Semaglutide alleviates osteoarthritis independent of weight loss via GLP-1R–mediated activation of autophagy through AKT/mTOR inhibition

Abstract

Objective

The development of osteoarthritis (OA) is closely associated with systemic metabolic disorders, yet there remains a lack of disease-modifying therapeutic strategies that simultaneously target metabolic abnormalities and inflammatory responses. This study aims to systematically evaluate the therapeutic potential of semaglutide, a long-acting glucagon-like peptide-1 receptor (GLP-1R) agonist used for diabetes management, in OA and to elucidate its underlying molecular mechanisms.

Methods

We utilized a zebrafish cartilage injury repair model to screen and assess the impact of several hypoglycemic drugs on cartilage regeneration. OA was induced in C57BL/6 mice by destabilization of the medial meniscus (DMM) surgery. Using systemic Glp-1r knockout mice, we systematically evaluated the effects of semaglutide on joint structure, function, and pain-related behaviors in vivo. RNA sequencing was performed to explore the regulator effects of semaglutide on extracellular matrix metabolism, associated signaling pathways, and autophagy in IL-1β-stimulated primary mouse chondrocytes. To verify the functional loss, the GLP-1R antagonist Exendin (9-39) and the autophagy inhibitor Bafilomycin A1 were employed.

Results

Drug screening using a zebrafish cartilage injury model demonstrated that semaglutide exerted the most significant pro-regenerative effects, markedly promoting cartilage repair. In wild-type (WT) mice with DMM-induced OA, semaglutide treatment significantly improved gait abnormalities and mechanical hyperalgesia without significantly affecting body weight, and alleviated cartilage destruction, synovitis, and subchondral bone sclerosis associated with abnormal chondrocyte metabolism. However, GLP-1R inhibition or Glp-1r knockout completely abolished the protective effects of semaglutide on chondrocyte metabolism and its therapeutic efficacy in OA. Moreover, Glp-1r gene deficiency exacerbated cartilage degeneration and bone structural damage, indicating that GLP-1R signaling is indispensable for maintaining cartilage homeostasis. Mechanistically, semaglutide inhibited the AKT/mTOR pathway through GLP-1R activation, thereby reversing IL-1β- and DMM-induced autophagy suppression and restoring the balance of extracellular matrix metabolism in chondrocytes.

Conclusion

Semaglutide exerts protective effects against OA by activating GLP-1R in chondrocytes, inhibiting the AKT/mTOR pathway, and enhancing chondrocyte autophagy. It alleviates abnormal cartilage metabolism in OA independently of body weight changes.

Translational potential of this study: This study demonstrates, for the first time, that semaglutide exerts protective effects against OA independent of weight loss by directly activating chondrocyte GLP-1R, inhibiting the AKT/mTOR pathway, and enhancing autophagy. These findings provide robust preclinical evidence supporting the repositioning of semaglutide as a disease-modifying therapy, particularly for patients with OA and comorbid metabolic disorders such as diabetes. Furthermore, they indicate that GLP-1R and its downstream signaling pathways may represent potential therapeutic targets for OA, thereby opening new avenues for drug repurposing and precision interventions in metabolic OA.

Graphical abstract

1. Introduction

Osteoarthritis (OA) is a highly prevalent degenerative joint disease worldwide and a leading cause of functional impairment and disability in middle-aged and older adults, affecting over 500 million individuals globally [1,2]. Its hallmark pathological features encompass progressive articular cartilage degradation, subchondral bone sclerosis, and osteophyte formation. At the molecular level, OA is characterized by aberrant elevations in intra-articular inflammatory cytokines (e.g., IL-1β, IL-6, TNF-α) and matrix-degrading enzymes, including matrix metalloproteinases (MMPs) and a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) [3,4]. Clinically, conventional treatments such as nonsteroidal anti-inflammatory drugs (NSAIDs) and physical therapy primarily alleviate symptoms but exhibit limited efficacy in halting or reversing OA progression [5,6]. Thus, transitioning from symptomatic management to disease-modifying strategies and identifying novel therapies that target the underlying pathogenesis of OA remain critical challenges and promising avenues for future research.

In recent years, accumulating evidence has underscored the pivotal role of metabolic dysfunction in the pathogenesis and progression of OA, highlighting a strong pathophysiological association between metabolic syndrome and OA [7,8]. For instance, diabetic patients exhibit a markedly elevated risk of OA, wherein metabolic aberrations such as hyperglycemia and insulin resistance exacerbate joint degeneration via systemic and local mechanisms [9,10]. Conditions like obesity and type 2 diabetes facilitate the release of pro-inflammatory mediators (e.g., adipokines and cytokines) from adipose tissue, inducing metabolic stress that indirectly or directly accelerates cartilage matrix degradation. Building on this, established hypoglycemic agents like metformin have been shown to modulate chondrocyte inflammatory responses and maintain matrix metabolic homeostasis, indicating their repurposing potential [9,11,12]. Moreover, emerging studies have evaluated the therapeutic implications of various hypoglycemic drugs in OA, with certain targets potentially altering disease trajectories [13].

In recent years, the zebrafish model has emerged as a pivotal platform for high-throughput in vivo drug screening in skeletal research, owing to its transparent embryos, facile genetic manipulation, and highly conserved signaling pathways with mammals [14,15]. To systematically evaluate the chondrogenic repair effects of various hypoglycemic agents, we utilized a chemically induced cartilage injury repair model in zebrafish for initial screening, including semaglutide. Our results revealed that semaglutide demonstrated markedly superior chondrogenic repair capacity relative to the other tested compounds. As a long-acting glucagon-like peptide-1 receptor (GLP-1R) agonist, semaglutide is extensively employed in the clinical management of type 2 diabetes and obesity, with a well-established safety profile. The recent STEP-9 clinical trial reported that semaglutide not only effectively reduces body weight in obese patients with knee OA but also significantly mitigates joint pain and enhances physical function [16]. Nevertheless, the extent to which this protective effect stems from weight loss versus direct joint-protective mechanisms remains uncertain. Notably, GLP-1R is expressed not only in metabolic organs such as the pancreas but also in articular cartilage [17]. GLP-1R activation has been shown to ameliorate glucose and lipid metabolism while exerting pronounced anti-inflammatory effects [18,19], thereby providing a theoretical foundation for GLP-1R agonist-based interventions in OA.

In this study, utilizing a Destabilization of the Medial Meniscus (DMM)-induced mouse osteoarthritis model in conjunction with Glp-1r knockout (Glp-1r KO) animals, we demonstrated that semaglutide ameliorates key pathological features of osteoarthritis, including cartilage degeneration, subchondral bone sclerosis, and joint dysfunction, in a GLP-1R-dependent manner. Furthermore, it regulates chondrocyte metabolic homeostasis, with these improvements occurring independently of weight loss. Mechanistically, semaglutide confers OA-protective effects through GLP-1R activation, AKT/mTOR pathway inhibition, and autophagy enhancement. These findings furnish robust preclinical evidence and a theoretical framework for repurposing semaglutide as an osteoarthritis therapy and advancing targeted interventions for metabolic osteoarthritis.

