Exercise alleviates age-related osteoarthritis associated with PGC-1α activation to suppress chondrocyte senescence and aberrant glycolysis

Abstract

Background

Age-related osteoarthritis (OA) is a leading global disability cause. Exercise is recommended as first-line therapy but poorly tolerated by elderly patients. This study elucidated the mechanism of low-intensity exercise in OA and develop a targeted exercise-mimetic nanotherapy.

Methods

The effects of low-intensity swimming exercise on age-related OA were investigated using aged mouse models. A senescence-targeted nanoparticle system (NP-B2M) was fabricated via a double-emulsion solvent evaporation method combined with EDC/NHS conjugation. The mechanisms underlying exercise and PGC-1α–mediated metabolic reprogramming were explored using transcriptomic sequencing, glycolysis analysis, RT-qPCR, and Western blotting in IL-1β–induced senescent ATDC5 chondrocytes. The therapeutic efficacy of intra-articular delivery of NP-B2M was systematically evaluated using micro-CT, histological staining, and immunohistochemistry in aged mice.

Results

Low-intensity swimming attenuated OA by suppressing chondrocyte senescence and pathological glycolysis via activating PGC-1α, effects that were recapitulated by ZLN005. NP-B2M enabled targeted intra-articular drug release and showed superior efficacy over free ZLN005.

Conclusion

This study demonstrates that low-intensity exercise alleviates age-related OA via PGC-1α–mediated metabolic reprogramming. NP-B2M provides a biocompatible, precisely targeted, long-acting platform that mimics exercise benefits, offering promising translational potential as a precision therapeutic strategy for immobile elderly OA patients.

The translational potential of this article

In this study, we developed NP-B2M, a PLGA-based nanoplatform, achieves prolonged joint retention and enhanced efficacy versus free drug, representing a promising targeted approach for OA patients with limited mobility.

Graphical abstract

1. Introduction

Osteoarthritis (OA) is a major chronic degenerative joint disorder and a leading contributor to global disability, with an estimated prevalence exceeding 500 million cases worldwide [1]. This progressive joint disease commonly affects multiple anatomical sites, with the knee joint being particularly vulnerable, primarily due to its weight-bearing function and sustained exposure to mechanical stress during daily activities [2]. The prevalence of OA continues to increase steadily, driven by population growth, aging demographics, and the absence of disease-modifying therapies [3]. Epidemiological projections indicate that, compared to 2020, the total global number of knee OA cases is expected to increase by 74.9% by 2050 [4].

Exercise, as a first-line non-pharmacological intervention for OA prevention, is consistently recommended in clinical practice guidelines [5]. However, the optimal exercise dosage is difficult to establish. Recent RCT on knee OA, comparing high-intensity to low-intensity resistance training, found no between-group differences in pain or muscle strength [6]. Notably, aquatic exercise provides bonus therapeutic benefits for OA patients through the weight-relieving properties of water. This buoyancy effect facilitates pain reduction and enhances joint mobility, thereby improving physical function, particularly in elderly OA populations [7]. However, current evidence remains insufficient to establish its long-term therapeutic benefits [8]. Furthermore, the mechanisms underlying long-term aquatic exercise benefits in age-related OA require further elucidation.

Exercise therapy has a positive impact on systemic metabolic reprogramming [9]. Emerging evidence suggests that exercise suppresses apolipoprotein E (APOE) transcription, thereby attenuating APOE-mediated cholesterol transport and exerting a critical regulatory influence on lipid metabolism [10]. In addition, physical exercise can decrease glutamate-induced nociception through G-protein-coupled receptor activation to reduce pain transmission [11]. Despite these advances, the critical relationship between exercise-induced metabolic shifts and glycolytic regulation in age-related OA remains poorly understood, particularly given the well-documented glycolytic phenotype characteristic of OA pathogenesis. Under conditions of aging, chondrocytes tend to adapt their metabolism to microenvironmental changes by shifting from oxidative phosphorylation to glycolysis; meanwhile, senescence markers are abnormally elevated alongside enhanced glycolysis, indicating a close correlation between glycolytic reprogramming and chondrocyte senescence [[12], [13], [14]]. This metabolic reprogramming has prompted investigations into developing multiple therapies that target glycolysis-related metabolism in OA management [15]. Given this context, elucidating exercise-mediated regulation of chondrocyte glycolysis emerges as a crucial yet underexplored research frontier with significant therapeutic implications.

The emerging concept of exercise mimetics has garnered significant research attention. Given the well-established health benefits of physical exercise, there is growing interest in identifying compounds that can replicate exercise-induced metabolic adaptations for therapeutic applications. This approach holds particular promise for patients with limited exercise capacity, including elderly populations. In 2008, researchers have identified AMPK and PPAR agonists as prototypical exercise mimetics [16]. Exercise exerts systemic beneficial effects across multiple mammalian organ systems, many of which are mediated through the transcriptional coactivator PPAR-γ coactivator-1α (PGC-1α) [17]. Consequently, PGC-1α agonists represent promising candidates for developing exercise mimetics in OA treatment. However, current intra-articular drug administration strategies require high-dose regimens and frequent injections, potentially causing adverse effects including pain, infection, inflammation, and poor patient compliance [18]. To overcome these clinical limitations, nanoparticle-based delivery systems offer a promising alternative. Poly (D, L-lactic-co-glycolic acid) (PLGA) nanoparticles have emerged as a particularly effective drug delivery platform, owing to their controlled biodegradation profile that allows slow release [19]. These properties suggest that PLGA nanoparticle-mediated delivery of exercise mimetics may offer a novel therapeutic strategy for OA management with enhanced efficacy and reduced side effects.

In this study, we demonstrate that long-term, low-intensity swimming intervention provides superior protective effects against age-related OA compared to moderate-intensity exercise. Mechanistically, exercise modulates articular cartilage homeostasis by suppressing pathological glycolytic activation in chondrocytes. Furthermore, we developed a PLGA-based nanoparticle (NP) delivery system that selectively targets senescent chondrocytes and enables sustained release of the PGC-1α agonist ZLN005, achieving precise intra-articular delivery of exercise mimetics. This therapeutic strategy effectively ameliorates age-related OA and may benefit elderly populations with limited exercise capacity.

