Targeted Blockage of Pathological Extracellular Vesicles and Particles From Fibroblast‐Like Synoviocytes for Osteoarthritis Relief: Proteomic Analysis and Cellular Effect

Abstract

Osteoarthritis (OA), the prevalent debilitating joint disorder, is accelerated by dysregulated intercellular crosstalk, yet the role of fibroblast‐like synoviocyte (FLS)‐derived extracellular vesicles and particles (EVPs) in disease progression remains to be elucidated. Here, integrative analysis of clinical specimens, animal models, and publicly available datasets revealed significant alterations in exosomal pathways within OA synovium. Proteomic profiling revealed distinct molecular signatures in EVPs derived from inflammatory and senescent FLSs, reflecting the pathophysiological status of their parent cells. We demonstrated that FLSs under inflammatory and senescent states in OA secreted pathogenic EVPs that propagated joint degeneration by disrupting chondrocyte homeostasis, polarizing macrophages towards a pro‐inflammatory phenotype, and impairing chondrogenesis of mesenchymal stem cells. To therapeutically target these pathogenic EVPs, we engineered an adeno‐associated virus 9 (AAV9) vector fused with a synovium‐affinity peptide (HAP‐1) to deliver shRNA against Rab27a, a key regulator of EVP secretion. Intra‐articular administration of the engineered AAV9 in a murine OA model induced by destabilization of the medical meniscus significantly reduced synovial hyperplasia, cartilage degradation and inflammatory responses, while demonstrating satisfactory systemic biosafety. Our findings establish FLS‐derived EVPs as critical mediators of OA pathogenesis and propose a targeted strategy to block their secretion, offering a promising disease‐modifying therapeutic avenue for OA.

Osteoarthritis (OA) progression involves pathogenic extracellular vesicles and particles (EVPs) from inflamed/senescent fibroblast‐like synoviocytes (FLSs). Proteomic profiling revealed that pathogenic EVPs contribute to OA progression by disrupting the homeostasis of macrophages, chondrocytes and mesenchymal stem cells. Targeted inhibition of FLS‐derived exosomes effectively alleviated OA progression with good biosafety.

Abbreviations

AAV

adeno‐associated virus

ADSC

adipose‐derived mesenchymal stem cell

ALT

alanine aminotransferase

ANOVA

analysis of variance

AST

aspartate aminotransferase

BMSC

bone marrow mesenchymal stem cell

BUN

blood urea nitrogen

CK‐MB

creatine kinase‐MB

CREA

creatinine

DAPI

4′,6‐diamidino‐2‐phenylindole

DAVID

database for annotation, visualization, and integrated discovery

DEG

differentially expressed gene

DMOAD

disease‐modifying osteoarthritis drug

EdU

5‐ethynyl‐2‐deoxyuridine

FLS

fibroblast‐like synoviocyte

GEO

gene expression omnibus

GO

gene ontology

H&E

haematoxylin and eosin

HRP

horseradish peroxidase

IF

immunofluorescence

IHC

immunohistochemical

OA

osteoarthritis

PBS

phosphate‐buffered saline

PCA

principal component analysis

qPCR

quantitative polymerase chain reaction

SDS‐PAGE

sodium dodecyl sulphate‐polyacrylamide gel electrophoresis

SO

Safranin O

TB

toluidine blue

TUNEL

terminal deoxynucleotidyl transferase dUTP nick‐end labelling

1. Introduction

As the most prevalent joint disease, osteoarthritis (OA) is characterized by joint stiffness and pain, leading to impaired mobility and diminished working capacity (Hunter and Bierma‐Zeinstra 2019). Moreover, the prevalence of OA keeps expanding owing to the aging population, imposing a notable socioeconomic burden (GBD 2021 Osteoarthritis Collaborators 2023). Current treatment for OA primarily focuses on pain relief and surgical interventions, with no disease‐modifying osteoarthritis drugs (DMOADs) available, leaving an urgent need for research into the complex etiopathogenesis of OA.

While cartilage degradation is a hallmark of OA, it has been typically associated with marked synovial inflammation and subchondral bone remodelling, demonstrating as progressive degeneration of the whole joint tissue (Sanchez‐Lopez et al. 2022). Studies have shown that synovial inflammation can occur prior to visible cartilage degradation, suggesting that it may precede other pathological changes during OA (Robinson et al. 2016; Liao et al. 2020; Mathiessen and Conaghan 2017; Roemer et al. 2011). According to the clinical trials, including the Multicenter Osteoarthritis Study and the Osteoarthritis Initiative study, synovitis is also an indicator of more intense cartilage loss and more aggressive disease course (Liu et al. 2010; Boutet et al. 2024; Thoenen et al. 2022; Riis et al. 2017). However, the specific molecular mechanisms governing synovitis remain unclear.

To further elucidate the pathogenic mechanisms of synovitis, fibroblast‐like synoviocytes (FLSs), the predominant cell type in the synovial membrane, have garnered our attention. FLSs undergo extensive proliferation and exhibit multiple pathological changes, including inflammation and senescence (Zhang et al. 2018; Chen et al. 2022). Meanwhile, FLSs interact with other joint cells through soluble factors, affecting chondrocyte homeostasis, immune cell infiltration and promoting angiogenesis as well as nerve growth. Single‐cell transcriptomic sequencing‐based analysis has also revealed that osteoarthritic FLSs diversify into a heterogeneous population, including distinct subpopulations exerting regulatory effects on various pathological aspects of OA. This finding further highlights the critical role of FLSs in OA pathogenesis (Knights et al. 2023; Collins et al. 2023; Roelofs et al. 2017).

Beyond the interactions mediated by secretory factors and receptors, extracellular vesicles and particles (EVPs) that transfer bioactive molecules also serve as crucial mediators of intercellular communication in both physiological and pathological conditions, facilitating our understanding of the complexity and multifactorial nature of OA pathology. EVPs constitute a heterogeneous population that includes vesicular subtypes such as exosomes along with non‐vesicular particles like exomeres (Wang et al. 2023). Exosomes secreted from OA cartilage‐derived chondrocytes could spread senescence and inflammation to neighbouring tissues within the joint via transferring encapsulated CX43 and other cargoes (Ni et al. 2019; Varela‐Eirín et al. 2022). Exosomal transfer of miRNAs from osteoclasts to chondrocytes reduces cartilage resistance to matrix degeneration, osteochondral angiogenesis and sensory nerve growth during OA (Liu et al. 2021). Simultaneously, exosomes derived from unmodified or engineered mesenchymal stem cells have shown promise for treating OA by modulating the inflammatory microenvironment, improving chondrocyte anabolic metabolism and promoting cartilage repair (Hanai et al. 2023; Tevlin et al. 2022; Yu et al. 2022; Cao et al. 2024).

Concerning EVPs derived from FLSs, their effects on OA progression vary depending on the state of the original FLSs. Exosomes derived from normal FLSs can alleviate chondrocyte pyroptosis and reduce extracellular matrix degradation by delivering non‐coding RNAs, whereas exosomes from osteoarthritic FLSs conversely promote ferroptosis and apoptosis in chondrocytes (Wang and Sun 2023; Zeng et al. 2022; Kong et al. 2023; Tan et al. 2020). Our previous study has also confirmed that EVPs derived from inflammatory FLSs accelerate the progression of experimental OA via promoting metabolic reprogramming in macrophages (Liu et al. 2024). However, whether inhibiting the pathogenic EVPs from osteoarthritic FLSs could alleviate OA remains to be investigated.

