Introduction
Osteoarthritis (OA) is a major cause of functional impairment and disability among the elderly [
1], yet current therapies predominantly target symptoms rather than providing prevention or curative treatment. Animal models of OA have been used extensively for studying the pathogenesis of cartilage degradation as well as the efficacy of potential therapeutic interventions [
2]. However, most of the currently available models only approximate the mechanisms underlying the human disease. Although several animal species – such as mice, Syrian hamsters, guinea pigs, and nonhuman primates – can develop spontaneous OA, the development of disease in these models is slow; typically, more than 9 to 12 months is required for significant cartilage erosion to occur [
2]. Consequently, these spontaneous models are cumbersome and time-consuming to use in arthritis research and drug development. Transgenic mice models have been of great help in clarifying the role of numerous pathogenic factors (matrix metalloproteinases, transforming growth factor β, nitric oxide) in the development of OA, yet these models may not be applicable for studies testing the therapeutic potentials of chondroprotective agents [
3,
4]. Surgically induced joint damage has also been used extensively as a model of OA, though this condition more nearly approximates a traumatic form of OA than it does the natural, spontaneously evolving form [
5]. Thus, there is an apparent need for an OA model that directly mimics a human form of the disease and at the same time provides a convenient methodological tool for preclinical investigations.
Development of such a generally applicable and convenient animal model of OA is complicated by the fact that our current understanding of the pathophysiology of the human disease is incomplete. However, one factor thought to affect the regulation of cartilage turnover is estrogen. The putative role of estrogens is corroborated by the fact that the prevalence of OA is higher in post-menopausal women than in men [
6‐
8]. Furthermore, the recent finding that ovariectomized (OVX) cynomolgus monkeys show OA-like pathological changes within articular joints [
9], as well as the chondroprotective effects of hormone replacement therapy proposed by some epidemiological observations [
10,
11], also argues for the involvement of estrogen deficiency in female OA.
The present study was designed to evaluate the role of estrogen in regulating cartilage turnover, by investigating the effects of ovariectomy on cartilage. Histological analysis of the knee joint was used to assess the pathological changes of the articular cartilage erosions. Furthermore, the effects of cessation of endogenous estrogen production on bone and cartilage turnover were assessed using biochemical markers of collagen type I and II degradation (CTX-I and CTX-II). An additional aim was to clarify whether OVX rats could provide a useful model of post-menopausal OA for future preclinical studies assessing the chondroprotective effects of exogenously administered estrogens and estrogen-like substances such as selective estrogen receptor modulators (SERMs).
Materials and methods
Animals and study design
Sprague–Dawley rats (Crl:CD®(SD)IGS.BR) obtained from Charles River Laboratories, Kisslegg, Germany, were used. Experiments were approved by the Experimental Animal Committee, Danish Ministry of Justice (Slotsholmsgade 10, DK-1216, Denmark) (approval number 2002/561-566) and were done in accordance with the European Standard for Good Clinical Practice. The animals were maintained at the Animal Research Facilities at Nordic Bioscience for 1 month before the start of experiments. They were housed, two per cage, in a room maintained at 20°C with a 12-hour/12-hour light/dark cycle and given food (Altromin 1234, Lage, Germany) and Milli Q water (Millipore, Glostrup, Denmark) ad libitum.
To assess age-related changes in cartilage turnover, we measured the creatinine-corrected excretion of CTX-II (for details see below) in the urine of six male and six female rats sampled at 1, 2, 3, 6.5, and 9.5 months of age. Urine samples were obtained as spot samples by placing the rats in a metabolic cage for 30 to 60 min and waiting for them to urinate.
Study of the effect of ovariectomy in OVX rats
For these studies, two cohorts of 20 virgin female Sprague–Dawley rats were used. At the start of the study they were either 5 months old (cohort A) or 7 months old (cohort B). At this baseline, body weight was determined and the animals were randomly stratified into two groups to undergo either bilateral ovariectomy using a dorsal approach or a standard sham operation under general anesthesia induced by Hypnorm-Dormicum (1 part Hypnorm® + 1 part Dormicum® + 2 parts sterile deionized water; dose 0.2 ml/100 g body weight). During the 9 weeks of follow-up, body weight was determined weekly; urine samples were obtained at baseline and weeks 2, 4, 6, and 9 after ovariectomy. At study termination, the knees were isolated and kept in 4% formaldehyde until further quantification of surface erosion in the articular cartilage by histological measurements as outlined below.
Study of the effect of exogenous estrogen and SERM
For this purpose, a cohort of 60 5-month-old virgin female Sprague–Dawley rats was included. At baseline, body weight was determined and the animals were randomly stratified into five groups with 12 rats in each group. One group was subjected to sham operation and the remaining four groups were ovariectomized as described above. The four equal groups received treatment either with the vehicle (50% Propylene Glycol [Unikem, Copenhagen, Denmark], 0.075 M NaCl), or with 17α-ethinylestradiol (E-4876, Sigma, St Louis, MO, USA) (0.1 mg/kg per day), or with the SERM (-)-
cis-3,4-7-hydroxy-3-phenyl-4-(4-(2-pyrrolidinoethoxy)phenyl)chromane [
12] given as an oral suspension in the vehicle from day 1 by gavage 5 days a week for 9 weeks, in either a low or a high dose (0.2 or 5 mg/kg per day, respectively). Animals were weighed and sampled for spot urine and serum at regular intervals. At study termination, knee joints were prepared for histology as described below.
Materials and buffers
All chemicals were analytical grade and purchased from either Sigma or Merck (Darmstadt, Germany). Peptides, from Chimex Ltd (St Petersburg, Russia), were >95% pure. Cell-culture reagents were obtained from Life Technologies, UK. The buffers used in the immunoassays have been described elsewhere [[
13]; P Qvist and colleagues, unpublished].
