Stage-Dependent Effects of Moderate Treadmill Exercise on Cartilage Preservation and Subchondral Bone Remodeling in Mouse Osteoarthritis Progression

Abstract

Objective

Exercise is widely advocated for osteoarthritis (OA) treatment; however, its effectiveness across disease stages, particularly in advanced OA, remains inconclusive. This study assessed the impact of treadmill exercise at distinct OA stages to determine optimal intervention timing.

Methods

Following validation of a moderate treadmill protocol, 96 male C57BL/6J mice underwent destabilization of the medial meniscus (DMM) surgery on the right knee and sham surgery on the left. Mice were randomly assigned to sedentary (SED) or treadmill exercise (RUN) groups, starting at 3 days, 4, 8, or 12 weeks postoperatively. At each timepoint, bilateral knees were analyzed by micro-computed tomography, histo-morphometry, immunohistochemistry, and osteoblast/osteoclast function.

Results

Exercise initiated 3-day post-surgery preserved cartilage integrity (median OARSI: 1.0 in RUN vs. 2.0 in SED) and reduced subchondral bone (SB) lesions (BV/TV: 76.6% in RUN vs. 81.8% in SED) and osteophyte volume (0.015 mm³ in RUN vs. 0.026 mm³ in SED). At four weeks, articular cartilage thickness remained higher in exercised joints (86.1 μm in RUN vs. 78.5 μm in SED), though SB deterioration and osteophyte formation persisted. By eight weeks, exercise exacerbated articular cartilage degeneration (41.5 μm in RUN vs. 62.3 μm in SED). At twelve weeks, further deterioration was observed, including reduced calcified cartilage cellularity and increased SB thickness (541.9 μm in RUN vs. 519.4 μm in SED), suggesting delayed exercise may worsen joint pathology.

Conclusions

Treadmill exercise offers protective effects when initiated early after OA onset but may worsen outcomes if delayed. Tailored exercise protocols based on disease stage are essential for optimal OA management.

Introduction

Osteoarthritis (OA) is a chronic degenerative joint disorder characterized by the progressive deterioration of cartilage and concurrent alterations in subchondral bone architecture, culminating in impaired joint function¹. With demographic shifts toward an aging populace and increasing rates of obesity, OA has emerged as the predominant arthritic form, afflicting a substantial portion of the population, and posing a significant burden on societal and individual levels^{2, 3}. Current treatments for OA are primarily palliative, with surgical interventions remaining the sole option for managing advanced stages of the disease^{2, 3}. Consequently, there exists a pressing need for advancements in strategies aimed at mitigating OA progression, deepening understanding of its intrinsic pathological mechanisms, and identifying novel therapeutic modalities to enhance treatment outcomes.

In addressing the imperative of impeding OA advancement, clinical guidelines endorse therapeutic exercise interventions^1–3. Various physical exercise regimens, notably aerobic and resistance training exercises, have demonstrated efficacy in ameliorating symptoms, augmenting joint functionality, and enhancing overall quality of life for OA sufferers^{2, 4–7}. Recent scholarly endeavors have witnessed a surge in investigations employing animal models of OA to elucidate the effects and mechanisms of exercise on osteoarthritic joints^8–12. Accumulated evidence suggests that mechanical loading, particularly through moderate exercise, exerts modulatory effects on joint constituents, thereby influencing joint remodeling processes and conferring protective benefits against cartilage degradation and subchondral bone remodeling^{10, 12–14}. Meanwhile, excessive mechanical loading, typified by high-impact sports activities, may exacerbate damage in cartilage and subchondral bone, accelerating to the progression of OA^{9, 12, 15–17}.

Nevertheless, a significant gap exists in the literature regarding the impact of exercise interventions on different OA stages. Current animal studies predominantly initiate exercise protocols immediately or shortly after OA induction, which do not faithfully replicate the clinical chronic progression of the disease^{11, 12}. Exercise interventions enacted within such truncated timeframes are more aptly construed as preemptive measures against the advancement of early-stage osteoarthritis or as rehabilitative strategies following acute ligamentous or meniscal injuries, rather than addressing the natural development of OA¹³. There is limited understanding of the efficacy of regular exercise protocols, even moderate ones, across various stages of OA progression, particularly in late or end-stage OA, which holds greater clinical significance as the majority of symptomatically diagnosed OA cases occur during these advanced stages^{1, 5, 18}.