In this study, we found that elevated amounts of EVPs were secreted by OA FLSs in both inflammatory and senescent states, along with altered proteomic profile, which propagated the pathological status of FLSs to other cells within the joint and lead to phenotypes characteristic of OA. We engineered an adeno‐associated virus 9 (AAV9) with capsid protein 2 fused with the synovium‐affinity peptide HAP‐1 for targeted delivering shRNA against Rab27a to FLSs. We demonstrated that intra‐articular injection of the engineered AAV9 inhibited EVP secretion from FLSs and displayed therapeutic effects in the murine model. Additionally, proteomic analysis of EVPs derived from control, inflammatory and senescent FLSs revealed differences in protein composition, confirming that EVPs reflect the disease state of the originated FLSs. In summary, this study systematically demonstrated the pathogenic effects of EVPs from diseased FLSs on various cell types and validated that impairing pathogenic EVP secretion can alleviate OA, suggesting a potential therapeutic approach for clinical OA therapy.

2. Materials and Methods

2.1. Cell Culture

Primary chondrocytes were isolated from the tibial plateau and femoral condyles of neonatal C57BL6/J mice according to previously published protocols (Gosset et al. 2008). Briefly, cartilage was minced into small fragments and digested in DMEM medium (SH30243.01, HyClone) supplemented with 1% penicillin‐streptomycin (BC‐CE‐007, Biochannel) and 3 mg/mL collagenase D (COLLD‐RO, Roche, Merck) at 37°C for 45 min, with this step repeated. The fragments were then subjected to repeated pipetting and washing to remove attached soft tissues. The remaining cartilage pieces were transferred to a new centrifuge tube and incubated overnight at 37°C in 0.5 mg/mL collagenase D solution. After digestion, the suspension was filtered through a 70 µm mesh (352350, Corning), centrifuged, resuspended and seeded for subsequent culture in low‐glucose (1.0 g/L) DMEM medium (319‐010‐CL, WISENT) supplemented with 10% FBS (Fetal Bovine Serum, Prime, F103, Vazyme Biotech Co. Ltd.) and 1% penicillin‐streptomycin. Mouse FLSs were isolated from the synovial tissue of C57BL/6J mice according to the previous publications (Zhao et al. 2016). Briefly, the isolated and cutted synovium was digested in DMEM medium containing 1% type IV collagenase (17104019, Gibco) at 37°C for 1 h, accompanied by vortexed intermittently using a vortex mixer (MX‐S, DLAB Scientific Co. Ltd.). The detached cells were collected and resuspended in fibroblast culture medium (CTCC‐009‐481, Meisen Chinese Tissue Culture Collections) supplemented with 10% FBS and 1% penicillin‐streptomycin. IL‐1β recombinant protein (10 ng/mL, 50101‐MNAE, SinoBiological) and bleomycin (25 µg/mL, T6116, Targetmol, USA) were employed to induce inflammation and senescence in FLSs.

RAW264.7 cells (SNL‐112, SUNNCELL) were cultured in DMEM medium supplemented with 10% FBS (JYK‐FBS‐303, Jin Yuan Kang Biotechnology) and 1% penicillin‐streptomycin. During passaging, cells were detached from the culture dish using phosphate‐buffered saline (PBS, AC08L011, Life‐iLab, Shanghai, China). ATDC5 cells were maintained in the complete culture medium (ZQ0938 and ZM0938, ZQXZ‐bio) and passaged with trypsin‐EDTA. For chondrogenic differentiation experiments, the medium was further supplemented with 100 nM dexamethasone (purchased from MedChemExpress, Monmouth Junction, NJ, USA; Cat# HY‐14648), 1% Insulin‐Transferrin‐Selenium (ITS) supplement (PB180431, Procell), 50 ng/mL ascorbic acid (A4544, Merck) and 10 ng/mL TGF‐β1 (CK33, Novoprotein, Shanghai). The immortalized cells were cryopreserved in cryopreservation tubes (607402, NEST Biotechnology) using serum‐free freezing medium (C40100, New Cell & Molecular Biotech Co. Ltd.) and stored at –80°C for subsequent use.

2.2. Induction of Chondrogenesis for BMSCs and ADSCs

Bone marrow mesenchymal stem cells (BMSCs) were isolated according to the publicly published protocol (Soleimani and Nadri 2009). Briefly, after euthanasia, the femurs and tibias of 6‐week‐old C57BL6J mice were completely harvested and immersed in DMEM medium supplemented with 1% penicillin‐streptomycin (AC03L332, Life‐iLab). Both ends of the bones were cut open, and a syringe was inserted to flush out the bone marrow with DMEM medium. The cell suspension was filtered through a 70 µm mesh to remove debris before adding to cell culture dishes in a 5% CO₂ incubator at 37°C (Thermo Scientific Forma 3111). After 3 h, the non‐adherent cells were removed by replacing them with fresh DMEM culture medium containing 10% FBS, and this step was repeated every 8 h. After 72 h, the medium was changed every 3 days and passaged using the Accutase Cell Detachment Solution (ZYFB003‐0100, ZUNYAN, Nanjing, China).

Adipose tissue‐derived stem cells (ADSCs) were isolated from the inguinal white adipose tissue as described in previous publications (Bai et al. 2016; Taha and Hedayati 2010). The isolated adipose tissue was digested with Krebs‐Ringer Bicarbonate buffer containing 2% bovine serum albumin (BSA) and 2 mg/mL collagenase (collagenase A, Roche) for 20 min at 37°C with gentle agitation. After filtration through 40 µm mesh, isolated cells were suspended in the DMEM culture medium containing 10% FBS, 5 ng/mL basic fibroblast growth factor (C044, Novoprotein, Shanghai), and 2 mM L‐glutamine (A502‐100, BaiDi Biotechnology Co. Ltd. [BDBIO]) before seeding onto the cell culture dishes, with the culture media changed every 3 days.

Both BMSCs and ADSCs were induced for chondrogenic differentiation using commercially available induction media (OriCell MUXMX‐90041 and MUXMD‐90041 from Cyagen Biosciences Inc. Guangzhou). For cells cultured in the plates, when reaching approximately 80%–90% confluency, the DMEM maintenance medium was replaced with chondrogenic differentiation medium. Cells were fixed with 4% paraformaldehyde (PFA, M40958, AbMole, USA) and subsequently subjected to Alcian blue staining (ALCB‐10001, OriCell). For chondrogenesis of pellet culture, 250,000 cells were counted (IC1000, Countstar) and centrifuged at 300 g for 5 min to form a pellet at the bottom of 15 mL polypropylene conical tubes (Estes et al. 2010). The conical tubes were kept at the 5% CO₂ incubator with the tops loosened for gas exchange and then transferred to culture in the low‐adherence spheroids culture plate 96U well kit (EFL‐SP101, Engineering for Life) after forming pellet‐like structures. The chondrogenic medium was changed every other day for 3 weeks.

2.3. Animal Experiments

C57BL/6J mice were purchased from GemPharmatech Co. Ltd. and maintained in a specific pathogen‐free facility under standardized conditions, including free access to food and water, a 12‐h light‐dark cycle, and controlled temperature and humidity. At 12 weeks of age, mice underwent sham surgery or destabilization of the medial meniscus (DMM) on the right knee, following established protocols described in prior publications. The animal experimental protocols were strictly designed in compliance with animal welfare principles and received ethical approval from the Animal Ethics Committee of Nanjing Drum Tower Hospital, the Affiliated Hospital of Nanjing University Medical School. Eight weeks after DMM surgery, knee joint and visceral samples were collected for section preparation, and blood samples were collected for serum biochemical testing.

