metabolites (DEMs) between OA and control groups. F. Expression profiles of three pathways (ferroptosis, cysteine/methionine metabolism and glutathione metabolism) across DEPs in OA vs. control human cartilage tissues. G. Protein-pathway interaction Cnet plot integrating ferroptosis pathway, cysteine/methionine metabolism pathway and glutathione metabolism pathway. H. Schematic of ferroptosis pathway dysregulation in OA cartilage, integrating proteomic (rectangles) and metabolomic (circles) alterations (red: OA-upregulated, blue: OA-downregulated elements).

In parallel, untargeted metabolomic profiling revealed 429 metabolites with significantly altered tissue level, including 251 downregulated and 178 upregulated metabolites in the OA group (Fig. 3D, Supplementary Fig. S3B). Complementary analysis of differentially expressed metabolites (DEMs) further underscored the metabolic reprogramming in OA. The KEGG pathway enrichment network illustrated that the most significantly altered metabolites were closely linked to amino acid metabolism (Fig. 3E). This network-based representation emphasized the centrality of cysteine and glutathione dynamics in distinguishing severely damaged cartilage of OA joints from relatively preserved cartilage.

We next examined the expression profiles of DEPs within three key pathways: ferroptosis, cysteine and methionine metabolism, and glutathione metabolism. Pathway enrichment analysis revealed a significant enrichment for ferroptosis in the GSE118559 dataset (18 DEGs; Supplementary Table S5), but not in the GSE112641 or GSE114007 datasets. To ensure a robust investigation of ferroptosis, we supplemented the analysis of GSE118559 with a manually curated set of established ferroptosis regulators [30,[34], [35], [36]]. This allowed us to focus on key genes across core regulatory, iron metabolism, and lipid peroxidation pathways. All three pathways exhibited coordinated dysregulation in OA cartilage, indicating systematic disruption of redox homeostasis and antioxidant defense mechanisms (Fig. 3F). To visualize the interplay among proteins and their associated pathways, a Cnet plot was constructed integrating these three pathways. Combined with the previous established ferroptosis signaling cascade, the interaction map highlighted several hub proteins that were involved in multiple pathways, such as GPX4 and SLC family transporters including SLC7A11 and SLC3A2, suggesting their integrative role in mediating oxidative stress and ferroptosis susceptibility (Fig. 3G).

Finally, we constructed a schematic model integrating proteomic and metabolomic alterations to depict the molecular dysregulation of the ferroptosis pathway in OA cartilage (Fig. 3H). In this model, upregulated elements included pro-ferroptosis regulators and iron transport proteins, while downregulated components involved key antioxidant enzymes and cysteine transporters. This integrative view illustrates how impaired cysteine metabolism and glutathione depletion collectively contribute to ferroptosis activation, leading to oxidative damage and cartilage degeneration in OA.

Taken together, our proteomic and metabolomic analyses provide robust confirmation of transcriptomic findings, consolidating the critical role of the cysteine metabolism-ferroptosis axis in OA progression. These multi-omics insights highlight potential therapeutic targets aimed at restoring redox balance and preventing ferroptotic cell death in OA cartilage.