2. Materials and methods

2.1. Establishment of the zebrafish cartilage injury model and drug treatment

All antidiabetic agents, including metformin (Met) (5 μM, HY-B0627), empagliflozin (Emp) (1 μM, HY-15409), rosiglitazone (Ros) (1 μM, HY-17386), gliclazide (Gli) (5 μM, HY-B0753), and semaglutide (Se) (0.5 μM, HY-114118), were purchased from MedChemExpress (MCE, USA). The transgenic zebrafish line Tg(col2a1a:dendra2-NTR) was generated by embryonic microinjection of plasmids (Beijing Tsingke Biotech, China) into zebrafish embryos (Fig. 1A). This strain specifically expresses a fusion protein of Dendra2 green fluorescent protein and nitroreductase (NTR) in chondrocytes. At 3.5 days post-fertilization (dpf), zebrafish were treated with 12 mM metronidazole (MTZ) for 20 h. After regeneration for 3 days using different hypoglycemic drugs, including Met (5 μM), Emp (1 μM), Ros (1 μM), Gli (5 μM), and Se (0.5 μM), cartilage repair was assessed via confocal microscopy (n = 10) (LSM 880 NLO, Carl Zeiss, Germany) (Fig. 1B). Each group contained 10 zebrafish larvae, and 5 larvae per group were randomly selected for confocal imaging and quantitative analysis. Two control groups were included: an uninjured control group without MTZ exposure and a vehicle control group in which MTZ-injured zebrafish received DMSO alone. For Fig. 1D, Z-stack images were acquired at 1 μm intervals across 15 optical sections, and the mean fluorescence intensity was calculated from the full Z-stack of each larva for statistical comparison.

Fig. 1.

Establishment of zebrafish cartilage injury and mouse OA models and therapeutic evaluation of antidiabetic drugs, including semaglutide.

A. The diagrams show the Tg(col2a1a:dendra2-NTR) transgenic line and its labeling of mandibular cartilages, including Meckel's cartilage (M), palatoquadrate cartilage (Pq), and ceratohyal cartilage (Ch). B. The flowchart illustrates the process of cartilage injury and drug treatment in zebrafish: at 3.5 days post-fertilization (dpf), zebrafish larvae were exposed to 12 mM metronidazole (MTZ) for 20 h, followed by drug intervention, with phenotypic analysis performed 3 days later. C. Representative gross morphology of zebrafish in the MTZ/NTR-induced cartilage injury repair model. D. Confocal microscopy analysis showing the repair effects of different antidiabetic drugs on zebrafish cartilage injury (n = 10). The antidiabetic agents included metformin (Met), empagliflozin (Emp), rosiglitazone (Ros), gliclazide (Gli), and semaglutide (Se). Working concentrations were selected based on published doses and further refined in pilot assays to identify the highest effective concentration that did not cause overt toxicity or teratogenic effects. E. Experimental design of Se administration in wild-type (WT) mice with DMM-induced osteoarthritis (OA). F. Body weight changes in WT mice during treatment with low- or high-dose Se following destabilization of the medial meniscus (DMM) surgery (n = 6). G. Mechanical allodynia in each group of WT mice was assessed using the von Frey test during Se treatment after DMM surgery (n = 6). H. Effects of Se on pain-related behavior in OA mice, along with representative gait footprints from each group. I. Quantitative gait analysis based on H-track parameters, expressed as the ratio of right hindlimb to left hindlimb (RH/LH) (n = 6). J. Representative micro-computed tomography (micro-CT) images of knee joints from each group, including sagittal, coronal, and transverse reconstructions, illustrate overall subchondral bone alterations following Se treatment. K. Quantitative analysis of subchondral bone structural parameters derived from micro-CT images in (J), including subchondral bone plate (SBP) thickness, bone volume fraction (BV/TV), and trabecular thickness (Tb.Th) (n = 6). Scale bar represents 100 μm. Data are presented as the means ± SD. ns. Non-significant. ∗P value < 0.05. ∗∗P value < 0.01. ∗∗∗P value < 0.001.

2.2. Mouse osteoarthritis model and grouping

Wild-type (WT) C57BL/6 mice were purchased from Beijing Baishitong Biotechnology Co., Ltd. (Beijing, China), and the Glp-1r KO mouse strain was obtained from Shanghai Model Organisms Center, Inc. (Shanghai, China). All mice were maintained under specific pathogen-free conditions with a 12-h light/dark cycle and had free access to food and water. After arrival, the mice were acclimatized for 7 days before the experiments. All animal procedures were approved by the Ethics Committee and were conducted in strict accordance with the Guidelines for Ethical Review of Animal Welfare (GB/T 35892-2018) and the ARRIVE reporting guidelines.

Ten-week-old male mice were selected, and osteoarthritis was induced via DMM surgery on the right knee joint [20]. The brief procedure is as follows: Mice were anesthetized with inhaled isoflurane, a longitudinal incision was made on the right knee joint, and the tibial ligament of the medial meniscus was transected under a surgical microscope to establish the DMM model. The sham group underwent only capsular opening without ligament injury. Two weeks post-surgery, mice were randomly assigned to receive subcutaneous injections of semaglutide (15 μg/kg for the low-dose group and 60 μg/kg for the high-dose group) or an equal volume of PBS twice weekly for 8 weeks. At the endpoint, mice were deeply anesthetized with 5% isoflurane and euthanized by cervical dislocation, ensuring death. Every effort was made throughout the study to minimize animal pain and discomfort. Knee joints and visceral organs were harvested for subsequent analysis.

2.3. Behavioral assessments of pain and motor function

Mechanical Pain Threshold Testing: Mechanical pain hypersensitivity in mice was assessed using calibrated Von Frey filaments. After 30 min of acclimation in a test cage with a metal mesh floor, the plantar surface of the right hind paw was stimulated with a series of filaments of varying intensities (North Coast Medical Inc., CA, USA, EXACTA, NC12775-99). The 50% mechanical paw-withdrawal threshold was determined using the Dixon method [21], with a positive response defined as rapid paw withdrawal, paw licking, or paw shaking.

Gait Analysis: Motor function was assessed using the VisuGait gait analysis system (Shanghai Xinruan Company). Mice were placed on a transparent glass treadmill, and high-speed video recordings of their free-walking gait were captured. The following parameters were analyzed via software: right hind paw footprint area, swing phase duration, swing velocity, and duty cycle (ratio of stance phase duration to total gait cycle duration) [22]. To minimize individual variability, parameter ratios between the right and left hind limbs were calculated for analysis [23].