2.4. Preparation of Paraffin and Frozen Sections

All specimens were sectioned at a thickness of 5 µm and prepared as previously reported (Xian et al. 2025; Wu et al. 2025). Briefly, the right knee joints and major organs were harvested after euthanasia and immediately fixed in 4% PFA for 24 h at room temperature. Knee joint specimens were subsequently decalcified in 0.5 M EDTA solution (diluted using Milli‐Q water, pH 7.4, HY‐Y0682, MedChemExpress) at room temperature for 14 days, with daily solution changes. For paraffin section, the samples were dehydrated through a gradient of alcohol (E809061, Macklin), cleared with xylene (X821391, Macklin) and embedded in paraffin (P434228, Aladdin). For frozen section, the samples were dehydrated in a 30% sucrose solution for 48 h, and embedded in OCT compound (4583, SAKURA).

For clinical samples, the synovium of OA patients who underwent total knee arthroplasty and patients diagnosed without OA who underwent arthroscopy surgeries at the Affiliated Nanjing Drum Tower Hospital of Nanjing University Medical School, with approval by the Ethics Committee of Nanjing Drum Tower Hospital and written informed consent obtained from all patients. The synovium was fixed with 4% PFA as soon as the operation was completed. The subsequent procedure of section preparation was the same as that of mouse tissues described above.

2.5. Chemical Staining for Tissue Sections

For chemical staining, sections were deparaffined and rehydrated. Hematoxylin & Eosin (H&E) staining was performed according to manufacturer's instruction (HY‐K0315, MedChemExpress). Briefly, sections were stained with haematoxylin for 4 min, differentiated in 1% acid ethanol, counterstained with eosin for 1 min, then dehydrated and mounted with resin. For cartilage‐specific histological evaluation, both murine cartilage sections and stem cell pellet sections were processed according to the manufacturer's protocols, including Alcian Blue staining (KGA4102‐50, Keygen BioTECH), Toluidine Blue staining, and Safranin O/Fast Green staining (G1371 and G2543, Beijing Solarbio Science & Technology Co. Ltd.).

2.6. Immunohistochemical (IHC) and Immunofluorescence (IF) Staining

For IHC and multiplex IF staining, paraffin‐embedded sections were deparaffined and rehydrated, while frozen sections were directly immersed in PBS (pH 7.4). Heat‐induced epitope retrieval was performed by incubating tissue sections in 0.5 M sodium citrate buffer (HY‐B1610J, MedChemExpress) at 95°C for 20 min, followed by cooling to room temperature. Endogenous peroxidase activity was quenched by treating sections with 3% hydrogen peroxide for 15 min at room temperature. Non‐specific binding sites were blocked with 10% normal goat serum for 1 h at 37°C. Sections were incubated with primary antibodies diluted in antibody dilution buffer (1% BSA in PBS, HY‐D0842, MedChemExpress) overnight at 4°C and incubated with either horseradish peroxidase (HRP)‐conjugated secondary antibodies (for IHC) or fluorochrome‐conjugated secondary antibodies (for IF) for 1 h at 37°C. For IHC detection, chromogenic development was performed using 3,3′‐diaminobenzidine substrates according to the manufacturer's protocol (abs996, Absin). IF‐stained sections were counterstained with DAPI (HY‐K1047, MedChemExpress) for nuclear visualization. Multiplex immunofluorescence staining was performed according to the manufacturer's instructions of the kit (AFIHC024, AiFang Biological). All sections were mounted with anti‐fade mounting medium and imaged using either brightfield microscopy (Olympus BX53) for IHC or fluorescence microscopy (Leica DMi8) for IF. The primary antibodies utilized were validated by manufacturers or previously publications: Vimentin (#201158) from zen‐bioscience, MMP13 (18165‐1‐AP) from Proteintech, P16 (ab211542) from Abcam, MMP3 (A11418), iNOS (A3774) and COL2A1 (A1560, IF) from ABclonal, IL‐1β (bs‐25615R) from Bioss, COL2A1 (BA0533) from BOSTER (IHC), RAB27A (DF6702) and SOX9 (AF6330) from Affinity Biosciences.

2.7. Isolation and Characterization of EVPs

EVPs were isolated from cell culture supernatants using ultracentrifugation. Following induction of inflammation or senescence for 24 h, the induction medium was completely removed and cells were washed three times with PBS to minimize residual factors. The FLS cells were then cultured in DMEM medium supplemented with 10% exosome‐depleted FBS for 48 h. All media processing was performed immediately post‐collection without storage or freeze to avoid degradation. Supernatants were filtered through a syringe filter and then sequentially centrifuged at 300 × g for 5 min, 2000 × g for 10 min and 10,000 × g for 30 min at 4°C to remove cells and debris, with pellets discarded after each centrifugation. The resulting supernatant was subjected to ultracentrifugation at 100,000 × g for 2 h at 4°C (L‐80 XP, Beckman Coulter) and the final pellet containing EVPs was resuspended in PBS. EVP morphology was examined by transmission electron microscopy and size distribution was analyzed using nanoparticle tracking analysis. Exosomal surface markers were verified by Western blotting following lysis on ice (UR33101, Umibio (Shanghai) Co. Ltd.).

For EVP labelling and uptake studies, isolated EVPs (2 × 10¹⁰ particles/mL in PBS) were incubated with 40 nM MemGlow dye (MG02, Cytoskeleton) at room temperature, protected from light. The reaction was quenched with exosome‐depleted serum‐containing medium after 30 min incubation. Recipient cells (10⁵ cells/confocal dish, 1092000, SAINING Biotechnology) were treated with 10³ EVPs/cell for 6 h at 37°C, washed with ice‐cold PBS three times, fixed with 4% PFA and counterstained with DAPI for imaging.

For other in vitro experiments, EVPs were administered at a concentration of 2 µg/mL (quantified based on protein content) following previous published literature (Liu et al. 2021), with EVPs present throughout the entire process of in vitro culture and differentiation.

2.8. Protein Extraction and Expression Analysis

Total protein was extracted using RIPA lysis buffer supplemented with protease inhibitor cocktail, phosphatase inhibitors and PMSF (HY‐K1001, HY‐K0010 and HY‐B0496, MedChemExpress). Cells were lysed on ice for 15 min with periodic vortexing and detached from the culture plates using a cell scraper (CSC011025, Jet Biofil), followed by centrifugation at 15,000 × g for 15 min at 4°C. The supernatant was collected and mixed with 5× SDS‐PAGE loading buffer (G2527, Beijing LABLEAD Inc.), then denatured at 95°C for 10 min. After measuring protein concentrations by BCA assay (KGB2101‐500, KeyGEN BioTECH), 20 µg of exosomal protein and 10 µg of cellular protein were loaded per lane of the SDS‐PAGE gels (M00668 from GenScript, PG112 from Epizyme). Equal amounts of protein samples were separated by electrophoresis. Coomassie Brilliant Blue staining solution (AP11L014, Life‐iLab) was used for staining proteins on SDS‐PAGE gels. For Western blot analysis, protein was transferred onto PVDF membranes (IPVH00010, Millipore), followed by blocking with a commercial blocking buffer (AP36L108, Life‐iLab) for 1 h at room temperature. Membranes were incubated with primary antibodies overnight at 4°C and then incubated with HRP‐conjugated secondary antibodies (8018021, Dakewe) for 1 h at room temperature. The primary antibodies utilized were validated by manufacturers or previously publications and diluted at 1:1000 using a universal dilution buffer (E182‐05, GenStar, Beijing, China): anti‐P16 (ab211542, Abcam), anti‐Phospho‐Histone H2A.X (γH2AX, ET1602‐2, HUABIO), anti‐COL10A1(RM5026, Biodragon), anti‐CD9 (CY5337, Abways), anti‐CD81 (ET1611‐87, HUABIO), anti‐β‐actin, anti‐CD63 and anti‐Calnexin (AC026, A19023 and A4846, Abclonal). Protein bands were visualized using enhanced chemiluminescence substrate (KF8001, Affinity). For ELISA assays, cells were seeded in insert culture dishes (TCS001006, Jet Biofil), and the conditioned medium from the lower chamber of the culture plate was subsequently collected to measure cytokine concentrations. The concentrations of IL‐6 (RK04845, Abclonal) and TNF‐α (JL10484, Jianglai biology, Shanghai) in cell culture supernatants were quantified using commercial ELISA kits according to the manufacturer's protocols.

