Introduction

Articular joints consist of cartilage and synovium. Degeneration of the cartilage is a representative feature of osteoarthritis (OA), the most prevalent joint disease. The status of the synovium is also closely associated with OA. The severity of joint pain is significantly related to the presence of synovitis1. Synovitis is also significantly associated with subsequent progression of OA2. Although these studies suggest an essential role of the synovium in cartilage maintenance, the mechanisms underlying synovium–cartilage interactions remain unclear.

OA is a typical multifactorial disease. Diverse factors are involved in OA pathogenesis, including aging, genetic background, comorbidities, obesity, joint instability, and joint overuse3,4. Excessive mechanical loading on cartilage is considered to be a common feature underlying these pathogenic factors. Previous studies have indicated that overloading of mechanical stress decreases sex-determining region Y-box 9 (Sox9), essential chondrogenic transcription factor, and representative chondrocyte extracellular matrix (ECM) genes such as collagen type II alpha 1 chain (Col2a1) and aggrecan (Acan)5,6,7. Additionally, such overloading induces inflammation in articular chondrocytes via nuclear factor-kappa B (NF-κB), which enhances the expression of catabolic enzymes, including family members of a matrix metalloproteinase (Mmp) and a disintegrin and metallopeptidase domain metallopeptidase with thrombospondin type 1 motif (Adamts)5,6,7. However, the acute response of chondrocytes to excessive mechanical loading in vivo, the effects of repeated overloading, and the involvement of the synovium remain poorly understood.

Here, we prepared a mouse model to investigate excessive mechanical loading on knee joints, combining treadmill exercise with weight attachment. We used RNA sequencing (RNA-seq) to investigate alterations in the gene expression profiles of the synovium and cartilage at 1, 4, and 24 h after single overloading. We also investigated the effects of repeated overloading using tissue samples obtained from mice that underwent four overloading events in four days.

Finally, time-course RNA-seq of the synovium and cartilage revealed that inflammation was transiently upregulated following single overloading, but became prolonged after repeated overloading for four consecutive days. We found Inhba as a candidate molecule responsible for the synovium–cartilage interaction following overloading and investigated its role using mouse primary chondrocytes.

Results

Development of a mouse model to explore overloading by combining treadmill exercise with weight attachment

To investigate short-term transcriptomic changes and cross-talk between the synovium and cartilage following overload of articular joints, we established a mouse model that combined treadmill exercise with weight attachment (Fig. 1a, Supplementary Fig. 1). A weight (20 g) was attached to the trunk of 11-week-old C57BL/6 mice, which had been gradually acclimated to a treadmill for 1 week. The mice were then subjected to 1 h of running on the treadmill, according to the protocol (Fig. 1a). To examine gene expression changes in the synovium and cartilage after a single overloading, we harvested joint tissues at 1, 4, and 24 h following the treadmill running (S1, S4, and S24, Fig. 1a) and compared them with those obtained from control mice without overloading (control, Fig. 1a). To investigate the effects of repeated overloading, we euthanized the mice 24 h after 4 consecutive days of treadmill running (R24, Fig. 1a). Histological analyses showed no obvious changes in the synovium and cartilage between the R24 and control groups (Fig. 1b).

Fig. 1
figure 1

A mouse overload model combining treadmill exercise with weight attachment. (a) The protocol for the mouse overload model. The mice were euthanized 1, 4, or 24 h after single running (S1, S4, and S24), 24 h after the final running of 4 days (R24), or after 1 day without running (control). (b) Safranin O staining of the intercondylar region for the assessment of the synovium (left) and the medial compartment for the assessment of the cartilage in the control and R24 mice. Scale bars, 250 μm.