2.4. Radiological and histological assessment

Radiological Analysis: After fixation of mouse knee joints in 4% paraformaldehyde (Solarbio Life Sciences, P1110), specimens were scanned using a small-animal X-ray imaging system (FAXITRON MX-20, USA) and a μ-CT scanner (vivaCT40, Scanco Medical AG, Switzerland). Images were reconstructed in three dimensions with Scanco Medical software. The region of interest (ROI) was defined as the entire subchondral bone area of the medial tibial plateau. Subchondral bone plate (SBP) thickness was measured within this ROI and reported as the mean value across the analyzed area. Bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), and trabecular number (Tb.N) were also quantified using Scanco Medical software [24,25].

Histological and Immunohistochemical Analysis: Knee joint samples underwent decalcification with ethylenediaminetetraacetic acid (EDTA, Solarbio Life Sciences, E1171), dehydration with graded ethanol, and paraffin embedding. Continuous sagittal sections 4μm thick were prepared. Sections were stained with Hematoxylin and Eosin (H&E) staining and Safranin O/Fast Green (S-F) staining, respectively (Solarbio Life Sciences, G1120, G1371). Cartilage damage severity was assessed using the OARSI scoring system (0-6 points) [26], while synovitis was evaluated using a semiquantitative histopathological scoring system in which three synovial features were each scored on a 0-3 scale and summed to yield a total synovitis score ranging from 0 to 9 [27].

Immunohistochemical (IHC) and immunofluorescence (IF) staining followed standard protocols [28]. Sections underwent dewaxing, rehydration, and overnight antigen retrieval in citrate/EDTA buffer. After 1 h of blocking with immunostaining blocking solution, sections were incubated overnight at 4°C with specific primary antibodies. Antibodies and dilutions were as follows: GLP-1R (1:200; 26196-1-AP,Proteintech), MMP13 (1:200; ab39012, Abcam), ADAMTS5 (1:100; ab182795, Abcam), COL2A1 (1:200; ab34712, Abcam), Aggrecan (1:500; 13880-1-AP, Proteintech), Phospho-AKT (Ser473) (1:200; 66444-1-Ig,Proteintech), AKT (phosphoT308) (1:200; ab38449, Abcam), and Phospho-mTOR (Ser2448) (1:500; 67778-1-Ig,Proteintech). The following day, sections were incubated with the corresponding HRP-labeled secondary antibody (1:5000; Proteintech) at room temperature for 1 h. Finally, images were captured and analyzed using an optical microscope.

2.5. Human cartilage explant culture and primary chondrocyte isolation

Human knee articular cartilage specimens were obtained from patients with OA undergoing total knee arthroplasty. Immediately after excision, the specimens were placed in prechilled DMEM and transported on ice. Macroscopically intact cartilage from the non-lesional areas of OA knees was selected for explant culture and cut into small pieces measuring 4 mm × 3 mm × 2 mm. The explants were cultured in F-12 medium supplemented with 10% fetal bovine serum (FBS) and treated with or without 10 ng/mL IL-1β in combination with Se. After 7 days of culture, the samples were collected for histological analysis. Primary chondrocytes were isolated and cultured according to standard procedures. Mouse knee articular cartilage was harvested from 7-day-old C57BL/6 mice [29]. Briefly, cartilage tissue blocks were digested in DMEM/F-12 (Gibco,11320033) medium containing type II collagenase. Isolated cells were resuspended in complete DMEM/F-12 medium supplemented with 10% fetal bovine serum (FBS, Gibco, 10099141C) and 1% penicillin/streptomycin, and cultured at 37°C under 5% CO₂ conditions with daily medium changes. All primary chondrocyte experiments were performed with three independent biological replicates, and each biological replicate was analyzed in triplicate technical replicates.

Cell Treatment and Viability Assay: For experimental manipulation, chondrocytes were stimulated with recombinant human IL-1β to mimic an inflammatory environment. Se, the GLP-1R antagonist Exendin (9-39) amide (MCE, HY-P0264; used to block GLP-1R signaling for receptor-dependence validation), or the autophagy inhibitor Bafilomycin A1 (MCE, HY-100558; used to inhibit autophagosome-lysosome fusion for autophagic flux assessment) were added to the medium alone or in combination for 48 h.

Cell viability was assessed using the CCK-8 assay (MCE, HY-K0301). Primary chondrocytes were seeded at 1 x10⁴ cells/well in a 96-well plate and incubated with different concentrations of semaglutide for 48 h. Subsequently, the medium was removed, cells were washed twice with PBS, and medium containing CCK-8 reagent was added for a further 1-h incubation. Absorbance at 450 nm was measured using a microplate reader to calculate relative cell viability.

2.6. Western blotting analysis

Cells were washed with pre-chilled PBS and subsequently lysed on ice in RIPA lysis buffer containing 1% benzenesulfonyl fluoride. After determining protein concentration via the BCA method, samples were mixed with SDS-PAGE loading buffer and denatured by boiling. Equivalent amounts of protein were separated by SDS-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane (Bio-Rad, USA). The membrane was blocked with TBST containing 5% bovine serum albumin, then incubated overnight at 4°C with the corresponding primary antibodies. Primary antibodies used included: GAPDH (1:5000; 60004-1-Ig, Proteintech), MMP13 (1:3000; ab39012, Abcam), ADAMTS5 (1:1000; ab41037, Abcam for Figs. 3B and 6A; and 1:2000; HA722011, HUABIO for Fig. 3K), COL2A1 (1:1000; ab34712, Abcam), GLP-1R (1:1000; 26196-1-AP, Proteintech), LC3B (1:5000; 14600-1-AP, Proteintech), Phospho-AKT (Ser473) (1:5000; 66444-1-Ig, Proteintech), AKT (1:5000; 10176-2-AP, Proteintech), Phospho-AKT (T308) (1:1000; ab38449, Abcam), mTOR (1:5000; 66888-1-Ig, Proteintech), Phospho-mTOR (Ser2448) (1:5000; 67778-1-Ig, Proteintech), P62 (1:10000; ab109012, Abcam), and ATG5 (1:1000; 10181-2-AP, Proteintech). Subsequently, membranes were incubated with the corresponding HRP-labeled secondary antibody (1:5000; SA00001-1, SA00001-2, Proteintech) at room temperature for 1 h. Finally, the membranes were visualized using an enhanced chemiluminescence system, and band intensities were quantified using ImageJ software.

Fig. 3.

Semaglutide regulates chondrocyte activity through GLP-1R signaling in vitro and in vivo.