2. Results

2.1. Exercise alleviates age-related OA

Given the limited therapeutic efficacy of high-intensity exercise in OA management, particularly in elderly populations [6,20], we systematically evaluated whether moderate- or low-intensity swimming exercise could more effectively attenuate age-related OA progression in mice. Based on established protocols where 30-min swimming represents moderate-intensity exercise [21], we adapted the regimen for aged mice to minimize aquatic stress and exercise-induced physiological burden (Fig. 1a). Following 8-weeks of intervention, mice in the exercise group exhibited a trend toward reduced body weight, yet no significant differences in body weight were observed across all experimental groups (Fig. S1b). MicroCT analysis revealed that aged mice showed significantly increased knee osteophyte formation and decreased subchondral bone volume/tissue volume (BV/TV), with trabecular thickness (Tb.Th), separation (Tb.Sp), and number (Tb.N) unaltered (Fig. 1b, c, Fig. S1a). This isolated BV/TV decline with stable Tb.Th and Tb.Sp indicates early-stage bone loss instead of mature trabecular remodeling, consistent with age-related skeletal alterations in mice where bone mass diminishes prior to severe trabecular microarchitectural damage [14]. Notably, swimming exercise effectively reversed the above aging phenotypes: it suppressed osteophyte generation and preserved subchondral BV/TV without shifting trabecular microstructural parameters, implying exercise exerts protective effects mainly via maintaining joint structural integrity and bone metabolic homeostasis, rather than modulating trabecular remodeling itself. Based on hematoxylin-eosin (HE) and safranin O staining of knee joint coronal sections, aged mice exhibited severe osteoarthritic degeneration, characterized by fissures in the articular cartilage, reduced safranin O staining within the cartilage matrix, and loss of chondrocytes (Fig. 1d). These pathological alterations were quantitatively assessed using the OARSI scoring system (Fig. 1e) [22]. Similarly, swimming exercise was found to ameliorate the age-related OA. Immunohistochemical analysis revealed an increase in the positive area for Aggrecan and COLII, accompanied by a decrease in the number of Matrix Metalloproteinase 13 (MMP13) and A Disintegrin And Metalloproteinase with Thrombospondin Motifs 5 (ADAMTS5)-positive cells following exercise (Fig. 1f–j). In addition, inflammatory scoring of mouse joint synovium demonstrated that exercise mitigated senescence-induced synovial inflammation and aberrant synovial hyperplasia, with the most prominent therapeutic effects detected in the ALS group, which indicated that the protective effects are not restricted to cartilage tissue alone (Fig. 1k and l). These collective findings suggest that swimming contributes to the restoration of extracellular matrix homeostasis in articular cartilage, with the low-intensity protocol yielding superior outcomes.

Fig. 1.

Exercise alleviates age-related OA. a Schematic of the physical exercise cohort. YS: young sedentary; AS: aged sedentary; ALS: aged low-intensity swimming; AMS: aged moderate-intensity swimming b Representative three-dimensional reconstructed micro-CT images of the mouse knee joint and subchondral bone. c bone volume over total volume (BV/TV) of subchondral bone according to μCT scanning. d Representative HE and Safranin O staining of knee joints (bar: 200 and 50 μm). e OARSI scores based on HE and Safranin O staining. f Representative immunohistochemical images of mouse knee joint (bar:50 μm). g-j Quantitative analysis of Aggrecan, COL II, MMP13 and ADAMTS5. k, l Representative HE staining images and quantitative analysis of Synovium (bar:50 μm). All data are shown as the mean ± SD, n = 3 mice per group in each biological experiments.

2.2. Exercise inhibited chondrocyte senescence and the concomitant increase in glycolysis

To determine whether exercise could reverse the senescent phenotype of chondrocytes, we first performed IHC staining for p16 and p21, which were well-established biomarkers of cellular senescence (Fig. 2a) [23]. Quantitative analysis revealed that low-intensity exercise was more effective than moderate-intensity exercise in reducing the percentage of p16-and p21-positive chondrocytes (Fig. 2b). To further investigate the mechanism by which exercise inhibits chondrocyte senescence, we performed RNA sequencing on articular cartilage from each group. Principal component analysis (PCA) revealed high intra-group similarity and clear segregation between AS group and both YS and ALS groups, suggesting that low-intensity swimming can reverse the transcriptomic alterations associated with cartilage aging (Fig. 2c). Of note, according to KEGG and GSEA analysis, the enrichment of Glycolysis/Gluconeogenesis drew our attention (Fig. 2d–f). Multiple studies have indicated that aberrant increases in glycolysis across various cell types can directly or indirectly exacerbate the progression of OA [24,25]. Furthermore, under aging conditions, chondrocytes tend to undergo a metabolic shift from oxidative phosphorylation to glycolysis as an adaptation to the altered microenvironment [13]. A detailed examination of the data through heat map analysis revealed a significant impact of exercise on glycolysis-related gene expression (Fig. 2e). Meanwhile, RT-qPCR analysis confirmed that the mRNA expression of four key glycolytic enzymes was significantly downregulated in the articular cartilage of exercised mice (Fig. 2g). Collectively, these findings demonstrate that low-intensity swimming exercise ameliorates chondrocyte senescence and concomitantly suppresses aberrant glycolytic activation, thereby rectifying the aging-related transcriptomic landscape in articular cartilage.

Fig. 2.

Exercise inhibited chondrocyte senescence and the concomitant increase in glycolysis. a Representative immunohistochemical images of p16 and p21 (bar:50 μm). b Quantitative analysis of p16 and p21. c Principal component analysis (PCA) of YS, AS, ALS and AMS group. d Transcriptome KEGG pathway ranking between ALS group and AS group. e Gene heatmap of glycolysis-related gene expression. f GSEA enrichment of Glycolysis/Gluconeogenesis between ALS and AS group. g RT-qPCR validation of four key glycolytic enzymes. Hk2: Hexokinase 2; Ldha: Lactate Dehydrogenase A; Pfkm: Phosphofructokinase; Pkm: pyruvate kinase M1/2. h Schematic of the IL-1β-induce ATDC5 senescence. i, j Representative SA-β-Gal staining image and quantitative analysis of the proportion of SA-β-gal-positive cells in control and IL-1β group. (bar:50 μm). k RT-qPCR validation of p21 and p53 between control and IL-1β group. l RT-qPCR validation of four key glycolytic enzymes. m The expression of interested proteins in control and IL-1β group. All data are shown as the mean ± SD, n = 3 mice per group in RNA-seq. n = 3 biologically independent experiments in other experiments.

2.3. Exercise-induced PGC-1α Attenuates Chondrocyte Senescence and Glycolysis in Articular Cartilage

To explore the mechanism by which exercise counteracts chondrocyte senescence, we established an in vitro senescence model by treating ATDC5s with IL-1β [26]. The successful induction of senescence was confirmed at multiple levels, including SA-β-Gal staining, gene transcription, and protein expression of SASP factors and biomarkers of cellular senescence (Fig. 2h–k, m; Fig. S1d–g). Furthermore, RT-qPCR results indicated that senescent chondrocytes exhibited aberrant activation of glycolysis, consistent with our in vivo findings (Fig. 2l). In addition, primary chondrocytes were isolated from the knee joints of 8-week-old C57BL/6 mice, and two additional in vitro chondrocyte senescence models were established by serial passaging and H₂O₂ stimulation. The results revealed that the metabolic and senescent phenotypes of chondrocytes induced by these two approaches were consistent with those of IL-1β-induced model cells, and the expression level of PGC1α was significantly downregulated as cellular senescence progressed (Fig. S2). Given that RNA-seq data indicated a suppressive effect of exercise on glycolytic flux, we sought to identify the key targets activated by exercise in cartilage that could reprogram this aberrant metabolic state. Substantial evidence from various tissues has established PGC-1α as an exercise-responsive master regulator of mitochondrial biogenesis and oxidative metabolism (Fig. 3a) [27]. Also, STRING analysis showed close interactions between PGC-1α and glycolytic enzymes (Fig. S3a). These lines of evidence led us to hypothesize that exercise may alleviate chondrocyte senescence associated with PGC-1α activation in cartilage, thereby inhibiting excessive glycolysis and restoring metabolic homeostasis. Firstly, IHC staining of human (Fig. 3b and c) and mouse articular cartilage tissues (Fig. 3d and e) indicated that aging significantly reduced the proportion of PGC-1α-positive chondrocytes, a decrease that was rescued by exercise intervention. Furthermore, our in vitro experiments confirmed this observation (Fig. 3f, Fig. S1d). Next, we sought to investigate the role of PGC-1α in chondrocyte senescence and glycolysis by performing both overexpression (Fig. S3b and c) and knockdown (Fig. S3d and e) of PGC-1α in ATDC5 cells. SA-β-gal staining revealed that overexpression of PGC-1α rescued IL-1β-induced chondrocyte senescence (Fig. 3g and h), while its knockdown exacerbated the senescent phenotype (Fig. S3g and h). Consistent with these findings, analyses at both the transcriptional and translational levels confirmed the anti-senescence effects (Fig. 3i, j, n). Furthermore, PGC-1α expression was positively correlated with the homeostasis of ECM metabolism in chondrocytes (Fig. 3i, j, m). Glycolytic rate assay and RT-qPCR confirmed PGC-1α′s ability to suppressed the senescence induced aberrantly elevated basal and compensatory glycolytic capacity in chondrocytes (Fig. 3k, l, o). Consistently, qPCR quantification of mitochondria-associated genes and ATP level detection revealed that PGC-1α activation partially restores mitochondrial function and drives metabolic reprogramming, rather than solely suppressing glycolytic activity (Fig. S3f and i). In summary, PGC-1α serves as a key mediator through which exercise ameliorates senescence and normalizes glycolytic metabolism in chondrocytes.