Full size image

Overview of changes in gene expression profiles over time in the synovium and cartilage after overloading

We conducted RNA-seq of the synovium and cartilage samples obtained from the control, S1, S4, S24, and R24 groups (Fig. 2a, Supplementary Fig. 2). Heatmaps and principal component analysis (PCA) of both the synovium and cartilage samples indicated that the control, S4, and S24 groups were similar, whereas the S1 and R24 groups differed from the control (Fig. 2a, Supplementary Fig. 2). In both types of tissues, acute transcriptomic changes at 1 h had been attenuated by 4 h (Fig. 2a). Notably, the R24 group exhibited a different expression profile from the control, S1, and S24 groups in both the synovium and cartilage (Fig. 2a). In the synovium, the lowest number of differentially expressed genes (DEGs) was observed between the S24 and the control groups (Fig. 2b). In the cartilage, the numbers of DEGs between the S4 vs. control, S24 vs. control, and S24 vs. S4 groups were similar and lower than those of other comparisons (Fig. 2b). These findings align with the variations in expression profiles among the five groups in both tissues, as shown in the PCA plots (Fig. 2a).

Fig. 2
figure 2

RNA sequencing revealed temporal changes in gene expression in synovium and cartilage after overloading. (a) PCA showing gene expression changes over time in synovium and cartilage samples. Arrows indicate the time course of tissue sampling; while dashed arrows represent the R24 group compared with the S24 group. (b) Number of differentially expressed genes (DEGs) between different time points in synovium and cartilage samples.

Full size image

Transient upregulation of immune cell response-related terms in the synovium after single overloading

Next, we examined the trends in transcriptomic alterations through enrichment analysis of up- or down-regulated DEGs in the synovium following single overloading (Fig. 3a–e). Between the S1 vs. control groups, terms such as “cell chemotaxis”, “positive regulation of cytokine production”, “positive regulation of response to external stimulus”, and “leukocyte differentiation” were enriched (Fig. 3a), suggesting a rapid response of the synovium. In contrast, “muscle system process”, “muscle structure development”, and vascular-associated terms were downregulated at 1 h (Fig. 3a). Many of the terms that were upregulated or downregulated at 1 h were downregulated or upregulated respectively at 4 h (Fig. 3b). “Wound healing” was enhanced in the S4 group compared with the S1 group (Fig. 3b). In the S24 group, terms related to metabolic processes were upregulated while those related to chemotaxis were suppressed compared with the S4 group (Fig. 3c). When comparing the S4 group with the control, inflammation-, chemotaxis-, and wound-related terms were upregulated (Fig. 3d). In the S24 group, metabolic process-related terms were upregulated, whereas “nuclear chromosome segregation” was downregulated compared with the control group (Fig. 3e). These results of the enrichment analyses suggested a transient upregulation of immune cell-related and inflammatory responses in the synovium following single overloading.

Fig. 3
figure 3

GO analysis of temporal changes in the synovium following a single overloading. (a–e) GO biological processes enrichment analysis of upregulated or downregulated DEGs between two time-points following single overload running. The top 20 enriched terms are shown. Comparisons between the S1 vs. control groups (a), S4 vs. S1 groups (b), S24 vs. S4 groups (c), S4 vs. control groups (d), and S24 vs. control groups (e).

Full size image

We further displayed DEGs in the synovium after single overloading (Supplementary Fig. 3a–e). Nr4a3 was identified as the most significantly upregulated gene 1 h after single overloading (Supplementary Fig. 3a). The upregulation of Nr4a3 was also detected in the S4 group compared with its expression in the control group (Supplementary Fig. 3d), while it was downregulated in the S24 group (Supplementary Fig. 3c). Although there were 52 DEGs in common between the S1 vs. control and S4 vs. control comparisons, there were only 9 DEGs in common between the S4 vs. control and S24 vs. control comparisons (Supplementary Fig. 3f.). DEGs in common between the S1 vs. control, S4 vs. control, and S24 vs. control comparisons were Clec4d, Calca, Ecel1, and Saa3 (Supplementary Fig. 3f.).

Transient upregulation of inflammatory response-related terms in the cartilage after single overloading

We also performed an enrichment analysis of up- or down-regulated DEGs in the cartilage following single overloading (Fig. 4a–e). At 1 h, “inflammatory response”, “leukocyte activation”, and immune cell activation-related terms were prominently upregulated, while terms closely associated with cartilage or chondrocytes, such as “skeletal system development” and “regulation of chondrocyte differentiation”, were downregulated (Fig. 4a). Many of these changes seen at 1 h were reversed at 4 h (Fig. 4b). Several terms, including “muscle”, were upregulated at 24 h (Fig. 4c). When comparing the S4 group with the control group, “nitric oxide-mediated signal transduction” and inflammation-related terms were upregulated (Fig. 4d). In the S24 group, muscle-related terms were clearly upregulated, while “regulation of chondrocyte differentiation” and “connective tissue development” were downregulated compared with the control group (Fig. 4e). The results of these enrichment analyses suggested that the suppressed anabolism of articular chondrocytes did not recover to the level of the control group within 24 h, although there was transient upregulation of inflammatory responses in the cartilage following single overloading.

