Interplay between lipid dysregulation and ferroptosis in chondrocytes and the targeted therapy effect of metformin on osteoarthritis

Graphical abstract

Highlights

• •Using Genome-wide RNA-Seq data, biochemical, and histochemical assays in human and murine OA cartilage as well as primary chondrocytes to demonstrate that the interplay between lipid metabolism and ferroptosis is involved in cartilage homeostasis.
• •Utilizing biochemical and histological approaches to reveal that metformin reshapes lipid availability and ameliorates chondrocyte ferroptosis sensitivity via the AMPK/ACC signaling pathway.
• •Demonstrated that ferroptotic chondrocytes induce the recruitment and chemotaxis of synovial macrophages through CCL2 and metformin diminishes inflammation in the joint microenvironment of osteoarthritis model mice.

Abstract

Introduction

Osteoarthritis (OA) is a devastating whole-joint disease affecting a large population worldwide; the role of lipid dysregulation in OA and mechanisms underlying targeted therapy effect of lipid-lowering metformin on OA remains poorly defined.

Objectives

To investigate the effects of lipid dysregulation on OA progression and to explore lipid dysregulation-targeting OA treatment of metformin.

Methods

RNA-Seq data, biochemical, and histochemical assays in human and murine OA cartilage as well as primary chondrocytes were utilized to determine lipid dysregulation. Effects of metformin, a potent lipid-lowering medication, on ACSL4 expression and chondrocyte metabolism were determined. Further molecular experiments, including RT-qPCR, western blotting, flow cytometry, and immunofluorescence staining, were performed to investigate underlying mechanisms. Mice with intra-articular injection of metformin were utilized to determine the effects on ACLT-induced OA progression.

Results

ACSL4 and 4-HNE expressions were elevated in human and ACLT-induced mouse OA cartilage and IL-1β-treated chondrocytes (P < 0.05). Ferrostatin-1 largely rescued IL-1β-induced MDA, lipid peroxidation, and ferroptotic mitochondrial morphology (P < 0.05). Metformin decreased the levels of OA-related genes (P < 0.05) and increased the levels of p-AMPK and p-ACC in IL-1β-treated chondrocytes. Intra-articular injection of metformin alleviated ACLT-induced OA lesions in mice, and reverted the percentage of chondrocytes positive for MMP13, Col2a1, ACSL4 and 4-HNE in ACLT mice (P < 0.05). Ferroptotic chondrocytes promoted the recruitment and chemotaxis of RAW264.7 cells via CCL2, which was blocked by metformin in vitro (P < 0.05).

Conclusion

We establish a critical role of polyunsaturated fatty acids metabolic process in OA cartilage degradation and define metformin as a potential OA treatment. Metformin reshapes lipid availability and ameliorates chondrocyte ferroptosis sensitivity via the AMPK/ACC pathway. In the future, gene-edited animals and extensive omics technologies will be utilized to reveal detailed lipids’ involvement in cartilage lesions.

Introduction

Osteoarthritis (OA) is now the most common form of degenerative joint disease among middle-aged and older individuals due to its fast rise in occurrence [1], [2]. Since 1990, the incidence rate of OA has increased by 9.3 %. According to estimates in 2020, approximately 240 million people worldwide have symptomatic OA, including 10 % of men and 18 % of women age 60 and older [3]. OA affects all joint tissues causing articular cartilage deterioration, osteophyte formation, synovial inflammation, and subchondral bone remodeling [4], [5]. Among the intrinsic pathological factors of OA, inflammation primarily contributes to cartilage degradation and inflammatory hyperplasia of the synovium [6], [7]. It was reported that the crosstalk between articular cartilage and periarticular synovium, called the cartilage-synovium axis, orchestrates the inflammatory environment in OA joints through communication at the cellular level [8], [9], [10]. However, the process via which inflammation accumulates in the joint environment and the exact mechanism by which inflammation amplifies osteoarthritic phenotypes have not been fully elucidated.

The homeostasis of articular cartilage is preserved by the unique resident cell type chondrocytes in cartilage tissues [11]. Under pro-inflammatory micro environmental conditions in osteoarthritic joints, chondrocytes transit their cellular metabolism from a dormant regulatory state to a much more active metabolic state [12], [13]. Among metabolic regulators involved in OA characteristics, the glycolytic pathway has been the most extensively studied [14], [15], but how lipid homeostasis is broken in chondrocytes and disturbs cartilage homeostasis is poorly understood [16], [17]. Emerging evidence demonstrates that patients with OA have elevated in free fatty acid levels [18], [19], [20] prior to the appearance of pathological characteristics [21], [22]. It has been shown that a substantial accumulation of lipid droplets (LDs) in articular cartilage is strongly related to matrix degradation and loss of chondrocyte cellularity [23], [24], [25], suggesting that accumulation of lipids in the cytoplasm of chondrocytes may be the cause of OA development. Thus, lipid metabolic reprogramming in chondrocytes from normal to pathological conditions is modulated by certain enzymatic status and may provide a promising strategy for addressing degenerative joint disorders.

Recent evidence has shown lipid metabolism (LM) as a crucial mediator of ferroptosis, suggesting a possible interplay with ferroptosis [26]. Ferroptosis is a kind of cell death that relies on iron and is distinguished by the lipid peroxides buildup and oxidative stress [27], [28]. During the catalytic role of acyl-CoA synthetase long-chain family member 4 (ACSL4), polyunsaturated fatty acids (PUFAs), especially adrenoyl (AdA) and arachidonoyl (AA) moieties produce acyl Co-A derivatives. Then, lysophosphatidylcholine acyltransferase 3 (LPCAT3) esterifies these derivatives into phosphatidylethanolamines (such as AdA-PE and AA-PE), and subsequently, the complex formed by 15-lipoxygenase (15-LOX) and phosphatidylethanolamine-binding protein 1 (PEBP1) selectively catalyzes AA-PE, involving in generating and accumulating lipid peroxidation product 15-hydroperoxyeicosatetraenoic acid-phosphatidylethanolamine (15HpETE-PE). Clearance obstacle of these products leads to cellular dysfunction and eventually results in cell death [29], [30], suggesting that LM is strongly connected to ferroptosis sensitivity in certain cells. However, the molecular basis for lipid deregulation-driven chondrocyte ferroptosis in the OA pathogenesis is not fully understood, and targeting the LM’s interaction with ferroptosis could make for the creation of new therapeutic methods for OA treatment.