The aim of this study was to explore the effects of moderate exercise across distinct stages of OA progression utilizing the established mouse destabilization of the medial meniscus (DMM) model¹⁹. It is postulated herein that moderate exercise may confer advantageous effects during the early stages of OA, while offering limited utility or potentially adverse consequences during later and end-stage OA.

Methods

Animal models

All animal procedures were approved by the IACUC of Thomas Jefferson University (Protocol No. 01830) and adhered to NIH and ARRIVE guidelines. A total of 112 male C57BL/6J mice (12 weeks old, ~25 g) from Jackson Laboratory (Bar Harbor, ME) were housed in an AAALAC-accredited facility. DMM model was induced in the right knee via transection of the medial menisco-tibial ligament (MMTL) through a 3–5 mm medial parapatellar incision^{19, 20}. The contralateral knee received the same exposure without ligament transection and served as a sham control.

Validation of the moderate treadmill exercise protocol in normal mice

To evaluate the impact of the treadmill exercise protocol on sham-operated knees, 16 male mice were randomly assigned to either the Sham + Treadmill Exercise (RUN, R) group (n = 8) or the Sham + Sedentary (SED, S) group (n = 8). Mice were acclimated to treadmill running (GUO BASILE S.R.L., Italy, Cat. No. 47300) using a mild electric stimulus (1 mA, 3 Hz). Those receiving >20 shocks in three sessions across the acclimation were excluded. A 4-week moderate treadmill protocol (12 m/min, 30 min/day, 5 days/week, 10° incline) began three days postoperatively²¹. Upon completion right knees were harvested and assessed via micro-computed tomography and histology for OARSI, subchondral bone volume fraction (SB.BV/TV), and subchondral bone thickness (SB.Th) at the medial tibial plateau (MTP) based on a defined region of interest (ROI)²² (Supplementary Fig. 1).

Treadmill exercise protocol after various stages of DMM surgery

To investigate the temporal effects of exercise on OA progression, 96 male mice underwent DMM surgery and were randomized into exercise or sedentary groups. The mice in the treadmill exercise groups (n = 12/group) initiated a 4-week treadmill running (4wR) protocol at 3 days (DM3d+4wR), 4 weeks (DM4w+4wR), 8 weeks (DM8w+4wR), and 12 weeks (DM12w+4wR) post-DMM surgery. In contrast, sedentary groups (n = 12/group) remained inactive with corresponding time-points (DM3d+4wS, DM4w+4wS, DM8w+4wS, and DM12w+4wS). (Fig. 1).

Fig. 1. Flowchart of DMM and treadmill exercise protocols.

Eight groups of 12-week-old male C57BL/6J mice (n=12/group) underwent DMM surgery on the right knee and sham surgery on the left. The mice were randomly assigned to sedentary groups (SED) and treadmill exercise (RUN) groups. A 4-week treadmill running protocol at 12 m/min, 30 min/d, 5d/wk, with a 10° incline, was initiated at 3 day, 4 weeks, 8 weeks, and 12 weeks post-DMM surgery in the treadmill exercise groups, allowing for analysis of the effects of different exercise initiation times on knee joint health.

At the end of each 4-week intervention, knees were collected for the following evaluations: (1) Safranin O/Fast Green and Hematoxylin/Eosin staining (n = 12/group), and immunohistochemical (IHC) analysis of Type II collagen (Col II) and matrix metalloproteinase-13 (MMP13) (n = 8/group) to assess cartilage changes; (2) micro-CT imaging (n = 12/group), and tartrate-resistant acid phosphatase (TRAP) and alkaline phosphatase (ALP) staining (n = 8/group) to analyze subchondral bone remodeling.

Micro-CT scanning

Knee joints were stored at −80 °C prior to imaging. High-resolution scans were obtained using a SkyScan 1275 system (Bruker, Kontich, Belgium), and data were reconstructed and analyzed with CT Analyzer software (Version 2.3.2.0) based on prior methodologies²³. The ROI was defined as the medial 20% of the whole width of the proximal tibial plateau in the coronal plane²² (Supplementary Fig. 1). Parameters assessed included SB.BV/TV, SB.Th, and osteophyte volume²⁴.