A. Effects of semaglutide (Se) treatment on the viability of primary mouse chondrocytes following 48-h exposure (mean ± SEM, n = 3). B-C. Western blot (WB) analysis of GLP-1R, MMP13, ADAMTS5, and COL2A1 expression levels in primary mouse chondrocytes treated with 0.3 μM Se for 48 h. D. Immunohistochemical (IHC) analysis of GLP-1R expression in articular cartilage of WT DMM mice treated with Se for 8 weeks. E. Quantification of GLP-1R-positive cells in articular cartilage based on IHC results in (D) (n = 6). F-G. Human cartilage explants treated with IL-1β and semaglutide. Representative images of H&E staining, Safranin O staining, and immunohistochemical staining for COL2A1 and MMP13, with quantification (n = 6). H-K. WB analysis of MMP13, ADAMTS5, and COL2A1 expression in primary mouse chondrocytes treated with the GLP-1R antagonist Exendin (9-39) (En9-39). L. Quantitative real-time PCR (qRT-PCR) analysis of Mmp13, Adamts5, Col2a1, and Acan mRNA expression in primary mouse chondrocytes treated with En9-39 (n = 3). Scale bar: 50 μm in D. Data are presented as the means ± SD. ns. None significant. ∗P value < 0.05. ∗∗P value < 0.01. ∗∗∗P value < 0.001.

Fig. 6.

Semaglutide may regulate chondrocyte matrix metabolic balance by improving autophagy in OA chondrocytes

A-B. Western blot analysis of MMP13, ADAMTS5, COL2A1, P62, ATG5, and the LC3-II/LC3-I ratio in IL-1β-stimulated chondrocytes treated with semaglutide and Bafilomycin A1(Baf-A1). C. Immunofluorescence (IF) analysis of LC3B puncta in IL-1β-stimulated chondrocytes treated with semaglutide and Baf-A1. Scale bar, 50 μm. D. Representative quantitative analysis of (C) performed using FIJI/ImageJ software. 5 to 10 fields of view were taken, and three replicate experiments were performed. E-F. IF analysis of LC3B and P62 in articular cartilage of wild-type (WT) mice after 8 weeks of semaglutide treatment following DMM surgery. Scale bar, 50 μm. (n = 6). Data are presented as the means ± SD. ns. Non-significant. ∗P value < 0.05. ∗∗P value < 0.01. ∗∗∗P value < 0.001.

2.7. RNA extraction, real-time quantitative PCR, and RNA sequencing

Total cellular RNA was extracted using TRIzol reagent (Thermofisher, 15596018). RNA concentration and purity were assessed via a NanoDrop spectrophotometer. RNA was subsequently reverse transcribed into cDNA (MCE, HY-K0511A), and qRT-PCR reactions were performed using SYBR Green Q-PCR kits on a Step One Plus real-time PCR system (Bio-Rad CFX, ver.5.0.021.0616, Singapore). Relative mRNA expression levels were calculated using the 2^−ΔΔCT method, with GAPDH used as the internal reference gene. Each sample included three replicate wells, with data expressed as mean ± standard deviation. Primer sequences are listed in Appendix Table 1.

For RNA sequencing, primary chondrocytes treated with IL-1β alone or IL-1β plus semaglutide for 48 h were collected separately. Total RNA was extracted using TRIzol reagent, and cDNA libraries were prepared using the NEBNext Ultra II RNA Library Prep Kit for Illumina (NEB, USA) according to the manufacturer's instructions. After RNA quality assessment, the libraries were sequenced on an Illumina NovaSeq 6000 platform by Shanghai Majorbio Bio-pharm Biotechnology Co., Ltd. (Shanghai, China). Raw sequencing reads were subjected to quality control to remove adapter sequences and low-quality reads, and clean reads were retained for downstream analysis. The resulting clean reads were aligned to the mouse reference genome (GRCm39, Ensembl release 109) using HISAT2 software, and gene-level quantification was performed using featureCounts based on the corresponding annotation file. Differentially expressed genes between groups were analyzed using edgeR software [30], and enrichment analysis of differentially expressed genes was performed using the online platform https://www.bioinformatics.com.cn [31].

2.8. Statistical analysis

All data were statistically analyzed using SPSS 22.0 and GraphPad Prism 8.0 software. Quantitative data are expressed as mean ± standard deviation from at least three independent experiments. Comparisons between groups were performed using two-tailed Student's t-tests, one-way analysis of variance (ANOVA) with LSD post hoc multiple comparisons, or Mann–Whitney U tests, depending on the data distribution. Differences were considered statistically significant at P < 0.05.

3. Results

3.1. Semaglutide improves motor function and alleviates subchondral bone sclerosis in DMM model mice

To systematically analyze the effects of various hypoglycemic drugs on cartilage repair, we employed a zebrafish cartilage injury repair model for drug treatment evaluation. We first established a Tg(col2a1a:dendra2-NTR) transgenic zebrafish line that expresses a fusion protein of fluorescent protein Dendra2 and nitroreductase (NTR) driven by the chondrocyte-specific promoter col2a1a (Fig. 1A). Using the metronidazole (MTZ)/NTR system [32], cartilage-specific injury could be induced by treatment with 12 mmol/L MTZ for 20 h in Tg(col2a1a:dendra2-NTR) zebrafish. Subsequently, various antidiabetic drugs were administered for screening (Fig. 1B). The results showed that semaglutide demonstrated the most significant cartilage repair-promoting effect among all tested drugs at 3 days post-injury (Fig. 1C and D, Fig. S1E).

To investigate the effect of semaglutide on the development of OA in mice, we further treated normal-weight WT mice in the DMM model with different doses of semaglutide (Fig. 1E). We found that mice in the high-dose group exhibited a gradual decrease in body weight, with the most pronounced reduction observed at 8 weeks post-injection, whereas the low-dose group showed no significant weight loss (Fig. 1F).

Using CatWalk gait analysis to assess behavioral changes revealed that semaglutide treatment significantly increased paw contact area on the right hindlimb, elevated swing velocity, and improved the stance-to-swing ratio (Fig. 1H and I) (Fig. S1A and B) compared to the untreated group. Von Frey filaments test results demonstrated that semaglutide treatment significantly elevated the mechanical foot-withdrawal threshold, effectively alleviating DMM surgery-induced mechanical hyperalgesia (Fig. 1G). X-ray and micro-CT analyses revealed marked subchondral bone sclerosis in the DMM group, whereas semaglutide intervention significantly reduced subchondral bone plate thickness and showed improvement trends in bone volume fraction and trabecular thickness (Fig. 1J and K) (Fig. S1C). These findings suggest that both low- and high-dose semaglutide partially improve gait abnormalities, reduce pain sensitivity, and mitigate pathological subsurface bone sclerosis in OA mice. Although the high-dose group exhibited weight loss, the low-dose group demonstrated symptomatic improvement without significant body weight alteration.

