Resveratrol offers therapeutic potential in obesity-related osteoarthritis by targeting ferroptosis via G3BP1-SGs/mTOR modulation

Abstract

The increased ferroptosis by obesity contributes to the progression of osteoarthritis (OA). Resveratrol has protective effects against obesity-related OA. However, the underlying mechanisms remain unclear. Here, we investigated the mechanism of resveratrol against obesity-related OA using male C57BL/6 mice and chondrocyte models. The results demonstrated that high-fat diet (HFD) and IL-1β exacerbated cartilage damage in mice and chondrocytes, accompanied by worsening indicators of ferroptosis and upregulation of Ras-GTPase-activating protein binding protein 1 (G3BP1) expression and stress granules (SGs) assembly, which responds to environmental stressors such as obesity. Resveratrol alleviated these effects. After G3BP1 was silenced, resveratrol no longer had the effect of inhibiting ferroptosis and reducing cartilage degradation; while mTOR inhibition rescued ferroptosis-related protective proteins. This study suggests that resveratrol exerts protective effects against obesity-related OA, at least in part, by modulating ferroptosis-related changes in chondrocytes through the G3BP1-SGs/mTOR pathway. These findings provide new insights into OA pathogenesis and highlight resveratrol as a promising therapeutic strategy for obesity-related OA.

Resveratrol alleviates obesity-related osteoarthritis by suppressing chondrocyte ferroptosis through G3BP1 stress granule-dependent modulation of mTOR signaling, supporting a potential therapeutic strategy for metabolic osteoarthritis.

Introduction

Osteoarthritis (OA) is a common chronic degenerative joint disease, mainly associated with factors such as age, genetics, obesity, and trauma. With the aging population and rising obesity rates, the prevalence of OA has been increasing, especially among the elderly and obese populations. The main pathological characteristics of OA include articular cartilage degeneration, abnormal subchondral bone calcification, and synovitis, which lead to pain, stiffness, joint deformity, and functional limitations, severely impacting the quality of life¹. Currently, no curative treatment is available and clinical treatment primarily relies on non-steroidal anti-inflammatory drugs (NSAIDs) and analgesics, which are often associated with significant side effects. Long-term use may result in serious complications. Therefore, to clarify the mechanisms of OA pathogenesis and explore safe and effective early treatments are of great clinical significance.

Ferroptosis is a kind of programmed cell death which is characterized by iron-dependent lipid peroxidation, mainly triggered by the consumption of glutathione (GSH), decreased activity of glutathione peroxidase 4 (GPX4), peroxidation of polyunsaturated fatty acids (PUFAs), and elevated intracellular ferrous ion (Fe²⁺) levels^2,3. Ferroptosis-related iron metabolism, lipid peroxidation, and amino acid metabolism pathways play important roles in OA development⁴, making them potential therapeutic targets.

Obesity accelerates tissue damage by promoting ferroptosis in multiple organs, such as the liver, heart, and kidneys. For example, obesity-related iron overload induces oxidative stress in the liver, triggering ferroptosis and exacerbating liver damage⁵. Ferroptosis has been confirmed in obesity-induced damage to muscles and other organs, contributing to the worsening of associated pathological conditions⁶. Obesity is a major risk factor for OA. Some studies have suggested that obesity-induced chronic inflammation and lipid metabolism disorders promote ferroptosis in chondrocytes, accelerating cartilage degradation, disrupting cartilage homeostasis, and exacerbating OA cartilage degeneration^7,8. However, the underlying mechanisms between obesity, ferroptosis, and OA remain to be fully elucidated.

Resveratrol, a natural non-flavonoid polyphenol compound, has garnered attention for its broad biological activities, including anti-inflammatory, antioxidant, and immunomodulatory effects^9,10. Resveratrol has been shown to have significant protective effects on articular cartilage by effectively inhibiting the catabolism of chondrocytes and synovial cells¹¹. Furthermore, resveratrol has no obvious toxic effects, making it a promising candidate for early treatment in OA. It is reported that resveratrol can regulate ferroptosis in various cells, such as cardiomyocytes, hepatocytes, and cancer cells, primarily through the Nrf2/GPX4 pathway. Moreover, resveratrol may regulate ferroptosis through the p53 pathway, supporting its therapeutic potential in cancer. Our previous study has shown that resveratrol can improve obesity-related OA in rats and mouse model¹². So, can resveratrol mitigate the obesity-related OA by inhibiting chondrocyte ferroptosis? What is the specific regulatory mechanism? It is not yet clear. In the present study, we performed bioinformatics analysis using the GSE254844 dataset, incorporating obesity, ferroptosis, and OA-related constraints to identify differentially expressed genes (DEGs). Preprocessing was done using the Limma package in R, and p-values were adjusted using the Benjamini & Hochberg method. Ultimately, we predicted that Ras-GTPase-activating protein binding protein 1 (G3BP1), a key regulator of stress granules (SGs) formation, may be a crucial target for resveratrol in inhibiting ferroptosis in obesity-related OA.

G3BP1 is a multifunctional binding protein closely associated with the assembly of SGs¹³, and SGs formation is an important cellular mechanism for responding to external stress. Studies showed that oxidative stress and inflammatory responses caused by obesity may promote the expression of G3BP1 and the formation of SGs¹⁴. Therefore, we speculated that resveratrol may exert its protective effect against obesity-related OA by modulating G3BP1 expression and SGs assembly, thereby inhibiting ferroptosis.

Thus, in this study, we investigated the mechanism of obesity-related OA progression linked to ferroptosis and of resveratrol alleviating obesity-related OA via the G3BP1/SGs-mediated ferroptosis. This study provides theoretical insights into OA pathogenesis and identifies potential therapeutic targets for future clinical strategies.

Results

Resveratrol alleviated cartilage damage, improved motor function and regulated cartilage matrix metabolism in obesity-related osteoarthritis mice

In the HFD group (mice treated with HFD and DMM surgery), marked pathological alterations in cartilage were observed. Histological Safranin O and Fast Green staining showed disrupted chondrocyte alignment at the cartilage surface, blurred tide lines, and substantial cartilage erosion (Fig. 1A). Quantitative scoring confirmed these findings, with significantly elevated OARSI and Modified Mankin scores (Fig. 1B, C). The HFD + RES group exhibited more orderly chondrocyte alignment, clearer tide lines, and notably reduced cartilage damage, reflected in lower histological scores (Fig. 1A–C). Gait analysis revealed substantial motor impairment in the HFD group, including shortened ground contact time, decreased average speed, prolonged four-paw contact time, and a lower contact rate index. These impairments were partially reversed by resveratrol treatment, which improved all four parameters to varying degrees (Fig. 1D–G).

Fig. 1. Resveratrol alleviated obesity-related OA in mice.