Recently, several studies have reported that metformin not only has therapeutic effects in type 2 diabetes but also shows potential in treating metabolic disorders encompassing non-alcoholic fatty liver disease (NAFLD), vascular calcification linked to hyperlipidemia, and polycystic ovary syndrome (PCOS) [31]. It was shown that metformin could lead to a reduction in adenosine triphosphate (ATP) production through interference with mitochondrial respiratory complex I [18], thereby activating AMP-activated protein kinase (AMPK) [32]. Emerging evidence expands the kinase’s functions of AMPK as a master energy sensor that determines cell fate by regulating the phosphorylation of different downstream substrates [33], [34]. In addition, the latest study confirmed that AMPK activation by energy stress could be one of the mechanisms regulating ferroptosis [35]. However, the coordination of the metformin-responsive lipid metabolic pathway by AMPK signaling in chondrocytes and the exact mechanism by which AMPK restores lipid metabolic balance to reverse an alternative source of PUFAs for lipid peroxidation (LPO) during ferroptosis remain unknown.

While previous reports have indicated that lipid accumulation exacerbates arthritis [19], our finding sheds new light on the mechanisms through which lipid accumulation influences the progression of arthritis. Herein, researchers found that inflammation triggered the LM interaction with ferroptosis in human and mouse OA cartilage, as well as mouse chondrocytes by interleukin-1β (IL-1β) stimulation. In addition, metformin exerted protective inflammatory effects in joints by alleviating the sensitivity to chondrocyte ferroptosis. The pharmacological mechanisms of metformin could underlie lipid metabolic reprogramming via the AMPK/acetyl-CoA carboxylase (ACC) signaling pathway. Researchers also demonstrated that ferroptotic chondrocytes amplified joint inflammation through the C–C motif chemokine ligand 2 (CCL2)-mediated chemotaxis of synovial macrophages, which was diminished by metformin in vitro and in the OA mouse model. The elucidation of the biological role and function of ACSL4 could provide a molecular basis and mechanism for deregulation of lipid homeostasis involved in the OA pathogenesis and could contribute in developing a new therapeutic approach for OA pathogenesis. The pharmacological mechanisms of metformin could underlie lipid metabolic reprogramming via the AMPK/ACC signaling pathway. Our studies suggest that metformin may be used in clinical interventions of OA disease.

Materials and methods

Chemicals and materials

Erastin (HY-15763), ferrostain-1 (Fer-1, HY-100579), metformin (HY-B0627), and RS102895 hydrochloride (HY-18611) were purchased from MCE (Shanghai, China). C11 BODIPY Lipid Peroxidation Sensor (D3861) was purchased from Thermo Fisher (Shanghai, China). Recombinant mouse IL-1β (401-ML-010/CF) was obtained from R&D Systems (Minneapolis, USA). Malondialdehyde (MDA) Assay Kit (E-BC-K028-M) was obtained from Elabscience (Wuhan, China), and the Alcian Blue Stain Kit (G1285) from Solarbio (Wuhan, China).

Isolation and culture of murine chondrocytes

Five-day-old C57BL/6J mice were employed to harvest primary murine chondrocytes from their knee joint cartilage [36]. The Experimental Animal Ethics Committee of Third Military Medical University granted authorization for the animal studies. After dissecting the tissue, the cartilage tissues were subjected to treatment with 0.25 % trypsin (Gibco) for 30 min to break them down enzymatically. They were then treated with 0.5 mg/ml collagenase type II (Servicebio) for 5 h at a temperature of 37 °C. The cells were suspended again and cultivated in DMEM/F12 media (Gibco) enriched with 1 % penicillin–streptomycin (Gibco) and 10 % fetal bovine serum (Gibco) in a humidified environment with 5 % CO2 at a temperature of 37 °C. To maintain phenotype integrity, we only used chondrocytes of the first and second passages.

Collection of human tissue samples

Patients from Army Medical University's Xinqiao Hospital's Department of Orthopedics provided samples of human OA cartilage. The acquisition and use of human biological tissue specimens were authorized by the Ethics Committee of Xinqiao Hospital of Army Medical University. The data on patients, such as their gender, age, height, weight, and BMI, are consolidated in Supplementary Table 2. All individuals provided written informed permission prior to undergoing complete knee replacement surgery. The human cartilage tissue slices were subjected to staining employing safranin O/fast green (SO/FG) and determined based on the Osteoarthritis Research Society International (OARSI) grading system [37].

Mice and experimental surgery

All animal experiments were approved by the Ethics Committee of Third Military Medical University (AMUWEC2019235). The study complied with all relevant ethical regulations for animal testing and research. The design, analysis, and reporting of animal experiments were performed following the Animals in Research: Reporting of In Vivo Experiments guidelines (ARRIVE; https://www.nc3rs.org.uk/arrive-guidelines). We purchased 10-week-old C57BL/6J (WT) male mice from Vital River Laboratory Animal Technology Co., Ltd.. The mice were housed in a specific pathogen free animal facility. Animals were maintained under constant temperature (23–25 °C) and humidity (45–65 %) with controlled light–dark cycles (12:12 h). We anesthetized the mice with an intraperitoneal injection using 1 % pentobarbital sodium (40 mg/kg) and then transected the ACL surgically to induce mechanical instability-associated osteoarthritis. In the sham groups, the knee capsule and infrapatellar fat pad were incised but no ACL transection was performed [38]. The criteria used for including and excluding animals during the experiment were established a priori.

For the experiment to observe the pathology of disease, mice were euthanized at 8 weeks after surgery (n = 5 per group for analysis). ACLT surgery was performed in the right knees of mice. For the experiment with drug interventions, mice were euthanized at 8 weeks after surgery (n = 5 per group for analysis). The sham group received a skin incision and suturing without patellar dislocation or ligament transection. For intra-articular injections, the stock solution of metformin (1.65 g/ml) was diluted in PBS (1:100) (ACLT + Metformin group) or vehicle (ACLT group) and administered in the knee every three days for 8 consecutive weeks [39]. At the end of the procedure, the mice were euthanized, and their knee joints were collected and fixed in 4 % paraformaldehyde for further experiments.

Alcian blue staining

Primary chondrocytes were rinsed with PBS and treated with 4 % formaldehyde for 10 min. Subsequently, they were exposed to Alcian blue staining solution (1.0 % Alcian blue in 0.1 M HCl) (Solarbio, China) and incubated overnight at 4 °C. The cells underwent three rounds of washing with PBS and were then imaged using a microscope.

Detection of lipid peroxidation

Primary chondrocytes were exposed to 5 µM BODIPY581/591 C11 (Thermo Fisher, D3861) for a period of 40 min. Following this, the cells were rinsed with PBS, treated with trypsin, and then passed through a filter to get individual cell suspensions. The BD FACS Aria II (Becton Dickinson) was employed to conduct flow cytometry analysis. After oxidation, the excitation maximum moves from 581 nm to 500 nm, and the emission maximum moves from 591 nm to 510 nm. The FITC filter was deployed to detect oxidized BODIPY-C11, with an emission wavelength of 510 nm. Data analysis was imaged using a microscope or conducted employing the FlowJo v10 (BD Bioscience).