3.2. Semaglutide alleviates abnormal OA cartilage metabolism and improves cartilage degeneration phenotype independently of body weight changes

We further assessed the therapeutic efficacy of semaglutide in OA through histopathological evaluation. Hematoxylin and Eosin (H&E) staining (Fig. 2A) and Safranin O/Fast Green (S-F) staining (Fig. 2C) revealed significant loss of cartilage matrix, thinning of the cartilage layer, and marked synovial inflammation in 8-week DMM mice. In contrast, semaglutide treatment better preserved cartilage structural integrity and enhanced matrix staining intensity. Semi-quantitative analysis based on the OARSI scoring system indicated that both low- and high-dose semaglutide treatment groups exhibited significantly lower cartilage damage scores and synovitis scores compared to the DMM model group, confirming the drug's efficacy in delaying the progression of cartilage degeneration (Fig. 2B and D).

Fig. 2.

Histological and immunohistochemical evaluation of articular cartilage in WT mice treated with semaglutide after DMM surgery.

A. Hematoxylin-eosin (H&E) staining results of articular cartilage in wild-type (WT) DMM mice treated with low- and/or high-dose semaglutide (Se) for 8 weeks. B. Summed synovitis scores based on H&E staining (n = 6). C. Safranin O/Fast Green (S-F) staining results of articular cartilage in WT DMM mice treated with low- and/or high-dose Se for 8 weeks. D. Summed Osteoarthritis Research Society International (OARSI) scores based on S-F staining in (C) (n = 6). E. Immunohistochemical (IHC) analysis of the expression of the COL2A1, MMP13, ADAMTS5, and Aggrecan in articular cartilage of WT DMM mice treated with Se for 8 weeks. F-I. Quantification of the percentage of positive cells for COL2A1, MMP13, ADAMTS5, and Aggrecan in the articular cartilage based on the IHC results in (E) (n = 6). Scale bar represents 50 μm in A, C, E. Data are presented as the means ± SD. ∗P value < 0.05. ∗∗P value < 0.01. ∗∗∗P value < 0.001.

IHC analysis further revealed that after low- and high-dose semaglutide treatment (Fig. 2E), the expression of matrix-degrading enzymes MMP13 (Fig. 2G) and ADAMTS5 (Fig. 2H) was significantly reduced compared to the untreated group, while the expression of key cartilage matrix components COL2A1 (Fig. 2F) and Aggrecan (ACAN) (Fig. 2I) proteoglycans was markedly upregulated. These findings indicate that semaglutide mitigates DMM-induced degradation of the cartilage matrix and exerts a protective effect on articular cartilage matrix synthesis. Moreover, this therapeutic efficacy was observed even in the low-dose group without significant body weight changes, suggesting that semaglutide can modulate cartilage metabolism and alleviate the cartilage degeneration phenotype of OA independently of body weight alterations.

3.3. Semaglutide may directly protect chondrocytes via GLP-1R-dependent pathways

To investigate whether semaglutide exerts direct regulatory effects on chondrocytes, we conducted in vitro cell experiments. CCK-8 assays indicated that semaglutide concentrations below 0.6 μM did not exhibit significant toxicity toward primary mouse chondrocytes; thus, a 0.3 μM concentration was selected for subsequent experiments (Fig. 3A). Western blot analysis revealed that IL-1β stimulation significantly upregulated MMP13 and ADAMTS5 expression while downregulating COL2A1 expression in chondrocytes. Co-treatment with semaglutide significantly reversed the IL-1β-induced imbalance in chondrocyte matrix metabolism (Fig. 3B and C). Notably, IL-1β itself suppressed GLP-1R expression in chondrocytes, while semaglutide intervention partially restored its expression levels. Consistent with this, IHC analysis of mouse articular cartilage confirmed decreased GLP-1R expression in DMM mouse joints, which was significantly increased by semaglutide treatment (Fig. 3D and E). To simulate the inflammatory microenvironment of OA chondrocytes, cartilage explants were collected from patients undergoing total knee arthroplasty and treated with 10 ng/mL IL-1β. At the histological level, H&E staining and Safranin O staining showed that IL-1β accelerated proteoglycan loss in the cartilage extracellular matrix, whereas semaglutide attenuated this effect. Further IHC analysis revealed that, compared with the untreated group, semaglutide treatment significantly increased the expression of COL2A1, a major component of the cartilage matrix, while significantly decreasing the expression of MMP13, a matrix-degrading enzyme (Fig. 3F and G).

Further evidence was provided by the antagonist experiment. When treated with the GLP-1R-specific antagonist Exendin (9-39), semaglutide's regulatory effects on chondrocyte matrix metabolism were completely blocked (Fig. 3H–K). Further RT-qPCR analysis revealed that semaglutide treatment improved IL-1β-induced decreases in chondrocyte anabolic markers Col2a1 and Acan while suppressing increases in catabolic markers Mmp13 and Adamts5. Conversely, Exendin (9-39) treatment markedly reversed semaglutide's protective effects on chondrocyte metabolism (Fig. 3L). Collectively, these findings indicate that semaglutide directly regulates chondrocyte anabolic and catabolic processes, relying on activation of the GLP-1R signaling pathway.

3.4. Glp-1r gene deletion exacerbates OA phenotype and abolishes semaglutide's protective effect against OA

To determine whether the ameliorative effect of semaglutide on OA progression is mediated through GLP-1R, we generated Glp-1r knockout (KO) mice and induced OA via DMM surgery followed by drug intervention in these mice (Fig. 4A and B, Fig. S4A). H&E staining analysis of major organs (heart, liver, spleen, and kidney) revealed no significant pathological alterations in either semaglutide-treated Glp-1r KO or WT mice (Fig. S3A and B), indicating good biosafety at the experimental dose. Notably, in Glp-1r KO mice, low-dose semaglutide treatment did not induce significant weight loss, with its weight-reducing effect being completely suppressed (Fig. 4C).

Fig. 4.

Glp-1r gene deletion exacerbates OA phenotype and abolishes semaglutide's protective effect against OA

A. Construction of Glp-1r knockout (KO) mice. B. Experimental design of semaglutide (Se) administration in Glp-1r KO mice with OA. C. Body weight changes in Glp-1r KO and WT mice following DMM surgery and low-dose Se treatment (n = 5). D. Representative micro-CT images of knee joints from Glp-1r KO and WT mice after DMM surgery and low-dose Se treatment, including sagittal, coronal, and transverse reconstructions, illustrating overall alterations in articular cartilage. E. Quantitative analysis of subchondral bone remodeling parameters based on the micro-CT results in (D), including subchondral bone plate (SBP) thickness, bone volume fraction (BV/TV), and trabecular thickness (Tb.Th) (n = 5). F-G. Safranin O/fast green (S-F) staining of articular cartilage and summed OARSI scores in Glp-1r KO and WT DMM mice following 8 weeks of low-dose Se treatment. H-I. Immunohistochemical (IHC) analysis of COL2A1 and MMP13 expression in the cartilage matrix of DMM mice after 8 weeks of low-dose Se treatment. All quantitative analyses were performed using the articular cartilage of the medial tibial plateau as the uniform region of interest (ROI). The percentage of positive cells was calculated as the number of positive cells divided by the total number of chondrocytes. Imaging and scoring were conducted in accordance with the same criteria for all groups. Scale bar represents 50 μm in F and H. Data are presented as the means ± SD. ns. Non-significant. ∗P value < 0.05. ∗∗P value < 0.01. ∗∗∗P value < 0.001.

