Article Text


Extended report
Inhibition of bone resorption blunts osteoarthritis in mice with high bone remodelling
  1. Abderrahim Kadri1,
  2. Thomas Funck-Brentano1,
  3. Hilène Lin1,
  4. Hang-Korng Ea1,
  5. Didier Hannouche2,
  6. Caroline Marty1,
  7. Frédéric Lioté1,
  8. Valérie Geoffroy1,
  9. Martine E Cohen-Solal1
  1. 1INSERM U606, Centre Viggo Petersen and Université Paris-Diderot Paris 7, Hôpital Lariboisière, Paris, France
  2. 2UMR-CNRS 7052, Centre Viggo Petersen and Université Paris-Diderot Paris 7, Hôpital Lariboisière, Paris, France
  1. Correspondence to Dr Martine E Cohen-Solal, INSERM U606, Centre Viggo Petersen and Université Paris-Diderot Paris 7, Hôpital Lariboisière, 2 rue Ambroise Paré, Paris 75010, France; martine.cohen-solal{at}


Background Osteoarthritis (OA) is characterised by cartilage degradation and bone lesions. Subchondral bone may be involved in the pathogenesis of cartilage matrix breakdown.

Objective To assess the role of bone remodelling in OA by studying the effect of bisphosphonate on OA development in mice with high bone remodelling.

Methods Mice overexpressing Runx2 (Runx2-Tg) under the control of collagen type I that displayed high bone remodelling were used. Joint instability was performed by partial medial meniscectomy to induce OA.

Results Six weeks after surgery, tibial cartilage of Runx2-Tg mice displayed an increased number of ADAMTS-4- and ADAMTS-5-expressing chondrocytes compared with controls (p<0.05). This increase was higher in Runx2-Tg mice than in wild-type mice, although their OA score did not differ (2.5±0.6 vs 2.4±0.2, P=NS). Pamidronate reduced the OA score in Runx2-Tg mice but not in wild-type littermates (1.2±0.5 vs 2.7±0.4; p<0.05) despite the reduction of bone resorption and of the expression of cartilage proteases in both genotypes.

Conclusions These findings support the hypothesis that the level of bone resorption influences cartilage metabolism and that inhibition might prevent the progression of OA. Targeting bone resorption might therefore provide an approach to the treatment of high bone resorbing forms of OA.

Statistics from


Osteoarthritis (OA) is a disorder that leads to the degradation of cartilage and osteophyte formation as a result of accumulated mechanical stress. Remodelling of cartilage is regulated through the activity of chondrocytes, which produce both catabolic and anabolic proteins that maintain the balance between the breakdown and synthesis of extracellular matrix.1 In OA, cartilage loss results from several mechanisms, including biomechanical and metabolic factors.2 Little is known about the underlying molecular mechanisms, although proteases such as ADAMTS have been shown to have a major role in mechanical-induced OA.3 4

OA is a disease of the whole joint, and not just of the cartilage.5 It has previously been suggested that subchondral bone may be involved6 7 on the basis that changes in the bone quality might alter the impact of the mechanical load on the cartilage. Hence, Brittle IV mice, a murine model of osteogenesis imperfecta with collagen-I defect, develop spontaneous and progressive OA with age through loss of cartilage proteoglycan.8 Subchondral bone sclerosis is a hallmark of the disease, and is observed in the late stages of OA. However, the presence of a bone marrow signal in MRI is associated with the onset of knee OA symptoms before radiological lesions9,,11 and this signal is a predictor of fast cartilage loss.12 These data suggest that subchondral bone events might participate to the development of OA.