2.9. RNA Extraction and RT‐qPCR Analysis

Total RNA was extracted from cultured cells using the RNApure Fast Tissue&Cell Kit according to the manufacturer's instructions (CW0599S, CWBIO, China). For reverse transcription, 1 µg of total RNA was converted to cDNA using the Evo M‐MLV Plus 1st Strand cDNA Synthesis Kit (AG11615, ACCURATE BIOTECHNOLOGY (HUNAN) CO.LTD, ChangSha, China). Quantitative PCR (qPCR) was performed using the 2× BioZues HS Taq Universal SYBR Green qPCR Master Mix (SM0101, BiOligo Biotechnology (Shanghai) Co. Ltd., Shanghai, China) on a real‐time PCR system (QS5, Applied Biosystem). Relative gene expression was calculated using Actb as the internal reference gene.

2.10. Staining of Cultured Cells

Cellular senescence was determined using the β‐Galactosidase staining kit following the manufacturer's protocol. Cells were fixed with 2% PFA for 5 min at room temperature and washed using PBS buffer (AC08L011, Life‐iLab) three times, then incubated with β‐galactosidase staining solution (KTA3030, Abbkine) at 37°C for 24 h. Senescent cells (blue staining) were imaged under bright‐field microscopy.

EdU (5‐ethynyl‐2‐deoxyuridine) staining and terminal deoxynucleotidyl transferase (TdT) dUTP nick‐end labelling (TUNEL) assay were utilized for measuring proliferation and apoptosis. For EdU staining, the EdU working solution was added to the ATDC5 culture media 2 h prior to PFA fixation, and then the reaction solution was added to perform a click chemistry reaction to link the fluorescent dye using the EdU imaging kit (K1075, Apexbio). For TUNEL staining, the fixed cells were permeabilized with 0.1% Triton X‐100 for 10 min at room temperature and incubated with TUNEL reaction mixture (E‐CK‐A320, Elabscience) for 1 h at 37°C in the dark. Nuclei were stained with DAPI and images were captured using a fluorescence microscope with consistent exposure settings.

2.11. Lentiviral‐Mediated Gene Knockdown

Lentiviral particles encoding short hairpin RNA (shRNA) targeting Rab27a and non‐targeting control shRNA were generated using the pSLenti‐U6‐shRNA‐CMV‐EGFP‐F2A‐Puro‐WPRE vector (OBiO Technology, Shanghai) after validating knockdown efficiency using small interfering RNA (synthesized by Shanghai Generay Biotech Co. Ltd. and transfected using Hieff Trans in vitro siRNA/miRNA Transfection Reagent, 40806, Yeasen Biotechnology). Mouse FLSs were transduced with lentivirus at a multiplicity of infection of 50 in the presence of 8 µg/mL polybrene (SJ‐MB0061, Sparkjade Biotechnology Co. Ltd., Shandong). After 24 h, the medium was replaced with fresh complete medium containing 5 µg/mL puromycin (HY‐B1743, MedChemExpress) for 2 days to eliminate the uninfected cells before further culture.

2.12. Construction of Synovium‐Targeting Adeno‐Associated Virus Vector

To construct the synovium‐targeting AAV9 vector, the DNA sequence encoding the synovium‐targeting peptide motif HAP‐1 was codon‐optimized. First, the start codon of VP2 in the pAAV9 plasmid was mutated to generate a plasmid that exclusively expresses VP1 and VP3. In a second plasmid, the HAP‐1 sequence was fused to the N‐terminus of the AAV9‐VP2 open reading frame (ORF). A Kozak sequence and an ATG start codon were inserted immediately upstream of the HAP‐1 sequence to ensure optimal expression under the control of the CMV promoter. These two plasmids were then employed in the production of synovium‐targeting adeno‐associated virus vector. Intra‐articular injections were performed with 10 µL of AAV (1 × 10¹¹ genome copies) per joint, following our established protocol (Chen et al. 2022). The first treatment was administered 2 weeks after DMM surgery, followed by a booster injection at 4 weeks post‐surgery.

2.13. EVP Proteomics Analysis

Frozen samples of EVPs were thawed on ice and lysed with a pre‐mixed lysis buffer (8 M urea, 1% protease inhibitor cocktail) via sonication. Cellular debris was removed by centrifugation (4°C, 15,000 × g, 10 min). The supernatant was transferred to new tubes, and protein concentration was quantified using the BCA assay.

Equal protein aliquots were adjusted to uniform volume with lysis buffer, precipitated with 20% (v/v) trichloroacetic acid (4°C, 2 h), and centrifuged (4°C, 15,000 × g, 5 min). Pellets were washed thrice with ice‐cold acetone, air‐dried, and resuspended in 100 mM triethylammonium bicarbonate via sonication. Proteins were then digested with trypsin (1:50 enzyme‐to‐protein ratio, m/m) at 37°C overnight, followed by reduction with 5 mM dithiothreitol (56°C, 30 min) and alkylation with 15 mM iodoacetamide (RT, 15 min, dark). Acidification to pH 2–3 with 10% trifluoroacetic acid preceded desalting using Strata‐X columns (Phenomenex). Peptides were lyophilized and quantified via Pierce Quantitative Peptide Assay (Thermo Scientific).

Peptides were separated using a nanoflow Vanquish Neo system (Thermo Scientific) equipped with a µPAC Neo High Throughput Column (5.5 cm × 1850 µm; silicon micro‐pillar array with C18 phase; 5 µm pillar diameter, 2 µm interpillar distance). Eluents were analyzed on an Astral mass spectrometer (Thermo Scientific) in DIA mode with following parameters. MS1 settings: 380–980 m/z scan range, 240,000 resolutions at m/z 200, normalized AGC target 500%, maximum injection time 5 ms. MS2 settings: 299 variable windows (2 m/z isolation), 25 eV HCD collision energy, normalized AGC target 500%, maximum injection time 3 ms. DIA data were analyzed using DIA‐NN (v1.8.1) with parameters: trypsin/P specificity, max missed cleavages = 1, fixed modification: carbamidomethylation, variable modifications: oxidation, N‐terminal acetylation, and 1% protein FDR threshold.

2.14. Transcriptomic Analysis of Public Data

Transcriptomic sequencing data was downloaded from the Gene Expression Omnibus (GEO) database, including: GSE89408, GSE293276, GSE114007, GSE100748 and GSE61298. Differentially expressed genes (DEGs) were identified using the GEO2R tool with the following criteria: fold change >2 or <0.5, and adjusted p value < 0.05. Subsequently, enrichment analysis was performed using the Database for Annotation, Visualization and Integrated Discovery (DAVID) database and Metascape database. The online platform Hiplot and OmicStudio were utilized for data visualization.