Malonaldehyde (MDA) assay

The MDA concentration in primary murine chondrocytes was determined by employing the Lipid Peroxidation MDA Assay Kit (Elabscience, E-BC-K028-M) depending on the manufacturer's instructions. In summary, the MDA reaction in the specimen with thiobarbituric acid (TBA) produced an MDA-TBA compound, which was then measured employing fluorescence spectroscopy with an excitation wavelength of 532 nm and an emission wavelength of 553 nm.

Lipid droplet staining

The cells were subjected to fixation employing a 4 % paraformaldehyde solution for a duration of 15 min. Subsequently, they were stained with a concentration of 0.1 µg/ml Nile Red (MCE, HY-D0718) for a period of 30 min in order to see the lipid droplets. Subsequently, the nucleus was stained with 4′,6-diamidino-2-phenylindole (DAPI; Servicebio, G1012-100ML) as a counterstain. The acquisition of images was conducted employing a laser scanning confocal microscope, while the subsequent processing and analysis were carried out employing ImageJ software.

Transmission electron microscopy (TEM)

TEM was conducted employing standard procedures by the Biomedical Analysis Center of Animal Military University. Briefly, cells were quickly placed in an electron microscopy fixative mixture at 4 °C. Subsequently, the samples were impregnated with 1 % agarose, subjected to dehydration, and sliced into ultrathin sections (60–80 nm), deploying an ultramicrotome. The sections were exposed to treatment with uranyl acetate in 100 % ethanol for 10 min, followed by staining with lead citrate for another 10 min. The acquisition of images was performed by deploying a TEM.

Micro-computed tomography

In this experiment, knee joints removed from mice were fixed overnight using a 4 % solution of paraformaldehyde. The tissues underwent scanning using a Micro-CT (Skyscan-1272; Bruker Micro-CT, Belgium) with a pixel resolution of 6 µm and operated at a voltage of 60 kV and 166 µA.

Immunohistochemical and immunofluorescence staining

Standard protocols were followed for immunohistochemical and immunofluorescence staining. Briefly, after deparaffinization and rehydration, the tissue sections were made permeable using 0.1 % Triton X-100. To minimize staining that is not specific to the target, the sections were treated with 3 % BSA. Subsequently, the sections were exposed to the specified antibodies for immunohistochemical staining and immunofluorescence staining. The primary antibodies used for immunohistochemistry were 4-HNE (Bioss, bs-6313R; dilution, 1:200), ACSL4 (Affinity, DF12141; dilution, 1:100), p-ACC (Cell Signaling Technology, 11818S; dilution, 1:200), and F4/80 (Affinity, DF2789; dilution, 1:200). The primary antibodies employed for immunofluorescence were MMP13 (Affinity, AF5355; dilution, 1:200), Col2a1 (Santa Cruz, sc-52658; dilution, 1:200), F4/80 (Affinity, DF2789; dilution, 1:200), CD206 (Affinity, DF4149; dilution, 1:100), and iNOS (Abcam, ab178945; dilution, 1:200). The secondary antibodies used were goat anti-rabbit IgG (H&L) conjugated with Biotin used for immunohistochemistry, Cy3 conjugated Goat Anti-Rabbit IgG (H&L) or FITC conjugated Goat Anti-Mouse IgG (H&L) for immunofluorescence (all from Servicebio). The nucleus was stained with 4′,6-diamidino-2-phenylindole (DAPI; Servicebio, G1012-100ML) as a counterstain. Images were taken using fluorescent microscopy. Histomorphometry and quantification were processed and analyzed with ImageJ.

Bioinformatics analysis of transcriptomics datasets

Transcriptome data for OA or diseases related to OA were obtained from the Gene Expression Omnibus (GEO) database, using both in vivo and in vitro models. Transcriptomes of human OA cartilage (GSE64394 and GSE 169077), IL-1β-treated chondrocytes (GSE75181, GSE6119, and GSE104793), and cartilage from various post-traumatic OA animal models (GSE42295, GSE8077, and GSE28958) were analyzed. All the differentially expressed genes were subjected to heat map analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses and Gene Ontology (GO) analysis. For the KEGG enrichment analysis, a P-value < 0.05 was used as the threshold to determine significant enrichment of the gene sets.

Statistical analysis

The statistical analyses were conducted employing GraphPad Prism 8 software (GraphPad Software, San Diego, USA). The statistical technique of one-way analysis of variance (ANOVA) was used, followed by the application of Tukey's post hoc test, in order to conduct multiple comparisons. The comparison between the two groups was conducted using a Two-tailed Student's t-test. Statistical significance was defined as P < 0.05 (*P < 0.05, **P < 0.01, ***P < 0.001). The quantification was conducted using a minimum of three separate experimental groups, and the outcomes were manifested as the mean ± s.d.

Results

The LM interplay with ferroptosis in human and murine OA cartilage

To seek a critical regulator of metabolic pathways in chondrocytes that has specifically modified expression in the context of OA, researchers conducted a thorough analysis of the transcriptomics data obtained from the Gene Expression Omnibus (GEO) database. This data was derived from human OA cases as well as relevant OA conditions in animal models and in vitro. Interestingly, the GO enrichment and KEGG pathway analysis manifested that the differentially regulated genes in these OA contexts were mainly related to functional annotations related to ‘long-chain fatty-acyl-CoA metabolic process’ and ‘fatty acid metabolism’. Furthermore, the 'ferroptosis' pathway was manifested to have a significant impact on the condition related to OA, contributing to cell death (Fig. 1A, B and Supplementary Tables 6–8). Notably, although lipid peroxidation was identified as a determinant in stimulating ferroptosis [40], the critical regulators of LM in ferroptosis are still not well established. Indeed, the expression of genes linked to these processes was significantly enriched in these extensively analyzed transcriptome data. Among the regulators of these annotations, acyl-CoA synthetase long-chain family member 4 (ACSL4) expressions were consistently elevated in human OA transcriptome datasets (Fig. 1C), and ACSL4 was reported to shape cellular lipid composition to dictate ferroptosis sensitivity in emerging evidence [41].

Fig. 1.

Inflammation triggers the interplay between lipid homeostasis dysregulation and ferroptosis in OA human cartilage. (A) Gene ontology analysis of differentially expressed genes (DEGs) found in the OA-relevant condition from dataset GSE75181 in GEO. (B) KEGG pathway enrichment analysis of DEGs identified in the OA-relevant condition from dataset. (C) Heatmaps depicting the expression of components of long-chain fatty-acyl-CoA metabolic process, fatty acid metabolism, and ferroptosis pathway in OA-relevant conditions. Public transcriptome datasets obtained from human OA cartilage (GSE64394 and GSE169077) were analyzed. (D) Representative plain radiographic images from patients with knee OA. A-P, anterior-posterior. (E) Illustrative representations of the tibial plateau from patients of varying ages following complete knee arthroplasty. The cartilage on the medial side of the knee joints had evident subchondral bone exposure, with a coarser surface than the cartilage on the lateral side. Scale bar, 1 cm. (F) Cartilage sections from damaged or undamaged areas of human OA cartilage were stained with SO/FG (Scale bar, 100 μm) and immunostained against ACSL4 and 4-HNE (Scale bar, 50 μm). (G) The degree of cartilage degeneration was determined by the OARSI grading system. (H and I) ACSL4- and 4-HNE-positive cells were quantified. n = 7. Error bars are means ± s.d., *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with several comparisons employing the Tukey method.