X-ray (Fig. S1D) and micro-CT (Fig. 4D) analyses revealed that Glp-1r Ko mice exhibited more severe post-surgical joint structural damage compared to WT mice of equivalent weight that underwent DMM surgery. This included more pronounced subchondral bone sclerosis and further increased subchondral bone plate thickness (Fig. 4E). Interestingly, semaglutide treatment failed to produce any beneficial structural improvements in Glp-1r KO mouse joints. Their radiographic findings showed no significant difference compared to PBS-treated Glp-1r KO controls. Histological analysis confirmed these findings. H&E staining (Fig. S4B and C) and S-F staining (Fig. 4F) revealed markedly reduced staining intensity in the cartilage matrix of Glp-1r KO mice following DMM surgery, accompanied by more pronounced cartilage layer thinning. OARSI scores indicated significantly higher cartilage damage scores in Glp-1r KO mice compared to the WT DMM group (Fig. 4G). Treatment with semaglutide similarly failed to reverse this accelerated cartilage degeneration in Glp-1r KO mice. IHC further revealed that DMM surgery induced significantly increased expression of the cartilage matrix-degrading enzymes MMP13 and ADAMTS5 in Glp-1r KO mice compared to the WT group, while expression of the key cartilage matrix component COL2A1 was markedly decreased (Fig. 4H and I) (Fig. S4D and E).

The above results indicate that GLP-1R plays an important role in regulating the occurrence of OA. Glp-1r gene deletion not only exacerbates the severity of OA disease but also completely inhibits the therapeutic effect of semaglutide on OA. This suggests that the ameliorative effect of semaglutide on OA is dependent on the activation of the GLP-1R signaling pathway, rather than entirely on the systemic weight loss it induces.

3.5. Semaglutide exerts chondroprotective effects by inhibiting the AKT/mTOR signaling pathway

To further investigate the mechanisms underlying semaglutide's regulation of OA, we performed RNA sequencing analysis on chondrocytes stimulated with IL-1β and those treated with IL-1β in combination with semaglutide (Fig. 5A). A Venn diagram revealed 13,035 genes co-expressed between the two groups, among which 1137 differentially expressed genes were identified (498 upregulated, 639 downregulated) (Fig. 5B) (Fig. S2A). Cluster analysis suggested that some differentially expressed genes may participate in common signaling pathways (Fig. 5C). Gene Ontology (GO) enrichment analysis further revealed the biological functional characteristics of differentially expressed genes. These genes showed significant enrichment in extracellular matrix (ECM) and collagen synthesis across cellular components, molecular functions, and biological processes. This finding suggests that semaglutide may exert its chondroprotective effects by regulating chondrocyte metabolism through modulation of ECM modulators/degrading enzymes (Fig. S2B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed significant enrichment of differentially expressed genes in the PI3K-AKT signaling pathway (Fig. 5D). Further in vitro validation revealed that IL-1β stimulation significantly increased phosphorylation levels of p-AKT (Ser473/Thr308) and p-mTOR (Ser2448) in chondrocytes, whereas semaglutide treatment markedly suppressed phosphorylation at these critical sites. Moreover, the GLP-1R antagonist Exendin (9-39) blocked semaglutide's inhibition of the AKT/mTOR pathway (Fig. 5E and F). IHC similarly revealed significantly lower p-AKT levels in articular cartilage tissue from semaglutide-treated mice compared to DMM controls, with a decreasing trend in p-mTOR expression (Fig. 5G and H). Collectively, these data suggest that semaglutide may exert its chondroprotective effects through GLP-1R-mediated inhibition of the AKT/mTOR signaling pathway.

Fig. 5.

Semaglutide exerts chondroprotective effects by inhibiting the AKT/mTOR signaling pathway

A. Schematic illustration of the RNA sequencing (RNA-seq) analysis of IL-1β-stimulated chondrocytes treated with semaglutide. (n = 3). B. Venn diagram illustrating the strategy for screening DEGs. C. Hierarchical clustering analysis of DEGs identified by RNA-seq. D. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of differentially regulated transcriptional targets. E-F. Western blot (WB) analysis of phosphorylated AKT (p-AKT, Ser473 and Thr308) and phosphorylated mTOR (p-mTOR) protein expression in IL-1β-stimulated chondrocytes treated with semaglutide and/or the GLP-1R antagonist Exendin (9-39) (En9-39). G. Immunohistochemical (IHC) analysis of p-AKT (Ser473), p-AKT (Thr308), and p-mTOR in knee joint tissues of DMM mice after 8 weeks of Se treatment. H. Percentage of p-AKT(Ser473), p-AKT(Thr308), and p-mTOR positive cells (n = 6). Scale bar, 50 μm. Data are presented as the means ± SD. ns. Non-significant. ∗P value < 0.05. ∗∗P value < 0.01. ∗∗∗P value < 0.001.

3.6. Semaglutide may regulate chondrocyte matrix metabolic balance by improving autophagy in OA chondrocytes

Given that the AKT/mTOR pathway is a key negative regulator of autophagy, and autophagy plays an important role in maintaining extracellular matrix homeostasis in chondrocytes [33], we further investigated whether autophagy participates in the protective mechanism of semaglutide. Western blot analysis revealed that IL-1β stimulation impaired autophagy flux, as evidenced by accumulation of the autophagy substrate p62, decreased LC3-II/LC3-I ratio, and reduced expression of the autophagy-related protein ATG5. These alterations were accompanied by downregulation of COL2A1 and upregulation of MMP13 and ADAMTS5. Semaglutide treatment significantly reversed these effects, indicating restoration of chondrocyte autophagy activity. To confirm the causal role of autophagy, we employed the autophagy inhibitor Bafilomycin A1. Results showed that Bafilomycin A1 effectively blocked the semaglutide-induced increase in the LC3-II/LC3-I ratio and p62 degradation (Fig. 6A and B). Direct observation of autophagic spot formation via immunofluorescence staining revealed that semaglutide increased the number of LC3-positive spots in IL-1β-stimulated chondrocytes, an effect similarly inhibited by Bafilomycin A1 (Fig. 6C and D). This demonstrates that semaglutide indeed promotes autophagy levels in chondrocytes.

To validate semaglutide's regulation of chondrocyte autophagy in vivo, we performed IF staining on knee articular cartilage tissue from mouse DMM models (Fig. 6E). Compared to the sham-operated group, the DMM model group exhibited reduced LC3B fluorescence intensity in chondrocytes, suggesting suppressed autophagy activity. Semaglutide treatment, however, significantly enhanced LC3B signaling in chondrocytes, indicating activated autophagy. Correspondingly, p62 exhibited marked aggregation in chondrocytes of the DMM model group, whereas semaglutide intervention effectively reduced p62 fluorescence intensity, suggesting restored autophagic flux in chondrocytes (Fig. 6F).