For screening genes associated with cartilage degeneration, MSC differentiation and macrophage polarization, a Venn diagram was employed to identify common differentially expressed genes/proteins between the proteomic data of EVP lysis and transcriptomic data from public databases. Subsequently, a protein‐protein interaction network was constructed using the STRING database. The network was then imported into CytoScape software, where the cytoHubba plugin was applied in Maximal Clique Centrality (MCC) mode to screen for core hub genes.

2.15. Statistical Analysis

All experimental data derived from at least three independent replicates were presented as mean ± standard deviation and were statistically analyzed using GraphPad Prism software. For comparisons between two groups, an unpaired Student's t‐test was employed. For experiments comprising multiple treatment, analysis of variance (ANOVA) was performed, followed by post hoc test to evaluate differences among groups.

3. Results

3.1. Increased EVP Secretion Is a Typical Pattern of OA FLSs

To investigate the pathological changes in OA synovial tissue, we obtained transcriptome data of both normal and osteoarthritic synovial samples from the Gene Expression Omnibus (GSE89408 and GSE293276). Gene Ontology (GO) cellular component enrichment analysis of differentially expressed genes revealed significant alterations in the GO term (GO:0070062) extracellular exosomes (Figure 1A). We demonstrated that RAB27A, as a key regulator of EVP release, was extensively upregulated in OA synovium, along with elevated expression of inflammation and senescence markers (Figure 1B). Interestingly, we revealed that the expression of RAB27A was positively correlated with inflammatory and senescence‐related genes expression (Figure 1C), which indicated enhanced EVP secretion of the synovium in OA pathogenesis.

FIGURE 1.

FLS pathology was accompanied by increased EVP secretion in OA. (A) Gene ontology (GO) cellular component enrichment analysis of differentially expressed genes between normal and OA synovium. (B, C) Violin plots for the expression levels of the exosome secretion marker (RAB27A), senescence markers (CDKN1A, CDKN2A) and inflammation‐related markers (PTGS2, IL6, MMP13), along with Pearson correlation analyses between the expression levels of these markers and RAB27A. Data was normalized as fragments per kilobase of exon model per million mapped fragments (FPKM). (D–G) Multiplex immunohistochemical (mIHC) staining of normal and OA synovium, with magnified views of the boxed areas showing individual colour channels. Scale bar = 100 µm. Line graphs display the relative intensity and co‐localization of each fluorescence along the line in the magnified images. (H, I) Representative RAB27A fluorescence staining images and quantification in the synovial region of sham‐operated mice and mice at 4 and 8 weeks post‐DMM surgery. Scale bar = 25 µm. (J) Schematic diagram of in vitro EVP collection. Briefly, FLSs were pre‐treated with IL‐1β (10 ng/mL) and Bleomycin (25 µg/mL) for 24 h and changed into fresh medium without inducers. EVPs were then isolated from the conditioned medium through ultracentrifugation. (K, L*) Representative MMP13 fluorescence staining and SA‐β‐gal staining images for FLSs after inflammation and senescence induction. (M) Diameter distribution of EVPs from control and pathological FLSs by NTA, with screenshots of the particle flow. (N, O) Representative TEM images for EVPs and Western blot gels for EVP protein markers. Scale bar = 200 µm. (P, Q) Quantification and protein concentration of EVPs isolated from different culture medium, both of which were normalized to original cell counts. ** Indicates p < 0.01, * indicates p < 0.05, ns indicates p > 0.05, versus the indicated groups, one‐way ANOVA.

Considering that FLSs are the predominant cell type in both healthy and OA synovium, we performed multiplex immunofluorescence staining and observed high expression of RAB27A in both inflammatory and senescent FLSs (Figure 1D‐G). Similarly, increased expression of RAB27A was observed in the synovium of OA mice (Figure 1H,I). To directly characterize the secretion of EVPs from FLSs under pathological conditions, we induced inflammation and senescence phenotype of FLSs in vitro, respectively. EVPs were then isolated from the conditioned medium using ultracentrifugation, which was named as Ctr‐EVP, Inf‐EVP or Sen‐EVP separately (Figure 1J‐L). All the three types of EVPs exhibited typical size and morphology (Figure 1M,N). Western blot analysis showed that EVPs were enriched with positive markers CD9, CD63 and CD81, and lacked the negative marker Calnexin, confirming the relative purity of the isolation method (Figure 1O). Nanoparticle Tracking Analysis (NTA) analysis revealed an elevated quantity of EVPs derived from inflammatory and senescent FLSs, accompanied by increased protein abundance in EVPs (Figure 1P,Q). Taken together, these results suggested that pathological FLSs may regulate OA progression through increased EVP release.

3.2. EVPs Derived From Inflammatory and Senescent FLSs Exhibit Distinct Proteomic Profile

To elucidate the molecular mechanisms of FLS‐derived pathogenic EVPs on regulating OA development, EVPs isolated from the supernatant of FLS were subjected to proteomic analysis (Figure 2A). SDS‐PAGE analysis revealed distinct molecular weight distribution characteristics of EVPs vary among three groups (Figure 2B). According to mass spectrometry‐based proteomic analysis, the number of identified proteins in Inf‐EVP and Sen‐EVP was higher than in Ctr‐EVP, with the most abundant variety of proteins detected in Sen‐EVP (Figure 2C). Principal Component Analysis (PCA) and Pearson correlation coefficient heatmaps showed good internal correlation within each group, while there were significant differences between the groups (Figure 2D,E). Using a fold change threshold of >2 or <0.5 and p value < 0.01, numerous differentially expressed proteins between groups were identified between Inf‐EVP and Sen‐EVP compared to Ctr‐EVP (Figures 2F,G and S1A–D). Based on enrichment analysis using Metascape, the commonly enriched pathways included cell‐environment interactions (including adhesion, migration and extracellular matrix) and inflammatory processes (neutrophil degranulation, protein degradation, proteasome and IL17A signalling), which may play a critical role in OA pathogenesis and contribute to the spread of inflammation and senescence (Figure 2H). According to the gene set enrichment analyses, proteomics of Inf‐EVP indicated upregulation in inflammation‐related pathways and showed downregulation of oxidative phosphorylation and the TCA cycle, indicating a shift toward pro‐inflammatory signalling at the expense of energy metabolism (Figure 2I). Meanwhile, the differentially‐expressed proteins between Sen‐EVP and Ctr‐EVP were mainly related to the disturbance of cell cycle and DNA repair, which were critical patterns in senescent FLSs (Figure 2J). All of these proteomic analyses highlighted the context‐dependent nature of exosomal protein cargo and its potential role in mediating pathogenic intercellular communication.

FIGURE 2.

Proteomic profiling of pathogenic EVPs reflected the pathological changes of the source cells. (A) Schematic diagram for proteomics analysis of EVPs isolated from FLSs after inducing inflammatory and senescent phenotypes, created with Figdraw (www.figdraw.com). (B) SDS‐PAGE gel electrophoresis images of proteins lysed from abovementioned EVPs. (C) Number of proteins identified by mass spectrometry in EVPs secreted from control FLSs and FLSs induced with inflammation and senescence. (D, E) Principal component analysis plot and Pearson's Correlation Coefficient heatmap of the protein composition in EVPs from three groups (n = 3 samples per group). The gradient colours and annotated values represent the Pearson correlation coefficients. (F) Number of differentially expressed proteins in EVPs between the three groups, with screening criteria set at an adjusted p value < 0.01 and a fold change >2 or <0.5. (G) Circos plot visualization of the overlaps among significantly altered proteins that overlap in Inf‐EVP and Sen‐EVP, with lines connecting the commonly altered proteins. (H) Pathway heatmaps of significantly differentially expressed proteins in Inf‐EVP and Sen‐EVP compared to Ctr‐EVP, enriched using Metascape. (I, J) Representative gene set enrichment analysis of proteins in Inf‐EVP or Sen‐EVP relative to those in Ctr‐EVP.