To explore whether the interplay of LM and ferroptosis contributed to OA progression, researchers collected the cartilage tissues of OA patients who had undergone complete knee replacement surgery (Fig. 1D). The surfaces of cartilage from the medial side of knee joints were rougher than those from the lateral side and had the visible exposure of subchondral bone (Fig. 1E). Therefore, the cartilage tissues of each patient were divided into relatively undamaged or damaged regions, and cartilage destruction in OA cartilage was ascertained by SO/FG staining using the Osteoarthritis Research Society International (OARSI) grading. The ACSL4 expression, a LM regulator in ferroptosis, and 4-hydroxynonenal (4-HNE), an indication of lipid peroxidation, were strongly expressed in the OA-affected areas of human cartilage affected by OA, but was scarcely detectable in unimpacted regions of the cartilage tissues (Fig. 1F–I).

Analogously, among the regulators of these annotations, ACSL4 expressions were consistently upregulated in different OA animal models (Fig. 2A). Furthermore, an ACLT murine model was established, which truly reflected the clinical characteristics of OA [38]. In the mouse model of OA, the histological scores of cartilage destruction and synovitis determined by SO/FG staining were significantly increased after ACLT surgery. Consistently, the ACSL4 and 4-HNE expressions were significantly elevated at the monitoring point following the surgical induction of OA (Fig. 2A–F). Taken together, the interplay between LM and ferroptosis could be responsible for the exacerbation of cartilage degeneration in the human and murine OA cartilage.

Fig. 2.

Interplay of LM and ferroptosis is detected in the mouse OA cartilage. (A) Heatmaps depicting the expression of components of long-chain fatty-acyl-CoA metabolic process, fatty acid metabolism, and ferroptosis pathway in OA-relevant conditions. Public transcriptome datasets obtained from cartilage from OA animal models (GSE42295, GSE8077, and GSE28958) were analyzed. (B) Sham-operation or ACLT surgery was conducted in 10-week-old mice. Joint sections were stained with SO/FG. Scale bar, 100 μm; 50 μm. ACSL4 and 4-HNE were detected by immunohistochemistry of in cartilage sections of mice following Sham-operation or ACLT at 8 weeks following surgery. Scale bar, 50 μm; 50 μm. (C-F) OA manifestations were scored by including cartilage destruction and synovial inflammation. Positive cells of ACSL4 and 4-HNE were manifested employing a bar-dot plot. n = 5. (G) The cell viability was ascertained by employing the CCK-8 assay for 48 h following the specified treatments. n = 3. (H) Representative images of mitochondrial ultrastructure under indicated treatments were observed by TEM. Primary articular chondrocytes were treated with erastin as a positive control. Scale bar, 200 nm. (I) MDA contents were measured 48 h after indicated treatments. n = 3. (J) Lipid peroxidation was determined by C11 BODIPI 581/591 staining and assessed by flow cytometry in chondrocytes 48 h after indicated treatments. (K-L) Quantitative RT-PCR and western blotting analyses of ACSL4, MMP13, and Col2α1 of primary culture of articular chondrocytes 48 h post indicated treatments. Error bars are means ± s.d., *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with multiple comparisons using the Tukey method.

Inflammation triggers lipid homeostasis dysregulation and evokes ferroptosis in chondrocytes

Inflammatory mediator IL-1β serves as a key pathogenic factor of OA under the pro-inflammatory micro environment of osteoarthritic joints [42]. The pathogenesis of OA includes a transition in the cellular metabolism of chondrocytes from a quiescent regulatory state to a metabolically active one, resulting in metabolic abnormalities characterized by oxidative stress, mitochondrial malfunction, and cell death [12], [13]. Following the descriptions of reports, primary murine chondrocytes were exposed to varying doses of IL-1β for 48 h. The results showed that IL-1β significantly reduced cell viability, as ascertained by cell counting kit (CCK-8) assay (Supplementary Fig. 1). To examine if the suppression in cell viability caused by IL-1β was due to lipid peroxidation-driven ferroptosis, the cells were treated with erastin, an effective classic inducer of ferroptosis, or Fer-1, a specific small-molecule inhibitor of lipid peroxidation, and measured cell death. Researchers found that IL-1β led to a significant increase in cell death of chondrocytes, comparable to the cytotoxicity induced by erastin, suggesting the existence of ferroptosis in IL-1β-stimulated chondrocytes. In contrast, when the cells were treated with Fer-1, which inhibited lipid peroxidation, it prevented cell death mediated by IL-1β (Fig. 2G). Ferroptosis is a lipid peroxidation-driven cell death cascade with special mitochondria characteristics [43]. The change of mitochondrial ultrastructure was observed in murine primary chondrocytes treated with IL-1β. The images revealed that mitochondria generally appeared smaller and less tubular, with darker-stained membranes and distinct disrupted inner membrane foldings, consistent with ferroptosis-related ultrastructure features induced by erastin (Fig. 2H).

To further confirm the triggerable regulator of chondrocyte ferroptosis by an inflammatory mediator, researchers investigated whether IL-1β interfered in lipid homeostasis during chondrocyte ferroptosis in vitro. C11-BODIPY staining indicated that lipid peroxidation was significantly upregulated by IL-1β stimulation, and measurement of lipid metabolite MDA contents showed similar results: intracellular level of MDA was increased by IL-1β. However, the treatment of Fer-1 protected against lipid peroxidation induced by IL-1β, as manifested by C11 BODIPY 581/591 fluorescence, and the reduced lipid metabolite level of MDA (a by-product of lipid peroxidation) (Fig. 2I, J and Supplementary Fig. S2). Next, researchers investigated the protective implications of targeting the lipid deregulation-driven chondrocyte ferroptosis on IL-1β-induced catabolism in vitro. Accumulating evidence has shown that shows that ACSL4 is found as an indication and contributes to ferroptosis and could provide an alternative source of lipid biosynthesis for peroxidation [44], [45], [46], [41], [47]. qRT-PCR and western blotting assay showed a high expression of ACSL4 in murine articular chondrocytes following treatment with IL-1β. qRT-PCR and western blotting analyses confirmed that Fer-1 effectively downregulated matrix metallopeptidase 13 (MMP13) expression level and upregulated Col2α1 expression level (Fig. 2K, L). Immunofluorescence staining of murine articular chondrocytes was conducted to further detect the expression of catabolic/anabolic markers at the protein level. Contrasted to the IL-1β group, the numbers of cells positive for MMP13 and Col2α1 were reversed by Fer-1 (Supplementary Fig. S3). Consistently, Fer-1 significantly reduced glycosaminoglycan (GAG) contents deposited in the ECM, evidenced by Alcian blue staining, demonstrating that restoring lipid homeostasis could improve cartilage homeostasis mainly by suppressing lipid peroxidation in chondrocytes (Supplementary Fig. S3). Overall, these results implied that inflammation dictates chondrocyte ferroptosis sensitivity mainly by promoting lipid peroxide accumulation.