Fig. 4
figure 4

GO analysis of temporal changes in the cartilage following a single overloading. (a) Number of DEGs between different time points. (b–d) GO biological processes enrichment analysis of upregulated or downregulated DEGs between two time-points following single overload running. The top 20 enriched terms are shown. Comparisons between the S1 vs. control groups (a), S4 vs. S1 groups (b), S24 vs. S4 groups (c), S4 vs. control groups (d), and S24 vs. control groups (e).

Full size image

Similar to the results seen in the synovium, Nr4a3 was identified as the most significantly upregulated gene 1 h after single overloading (Supplementary Fig. 4a); however, Nr4a3 was downregulated at S4 (Supplementary Fig. 4b). There were 32 DEGs in common between the S1 vs. control and S4 vs. control comparisons, while there were only 5 DEGs in common between the S4 vs. control and S24 vs. control comparisons (Supplementary Fig. 4f.). Wnt1, Tnf, and Saa3 were among the 11 DEGs in common between the S1 vs. control, S4 vs. control, and S24 vs. control comparisons (Supplementary Fig. 4f.).

Effects of repeated overloading on the transcriptome in the synovium and cartilage

Given these short-term changes after single overloading, we next examined the effects of repeated overloading. In both the synovium and cartilage, many more DEGs were detected between the R24 and control groups than between the S24 and control groups (Fig. 2b). In the synovium, upregulation of immune cell-related and inflammatory terms was prominent between the R24 compared with the control group (Fig. 5a), unlike in the comparison between the S24 vs. the control group (Fig. 3e). In the cartilage, upregulation of “inflammatory response” and downregulation of cartilage anabolism-related terms were more prominent between the R24 vs. control groups (Fig. 5b) compared with the S24 vs. control groups (Fig. 4e). Notably, most of the DEGs detected in the R24 vs. control groups were not detected in the comparison between the S24 vs. control groups in both the synovium and cartilage (Supplementary Fig. 5a–d). However, Saa3 was identified as a common DEG between both comparisons in both the synovium and cartilage (Supplementary Fig. 5b,d). These data suggested that repeated overloading enhanced inflammation of the synovium and impaired the characteristics of the cartilage.

Fig. 5
figure 5

Enrichment analysis of the synovium and cartilage following repeated overloading. (a, b) GO biological processes enrichment analysis of upregulated or downregulated DEGs following repeated overload running. The top 20 enriched terms in the synovium (a) and cartilage (b) are shown.

Full size image

We further investigated DEGs related to inflammation or NF-κB in the synovium and cartilage. According to the changes in Gene Ontology (GO) terms over time described above, many of the DEGs were highly upregulated at 1 h and decreased thereafter in both the synovium and cartilage (Fig. 6a,b, 7a,b). Notably, the expression levels of these DEGs were upregulated again in the R24 group (Fig. 6a,b, 7a,b). We further examined ECM- and cartilage development-related DEGs in the cartilage. Representative cartilage ECM-related genes, including Col2a1, Acan, Comp, and Dcn, were downregulated following both single and repeated overloading, whereas Col1a1 and Col1a2 were highly upregulated after repeated overloading (Fig. 7c,d). Overloading also decreased the expression of chondrogenic transcription factors, including Sox9 and Sox5 (Fig. 7d). Many MMP and Adamts family members were upregulated by single or repeated overloading, while the anti-catabolic factors Timp1 and Timp3 were decreased by repeated overloading (Fig. 7c,d).

Fig. 6
figure 6

Inflammation- or NF-κB-related DEGs in the synovium. (a, b) Heatmaps of inflammation- (a) or NF-κB- (b) related DEGs in the synovium.