3.3. Exosomal Proteomics Suggests That Pathogenic EVPs May Regulate Inflammation and Cartilage Homeostasis

Gene ontology enrichment analyses were performed on differentially expressed proteins in Sen‐EVP and Inf‐EVP compared to Ctr‐EVP (fold change >2 or <0.5, p value < 0.01) to further interpret the proteomic characters. The results demonstrated significant enrichment of pathways related to cartilage development and collagen metabolism (Figure 3A). Concurrently, multiple pathways associated with inflammatory response and inflammatory cell chemotaxis were also significantly enriched, suggesting that differentially expressed proteins in EVPs might regulate both cartilage metabolism and inflammation (Figure 3B). To further elucidate the association between exosomal proteins and OA pathology, public sequencing data (OA cartilage, chondrogenic MSCs, polarized macrophages) were analyzed to identify differentially expressed genes. These gene lists were intersected with differentially expressed exosomal proteins and protein‐protein interaction networks were constructed, revealing that differentially expressed proteins in pathogenic EVPs linked to OA cartilage degradation, MSC chondrogenesis and macrophage M1 polarization (Figure 3C–E).

FIGURE 3.

Inf‐EVP and Sen‐EVP contributed to OA progression through multifaceted mechanisms. (A, B) Bubble plots of GO enrichment analysis for differentially expressed proteins in Sen‐EVP and Inf‐EVP compared to Ctr‐EVP, highlighting inflammation‐ and cartilage‐related pathways. (C–E) Venn diagrams showing the overlap between differentially expressed proteins in Inf‐EVP and Sen‐EVP (vs. Ctr‐EVP) and genes associated with (C) OA cartilage degeneration, (D) chondrogenic differentiation of mesenchymal stem cells and (E) macrophage M1 polarization. Protein‐protein interaction networks were constructed using the top 20 hub proteins ranked by the MCC algorithm of the CytoHubba plugin in Cytoscape.

3.4. Pathogenic FLS EVPs Disrupt the Homeostasis of Chondrocytes and Macrophages

To elucidate the effects of different EVPs on chondrocyte homeostasis, three types of EVPs were labelled with fluorescent dye and added to the culture medium of primary chondrocytes. Our results revealed that all the three types of EVPs were internalized by the chondrocytes (Figure 4A). We observed that the SA‐β‐gal‐positive chondrocytes were profoundly increased with the treatment of Inf‐EVP and Sen‐EVP, as well as increased protein level of P16 and γ‐H2AX, which indicated that pathogenic FLS EVPs could contribute to the senescence of chondrocytes (Figure 4B–D). Additionally, compared to the chondrocytes stimulated with Ctr‐EVP, the expression of anabolic genes (Sox9 and Col2a1) was significantly decreased in chondrocytes treated with Inf‐EVP and Sen‐EVP, along with elevated levels of the matrix‐degrading enzyme Mmp13 (Figure 4E, F and I). In parallel, we further demonstrated that both Inf‐EVP and Sen‐EVP could result in decreased proliferation and increased apoptosis of chondrocytes, as evaluated by the TUNEL and EdU staining (Figure 4G, H, J and K).

FIGURE 4.

Pathogenic FLS EVPs disrupt chondrocyte and macrophages homeostasis in vitro. (A) Representative fluorescence for the internalization of EVPs by mouse chondrocytes. Scale bar = 20 µm. The red fluorescence represents EVPs labelled with the MemGlow fluorescent dye, and the blue fluorescence represents nuclei stained with DAPI. (B, C) Representative images and quantification of SA‐β‐gal staining for mouse chondrocytes after co‐culturing with different EVPs for 48 h, blue arrow indicting the SA‐β‐gal positive cells. Scale bar = 200 µm. (D) Representative Western blot images showing the senescence‐associated markers P16 and γ‐H2AX in chondrocytes treated with three types of EVPs for 48 h. (E) mRNA expression for OA‐related genes of mouse chondrocytes after co‐culturing with different EVPs for 24 h. (F–K) Representative images and quantification of COL2A1 and TUNEL staining for mouse chondrocytes, as well as EdU staining for ATDC5 cell line after stimulation with different EVPs for 48 h. Scale bar = 200 µm. (L) Internalization of EVPs by RAW264.7 macrophages. Scale bar = 200 µm. (M) mRNA expression for senescence marker (Cdkn1a), M1 polarization‐related genes (Il6, Tnf, Nos2 and Ptgs2), and M2 polarization‐related genes (Arg1, Cd163 and Cd206) of RAW264.7 macrophages after co‐culturing with different EVPs for 24 h. (N–Q) Representative images and quantification of iNOS and P16 fluorescence staining for RAW264.7 macrophages after stimulation with different EVPs for 48 h. Scale bar = 50 µm. (R, S) Concentration of TNF‐α and IL‐6 in the cell culture supernatant of EVP‐stimulated RAW264.7 macrophages. ** Indicates p < 0.01, * indicates p < 0.05, ns indicates p > 0.05, versus the indicated groups, one‐way ANOVA.

To further investigate the role of pathogenic FLS EVPs on macrophage function, we found that macrophages could also internalize EVPs effectively (Figure 4L). In addition, we identified that Inf‐EVP and Sen‐EVP treatment significantly elevated the expression of senescence‐associated gene (Cdkn1a), accompanied with increased levels of M1 polarization and inflammatory genes (Il6, Tnf, Nos2 and Ptgs2) in macrophages. However, there was no significant difference in the expression of M2 polarization‐related genes (Arg1, Cd163 and Cd206) among macrophages treated with the three types of EVPs (Figure 4M). The immunofluorescence staining of the macrophages revealed that the proportion of iNOS‐positive (M1‐polarized) macrophages was highest in the Inf‐EVP‐treated group, while the proportion of P16‐positive cells was highest in the Sen‐EVP‐treated group (Figure 4N–Q). In addition, we observed an enhanced secretion of inflammatory factors including TNF‐α and IL‐6 in the macrophages treated with Inf‐EVP and Sen‐EVP (Figure 4R,S). These results indicated that pathogenic FLS EVPs could induce the senescence and M1 polarization of macrophages in OA progression.