Fig. 3.

Exercise-Induced PGC-1α Attenuates Chondrocyte Senescence and Glycolysis in Articular Cartilage. a Schematic diagram illustrating the mechanism by which exercise promotes PGC-1α expression to restore metabolic homeostasis [27]. b, c Representative immunohistochemical images and quantitative analysis of PGC-1α in human cartilage tissues (bar:50 μm). d, e Representative immunohistochemical images and quantitative analysis of PGC-1α in mouse articular cartilages (bar:50 μm). g, h Representative SA-β-Gal staining image and quantitative analysis of the proportion of SA-β-gal-positive cells with or without the treatment of IL-1β or plasmid of PGC-1α (bar:50 μm). i The expression of interested proteins with or without the treatment of IL-1β or plasmid of PGC-1α. j The expression of interested proteins with or without the treatment of IL-1β or siRNA of PGC-1α. k Results from Seahorse glycolytic rate assay; ECAR: Extracellular Acidification Rate. l Compensatory and basal glycolysis analysis based on the ECAR data. m The mRNA expression of ECM metabolism-related genes in ATDC5s treated as described above. n The mRNA expression of cellular senescence-related genes in ATDC5s treated as described above. o The mRNA expression of four key glycolytic enzymes in ATDC5s treated as described above. n = 8 biologically independent experiments in seahorse experiments. n = 3 biologically independent experiments in the other experiments.

2.4. The PGC-1α Agonist Ameliorates Chondrocyte Senescence and glycolysis

Given our findings that PGC-1α is an exercise-activated target in chondrocytes with anti-senescence benefits, we next explored whether pharmacological activation of PGC-1α could mimic the protective effects of exercise and mitigate OA progression. To this end, we identified 2-(4-tert-butylphenyl) benzimidazole (ZLN005) (Fig. 4a), a known agonist demonstrated to upregulate PGC-1α expression in various cell types and shown to have therapeutic potential in neurological and circulatory diseases [28,29]. To assess the cytocompatibility of ZLN005, we first performed a CCK-8 assay on chondrocytes. The results indicated that a concentration of 20 μM was non-cytotoxic and suitable for subsequent in vitro experiments (Fig. 4b). To investigate its inhibitory effect on chondrocyte senescence, we repeated the experimental paradigm described above. Consistent with our hypothesis, ZLN005 significantly upregulated PGC-1α expression at both the transcriptional and translational levels (Fig. 4e and f). Meanwhile, ZLN005 treatment significantly reduced the percentage of SA-β-gal-positive cells and downregulated the expression of p16, p21 and γ-H2AX (Fig. 4c–f). Furthermore, it enhanced the metabolic homeostasis of the ECM in chondrocytes (Fig. 4e and f). Additionally, RT-qPCR analysis confirmed that ZLN005 reversed the senescence-associated aberrant expression of key glycolytic enzymes (Fig. 4g). This metabolic reprogramming was accompanied by improved mitochondrial function, as evidenced by JC-1 staining, indicating a restoration of the healthy chondrocyte metabolic profile (Fig. 4h). Meanwhile, to rule out potential off-target effects of ZLN005, we performed rescue experiments using siRNA. The results demonstrated that PGC-1α knockdown completely abolished the aforementioned protective effects of ZLN005 on chondrocytes (Fig. S4a–d).

Fig. 4.

The PGC-1α Agonist Ameliorates Chondrocyte Senescence and Glycolysis a The chemical structure of ZLN005. a The chemical structure of ZLN005. b Cell viability analysis of ATDC5 under different concentration of ZLN005 with CCK-8 c, d Representative SA-β-Gal staining image and quantitative analysis of the proportion of SA-β-gal-positive cells with or without the treatment of IL-1β or ZLN005 (bar:50 μm). e The expression of interested proteins in ATDC5s treated as described above. f, g The mRNA expression of relative genes of ECM homeostasis, cellular senescence and glycolysis in ATDC5s treated as described above. h Representative Mitochondrial membrane potential image using the JC-1 staining of ATDC5s (bar:40 μm). i Representative multiplex immunofluorescence images of knee joints of YS group and AS group (bar:20 μm and 2 μm). j, k The expression and quantitative analysis of B2M protein in IL-1β and control group. l Representative multiplex immunofluorescence images of ATDC5 with or without the treatment of IL-1β (bar:20 μm). m, n Fluorescence intensity distribution profiles along linear scanning distance of representative multiplex immunofluorescence images from l. Green curves represent B2M; red curves represent p21. n = 3 biologically independent experiments in all the experiments.

2.5. Development and Functional Validation of a Senescence-Targeting Nanoparticle for Exercise-Mimetic Therapy

Despite the established role of PGC-1α activation in mitigating chondrocyte senescence and aberrant glycolysis, the clinical translation of exercise mimetics remains challenging due to the limitations of conventional intra-articular drug delivery, including rapid clearance, frequent dosing, and systemic side effects [30]. To overcome these barriers, we developed a senescence-targeting nanoparticle system (NP-B2M) for intra-articular delivery of ZLN005. This system leverages the properties of PLGA to provide controlled drug release and excellent biocompatibility, while being functionalized with a B2M antibody to enable specific recognition of senescent chondrocytes (Fig. 5a). This design is supported by previous studies documenting elevated B2M levels in articular cartilage and synovial fluid in aging-related OA [31], and demonstrating the targeting capability of B2M antibody-conjugated nanoparticles to senescent chondrocytes [32]. By combining sustained drug release with cellular targeting, this nanotherapeutic strategy aims to enhance therapeutic efficacy while minimizing dosing frequency.

Fig. 5.