Full size image
Fig. 7
figure 7

Inflammation-, NF-κB-, ECM-, or cartilage development-related DEGs in the cartilage. (a–d) Heatmaps of inflammation- (a), NF-κB- (b), ECM-(c), or cartilage development-related DEGs (d) in the cartilage.

Full size image

Identification of inhibin subunit beta A (Inhba) as a candidate cytokine for the synovium–cartilage interaction

To further investigate synovium–cartilage interactions, we focused on secreted molecules in the synovium. In the synovium, 29 cytokine genes exhibited increased expression with a fold change greater than 2 following overloading (Fig. 8a). Many of these secreted molecule genes, including Il1b, were induced 1 h after single overloading, subsided after 4 h, and were induced again after 4-day overloading (Fig. 8a). Among these cytokines and chemokines, we detected Inhba, which encodes inhibin subunit beta A (Fig. 8a). Inhba is a preproprotein that is proteolytically processed to generate a subunit of the dimeric activin and inhibin protein complexes, which play essential roles in diverse phenomena8. When we sorted the DEGs of cytokines in order of the greatest variation in expression between two groups, there were large differences in Inhba expression between the control vs. S1, control vs. R24, and S24 vs. R24 groups (Fig. 8b).

Fig. 8
figure 8

Identification of Inhba as an upstream synovial factor for transcriptomic changes in cartilage following repeated overloading. (a) Temporal changes in the expression of cytokine genes in the synovium with fold change > 2 at any time point. The analysis was performed on genes associated with the “cytokine activity” term in GO molecular functions. (b) Heatmaps of the top-30 upregulated cytokines based on comparisons between two time-points, ordered by fold-change values. Comparisons between the S1 vs. control groups (left), the R24 vs. control groups (middle), and the R24 vs. S24 groups (right). (c) Upstream cytokines or growth factor candidates predicted by ingenuity pathway analysis to regulate gene expression changes in cartilage in the R24 group.

Full size image

To identify upstream regulators for the transcriptomic changes we observed in the cartilage following repeated overloading, we performed ingenuity pathway analysis (IPA) (Fig. 8c). Based on the DEGs between the R24 vs. control groups, IPA identified 12 genes as candidate upstream regulators (Fig. 8c). Among them, Nrg2, Ccl4, Pgf, and Cxcl2 were significantly upregulated in the cartilage (Fig. 8c). Notably, only Inhba was identified as a candidate upstream regulator that was induced in the synovium by repeated overloading (Fig. 8c). These findings suggest that Inhba may play a role in the synovium–cartilage interaction in response to overloading.

Inhba is expressed in the lining layer of synovium following repeated overloading

Members of the activin and inhibin family comprise an alpha subunit (Inha) and/or three beta subunits (Inhba, Inhbb, and Inhbc)8. Activin A is a homodimer of Inhba. Activin AB is a heterodimer of Inhba and Inhbb, and inhibin A is a heterodimer of Inha and Inhba. Among the subunit genes, only the expression of Inhba was markedly increased in the synovium by both single and repeated overloading (Supplementary Fig. 6a,b). Inhbc was not detected in the synovium. We then performed immunofluorescence to analyze the localization of Inhba in the articular joints. Inhba was not detected to any great extent in the synovium or cartilage of the joints of mice in the control group (Fig. 9a). Although Inhba expression was unchanged in the S24 synovium, Inhba was clearly detected in the lining layer of the R24 synovium (Fig. 9a).

Fig. 9
figure 9

Upregulation of Inhba expression in the synovial lining layer and time course-dependent effects of activin A on primary cultured chondrocytes. (a) Immunofluorescence of Inhba in the control , S24, and R24 groups. Inset boxes in the left-hand panels indicate the locations of the enlarged images of the lining and sublining layers shown in the right-hand panels. The percentage of the Inhba-positive area in the lining region is shown. (b) Time course of mRNA levels of marker genes in the cartilage following recombinant human activin A treatment. P-values were determined by one-way ANOVA followed by Dunnett’s test. Data are expressed as dot plots and mean ± SD.