A Representative H&E, Safranin O, and Safranin O–Fast Green staining of knee cartilage; arrows in the H&E staining indicate chondrocyte defects or abnormal proliferation (scale bar = 100 μm, magnification = 400×); arrows in the Safranin O staining indicate irregular tidemark; arrows in the Safranin O-Fast Green staining indicate cartilage defects and matrix loss (scale bar = 100 μm, magnification = 40×); B OARSI scoring; C Modified Mankin scoring; D Stance time; E Average speed; F Simultaneous four-paw contact time; G Stance rate index; H Expression of MMP13 protein in articular cartilage; I Quantification of MMP13-positive cells from immunohistochemistry; J Expression of type II collagen protein in articular cartilage; K Quantification of type II collagen-positive cells from immunohistochemistry. Compared with the CON group, *P < 0.05, **P < 0.01; compared with the HFD group, #P < 0.05, ##P < 0.01; compared with the Sham joint, &P < 0.05, &&P < 0.01; n = 6 biologically independent animals per group for histology, immunohistochemistry and gait analysis; Data are presented as mean ± SD.

Moreover, the HFD group showed increased expression of the matrix-degrading enzyme MMP13, as demonstrated by immunohistochemical staining and quantification. This elevation was attenuated in the HFD + RES group (Fig. 1H, I). In parallel, expression of type II collagen, a key component of the cartilage extracellular matrix, was reduced in the HFD group, while resveratrol treatment restored its expression (Fig. 1J, K). The effect of resveratrol on basal metabolic parameters (body weight, body fat, food intake and energy intake) in obese mice was shown in Supplementary Fig. 1.

Resveratrol decreased serum lipid peroxidation and inhibited ferroptosis of cartilage in obese mice with OA

In the HFD-induced obesity-related OA mouse model, serum levels of malondialdehyde (MDA) and triglycerides (TG) were significantly increased, reflecting metabolic disturbance and enhanced lipid peroxidation (Fig. 2A, B). In the HFD + RES group, both TG and MDA levels were markedly reduced. Expression of acyl-CoA synthetase long-chain family member 4 (ACSL4) in cartilage was significantly elevated in the HFD and Erastin groups, particularly in DMM joints. This increase was attenuated in the HFD + RES group (Fig. 2C, D). In contrast, expression of glutathione peroxidase 4 (GPX4) was significantly decreased in the HFD and Erastin groups and was partially restored following resveratrol treatment (Fig. 2E, F). Furthermore, fluorescence staining showed increased Fe²⁺ accumulation in articular cartilage of the HFD group in both Sham and DMM joints. In the HFD + RES group, Fe²⁺ signals were substantially reduced (Fig. 2G, H). Oil Red O staining showed increased lipid droplet accumulation in the DMM joints of CON mice. In HFD-fed mice, lipid deposition was already elevated in Sham joints and was further increased in DMM joints. Resveratrol treatment reduced lipid droplet accumulation in both Sham and DMM joints (Fig. 2I, J).

Fig. 2. The effects of resveratrol on lipid peroxidation and ferroptosis in obese mice with OA.

A Serum MDA levels; B Serum TG levels; C Expression of ACSL4 in articular cartilage, arrows point to brown particles representing positive cells (scale bar = 100 μm; magnification = 40×), rectangles indicate regions magnified to 400×; D Quantification of ACSL4-positive cells from immunohistochemistry; E Expression of GPX4 in articular cartilage; F Quantification of GPX4-positive cells; G Fluorescence staining of Fe²⁺ in cartilage sections (scale bar = 100 μm; magnification = 40×); H Quantification of Fe²⁺ fluorescence intensity; I Representative Oil Red O staining of knee joint sections (scale bar = 50 μm; magnification = 40×); J Quantification of the relative area of lipid droplets. Compared with the CON group, *P < 0.05, **P < 0.01; compared with the HFD group, ##P < 0.01; compared with the Sham joint, &&P < 0.01; n = 12 biologically independent animals per group for serum analyses; n = 6 biologically independent animals per group for histological and fluorescence analyses; Data are presented as mean ± SD.

Resveratrol suppressed ferroptosis in IL-1β-treated chondrocytes

Normal mouse primary chondrocytes exhibited typical cellular morphology (Fig. 3A), with strong proteoglycan staining shown by toluidine blue (Fig. 3B) and robust type II collagen expression confirmed by immunohistochemistry (Fig. 3C). Cell viability of IL-1β, resveratrol and Erastin was shown in Fig. 3D–F. In this in vitro model, IL-1β was used to mimic an OA-like inflammatory microenvironment. Fluorescence staining showed increased Fe^2⁺ accumulation and ROS signals following IL-1β and Erastin exposure (Fig. 3G). Quantitative analysis further showed that both Fe^2⁺ and ROS fluorescence intensities were increased in the IL-1β and Erastin groups and were reduced after resveratrol treatment (Fig. 3H, I).

Fig. 3. The effects of resveratrol on ferroptosis in IL-1β-treated chondrocytes.

A Morphology of mouse primary chondrocytes under the microscope (scale bar= 100 μm; magnification = 200×); B Toluidine blue staining for proteoglycans in primary chondrocytes; C Type II collagen staining in primary chondrocytes; D Effects of different concentrations of IL-1β on chondrocyte viability; E Effects of different concentrations of resveratrol on chondrocyte viability; F Effects of different concentrations of Erastin on chondrocyte viability; G Fluorescence images of Fe²⁺ (scale bar = 50 μm; magnification = 200×) and ROS (scale bar = 50 μm; magnification = 100×) in chondrocytes; H Quantification of Fe²⁺ fluorescence intensity; I Quantification of ROS fluorescence intensity; J Western blot analysis of MMP13, GPX4, and SLC7A11 protein levels; K Quantification of MMP13 expression; L Quantification of GPX4 expression; M Quantification of SLC7A11 expression. Compared with the CON group, *P < 0.05, **P < 0.01; compared with the IL-1β group, #P < 0.05, ##P < 0.01; n = 3 biologically independent experiments; Data are presented as mean ± SD.

In line with these findings, ferroptosis-related markers also exhibited notable alterations. IL-1β and Erastin induced a significant upregulation of MMP13 and downregulation of GPX4 and SLC7A11 in primary chondrocytes, as shown by immunoblotting and quantitative analysis. Resveratrol treatment suppressed MMP13 expression while restoring GPX4 and SLC7A11 levels (Fig. 3J–M). Similar results were observed in C28/I2 chondrocytes, confirming the consistency of resveratrol-mediated effects across different chondrocyte models. Details were provided in Supplementary Figs. 2–3.

G3BP1 and SGs assembly were upregulated in mice with obesity-related OA, while resveratrol mitigated OA by targeting G3BP1

Using bioinformatics analysis of the GEO dataset GSE254844, a total of 521 differentially expressed genes (DEGs) associated with obesity and OA were identified, including 327 upregulated and 194 downregulated genes (Fig. 4A). Among these, G3BP1 was highlighted as a potential key regulator involved in metabolic pathways, RNA processing, and cellular stress responses. GO enrichment analysis linked G3BP1 to RNA binding (GO:0003723) and SGs assembly (GO:0034063), while KEGG analysis indicated enrichment in the MAPK signaling pathway (ko04013) and protein processing in the endoplasmic reticulum (ko04141) (Fig. 4B, C). The molecular docking prediction results showed that the crystal structure of the NTF2-like domain of G3BP1 can be ligand docked with resveratrol. Resveratrol forms hydrophobic interactions with G3BP1 at Tyr19, Phe131, and Tyr133 sites, and its hydroxyl group forms hydrogen bonds with the Asn100 and Leu98 side chains of G3BP1 (Fig. 4D).