Metformin protects chondrocytes affected by osteoarthritis by reducing their sensitivity to ferroptosis

Here, researchers determined if chondrocyte ferroptosis sensitivity by IL-1β-induced-dysregulation of lipid homeostasis was reversed through metformin supplementation. Primary mouse chondrocytes were extracted and incubated in a medium enriched with metformin or PBS. The CCK-8 assay demonstrated that appropriate concentrations (≤10 mM) of metformin within our observation time were non-toxic to chondrocytes (Supplementary Fig. S4A). Chondrocyte ferroptosis contributes to cartilage degeneration in OA [48]. Then, researchers assessed the implications of metformin on IL-1β-induced chondrocyte catabolism in vitro and found that the mRNA level of catabolic gene MMP13 in IL-1β-treated chondrocytes was strongly downregulated. However, the mRNA level of the anabolic gene Col2a1 was significantly increased after metformin administration (Fig. 3A). Western blotting analysis demonstrated that metformin effectively reduced the MMP13 protein and increased the Col2α1 protein level (Fig. 3B). Consistently, metformin also regulated cartilage homeostasis by restoring IL-1β-caused matrix metabolism in articular chondrocytes in vitro, as demonstrated by immunofluorescence staining and Alcian blue staining (Supplementary Fig. S5). Also, the ACSL4 mRNA and protein levels, a supplier of alternative source of lipid biosynthesis for peroxidation [49], were decreased in metformin-treated chondrocytes relative to chondrocytes treated with IL-1β (Fig. 3A, B).

Fig. 3.

Metformin shields chondrocytes affected by osteoarthritis by hindering their susceptibility to ferroptosis. (A-B) Quantitative RT-PCR and western blotting were employed to analyze the expression of ACSL4, MMP13, and Col2α1 in primary articular chondrocytes following specific treatments. (C-D) Primary articular chondrocytes were treated with Fer-1 or metformin incubation with or without IL-1β and stained with Nile Red to observe lipid droplets (LDs). Dotted areas were magnified at the bottom left. Relative LDs areas were quantified. n = 3. Scale bar, 20 μm. (E) Lipid peroxidation was determined by C11 BODIPI 581/591 staining by flow cytometry in chondrocytes 48 h after indicated treatments. (F) BODIPY staining of murine articular chondrocytes. Primary articular chondrocytes were incubated in a medium enriched with metformin or PBS in the existence of IL-1β, and lipid peroxidation was determined by C11 BODIPI 581/591 staining. Scale bars, 100 μm. (G) MDA contents were quantified 48 h following specified treatments. n = 3. (H) TEM images showing representative mitochondria in primary culture of articular chondrocytes 48 h after indicated treatments. Scale bar, 200 nm. Error bars are means ± s.d., *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with several comparisons employing the Tukey method. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To determine the effects of metformin on osteoarthritic chondrocytes, researchers investigated whether metformin interfered in chondrocyte sensitivity to ferroptosis in vitro. Accordingly, ferroptosis was closely linked with LDs, which accumulated and sensitized osteoarthritic chondrocytes to ferroptosis. Subsequently, Nile Red was employed as a stain for LDs and it was discovered that the quantity of LDs was hindered in cells treated with metformin, as contrasted to cells treated with IL-1β (Fig. 3C, D). Moreover, Fer-1, an inhibitor of lipid peroxide accumulation, attenuated the IL-1β cytotoxicity to chondrocytes, suggesting the harmful effect of lipid peroxidation on IL-1β-treated chondrocytes. Metformin elevated the resistance of chondrocytes to IL-1β-potentiated ferroptosis, as evidenced by similar results on Fer-1 in cell viability assay (Supplementary Fig. S4B). Moreover, C11-BODIPY staining manifested that lipid peroxidation was upregulated by IL-1β stimulation and reversed by metformin (Fig. 3E). The significant decrease was also observed in the percentage of BODIPY-positive cells in chondrocytes induced by metformin compared to the control (Fig. 3F). Moreover, as for the by-product of lipid peroxidation, metformin significantly reduced the content of MDA (Fig. 3G). TEM analysis showed a classical ferroptosis-like mitochondrial morphologic feature in chondrocytes upon treatment with IL-1β. Notably, metformin treatment improved IL-1β-mediated mitochondrial impairment (Fig. 3H).

In addition, researchers quantified the alterations in the expression of genes linked to iron homeostasis in chondrocytes and quantified the amount of cellular iron employing the Phen Green FL fluorescent probe. The outcomes manifested that compared to the control group, chondrocytes stimulated with IL-1β exhibited upregulation of divalent metal transporter 1 (DMT1), transferrin receptor (TFRC), and ferroportin (FPN) gene expression, as well as an increase in iron content. However, when exogenous metformin was applied in the existence of IL-1β, the DMT1, TFRC, and FPN gene expression and the iron content in chondrocytes were downregulated (Supplementary Fig. S6A–E). Moreover, compared to the control group, IL-1β treatment upregulated the expression of solute carrier family 7 member 11 (SLC7A11) in chondrocytes, while glutathione peroxidase (GPX4) expression was downregulated. Exogenous metformin downregulated the SLC7A11 expression in chondrocytes stimulated by IL-1β but had no effect on GPX4 (Supplementary Fig. S7). Taken together, these findings demonstrate that metformin prevents lipid storage and lipid peroxide accumulation to protect chondrocytes from ferroptosis.