3.5. Pathogenic FLS EVPs Impede the Chondrogenic Differentiation of Mesenchymal Stem Cells

Mesenchymal stem cells (MSCs) possess multipotent differentiation potential and ameliorate the inflammatory microenvironment, making them a promising choice as seed cells for cartilage repair (Yu et al. 2022). Thus, we further evaluated whether EVP from pathological FLSs could impact the chondrogenic differentiation of MSCs (Figure 5A). Bone marrow mesenchymal stem cells (BMSCs) were isolated from mouse bone marrow, which could internalize FLS EVPs in vitro (Figure 5B). The expression of chondrogenic differentiation markers (Sox9, Col2a1, Acan) was significantly suppressed in the chondrogenic differentiated BMSCs treated with Inf‐EVP and Sen‐EVP, consistent with the results of Alcian Blue staining (Figure 5C,E). However, the expression level of COL10A1, a marker of chondrocyte hypertrophy, was significantly increased, indicating an impairment in chondrogenic differentiation (Figure 5D). Furthermore, the proportion of SA‐β‐gal‐positive BMSCs was increased in the Inf‐EVP and Sen‐EVP treated groups, indicating exacerbated cellular senescence (Figure 5F). To further confirm this finding, BMSCs were induce to form pellets for 3D culture in the presence of EVPs (Figure 5G). In the Ctr‐EVP treated group, the pellets exhibited a well‐organized structure, with high expression of the SOX9 transcription factor and extracellular matrix components that could be stained with safranin O (SO), alcian blue (AB) and toluidine blue (TB). However, in the Inf‐EVP and Sen‐EVP treated groups, the chondrogenesis‐related staining was diminished, and the structure was disorganized, along with decreased SOX9 expression (Figure 5G–I). Additionally, we isolated adipose‐derived mesenchymal stem cells (ADSCs) from the inguinal white adipose tissue (iWAT) of mice and confirmed their uptake of the EVPs (Figure 5J,K). Both Inf‐EVP and Sen‐EVP treatments led to an increase in SA‐β‐gal‐positive cells and a reduction in alcian blue staining (Figure 5L,M). In addition, the SOX9 fluorescence staining of the ADSC‐derived pellets was reduced in the Inf‐EVP and Sen‐EVP treated groups compared to the Ctr‐EVP group after 21‐day chondrogenic induction, which was consistent with the results observed in BMSCs (Figure 5N). These results indicated that pathogenic FLS EVPs may hinder the potential of utilizing stem cells for cartilage restoration in OA progression.

FIGURE 5.

Pathogenic FLS EVPs impair chondrogenic differentiation of mesenchymal stem cells. (A) Schematic diagram of EVP stimulation on mouse BMSCs isolated from the femoral bone marrow cavity of mice. (B) Internalization of EVPs by BMSCs. Scale bar = 50 µm. (C) Expression of chondrogenic differentiation‐related genes in chondrogenesis‐induced BMSCs after 7 days of treatment with different EVPs. (D) Representative Western blot images of COL10A1, the marker of chondrocyte hypertrophy. (E, F) Representative images of alcian blue staining and SA‐β‐gal staining in BMSCs after 7 days of stimulation with different EVPs. Scale bar = 1 mm and 200 µm separately. (G) Schematic diagram of section staining observation after inducing BMSCs to form chondrocyte pellets for 21 days while simultaneously stimulating with different EVPs. (H, I) Representative images of SOX9 fluorescence staining, safranin O (SO) staining, alcian blue (AB) staining and toluidine blue (TB) staining of BMSC‐differentiated chondrocyte pellet sections. Scale bar = 50 µm. (J) Schematic diagram of EVP treatment on mouse ADSCs isolated from the iWAT of mice. (K) Internalization of EVPs by ADSCs. Scale bar = 50 µm. (L, M) Representative images of SA‐β‐gal staining and alcian blue staining of ADSCs after 7 days of chondrogenic induction and treatment with different EVPs. (N) Representative images of SOX9 staining in sections of chondrocyte pellets formed by ADSC after 21‐day induction.

3.6. HAP‐1 Peptide‐Expressing AAV Targets FLS for Rab27a Inhibition to Reduce EVP Secretion

Considering the impact of pathogenic FLS EVPs on OA progression, we next investigated whether targeted intervention of EVP secretion could serve as a potential strategy for OA therapy. The small GTPase family member RAB27A is a key regulator of multivesicular body fusion with the plasma membrane for EVP releasing (Ostrowski et al. 2010). We constructed shRNA to knock down Rab27a and observed a significant reduction in the number of EVPs, while the diameter distribution and surface marker expression of the EVPs remained unchanged (Figures 6A–D and S2A). In addition, knockdown of Rab27a using shRNA did not exacerbate inflammatory and senescent phenotypes in FLS as evidenced by qPCR, demonstrating the safety of shRNA‐mediated Rab27a inhibition in reducing EVP secretion (Figure S2B–D).

FIGURE 6.

Intra‐articular injection of FLS‐targeting AAV for delivering Rab27a‐shRNA to specifically reduce EVP secretion. (A) Rab27a expression levels in FLSs after transfection with control shRNA and Rab27a knockdown shRNA. (B, C) EVP size distribution and quantification after transfection with control shRNA and Rab27a knockdown shRNA. ** Indicates p < 0.01, versus the indicated groups, student’s t‐test. (D) Representative Western blot images for positive and negative surface markers of EVPs at equal concentration after shRNA transfection. (E) Schematic diagram of constructing a virus targeting FLS to inhibit Rab27a for intra‐articular injection. The synovium‐affinity peptide HAP‐1 is fused with the AAV9 viral capsid viral protein 2 (VP2), and the AAV9 vector is designed to simultaneously carry the gene encoding the mScarlet fluorescent protein and an expression cassette for shRNA. (F) Representative images of mScarlet fluorescence in the synovium and cartilage regions of mouse knee joints. Scale bar = 200 µm and 100 µm separately. (G) Representative images of mScarlet fluorescence in sections of multiple mouse organs. Scale bar = 200 µm. (H) Representative images and quantitative results of RAB27A expression in the synovial region after intra‐articular AAV injection. ** Indicates p < 0.01, versus the indicated groups, two‐way ANOVA.

We engineered an AAV9 capsid protein to express the synovium‐affinity peptide HAP‐1 for targeted inhibition of Rab27a (Figure 6E). Eight weeks after DMM or sham operation, the mScarlet fluorescence was primarily localized to the synovial region of the joint, with few red fluorescence observed in the cartilage area and only a few mScarlet‐positive cells detected in the liver (Figure 6F,G). Additionally, IF staining showed that the AAV virus effectively downregulated the abnormally elevated RAB27A levels in the synovial region following DMM surgery (Figure 6H). These results demonstrated that HAP‐1 peptide‐modified AAV could targeted suppress RAB27A expression in OA synovium.

3.7. Inhibition of Pathogenic FLS EVPs Alleviates Murine OA and Demonstrates Good Biosafety

To evaluate the therapeutic effects of targeted inhibition of FLS‐derived EVPs in OA mice, histological analyses were performed at 8 weeks post DMM operation. Safranin O and fast green (SO), haematoxylin and eosin (HE) and toluidine blue (TB) staining showed that inhibiting EVP secretion of FLSs by shRab27a effectively ameliorated cartilage wear and structural degeneration. Immunostaining for COL2A1 and MMP13 indicated that cartilage anabolism was partially restored, and the hyperactive catabolic processes were alleviated (Figure 7A, C and E). Inhibition of FLS‐derived EVPs also led to a reduction in synovial hyperplasia in DMM mice, as well as decrease in the expression of M1 macrophage marker iNOS, the inflammatory cytokine IL‐1β and the matrix‐degrading enzyme MMP13 in the synovial region (Figure 7B, D and E). To confirm the safety of intra‐articular injection of the knockdown virus, serum from four groups of mice were collected for analysis. The liver function markers (ALT, AST), kidney function markers (BUN, CREA) and cardiac injury marker (CK‐MB) showed no significant differences among the groups, indicating no systemic toxicity on major organs (Figure 7F). Our results suggested that targeted inhibition of RAB27A in FLSs could effectively suppress OA progression via attenuating cartilage degradation and synovial inflammation.

FIGURE 7.