The results above indicate that semaglutide partially reverses the inhibition of chondrocyte autophagy induced by IL-1β and DMM surgery, thereby improving the metabolic balance of the chondrocyte matrix and exerting its chondroprotective efficacy.

4. Discussion

This study systematically elucidated the therapeutic potential of the antidiabetic drug semaglutide in treating OA and its distinct mechanism of action through an integrated approach combining in vivo zebrafish screening, mouse disease models, and cellular and molecular mechanistic analyses. Our study not only demonstrates that semaglutide effectively mitigates joint structural damage, improves functional status, and reduces pain behavior during OA progression, but more importantly, we discovered that low-dose semaglutide-without significantly altering body weight-also exerts therapeutic effects against OA. Knocking out Glp-1r suppressed semaglutide's therapeutic effects, indicating that semaglutide can regulate cartilage metabolism and alleviate OA cartilage degeneration phenotypes by directly activating chondrocyte GLP-1R, independent of body weight changes. At the molecular level, we found that semaglutide inhibits the AKT/mTOR signaling pathway in chondrocytes via GLP-1R, thereby releasing the negative regulation of this pathway on cellular autophagy. Ultimately, enhancing autophagy helps chondrocytes maintain homeostasis under inflammatory stress (Fig. 7).

Fig. 7.

The schematic diagram illustrates the therapeutic effects and underlying mechanisms of semaglutide in the alleviation of osteoarthritis.

Under pathological conditions (left), IL-1β or DMM surgery induces downregulation of GLP-1R expression, leading to hyperactivation of the AKT/mTOR signaling pathway, which subsequently suppresses the cytoprotective autophagy process. This ultimately results in extracellular matrix (ECM) degradation (upregulated expression of MMP13 and ADAMTS5) and reduced synthesis (downregulated expression of COL2A1 and Aggrecan), thereby accelerating cartilage destruction. Following semaglutide treatment (right), it binds to GLP-1R on the surface of chondrocytes, inhibiting the AKT/mTOR signaling pathway and alleviating the suppression of autophagy. Enhanced autophagy activity contributes to restoring metabolic homeostasis in chondrocytes, promoting ECM synthesis while reducing its degradation, thus exerting a chondroprotective effect.

In recent years, osteoarthritis has been redefined as a disease closely linked to systemic metabolic status [7,34]. Hyperglycemia interacts with OA at both local and systemic levels, collectively forming a complex pathological network [35]. This understanding has sparked renewed interest in repurposing existing hypoglycemic drugs. A recent large-scale Mendelian randomization study revealed a significant genetic association between the GLP-1R target of semaglutide and reduced knee osteoarthritis risk [13], providing crucial human genetic evidence for this research. Although GLP-1R has been extensively studied in the pancreas and central nervous system [19], its function in articular cartilage has only recently gained attention. Yang et al. elucidated a functional “gut-joint axis” pathway, dominated by the gut microbiota metabolite glycocholic acid (GUDCA). This pathway antagonizes the intestinal farnesoid X receptor (FXR), thereby promoting glucagon-like peptide-1 (GLP-1) secretion and activating GLP-1R, ultimately exerting osteoarthritis protective effects [36]. This discovery reveals a novel protective mechanism against osteoarthritis initiated by gut microbiota and mediated through systemic circulation. Concurrently, Chen et al. reported an alternative mode of action for GLP-1-related drugs: in animal models of obesity-associated diabetes, the dual GIP/GLP-1R agonist Tirzepatide induced gut dysbiosis and altered metabolite profiles, ultimately leading to adverse bone mass reduction [37]. These studies highlight the complexity of GLP-1R regulation in bone and joint homeostasis, where its final effects are modulated by systemic metabolic status and the intestinal microenvironment.

We first conducted a screening using transgenic zebrafish models and found that semaglutide demonstrated significant cartilage repair and regenerative capacity among all tested hypoglycemic drugs, providing preliminary evidence for subsequent in-depth research. As a long-acting GLP-1R agonist, semaglutide has shown significant weight loss effects in the treatment of type 2 diabetes and obesity [38,39]. The recent STEP-9 clinical trial demonstrated that semaglutide mediated approximately 13.7% weight loss in obese patients with knee OA, while significantly improving joint pain and function; the study primarily attributed the observed joint benefits to weight reduction [16]. Our study found that low-dose semaglutide had no significant effect on body weight, whereas high-dose semaglutide markedly reduced mouse body weight. This result aligns with the dose-dependent weight loss effect observed in clinical studies. In the STEP UP clinical trial, semaglutide 7.2 mg produced greater weight loss than 2.4 mg in obese adults [40]. This indicates that the weight-reducing effect of semaglutide is indeed dose-dependent. Lower doses may be insufficient to activate sufficient metabolic responses or central appetite suppression pathways, whereas higher doses more effectively promote weight loss. In DMM-induced WT mouse (normal weight), we indeed observed that high-dose semaglutide effectively improved locomotor function and pain hypersensitivity while reducing body weight, and delayed cartilage degeneration and subchondral bone sclerosis. However, the low-dose group also demonstrated improvements in OA progression without significant changes in body weight, indicating that semaglutide's OA protective effects do not entirely depend on weight reduction. This provides important additional insight into the mechanism underlying semaglutide's protective effects against osteoarthritis.

We also found that GLP-1R expression in chondrocytes was significantly downregulated under inflammatory conditions, while semaglutide intervention restored its expression levels. This suggests that GLP-1R not only serves as a target for drug action but also participates in regulating the pathological process of OA. Further studies in Glp-1r KO mice revealed that the weight-reducing effect of semaglutide was completely absent in these mice. This observation aligns with earlier reports, where Scrocchi et al. first observed minimal weight changes in Glp-1r-deficient mice on a standard diet [41]. Simultaneous knockout of both Glp-1r and Gipr resulted in more pronounced metabolic abnormalities and pancreatic dysfunction, suggesting compensatory pathways exist in vivo following single receptor deficiency [42]. Concurrently, we observed that semaglutide's joint-protective effects also vanished. Even in DMM-induced Glp-1r KO mice, the OA phenotype was more severe than in WT mice, occurring despite no difference in body weight. These findings strongly indicate that the GLP-1R signaling pathway itself exerts a direct protective effect on cartilage homeostasis. Semaglutide's efficacy is achieved through direct activation of this pathway, rather than solely relying on reduced mechanical load from weight loss.