Fig. 4. Resveratrol attenuated obesity-related OA by downregulating G3BP1 and inhibiting SGs assembly.

A Venn diagram and volcano plot of differentially expressed genes; B GO analysis for molecular functions; C KEGG pathway enrichment analysis; D Molecular docking prediction of binding sites between G3BP1 and resveratrol; E G3BP1 protein bands; F Quantification of G3BP1 protein expression; G Co-expression of G3BP1 and TIA1 in immunofluorescence, magnification 100×, scale bar = 100 μm. Green fluorescence represents G3BP1 protein, red fluorescence represents TIA1 protein, blue fluorescence represents DAPI-stained nuclei, and yellow fluorescence indicates SGs formed by the co-expression of G3BP1 and TIA1 proteins; H Quantification of G3BP1 and TIA1 co-expression in immunofluorescence; I Western blot of G3BP1 protein; J Quantification of G3BP1 protein expression; K Co-expression of G3BP1 and TIA1 in mouse primary chondrocytes using immunofluorescence, 100×, scale bar = 100 μm. Green fluorescence indicates G3BP1 protein, red fluorescence indicates TIA1 protein, blue fluorescence indicates DAPI-stained nuclei, and yellow fluorescence indicates SGs formed by the co-expression of G3BP1 and TIA1 proteins; L Quantification of G3BP1 and TIA1 co-expression in immunofluorescence; M Effect of resveratrol on heat-induced degradation of G3BP1 protein; N Quantification of G3BP1 degradation dynamics under resveratrol treatment. Compared with the CON group, **P < 0.01; compared with the HFD group, ##P < 0.01; compared with the sham joint, &P < 0.05, &&P < 0.01; compared with the IL-1β group, #P < 0.05; compared with the IL-1β + RES group, ##P < 0.01; n = 6 biologically independent animals per group for tissue western blotting and immunofluorescence analyses; n = 3 biologically independent experiments for cell western blotting, immunofluorescence and cellular thermal shift assays; Data are presented as mean ± SD.

Western blot analysis showed that G3BP1 expression was significantly elevated in HFD group, while resveratrol treatment reduced its expression to near-control levels (Fig. 4E, F). Immunofluorescence revealed increased co-localization of G3BP1 and TIA1 in these groups, which was attenuated following resveratrol treatment (Fig. 4G, H). In IL-1β-induced chondrocytes, similar trends were observed. Resveratrol treatment decreased G3BP1 expression and reduced SGs formation, as evidenced by decreased G3BP1/TIA1 co-staining intensity (Fig. 4I–L). Similar regulatory effects of resveratrol on G3BP1 expression and G3BP1/TIA1 co-localization were also observed in C28/I2 chondrocytes, as shown in Supplementary Fig. 4. The thermal shift assay showed that in the DMSO group, G3BP1 protein levels decreased progressively as temperature increased from 35 °C to 70 °C, with substantial degradation observed at 60 °C and above. In contrast, resveratrol -treated cells exhibited delayed G3BP1 degradation under thermal stress, indicating enhanced thermal stability of the protein (Fig. 4M, N).

Resveratrol inhibited ferroptosis by modulating G3BP1/mTOR signaling pathway

In the HFD group, the p-mTOR/total mTOR protein ratio was elevated in both Sham and DMM joints compared to the CON group, with a more pronounced increase observed in DMM joints (Fig. 5A, B). In primary mouse chondrocytes, IL-1β stimulation led to increased expression of G3BP1 and MMP13 and decreased levels of GPX4 and SLC7A11 relative to the NC group. These alterations were reversed by resveratrol treatment, which decreased G3BP1 and MMP13 levels and increased GPX4 and SLC7A11 expression (Fig. 5C–G). To investigate whether G3BP1 contributes to mTOR regulation, we examined p-mTOR expression in G3BP1-silenced chondrocytes. Compared to the IL-1β + RES group, G3BP1 knockdown resulted in elevated p-mTOR levels and an increased p-mTOR/total mTOR ratio (Fig. 5H, I), indicating attenuation of resveratrol mediated mTOR inhibition. The knockdown efficiency of G3BP1 is shown in Supplementary Fig. 5. Further analysis showed that G3BP1 knockdown increased G3BP1 and p-mTOR expression compared with the IL-1β+RES group, whereas rapamycin treatment reduced p-mTOR activation without restoring G3BP1 expression (Fig. 5J–L).

Fig. 5. Resveratrol suppressed mTOR signaling by downregulating G3BP1.

A Phosphorylated mTOR (p-mTOR) and total mTOR protein expression in cartilage; B Quantification of the p-mTOR/total mTOR protein ratio; C G3BP1, MMP13, GPX4, and SLC7A11 protein expression in primary chondrocytes; D Quantification of G3BP1 protein expression; E Quantification of MMP13 protein expression; F Quantification of GPX4 protein expression; G Quantification of SLC7A11 protein expression; H p-mTOR and total mTOR protein expression following G3BP1 knockdown in IL-1β-treated primary chondrocytes; I Quantification of p-mTOR/total mTOR ratio; J G3BP1 and p-mTOR expression in cells treated with IL-1β + RES with or without G3BP1 knockdown and rapamycin; K Quantification of G3BP1 protein expression; L Quantification of p-mTOR/total mTOR ratio; M MMP13, GPX4, and SLC7A11 expression under G3BP1 knockdown with or without rapamycin treatment; N Quantification of MMP13 protein expression; O Quantification of GPX4 protein expression; P Quantification of SLC7A11 protein expression. Compared with the CON group, **P < 0.01; compared with the HFD group, ##P < 0.01; compared with the Sham joint, &&P < 0.01; compared with the NC group, *P < 0.05, **P < 0.01; compared with the IL-1β group, #P < 0.05, ##P < 0.01; compared with the IL-1β + RES group, & &P < 0.01; compared with the IL-1β + RES + siG3BP1 group, $$P < 0.01; ;n = 6 for animal samples; n = 3 biologically independent cell experiments; Data are presented as mean ± SD.

The expression patterns of MMP13, GPX4, and SLC7A11 under extended treatment conditions were further assessed (Fig. 5M). G3BP1 knockdown diminished the regulatory effects of resveratrol, with increased MMP13 (Fig. 5N) and decreased GPX4 (Fig. 5O) and SLC7A11 (Fig. 5P) expression compared to the IL-1β + RES group. Rapamycin treatment reversed these changes in both the IL-1β+siG3BP1+rapa and IL-1β + RES + siG3BP1 + rapa groups, partially restoring the molecular profiles observed under resveratrol treatment. Similar results were observed in C28/I2 chondrocytes, and details are provided in the Supplementary Fig. 6.