Full size image

Next, to confirm the subset of synovial cells expressing Inhba, we used deposited single-cell RNA-seq (scRNA-seq) datasets from acute injury model mice employing anterior cruciate ligament resection (ACLR) (GSE211584)9. Synovial cells of sham and ACLR models at day 7 were categorized into 10 clusters (Supplementary Fig. 7a, b). Among them, Inhba expression was detected in fibroblasts, particularly in lining fibroblasts (Supplementary Fig. 7c), consistent with the immunofluorescence findings (Fig. 9a).

Activin A exerts chondrogenic effects in the early phase and subsequently induces catabolic effects in primary chondrocytes

Given the expression of Inhba, we further investigated the effects of activin A on cartilage. We administered recombinant human activin A protein to primary cultured chondrocytes isolated from P5 mice and examined the expression of marker genes over time. The recombinant human activin A treatment enhanced the expression of Sox9, Col2a1, and Acan at 1 h (Fig. 9b). However, the mRNA levels began to decrease after 4 h and were significantly suppressed at 24 and 48 h compared with their levels at 0 h (Fig. 9b). Timp1 expression was also significantly increased by recombinant human activin A and sustained until 48 h (Fig. 9b).

Discussion

In this study, we first revealed the transcriptomic changes in the synovium and cartilage within 24 h after single overloading. Inflammation-related terms were upregulated in both the synovium and cartilage, and cartilage-related terms were downregulated in the cartilage, 1 h after the single loading. These changes subsided within 24 h; however, many of these restorations did not occur following four-day repeated overloading. We performed IPA and identified Inhba as a candidate upstream molecule for the transcriptomic changes in cartilage. Inhba was induced in the synovium by single and repeated overloading, while the expression of other activin/inhibin subunit genes was not enhanced. Inhba was detected in the lining layer of synovium following overloading. In primary chondrocytes, recombinant human activin A treatment first transiently enhanced the expression of chondrogenic and anti-catabolic factors before subsequently reducing the expression of chondrocyte marker genes. These findings have important implications, as they highlight how repeated overloading can induce chronic inflammation and the degeneration of cartilage in the joints.

An increasing number of studies have shown the diverse effects and roles of activin A and Inhba. They are involved in various cancers, and several studies have reported their effects on cartilage. Hermansson et al. reported that activin A induces Timp1 expression in human chondrocytes in a dose-dependent manner10. Similarly, Alexander et al. demonstrated that low concentrations of activin A inhibit aggrecan degradation in human articular cartilage11. These findings provide evidence for the anti-catabolic effects of activin A. Furthermore, Diller et al. reported that rheumatoid synovial cell infiltration into cartilage is suppressed by activin A overexpression12. In contrast, activin A likely has catabolic effects on chondrocytes. Yu et al. reported that the concentration of activin A in synovial fluid, as well as its expression in synovium and cartilage, increases in OA and rheumatoid arthritis13. Zou et al. found that concentrations of activin A in serum and synovial fluid showed a positive correlation with the occurrence and severity of knee OA14. Ota et al. demonstrated that in rheumatoid arthritis, activin A promotes synovial fibroblast proliferation, which could be partially inhibited by administration of the activin A antagonist follistatin15. Mitugi et al. reported that activin A suppresses chondrogenic differentiation of ATDC5 cells16, while Jeon et al. showed that activin A induces Mmp3 and Mmp13, catabolic enzymes that contribute to the development of OA17. These contradictory effects of activin A on the cartilage and chondrocytes may be explained by our current findings, i.e., anabolic effects in the earlier phase and catabolic effects in the later phase. Additionally, activin A is produced by cardiac muscle in response to acute exercise18. Activin A is also known to be induced by inflammatory cytokines18. Activin A may have time-dependent biphasic effects on cartilage, with its original role being to restore the expression of anabolic factors that are reduced in chondrocytes exposed to excessive mechanical loading during the acute phase.

We previously reported that excessive mechanical loading enhances cartilage degeneration via NF-κB7, and that the decreased expression of chondrocyte matrix genes due to inflammation is attenuated by Runt-related transcription factor 219. Upregulation of inflammation- and NF-κB-related genes by mechanical overloading in the cartilage is consistent with previous studies and the present findings (Fig. 7a,b). Notably, a similar level of upregulation was observed in the synovium (Fig. 6a,b). The inflammatory response to mechanical overloading in the synovium may contribute to the induction of Inhba. In both the synovium and cartilage, the restoration of the enhanced inflammation was impaired by repeated overloading compared with that following single overloading. Given these findings, taking appropriate rest following single overloading, and the avoidance of repeated overloading, may contribute to inhibiting the accumulation of inflammation and maintaining the articular cartilage.