Development and Functional Validation of a Senescence-Targeting Nanoparticle for Exercise-Mimetic Therapy. a Schematic illustration of the NP-B2M fabrication process. b TEM image of Blank NP and NP-B2M (bar:200 nm) c The diameter range of Blank NP and NP-B2M. d Zeta potential of Blank NP and NP-B2M. e Release profile of ZLN005 from NP in PBS at pH 5.4 and 7.0 (37 °C). f FTIR spectra of Blank NP and NP-B2M. g Cell viability analysis of ATDC5s under different concentration of NP-B2M from day1-5 with CCK-8. h Immunofluorescence staining of NP-B2M (bar:100 μm). i Representative confocal microscopy images of ATDC5 cells, with or without IL-1β to induce senescence upon treatment with free Nile Red, NP, or NP-B2M (bar:40 μm). j Representative SA-β-Gal staining image of the proportion of SA-β-gal-positive cells in ATDC5s treated as described above. k The expression of interested proteins in ATDC5s treated as described above. l-n The mRNA expression of relative genes of cellular senescence, ECM homeostasis and glycolysis in ATDC5s treated as described above. n = 3 biologically independent experiments in all the experiments.

To achieve targeted delivery of ZLN005 to senescent chondrocytes, we developed NP using a double-emulsion method as previously described [33]. The resulting NP-B2M conjugates were synthesized by covalently linking B2M antibody to the surface of NP loaded with ZLN005 by 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide/N-hydroxy succinimide (EDC/NHS) method (Fig. 5a). Initially, we confirmed by immunoblot that B2M expression was significantly upregulated in senescent chondrocytes (Fig. 4j and k). Meanwhile, to verify whether B2M serves as a therapeutic target for senescent chondrocytes, multiple immunofluorescence staining assays were performed both in vitro and in vivo. In vivo, p21 and Aggrecan were used as canonical markers to label senescent chondrocytes. The results revealed that B2M was highly expressed in p21/Aggrecan double-positive chondrocytes from the articular cartilage of the AS group (Fig. 4i). Consistently, robust co-localization between B2M and p21 was observed in cultured chondrocytes in vitro, and the fluorescence intensity of B2M was markedly elevated following IL-1β stimulation (Fig. 4l–n). Therefore, B2M may act as a characteristic biomarker for senescent chondrocytes within osteoarthritic lesions, which lays a foundation for our subsequent development of NP targeting senescent chondrocytes. Transmission electron microscopy (TEM) revealed the spherical morphology of both NP and NP-B2M. A distinct corona surrounding the NP-B2M core was observed, suggesting successful surface conjugation of the antibody, a phenomenon consistent with previous reports on protein-coated nanoparticles (Fig. 5b) [34]. Zeta potential measurements revealed that after antibody conjugation onto blank NP, abundant negatively charged surface carboxyl groups of PLGA were consumed and replaced with weakly anionic antibody proteins. Consequently, the overall surface charge density decreased, accompanied by a positive shift in zeta potential, which verified the successful immobilization of antibodies on PLGA nanoparticles (Fig. 5d). We next evaluated the drug release profile of NP under physiologically relevant pH conditions at 37 °C. As shown in Fig. 5e, the NP exhibited sustained ZLN005 release over time, with a more rapid and complete release observed at pH 5.4 compared to pH 7.0, suggesting a responsive release behavior in the inflammatory OA microenvironment. Nanoparticle tracking analysis (NTA) revealed that the average diameters of the Blank NP and NP-B2M were 156.4 ± 50.2 nm and 209.0 ± 55.7 nm, respectively (Fig. 5c). Meanwhile, the NP-B2M were incubated in 50% serum at 37 °C to evaluate serum stability. The particles exhibited favorable serum stability, with the hydrodynamic size remaining above 100 nm after three days of incubation (Fig. S4f), and the endotoxin level was maintained within a low range (0.603 ± 0.052EU/ml) [35]. The encapsulation efficiency (EE%) and drug loading capacity (LC%) of NP-B2M were determined to be 80.36 ± 2.15% and 8.20 ± 0.21% as measured using UV–vis spectrophotometer. The FTIR spectra of NP and NP-B2M showed a small peak at 1625 cm−1 in the latter, which is characteristic of the amide bond formation, confirming the successful conjugation of the anti-B2M antibody to the nanoparticle surface (Fig. 5f) [36]. We further labeled the NP-B2M with Nile Red and subsequently probed with a FITC-conjugated secondary antibody. The strong colocalization of red (Nile Red) and green (FITC) fluorescence signals served as critical validation of the surface-functionalized anti-B2M antibody (Fig. 5h). Quantitative analysis revealed that the proportion of dual-positive spots reached 90.89% ± 1.9%, indicating a conjugation efficiency exceeding 90%. Furthermore, CCK-8 assays conducted over 1–5 days demonstrated that NP-B2M, at concentrations ranging from 0 to 400 μg/ml, exhibited excellent biocompatibility with ATDC5s (Fig. 5g). To assess the cellular targeting and internalization efficiency of NP-B2M, we performed confocal microscopy using Nile red-labeled nanoparticles. In IL-1β-induced senescent chondrocytes over control chondrocytes, NP-B2M showed significantly enhanced cellular uptake compared to non-targeted NPs or free Nile red, demonstrating its specificity toward senescent cells (Fig. 5i). We further evaluated the functional efficacy of NP-B2M in rescuing senescence-associated phenotypes. Consistent with the role of PGC-1α in metabolic reprogramming, treatment with NP-B2M markedly attenuated chondrocyte senescence and suppressed aberrant glycolysis, and restored ECM metabolic homeostasis as well as mitochondrial function (Fig. 5j–n, Fig. S4e and g). Finally, we further performed in vitro rescue experiments using siRNAs to verify that the protective effects of NP-B2M against chondrocyte senescence and its role in maintaining ECM homeostasis are mediated by PGC-1α, thereby ruling out potential off-target effects (Fig. S4h–k). Collectively, these results validate NP-B2M as an effective, senescent cell-targeted delivery system that not only enhances cellular uptake and retention but also recapitulates the anti-senescence and metabolic-normalizing effects of exercise mimetics.

2.6. Intra-articular delivery of NP-B2M mitigated age-related OA

The therapeutic potential of NP-B2M was assessed in 18-month-old mice via intra-articular injection, followed by sacrifice and analysis after 8 weeks (Fig. 6a). MicroCT analysis revealed that NP-B2M treatment more effectively increased the subchondral BV/TV and reduced osteophyte formation compared to free ZLN005, although it did not significantly alter Tb.Th and Tb.Sp (Fig. 6b, Fig. S5e and f). Histological analysis (HE and Safranin O staining) demonstrated that NP-B2M treatment resulted in improved tissue integrity, characterized by reduced surface fissures, greater cartilage area (Fig. 6d, e), and a lower OARSI score (Fig. 6f). To assess potential NP-B2M retention, we performed confocal imaging on knee sections. Notably, Nile Red fluorescence was still observed within the articular cartilage after the 8-week intervention, indicating a prolonged intra-articular retention of NP-B2M (Fig. 6g). Finally, we validated the expression of key proteins by IHC and immunoblot. The agonistic effect on PGC-1α observed in the free ZLN005 group was largely diminished after 8 weeks. In contrast, the NP-B2M group maintained a sustained activation of PGC-1α (Fig. 6h, i, Fig. S5g). Furthermore, NP-B2M more effectively mitigated age-related OA progression than free ZLN005, as evidenced by greater improvements in ECM metabolic homeostasis and stronger suppression of senescence markers (Fig. 6j–n, S5a-d, g). Consistently, NP-B2M alleviated senescence-triggered synovial inflammation and aberrant synovial hyperplasia, with superior efficacy compared to free ZLN005. These observations suggest that NP-B2M recapitulates the therapeutic benefits of exercise (Fig. S5h and i). Collectively, these results demonstrate that NP-B2M exerts superior therapeutic effect against age-related OA by enhancing drug retention and persistently associated with PGC-1α activation.