Discussion

This study investigated the involvement of the G3BP1-SGs/mTOR axis in chondrocyte ferroptosis during obesity-related OA and the protective effect of resveratrol. We found that obesity-related metabolic stress was associated with increased G3BP1 expression, enhanced SGs assembly, aggravated ferroptosis-related changes, and disrupted cartilage homeostasis. Resveratrol attenuated these alterations. Collectively, our findings support a role for the G3BP1-SGs/mTOR axis in obesity-related OA.

Ferroptosis has been increasingly recognized as a contributor to OA progression. Under inflammatory and iron-overload conditions, chondrocytes are prone to ferroptotic cell death, which may exacerbate cartilage degeneration¹⁵. Studies reported that Erastin treatment of SW1353 cells increased MMP13 expression and decreased type II collagen levels, while ferroptosis inhibition reversed these effects and restored matrix protein synthesis^16,17, suggesting that ferroptosis plays a critical role in OA pathogenesis. Metabolic alterations caused by obesity, such as hyperlipidemia and elevated lipoxygenase (LOX) activity (facilitated the oxidation of free fatty acids) may increase ROS accumulation and activate redox-sensitive signaling pathways, which can ultimately trigger ferroptosis¹⁸. Despite obesity showed an obvious effect of promoting ferroptosis, the specific role and regulatory mechanisms to ferroptosis in cartilage remain poorly understood. In this study, Erastin was intra-articularly injected to induce ferroptosis in knee joints, not only serving as a positive control for ferroptotic induction in chondrocytes, but also to assess the contribution of ferroptosis to OA progression. The HFD model exhibited hallmark features of chondrocyte ferroptosis, including increased ACSL4 expression, decreased GPX4 expression, and elevated Fe²⁺ accumulation and lipid peroxidation levels. These data suggested that obesity was accompanied by ferroptosis-related changes in chondrocytes, which were associated with the worsening of obesity-related OA. Previous studies have shown that ferroptosis inhibition can alleviate OA-related cartilage degeneration and that GPX4-dependent ferroptosis regulation is involved in OA pathology¹⁹. However, in the absence of in vivo rescue experiments using ferroptosis inhibitors or genetic anti-ferroptotic strategies, the current data should be interpreted as supporting the involvement of ferroptosis rather than proving it as a definitive driving mechanism in this model. Consistent with these findings, Oil Red O staining indicated that HFD was associated with increased local lipid deposition in the knee joint, whereas DMM further aggravated this change under the HFD background. These findings suggested that HFD and DMM may act as parallel contributors with additive effects in this model, with HFD providing a metabolic stress background and DMM serving as an additional local mechanical insult. This interpretation is also supported by previous studies showing that high-fat diet-induced obesity and metabolic inflammation can independently promote OA-like changes, while high-fat feeding may further prime the joint for more severe pathological changes after injury^20,21.

Next, we continued to explore how resveratrol reduced ferroptosis and improved obesity-related OA. Although the mechanisms by which resveratrol alleviates OA have been extensively studied, this study mainly focused on the role of obesity in the pathogenesis of OA. Through bioinformatics analysis of public databases and molecular docking prediction, we identified G3BP1 as a potential target associated with the anti-ferroptotic effect of resveratrol in chondrocytes. Molecular docking suggested a potential interaction between resveratrol and the NTF2-like domain of G3BP1. In addition, the in vitro binding assay reported by Naomi et al. using resveratrol-coupled agarose magnetic beads also implied an interaction between resveratrol and G3BP1²². Our cellular thermal shift assay further showed that resveratrol altered the thermal stability of G3BP1 in cells, supporting potential target engagement rather than directly proving physical binding. Further studies, such as SPR, ITC, or pull-down assays, are still needed to determine whether resveratrol directly binds to G3BP1.

We observed that a HFD not only promoted ferroptosis but also significantly upregulated G3BP1 expression in mouse articular cartilage, suggesting a potential link between ferroptosis and G3BP1 activity in the context of obesity-related OA. This HFD-induced upregulation of G3BP1 may be associated with systemic low-grade inflammation and mechanical stress induced by obesity²³. Consistent with our findings, Bai et al. reported elevated G3BP1 expression in macrophages and liver tissues of HFD-fed mice²⁴. In our study, resveratrol treatment suppressed the aberrant G3BP1 upregulation in articular cartilage, paralleling its inhibitory effect on ferroptosis. Li et al. previously demonstrated that G3BP1 contributed to ferroptosis by increasing intracellular iron levels and promoting lipid peroxidation²⁵. Furthermore, resveratrol has been reported to regulate the p53 signaling pathway via G3BP1 to exert anti-cancer effects²⁶. In the present study, IL-1β was used to establish an OA-like inflammatory microenvironment in vitro for mechanistic verification and evaluation of resveratrol intervention, rather than to represent a specific upstream insult such as metabolic stress or mechanical injury. This is consistent with previous studies in which IL-1β was widely used to induce an OA-like catabolic phenotype in chondrocytes and to assess the protective effects of resveratrol in vitro²⁷. In vitro, we found that G3BP1 silencing abolished the anti-ferroptotic and chondroprotective effects of resveratrol, indicating that G3BP1 is essential for mediating these actions. Collectively, our findings demonstrated that resveratrol alleviated obesity-related OA by inhibiting G3BP1-mediated chondrocyte ferroptosis.

G3BP1 is a core scaffold protein essential for SGs assembly in response to various cellular stresses such as oxidative stress, heat shock, and endoplasmic reticulum stress²⁸. It plays a central role in initiating and stabilizing SGs by interacting with untranslated mRNAs and other RNA-binding proteins like TIA1, with which it functionally competes to regulate SGs dynamics^29,30. Under stress, G3BP1 primarily promotes the nucleation of SGs, while TIA1 supports SGs stability^30,31. In our study, we observed that HFD treatment significantly increased the co-expression of G3BP1 and TIA1 in cartilage tissue, suggesting enhanced SGs formation. Resveratrol treatment reduced G3BP1 expression and decreased G3BP1-TIA1 co-localization, indicating that the anti-ferroptotic effect of resveratrol may be partially attributed to suppression of SGs assembly. Notably, G3BP1 function appears to depend on its competitive interaction with other RNA-binding proteins (e.g., TIA1 or FMRP), which may determine whether SGs serve a protective or deleterious role. Recent evidence in non-cancerous tissues showed that G3BP1-mediated SGs could act as regulators of oxidative stress, inflammation, and cell death pathways, with either protective or pathological outcomes depending on context and duration^14,32. In general, SGs formed during acute stress are transient and adaptive structures associated with translational repression, which help cells temporarily reorganize mRNA metabolism and maintain survival under stress^33,34. However, under chronic stress conditions, SGs may become more persistent and less readily cleared, and their prolonged existence has been linked to disturbed proteostasis and pathological cellular responses. Therefore, the biological significance of SGs should be interpreted in a context- and duration-dependent manner rather than as a uniformly deleterious event. In this study, the increased G3BP1/TIA1 co-localization observed under the HFD background is more likely to reflect a chronic stress-associated imbalance than a simple acute adaptive response.