Metformin reshapes lipid availability and ameliorates chondrocyte ferroptosis sensitivity via the AMPK/ACC signaling pathway

The genes profile in LM employing the GSE dataset indicated the contribution of the AMPK pathway in the pathogenesis of OA (Fig. 4A). ACC, a downstream effector of AMPK, regulates fatty acid metabolism and has been implicated in metabolic diseases [35]. To further elucidate the impact of AMPK/ACC pathway activation on IL-1β-induced lipid availability and ferroptosis sensitivity in chondrocytes, researchers employed compound C, a potent antagonist of AMPK that disrupts the interaction between the γ subunit and the α subunit of AMPK, thereby inhibiting AMPK activation [50], [51]. Compound C was administered to murine articular chondrocytes in the existence or lack of IL-1β. The p-AMPK and p-ACC levels were reduced during IL-1β-induced ferroptosis, as shown by western blotting. Western blotting validated a significant reduction in p-AMPK expression following treatment (Fig. 4B). In contrast, the phosphorylation levels of these proteins were significantly upregulated by metformin treatment, indicating that metformin inhibited ferroptosis by activating the AMPK/ACC signaling pathway (Fig. 4B). As indicated above, the protein concentrations of ACSL4 demonstrated characteristics of a dual indicator of lipid availability and ferroptosis. Therefore, researchers detected the protein by western blotting and found that metformin-induced downregulation of ACSL4 at the protein level was eliminated by exposure of compound C into chondrocytes (Fig. 4C). Moreover, compound C in chondrocytes nullified the beneficial implications of metformin on IL-1β-impaired chondrocyte homeostasis and osteoarthritic cartilage degeneration. (Fig. 4C). In line with that, metformin failed to improve LD deposition in compound C-treated chondrocytes (Fig. 4D, E). Collectively, these data manifested that the active expression of the AMPK/ACC signaling pathway was an important prerequisite for reshaping lipid availability and ameliorating chondrocyte ferroptosis sensitivity by metformin.

Fig. 4.

Metformin reshapes lipid availability and ameliorates chondrocyte ferroptosis sensitivity via the AMPK/ACC signaling pathway. (A) The regulation of LM by the AMPK signaling pathway in chondrocytes. KEGG analyses of RNA-seq data showing enriched pathways based on dataset GSE41342 in GEO. (B-C) Western blotting analysis of total AMPK, p-AMPK, ACC, p-ACC, ACSL4, MMP13, and Col2α1 expression in chondrocytes treated as indicated. (D-E) Images of chondrocytes after certain treatments stained with Nile Red to identify LDs and were quantified. The dotted sections are enlarged in the bottom left. n = 3. Scale bars, 20 μm. Error bars are means ± s.d., *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with multiple comparisons using the Tukey method. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Metformin alleviates the progression of osteoarthritis by regulating chondrocyte LM and suppressing ferroptosis

To investigate the metformin function in OA therapeutics, researchers ascertained the effects of improving the osteoarthritic joint after ACLT surgery with or without metformin application. Metformin was given through intra-articular injections every three days for 8 consecutive weeks beginning two weeks following ACLT surgery (Fig. 5A). Three-dimensional reconstructions of knee joints revealed that ACLT mice treated with the vehicle had elevated joint mineralization, namely on the medial side and the irregular bone surface of the tibia and femur. Metformin-treated mice exhibited ameliorated pathological changes (Fig. 5B). SO/FG staining of cartilage and OARSI score analysis showed that cartilage degeneration appeared at 8 weeks post-surgery. Based on histological assessment of cartilage destruction, intra-articular injections of metformin attenuated articular cartilage degradation together with a reduced OARSI grading score (Fig. 5C, D). Histological observation of subchondral bone sclerosis and osteophyte maturation showed that metformin also improved ACLT-induced post-traumatic OA development (Fig. 5C, E, F). The results showed that treating ACLT mice with metformin diminished synovial inflammation compared to vehicle-treated ACLT mice (Fig. 5C, G). Therefore, these findings indicated that metformin relieved OA progression mainly by improving cartilage degeneration.

Fig. 5.

Metformin ameliorates OA progression and restores joint bone remodeling in ACLT mice. (A) Illustrative model of the time course for establishing the ACLT model of OA mice, both with and without the injection of metformin. (B) Illustrative micro-CT scans show the knee joint from the AP and LAT views, emphasizing the alterations seen in the femoral and tibial surfaces, and the medial and lateral menisci. A-P, anterior-posterior. (C) Representative SO/FG staining and H&E staining of articular cartilage of the WT mice intrarticularly treated daily with vehicle or metformin 8 weeks following ACLT surgery. The inset in the images is shown as magnified images in the bottom row. Scale bar, 50 μm (second row); 100 μm (third row); 100 μm (fourth row). (D-G) Scores of OA manifestations include cartilage destruction, subchondral bone sclerosis, osteophyte formation, and synovitis. (H) Immunohistochemistry assays were performed using histological sections of mice with sham operation or ACLT surgery with/without metformin treatment at week 8 post-treatment. Scale bar, 50 μm. (I-J) Positive cells of ACSL4 and 4-HNE were indicated by a bar-dot plot. n = 5. Error bars are means ± s.d., *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with multiple comparisons using the Tukey method.

Since metformin impeded the activation of the lipid biosynthetic process and ferroptosis pathway in vitro, as well as subsequent cartilage degeneration, researchers examined whether lipid dysregulation-driven chondrocyte ferroptosis was involved in metformin-mediated protection of osteoarthritic cartilage and investigated the expression of lipogenesis-related ACSL4 and lipid peroxidation marker 4-HNE in response to metformin in the OA mouse model. The outcomes manifested that contrasted to the sham group, the percentage of chondrocytes positive for ACSL4 and 4-HNE in the ACLT group increased. Metformin reverted it to a level similar to that of sham mice (Fig. 5H–J). ACC functions as the enzyme that controls the speed of the key process in the biosynthesis of long-chain fatty acids and acts as the enzyme that limits the pace of lipogenesis. ACC is rapidly controlled by reversible phosphorylation; phosphorylation inactivates this enzyme, whereas dephosphorylation activates itself [52]. The result showed that the increase of p-ACC indicated that metformin could induce downregulation of lipogenesis in vivo (Supplementary Fig. S8A, B). Immunofluorescence staining revealed that metformin treatment heightened the level of the matrix protein Col2α1 and significantly reduced the MMP13 level compared to vehicle-treated ACLT mice (Supplementary Fig. S8C–E). Together, these in vivo data implied that metformin alleviated OA progression and played a chondroprotective role by regulating lipogenesis and chondrocyte sensitivity to ferroptosis.

Ferroptotic chondrocytes cause the recruitment and chemotaxis of synovial macrophages via CCL2