At the mechanistic level, our study revealed that semaglutide activates GLP-1R in chondrocytes, thereby inhibiting the AKT/mTOR signaling pathway and enhancing autophagic activity. This finding holds dual significance. First, autophagy-a crucial catabolic process within cells-is typically suppressed in OA chondrocytes, and restoring autophagy is considered a potential therapeutic strategy for OA [43]. Our results demonstrate that semaglutide effectively reverses IL-1β-induced autophagy suppression, manifested by increased LC3-II/LC3-I ratio, enhanced p62 degradation, and restored ATG5 expression. Recent studies have also shown that GLP-1R agonists can delay systemic aging through a brain-dependent pathway, and that the resulting multi-omics signature closely resembles that induced by mTOR inhibition [44]. Together with these findings, our data suggest that semaglutide maintains cartilage homeostasis not only by inhibiting the AKT/mTOR pathway, but also by relieving mTOR-mediated suppression of autophagic activity. This implies that GLP-1R activation may initiate a broader homeostatic program that couples metabolic reprogramming with restoration of autophagy in OA chondrocytes. Second, we identified distinct signaling regulation between semaglutide and the short-acting GLP-1R agonist liraglutide. While literature reports liraglutide primarily exerts cytoprotective effects by activating the PI3K/AKT pathway [45], our data demonstrate that semaglutide, as a long-acting agonist, conversely inhibits AKT/mTOR pathway activity. This signaling divergence likely stems from differences in the pharmacokinetics and receptor occupancy duration between long-acting and short-acting GLP-1R Agonists (GLP-1RAs) in vivo. Sustained receptor activation may trigger distinct downstream signaling feedback and reprogramming compared with transient activation. In addition, GLP-1R signaling may also be modulated by other ligands and by changes in receptor occupancy over time. Therefore, the differences observed between semaglutide and liraglutide may reflect not only their distinct pharmacokinetic profiles, but also broader GLP-1R signaling dynamics. This finding offers a novel perspective for understanding the tissue-specific effects of different GLP-1Ras.

From a translational medicine perspective, our study provides robust preclinical evidence supporting the use of semaglutide for OA treatment. First, we employed a subcutaneous dosing regimen consistent with clinical practice, enhancing the translational relevance of our findings. Second, we observed that semaglutide not only improved cartilage status but also alleviated subchondral bone sclerosis, suggesting its effects may extend beyond chondrocytes to include multi-target regulation of the osteochondral unit. This comprehensive structural improvement may further promote pain relief and functional recovery. Regarding safety, extensive clinical trials have demonstrated that semaglutide is generally well tolerated [46,47], and our study did not observe pathological changes in major organs, supporting its safety for long-term use. Importantly, a recent study reported that semaglutide reduced lean mass and weakened contractile force in some skeletal muscles in mice, suggesting that muscle-related safety is a clinical concern during GLP-1R agonist therapy [48]. In this context, the low-dose regimen used in our study achieved significant chondroprotective effects without affecting body weight, indicating that semaglutide may confer cartilage benefits at a dose that is less likely to impose an obvious systemic catabolic burden.

While we were completing this study, a recent publication also confirmed the chondroprotective effects of semaglutide in osteoarthritis [49]. Consistent with these findings in obese mouse models, we demonstrate in non-obese DMM mice that semaglutide protects cartilage independently of weight loss. That study identified AMPK- and PFKFB3-mediated metabolic reprogramming as a key mechanism, whereas our work reveals a complementary pathway involving suppression of AKT/mTOR signaling and restoration of autophagy. Collectively, these findings across distinct OA models suggest that semaglutide may preserve cartilage homeostasis through coordinated regulation of metabolic optimization and autophagic clearance.

This study still has several limitations. First, the DMM model is essentially a highly aggressive post-traumatic osteoarthritis model, in which persistent joint mechanical abnormalities induced by destabilization of the medial meniscus lead to progressive cartilage damage. Therefore, based on the current experimental design, it remains difficult to precisely distinguish whether the joint-protective effects observed with semaglutide are mainly attributable to its anti-catabolic actions or whether they also involve promotion of cartilage anabolic metabolism. Second, this study used a DMM-induced traumatic OA model, which has inherent limitations in simulating the pathological progression of age-related, degenerative OA and may, to some extent, restrict the direct extrapolation of our findings to elderly patients with clinical OA. Future studies should further validate these findings in a broader range of OA models, including cartilage defect repair models, joint immobilization models after DMM surgery, and natural aging models, so as to more comprehensively and precisely evaluate the functional characteristics and therapeutic potential of semaglutide across different OA subtypes.

OA is a heterogeneous whole-joint disease with distinct clinical phenotypes and molecular endotypes, and radiographic KL grading alone cannot fully capture the coexistence of different cartilage damage stages within a joint [50]. Therefore, future clinical translation of semaglutide should consider phenotype- or endotype-based patient stratification rather than relying solely on KL grade. A recent systematic review by Cheng et al. further summarized the pre-clinical and human evidence for GLP-1 receptor agonists in OA, supporting the need for stratified, phenotype-driven clinical development [51].

In summary, this study systematically reveals the distinct therapeutic efficacy of the antidiabetic drug semaglutide in osteoarthritis and its unique mechanism of action. Our findings indicate that semaglutide directly activates GLP-1 receptors on chondrocytes, thereby inhibiting the AKT/mTOR signaling pathway. This action enhances cellular autophagy activity, improves metabolic homeostasis in chondrocytes, and ultimately exerts joint-protective effects. This mechanism operates independently of its systemic weight-loss effects, offering a novel perspective on the role of GLP-1R agonists within the skeletal system. Given semaglutide's established safety profile and broad metabolic benefits in clinical practice, our findings strongly support its strategic repositioning as a potential disease-modifying agent for metabolic osteoarthritis, laying a robust theoretical foundation for subsequent clinical investigations.

Availability of data and materials

The data and materials supporting the findings of this study are available from the corresponding author upon reasonable request. The raw RNA sequencing data generated during the current study have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession number PRJNA1455299.

Ethics Committee

All animal procedures were approved by the Ethics Committee (IACUC-SAHCQMU-2025-0161)

Author contributions

X.D.S., M.N., W.H.X. and J.M.L. designed the study. J.M.L., M.L.L., H.X.T. and K.F.L. acquired the data. X.J.Y., Y.X.M. and Q.L.J. analyzed the data. J.M.L., M.L.L., and X.D.S. wrote the manuscript. Z.L.D., X.D.S., M.N. and W.H.X. reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.

Declaration of generative AI in scientific writing

During the preparation of this work, the authors used Grammarly for language polishing and grammar checking. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Funding

The study was supported by research funding from the National Natural Science Foundation of China (No.82372397), Chongqing Science and Technology Bureau (CSTB2025TIAD-STX0001), Chongqing Medical Scientific Research Project (Joint project of Chongqing Health Commission and Science and Technology Bureau) (2025QNKX005), Senior Medical Talents Program of Chongqing for Young and Middle-aged (YXGD202525) and China Postdoctoral Science Foundation (2022M720611, 2024T171107).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have influenced the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

figs1.

figs2.

figs3.

figs4.