While excessive and persistent SGs assembly has been shown to increase neuronal sensitivity to oxidative stress, thereby contributing to neurodegenerative disorders³⁵. In contrast, complete loss of G3BP1 did not appear to impair acute stress responses (e.g., heat shock or arsenite exposure), but resulted in long-term disruption of proteostasis, highlighting its essential role in maintaining cellular homeostasis³⁶. These findings suggest that G3BP1 may need to be maintained within a narrow range to support stress adaptation, and that either excessive upregulation or depletion may be detrimental to chondrocyte homeostasis. This interpretation is biologically plausible, as previous studies have shown that G3BP1 acts as a tunable switch for SG assembly and that its intracellular abundance can influence SG formation thresholds and phase separation behavior³⁷. In our study, obesity-related upregulation of G3BP1 was associated with ferroptosis and cartilage damage, whereas G3BP1 silencing impaired the protective effect of resveratrol. The relationship between SGs and mTOR signaling is unlikely to be unidirectional, as previous studies have shown that the mTOR-S6 kinase pathway can promote SG assembly and that mTORC1 signaling may crosstalk with SG dynamics^38,39. Given that chronic metabolic abnormalities and metabolic stress-induced joint inflammation are important contributors to obesity-related OA⁴⁰, the observed increase in p-mTOR in our model may be more closely related to chronic metabolic overload than to SG assembly itself, whereas SG abnormalities may represent a concomitant stress response or a potential amplifying event. Our findings aligned with these observations. G3BP1 was required to maintain basal stress response capacity, and its excessive upregulation in obesity promoted ferroptosis and cartilage damage. Resveratrol mitigated this by rebalancing SGs dynamics. Furthermore, when G3BP1 was silenced, the protective effect of resveratrol was markedly attenuated, supporting an important role for G3BP1 in SG assembly and chondroprotection. The observed reduction in SGs after resveratrol treatment may result from either direct inhibition of G3BP1 expression or competitive interference with its interaction with other RNA-binding proteins such as TIA1.

Previous studies have shown that mTOR, a central regulator of cellular metabolism, is closely related to ferroptosis^41,42. In addition, G3BP1-driven SG assembly may influence the translation of metabolism-related transcripts and the balance of ferroptosis-associated proteins, suggesting a potential link between SG dynamics and ferroptosis-related regulation^43,44. G3BP1 has also been reported to participate in the regulation of the lysosomal TSC/mTOR complex^45,46, while mTOR activity may in turn affect SG dynamics and stability^38,47,48. Together, these findings provide a potential structural and functional basis for crosstalk between G3BP1/SGs and mTOR signaling. In the present study, IL-1β increased p-mTOR expression, whereas resveratrol attenuated this change. G3BP1 silencing weakened the inhibitory effect of resveratrol on mTOR signaling, and rapamycin partially restored the protective phenotype after G3BP1 knockdown. These findings suggest that mTOR signaling is functionally involved in the G3BP1-associated effects observed in this study. However, our data do not directly demonstrate that excessive SG assembly activates mTOR by disrupting the TSC-mTOR interaction. Therefore, this part should be regarded as a working model, and further studies examining mTOR localization and G3BP1-TSC complex dynamics are required.

It should also be noted that resveratrol is a pleiotropic molecule with broad anti-inflammatory, antioxidant, and metabolic regulatory activities, and its protective effects in obesity-related OA are unlikely to be explained by a single pathway alone⁴⁹. In addition to the G3BP1-SGs/mTOR axis highlighted in this study, other mechanisms may also contribute to the anti-ferroptotic effect of resveratrol. For example, the Nrf2/GPX4 axis is a key antioxidant pathway involved in ferroptosis regulation and OA progression, whereas SIRT1/p53 signaling has also been implicated in ferroptosis-related regulation and may represent another route through which resveratrol exerts protective effects^50,51. Therefore, the current findings support the involvement of the G3BP1-SGs/mTOR axis, but do not exclude the contribution of other parallel pathways.

Although the present study provides mechanistic evidence for the G3BP1-SGs/mTOR axis in obesity-related OA, this evidence was obtained mainly from animal and cell models. Therefore, the current findings should be interpreted as supporting a potential mechanistic axis rather than providing direct clinical confirmation. Nevertheless, existing human evidence supports the translational relevance of the present study. Previous epidemiological and clinical studies have shown that metabolic abnormalities and metabolic syndrome are associated with an increased risk of OA, particularly knee OA, supporting the clinical relevance of combining metabolic stress and local joint injury in the HFD-DMM model⁵². In addition, human OA-related transcriptomic evidence has implicated ferroptosis-related alterations in OA pathology, and previous studies in human chondrocyte-related systems have also shown protective effects of resveratrol against IL-1β-induced inflammatory and catabolic responses⁵³. Future studies incorporating human OA cartilage samples or primary human chondrocytes will be important for further validating the G3BP1-SGs/mTOR axis and ferroptosis-related changes in a clinically relevant setting.

In summary, this study suggests that resveratrol alleviates obesity-related OA by modulating chondrocyte ferroptosis and improving cartilage matrix metabolism through the G3BP1-SGs/mTOR signaling pathway, as summarized in Fig. 6. Using a mouse model combining HFD-induced obesity with DMM surgery, together with chondrocyte-based experiments, we show that obesity-related stress is associated with aggravated cartilage degeneration, enhanced ferroptosis-related changes, increased G3BP1 expression and SGs assembly. Resveratrol attenuates these pathological alterations, whereas G3BP1 silencing weakens its anti-ferroptotic and chondroprotective effects, supporting an important role for G3BP1-SGs/mTOR signaling in this process. These findings provide mechanistic insight into obesity-related OA and support resveratrol as a potential therapeutic strategy, while further targeted and clinically relevant studies are needed to validate this pathway and its translational potential.

Fig. 6.

Schematic of resveratrol inhibition of cartilage ferroptosis via G3BP1-SGs/mTOR signaling. The schematic summarizes the proposed mechanism by which obesity-related stress promotes G3BP1 expression, SGs assembly, mTOR activation and chondrocyte ferroptosis, whereas resveratrol attenuates these changes and protects cartilage matrix homeostasis. Figure 6 was created by the authors using Figdraw under license ID [IAARWdba40]. No BioRender or generative AI-created elements were used.

Methods

Animal experimental design and handling

All animal experiments were approved by the Experimental Animal Ethics Committee of China Medical University (Approval No.: CMU20222005). We have complied with all relevant ethical regulations for animal use. Forty-eight male C57BL/6 mice, aged 8 weeks and weighing 22–24 g, were purchased from Beijing HFK Bioscience Co., Ltd. Mice were housed in a specific pathogen-free (SPF) environment with controlled temperature (22 °C ± 2 °C), humidity (60–70%), and a 12-hour light-dark cycle, with free access to water and food.