Fig. 6.

Intra-articular delivery of NP-B2M mitigated age-related OA. a Schematic illustration of NP-B2M administration in vivo. b Representative three-dimensional reconstructed micro-CT images of the mouse knee joint and subchondral bone. c BV/TV according to μCT scanning. d, e Representative HE and Safranin O staining of knee joints (bar: 200 and 50 μm). f OARSI scores based on HE and Safranin O staining. g Representative images of DAPI staining of knee joints. h, i Representative immunohistochemical images and quantitative analysis of PGC-1α of mouse knee joint (bar:50 μm). j-l Representative immunohistochemical images and quantitative analysis of COL II and MMP13 of mouse knee joint (bar:50 μm). m, n Representative immunohistochemical images and quantitative analysis of p16 of mouse knee joint (bar:50 μm). All data are shown as the mean ± SD, n = 3 mice per group in each biological experiments.

3. Discussion

This study systematically elucidates the mechanism by which low-intensity exercise alleviates age-related OA through activation of PGC-1α to suppress chondrocyte senescence and aberrant glycolysis, through integration of in vivo and in vitro experiments. More importantly, we successfully developed and validated a nanoparticle, NP-B2M, that targets senescent chondrocytes and enables the sustained intra-articular delivery of the exercise mimetic ZLN005, thereby mimicking the therapeutic benefits of exercise in elderly individuals who cannot benefit from physical activity.

OA guidelines recommend exercise as the core form of therapeutic OA management [37]. However, especially for the elderly, the optimal type or dose of therapeutic exercise that most effectively provides OA management is not currently known [38]. Wu et al. [10] have indicated that swimming exercise confers beneficial effects by mitigating local joint damage through enhanced systemic antioxidant capacity, the optimal exercise intensity specifically tailored for the elderly population remains to be determined. Here, our research provides new scientific evidence for the suitable exercise intensity for the elderly in OA management. We found that long-term, low-intensity swimming exercise demonstrated superior effects compared to the moderate-intensity regimen in improving subchondral bone architecture, reducing osteophyte formation, maintaining cartilage ECM homeostasis, and suppressing cellular senescence. This may be attributed to the lower mechanical load of low-intensity swimming exercise, as well as the composite nature of aquatic intervention—in which mild hydrostatic pressure promotes peripheral circulation and alleviates synovial swelling, while buoyancy reduces weight-bearing joint compression, enabling safe long-term exercise without impact injury. This aligns with clinical observations showing no significant differences in pain relief or muscle strength between high- and low-intensity resistance training in knee OA patients [20], while our findings further provide mechanistic insights at the molecular level of cellular senescence and metabolic reprogramming, highlighting the unique advantages of long-term, low-intensity exercise.

Mechanistically, we linked the beneficial effects of exercise to a key pathological feature of OA—cellular senescence and its associated metabolic dysregulation [13,39]. RNA sequencing and functional experiments collectively demonstrated that the protective role of exercise is closely associated with the suppression of the senescent phenotype and the correction of aberrant glycolytic activation in chondrocytes. We further identified PGC-1α as the central mediator of this process. As a master regulator of the exercise response, PGC-1α activation has been shown to ameliorate osteoarthritis pathologies, including impaired mitochondrial biogenesis, oxidative stress, and inflammation [40,41]. However, its role in age-related OA, particularly the mechanistic links among PGC-1α activation, glycolytic reprogramming, and chondrocyte senescence, remains poorly understood. In this study, we demonstrate that PGC-1α activation not only directly ameliorated cellular senescence but also successfully reversed the cellular metabolic profile from a glycolytic-dominant state towards a healthier metabolism model, thereby improving the survival environment and function of chondrocytes. This discovery tightly connects the three critical fields of exercise, cellular senescence, and metabolic regulation, deepening our understanding of how exercise fundamentally mitigates OA progression.

Addressing the clinical translation challenges of exercise mimetics, the NP-B2M nano-delivery system developed in this study offers an innovative solution. Conventional intra-articular injections face challenges such as rapid clearance, frequent dosing, and significant side effects [42]. Antibody-drug conjugates (ADCs), which leverage the specific binding between antibodies and target antigens, have been extensively investigated as a strategy to achieve tissue-specific drug delivery—such as in oncology—thereby minimizing systemic exposure and associated side effects [42]. Guided by this concept, we designed our nanotherapeutic to achieve targeted drug delivery by leveraging the high expression of B2M on senescent chondrocytes, and antibody-mediated targeting enhances therapeutic specificity and reduces off-target risks. PLGA is a synthetic copolymer approved by the FDA and the European Medicine Agency for pharmaceutical and biomedical applications due to its excellent biocompatibility and predictable biodegradability [43]. Firstly, the terminal carboxyl groups (-COOH) on PLGA nanoparticles provide an accessible chemical handle for covalent conjugation of targeting ligands. This was essential for our design, enabling us to functionalize the nanoparticle surface with the anti-B2M antibody via EDC/NHS chemistry to generate NP-B2M, thereby conferring specific targeting capability to senescent chondrocytes (Fig. 5a). Secondly, leveraging the controlled-release capability of PLGA, the NP-B2M formulation achieved a prolonged intra-articular retention of ZLN005, which was maintained for up to 8 weeks in vivo (Fig. 6g). This property effectively addresses the rapid clearance limitation of conventional intra-articular injections, thereby significantly extending the dosing interval. Finally, in vivo efficacy evaluation demonstrated that NP-B2M was significantly superior to free ZLN005 in improving OA structural pathology, restoring cartilage integrity, and sustaining PGC-1α pathway activation, which was capable of mimicking the positive effects of low-intensity swimming against OA progression. Taken together, these findings validate the synergistic advantages of targeted and sustained release formulations. Such a translational platform can effectively mimic the chondroprotective outcomes of physical exercise and provide therapeutic benefits to immobile aged populations.

Our study has several limitations. First, the long-term biosafety and potential immunogenicity of NP-B2M require more comprehensive evaluation in larger animal models. Further systematic investigations are needed to determine its in vivo safety profile, including histopathological assessment of major organs, serum biochemical parameters, and immunogenic responses, prior to potential clinical translation. Second, the specific upstream signaling pathways through which exercise activates PGC-1α, such as the potential involvement of mechanosensitive ion channels or specific kinases, represent an important direction for future investigation. Third, a direct in vivo comparison between the effect size of the selected ZLN005 dose and that of low-intensity swimming has not been performed. Although our in vitro data suggest functional convergence, we acknowledge that establishing dose-to-intensity equivalence in vivo remains an open question. Future studies incorporating dose titration alongside exercise protocols are warranted to more precisely define the functional equivalence of pharmacological and physiological PGC-1α activation. Furthermore, applying this senescent cell-targeting nano-strategy to other aging-related joint diseases offers broad exploration potential. Additional in vivo control groups, such as blank NP-B2M, non-targeted NP-ZLN005 or repeated free ZLN005 injections, would also help further dissect the relative contributions of targeting versus sustained release. Finally, Future studies using genetic model will be valuable to further establish the causal role of cartilage PGC-1α in exercise-mediated OA protection and to strengthen the translational rationale for PGC-1α-targeted therapies.