In humans, decreased connectivity of trabeculae has been identified in the early stages, suggesting that high bone remodelling might be one of the early events of cartilage damage.13 Several animal studies have confirmed that enhanced bone resorption takes place in the early stages, and might trigger the onset of OA.6 14 15 The inhibition of bone resorption by alendronate prevents cartilage degradation in a model of joint instability in the rat,16 but bisphosphonates failed to demonstrate a prevention of joint space width narrowing in humans.17 However, cross-talk may occur between bone cells and chondrocytes, thereby stimulating cartilage breakdown.6 18,,20 Osteoprotegerin (OPG) reduces the development of OA in mice.18 21 The increase in bone resorption might result from disequilibrium in bone remodelling and/or cytokine imbalance in the local bone environment in patients with OA. We have shown that disrupting the OPG/RANKL balance reduces bone remodelling and prevents OA in a joint instability model in mice.18 Therefore, we hypothesised that the rate of turnover in subchondral bone may affect the occurrence of OA. Here, we used a mice model with increased bone turnover22 to assess, first, whether bone turnover affects cartilage metabolism and second, to evaluate the impact of bisphosphonate on cartilage damage on high bone remodelling.

Subjects and methods


The experiments complied with the Guidelines for Animal Experimentation issued by the local Ethics Committee on Animal Care and Experimentation. Male mice overexpressing Runx2 under the control of the collagen type I (Colla1) promoter were produced as previously described.22 The genetic background of the wild-type and transgenic mice was B6CF strain (C57BL/6 × BALB/c). The mice used in the study were from the 15th generation. Four groups of 10-week-old male mice were used in the study. Joint instability of the knee was induced surgically by medial meniscectomy (Mnx) using microscopic lenses.23 The right knee was opened, and the meniscotibial ligament was sectioned in order to access the medial meniscus. After being freed, the medial meniscus was gently pulled out using surgical forceps and partially sectioned using an ophthalmic scalpel. Muscles and skin were then stitched up to restore muscle integrity. The animals' left knees were sham-operated by performing a single cutaneous incision, and then stitched up. The day after surgery, Runx2-Tg mice and wild-type (WT) littermates received intraperitoneal injections of pamidronate, an osteoclast inhibitor (Pam 10 mg/kg; Sigma, Saint-Quentin Fallavier, France) once a week for 6 weeks. Ten animals were included in each group, and mice receiving phosphate-buffered saline were used as controls for each group. The animals were then killed 6 weeks after Mnx as previously described.18 At that time point, cartilage damage had started but was moderate, and therefore it was an appropriate time at which to examine the early events occurring during progression of the disease.24

Preparation of joint samples

At death, whole knee joints were dissected free from soft tissues. Specimens were fixed with 4% paraformaldehyde (pH 7.4) for 48 h at 4°C, and then decalcified with 1% paraformaldehyde–0.2M EDTA (pH 7.4, 4°C) for 2 weeks, with the medium changed twice a week. The specimens were then dehydrated using increasing concentrations of ethanol, before being embedded in paraffin. Each knee was embedded with the same orientation. Serial sagittal sections (5 µm thick) were cut from the medial femoral compartment up to the central part of the femorotibial compartment of each animal. Sections were stained for immunohistochemistry or bone histomorphometric analysis.


Sections were deparaffinated by two successive 20 min immersions in a xylene bath. The sections were then hydrated in three successive alcohol baths at decreasing concentrations of 100%, 70% and 40% for 1 min, before being washed twice by immersing for 5 min in distilled water baths. The sections were stained using Mayer's Haemalum stain for 5 min to stain the nuclei, and the sections were then counterstained with 0.125% Fast Green for 10 min to visualise bone tissue. The sections were rinsed in two successive baths of 1% acetic acid, then stained in 0.5% Safranin-O in the same bath and rinsed in 95% ethanol. All the stained sections were then rinsed with distilled water.

Histology OA score

Histological scores were recorded for the tibial plateaus. Two blinded observers used a Mankin modified version of a semiquantitative scoring system to rate the cartilage destruction as follows:

  1. 0.. No damage

  2. 1.. Irregular cartilage surface

  3. 2.. Loss of Safranin-O staining without structural changes

  4. 3.. Loss of surface <20%

  5. 4.. Severe loss of joint cartilage (>80%)

  6. 5.. Erosion down to the bone.