Mice were allowed to acclimatize for 1 week before experimental procedures. Mice were randomly divided into four groups (n = 12): control group (CON), high-fat diet group (HFD), resveratrol treatment group (HFD + RES), and ferroptosis positive control group (Erastin). The CON and Erastin groups were fed a standard diet (10% energy from fat), while the HFD and HFD + RES groups were fed a HFD (60% energy from fat). Mice in the HFD + RES group were received resveratrol (45 mg/kg/day) via gavage, dissolved in 0.5% carboxymethylcellulose sodium solution. Other groups were gavaged daily with the same volume of 0.5% carboxymethylcellulose sodium solution. Mice in the Erastin group were received intra-articular injections of Erastin (0.5 μg/joint) twice a week for eight consecutive weeks to induce ferroptosis. The right hind leg of all mice underwent destabilization of the medial meniscus (DMM) surgery, while the left hind leg received a sham surgery. The animal experimental workflow is summarized in Scheme 1. All experimental groups were maintained on their respective diets for 18 weeks. Throughout the experiment, body weight, food intake, and behavioral activities of all animals were regularly monitored and recorded. Humane endpoints included severe weight loss, persistent inability to access food or water, severe wound infection, marked lameness or distress, or moribund condition. At the 18th week, all mice underwent body composition analysis using a small animal body composition analyzer (Bruker Minispec LF90II) while conscious.

Scheme 1.

Experimental workflow of animal handling. The schematic illustrates the timeline of high-fat diet feeding, resveratrol administration, Erastin injection, DMM surgery, sham surgery and sample collection. Scheme 1 was created by the authors using Microsoft PowerPoint. The mouse illustration was drawn by the authors. No third-party, BioRender or generative AI-created elements were used.

Knee joint DMM surgery

All mice underwent bilateral procedures during the seventh week of diet intervention (15 weeks of age), with DMM surgery performed on the right knee and sham surgery conducted on the left knee. During the procedure, the mice were anesthetized using an isoflurane anesthesia system (RWD Life Science, Shenzhen, China), and under sterile conditions, the medial meniscotibial ligament was exposed and transected to induce knee joint instability. Postoperatively, each mouse received analgesics and antibiotics at the surgical joint site.

Gait analysis

Starting from the eighteenth week, gait analysis was performed using a gait analysis system (CatWalk XT 9.1, Noldus Information Technology, Wageningen, Netherlands) to evaluate stride length, stride frequency, and weight distribution. Gait data were automatically collected by the software in a single assessment conducted before sacrifice.

Tissue sample collection and processing

At the end of the experiment (week 18), the mice were fasted for 12 h to minimize interference from food intake on serum biochemical parameters. Subsequently, the mice were anesthetized with isoflurane gas, the eyeballs were removed to collect blood samples. The removed eyeballs rapidly released blood from the ocular venous plexus and were collected into non-anticoagulant blood collection tubes. After standing for 30 minutes, the samples were centrifuged at 3000 rpm for 15 min (4 °C, 5804 R refrigerated centrifuge, Eppendorf, Hamburg, Germany) to separate the serum. The separated serum was transferred to cryogenic tubes and stored in a −80 °C ultralow temperature freezer (DW-86L828J, Haier, Qingdao, China) for subsequent oxidative stress and lipid metabolism testing (using MDA assay kit and TG assay kit, Biotime Biotechnology, Shanghai, China; Nanjing Jianchen Bioengineering Institute, Nanjing, China).

After the eyeball removal, the mice were euthanized by cervical dislocation. The knee joint tissues were collected and processed as follows: part of the tissue was fixed in 4% paraformaldehyde solution for 24 h (Sigma-Aldrich, USA), then decalcified in 10% EDTA solution (pH 7.4, Solarbio, Beijing, China), with the decalcification solution replaced every three days until the bone tissue softened (approximately 6-8 weeks). After decalcification, the samples were dehydrated using an ASP300S automatic dehydrator (Leica Microsystems, Wetzlar, Germany), and then embedded in paraffin using an EG1150H paraffin embedding machine (Leica Microsystems, Wetzlar, Germany), sectioned to a thickness of 5 μm using an RM2235 rotary microtome (Leica Microsystems, Wetzlar, Germany) for HE staining and immunohistochemical analysis. Another part of the cartilage tissue was rapidly frozen in liquid nitrogen and stored in a –80 °C ultralow temperature freezer for subsequent protein extraction and detection.

Histological staining and pathological evaluation

Paraffin sections of knee joints were sequentially subjected to Hematoxylin and Eosin (H&E) staining (Sigma-Aldrich, St. Louis, USA), Safranin O staining (Sigma-Aldrich, St. Louis, USA), and Safranin O-Fast Green staining (Solarbio, Beijing, China) to assess changes in cartilage morphology and matrix composition. Oil Red O staining was additionally performed to evaluate the morphology and distribution of lipid droplets in mouse knee joint tissues. Frozen sections were brought to room temperature, rinsed in distilled water for 10 min, and briefly washed with 60% isopropanol to optimize staining. The sections were then stained with freshly prepared Oil Red O working solution for 15 min in the dark, followed by a brief rinse in 60% isopropanol to remove excess dye. After mounting with glycerol gelatin, the sections were scanned and imaged using an Aperio CS2 compact pathology scanner (Leica Biosystems,Wetzlar,Germany). For quantitative analysis, five random high-power fields (40×) were selected from the articular cartilage and adjacent subchondral region of each section, and the Oil Red O-positive area within the ROI was quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA) and expressed as the relative area of lipid droplets. Pathological sections were observed under an optical microscope (Olympus BX53, Olympus, Tokyo, Japan). Analysis of pathological sections was conducted using the Osteoarthritis Research Society International (OARSI) scoring system (OARSI, USA) and the Mankin scoring system. The OARSI scoring system ranges from 0 to 6, where 0 indicates no cartilage degeneration, and scores of 4-6 indicate moderate to severe cartilage degeneration. The Mankin scoring includes parameters such as cartilage structure, cell arrangement, proteoglycan staining intensity, and pathological changes in subchondral bone, with a total score range of 0-14; higher scores indicate more severe cartilage damage. Each slide was evaluated by two researchers in a double-blind manner.

Isolation and culture of primary mouse chondrocytes

Primary mouse chondrocytes were isolated from the knee joint cartilage of 5-day-old C57BL/6 J mice. Cartilage tissues were digested in 0.2% collagenase II (Sigma-Aldrich, St. Louis, USA) at 37 °C in a water bath for 4 h. The digested cells were filtered through a 70 μm cell strainer (BD Falcon, Franklin Lakes, USA), then centrifuged at 1200 rpm for 5 min. Cells were resuspended in high-glucose DMEM medium (Gibco, Waltham, USA) containing 10% fetal bovine serum (FBS) (Gibco, Waltham, USA) and penicillin-streptomycin solution (100 U/mL penicillin and 100 μg/mL streptomycin) (Gibco, Waltham, USA) and cultured in a 37 °C, 5% CO₂ incubator (Thermo Fisher, Waltham, USA). When cells reached 80–90% confluence, they were digested with 0.25% trypsin-EDTA (Gibco, Waltham, USA) for passage. First-passage cells were used for subsequent experiments.

Culture of human C28/I2 chondrocytes

Human immortalized C28/I2 chondrocytes were purchased from Bluefcell Biotechnology Co., Ltd. (Shanghai). Cells were cultured in DMEM containing 10% fetal bovine serum (FBS) and penicillin-streptomycin (100 U/mL and 100 μg/mL, respectively), at 37 °C in a 5% CO₂ incubator.