In addition to Inhba in the synovium, four genes were identified in the cartilage as upstream factors that induced changes in cartilage gene expression in the R24 group. Cxcl2 encodes a well-known chemokine that enhances inflammatory responses and recruits neutrophils. It is believed to contribute to the onset of arthritis and cartilage degeneration20. Ccl4 has been reported to be upregulated in the cartilage tissue of patients with OA and to promote chondrocyte apoptosis and ECM degradation via the NF-κB pathway21. While there have been no reports regarding the specific role Nrg2 plays in cartilage, it has been identified as a ligand for ErbB3 and ErbB422, suggesting a potential role in regulating chondrocyte proliferation and differentiation through these receptors23. Pgf has been implicated in cartilage turnover and repair, as studies using fracture models in Pgf-deficient mice suggest it plays a role in cartilage regeneration24. However, as a member of the vascular endothelial growth factor (VEGF) family, Pgf also contributes to inflammation by promoting angiogenesis and the recruitment of inflammatory cells. Taken together, these findings suggest that repeated overloading triggers degenerative processes leading to cartilage degradation, although it may also activate mechanisms that can potentially contribute to cartilage repair.

For single overloading, Nr4a3 was identified as the most significantly upregulated gene in both the synovium and cartilage 1 h after running. Nr4a3, a protein of the orphan nuclear receptor Nr4a family, is activated as an early stress factor under various pathological conditions and plays a regulatory role in various tissues and cells, participating in processes such as cell differentiation, apoptosis, metabolism, and homeostasis25. Ma et al. reported high expression levels of NR4A3 in cartilage samples from human patients with OA26. They also demonstrated that Nr4a3 enhanced the inflammatory responses of chondrocytes in vitro, primarily via the NF-κB pathway26. Although the role of Nr4a3 in OA pathophysiology remains unclear, it is intriguing that the mRNA level of Nr4a3 increased in both the synovium and cartilage immediately after single overloading in the present study (Supplementary Fig. 3a, 4a). Notably, Saa3 was identified as an upregulated DEG from S1 to R24 in both the synovium and cartilage (Supplementary Fig. 3f., 4f., 5b, d). Serum amyloid A3, encoded by Saa3, is induced by IL-1β in rabbit chondrocytes, and it promotes Mmp13 expression and triggers cartilage degeneration27. Considering the other genes that are upregulated following single overloading, further investigations should be undertaken to explore the complete effects of single overload on joints.

Our study had several limitations. First, we did not investigate further the effects of treadmill running without attachment of weight, which might have allowed us to distinguish whether the observed changes were caused solely by running or by the magnitude of the weight load. Second, the sample size was small. To harvest sufficient mRNA for RNA-seq, we had to combine synovial tissues from six knee joints to make up one sample and cartilage tissues from nine knee joints to make up one sample. However, we were unable to increase the number of mice due to the limited number of treadmill lanes. Finally, we could not explore the mechanisms underlying the induction of Inhba by overloading and the duration of Inhba upregulation following the final session of running on the treadmill. Future investigations into the upstream regulators and precise time-course alterations of Inhba expression may contribute to identifying therapeutic targets for mitigating inflammation-related upregulation following excessive exercise.

In conclusion, the present study revealed the transcriptomic changes that occur in the synovium and cartilage following single and repeated mechanical overloading, with prolonged inflammation in both types of tissues and downregulation of chondrogenic genes in the cartilage. Inhba may be responsible for some of the synovium-induced changes in the cartilage. Although we were unable to perform experiments to confirm the roles of Inhba or activin A in overloaded joints in vivo, further studies into the mechano-responses of the articular joints and the role played by Inhba may suggest novel therapeutic approaches to maintaining joint homeostasis.

Methods

Animals

The mice used in this study were housed in a specific pathogen-free facility under the supervision of the Institutional Animal Care and Use Committee (IACUC). The environment was maintained at a constant temperature (18–22 °C) with a 12-h light/dark cycle. No exclusion criteria were applied in this study.