Immunohistochemistry was performed on decalcified serial sections using a Vector kit (Vectastain ABC kit purchased from Abcys, Paris, France). The protocol was applied according to the manufacturer's instructions. Immunohistochemistry was performed using primary polyclonal antibodies for ADAMTS-4 and ADAMTS-5 (RP1-ADAMTS-4, RP1-ADAMTS-5, Triple Point Biologics, Forest Grove, Oregon, USA), at a concentration of 1/100 and rabbit IgG as secondary antibodies provided in the Vector kit. The sections were then counterstained with toluidine blue. Positive cells were counted on the same defined area (×25), and expressed as a percentage of the total cell count.

Identification of Runx-2 transgene in bone cells and chondrocytes was performed in sections using primary DYKDDDDK Tag antibody (Ozyme, Saint-Quentin en Yveline, France), according to the manufacturer's protocol.

Histomorphometry of the epiphysial bone

Modifications of the bone were measured on the same slides that had been used for the protease measurements. Microarchitectural indices of the underlying bone were assessed in the total epiphysis by image analyser (Microvision, Evry, France) using specially designed software (Bonolab). On each slide, the underlying trabecular zone was identified between the cartilage and the growth plate (see figure 4). The following parameters were measured:

  • ▶. Bone volume/tissue volume (BV/TV, %)

  • ▶. Trabecular thickness (Tb.Th, µm) as a parameter of bone formation

  • ▶. Trabecular separation (Tb.Sp, µm) as a parameter of bone resorption

  • ▶. Osteoclast number (N.Oc/BV).

Figure 4

Osteoarthritis (OA) score in meniscectomised Runx2-Tg and wild-type (WT) mice with pamidronate (PAM). Runx2-Tg and WT mice underwent medial meniscectomy (Mnx) of their right knees, and sham surgery in their left knees. Either pamidronate or phosphate-buffered saline was injected intraperitoneally once a week for 6 weeks, and then the mice were killed. The OA score and number of ADAMTS-5 expressing cells were determined on sagittal sections of the Mnx knees of Runx2-Tg and WT mice. (A) OA was not affected by pamidronate in WT mice, whereas it was significantly reduced in Runx2-Tg mice (upper panel). (B) However, the number of ADAMTS-5(+) cells was lower after pamidronate in both WT and Runx2-Tg mice (upper panel). Illustrations of the protective effect of pamidronate in Runx2-Tg mice as shown by proteoglycan loss (SO staining) (A, lower panel), and ADAMTS-5 expressing cells (immunofluorescence) (B, lower panel) are shown. *p<0.05; **p<0.01..

Osteoclasts were detected by enzyme histochemistry in serial decalcified sections by tartrate-resistant acid phosphatase (TRAP) staining. Briefly, sections were stained for acid phosphatase using naphthol ASTR phosphate as substrate in the presence of 50 mM tartrate with hexazotised pararosaline, and counterstained with methyl green. TRAP+ cells were counted in the whole epiphysis (×25), and expressed as the number of osteoclasts per bone volume.

Cartilage explant cultures

To assess whether pamidronate has a direct effect on matrix degradation, femoral heads were harvested from 10-week-old WT mice.3 25 Cartilage was easily separated from the underlying subchondral bone. All cartilage explants (weight 0.5 mg) were cultured at 37°C in a humidified atmosphere of 5% CO2 and 95% air in Dulbecco's modified Eagle's medium (DMEM: Gibco, Invitrogen, Cergy-Pontoise, France) containing 1% antibiotic/antimycotic solution (Gibco). After prestimulating with interleukin 1 (IL1) for 24 h, the cartilage explants were washed three times, and cultured for 72 h in serum-free DMEM. Each condition was tested in triplicate wells (one explant per well) and the experiments performed in at least three different explants. The supernatants were collected at the end of culture to measure degraded proteoglycans released in the medium. Proteoglycan content was measured by a colorimetric assay using dimethylmethylene blue with shark chondoitin-6-sulphate as a standard, according to a previously reported procedure.26

Statistical analysis

Results were expressed as mean±SD Statistical analysis was performed using Statview software version 5. Parameters were compared by analysis of variance (ANOVA) followed by Fisher's least significant difference test when the p value of an ANOVA test was <0.05.


Localisation of the transgene

The presence of the transgene was verified by immunohistostaining using an anti-flag antibody. As expected, we confirmed that the transgene was expressed in osteoblasts along the bone surface and in the bone marrow precursors as overexpression of Runx2 was under the control of the Colla1 promoter. Moreover, there was no expression in articular chondrocyte (figure 1).

Figure 1

Expression of the Runx2 transgene in bone and cartilage. Expression of the Runx2 transgene was verified at the bone and cartilage levels. Immunohistochemistry was performed in knee section of Runx2-Tg mice using an anti-flag antibody. Runx2 transgene was present in mature osteoblasts and osteoblast precursors (arrows), but not in articular chondrocytes.

Mnx enhances ADAMTS expression in transgenic mice without affecting the OA score

Biochemical changes induced by Mnx were assessed in 10-week-old Runx2-Tg mice and WT littermates. Mnx significantly increased the number of ADAMTS-4-expressing cells in both WT (53.81%±2.59% vs 37.9%±1.31%; p<0.001) and Runx2-Tg mice (62.46%±2.81%, vs 33.4%±2.89%; p<0.001; figure 2). The number of ADAMTS-5-expressing cells was also greater in Mnx knees than sham-operated knees in both WT mice (37.06%±1.84% vs 30.4%±2.19%; p<0.05) and Runx2-Tg mice (47.53%±3.51% vs 32.44%±3.14%; p<0.05). However, in Mnx knees, the number of ADAMTS-4(+) cells was significantly higher in Runx2-Tg than in WT mice (62.46%±2.81% vs 53.81%±2.59%; p<0.05) as was the number of ADAMTS-5(+) cells (47.53%±3.51% vs 37.06%±1.84%; p<0.05). In contrast, the OA score in Mnx knees was no different in WT and Runx2-Tg mice (2.4±0.2 vs 2.5±0.6, p=NS) (figure 2).

Figure 2

Cartilage indices in meniscectomised Runx2-Tg and wild-type (WT) mice. Runx2-Tg and WT mice underwent medial meniscectomy (Mnx) of their right knees, and their left knees were sham-operated. After 6 weeks, mice were killed to assess cartilage changes in both knees. ADAMTS-4(+) and ADAMTS-5(+) cells were counted in joint cartilage, and osteoarthritis (OA) was scored according to Mankin. (A) ADAMTS-4(+) (left panel) and ADAMTS-5(+) (right panel) cells increased after Mnx in both Runx2-Tg and WT mice, but were significantly higher in Runx2-Tg than WT mice. (B) The OA score was no different in Runx2-Tg and WT mice (left panel), as shown in representative safranin-O-stained sections of Runx2-Tg and WT Mnx knees (right panel). *p<0.05; **p<0.001.

Mnx affects subchondral bone microarchitecture differently in Runx2-Tg and WT mice

Bone remodelling is known to be different at the epiphysis and metaphysis of the long bones. Indeed trabecular bone volume (BV/TV) has previously been shown to be significantly lower at the metaphysis in Runx2-Tg mice than in their WT littermates.27 However, we found here, in sham-operated mice, that BV/TV at the epiphysis was comparable in Runx2-Tg and WT mice (23.7%±0.2% vs 21.2%±0.2%; p=0.63) (figure 3). Six weeks after Mnx, BV/TV at the epiphysis was significantly decreased in Runx2-Tg mice (15.2%±0.01% vs 23.7%±0.2%; p<0.05) but not in WT mice (19.5%±0.2% vs 21.2%±0.2%; p=0.32). Moreover, the higher number of osteoclasts in Runx2-Tg mice compared with WT mice (65.6±7.2 vs 25.4±3.7; p<0.01, table 1) suggests that bone resorption is indeed increased in transgenic animals. Tb.Sp was higher in Runx2-Tg mice than in WT mice, although the difference was not significant (124.5±11.43 vs 103.15±10.75 µm; p=0.2). Trabecular number and thickness were not affected by genotype.

Figure 3

Bone volume at the epiphysis in meniscectomised Runx2-Tg and wild-type (WT) mice. After meniscectomy (Mnx) of their right knees, and sham operation (sham) of their left knees, mice received an intraperitoneal injection of pamidronate (10 mg/kg) or phosphate-buffered saline once a week for 6 weeks. After 6 weeks, the mice were killed and sagittal sections of knees were cut. Bone volume/tissue volume (BV/TV) was measured at the epiphysis. BV/TV was significantly reduced after Mnx in the Runx2-Tg mice, but not in the WT mice. Pamidronate increased BV/TV in Mnx knees in all mice. *p<0.05; **p<0.001.

Table 1

Bone resorption indices after pamidronate (PAM) in meniscectomised Runx2-Tg and wild-type (WT) mice

Pamidronate inhibits bone resorption in epiphysial bone in both Runx2-Tg and WT mice

Six weeks after Mnx, pamidronate induces an increase in BV/TV in both Runx2-Tg (32.9%±0.5% vs 15.2%±0.01%; p<0.05) and WT mice (44.8%±0.4% vs 19.5%±0.2%; p<0.01) (figure 3). The high initial Oc.N/BV observed in Runx2-Tg mice decreased after pamidronate, and there was also a significant reduction in Tb.Sp (table 1). A similar decrease in the number of osteoclasts and in Tb.Sp was seen in WT mice receiving pamidronate, which were of same magnitude. However, BV/TV was lower in Runx2-Tg than in WT mice receiving pamidronate, although this difference was not statistically significant (p=0.07). Trabecular thickness was not affected by either genotype or treatment, but the trabecular number was significantly increased in the Mnx knees of pamidronate-treated mice, irrespective of their genotype (table 1).

Blocking bone resorption by pamidronate prevents the increase in the OA score

The OA score was reduced by pamidronate in Runx2-Tg mice compared with Runx2-Tg phosphate-buffered saline-treated mice (1.2±0.5 vs 2.4±0.25; p<0.05), but not in WT mice (2.7±0.4 vs 2.4±0.7; p=NS). However, pamidronate dramatically decreased ADAMTS-5 expression in both WT (6.95%±0.40% vs 37.06%±1.84%; p<0.01) and Runx2-Tg mice (7.62%±0.41% vs 47.53%±3.52%; p<0.01) (figure 4B).

Proteoglycan release from Runx2-Tg cartilage explants is not affected by pamidronate ex vivo

We then investigated whether pamidronate had any direct effect on proteoglycan release in cartilage. In cartilage explants, IL1 significantly enhanced the proteoglycan content in the supernatant compared with controls (figure 5). Pamidronate reduced the IL1-induced proteoglycan content (253±5.5 vs 326±23 µg/ml; p<0.05), but had no direct effect on its own (279±15 vs 227±28 µg/ml; p=NS).

Figure 5

Proteoglycan release in cartilage explants from wild-type knees. Proteoglycan content was measured in the supernatant of cartilage explant cultures in the presence or absence of pamidronate (PAM). Cartilage explants from wild-type mice were stimulated with interleukin 1 (IL1) for 24 h, washed three times and cultured in the presence of pamidronate for another 72 h. The proteoglycan content in the supernatant was increased by IL1, and significantly reduced by pamidronate. *p<0.05.