Synovitis is a common clinical symptom for OA patients, characterized by synovial membrane thickening [53]. Increased infiltration of inflammatory macrophages into the synovium elevates the production of inflammatory mediators, which trigger chondrocyte cell death signaling and break the matrix homeostasis [54]. In turn, ferroptosis was identified as extremely pro-inflammatory [55], and how ferroptotic chondrocytes amplified the joint inflammation remains poorly understood. Based on transcriptome data and enrichment analysis of KEGG signaling pathways of OA or OA-relevant conditions, the result indicated that changed genes were enriched in various inflammatory pathways, such as the ‘‘response to cytokine”, ‘‘response to tumor necrosis factor,” ‘‘response to lipopolysaccharide,” and ‘‘response to molecules of bacterial origin” (Fig. 6A). Various inflammation-linked genes were upregulated in IL-1β-treated chondrocytes, such as CCL7, LCN2, and PTGS2. Macrophage chemokines, especially CCL2, were also significantly increased (Fig. 6B). As expected, ELISA results showed an increase in the secretion of CCL2 in primary cultured chondrocytes induced by erastin and IL-1β, whereas metformin treatment significantly decreased the secretion level of CCL2 (Fig. 6C). To determine whether ferroptotic chondrocytes induce macrophage migration, researchers introduced murine chondrocytes treated with erastin or IL-1β in the lower chambers of the transwells and the monocyte/macrophage cell line RAW264.7 in the upper chambers (Fig. 6D). The outcomes manifested a significant rise in the migration of cells caused by ferroptotic chondrocytes. However, when RS102895, a CCR2 inhibitor, was administered to the top chamber, it effectively reduced the RAW264.7 cell migration (Fig. 6E). Further, consistent with the inhibition of monocyte chemoattractant in vitro, macrophage numbers were hindered in the knee joints of metformin-treated ACLT mice compared to vehicle-treated ACLT mice (Fig. 6E and Supplementary Fig. S9). Collectively, our results showed that ferroptotic chondrocytes activated a pro-inflammatory response by attracting macrophages via the release of CCL2. Additionally, researchers observed that metformin effectively suppressed inflammation caused by ferroptosis in vitro.

Fig. 6.

Metformin diminishes joint inflammation amplified by ferroptotic chondrocytes-derived CCL2 in vitro and in mice of ACLT model. (A) Enrichment analysis of KEGG signaling pathway in the OA-relevant condition based on dataset GSE75181 in GEO. (B) Altered inflammation-correlated genes in OA-relevant conditions. Transcriptome datasets obtained from IL to 1β-treated chondrocytes (GSE75181, GSE6119, and GSE104793) were analyzed. (C) ELISA analysis of CCL2 expressed in primary chondrocytes post-indicated treatments. n = 3. (D-E) Monocyte/macrophage cell line RAW264.7 cells migration assay (visualized by crystal violet staining) cocultured with primary articular chondrocytes post-indicated treatments. (F) Immunofluorescence staining of iNOS and F4/80 in synovium of mice with sham operation or ACLT surgery with/without metformin treatment at week 8 post-treatment. Scale bar, 100 μm. (G) Immunofluorescence staining of CD206 and F4/80 in synovium of mice with sham operation or ACLT surgery with/without metformin treatment at week 8 post-treatment. Scale bar, 100 μm. n = 5. Error bars are means ± s.d., *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with multiple comparisons using the Tukey method. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Metformin diminishes inflammation in the joint micro environment of OA model mice

Synovitis peaks at early-stage before declining after ACLT surgeries but throughout the mid and late stages of OA it keeps at a degree greater than that in sham mice. Synovial macrophage infiltration is essential in amplifying the inflammation in OA [56]. Researchers confirmed the pro-inflammatory effects of ferroptotic chondrocytes by inducing the chemotaxis of synovial macrophages and determined whether metformin exerted a therapeutic effect on the infiltration of inflammatory macrophages in the joints in vivo. The immunofluorescent staining findings demonstrated a significant elevation in the M1 macrophage marker iNOS within the F4/80+ cells present in the synovial tissue of the vehicle-treated ACLT animals. Nevertheless, the proportion of iNOS-expressing cells within the F4/80 positive cell population was decreased in the metformin-treated ACLT mice in comparison to the ACLT mice (Fig. 6F and Supplementary Fig. S11A). Conversely, the percentage of cells expressing the M2 macrophage marker CD206 within the F4/80+ cell population was greater in the metformin therapy group contrasted to the ACLT group (Fig. 6G and Supplementary Fig. S11B). These outcomes manifested that metformin significantly regulated synovial macrophage polarization in the joint micro environment of OA model mice. Taken together, ferroptotic chondrocytes could trigger inflammation in the joint by regulating pro-chemoattract cytokine release and macrophage recruitment, and the suppression of ferroptosis by metformin could ameliorate inflammation-induced joint injury.

Discussion

Given the complicated etiology of OA, multiple risk factors have been identified in this process, including inflammation, imbalance in cellular redox, mechanical loading, chondrocyte death, and extracellular matrix degradation. Inflammation in the joint environment is essential in OA progression [57], [58]. For the first time, this paper showed a metabolic association between inflammation and ferroptosis in OA. Researchers also found that metformin could attenuate chondrocyte ferroptosis by modulating lipid metabolic reprogramming. Mechanistically, AMPK signaling was activated and negatively regulated fatty acid synthesis by inhibitory phosphorylating ACC, leading to decreased PUFAs for lipid peroxidation. Additionally, metformin was beneficial to osteoarthritis by blocking the CCL2/CCR2 signaling and reducing macrophage accumulation, synovial inflammation, and cartilage injury during OA development.

Fine regulation of cellular metabolism could maintain cartilage homeostasis, but metabolic imbalances triggered various pathological conditions by stimulating pro-inflammatory cytokines [21]. Accumulated evidence confirms the essential roles of inflammation in OA pathogenesis, and there are many reports on the crosstalk between inflammation and metabolic factors in cartilage [59]. However, the direct relationship between the inflammation process and metabolic pathways and how it determines chondrocyte fate remains incompletely comprehended. In this current study, researchers found that ferroptosis-related LM was dysregulated in the human OA cartilage, the OA-induced animal cartilage based on bioinformatics analysis of transcriptomics datasets and Immunohistochemistry of regulator of LM in ferroptosis (such as ACSL4) and lipid peroxidation marker (such as 4-HNE). Additionally, researchers demonstrated that the suppressive implications of IL-1β on chondrocyte survival have the potential to be blocked by Fer-1, a lipid peroxidation inhibitor. Electron microscopy showed that IL-1β affected the mitochondrial morphology in chondrocytes, which was similar to the effects seen with erastin, a positive control. Overall, these results indicated that inflammation triggered the disorder of lipid homeostasis and promoted ferroptosis in chondrocytes, indicating the interaction of LM and ferroptosis during OA.

Earlier investigations revealed ferroptosis-related features, including abnormalities of iron metabolism, lipid peroxidation, and mitochondrial disorders, that were closely correlated with accelerating OA features [60], [61], [62]. Recently, several studies have highlighted that chondrocyte ferroptosis contributed to OA development [55]. However, the link between OA and ferroptosis was determined depending on observations that OA progression was obstructed by DFO or loss of activity of LPO scavenger GPX4 and iron overload-induced chondrocyte ferroptosis in vivo [63], [64]. Our work not only provided an intriguing finding that lipid availability mediated cartilage destruction by triggering ferroptosis susceptibility in chondrocytes, which opened the way into mechanisms underlying ferroptosis-susceptible state in the OA micro environment and expanded our knowledge of potential therapeutic approaches for treating ferroptosis in OA pathogenesis. ACSL4 and LPCAT3 were first found as pro-ferroptotic proteins attributed to their ability to enhance the PUFAs integration into membrane lipids, even though GPX4 and SLC7A11 typically inhibit ferroptosis. Furthermore, the inactivation of ACSL4 is recognized as a crucial mechanism for suppressing ferroptosis in many circumstances, as determined by studying cell lines that are resistant to ferroptosis and conducting a CRISPR suppression screen [41]. Determination of the exact source and function of LPO in ferroptosis and detailed underlying mechanisms driving the LM interaction and ferroptosis require further exploration. In this study, researchers observed that IL-1β also increased lipogenic ACSL4 protein expression and intracellular lipid depositions and promoted the accumulation of LPO.