3.4. Extracellular L-cystine deficiency drives ferroptosis in inflammatory chondrocytes

To investigate the role of L-cystine/cysteine deficiency in ferroptosis and its potential contribution to OA pathogenesis, we employed an in vitro model of IL-1β-induced inflammation in murine-derived primary chondrocytes. L-cystine/cysteine metabolism dysfunction was modeled by culturing chondrocytes in the presence of IL-1β with or without extracellular L-cystine deficiency (Fig. 4A). Extracellular L-cystine deficiency effectively depletes the essential precursor for intracellular cysteine pools. IL-1β treatment severely reduced cell viability of chondrocytes by 45.93% (95%CI 41.08-50.78%). While L-cystine/cysteine deficiency further exaggerated this reduction by 20.97% (95%CI 11.39-30.55%) (Fig. 4B), highlighting the vulnerability of chondrocytes to oxidative stress and ferroptosis under L-cystine/cysteine deficiency conditions. Gene expression analysis demonstrated a remarkable downregulation of ferroptosis-related genes, including Slc7a11 (encoding the xCT cystine transporter), Slc3a2 (encoding the light chain of system Xc⁻), and Gpx4 (encoding glutathione peroxidase 4), in the L-cystine/cysteine deficiency group (Fig. 4C). At the protein level, L-cystine/cysteine deficiency significantly reduced the expression of SLC7A11, SLC3A2 and GPX4, while increasing the expression of TFRC (transferrin receptor), an iron uptake regulator and marker of ferroptosis susceptibility [30] (Fig. 4D, Supplementary Fig. S4). These findings indicate that L-cystine/cysteine deficiency impairs critical antioxidant defenses, driving ferroptosis in inflammatory chondrocytes. Biochemical analyses further elucidated that IL-1β treatment significantly reduced glutathione (GSH) levels, a key antioxidant molecule, compromising the cellular ability to mitigate oxidative stress (45.87%, 95%CI 29.22-69.52%), while L-cystine/cysteine deprivation intensified this reduction (74.64%, 95%CI 46.82%-109.46%) (Fig. 4E). Additionally, total iron content, a critical driver of ferroptosis [30], was significantly elevated upon L-1β treatment (47.26%, 95%CI 17.70-76.83%), and was further increased in L-cystine/cysteine deficient conditions (76.79%, 95%CI 30.57-123.02%) (Fig. 4F). Elevated iron levels promoted reactive oxygen species (ROS) generation and lipid peroxidation, as evidenced by the substantial increase in malondialdehyde (MDA) levels upon L-1β treatment (43.86%, 95%CI 15.69-72.30%) and was further strengthened by L-cystine/cysteine deficiency (77.74%, 95%CI 36.77-118.99%) (Fig. 4G). Collectively, these results suggest that L-cystine/cysteine deficiency sensitizes chondrocytes to ferroptosis by depleting GSH, increasing intracellular iron levels, and promoting lipid peroxidation.

Fig. 4.

Cysteine downregulation drives ferroptosis induction and contributes to OA pathogenesis A. Schematic representation of L-cystine/cysteine depletion in an IL-1β-induced in vitro inflammatory model. B. Cell viability of murine-derived primary chondrocytes with or without L-cystine/cysteine depletion in the presence IL-1β treatment. C. Gene expression levels of Slc7a11, Slc3a2, and Gpx4 in murine-derived primary chondrocytes with or without L-cystine/cysteine depletion in the presence IL-1β treatment. D. Protein expression levels of SLC7A11, SLC3A2, GPX4, and TFRC in murine-derived primary chondrocytes with or without L-cystine/cysteine depletion in the presence IL-1β treatment. E-G. Measurement of GSH levels (E), total iron content (F), and MDA levels (G) in murine-derived primary chondrocytes with or without L-cystine/cysteine depletion in the presence IL-1β treatment. H. Schematic representation of L-cystine/cysteine supplementation in an IL-1β-induced in vitro inflammatory model. I. Cell viability of murine-derived primary chondrocytes with or without L-cysteine supplementation in the presence IL-1β treatment. J. Gene expression levels of Slc7a11, Slc3a2, and Gpx4 in murine-derived primary chondrocytes with or without L-cysteine supplementation in the presence IL-1β treatment. K. Protein expression levels of SLC7A11, SLC3A2, GPX4, and TFRC in murine-derived primary chondrocytes with or without L-cysteine supplementation in the presence IL-1β treatment. L-N. Measurement of GSH level (L), total iron content (M), and MDA level (N) in murine-derived primary chondrocytes with or without L-cysteine supplementation in the presence IL-1β treatment. Data are presented as mean ± 95% confidence interval (n = 4; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Student's two tailed unpaired t-test for pairwise comparisons, one-way ANOVA for multiple comparisons.

To test whether L-cysteine supplementation could rescue chondrocytes from ferroptosis, we supplemented extracellular L-cysteine to the culture medium in the presence of IL-1β (Fig. 4H). IL-1β treatment significantly reduced chondrocytes viability (48.18%, 95%CI 41.65-54.71%), while extracellular L-cysteine supplementation significantly rescued the viability of IL-1β-treated chondrocytes (43.34%, 95%CI 21.32-65.35%) (Fig. 4I). Gene expression analysis revealed that extracellular L-cysteine supplementation restored the expression of Slc7a11, Slc3a2, and Gpx4, rescuing key components of the antioxidant system in IL-1β-treated chondrocytes (Fig. 4J). Similarly, extracellular L-cysteine supplementation restored the protein levels of SLC7A11, SLC3A2, and GPX4, while reducing the elevated expression of TFRC to near-control levels (Fig. 4K). Further biochemical analyses confirmed the protective effects of extracellular L-cysteine supplementation. GSH level was significantly repressed by IL-1β treatment (47.87%, 95%CI 30.31-65.43%) but significantly restored by L-cysteine (59.43%, 95%CI 24.97-93.90%) (Fig. 4L), while total iron content and MDA levels were markedly increased in IL-1β group (61.47%, 95%CI 25.21-97.73% and 93.85%, 95%CI 49.42-138.28%, respectively) but reduced by L-cysteine addition (20.16%, 95%CI 5.19-35.14% and 28.53%, 95%CI 1.54-55.52%, respectively) (Fig. 4M and N). These findings indicate that extracellular L-cysteine supplementation mitigates ferroptosis induction by restoring antioxidant defenses, reducing iron accumulation, and preventing lipid peroxidation. Taken together, these results demonstrate that extracellular L-cystine deficiency drives ferroptosis induction in inflammatory chondrocytes, while extracellular L-cysteine supplementation effectively protects against ferroptosis and oxidative damage. These findings underscore the critical role of cystine/cysteine metabolism in OA and suggest that targeting cystine/cysteine deficiency may offer therapeutic potential.

3.5. L-cysteine supplementation mitigates in vivo ferroptosis and OA pathology

To further validate the role of cysteine dysregulation and ferroptosis in OA pathogenesis, we extended our investigation on in vivo OA murine model. We also assessed the therapeutic potential of L-cysteine supplementation in mitigating cartilage degeneration and ferroptosis (Fig. 5A). In OA cartilage, safranin-O and toluidine blue staining showed severe cartilage degradation, marked by loss of proteoglycan content and disrupted tissue architecture in OA group. Notably, both low- and high-dosage administration of L-cysteine partially restored cartilage integrity, as evidenced by improved proteoglycan retention and reduced tissue damage. These findings suggest that L-cysteine supplementation exerts protective effects on cartilage structure in OA (Fig. 5B). This was also demonstrated by increased OARSI (Osteoarthritis Research Society International) score in OA mice compared to that of sham mice, while L-cysteine could lower the OARSI score (Fig. 5C). To investigate the regulation effect of L-cysteine on ferroptosis in OA, the immunohistochemistry analysis to assess the expression of ferroptosis-related proteins, including GPX4, SLC7A11, and TFRC, in cartilage tissues. OA cartilage exhibited significantly reduced levels of GPX4 and SLC7A11, indicating impaired antioxidant defenses, along with elevated expression of TFRC, which consistent with increased ferroptosis susceptibility. Strikingly, L-cysteine low- and high-dosage supplementation restored GPX4 and SLC7A11 expression while reducing TFRC levels in a dose-dependent manner, suggesting that cysteine supplementation mitigates in vivo ferroptosis (Fig. 5D–G, Supplementary Fig. S5). To further confirm the effects of L-cysteine supplementation on ferroptosis-related biochemical markers, we measured glutathione (GSH) levels, total iron content, and malondialdehyde (MDA) levels in cartilage tissues. OA cartilage showed significantly reduced GSH levels (42.83%, 95%CI 31.45-54.21%), reflecting weakened antioxidant defenses. Concurrently, total iron content was elevated (228.56%, 95% CI 79.72-377.41%), and MDA levels were markedly increased (91.38%, 95% CI 15.50-167.27%), which indicate the activation of ferroptosis-mediated lipid peroxidation. Importantly, low- and high-dosage L-cysteine supplementation rescued GSH levels (low: 42.22%, 95% CI 4.77-79.68%; high: 60.43%, 95%CI 30.22-90.65%), reduced iron content (low: 33.64%, 95% CI 20.10-47.19%; high: 62.48%, 95% CI 52.96-72.01%), and decreased MDA levels (low: 34.34%, 95%CI 24.33-44.34%; high: 35.47%, 95%CI 10.94-59.99%) in OA cartilage tissues to levels comparable to sham controls (Fig. 5H–J). These findings collectively demonstrate that cysteine dysregulation contributes to ferroptosis induction and cartilage degeneration in OA. L-cysteine supplementation effectively protects against ferroptosis by restoring cysteine metabolism, improving antioxidant defenses, and reducing oxidative damage in cartilage tissues. These results highlight the therapeutic potential of targeting cysteine metabolism as a strategy to mitigate cartilage damage and ferroptosis in OA pathogenesis.

Fig. 5.

Cysteine downregulation drives ferroptosis induction and contributes to OA pathogenesis. A. Schematic showing L-cysteine supplementation in the murine OA model. B&C. Representative safranin-O/fast green staining, toluidine blue staining (B) and OARSI scores (C) in sham and OA cartilage tissues with or without L-cysteine supplementation. Scale bar = 100 μm (n = 6; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). D-G. Immunohistochemistry staining (D) and semi-quantitative analysis of the percentage of GPX4 (E), SLC7A11 (F) and TFRC (G) positive chondrocytes of sham and OA cartilage tissues with or without L-cysteine supplementation. Scale bar = 100 μm (n = 6; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). H-J. Measurement of GSH levels (H), total iron content (I), and MDA level (J) in sham and OA cartilage tissues with or without L-cysteine supplementation. Data are presented as mean ± 95% confidence interval (n = 6; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Student's two tailed unpaired t-test for pairwise comparisons, one-way ANOVA for multiple comparisons.

4. Discussion

This study uncovered the critical role of cysteine metabolism and ferroptosis in OA pathogenesis, providing a novel perspective on the role of metabolic regulation during cartilage degeneration. By integrating transcriptomic across species, proteomic, and metabolomic data, along with in vitro and in vivo validation, we demonstrated that cysteine dysregulation is not merely a consequence of OA but actively drives OA progression by impairing antioxidant defenses, promoting iron accumulation, and inducing lipid peroxidation. These findings supplemented the traditional view of OA as a disease primarily driven by mechanical stress and inflammation [37,38], highlighting the importance of metabolic dysfunction in its pathophysiology and opening new avenues for therapeutic intervention [10].

The consistent alteration of metabolic pathways across three independent datasets from different species underscores the conserved nature of metabolic dysregulation in OA. While previous studies have predominantly focused on inflammation and extracellular matrix (ECM) remodeling [39,40], our results positioned metabolic reprogramming as a critical feature of OA. Among the dysregulated pathways, cysteine and methionine metabolism, along with glutathione metabolism, emerged as key contributors, linking metabolic dysfunction to oxidative stress and ferroptosis [15,25]. This aligned with emerging evidence that ferroptosis plays a critical role in degenerative diseases [41]. Identification of cysteine metabolism as a shared dysregulated pathway across species suggested its fundamental role in cartilage homeostasis and disease progression. Our multi-omics validation through proteomic and metabolomic analyses further reinforced these findings. These findings raised important questions about the interplay between metabolic, inflammatory, and mechanical factors in OA, suggesting that metabolic dysfunction may act as a unifying mechanism underlying cartilage degeneration.

The dysregulation of key ferroptosis-related genes, including SLC7A11, GPX4, and TFRC, provides mechanistic insights into how cysteine depletion drives ferroptosis in OA. The downregulation of SLC7A11 and GPX4 compromises the antioxidant capacity of chondrocytes, while the upregulation of TFRC indicates iron overload, a hallmark of ferroptosis. To definitively decouple the effects of inflammation from those of metabolic deficiency, we investigated the impact of L-cystine in the absence of inflammatory stimuli. L-cystine is the oxidized and essential form of cysteine for cellular uptake. Our results demonstrate that L-cystine deficiency alone is sufficient to initiate the ferroptotic cascade, as evidenced by the downregulation of the system Xc-subunits Slc7a11 and Slc3a2, downregulation of Gpx4, a decrease in GSH, and an increase in lipid peroxidation and iron concentration (Supplementary Fig. S6A–6D). This confirms that cystine deprivation, even without an inflammatory trigger, can directly engage the molecular machinery of ferroptosis in chondrocytes. It is important to note that while these changes were significant, the ferroptotic phenotype was markedly amplified in the presence of IL-1β, indicating that inflammation and metabolic stress synergize to drive a more robust ferroptotic response. This progressive dysfunction was further evidenced by a slight but significant decrease in viability observed at a later timepoint of 72 h (Supplementary Fig. S6E), consistent with the progressive nature of ferroptosis. Furthermore, our direct measurement revealed that IL-1β treatment influences intracellular cystine levels, causing a significant decrease of approximately 10% (Supplementary Fig. S7). This finding indicates that IL-1β treatment not only creates an inflammatory and oxidative environment but also directly weakens the metabolic foundation by reducing cystine availability. Together, these effects synergistically exacerbate the consequences of cystine deficiency, thereby potently potentiating ferroptosis. These molecular alterations create a pro-ferroptotic environment characterized by depleted glutathione levels, excessive lipid peroxidation, and oxidative damage, ultimately leading to chondrocyte death and cartilage degeneration. Our findings suggested that ferroptosis is not merely a pathological phenomenon in OA but a central mechanism driving cartilage degeneration. This finding was further supported by the protective effects of L-cysteine supplementation on chondrocyte dysfunction and cartilage degeneration, which restored antioxidant defenses and mitigated ferroptosis in both in vitro and in vivo models. These results corresponded with studies on other degenerative diseases, such as Alzheimer's disease (AD) [42,43], Huntington's disease (HD) [44], Parkinson's disease (PD) [45,46] and amyotrophic lateral sclerosis (ALS) [47], where ferroptosis inhibition has shown therapeutic promise. While P2 chondrocytes in in vitro study were necessitated by cell yield constraints, SOX9 verification confirmed maintained chondrogenic identity. Future studies will employ explant models to avoid passaging. Our study comprehensively analyzed the interaction among cysteine metabolism, ferroptosis, and OA progression, providing a novel framework for understanding and treating this debilitating disease.

The therapeutic potential of L-cysteine supplementation on OA is a key finding of this study. Previous study results supported the therapeutic potential of targeted amino acid interventions on neurodegenerative and protein-misfolding disorders. For instance, L-serine supplementation could ameliorate neuropathology in amyotrophic lateral sclerosis/motor neuron disease (ALS/MND) [48], while N-acetylcysteine (NAC), a glutathione precursor, demonstrated neuroprotection effect on AD models by suppressing amyloid-β secretion and counteracting oxidative stress [49,50]. Notably, the study in tauopathy models revealed that essential amino acid (EAA) supplementation not only restores proteostasis but also reduces pathological tau aggregation and enhances neuronal survival. Similarly, L-arginine supplementation has been explored in mitochondrial encephalopathies to be effective in improving symptoms [51]. Collectively, these studies demonstrated the versatility of amino acid-based strategies in degenerative diseases. In our study, the in vitro cysteine supplementation rescued chondrocyte viability, restored glutathione levels, and reduced lipid peroxidation by mitigating ferroptosis. In in vivo studies, L-cysteine supplementation improved cartilage integrity, restored antioxidant defenses, and reduced iron accumulation in a surgery-induced OA murine model. These findings suggested that cysteine metabolism plays a significant role in chondrocyte dysfunction in OA, and cysteine supplementation offers a more comprehensive approach compared with current strategies focusing on inflammation or ECM degradation. However, clinical translation of cysteine supplementation therapeutics required a careful evaluation of the pharmacokinetics, safety, and long-term effects in human patients. Additionally, combining cysteine supplementation with other therapeutic strategies, such as ferroptosis inhibitors or anti-inflammatory agents [24,52,53], may enhance its efficacy and provide a synergistic approach to OA treatment.

While our study provides robust evidence for the role of cysteine metabolism and ferroptosis in OA, several questions still remain. First, while our proteomics analysis (n = 3 patients) aligns with previous small-scale OA studies [54], limited sample size constrains generalizability. Future studies should validate findings in larger cohorts. Second, our use of a relatively lenient FC threshold (1.2) may have increased the risk of false-positive gene identifications. However, cross-dataset consistency and experimental validation reduced this risk. Future studies with larger sample sizes will enable more stringent cutoffs without sacrificing biological discovery. Third, oral L-cysteine supplementation may exert systemic metabolic effects beyond the joint. While our in vivo data show clear protection against cartilage degeneration, we acknowledge that the observed benefits could be mediated partly by systemic changes. Future studies should explore cartilage-targeted delivery strategies (e.g., intra-articular injection of cysteine-loaded nanoparticles) to achieve local cysteine enrichment while minimising systemic exposure. Fourth, it is unclear whether the observed metabolic changes are primary drivers of OA or secondary responses to mechanical stress and inflammation. Further studies are needed to disentangle these complex interactions and determine whether targeting metabolic dysfunction can prevent OA progression. Fifth, the upstream regulators of cysteine metabolism in OA were still poorly understood. Identifying the signaling pathways and molecular mechanisms controlling cysteine uptake and utilization in chondrocytes could uncover additional therapeutic targets. Finally, the translational potential of metabolic interventions in OA warrants further exploration, including clinical trials to evaluate the efficacy and safety of cysteine-based therapies.

In conclusion, our study demonstrated that cysteine dysregulation and ferroptosis are central to OA pathogenesis, providing new insights into the metabolic basis of cartilage degeneration. By linking cysteine depletion to ferroptosis induction and oxidative damage, we highlighted the significance of cysteine metabolism in maintaining cartilage homeostasis and protecting against OA progression. The therapeutic effects of L-cysteine supplementation observed in both in vitro and in vivo models indicated its potential as a novel intervention for OA. These findings not only advance our understanding of OA pathophysiology but also open avenues for developing metabolic-targeted therapies to mitigate cartilage damage and improve patient outcomes. Future research will be conducted to further elucidate the regulatory mechanisms of cysteine metabolism and ferroptosis in OA and explore the translational potential of metabolic interventions in clinical settings.

Author contribution

G. L., L. F. and J. Z. designed the study. L. Z., Z. Y., S. C., G. C., N. H., W. L., C. H. P. C., Y. W., H. C., performed the experiments and acquired the data. Y.F. W., L. Z., Z. Y. and L. F. analyzed and interpreted the data and drafted the manuscript. G. L., L. F. and J. Z. finalized and approved the manuscript for submission.

Ethical approval and patient inclusion

This study was approved by the Ethics Committee of Jiaxing First Hospital (2024-LP-032).

Declaration of generative AI in scientific writing

During the preparation of this work, the authors used ChatGPT-5 in order to enhance language and readability. After using this tool, the authors reviewed and edited the content as needed and took full responsibility for the content of the publication.

Funding statement

This work was partially supported by grants from the National Natural Science Foundation of China (81772322); University Grants Committee, Research Grants Council of the Hong Kong Special Administrative Region, China (14120118, 14108720, C7030-18G, T13-402/17-N and AoE/M-402/20); Hong Kong Innovation Technology Commission Funds (PRP/050/19FX); Hong Kong Medical Research Funds (16170951 and 17180831); Key Departments of Jiaxing, China (2023-ZC-012); Medical and Health Science and Technology Plan of Zhejiang, China (2025KY1592); Shenzhen Science and Technology Program (No. JCYJ20220530144016037). This study also received support from the research funds from Health@InnoHK program launched by Innovation Technology Commission of the Hong Kong SAR, PR China.

Competing interests

The authors declare no competing interests.

Appendix A. Supplementary data

The following are the Supplementary data to this article.