In conclusion, our work not only reveals a novel mechanism by which low-intensity exercise protects joint health via PGC-1α-mediated coordination of senescence and metabolism but, more importantly, successfully develops a biomimetic nanotherapeutic strategy that mimics the benefits of exercise. By enabling the targeted and sustained delivery of an exercise mimetic, NP-B2M provides a promising and translatable precision therapeutic option for elderly OA patients who are unable to benefit from exercise rehabilitation due to mobility limitations.

4. Materials and methods

4.1. Ethics approval statements

All animal experiments were performed with protocols approved by the Tongji University Ethnics Committee on the Use and Care of Animals (Approval NO. TJAA13023101). Human cartilage samples were obtained from patients undergoing knee replacement surgery and supernumerary finger specimens were collected from pediatric patients with polydactyly undergoing surgical resection at Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine. The study was approved by the Institutional Review Board of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine (Approval No. SH9H-2021-T401-2). All patients provided written informed consent prior to sample collection. Human samples were anonymized and used strictly for research purposes in accordance with the Declaration of Helsinki.

4.2. Animal model and exercise cohort

Eighteen-month-old male C57BL/6 mice underwent a 1-week swimming adaptation period to minimize water-associated stress prior to formal exercise training. The exercise protocol consisted of two phases: (1) During weeks 2-5, the acute low-intensity swimming (ALS) group (n = 6) performed 10-min sessions (3×/week), while the acute moderate-intensity swimming (AMS) group (n = 6) completed 20-min sessions; (2) From week 6 onward, session durations were increased to 15 min (ALS) and 30 min (AMS) to maintain training efficacy (Fig. 1a). All swimming exercises were conducted in acrylic tanks (41 × 27 × 35 cm) with water temperature maintained at 35 ± 1 °C during the early active phase of the mice. Eighteen-month-old aged mice were utilized as aged sedentary (AS) group (n = 6). And eight-week-old young mice were utilized as the young sedentary (YS) control group (n = 6).

To evaluate the therapeutic effect of NP-B2M in vivo, eighteen-month-old male C57BL/6 mice were randomly assigned to four groups: Control, Vehicle, ZLN005, and NP-B2M. The mice received a single intra-articular injection (10 μl) into the right knee joint of PBS, ZLN005 (1.5 mg/ml), or NP-B2M (15 mg/ml). After eight weeks of routine maintenance, all mice were euthanized for subsequent analysis.

All mice were randomly assigned to groups via a GraphPad Prism-generated random number table prior to OA modeling and treatment to avoid subjective grouping bias. All operators and analysts remained blinded to sample grouping during micro-CT scanning, tissue sectioning, staining and quantitative analysis. Samples were labeled with random codes, and group identities were revealed only after all quantification was finished. Animals were excluded if they had severe infection or death, unrelated limb trauma, spontaneous advanced cartilage damage detected by baseline micro-CT, or damaged tissue sections unfit for quantitative analysis. Each mouse represented one biological replicate. Multiple serial knee sections per mouse served as technical replicates for histology. Technical replicate values from each mouse were averaged before intergroup statistical comparison.

4.3. Cell culture and treatment

Primary chondrocytes were isolated from the knee joints of 8-week-old C57BL/6 mice. Briefly, mice were euthanized via cervical dislocation under sterile conditions, and the bilateral knee joint tissues were rapidly dissected. The intact femoral condyles and tibial plateau cartilages were exposed and collected. The harvested cartilage tissues were rinsed repeatedly with pre-cooled PBS containing 1% penicillin–streptomycin to remove residual blood cells and tissue debris.

Subsequently, the cartilage tissues were cut into fine pieces and subjected to two-step enzymatic digestion. Firstly, the tissue fragments were digested with 0.25% trypsin for 30 min at 37 °C to remove superficial fibrous tissues. After discarding the trypsin solution, the samples were further digested with 0.2% type II collagenase for 4–6 h at 37 °C. The digested cell suspension was filtered through a 70 μm cell strainer to eliminate undigested tissue masses. After centrifugation at 1000 rpm for 5 min, the cell pellet was resuspended in complete DMEM/F-12 (Gibco, Thermo Fisher Scientific, Inc., 11320033) supplemented with 10% FBS (Cyagen Biosciences, Guangzhou, China, FBSSR-01021-500) and 1% penicillin–streptomycin (Pricella Life Science & Technology, PB180120). Finally, chondrocytes were seeded in culture dishes and incubated at 37 °C in a humidified atmosphere containing 5% CO₂. Chondrocytes at passages 1–3 were used for subsequent senescence induction experiments.

The ATDC5 chondrogenic cell lines were purchased from KeyGEN BioTECH (KeyGEN, China, KG445) and cultured in DMEM/F-12 supplemented with 10% FBS, 1% penicillin-streptomycin at 37 °C in 5% CO₂. The cells were passaged once after three days.

In vitro senescence of chondrocytes was induced by treatment with IL-1β (10 ng/ml, TargetMol, China, TMPY-02134) in serum-free medium (Fig. 2g). Meanwhile, the ATDC5 were treated with or without ZLN005 (20 μM, TargetMol, China, T1802) and NP-B2M (200 μg/ml). Two senescent chondrocyte models were constructed separately via stimulation with 200 μM H₂O₂ and continuous serial passaging up to passage 11.

4.4. Cell transfection

Transient transfection of ATDC5s was performed using Lipo3000 (Invitrogen, USA, L3000075), siRNA and plasmids (Sangon, China). All processes followed the manufacturer's standard operating procedures (siRNA : lipo3000 = 10 : 1 (pmol : μl); plasmid : lipo3000 = 1 : 3 (μg : μl)). The effect of gene knockdown or overexpression were confirmed by WB and Real-time qPCR analysis. The siRNA and plasmid sequences were shown in the Extended data materials (Supplementary Tables 2 and 3).

4.5. Micro-CT analysis

Mice were euthanized after exercise. The knee-joint tissues were fixed in 4% paraformaldehyde, and microcomputed tomography (Quantum GX, PerkinElmer, USA) was used to evaluate the acquisition parameters were set as follows: 90 kV (voltage), 88 μA (current), and 25 mm (field of view), with a resolution of 27.5 μm. Next, bone volume/total volume (BV/TV), trabecular thickness (Tb. Th), trabecular separation (Tb. Sp), and trabecular number (Tb. N) analyses were conducted via Analyze 12.0 (PerkinElmer).

4.6. HE and safranin O staining

After Micro-CT analysis, the knee-joints of the mice were fixed with 4% paraformaldehyde (Servicebio, China, G1101), decalcified with 0.5 M EDTA (Servicebio, China, G1105), and embedded in paraffin. The sample was cut into 5 μm-thick sections along the coronal plane, and the sections were deparaffinized with xylene, hydrated with graded ethanol. H&E (Solarbio, China, G1076) and Safranin O staining (Solarbio, China, G1371) were performed according to the manufacturer's protocol. OA severity was quantitatively assessed using the OARSI histological scoring system [22]. The system for synovitis score system referred to the research conducted by Krenn V and the colleagues [44].