Cell viability assay

Primary mouse chondrocytes and C28/I2 chondrocytes were counted using a hemocytometer and seeded into 96-well plates at 5×10³ cells per well. After 24 h, the medium was replaced with 10 μL CCK-8 solution (Beyotime, Shanghai, China) and 90 μL complete culture medium. For treatments, recombinant mouse IL-1β (PeproTech, Rocky Hill, NJ, USA) was applied in a gradient (0, 5, 10, 20, 30, 40, 50 ng/mL). Erastin (Sigma-Aldrich, St. Louis, MO, USA), a ferroptosis inducer, was applied in a gradient (0, 5, 10, 15, 20, 25, 30 μM) and resveratrol (Sigma-Aldrich, USA) was used in a gradient (0, 6.25, 12.5, 25, 50, 100, 200 μmol/L). Plates were incubated at 37 °C, 5% CO₂ (Thermo Fisher, USA) for 2 h. All treatments were conducted with triplicate wells, and absorbance at 450 nm was measured using a microplate reader (BioTek, USA). The experiments were repeated three times.

Cell treatments

Based on the results of the cell viability assays, the optimal working concentrations for subsequent experiments were determined to be 10 ng/mL IL-1β, 50 μmol/L resveratrol, and 10 μmol/L Erastin. Primary mouse chondrocytes were assigned to four treatment groups: the CON group was treated with complete culture medium containing 0.1% DMSO; the IL-1β group was treated with 10 ng/mL IL-1β and 0.1% DMSO; the IL-1β + RES group was treated with 10 ng/mL IL-1β, 50 μmol/L RES, and 0.1% DMSO; and the Erastin group was treated with 10 μmol/L erastin and 0.1% DMSO. C28/I2 cells were subjected to the same treatment protocol for consistency. IL-1β was used to establish an OA-like inflammatory microenvironment in vitro. Additional experimental conditions, such as siRNA transfection and rapamycin treatment, are described in subsequent sections.

siRNA transfection

Small interfering RNA (siRNA) targeting G3BP1 was synthesized by General Bio (Anhui) Co., Ltd. When C28/I2 chondrocytes and mouse primary chondrocytes reached 70–80% confluence, siRNA transfection was performed using Lipofectamine™ 2000 (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. Briefly, 10 nM siRNA and Lipofectamine™ 2000 were diluted in 50 μL Opti-MEM (Gibco, Waltham, MA, USA) separately, incubated for 5 minutes, then mixed and incubated for another 20 min to form siRNA-lipid complexes. The complexes were added to each well (final volume 2 mL), and cells were incubated in serum-free medium for 6 hours. After transfection, the medium was replaced with complete culture medium containing 0.1% DMSO, and cells were further treated as designed. The specific sense sequences of the siRNAs used were as follows: siRNA-G3BP1-1: 5’-CUGCCACACCAAGAUUCGGTT-3’; siRNA-G3BP1-2: 5’-CAUUAACAGUGGUGGGAAATT-3’; siRNA-G3BP1-3: 5’-AGAAAGAAAUCCACAGGAATT-3’; Negative control: 5’-UUCUCCGAACGUGUCACGUTT-3’; FAM negative control: 5’-UUCUCCGAACGUGUCACGUTT-3’; GAPDH positive control: 5’-GUAUGACAACAGCCUCAAGTT-3’.

Rapamycin treatment

After siRNA transfection to knock down G3BP1, cells were treated according to experimental groupings. In the IL-1β+siG3BP1+rapa group, the culture medium was replaced with medium containing recombinant human IL-1β (10 ng/ml, PeproTech, Rocky Hill, NJ, USA) and 20 nM rapamycin (Selleck, Houston, TX, USA). In the IL-1β + RES+siG3BP1+rapa group, the culture medium was supplemented with recombinant human IL-1β, 50 μM resveratrol (Sigma-Aldrich, St. Louis, MO, USA), and 20 nM rapamycin, and the cells were incubated for 24 hours. Untransfected and negative control siRNA-transfected groups were included as controls to ensure the reliability and scientific rigor of the experimental data.

Immunohistochemistry

Paraffin-embedded cartilage tissue sections were dewaxed and rehydrated through a graded ethanol series. Antigen retrieval was performed using sodium citrate buffer (pH 6.0) (Solarbio, Beijing, China) at 95 °C for 10 min. After washing with PBS (Gibco, Waltham, MA, USA), sections were blocked with 3% BSA (Sigma-Aldrich, St. Louis, MO, USA) for 30 min to reduce nonspecific binding. Primary antibodies (G3BP1, type II collagen, MMP13, ACSL4, and GPX4, diluted 1:200, Abcam, Cambridge, UK) were added and incubated overnight at 4 °C. Then, sections were washed three times with PBS, each for 5 min, followed by incubation with HRP-conjugated secondary antibodies (Cell Signaling Technology, Danvers, MA, USA, diluted 1:1000) at room temperature for 1 h. Color development was achieved using DAB staining solution (Dako, Glostrup, Denmark), and nuclei were counterstained with hematoxylin (Sigma-Aldrich, St. Louis, MO, USA). Finally, sections were observed under an optical microscope (Olympus BX53, Tokyo, Japan), and positive signals were quantitatively analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Immunofluorescence analysis

Cells were fixed and permeabilized with 0.5% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA) for 10 min, then blocked with 5% BSA (Sigma-Aldrich, St. Louis, MO, USA) for 30 minutes to reduce nonspecific binding sites. Primary antibodies (Proteintech, Rosemont, IL, USA,G3BP1 1:500, TIA1 1:500 dilution) were added and incubated overnight at 4 °C. After washing fluorescently labeled secondary antibodies (Proteintech, Rosemont, IL, USA,FITC 488 nm anti-rabbit fluorescence secondary antibody 1:3000, and FITC 595 nm anti-mouse fluorescence secondary antibody 1:3000) were added and incubated at room temperature for 1 h. Finally, nuclei were stained with DAPI (Thermo Fisher, Waltham, MA, USA, diluted 1:1000). Cells were observed under a fluorescence microscope (Olympus IX83, Tokyo, Japan), and fluorescence intensity was analyzed using ImageJ software.