For all in vitro and in vivo experiments requiring wild-type (WT) mice, C57BL/6 J WT mice were purchased from Sankyo Labo Service Corporation (Tokyo, Japan). In the overload model experiments, we used male mice from 10 to 13 weeks of age; for the in vitro studies, we used 5-day-old mice.

Overload mice model

The overload model, using the 10-week-old male mice, was initiated after a one-week acclimation period. The mice were trained on a treadmill, starting at a speed of 12 m/min and gradually increasing to 20 m/min over 5 days. After a two-day rest period, the mice were anesthetized using inhaled anesthesia, and lead tape was wrapped around their trunk. They then underwent treadmill running at 12 m/min for 10 min, with progressive increases in duration, over a period of 5 days. After another two-day rest period, the mice ran for 60 min at 12 m/min. Samples were collected from the mice at 1, 4, and 24 h post-running. These mice comprised the single overloading group. For the repeated overloading group, the mice ran for 60 min at 12 m/min for 4 consecutive days, and samples were collected 24 h after their final run. All mice were subsequently euthanized by CO₂ inhalation.

Histological analysis and immunohistochemistry

Tissue samples were fixed in 4% paraformaldehyde in phosphate-buffered saline (pH 7.4) and subsequently degreased with 100% ethanol at room temperature. The samples were decalcified in EDTA (pH 7.4) at 4 °C for 5–6 days with continuous agitation. After decalcification, the tissue samples were embedded in paraffin.

Sagittal sections of 4-µm thickness were prepared from the medial compartment and intercondylar regions of the knee joint. These sections were used for Safranin O staining and immunohistochemical analysis. Safranin O staining was performed according to standard protocols, and images of all stained slides were captured using a NanoZoomer S60 (Hamamatsu Photonics, Shizuoka, Japan). For immunohistochemical staining, the tissue sections were incubated overnight at 4 °C with anti-activin A antibody (5 µg/mL, AF-338, R&D Systems, USA). After incubation with horseradish peroxidase-conjugated secondary antibodies for 30 min at room temperature, positive signals were visualized using Alexa Fluor (B40958, Thermo Fisher Scientific, Waltham, MA, USA). All images were captured using a BZ-X710 microscope (KEYENCE, Osaka, Japan). The signal-positive area in the lining layer of synovium was measured and compared using QuPath28.

RNA-seq analysis

Following euthanasia of the overload model mice via CO2 inhalation, synovium and cartilage tissues were harvested from their knees. For RNA-seq, synovium samples obtained from six knee joints were combined into one sample, while cartilage samples obtained from nine knee joints were combined into one sample. In total, 18 mice per group underwent the treadmill running, while 27 mice were used for the control group. Harvested tissues were homogenized, and total RNA was extracted using TRI Reagent (Molecular Research Center, Inc.). Total RNA was further purified using a Direct-zol RNA Kit (#R2062, Zymo Research, Irvine, CA), according to the manufacturer’s protocol.

Sequencing was performed on an Illumina NovaSeq 6000 system with two 150-bp paired-end reads. After adapter sequences had been removed using fastp (ver.0.23.4)29, the resulting RNA reads were mapped to the mouse reference genome using STAR version 2.7.8a30, with the reference index created from GRCm39.primaryassembly.genome.fa and gencode.vM33.annotation.gtf, both obtained from GENCODE31. Expression levels were calculated using RSEM (ver.1.3.1)32. Differential gene expression analysis was conducted using the web application iDEP33, comparing data across different time points, and PCA was performed. DEGs were identified using the default iDEP thresholds (FDR < 0.1, fold change > 2). GO analysis was performed via Metascape34, and IPA35 was employed to identify upstream cytokines involved in transcriptome changes in the cartilage.

Next, characteristic genes at each time point were extracted to investigate changes in their expression. Gene counts were normalized using DESeq236, and 17,316 genes from the synovium and 18,105 genes from the cartilage were identified. Further analyses focused on protein-coding genes, and genes with FDR < 0.1 and fold change > 1.5 or < 0.67 at any time point relative to the control were selected. Finally, 5,114 genes from the synovium and 4,904 genes from the cartilage were subjected to our analysis. For GO molecular function terms, genes associated with “Cytokine activity” (fold change > 2) were extracted. For GO biological processes terms, genes linked to “Inflammation” (fold change > 3 or < 0.33), “Nf-kappaB” (fold change > 2 or < 0.5), and “Cartilage development” (fold change > 1.5 or < 0.67) were selected and visualized as heatmaps. Additionally, heatmaps for representative ECM-related genes were generated (fold change > 1.5 or < 0.67).

Data visualization was performed using R version 4.1.1 (R Core Team, 2021). Heatmaps were generated using the circlize37 and ComplexHeatmap38 packages. Volcano plots were generated using the EnhancedVolcano package39.

scRNA-seq

Previously deposited scRNA-seq datasets from acute injury model mice (GSE211584)9 were utilized. Data from the sham and ACLR groups at day 7 were analyzed using Seurat (version 4.3.0.1). During Seurat object generation, cells with fewer than 200 detected genes and genes expressed in fewer than 3 cells were excluded. Next, low-quality cells were filtered out based on the criteria: 500 < nFeature_RNA < 6,000. Cells in which more than 10% of all genes were mitochondrial-derived (percent.mt > 10) were also excluded.

After merging the two Seurat objects, SCTransform40 was used to perform linear transformation and to regress out unwanted variation due to cell cycle and mitochondrial gene expression. Linear dimensionality reduction was performed using PCA. Unsupervised clustering was carried out using FindNeighbors (dims = 1:30) and FindClusters (resolution = 0.1) to identify distinct cell clusters. Dimensionality reduction was visualized using uniform manifold approximation and projection (UMAP).

Clusters corresponding to chondrocytes (characterized by high expression of Acan, Col2a1, and Sox9) were excluded. After re-clustering (dims = 1:30, resolution = 0.1) was performed, cell annotation for each cluster was conducted according to the previously described method9. Finally, cells expressing Inhba were visualized using FeaturePlot.

Cell culture

Articular chondrocytes were isolated from the cartilage of 5-day-old WT mice. Cartilage tissue specimens were subjected to consecutive enzymatic digestions with proteinase and collagenase I, as previously described41. Briefly, the distal femur and proximal tibia were dissected from 5-day-old mice, and ligaments and tendons were removed. The cartilage tissues were incubated with 0.25% trypsin (Thermo Fisher Scientific, Waltham, MA, USA) for 1 h, followed by digestion with 200 U/mL of type I collagenase (Worthington Biochemical Corporation, Lakewood, NJ, USA) for 1.5 h. Primary chondrocytes were cultured in Dulbecco’s modified Eagle medium (DMEM) with 10% fetal bovine serum, and the medium was changed every 3 days.

Once the cells had reached sufficient numbers, they were treated with recombinant human activin A (338-AC, R&D Systems, USA) at determined concentrations (0 ng/mL and 200 ng/mL) in DMEM containing HEPES. At 1, 4, 24, and 48 h following administration of recombinant human activin A, total RNA was extracted using TRI Reagent.

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

The extracted RNA was reverse transcribed into complementary DNA using ReverTra Ace qPCR RT Master Mix (#FSQ-201, Toyobo, Osaka, Japan). Then, qRT-PCR was performed using THUNDERBIRD SYBR qPCR Mix (#FSQ-201, Toyobo) and the Thermal Cycler Dice Real-Time System III (Takara Bio, Kusatsu, Japan). The mRNA levels of the target genes were normalized to those of hypoxanthine phosphoribosyltransferase 1 (Hprt1). Gene expression levels were calculated using the ΔΔCt quantification method. All reactions were performed in triplicate using the Thermal Cycler Dice System (Takara Bio). The primer sequences used in this study are provided in the Supplementary Information (Supplementary Table 1).

Statistics and reproducibility

The results were analyzed using a bell curve chart in Excel (Social Survey Research Information Co., Ltd., Japan). One-way ANOVA followed by Dunnett’s post hoc test was used to detect significant differences. P-values of less than 0.05 were considered to be significant. Biologically independent sample numbers are shown in the figures and figure legends.