4.7. Immunohistochemical staining

To detect the expression levels of interested proteins in the knee-joints, immunohistochemical staining was carried out using a DAB kit (ZSGB, China, ZLI-9018) and a rabbit two-step detection kit (Boster, China, SV0002) according to the manufacturer's protocol. Immunohistochemistry quantification was performed using ImageJ (Image Progressing and Analysis in Java, version 1.42q, https://imagej.nih.gov/ij/).

4.8. Immunofluorescence staining

As previously described [45], 2 mg of NP-B2M was incubated with a FITC-conjugated goat anti-rabbit IgG (H + L) secondary antibody (1:200 dilution, Affinity Biosciences, China, S0008) at room temperature for 2 h. The Samples were then centrifuged at 15,000 × g for 15 min and washed twice with PBS to eliminate the excess secondary antibody. Finally, the pellets were resuspended in 20 μL of anti-fluorescence quenching sealing agent (Servicebio, China, G1401) and visualized under an OLYMPUS IX73 fluorescence microscope.

The colocalization and expression levels of interested proteins in chondrocytes were visualized using tyramide signal amplification kits (Servicebio, China, G1226) according to the manufacturer's protocol. And the images were captured with Zeiss LSM 900 confocal microscope (Zeiss, Germany).

4.9. RNA-seq analysis

Cartilage tissue transcriptome sequencing was performed by BGI Tech Solutions Co., Ltd. (China). Raw sequencing reads were preprocessed to remove adapter sequences and low-quality reads (length < 30 bp, average quality score < 9). The remaining high-quality reads (length ≥ 30 bp, average quality score ≥ 9) were aligned to the Mus musculus reference genome. Gene expression levels were quantified and normalized as reads per kilobase per million mapped reads (RPKM). Subsequently, the p values and log2 of each gene were then calculated. For functional annotation, differentially expressed genes (DEGs) were subjected to BLASTX alignment against the NCBI non-redundant database, followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation using Blast2GO. Gene set enrichment analysis (GSEA) analysis was utilized for comparing pathways between different groups. Enrichment analysis of DEGs was conducted using the g: Profiler platform (https://biit.cs.ut.ee/gprofiler/gost). Data visualization was performed using R (version 4.4.1).

4.10. Real-time qPCR

RNA was extracted from cells and cartilage tissues after the indicated treatments. RNAfast200 purification kit (Fastagen Biotech, China, 220011) was used for RNA extraction. The extracted RNA was reverse transcribed with PrimeScript™ RT Master Mix (Takara, Japan, RR036A). Then, according to the manufacturer's protocols, cDNA was amplified with TB Green Premix EX Taq II (Takara, Japan, RR820A). Relative gene expression was calculated using the 2^−ΔΔCq method, and β-actin was used as the internal reference. All processes followed the manufacturer's standard operating procedures. The qPCR primers were designed using NCBI Primer-Blast and are listed in the Extended data materials (Extended data Table S1).

4.11. Immunoblot

Total proteins were extracted using RIPA lysis buffer and quantified with a BCA assay kit (Beyotime, China, P0012S). Western blot analysis was performed according to standard protocols. Briefly, NC membranes were incubated with primary antibodies at 4 °C overnight, followed by incubation with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature. Protein bands were visualized using an ECL chemiluminescence detection kit (Abclonal, China, RM02867) and imaged with a western blotting detection system (Tanon V8, Tanon, China). The antibodies used were listed as follows: PGC1-α (1:1000 dilution, Abclonal, China, A20995), beta 2 Microglobulin (1:1000 dilution, Abclonal, China, A23430), Aggrecan (1:1000 dilution, Proteintech, China, 13880), MMP13 (1:1000 dilution, Proteintech, China, 18165), COL II (1:1000 dilution, Abclonal, China, A1560), p16 (1:1000 dilution, Cell Signaling Technology, USA, 23200), p21 (1:1000 dilution, Abcam, USA, ab109199), HSP60 (1:3000 dilution, Servicebio, China, GB11243), HRP-labeled Goat Anti-Rabbit IgG(H + L) (1:1000 dilution, Beyotime, China, A0208).

4.12. SA-β-gal staining

Following the indicated treatments, ATDC5 cells were fixed with 4% paraformaldehyde for 10 min at room temperature. Senescence-associated β-galactosidase (SA-β-gal) staining was subsequently performed according to the manufacturer's instructions (Solarbio, China, G1580) by incubating the cells with staining solution at 37 °C for 24 h in a non-CO₂ environment. Cells exhibiting high SA-β-gal activity, indicative of senescence, were identified by the development of blue staining. The number of SA-β-gal-positive cells was quantified by counting five randomly selected fields per slide under an OLYMPUS IX73 microscope.

4.13. Cell viability assay

Cell viability of ATDC5s were assessed using the Cell Counting Kit-8 (BIOSS, China, BA00208) to determine optimal concentrations of ZLN005 and NP-B2M. Following the manufacturer's protocol, absorbance measurements were performed at 450 nm using a microplate reader (BioTek EPOCH, USA).

4.14. ATP concentration assay

Cell ATP concentrations were assessed using the Enhanced ATP Assay Kit (Beyotime, China, S0027). Following the manufacturer's protocol, intracellular ATP levels were quantified by measuring relative light unit values with a luminometer (Promega, GloMax® Discover (GM3000), USA).

4.15. Cellular metabolic analysis

The glycolytic capacity was measured using the Agilent Seahorse XF Glycolytic Rate Assay (Agilent Technologies, USA, 103344-10) following the manufacturer's instructions. Subsequently, the data were analyzed and represented using the Wave software (v2.6.3.5, Agilent Technologies, USA). In this assay, only two compounds—rotenone/antimycin A (Rot/AA) and 2-deoxy-D-glucose (2-DG)—are sequentially injected to delineate distinct components of extracellular acidification. Rot/AA (final concentration 0.5 μM) inhibits mitochondrial complex I and III, thereby blocking oxidative phosphorylation and enabling quantification of mitochondrial-derived proton production; the subsequent increase in glycolysis reflects compensatory glycolytic capacity. 2-DG (final concentration 50 mM), a competitive inhibitor of hexokinase, is then added to suppress glycolysis, confirming that the majority of extracellular acidification originates from glycolytic lactate production. This two-injection protocol allows for the calculation of the glycolytic proton efflux rate by subtracting mitochondrial CO₂-derived acidification from total proton efflux, providing a more accurate measurement of glycolytic activity.

4.16. Mitochondrial membrane potential

Mitochondrial membrane potential was assessed using the JC-1 MitoMP Detection Kit (Dojindo, Japan, MT09) according to the manufacturer's instructions. Briefly, ATDC5 cells were seeded in confocal dishes and subjected to experimental treatments. Subsequently, the cells were stained with JC-10 dye under standard conditions for 30 min. Fluorescence images of JC-10 aggregates were acquired using a Zeiss LSM 900 confocal microscope (Zeiss, Germany) at 488 nm and 590 nm.