Inhibition of FLS EVPs Alleviated OA in a Murine Model Without Toxicity. (A) Representative images of SO staining, HE staining, TB staining, COL2A1 immunohistochemistry staining and MMP13 immunofluorescence staining in the articular cartilage region after DMM operation and AAV virus injection. (B) Representative images of HE staining, and iNOS, IL‐1β and MMP13 immunofluorescence staining in the synovial region after DMM operation and AAV virus injection. (C–E) OARSI scores, synovitis scores and quantitative heatmaps of the immunostaining shown in (A) and (B), with six mice per group. ** Indicates p < 0.01, versus the indicated groups, two‐way ANOVA. (F) Levels of liver (ALT, AST), kidney (BUN, CREA) and cardiac injury (CK‐MB) markers in the serum of mice from each group after DMM operation and AAV injection (n = 6 per group). ns indicates p > 0.05, two‐way ANOVA.

4. Discussion

OA is characterized by the progressive deterioration of the whole joint tissue, highlighting the critical role of intercellular crosstalk in unravelling the complex pathogenesis. In this study, we systematically investigated the pathogenic roles of FLS‐derived EVPs on multiple intra‐articular cell populations. Our findings demonstrated that EVPs derived from senescent and inflammatory FLSs disrupted chondrocyte homeostasis and promoted macrophage polarization towards a pro‐inflammatory phenotype, contributing to the pathological microenvironment in OA. These observations are consistent with previous studies highlighting the role of EVP‐mediated spread of senescence and inflammation under multiple disease circumstances (Hou et al. 2024; Wozniak et al. 2020; Yanagawa et al. 2024; Mensà et al. 2020), which emphasizes the value of targeting EVPs for disease intervention.

MSCs have been widely explored as potential therapeutic agents for cartilage regeneration due to their potential for both chondrogenesis and immunomodulation (Copp et al. 2023). However, the efficacy of MSC‐based therapies is often limited by poor differentiation efficiency and functional instability in OA microenvironments (Yu et al. 2022; Boulestreau et al. 2024). Our results revealed that Inf‐EVP and Sen‐EVP impaired chondrogenesis and extracellular matrix synthesis of BMSC and ADSC in both 2D and 3D culture models. These findings indicate that combinatorial approaches that address synovial pathology or directly eliminate pathogenic EVPs may be necessary to optimize MSC‐based regenerative therapies.

Given the critical role of EVP‐mediated intercellular communication dysregulation in the pathogenesis of various diseases, directly blockage of EVPs or combination with existing therapies offers a promising therapeutic strategy for cancer, inflammatory diseases and viral infections (Kim et al. 2022). For instance, inhibition of EVP secretion can be achieved through small‐molecule inhibitors including GW4869 and genetic knockdown of critical regulators like RAB27A, both of which have been reported to enhance immunotherapy responsiveness (Yang et al. 2018; Poggio et al. 2019). In this study, we developed a targeted therapeutic strategy using an adeno‐associated virus (AAV)‐delivered shRNA system to inhibit RAB27A, a key regulator of EVP secretion. This approach significantly attenuated cartilage degradation in OA models while demonstrating a favourable biosafety profile, highlighting its promise as a disease‐modifying intervention for OA. Besides, numerous studies have investigated the regulatory mechanisms of RAB27A expression at transcriptional, translational and post‐translational modification levels. For instance, IRF‐1 mediates RAB27A upregulation through specific binding to its promoter (Yang et al. 2018), METTL3 promotes RAB27A translation via the m⁶A‐YTHDF1 axis (Li et al. 2025), and KIBRA directly interacts with RAB27A to prevent its ubiquitination‐mediated degradation (Song et al. 2019), thereby enhancing EVP secretion. However, the regulatory mechanisms governing RAB27A expression in FLS under inflammatory and senescent conditions during OA pathogenesis remain to be further elucidated.

Exosomal cargo is highly dynamic and reflects the functional and pathological state of the parent cell (Lee et al. 2024). Proteomic profiling of Inf‐EVP and Sen‐EVP revealed distinct molecular signatures in both the quantity and composition of exosomal proteins. Inf‐EVP exhibited dysregulation of proteins involved in inflammation signalling and repressed activity of the tricarboxylic acid cycle, while Sen‐EVP showed alterations in DNA repair and cell cycle. The analysis of exosomal proteomics not only aids in our understanding of the mechanisms by which Inf‐EVP and Sen‐EVP disrupt the homeostasis of other joint cells, but also suggests that EVPs may serve as indicators of synovial pathology by means of methods like liquid biopsy, paving the way for advancing both early diagnosis accuracy and personalized interventions of OA (Yu et al. 2022). Moreover, the cargo within extracellular vesicles also includes nucleic acids (such as mRNA, miRNA and lncRNA), metabolites (including amino acids and lipids) (Zhang et al. 2023; Wu et al. 2022). Together, these components constitute a complex intercellular communication network mediated by EVPs. This complexity makes therapeutic strategies targeting the inhibition of EVP secretion, rather than focusing on a specific cargo, a more rational approach.

Despite these conclusions, several limitations of this study warrant further investigation. First, the molecular mechanisms driving EVP overproduction in OA FLSs remain to be explored. Second, while proteomic analysis provided valuable insights, a comprehensive understanding of EVP cargo requires multi‐omic characterization, including miRNA and lipidomic profiling. Third, emerging single‐vesicle sequencing technologies could identify the heterogeneity of EVP subpopulations, enabling the distinction between physiological communication‐related EVPs and disease‐driving EVPs. This will enable targeted intervention of pathogenic EVP subpopulations while preserving EVPs required for normal physiological communication. In addition, the volume of mouse synovial fluid is too small to allow for EVP isolation and quantification, which limits the direct acquisition of FLS‐derived EVPs in vivo following shRNA intervention. Tackling these limitations will be helpful for advancing EVP‐based diagnostics and therapeutics in OA.

5. Conclusion

In this study, we demonstrated that inflammatory and senescent FLS‐derived EVPs served as pivotal mediators of OA pathogenesis, orchestrating multifaceted pathological processes including chondrocyte degradation, pro‐inflammatory macrophage polarization and MSC dysfunction. Comprehensive proteomic analysis revealed that these pathogenic EVPs carried diverse molecular cargo that reflected their parent cells' pathological state and contributed to OA progression through multiple mechanisms. This recognition led us to develop an alternative strategy focusing on global modulation of EVP secretion rather than interfering single component. Impairing EVP secretion through synovium‐targeted Rab27a knockdown effectively ameliorated murine OA pathology in vivo, without causing systemic toxicity (Figure 8). In summary, our findings propose a potential therapeutic strategy for clinical OA treatment.

FIGURE 8.

Damaging effects of pathogenic FLS EVPs on multiple joint cell types and alleviation of OA via targeted R a b 2 7 a inhibition using genetically‐modified AAV (created in biorender.com).

Author Contributions

B.L. conducted majority of the assays, acquired and analyzed data. X.C. drafted the manuscript and compeleted revision for peer review. T.S. and Y.B. participated in isolation and maintenance of primary cells. Y.X., Y.S. and X.A. assisted in project design and provided technique supports. W.W., W.G., R.P. and W.G. participated in animal experiments. B.G. and Q.J. conceived, designed the project, supervised experiments and provided research resources. All authors approved the final version of this manuscript.

Conflicts of Interest

The authors declared no conflicts of interest.

Supporting information

Supplementary Materials: jev270162‐sup‐0001‐SuppMat.docx

Liu, B. , Xian Y., Shen T., et al. 2025. “Targeted Blockage of Pathological Extracellular Vesicles and Particles From Fibroblast‐Like Synoviocytes for Osteoarthritis Relief: Proteomic Analysis and Cellular Effect.” Journal of Extracellular Vesicles 14, no. 9: 14, e70162. 10.1002/jev2.70162

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.