Safranin O/Fast green and Hematoxylin/Eosin staining

Following scanning, knees were fixed in 4% paraformaldehyde for 48 hours and decalcified in 12% EDTA for 2 weeks. Sagittal sections (6 μm) were stained with Safranin O and Fast Green, and separately with Hematoxylin and Eosin. Histological alterations in the MTP were evaluated by OARSI, as well as the thickness (Th) and cellularity of both articular cartilage (AC) and calcified cartilage (CC) in the delineated region (Supplementary Fig.2) using ImageJ, following previously established methods^{11, 12, 25}.

IHC staining for collagen II and MMP13

Paraffin sections were immunolabeled with anti-collagen II (Abcam, Cat. No. ab34712, 1:100) or anti-MMP13 (Abcam, Cat. No. ab39012, 1:100) antibodies. Detection was performed using the streptavidin-biotin-peroxidase system (Vector Laboratories) and ImmPACT DAB chromogen. Images were acquired using a light microscope (BX53; Olympus, Tokyo, Japan). Type II collagen expression was semi-quantitatively analyzed by measuring the pixel intensity values (0–255)¹³ in defined cartilage ROI (Supplementary Fig.2), while the percentage of MMP13-positive chondrocytes was calculated from at least three representative 20x fields per sample²⁷.

Assessment of osteoclast and osteoblast activity

Subchondral bone remodeling was evaluated via histological staining for TRAP (Sigma-Aldrich, Cat. No. 387A) and ALP (NBT/BCIP substrate, Thermo Scientific, Cat. No. 34042), with nuclear counterstaining using Nuclear Fast Red (Thermo Scientific, Cat. No. J61010.AP). Imaging was performed at 4×, 20×, and 40× magnifications (Supplementary Fig.2). TRAP-positive osteoclasts and ALP-positive osteoblasts were manually counted in three randomly selected high-power fields (20× for TRAP, 40× for ALP) and normalized to bone surface area (cells/BS)²⁸. Correlations between osteoblast/osteoclast parameters (Ob/BS, Oc/BS, Ob/Oc ratio) and SB.Th were further analyzed to evaluate the relationship between cellular activity and structural adaptation.

Data analysis and statistics

Data was independently collected by two investigators and analyzed in a blinded manner using GraphPad Prism (version 9) and JMP Student Edition (version 18.2.1, SAS Institute Inc.). Sample sizes were based on previous studies^{11, 29}. Data normality and variance homogeneity were assessed using the Shapiro–Wilk and Levene’s tests, respectively. Student’s t-test or Mann–Whitney U test was used for comparisons between sham groups. For post-DMM time points, mixed-effects models were employed, with “mouse ID” specified as a random effect to account for repeated measures within individual animals and “group” as a fixed effect³⁰. Post hoc analyses were conducted following significant main effects: Tukey’s honestly significant difference (HSD) test was used for normally distributed data, while Dunn’s test was applied for ordinal or non-normally distributed outcomes (Appendix table 2). Pearson correlation analysis was used to evaluate associations between continuous variables. A two-tailed p value < 0.05 was considered statistically significant.

Results

1. Moderate Treadmill Exercise Does Not Induce Structural Changes in Sham-Operated Mice

No significant differences were observed between the two sham-operated groups in OARSI, SB.BV/TV and SB.Th (Fig. 2). The lack of substantial changes in these critical metrics suggests that under the specified experimental conditions, moderate treadmill exercise protocol did not induce structural alterations in the MTP of otherwise healthy mice.

Fig. 2. Effects of 4-week sedentary vs. moderate treadmill exercise on MTP in sham surgery mice.

The results indicate that the treadmill exercise protocol had no significant impact on the medial tibial plateau (MTP) of knee joints in sham-operated mice, as evidenced by the lack of significant differences in the OARSI score (Top row, scale bar: 100 μm), subchondral bone volume fraction, and thickness (Middle row, scale bar: 500 μm ) between the sedentary and the treadmill exercise groups. n = 8/group, statistical significance was defined as p<0.05.

2. Differential Effects of Treadmill Exercise on Cartilage at Various Stages of OA

One mouse from each of the DM4w+4wR, DM8w+4wR, and DM12w+4wR groups (total n = 3) was excluded due to noncompliance with the exercise regimen. Histological analysis revealed that the effects of treadmill exercise after DMM surgery on cartilage integrity differed notably depending on the stage of OA progression (Fig. 3).

Fig. 3. Safranin O and H& E staining of the knee joints at different stages of DMM surgery.

Histological analyses of articular and calcified cartilage in sedentary (S) and treadmill exercise (R) groups at various stages following DMM surgery. The left panel displays representative images of safranin O (A) and hematoxylin and eosin (H&E) (B) staining of knee joints at different time points post-DMM surgery. The right panel quantitatively illustrates the OARSI scores, thickness, and cell density of articular cartilage (AC.Th, AC.Cell density), as well as the thickness and cell density of calcified cartilage (CC.Th, CC.Cell density) (C). (n = 12/group, n= 11 for the DM4w+4wR, DM8w+4wR and DM12w+4wR group due to one mouse in each group not participating in running. statistical significance was defined as p<0.05. Scale bar: 100 μm.)

Early initiation of treadmill exercise (DM3d+4wR, OARSI median: 1.0, IQR [0.5–1.5]) improved cartilage integrity compared to the sedentary group (DM3d+4wS, OARSI median: 2.0, IQR [2.0–2.5]). Although both DMM groups had higher OARSI than sham controls (OARSI median: 0.5, IQR [0.0–1.0]), treadmill exercise (DM3d+4wR) restored AC thickness (92.6±6.7μm) relative to DM3d+4wS group (85.5±7.0μm) and closely approximated that of the sham control (97.5±5.9μm). No significant differences were found among groups in the CC thickness (CC.Th) or cellularity in either the AC or CC layers, collectively supporting the protective role of early exercise intervention.

When exercise commenced 4 weeks after DMM, the protective effects were less pronounced. OARSI in both the exercise group (DM4w+4wR, median: 2.5, IQR [1.5–3.0]) and sedentary controls (DM4w+4wS, median: 2.75, IQR [2.0–3.5]) remained significantly elevated versus sham (median: 0.75, IQR [0.0–1.0]) without intergroup difference. However, a trend of preservation of AC.Th was noted in the exercise (DM4w+4wR, 86.1±4.4μm) group compared to the sham group (93.7±8.0μm), while AC.Cellularity were significantly higher in the DM4w+4wR group (65.4±4.8/10⁴μm²) than the sedentary group (54.5±10.4/10⁴μm²). CC parameters remained unchanged, indicating a limited impact of OA on deeper cartilage layers at this stage.

In contrast, treadmill exercise initiated at eight weeks post-DMM appeared to exacerbate joint damage. AC.Th was significantly reduced in the exercise group (41.5±10.7μm) compared to the sedentary group (62.3±12.3μm) with a trend toward lower AC.Cellularity (p=0.0582) (Fig.3), suggesting that delayed exercise may accelerate AC lesions. Both groups showed a reduction in CC.Th and CC.Cellularity compared to sham controls, indicating extensive CC damage by this stage.

At 12 weeks post-DMM (DM12w+4w), minimal AC remained in both exercise (25.0±17.2μm) and sedentary groups (34.2±18.3μm). However, while both groups still exhibited reduced CC.Th and CC.Cellularity compared to sham controls, the DM12w+4wR group (13.5±2.7/10⁴μm²) demonstrated significantly lower CC.Cellularity than the DM12w+4wS group (16.9±3.2/10⁴μm²), indicating detrimental effects of late-stage exercise on joint cartilage integrity (Fig.3).

3. Stage-Specific Effects of Treadmill Exercise on Subchondral Bone Remodeling

Micro-CT analysis revealed distinct effects of exercise on SB remodeling depending on the timing of intervention during different stages of OA (Fig. 4).

Fig. 4. Micro-CT Analysis of Subchondral Bone.

The micro-CT analysis of the medial tibial plateau (MTP) of the knee joints, examining the effects of treadmill exercise at various stages following DMM surgery. (A) The upper panel showcases representative images of coronal sections of the tibial plateau after 4 weeks of either sedentary behavior (4wS) or treadmill exercise (4wR) at specified time points (3 day, 4 weeks, 8 weeks, and 12 weeks post-surgery). The left knee joints of mice in the sedentary (SED) groups were used as controls. (B) The lower panel quantitatively measures subchondral bone volume fraction (SB.BV/TV), subchondral bone thickness (SB.Th) and osteophyte volume. (n = 12/group, n = 11 for the DM4w+4wR, DM8w+4wR and DM12w+4wR group due to one mouse in each group not participating in running. statistical significance was defined as p<0.05. Scale bar: 500 μm.)

In the early post-DMM phase (DM3d+4wS), sedentary mice showed increased SB.BV/TV (81.8±3.4%) and reduced SB.Th (446.8 μm) compared to sham controls, suggesting early remodeling driven by altered mechanical stress¹⁹. In contrast, early exercise (DM3d+4wR) preserved SB structure with SB.BV/TV of 76.6±4.6% and SB.Th of 463.5±12.3 μm, approximating sham values. Osteophyte volume was significantly lower in exercised animals (0.015±0.008mm³ vs. 0.026±0.011mm³ in sedentary mice), underscoring the protective effect of mechanical stimulation initiated soon after DMM.

At four weeks post-DMM, the sedentary group (DM4w+4wS) exhibited increases in SB.BV/TV (80.1±3.8%) and osteophyte volume (0.050±0.018mm³), but SB.Th remained relatively preserved (475.1±26.5μm) compared to the sham control. Exercise initiated at this time failed to confer meaningful structural protection, as SB parameters in the DM4w+4wR group remained comparable to sedentary animals. By eight weeks post-surgery, both sedentary (DM8w+4wS) and exercised (DM8w+4wR) groups exhibited pronounced SB sclerosis and osteophyte growth, with no measurable protective effect of exercise, indicating that exercise could not fully reverse the bone alterations at this time point.

At 12 weeks post-DMM, the sedentary group (DM12w+4wS) displayed severe degenerative changes, with significant increases in both SB.BV/TV, SB.Th and osteophyte volume. Treadmill exercise initiated at this stage (DM12w+4wR) failed to ameliorate these alterations but in contrast, exacerbated SB.Th increases (541.9±19.1μm vs. 519.4±21.2μm in sedentary mice), suggesting that treadmill intervention at this late OA stage may worsen subchondral sclerosis

4. Cartilage Matrix Alterations with Exercise Reflect Collagen II and MMP13 Expression Dynamics

Immunohistochemical analysis revealed that treadmill exercise modulated cartilage matrix composition in a time-dependent manner (Fig. 5A). In early OA (DM3d+4w), collagen II staining intensity in the exercise group (86.5±22.0) was significantly higher than in sedentary controls (60.7±10.4), approaching levels observed in sham mice. This finding indicates preservation of extracellular matrix under early mechanical loading. A similar, though nonsignificant, trend was noted in the DM4w+4wR group (p = 0.0749), suggesting that timely exercise may partially slow collagen II loss. In contrast, at later stages (DM8w+4w and DM12w+4w), all DMM groups exhibited reduced collagen II staining, with the DM12w+4wR group showing lower intensity (20.1±12.8) than corresponding sedentary mice (39.3±8.0), indicating exercise-induced exacerbation of matrix degradation in advanced OA.

Fig. 5. IHC staining of Collagen II and MMP13 of the knee joints in Sedentary (SED) and Treadmill Exercise (RUN) Groups at Various Stages Following DMM Surgery.

(A) Representative images of collagen II staining at various stages following DMM surgery were presented in the left top panel. Relative staining intensity of collagen II was illustrated in the right panel. (B) Representative images of MMP13 staining at various stages following DMM surgery were presented in the left bottom panel. The percentage of MMP13-expressing chondrocytes was calculated relative to the total number of cells in at least three 20x fields per sample and was illustrated in the right panel. n =8/group, statistical significance was defined as p<0.05. Scale bar: 100 μm.

MMP13 expression, a marker of catabolic activity, further supported these findings (Fig. 5B). In the DM3d+4wS group, the proportion of MMP13-positive chondrocytes was significantly elevated (12.4±3.6%) compared to sham (3.7±1.5%), whereas early exercise (DM3d+4wR) reduced this percentage to 5.0±1.3%. Similarly, in the DM4w+4w stage, exercise decreased MMP13-positive cells (6.7±1.6% in RUN vs. 14.3±4.7% in SED). In contrast, at eight and twelve weeks post-DMM, both exercised and sedentary groups showed persistent elevation of MMP13+ cells, indicating persistent catabolic activity with no beneficial modulation by exercise.

5. Subchondral Bone Cellular Responses to Treadmill Exercise: Osteoclast and Osteoblast Activity

Tartrate-resistant acid phosphatase (TRAP) staining revealed that early treadmill intervention significantly reduced osteoclast activity in the SB compartment (Fig. 6). Osteoclast numbers in the DM3d+4wR group (6.3±2.1/mm) were significantly lower than in sedentary controls (14.5±3.4/mm) and were comparable to sham levels (3.3±1.7/mm), suggesting exercise may suppress bone resorption in early OA. However, at four, eight, and twelve weeks post-DMM, exercise had no significant effect on osteoclast counts, which remained elevated compared to sham.

Fig. 6. TRAP staining of the knee joints in Sedentary (SED) and Treadmill Exercise (RUN) Groups at Various Stages Following DMM Surgery.

Representative images at various stages following DMM surgery were presented in the left panel. Number of osteoclasts per bone surface (N.Oc/BS) was illustrated in the right panel. n = 8/group, statistical significance was defined as p<0.05. Scale bar: 100 μm.

Conversely, osteoblast activity (ALP staining) (Fig. 7) remained comparable between the DM3d+4wR and sedentary (DM3d+4wS) groups, with no significant differences compared to sham controls. Notably, osteoblast activity significantly increased in both DM4w+4w and DM8w+4w groups relative to the sham control, independent of timing of exercise. At twelve weeks post-DMM, osteoblast counts were higher compared to sham, with a trend toward further elevation in the exercised (DM12w+4wR) group (p = 0.0686), implying that late-stage exercise may exacerbate osteoblast-driven bone formation and sclerosis.

Fig. 7. ALP staining of the knee joints in Sedentary (SED) and Treadmill Exercise (RUN) Groups at Various Stages Following DMM Surgery.

Representative images at various stages following DMM surgery were presented in the left panel. Number of osteoblasts per bone surface (N.Ob/BS) was illustrated in the right panel. n = 8/group, statistical significance was defined as p<0.05. Scale bar: 100 μm.

The osteoblast-to-osteoclast ratio (OB/OC), illustrated in Fig. 8, was significantly lower in the DM3d+4wS group compared to sham, while early exercise (DM3d+4wR) partially restored this ratio to near-sham levels. At four weeks post-surgery, the OB/OC ratio was similar across all three groups. In the DM8w+4wR group, however, the OB/OC ratio significantly exceeded that of the sham controls, although no notable difference was observed between the exercise and sedentary groups post-DMM surgery. In the DM12w+4wR group, OB/OC was markedly elevated, surpassing both sedentary and sham controls. While correlations between individual osteoclast or osteoblast activities and SB.Th were not statistically significant, OB/OC was positively and significantly associated with SB.Th, suggesting that a shift toward osteogenesis contributes to subchondral sclerosis during OA progression.

Fig. 8. Osteoblast/Osteoclast ratio at various stages of DMM surgery and correlation analysis with SB.Th.

The top panel illustrates the osteoblast-to-osteoclast ratio (Ob/Oc), calculated as the ratio of osteoblasts per bone surface area (Ob/BS) to osteoclasts per bone surface area (Oc/BS). n = 8/group, statistical significance was defined as p<0.05. The bottom panel presents a correlation analysis, examining the relationship between SB.Th and (i) the number of osteoblasts per bone surface area (Ob/BS), (ii) the number of osteoclasts per bone surface area (Oc/BS), and (iii) the Ob/Oc ratio.

Discussion

The present study investigated the effects of moderate exercise on joint health across various stages of OA progression using the DMM mouse model. Our results demonstrated that the impact of exercise was highly dependent on the timing of its initiation. Notably, the early initiation of treadmill exercise following DMM surgery, implemented as a preventive rather than therapeutic intervention, effectively preserved subchondral bone integrity and reduced cartilage degradation. Parameters such as SB.BV/TV, SB.Th, OARSI, and osteophyte volume closely resembled those of sham-operated controls, suggesting strong protective effects. Initiating exercise at 4 weeks post-surgery also yielded therapeutic benefits, including preserved cartilage thickness, cellularity, increased Col II content, and reduced MMP13 expression, although structural deterioration in subchondral bone and osteophyte formation remained evident compared to sham controls. In contrast, delaying exercise until 8 or 12 weeks post-DMM surgery eliminated its protective effects and, in some cases, worsened joint degeneration, as reflected by elevated OARSI, decreased cartilage integrity, increased SB.Th, and enlarged osteophyte volume. These findings underscore the stage-dependent nature of exercise-mediated joint protection and suggest that early intervention is critical. To our knowledge, this study is the first to explore exercise as a therapeutic intervention for late- or end-stage DMM-induced OA, a focus with significant clinical relevance as most symptomatic OA emerges at these stages^{1–3, 18}.

Extensive research has demonstrated the benefits of moderate running exercise in early-stage OA (Appendix table 1) using murine models; however, its clinical relevance may be limited, as OA is rarely diagnosed at such early stages. Studies by Iijima et al. reported that treadmill walking prevented subchondral bone deterioration in rats^{10, 11}, while Huesa et al. found that early moderate exercise in DMM mice reduced cartilage degradation, delayed osteosclerosis, and improved trabecular microarchitecture¹⁴. In line with these findings, our study demonstrated that early post-DMM treadmill exercise significantly preserved joint integrity by reducing cartilage loss, maintaining type II collagen levels, and suppressing MMP13 expression. Subchondral bone structure was also preserved, likely due to reduced osteoclast activity compared to sedentary controls. Since early OA is often associated with bone marrow lesions from abnormal loading^{31, 32}, our findings, in conjunction with literature^{7, 33}, suggest that early exercise may help normalize joint loading and thereby mitigate disease progression.

Fewer studies have examined the effects of exercise initiated approximately one month after OA induction^{13, 34, 35}. Hao et al. demonstrated that regular treadmill walking (15 m/min) alleviated cartilage-subchondral bone deterioration in post-traumatic osteoarthritis (PTOA)³⁴ rat model. Da Silva et al. found reduced inflammation and improved histology following treadmill intervention 30 days post-meniscectomy³⁵. Similarly, Iijima et al. reported that exercise promoted bone morphogenetic proteins and improved PTOA outcomes¹³. In our study, treadmill exercise initiated four weeks post-DMM preserved cartilage thickness, cellularity, and type II collagen, and reduced MMP13 expression. However, OARSI remained elevated relative to sham controls, and osteophyte volume and osteoclast activity were not significantly improved. These findings suggest that although delayed exercise confers partial protection, its effects are less robust than early intervention. Variations across studies may reflect species-specific disease dynamics and mechanical responses, emphasizing the need for model-specific considerations when translating preclinical findings^{32, 36}.

Although moderate exercise is widely considered beneficial for OA patients, mechanistic data, particularly regarding its impact on late- or end-stage disease, remain limited. Since most symptomatic OA cases are diagnosed at advanced stagesstages⁷, our study specifically evaluated the therapeutic efficacy of moderate treadmill exercise in this context. The results revealed that exercise failed to ameliorate degenerative changes in late-stage OA and, in some cases, even worsened joint pathology, as evidenced by a deterioration in cartilage integrity and an increase in SB.Th and osteophyte volume.

The worsening of joint structures following treadmill intervention in advanced OA likely stems from the excessive mechanical loading imposed on a structurally compromised osteochondral junction^37–39. In advanced OA, the disrupted integrity of the osteochondral junction, resulting in osteochondral leakage, facilitates aberrant fluid, cellular, and cytokine exchange between the subchondral bone and joint cavity^{2, 38, 40}. The mechanical loading introduced by treadmill exercise on the compromised osteochondral junction could potentially exacerbate this disruption by enhancing the exchange process^{38, 41}. This intensified substance exchange may, in turn, contribute to the amplification of the destructive inflammation process and acceleration of disease progression⁴². Additionally, as articular cartilage erodes, the underlying bone is directly exposed to mechanical forces, promoting subchondral sclerosis and osteophyte formation^{41, 43}.

Our findings also underscore the pathological shifts in subchondral bone remodeling that accompany OA progression. In early OA, increased bone resorption leads to thinning of the subchondral plate. However, as disease progresses, this is replaced by an imbalance characterized by excessive OB activity and insufficient OC function, resulting in subchondral sclerosis^{38, 44}. In our study, the reduced OB/OC ratio in early OA transitioned to be markedly elevated in advanced stages, reflecting a pathological shift toward aberrant bone formation. This imbalance is indicative of excessive osteoblastic activity relative to bone resorption, which likely contributes to the development of subchondral bone sclerosis and osteophyte formation, and in turn compromises joint flexibility and disrupts the mechanical environment of the overlying articular cartilage^{45, 46}. These results align with previous reports describing a spatiotemporal uncoupling of bone cell activities during OA progression, wherein initial OC dominance giving way to OB-mediated sclerosis in advanced stages^{37, 41}. This maladaptive remodeling disrupts joint biomechanics and contributes to cartilage degeneration. Collectively, our findings emphasize the active role of subchondral bone in OA pathogenesis and support the therapeutic potential of targeting bone remodeling processes to slow or prevent disease progression.

Clinically, moderate exercise is generally recognized for its benefits in reducing pain, inflammation, and joint dysfunction in OA patients^{2, 7, 49}. However, adherence to exercise guidelines can be challenging due to fears of symptom exacerbation or adverse outcomes⁴⁰. Although recent studies have shown that exercise does not necessarily increase pain or accelerate cartilage damage^{7, 49}, other literature has indicated a higher prevalence of OA, particularly in the lower limbs, among former elite athletes compared to the general population^{16, 50, 51}. Moreover, recent systematic reviews indicate that the effects of exercise on OA vary depending on the type and intensity of exercise regimen^{33, 49}. Although this study did not directly assess behavioral pain, our findings at least suggest that concerns regarding the potential worsening of symptoms in patients with advanced-stage OA are neither negligible nor unwarranted. Our findings, though derived from a preclinical DMM mouse model, support the notion that in the context of advanced joint degeneration, mechanical loading may elicit maladaptive responses. Consequently, these data underscore the need for refinement in clinical trial design, particularly in selecting appropriate patient populations and tailoring exercise modalities, to ensure efficacy while minimizing potential detrimental effects in patients with advanced-stage OA^{1, 33}.

This study has several limitations that warrant consideration. Firstly, the use of rodent models presents challenges, as the quadrupedal locomotion of rodents may not accurately replicate the mechanical loading patterns experienced by upright, bipedal humans³². However, while previous clinical studies have shown that individuals with knee OA are at increased risk of developing contralateral OA due to compensatory mechanical loading patterns^{52, 53}, recent gait analyses suggest that ground reaction forces in the contralateral limb of such individuals may not significantly differ from healthy controls⁴⁶. In our DMM mouse model, no evidence of contralateral knee OA was observed at any postoperative time point (Supplementary Fig. 3). However, caution should be exercised when extrapolating these findings to clinical contexts. In addition, only male mice were used in this study to minimize variability in OA phenotype and mechanical response following DMM surgery. While this approach facilitates greater consistency, it inevitably limits generalizability. Future investigations incorporating female models are essential for understanding sex-specific differences in disease progression and response to exercise interventions. Furthermore, this study employed only a single treadmill exercise protocol, while articular cartilage exhibits dose-dependent responses to mechanical stimuli^{54, 55}. Thus, exploring the effects of varied exercise protocols on OA could provide additional insights. Moreover, OA staging in this study was defined solely based on time post-DMM surgery. This temporal approach may not fully capture the histopathological or biomechanical complexity of OA progression in clinical populations. Additionally, the use of both knees (Sham and DMM) from the same animal introduces statistical non-independence due to repeated measures. Although this design reduces inter-animal variability and adheres to animal welfare principles, it introduces within-subject dependency that necessitates appropriate statistical handling³⁰. Future research may consider employing independent group designs to eliminate potential confounding due to repeated measures. Lastly, in vivo assessments of joint pain, function, and locomotor behavior were not included, which could provide a more comprehensive understanding of OA stages and should be incorporated in future research⁵⁶.

Conclusion

This study highlights the pivotal role of exercise timing in OA management. While early intervention preserves joint integrity and attenuates disease progression, delayed exercise, particularly in late-stage OA, may not offer the same advantages and can potentially be detrimental. These findings support the need for stage-specific, personalized exercise strategies to optimize therapeutic outcomes across the full spectrum of OA progression.

Data availability

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