Clinical studies have suggested that individuals with obesity have a significantly increased incidence of OA. Lipid accumulation in the joint is also detected during the first stages of osteoarthritis, before any histological changes occur [23]. This indicates that lipids contribute in developing OA and that there are mechanisms that regulate their presence. Metformin has shown promise as a therapeutic agent for inflammatory and lipid metabolic disorders, in addition to its established anti-diabetic properties [65]. The most well-known mechanism of metformin-mediated cytoprotection is AMPK pathway activation, which modulates cellular metabolism and protects cells from stress response [66]. It was shown that AMPK, a serine/threonine protein kinase, could enhance ATP preservation to reprogram cellular metabolism via phosphorylating its targets [67]. The important process regulated by AMPK is LM. AMPK-mediated phosphorylation of serine 79 of ACC inhibits fatty acid synthesis [68]. Acetyl-CoA is necessary for providing a carbon donor in de novo synthesis of fatty acids in LM. It is employed by ACC to facilitate the production of malonyl-CoA, which is a key component in fatty acids synthesis [69]. Lipid peroxidation in ferroptosis preferentially occurs on lipid bilayers of cell membranes, disrupting the integrity of cellular membranes and generating secondary toxic products, such as 4-HNE and MDA [70]. Our data support that metformin activated AMPK, which then phosphorylated and inactivated ACC1, resulting in restrained lipid biosynthesis and ferroptosis inhibition. These outcomes manifest that targeting the interaction between LM and ferroptosis could be an attractive approach for treating OA. Moreover, researchers also demonstrated that chondrocytes experienced disrupted iron homeostasis under IL-1β stimulation, while metformin can restore iron homeostasis in chondrocytes under IL-1β stimulation. In fact, to evaluate the metformin implications on the expression of ferroptosis-relevant genes and features, our findings suggested that the occurrence of ferroptosis in chondrocytes under IL-1β stimulation may increase the demand for cysteine and glutamate and the decrease in GPX4 expression confirms this possibility. The fact that metformin has no effect on GPX4 expression and downregulates SLC7A11 suggests that metformin does not alleviate chondrocyte ferroptosis through the SLC7A11-GPX4 pathway. Overall, this further confirmed that metformin could alleviate chondrocyte ferroptosis by regulating LM. While metformin has been shown to inhibit ferroptosis in the treatment of vascular calcification and polycystic ovary syndrome [71], [72], [73], its role in ferroptosis in OA and the mechanisms underlying how chondrocytes regulate ferroptosis have not been reported. At the same time, current studies on OA and ferroptosis mainly focus on independent indicators of ferroptosis, without elucidating upstream pathways or the use of potential therapeutic drugs. Our studies expanded application principle of metformin used clinically to treat OA patients with several metabolic diseases including dyslipidemia. Researchers found metformin reshaped lipid availability and ameliorated chondrocyte ferroptosis sensitivity via the AMPK/ACC pathway.

Chronic low-grade inflammation contributes to the OA development, even at the early stage of this disease [74]. The interaction between cartilage and synovium undergoes pathological alterations that are crucial in the development of this inflammatory process. The intricate biochemical interaction between cartilage and synovium is essential in the pathogenesis of OA [75]. It is increasingly acknowledged that dying cells release pro-inflammatory mediators, which worsen tissue damage. Like different types of controlled cell death, ferroptosis is recognized to be pro-inflammatory. However, the specific interactions between ferroptosis and inflammation are not well understood. The publically accessible statistics indicate that the inflammatory pathways were activated in chondrocytes during ferroptosis. Of the genes associated with inflammation, the macrophage chemokines CCL2 were particularly and strongly activated by ferroptosis. Furthermore, according to our experiments, ferroptotic chondrocytes did induce the recruitment and infiltration of synovial monocytes. Various monocyte chemokine systems were shown to be increased, including CCL2, CCL3, and CCL4, in the inflamed synovium and cartilage compared with control specimens [76]. A recent study showed that monocytes were recruited via CCL2/CCR2 and propagated the inflammatory response and cartilage damage [77]. Metformin and Fer-1 significantly reduced the migration of monocyte/macrophage cell line RAW264.7 and levels of inflammatory infiltration in the synovium. Herein, our findings linked ferroptosis-immune crosstalk in the inflammatory joint micro environment and defined important cellular and molecular mediators.

Taken together, this work presents evidence for the LM’s interaction with ferroptosis in the pathogenesis of OA. Importantly, our outcomes indicate that metformin exerts protective effects on osteoarthritis by alleviating sensitivity to chondrocyte ferroptosis. The underlying mechanism was related to lipid metabolic reprogramming via the AMPK/ACC signaling pathway and suppression of lipid availability. Researchers found that ferroptotic chondrocytes induced the recruitment of macrophages and amplified joint inflammation, which was diminished by metformin in vitro and in OA mice models. In human clinical studies, metformin application has been demonstrated to have a beneficial effect on long-term knee joint outcomes in those with knee OA and obesity [78]. Furthermore, our studies proved that ferroptosis was involved in human OA pathogenesis and contributed to OA progression by the evidence that lipid metabolism and ferroptosis could be responsible for the exacerbation of cartilage degeneration in the human and murine OA cartilage, and metformin reshapes lipid availability and ameliorates chondrocyte ferroptosis sensitivity via the AMPK/ACC signaling pathway. Our studies expanded application principle of metformin used clinically to treat OA patients with several metabolic diseases including dyslipidemia. While this study has identified alterations in lipid metabolism implicated in the development of osteoarthritic cartilage lesions, the specific molecular mechanisms remain unclear. In the future, we can utilize genetically edited animal models and extensive omics technologies to elucidate the detailed mechanisms by which lipids contribute to cartilage pathology in osteoarthritis.

Data sharing statement

The accompanying authors may provide all the data produced for this research upon a fair request.

Compliance with ethics requirements

All animal experiments in this study were performed in accordance with institutional guidelines and approved by the Experimental Animal Ethics Committee of Third Military Medical University (AMUWEC2019235). Patients involved in the study provided consent, and the study was approved by medical ethics regulations of the Department of Orthopedics, Xinqiao Hospital, Army Medical University (2023-YANDI056-01).

Declaration of competing interest

The authors 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 material

The following are the Supplementary data to this article: