Extracellular Vesicle-Mediated miR-150-3p Delivery in Joint Homeostasis: A Potential Treatment for Osteoarthritis?

Specific-pathogen-free female Wistar rats (8 weeks old and 180–220 g) (SLAC, Shanghai, China) were raised for one week under conventional housing conditions to adapt to the environment before the experiments. After being anesthetized via intraperitoneal injection of 2% pentobarbital sodium (Sinopharm, Beijing, China), the rats were fixed in a supine position. Unilateral knee OA was induced within the rats (model group) through anterior cruciate-ligament transection (ALCT) (Figure 1A). Then, the rats were intramuscularly administered 30,000 U/d penicillin (North China Pharmaceutical Co, Hebei, China) once a day for three consecutive days after the surgery to prevent infection. The mobility of the rats on the surgical side was maintained, and the food and water intake, changes in hair color, incision healing, as well as suture intactness, and incidence of infection were closely monitored. Healthy rats that were not subjected to the surgery were set as the control group and housed under the same conditions. All the animals were provided with water and the conventional pellet food and housed in a quiet and comfortable environment. The study was approved by the Animal Experiment Ethics Committee of China-Japan Friendship Hospital (No. 180117).

Whole blood was collected from the abdominal aorta of the rats in the two groups after 10 weeks and centrifuged at 1900× g for 10 min to separate the serum. These serum samples were then stored at −80 °C until needed. The entire knee joint of the rats in each group was harvested, fixed in 4% paraformaldehyde, and stained with hematoxylin and eosin (HE) to validate the successful establishment of the OA animal model.

The EVs were isolated from the serum samples from each group by using differential centrifugation as follows: 500× g for 10 min, 2000× g for 30 min, and 10,000× g for 30 min. Afterwards, the supernatants containing the EVs were collected, filtered through 0.22 µm sterile filters into centrifuge tubes, topped up to 20 mL with sterile PBS, and then centrifuged at 120,000× g for 1 h (Optima XE, Beckman, Fort Collins, CO, USA). All of the above procedures were conducted at 4 °C in sterile centrifuge tubes. Finally, after carefully removing the supernatant, the pellets were resuspended with sterile PBS and stored at −80 °C for subsequent experiments.

The extracted EVs were identified using transmission electron microscopy (TEM, Tecnai f20, Philips, Amsterdam, Netherlands), nanoparticle tracking analysis (NTA, NS300, NanoSight, Malvern, UK), and Western blotting (WB). After the extracted particles were stained with phosphotungstic-acid solution, TEM was used to observe and photograph the extracted particles and determine whether their morphology was consistent with the general characteristics of EVs. NTA was used to measure the diameter and concentration of the extracted particles (diluted for 30 folds in sterile phosphate-buffered saline [PBS, Corning Incorporated, Corning, NY, USA]). WB was used to detect EV marker proteins, including CD9, CD63, and HSP70 (Abcam, Cambridge, MA, USA), and the detailed specific procedures are illustrated in the “WB” section.

The EV miRNAs were analyzed using high-throughput sequencing. The procedures of the bioinformatic analysis were as follows: The raw data were processed and statistically analyzed; then, quality control, length-distribution analysis, database comparison and filtration, gene comparison, and analysis of the gene-expression level were conducted on known and predicted miRNAs to identify the differentially expressed (DE) miRNAs (YanzaiBio Co, Shanghai, China). The DE miRNAs were then distributed on a volcano plot, and a heatmap was employed to cluster these miRNAs according to their expression levels. The DE miRNAs were validated using reverse transcription-quantitative polymerase chain reaction (qPCR) and subjected to gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses to predict their target genes, associated pathways, and putative functional roles.

miR-150-3p mimics were labeled with Cy3 to observe their location. FLSs were transfected with the Cy3-labeled mimics (GenerayBiotech, Shanghai, China) according to the manufacturer’s instructions. The sequences were as follows: miR-150-3p mimic, forward 5′-CUGGUACAGGCCUGGGGGA-3′, reverse 5′-CCCCAGGCCUGUACCAGUU-3′.

The transfected FLSs were seeded in the upper chamber and co-cultured with chondrocytes pre-seeded in the lower chamber. To assess whether the miR-150-3p mimics transfected into the FLSs in the upper chamber had been released and then absorbed by the chondrocytes in the lower chamber, the localization of the red Cy3 fluorescence was characterized. As a control, another group of FLSs was treated with Cy3 that was not conjugated to miR-150-3p and then co-cultured with the chondrocytes. Under a fluorescence microscope, the chondrocytes were examined for the presence of the red Cy3 fluorescence signal to address the effect of Cy3 on them alone.

In addition, to assess whether EVs are involved in the delivery of miR-150-3p from FLSs to chondrocytes, we treated another set of the in vitro OA model transfected with the Cy3-labeled miR-150-3p mimics with the EV inhibitor GW4869 (Sigma-Aldrich, St. Louis, MO, USA) (10 µM).

Given the high levels of miR-150-3p in the circulating EVs of healthy rats, we considered that miR-150-3p might positively impact the injured joints in OA. However, to determine whether EVs-150 in the circulation are associated with FLSs, which participate in OA progression, and whether FLSs can secrete EVs, we isolated FLSs from healthy rats and cultured them as described above. The FLSs were passaged until they reached the required cell quantity. Then, they were collected, rinsed three times with PBS, and transferred to a fresh serum-free EV medium (Gibco, Carlsbad, CA, USA). After 48 h, the culture supernatant was collected and its EV content was extracted via differential centrifugation as described above. TEM, NTA, and WB (CD9, TSG101, and HSP70 [Abcam, Cambridge, MA, USA]) were used to characterize the EVs.

To detect the uptake of FLS–EVs by chondrocytes in vitro, we labeled the FLS–EVs using the PKH67 green-fluorescent kit (Sigma-Aldrich, St. Louis, MO, USA). For subsequent experiments, the EVs were re-extracted via differential centrifugation and quantitated using a bicinchoninic acid (BCA, Sigma-Aldrich, St. Louis, MO, USA) assay. To observe the uptake of EVs by chondrocytes, the chondrocyte nuclei were first labeled with the Hoechst stain (Sigma-Aldrich, St. Louis, MO, USA) for 15 min to aid the localization of the cells, and then, the cells were rinsed three times with PBS. With the addition of the EVs (50 mg/L final concentration), the EV uptake of the chondrocytes was observed and recorded after 1, 6, 12, and 24 h. The other group was pre-treated with GW4869 for 24 h before the addition of the EVs, and the change in fluorescence intensity after the EV blockage was observed.

To further study the functions of EVs, we treated the chondrocytes with GW4869 or FLS–EVs isolated from healthy rats and compared these cells with untreated chondrocytes. qPCR was used to evaluate the effect of the EV treatment on the expression of miR-150-3p in the chondrocytes, to confirm that this miRNA was enriched in EVs and transported by them and to determine whether GW4869 terminated this process.

To assess the regulatory effect of EVs-150 on the chondrocytes of OA, the expression levels of COLII and aggrecan (ACAN) were measured using qPCR and WB. The concentrations of the pro-inflammatory cytokines interleukin-1β (IL-1β), IL-6, and tumor necrosis factor-α (TNF-α) in the chondrocyte culture medium were measured using an enzyme-linked immunosorbent assay (ELISA).

Furthermore, to investigate the mechanism whereby FLS–EVs from healthy rats regulate OA, we predicted the target genes of miR-150-3p and associated pathways via GO enrichment analysis. Given the role of the innate immune response in OA that we showed in our previous study [34], the effects of EVs-150 on the expression of the target gene Trim14 and the downstream genes NF-κB and IFN-β, three genes involved in the innate immune response, were analyzed using qPCR and WB.

To further study the functions of EVs-150, we used TargetScan to predict the targets of miR-150-3p. Consequently, we predicted Trim14 as a potential miR-150-3p target and identified the putative binding site of miR-150-3p on Trim14. These results were validated using the dual luciferase assay. To this end, we constructed a set of pGL3 luciferase reporter vectors with the promoter region composed of the identified miR-150-3p binding sites on the Trim14 promoter. Next, 293T cells (Cell Bank of Chinese Academy of Sciences, Shanghai, China) were transfected with a vector expressing miR-150-3p or the negative control (NC) vector. The sequences (Generay Biotech, Shanghai, China) used for the dual-luciferase assay were: pGL3-Trim14-3′ UTR-WT, TTCTGAGCTGGGGTTTGTGCTGGC; pGL3-Trim14-3′ UTR-MUT, TAGTGAGGTGCCGTTACAGGTGCC; miR-150-3p mimic NC, forward 5′-UUCUCCGAACGUGUCACGUTT-3′, reverse 5′-ACGUGACACGUUCGGAGAATT-3′.

To elucidate the molecular mechanism underlying the effect of miR-150-3p on joint homeostasis during OA, we transfected FLSs with a miR-150-3p mimic, miR-150-3p inhibitor, or NC by using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Then, the cells were transferred to the upper chamber of the transwell system in the cell model and evaluated for their ability to influence the chondrocytes in the lower chamber. The pro-inflammatory cytokines IL-1β, IL-6, and TNF-α in the culture medium were quantitated using ELISA. The expression levels of COLII, ACAN, Trim14, NF-κB, and IFN-β in the chondrocytes of each group were determined via qPCR and WB. The sequences of the oligonucleotides (Generay Biotech, Shanghai, China) used were as follows: miR-150-3p inhibitor, 5′-UCCCCCAGGCCUGUACCAG-3′; miR-150-3p inhibitor NC, 5′-GGCCUCACCGGGUGUAAAUCAG-3′.

Healthy FLS-derived EVs (H-FLS–EVs) were labeled with the DiR fluorescent dye (KeyGEN, Nanjing, China). The labeled EVs were re-extracted to remove impurities and then injected into rats with OA via the caudal vein (the construction of the OA rat model is described above). The fluorescence distribution in the animals was observed after 1, 6, and 24 h via live imaging (IVIS Lumina XRMS Series III, PerkinElmer, Waltham, MA, USA). Likewise, the DiR fluorescent dye without EVs was injected into another set of healthy or OA rats, as a control.

The H-FLS–EVs were injected inside the rats through the tail vein, and four-week treatment in the early stages (early EV-treated group) of OA was compared with the same treatment period in the late stages (late EV-treated group). The rats were grouped and processed as follows: In the early EV-treated group, EV injection (500 µg/kg) was conducted on days 1 and 4, from the fourth to the seventh week after the surgery; in the late EV-treated group, EV injection was conducted on days 1 and 4, from the seventh to the tenth week after the surgery; in the model group, the animals were not subjected to any intervention after the surgery; and in the control group, healthy rats from the same batch of animals that had been used to construct the OA model were used. The behavior of the rats within each group was assessed based on the modified Lequesne MG index [35]. It had the stimulus–response of local pain, altered gait, joint motion range, and joint swelling degree. Blood and joint fluid were collected from each group on the last day of week 10, and the concentrations of the pro-inflammatory cytokines IL-1β, IL-6, and TNF-α were measured using ELISA. The cartilage tissues were harvested and stained with Masson to perform histological grading through the Wakitani scoring system. qPCR was used to measure the miR-150-3p level in the serum EVs of each group. The effects of EVs-150 on the expression levels of COLII, ACAN, Trim14, NF-κB, and IFN-β in the AC of each group were determined via qPCR, WB, and IF to investigate the regulatory functions of EVs-150 in OA.

The FLSs or chondrocytes were digested with 0.25% trypsin containing 0.05% EDTA (Gibco), centrifuged at 300× g for 5 min, collected, and then washed twice in the stain buffer (BD Pharmingen, San Diego, CA, USA). After the cell pellets were resuspended in the stain buffer, the cell concentration in each sample was adjusted to 1 × 10⁷ cells/mL. Then, 100 μL of each cell suspension (approximately 1 × 10⁶ cells) was transferred into an FCM tube. Subsequently, Annexin V-FITC/PI, FITC-labeled anti-CD3 antibody, PE-labeled anti-CD29 antibody (eBioscience, San Diego, CA, USA), PE-labeled anti-CD90 antibody, PE-labeled anti-CD68 antibody, and FITC-labeled anti-CD105 antibody (Abcam, Cambridge, MA, USA) were added into each sample. The samples were then incubated for 20 min at room temperature (20 ± 5 °C) in the dark. The isotype control was likewise prepared using the isotype antibody (BD Pharmingen) at a volume equal to the total volume of the anti-CD antibodies. After the incubation, the samples were mixed with 2 mL stain buffer and centrifuged at 300× g for 5 min. The supernatant was discarded, and each cell pellet was resuspended in 0.5 mL stain buffer. FCM analysis (LSRII, BD Biosciences, San Jose, CA, USA) was promptly performed, and approximately 10,000 cells per sample were analyzed. The collected data were exported in the FCS3.0 format and analyzed using the Flowjo software (Tree Star Inc., Ashland, OR, USA). The proportion and intensity of the antigen expression were determined based on the isotype controls.

The entire knee joint or femoral condyle with complete AC was harvested, fixed in 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO, USA) for 24 h, and then decalcified using the EDTA (Servicebio Tech., Wuhan, China) method for 28 d with daily replacement of the decalcifying solution. After dehydration by an ethanol gradient, vitrification by xylene, and paraffin embedding, the trimmed wax block was cut into 4 μm-thick sections using a microtome (RM2016, Leica, Wetzlar, Germany). The tissue sections were then dewaxed and rehydrated using conventional procedures for subsequent use.

The entire joints from the control and model groups were stained with HE (Servicebio Tech., Wuhan, China). The morphological characteristics of the tissues were observed under a microscope to validate the successful construction of the OA model. AC morphology and collagen maturity were assessed via Masson’s staining (Sigma-Aldrich, St. Louis, MO, USA) to evaluate the OA progression after the EV-150 treatment.

Chondrocyte suspension was collected, and 3.5 × 10⁴ cells were transferred into each well of a 24-well plate containing sterile coverslips. After overnight standing, the culture supernatant was removed, and the cells were incubated in 4% paraformaldehyde for 20 min at room temperature. Afterwards, the cells were washed with PBS and stained with toluidine blue (Solarbio, Beijing, China) for 30 min. Finally, they were rinsed with anhydrous ethanol and observed and photographed under a microscope.

The cells or tissues were lysed, and the total protein concentration was measured using the BCA assay. Equal quantities of proteins were resolved using SDS-PAGE and then transferred onto a PVDF membrane (Merck Millipore, Billerica, MA, USA). Next, the membrane was blocked at 37 °C for 2 h and then incubated with the primary antibody (anti-COLII, anti-ACAN, anti-NF-κB, anti-IFN-β, anti-GAPDH [all from Abcam, Cambridge, MA, USA], or anti-Trim14 [Novus Biologicals] antibody) at 4 °C overnight. The membrane was washed three times with PBST (PBS with Tween 20, Solarbio) and then incubated with the horseradish-peroxidase-labeled anti-mouse or anti-rabbit IgG secondary antibody (Jackson Immunoresearch, West Grove, PA, USA) (diluted 1: 10,000 with PBST) at 37 °C for 2 h. Finally, the membrane was washed five times with PBST and developed using chemiluminescence (BeyoECL Star [Beyotime, Shanghai, China]). The results were analyzed using the ImageJ software (NIH, Bethesda, MA, USA).

Total RNA was extracted, and cDNA was synthesized using the RNeasy Mini Kit (QIAGEN, Duesseldorf, Germany), PrimeScript RT Master Mix, PrimeScript II 1st Strand cDNA Synthesis Kit (TaKaRa, Tokyo, Japan), and Power SYBR Green PCR Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) according to the instructions of the manufacturers. The relative expression levels of miR-150-3p, COLII, ACAN, Trim14, NF-κB, and IFN-β were measured via qPCR using the ΔΔCt method. For normalization, U6 and cel-miR39 were used as the internal and external small RNA controls (for miR-150-3p), and GADPH was used as the internal control for the mRNAs. The results were normalized, and the specificity of the amplification was determined based on the presence of a single peak in the melting curve. The sequences of the primers (Sangon Biotech, Shanghai, China) used were as follows: miR-150-3p, stem-loop 5′-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACTCCCCC-3′, forward 5′-GCGCCTGGTACAGGCCT-3′; universal downstream primer for stem-loop method, 5′-GTGCAGGGTCCGAGGT-3′; U6, reverse transcription 5′-CGCTTCACGAATTTGCGTGTCAT-3′, forward 5′-GCTTCGGCAGCACATATACTAAAAT-3′, reverse 5′-CGCTTCACGAATTTGCGTGTCAT-3′; cel-miR-39, stem-loop 5′-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCAAGCT-3′, forward 5′-GGCCTCACCGGGTGTAAATCAG-3′; COLII, forward 5′-ACTTAACATCCAAGGCCGCT-3′, reverse 5′-ACAATATTTGCCTCAGTTTGTGC-3′; ACAN, forward 5′-AGAGGCAGAGGGACTTTCGGT-3′, reverse 5′-AGAGGCAGAGGGACTTTCGGT-3′; Trim14, forward 5′-GTGAATACTTACAGTGCCTTGC-3′, reverse 5′-GACCCAGACCAGAACCCT-3′; NF-κB, forward 5′-GCACGGATGACAGAGGCGTGTATAAGG-3′, reverse 5′-GGCGGATGATCTCCTTCTCTCTGTCTG-3′; IFN-β, forward 5′-ATTGCGTTCCTGCTGTGC-3′, reverse 5′-GTCCGAATGCTAGTGCTTTGTC-3′; GAPDH, forward 5′-AGACAGCCGCATCTTCTTGT-3′, reverse 5′-CTTGCCGTGGGTAGAGTCAT-3′.

The IL-1β, IL-6, and TNF-α levels in cell-culture media, sera, and synovial fluid were measured using ELISA according to the instructions of the manufacturer of the ELISA kit (Abcam, Cambridge, MA, USA).

All the data are expressed as mean ± SEM and statistically processed using the SPSS 22.0 software (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 9.0 software (GraphPad Software, Inc., La Jolla, CA, USA). Comparison between the two groups and among more groups was performed using the Student’s t-test and one-way ANOVA, respectively. p < 0.05 was considered to indicate statistical significance.

To identify the miRNAs loaded into EVs at different levels between OA and healthy rats, we first established a rat model of knee OA (described in the Materials and Methods section). After 10 weeks, we analyzed the knee joints of the rats in the control and model groups using HE staining. We found that the synovial tissues in the control group were smooth and shiny. In contrast, those in the model group were significantly thickened and had a rough texture and yellow appearance (Figure 1C). The results from the HE staining of the joints further validated that the animal model was successfully established. The rats in the control group had a healthy, continuous, and intact synovium that enveloped the tissues and structures within the joint cavity, and the synoviocytes were neatly arranged as a thin layer. However, the rats in the model group had thickened synovial tissues with proliferated, aggregated, and disorganized synoviocytes. Furthermore, the control group had a continuous and intact AC tidal-line layer, and the superficial, transitional, radial, and calcified zones were clear. In contrast, the AC in the model group had an abraded superficial layer and a disorganized, shifted, and thinned tidal-line layer. The cell layer was arranged in a disorderly manner and exhibited calcification (Figure 1D).

Next, we extracted circulating EVs from serum samples to characterize the miRNA content of these EVs. The characteristics of EVs include morphology, size distribution, quantification, and surface marker expression. TEM, NTA, and WB analysis of the three positive proteins was employed to identify the EVs extracted from the control and the model groups. Under TEM, cup-shaped or round, hollow spherical vesicles were revealed (Figure 2A), and the NTA results showed the sizes and concentrations of the extracted particles (Figure 2B). Moreover, the WB results showed that the extracted EVs were enriched in the EV markers CD9, CD63, and Hsp70 (Figure 2C). These data were consistent with the characteristics of EVs, depicting the successful extraction of circulating EVs.

miRNAs are the main molecules in EVs and are involved in pathways important in joint injury and repair in OA by regulating the protein levels of target mRNAs. To assess the OA-induced changes in the miRNA levels in EVs, we performed deep sequencing of the miRNAs in the circulating EVs from healthy and OA rats (BioProject ID: PRJNA753999). Among the 233 known and 344 predicted miRNAs in all the samples, we identified the 14 miRNAs that were the most significantly differentially represented between the two groups, including 12 upregulated and 2 downregulated miRNAs. GO enrichment analysis was performed to evaluate the functions of the DE miRNAs. The sequencing results showed that miR-150-3p was over-represented in the control (C) group compared with the level in the model (M) group (log₂ fold change: 8.43) (Figure 3A–D). The qPCR results confirmed that the miR-150-3p level in the circulating EVs was significantly different between the two groups (p < 0.05) (Figure 3E). These results indicated that EVs-150 might be associated with the pathogenesis of OA. Thus, in the subsequent experiments, we assessed the role of miR-150-3p in OA.

To clarify the substance delivery by EVs-150 and the regulatory relationship among different cell types in OA, we established an in vitro cell model that simulates OA.

We isolated FLSs and chondrocytes from the SM and AC tissues of OA rats through collagenase treatment, respectively (Figure 4A). The isolated synoviocytes were cultured, and the cultured cells primarily consisted of FLSs after several passagings. Under an inverted microscope, the FLSs exhibited columnar, spindle-shaped, and occasionally polygonal morphology, with an oval nucleus in the center. The FLSs in the culture formed clusters and grew in a polarized fashion. In contrast, the chondrocytes adhered to the substratum, extended pseudopodia, and exhibited a spindle, star, or polygon shape. Their nucleus was oval or round, clear, and located in the center. The chondrocytes in the culture showed fusion growth and grew in clusters (Figure 4B).

FCM was used to determine the viability of the FLSs and the isolated chondrocytes. The results revealed that the total number of viable cells from the two types of cells was more than 95%. Moreover, the total number of cells during early and late apoptosis or necrosis was less than 5%, which could be utilized for subsequent experiments (Figure 4C).

To further label FLSs, Vimentin, the characteristic labeling protein of FLSs, was labeled in the SM tissue required for the FLS culture. We observed that the SM in the field of view under the light microscope was all over the brown-yellow positive cells after the IHC staining of the SM (Figure 4D). This was established by the IF staining results of the SM. Moreover, it was observed under the fluorescence microscope that the red Vimentin was mainly located in the cytoplasm, especially in the lining layer of the SM, with a large amount of red fluorescence concentration. It indicated that FLSs were the primary cell type in the SM (Figure 4E), the same as our previous study [34]. The FLSs isolated from the SM were labeled with Vimentin, and the results consistently revealed that most cells were Vimentin-positive (Figure 5A).

FCM was employed to analyze several main cell types in the synoviocytes, namely FLSs (CD90), macrophages (CD68), and T cells (CD3), and the results showed that highly pure FLSs (99.29% of the total cells) could be obtained using our isolation and culture methods (Fig. 5B). In addition, the expression rates of the MSC surface markers CD29 and CD105 were 0.01% and 0.26%, respectively. These were negative, excluding the possibility that our isolated cells from the SM were MCSs. Proteoglycans and COLII are the main components of the cartilage matrix and thus the characteristic markers of chondrocytes. The basic toluidine blue stain combines with the acidic proteoglycans in chondrocytes to present a bluish-purple color (Figure 5C). COLII IF staining showed that the cytoplasm and nuclei of the chondrocytes displayed red and blue fluorescence, respectively. Accordingly, the chondrocytes were COLII-positive. These results demonstrated that the cells we isolated and cultured displayed the characteristics of chondrocytes (Figure 5D).

We established a transwell-based cell model in which FLSs and chondrocytes were co-cultured in the upper and lower chambers in a common liquid environment, respectively, to simulate the interaction between the two major cell types in a joint. The subsequent experiments were performed using this in vitro OA model.

Various cell types actively secrete miRNAs entering the blood circulation. They can be delivered into recipient cells by vectors, such as EVs, thereby affecting the functions of the recipient cells. Healthy synoviocytes, together with the synovial fluid they secret, play important roles in the protection of chondrocytes against injury. We determined whether FLSs, the largest cell population in synoviocytes, can secret extracellular miRNAs transported to chondrocytes.

FLSs transfected with Cy3-labeled miR-150-3p mimics were transferred to the upper chamber of a transwell to be co-cultured with chondrocytes plated in the lower chamber. The presence of the red Cy3 fluorescence in the chondrocytes suggested that the Cy3-labeled miR-150-3p mimics were transferred from the FLSs in the upper chamber to the chondrocytes in the lower chamber. FLSs can secrete extracellular miRNAs that can be absorbed by chondrocytes as the recipient cells, suggesting miRNA-mediated signaling from FLSs to chondrocytes.

We considered that FLS–EVs might carry the miRNAs to be transported to the chondrocytes and thus treated the transwell co-culture system with GW4869, an inhibitor of EV secretion. The GW4869 treatment reduced the intensity of the red fluorescence in the chondrocytes in the lower chamber, indicating that inhibiting the secretion of EVs decreased the delivery of miR-150-3p to the chondrocytes. Thus, FLS–EVs may serve as carriers for the miRNAs to be transported from FLSs to chondrocytes (Figure 6).

EVs are secreted by cells containing miRNAs and are the primary mechanism of intercellular miRNA transport. To determine whether FLSs can secrete EVs, we cultured FLSs isolated from the SM of healthy rats, extracted the EVs in the cell-culture supernatant via differential centrifugation, and then characterized them. These EVs were presented as hollow spherical vesicles under TEM (Figure 7A). In addition, NTA revealed that the main peak for the particle size corresponded to 116 nm (Figure 7B), and the particles were positive for the EV-specific protein markers CD9, TSG101, and HSP70, indicating that FLSs can secrete EVs (Figure 7C).

As the transwell assay can only prove that substance transport did occur between the cells in the upper and lower chambers, we conducted an EV-tracking experiment to assess whether chondrocytes could uptake EVs secreted by FLSs. The FLS–EVs were labeled with the fluorescent dye PKH67 and then co-cultured with chondrocytes. After 6 h, green fluorescence was observed in the perinuclear region of the chondrocytes, demonstrating that the PKH67-labeled EVs had been endocytosed by the chondrocytes. The intensity of the green fluorescence in the chondrocytes gradually increased with time, especially in the perinuclear region, and reached saturation at 24 h, indicating the high absorption of the FLS–EVs by the chondrocytes (Figure 7D).

Our results indicated that FLSs could secrete EVs that effectively transport miR-150-3p to chondrocytes.

Considering that H-FLS–EVs may have a positive impact on the maintenance of joint homeostasis, we co-cultured FLS–EVs isolated from the SM of healthy rats with OA chondrocytes for 24 h. The qPCR results showed that the miR-150-3p level in the chondrocytes was significantly increased compared with the control group (p < 0.05), further confirming that EVs deliver miR-150-3p from FLSs to chondrocytes in vitro. In contrast, the GW4869 treatment suppressed the increase in miR-150-3p level in the chondrocytes.

Moreover, the COLII and ACAN expression in the chondrocytes treated with the H-FLS–EVs was enhanced, and the concentrations of the pro-inflammatory cytokines IL-1β, IL-6, and TNF-α in the chondrocyte culture medium were decreased, demonstrating that EVs-150 may promote chondrocyte repair through the regulation of the expression levels of the target genes and related pathways. GO and KEGG enrichment analyses of the target genes and pathways revealed that miR-150-3p could target the Trim14 and NF-κB pathways of the innate immune response. Adding H-FLS–EVs decreased the mRNA and protein levels of Trim14 and its downstream genes NF-κB and IFN-β. However, the addition of GW4869 suppressed this effect, indicating that EVs-150 regulated Trim14, NF-κB, and IFN-β expression levels at the transcript level (Figure 8).

To elucidate the mechanism underlying the regulation of Trim14 expression by miR-150-3p and its relationship with the NF-κB signaling pathway, we analyzed the binding site between miR-150-3p and the 3′ end of Trim14 by using TargetScan. A plausible binding site was identified in the promoter region of Trim14. As shown in Figure 9A, the promoter fragments were significantly inhibited by miR-150-3p expression, indicating that Trim14 is a direct target of miR-150-3p.

To further investigate the mechanisms of the chondrocyte repair and maintenance of joint homeostasis induced by H-FLS–EVs, we examined the effect of miR-150-3p on the Trim14/NF-κB/IFN-β signal transduction in the OA-related innate immune response. FLSs transfected with plasmids overexpressing miR-150-3p mimics were co-cultured with OA chondrocytes in the in vitro model. The COLII, ACAN, Trim14, NF-κB, and IFN-β expression levels in the chondrocytes were measured using WB and qPCR. The miR-150-3p mimics significantly upregulated the COLII and ACAN protein levels but downregulated the Trim14, NF-κB, and IFN-β protein levels in the chondrocytes (Figure 9B,C). The corresponding mRNA levels had a similar trend with the protein levels. Furthermore, the miR-150-3p mimics also significantly decreased the release of the pro-inflammatory cytokines IL-1β, IL-6, and TNF-α into the culture medium.

We also assessed whether the miR-150-3p inhibitor had the opposite effect. It was found that the inhibitor enhanced the release of the pro-inflammatory cytokines, reduced the COLII and ACAN expression in the chondrocytes, and activated the Trim14/NF-κB/IFN-β pathway. These observations demonstrated that miR-150-3p could effectively regulate the Trim14/NF-κB/IFN-β axis of the innate immune response and stimulate FLSs to influence chondrocytes and the co-culture microenvironment, indicating that miR-150-3p may be a potential therapeutic target in OA and other joint diseases (Figure 9).

The miR-150-3p carried by H-FLS–EVs has an inhibitory effect on the Trim14/NF-κB/IFN-β signal transduction during the innate immune response in vitro. We evaluated the effects of this miRNA on joint homeostasis in OA in vivo. Fluorophore-labeled H-FLS–EVs were injected into the rats via the tail vein and tracked in vivo. Most fluorophores clustered around the liver in each group and gradually dimmed and disappeared by 24 h, in line with the metabolism process of medicines in rats. Any signal in other regions was assessed until the fluorophores disappeared from the imaging system with time quenching. This result is inconsistent with previous studies in the literature, which reported evident fluorophore distribution lasting several weeks in the injured joints [36]. Our results indicated a strong liver expression, and the small level of fluorescence that may be present at the joints was not enough to be detected. We deduced that H-FLS–EVs injected through the caudal vein could exert physiological effects such as circulating EVs-miRNA through systemic circulation. This is also one of the reasons why we selected blood as the most accessible sample to extract EVs for miRNA sequencing at the beginning of the study (Figure 10A).

Despite the loss of the evidence of intraarticular fluorescence, we have demonstrated the joint protection effect of FLS–EVs in vivo by other means, similar to the ones described in vitro. The joints of the rats treated with H-FLS–EVs (early EV-treated group and late EV-treated group) were compared with the joints of the rats in the model group (rats with OA) and the control group (healthy rats) without any EV intervention. We first measured the concentrations of inflammatory mediators in the synovial fluid, and the results showed lower concentrations in the early EV-treated and late EV-treated groups than in the model group, especially in the early EV-treated group, indicating an improved liquid-matrix environment in the joint. Before sample collection, various indices of the rats in each group were also consistent with the above trend, according to a Lequesne MG index, suggesting the improvement effect of H-FLS–EVs on behavior (Figure 10B). However, no significant differences were observed in the serum levels of the inflammatory mediators among the groups. Moreover, no abnormal level of circulating inflammatory mediator was detected in comparing the model group with the control group (Figure 10C). This result indicated that OA tends to cause sterile, low-grade joint inflammation. Even though the inflammatory joint in OA could deliver some substances to influence disease progression via blood circulation, OA is not a systemic inflammatory disease, based on the traditional definition. The widely recognized inflammatory mediators in the blood cannot be considered indicators for a sensitive diagnosis and prognosis in OA. However, the level of miR-150-3p in the serum EVs was also significantly upregulated in the EV-treated rats, similar to the findings in vitro. This result again emphasizes the importance of our studies on EVs-150.

In EV-treated rats, although the levels of some inflammatory markers in the bloodstream were not substantially changed compared with the control levels, the AC showed a clear improvement. We performed the histological and characteristic analysis of the AC in each group. Masson’s staining and IF revealed that, compared with the rats in other groups, the rats in group OA had severe joint wear and loss of the cartilage matrix, disordered chondrocyte arrangement, and significantly reduced the COLII and ACAN expression. Cartilage damage and matrix loss were also observed in the early EV-treated and late EV-treated groups, unlike in the control group. However, these changes were not as severe as those in the model group, which showed thinning of the cartilage matrix and chondrocyte clustering. They decreased the COLII and ACAN levels in the cartilage matrix. The AC injury severity in the early EV-treated group was less than that in the late EV-treated group. It showed a slightly thinner cartilage matrix compared with that in the control group, a lighter degree of decrease in the COLII and ACAN levels in the cartilage matrix, and significantly better AC conditions than those in the model and late EV-treated group (Figure 10D and Figure 11).

Treatment with H-FLS–EVs reduced the inflammation level and AC injury in OA rats. The early FLS–EV intervention significantly suppressed the progression of OA. It effectively prevented the exacerbation of the AC injury caused by the knee instability even more than the late intervention. This observation highlights the importance of starting the treatment as early as possible.

To further evaluate the effect of the EV-150 treatment on the innate immune signal transduction pathway, we measured the levels of miR-150-3p, its target gene Trim14, and the genes involved in the related pathways in the AC of each group. Compared with the levels in healthy rats, in the AC of OA rats the miR-150-3p level was significantly decreased. In contrast, the mRNA and protein levels of Trim14, NF-κB, and IFN-β were significantly increased, suggesting that the innate immune pathway was activated. After the injection of H-FLS–EVs, the abundance of miR-150-3p was partially restored. Concurrently, the miR-150-3p accumulation induced by EVs interrupted the Trim14/NF-κB/IFN-β axis and inhibited joint inflammation.

Co-staining, the AC for Trim 14 and NF-κB, showed the colocalization of these proteins and their upregulation during OA. However, the EV-150 treatment decreased their levels, consistent with the above-mentioned morphological observations on the AC in the different groups. We observed that several target proteins were expressed in the AC at different levels among the different groups. Thus, the levels of Trim14 and NF-κB in the model group were the highest, followed by those in the early EV-treated and the late EV-treated groups, whereas these proteins were barely detected in the joints of healthy rats. Conversely, the levels of COLII and ACAN showed the opposite expression pattern.

In addition, Trim14, COLII, and ACAN were found mainly concentrated in the cytoplasm. However, NF-κB was localized in the cytoplasm and nucleus to varying degrees among the different groups. In the control group, NF-κB was at the resting state, localized in the cytoplasm. At this state, NF-κB interacts with its suppressor IκBα to form a non-active complex. When OA occurs, activated IκB kinase (IKK) leads to the phosphorylation of IκBα, and consequently, NF-κB becomes activated and translocates into the nucleus. This process could be suppressed to varying degrees using EV-150 treatment. We also found that the region with the highest fluorescence intensity was concentrated in the wear and ruptured areas of the AC, indicating that they were all involved in joint injury and repair during OA (Figure 11 and Figure 12). The joint homeostasis and repair of the AC injury were stimulated via regulating the innate immune signal transduction pathways, demonstrating that EVs-150 obtained from healthy FLSs show great potential in joint protection. In Figure 13, we have summarized the therapeutic mechanisms of the FLS-derived EVs-150.

All data generated or analyzed during this study are included in this published article and are available from the corresponding author upon reasonable request.