Measurement of Fe²⁺ and ROS

Intracellular Fe²⁺ levels were measured using the specific fluorescent probe FeRhoNox-1 (10 μM; MedChemExpress, Monmouth Junction, NJ, USA). Cells were incubated with FeRhoNox-1 for 30 min, washed three times with PBS, and Fe²⁺ fluorescence signals were observed using a fluorescence microscope (Olympus IX83, Tokyo, Japan). For tissue sections, after similar preparation, Fe²⁺ fluorescence was also observed. Intracellular ROS levels were assessed using the DCFH-DA probe (10 μM; MedChemExpress, Monmouth Junction, NJ, USA). Cells were incubated with DCFH-DA for 30 min, washed three times with PBS, and ROS generation was observed under a fluorescence microscope. Quantification of fluorescence signals was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Protein extraction and western blot analysis

Total proteins from cartilage tissues and cells were extracted using RIPA lysis buffer (Thermo Fisher Scientific, Waltham, MA, USA) containing protease and phosphatase inhibitors. Samples were incubated on ice for 30 min and then centrifuged at 12,000 rpm for 10 min at 4 °C (Eppendorf 5804 R centrifuge, Hamburg, Germany). The supernatant was collected as the total protein sample. Protein concentrations were determined using a BCA protein assay kit (Shanghai Beyotime Biotechnology, Shanghai, China). After protein sample preparation, 15-25 μg of protein from each sample was loaded onto 10%-12% SDS-PAGE gels for electrophoresis, following standard protocols. The gel concentration used was adjusted according to the molecular weight of the target protein: for larger proteins (>70 kDa), a 10% gel was recommended, while a 12% gel was preferred for smaller proteins (<50 kDa) to achieve optimal resolution. After electrophoresis, proteins were transferred onto PVDF membranes (Thermo Fisher Scientific, 0.45 μm, Waltham, MA, USA) using the wet transfer method. Membranes were blocked in 5% non-fat milk (Sigma-Aldrich, St. Louis, MO, USA) for 1 h, then incubated with primary antibodies (Proteintech, Rosemont, IL, USA,G3BP1, MMP13, GPX4, SLC7A11, p-mTOR, mTOR, and β-actin were all diluted 1:1000) overnight at 4 °C. Then membranes were washed three times with PBS, and incubated with HRP-conjugated secondary antibodies (Proteintech, Rosemont, IL, USA,HRP-conjugated anti-rabbit and anti-mouse goat antibodies were all diluted 1:5000) at room temperature for 1 hour. Chemiluminescence was developed using ECL detection reagent (Shanghai Tianneng, Shanghai, China). Band signals were quantitatively analyzed using ImageJ software, with the control group serving as the standard for comparing the relative gray values of protein expression in each experimental group. β-actin was used as an internal reference.

Bioinformatics analysis

This study utilized publicly available knee joint cartilage gene expression datasets, mainly from the Gene Expression Omnibus (GEO) database. Specifically, the dataset GSE254844, based on the Illumina human genome array platform, contains whole-genome expression data of knee joint cartilage from OA patients and non-OA control groups. Data preprocessing was performed using the Limma package in R, adjusting P-values using the Benjamini & Hochberg method to screen DEGs with a criterion of adjusted P < 0.1. Volcano plots and Venn diagrams were used to visually display the DEGs. Molecular docking simulations were conducted for the selected differential gene G3BP1, using the CB-Dock2 tool to predict the binding mode between RES and G3BP1. Binding affinity was calculated using AutoDock Vina software to evaluate the potential regulatory effect of resveratrol on G3BP1 binding

Cellular thermal shift assay

To assess the effect of resveratrol on the thermal stability of G3BP1, a cellular thermal shift assay was employed. C28/I2 cells were treated with 50 μM RES or vehicle control (DMSO) for 24 h. Cell lysates were aliquoted and heated at a temperature gradient (35 °C, 40 °C, 45 °C, 50 °C, 55 °C, 60 °C, 65 °C, 70 °C) for 3 min using a thermal cycler, followed by immediate cooling on ice. Denatured proteins were pelleted by centrifugation (16,000 × g, 20 min, 4 °C), and supernatants containing thermostable proteins were analyzed by Western blot. Equal protein amounts (20 μg/lane) were separated by SDS-PAGE, transferred to PVDF membranes, and probed with anti-G3BP1 (1:1,000) and anti-β-actin (1:5,000) antibodies. Band intensities were quantified using ImageJ and normalized to the 35°C untreated control.

Statistics and reproducibility

No statistical method was used to predetermine sample size. For animal experiments, mice were randomly assigned to experimental groups using computer-generated random numbers. Animals from different groups were housed under the same SPF conditions, and treatments and measurements were performed in the same experimental period where applicable to minimize potential confounders.No animals, experimental units or data points were excluded from the analyses unless otherwise stated. Histological scoring was performed by two investigators in a double-blind manner. For animal experiments, n refers to biologically independent animals. For serum biochemical analyses, n = 12 biologically independent animals per group unless otherwise stated. For histological, immunohistochemical and fluorescence analyses of mouse joint tissues, n = 6 biologically independent animals per group unless otherwise stated. For cell culture experiments, n refers to biologically independent experiments performed on separate occasions. For animal experiments, the experimental unit was a single mouse. Cell-based assays were performed in triplicate wells and repeated three times unless otherwise stated; technical replicates were averaged within each biologically independent experiment before statistical analysis.

For microscopy-based quantification, multiple fields of view were analyzed per sample or experiment and averaged to generate one value for each biologically independent sample or experiment. Representative western blots, histological images and microscopy images were reproduced in at least three biologically independent experiments or biologically independent samples with similar results unless otherwise stated.

Data are presented as mean ± standard deviation (SD) unless otherwise indicated. Statistical analyses were performed using SPSS 21.0. Graphs were generated using GraphPad Prism 8.0. Normality and homogeneity of variance were assessed before parametric testing. If assumptions were not met, data were analyzed using appropriate non-parametric tests. Comparisons among more than two groups were performed using one-way analysis of variance (ANOVA) followed by LSD post hoc tests. Exact p values are provided in Supplementary Data 1 where possible. A p value < 0.05 was considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information ^{(9.2MB, pdf)}

42003_2026_10360_MOESM2_ESM.pdf ^{(74.7KB, pdf)}

Description of Additional Supplementary Materials

Supplementary Data 1 ^{(89.4KB, xlsx)}

Reporting Summary ^{(5.7MB, pdf)}

Author contributions

Jianyi He: Investigation, Validation, Data curation, Writing – original draft. Ze Li: Investigation, Validation, Data curation, Writing – original draft. Haocheng Zhang: Investigation. Ye Yuan: Investigation. Ruiqi Wang: Investigation. Min Liu: Data curation. Qiwei Ma: Data curation. Hailun Gu: Methodology, Investigation, Validation. Li Liu: Conceptualization, Funding acquisition, Methodology, Visualization, Writing – review & editing. All authors have read and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Aline Bozec and Joao Valente.

Funding

This study was supported by the National Natural Science Foundation of China (82073541). The funder had no role in study design, data collection, data analysis, data interpretation or manuscript preparation.

Data availability

The public dataset used in this study is available in the Gene Expression Omnibus (GEO) database under accession number GSE254844. Uncropped and unedited western blot images are provided in Supplementary Figs. 7-9. The numerical source data underlying the graphs and charts presented in the main and Supplementary Figs. are provided as Supplementary Data 1. The data supporting the remaining findings of this study are available from the corresponding author upon reasonable request.

Code availability

No custom computer code or mathematical algorithm central to the conclusions of this study was generated or used. Data analysis and visualization were performed using publicly available or commercial software, including R/limma for differential expression analysis, CB-Dock2 and AutoDock Vina for molecular docking, ImageJ for image quantification, GraphPad Prism 8.0 for graph generation and SPSS 21.0 for statistical analysis. Key analysis parameters are described in the Methods section. Therefore, no custom code has been deposited in a DOI-minting repository.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10360-3.