4.17. Preparation and characterization of NP-B2M

PLGA nanoparticles (NPs) encapsulating ZLN005 (TargetMol, China, T1802) and Nile red were synthesized via a double-emulsion solvent evaporation method [46]. Briefly, 13.5 mg of PLGA-COOH and 1.5 mg of ZLN005 were dissolved in 1 mL of dichloromethane and emulsified with 2 mL of deionized water by sonication (25 W, 1 s on/2 s off cycles) for 5 min in an ice-water bath. Subsequently, 5% sodium cholate solution was added, and the emulsion was further sonicated under the same conditions. The resulting emulsion was then added dropwise into 10 mL of 0.5% sodium cholate solution under continuous stirring for 10 min at room temperature. Organic solvents were removed by rotary evaporation. NPs were collected by centrifugation at 11,000 ×g for 15 min, washed three times with deionized water, and stored at 4 °C for further use. To conjugate the B2M antibody to NPs, an EDC/NHS coupling reaction was performed at a molar ratio of 1 : 3 (antibody : PLGA polymer chains) [47]. Unconjugated antibodies were removed through three cycles of centrifugation at 10,000 rpm for 10 min each. The morphology of the NPs was examined using transmission electron microscopy (TEM; Hitachi HT7800, Hitachi High-Technologies, Tokyo, Japan). Successful antibody conjugation was confirmed by Fourier transform infrared spectroscopy (FTIR; PerkinElmer Spectrum RX1 system) within the wavelength range of 4000–400 cm⁻¹.

4.18. Drug loading efficiency

The encapsulation efficiency (EE%) and loading capacity (LC%) of ZLN005 within the NPs were determined as follows. Briefly, 1 mg of NPs was accurately weighed and dissolved in PBS. The solution was sonicated for 30 min to ensure complete nanoparticle disruption. After centrifugation (10,000 rpm, 10 min), the supernatant was collected and analyzed using a UV–vis spectrophotometer (NanoPhotometer NP80, IMPLEN GmbH, Germany) at a wavelength of 310 nm [48]. The concentration of ZLN005 was quantified based on a pre-established standard calibration curve. All measurements were performed in triplicate. The LC% and EE% were calculated using the following equations:

$$EE(\%)=\frac{\text{Mass}\text{of}\text{encapsulated}\mathrm{Z}\mathit{LN}005}{\text{Total}\text{mass}\text{of}\mathrm{Z}\mathit{LN}005}\times 100\%$$

$$LC(\%)=\frac{\text{Mass}\text{of}\text{encapsulated}\mathrm{Z}\mathit{LN}005}{\mathrm{M}assofNPs}\times 100\%$$

4.19. In vitro drug release

To mimic the acidic environment of osteoarthritis and the release of drugs under normal physiological conditions, the in vitro release profile of ZLN005 from PLGA-ZLN005 was evaluated in PBS at pH 7.0 and 5.4 under gentle shaking at 37 °C [49,50]. Briefly, 1 mL of PLGA-ZLN005 suspension (15 mg/mL) was diluted with 1 mL of PBS at the respective pH values. At predetermined time intervals, 1 mL of the supernatant was collected after centrifugation (10,000 rpm, 10 min) and analyzed using a UV–vis spectrophotometer at a wavelength of 310 nm. An equal volume of fresh PBS was then replenished to maintain a constant total volume.

4.20. Nanoparticles size and zeta potential

For detection of size distribution, nanoparticle tracking analysis (NTA, NanoSight NS300, Malvern) was utilized. The zeta potential of blank NP and antibody-conjugated NP-B2M was characterized using a dynamic light scattering instrument equipped with a zeta potential analyzer (DLS, Malvern Zetasizer Nano ZS90, UK). Prior to detection, all microsphere samples were diluted with endotoxin-free PBS (pH 7.4) to an appropriate concentration to avoid signal overload. Each sample was loaded into a disposable zeta potential capillary cell, and three independent replicate measurements were performed at 25 °C. The average zeta potential value of each group was recorded and analyzed to compare the surface charge changes before and after antibody conjugation.

4.21. Endotoxin quantification

The endotoxin levels of NP-B2M suspensions were quantified using a One-step High-sensitivity Chromogenic LAL Endotoxin Assay Kit (Beyotime, China, C0275S) following the manufacturer's standard protocol. All experimental procedures were performed using endotoxin-free consumables under pyrogen-free conditions to avoid exogenous contamination. The absorbance at 405 nm was measured by a microplate reader. Endotoxin concentrations were calculated from the linear standard curve.

4.22. Confocal microscopy analysis of cellular targeting and internalization

To visualize the cell targeting and internalization ability of the different formulations to ATDC5s and senescent chondrocyte, the NPs were labeled by Nile red. ATDC5s were cultured in 24-well plates and subsequently incubated with NP, NP-B2M, or free Nile red. Following treatment, ATDC5s were fixed with 4% paraformaldehyde for 10 min. The treated cells were washed three times with PBS and stained with DAPI (Servicebio, China, G1012) and FITC-phalloidin (Beyotime, China, C1033), followed by imaging under a confocal microscope. Regular nuclear staining with DAPI was performed to analysis the coronal sections from the injected knee joints, so as to confirmed the retention and targeting of NP-B2M within the joint tissues.

4.23. Statistics and reproducibility

Statistical analysis was performed using GraphPad Prism 9.0 (San Diego, CA) and Origin Pro Version 2025 (OriginLab Corporation, Northampton, MA, USA). Data are represented as mean ± SD from a minimum of three independent experiments. (1) Normality was assessed using the Shapiro–Wilk test; (2) Homogeneity of variance was evaluated using Levene's test; (3) for multiple-group comparisons, one-way ANOVA was performed followed by Tukey's post hoc test; and (4) for two-group comparisons, unpaired two-tailed Student's t-test was applied. Statistical significance analysis was conducted using Excel or Prism software, and ∗ indicate the significance among groups: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Data availability

The datasets for the current study are available from the corresponding author upon reasonable request.

Author contributions

Original draft: Y.C. In vivo experiments: Y.C., S.Z. In vitro experiments: Y.C., A.L. and H.Z. Visualization: Y.C. and Z.S. Conception: K.L. Resource: R.Z., K.L, Y.Z., F.J. and H.L. Supervision: R.Z., H.L. and K.L. Review and editing: R.Z., K.L. and L.W. All authors have read and approved the article.

Ethics approval statements

All animal experiments were performed with protocols approved by the Tongji University Ethnics Committee on the Use and Care of Animals (Approval NO. TJAA13023101). Human cartilage samples were obtained from patients undergoing knee replacement surgery and supernumerary finger specimens were collected from pediatric patients with polydactyly undergoing surgical resection at Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine. The study was approved by the Institutional Review Board of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine (Approval No. SH9H-2021-T401-2). All patients provided written informed consent prior to sample collection. Human samples were anonymized and used strictly for research purposes in accordance with the Declaration of Helsinki.

Declaration of AI and AI-assisted technologies in the writing process

During the preparation of this manuscript, the authors used DeepSeek solely to improve the clarity, fluency, and readability of the language. No scientific ideas, interpretations, data analyses, or conceptual content were generated by AI. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Funding statement

This work is supported by the fundings of Noncommunicable Chronic Diseases‐National Science and Technology Major Project (2024ZD0530900), National Key Clinical Specialty Discipline Construction Project of China (Z155080000004), Shanghai Research Center of Rehabilitation Medicine (Top Priority Research Center of Shanghai) (2023ZZ02027).

Conflict of interest

The authors declare no competing interests.

Appendix A. Supplementary data

The following is the Supplementary data to this article.