In this study, we have found that a marked reduction in bone resorption blunted mechanically induced OA in mice with high bone remodelling. It has been suggested that subchondral bone participates in the early events of OA.6 Several authors have identified increased bone resorption underneath the cartilage at an early stage of OA in different animal models and in human osteoarthritic cartilage.2 14 28 29 To test whether the level of bone remodelling might modify the metabolism of cartilage, we used Runx2 overexpression driven by type I collagen promoter. The expression of the transgene leads to a bone phenotype as the transgene is only present in mature osteoblasts and their precursors, but not in articular chondrocytes. Therefore, this model is suitable for assessing the impact of high bone turnover on cartilage. In the absence of Mnx, the expression of ADAMTS was no different in Runx2-Tg and WT mice showing that there is no impact of Runx2 hyperexpression on articular chondrocytes. After Mnx, the number of osteoclasts are associated with an increased expression of ADAMTS in Tg mice, which may be due to the increased bone resorption itself and/or may promote the secretion of local factors that influence cartilage breakdown. Our data are in line with previous reports,2 14 28 29 strongly suggesting a cross-talk between bone and cartilage. Our results are different from the protective effect obtained in Runx2± mice against OA.30 The Runx2-Tg mice, in which expression of Runx2 was decreased in all cells, had a lower progression of OA because of a reduced chondrocyte hypertrophy. Moreover, such a protection might be related to an effect in chondrocytes and/or in osteoblasts in Runx2± mice. Our transgenic Runx2 mice display no matrix cartilage phenotype in basal conditions, but a higher expression of two major cartilage proteases after Mnx, which suggests that the level of bone resorption and not the level of Runx2 expression on itself might influence cartilage remodelling under mechanical conditions.

Pamidronate reduces the OA score only in Runx2 transgenic mice despite reduced osteoclast recruitment in all groups. The indirect role of osteoclasts in cartilage remodelling in response to mechanical stress has been observed using different osteoclast-blocking agents.18 28 31 Blocking bone resorption prevented OA progression in rats treated with a high dose of alendronate16 and in dogs receiving licofelone.2 Our data indicate that a marked suppression of bone resorption is required in order to achieve an effect on cartilage structure. Reducing bone resorption could inhibit the secretion of local factors, which may have a role in the development of OA.18 Indeed, cytokine disequilibrium in the bone environment might regulate the expression of the proteases, or protease inhibitors, which are involved in cartilage metabolism.32,,34 We demonstrated previously that Mnx-induced OA was prevented by systemic administration of OPG, a potent inhibitor of bone resorption.18 Indeed, osteoblast precursors from Runx2 transgenic mice have an enhanced expression and production of RANKL.22 35 Therefore, blocking bone remodelling by pamidronate may have decreased the secretion of RANKL by bone cells and thereby prevented the degradation of cartilage.

Overall, these findings suggest that it might be possible to prevent cartilage damage by achieving a low resorption threshold when bone resorption is high at baseline. The findings presented here are in line with observations in human clinical trials. Risedronate failed to slow the radiographic progression of OA based on joint space width. However, a dose-dependent reduction in the levels of collagen type II (CTX-II), a cartilage degradation marker, was observed with high doses of risedronate17 and patients with both low collagen type II and collagen type I levels had the lowest risk of OA progression.36 Therefore, slowing down bone resorption might be useful to reduce cartilage breakdown in a subgroup of patients with high bone remodelling.

In conclusion, we have shown that high bone remodelling in mice is responsible for increased protease expression in mechanically induced OA. These results show that bone tissue influences cartilage breakdown. Moreover, reducing bone resorption might blunt the severity of OA in mice with high bone remodelling. Therefore, targeting bone resorption might be used as a therapeutic approach to the treatment of high bone resorbing forms of OA.


The authors are grateful to the association ‘Rhumatisme et Travail’ for their help and financial support.


View Abstract


  • Competing interest None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.