Histology
After careful dissection, the knees were decalcified for 3 to 4 weeks in 10% formic acid, 2% formaldehyde. The decalcified knee joints were cleaved along the medial collateral ligament into two sections and embedded in paraffin. Coronal sections were then cut at three different depths (0, 250, and 500 μm) from the medial collateral ligament. Each section was stained in Toluidine blue and the section that comprised the most load-bearing region were used for measurements. The histological sections were assessed by a blinded observer.
In a preliminary study, we evaluated apparent histological features as well as applicable assessment methods for quantifying pathological changes in the knee joints. The previously described Mankin and Colombo score systems are used in analyzing known OA models such as the guinea pig, and may not fulfil the criteria for a reliable scoring system in this OVX rat model [
14]. In the preliminary study, we analyzed OVX and sham-operated rats by the Colombo method and found that erosion was the feature most readily influenced by the ovariectomy in the OVX rats in comparison with the sham-operated rats. In order to simplify evaluation protocols and increase the robustness of the scoring system, we found it more reproducible to concentrate evaluation on surface erosion as the main feature of cartilage damage. Exact numerical values were obtained by measuring the length of the erosion surface and dividing it by the total cartilage surface. This approach enabled us to quantify erosion in exact numerical values instead of scores relying on the observer. Furthermore, it relates to a feature that is directly relevant to development of OA lesions. We therefore decided to keep the analysis simple and focus on surface erosion.
RatLaps ELISA to assess bone resorption
The RatLaps ELISA (Nordic Bioscience Diagnostics A/S, Herlev, Denmark) measures collagen type I C-telopeptide degradation products (CTX-I) using a specific monoclonal antibody in a competitive ELISA form [P Qvist and colleagues, unpublished]. The assay is applicable for measurement of both urine and serum samples, but only serum samples were assessed in this study. All serum samples measured in the assay were from animals that had been fasting for at least 6 hours prior to the sampling. Briefly, the assay is performed by incubating a biotinylated form of a synthetic peptide representing the C-telopeptide epitope EKSQDGGR. This is followed by addition of sample and primary antibody and after overnight incubation the amount of bound antibody is made visible using a peroxidase-labeled secondary antibody and a chromogenic peroxidase substrate. The concentrations in the samples were determined from the construction of a calibration curve based on the measurement of synthetic peptide standards. Intra-assay and interassay variations were 6.9% and 10.4%, respectively. All samples were measured in duplicate and samples from the same animal were included on the same microtiter plate. Three genuine control samples were included on each microtiter plate to verify performance, and samples were remeasured if the coefficients of variation exceeded 15% or if any of the control samples measured more than 20% off the predetermined value.
CartiLaps ELISA to assess cartilage turnover
Monoclonal antibody mAbF46 specific for collagen type II C-telopeptide fragments (CTX-II) was used in a competitive ELISA format developed for measurement of CTX-II in urine samples (CartiLaps ELISA, Nordic Bioscience Diagnostics A/S) [
13]. The assay was performed by first incubating biotinylated collagen type II C-telopeptide-derived peptide (EKGPDP) on a streptavidine microtiter plate, and then the sample as well as the primary antibody were added. After overnight incubation, the plates were washed and a peroxidase-labeled secondary antibody was added, followed by a chromogenic peroxidase substrate. The concentrations of CTX-II (μg/l) were standardized to the total urine creatinine (mmol/l) (JAFFA method; Hoffmann-La Roche, Basel, Switzerland) giving concentration/creatinine (μg/mmol). The precision of the assay was 7.1% and 8.4% for intra-assay and interassay variations, respectively. Assay performance and quality assurance were treated as described above for the CTX-I assay.
Statistical analysis
Means and SDs were calculated using parametric statistics. Differences between groups were assessed with the Mann–Whitney U-test for unpaired observations. The association between the biomarkers and the histology data was calculated using Spearman's rank correlation.
Discussion
Estrogen receptors are found in a wide range of cell types in the body, explaining the pleiotropic effects of this hormone [
15]. The effect of estrogen on several estrogen-responsive tissues such as endometrium, bone, and breasts has been extensively studied. In the present study, cartilage turnover and morphology were assessed in sham-operated and OVX rats to investigate whether cessation of endogenous estrogen production may influence articular cartilage turnover and integrity. Our findings show that ovariectomy induces a significant increase in the breakdown of collagen type II and subsequent articular cartilage erosion. Furthermore, we demonstrate that administration of exogenous estrogen or a SERM to OVX rats suppresses the progression of these events.
The assessment of articular cartilage turnover in rodents is complicated by the fact that the growth plate in these animals remains present and is at least partly metabolically active, even at older age [
16]. The growth plate contains a significant amount of collagen type II, which undergoes constant remodeling during ectopic bone formation and thereby contributes to systemic levels of collagen type II metabolites [
17]. Accordingly, we observed high CTX-II levels in animals below 3 months of age (Fig.
1), suggesting that a significant fraction of the analytes obtained from young rats and measured in the assay originates from growth plate turnover and not from articular cartilage. A similar situation is observed in humans younger than 20 to 25 years of age, but in contrast to the situation in rodents, the growth plates in human adults close when skeletal growth has ceased [
18]. In 6-month-old rats, the CTX-II levels decreased by 86% compared to 3-month-old rats (Fig.
1), suggesting that skeletal growth and thereby growth-plate turnover at this time are minimized. These observations formed the rationale for assessing the effects of ovariectomy on cartilage turnover and structural integrity in 5- and 7-month-old rats.
In OVX rats from all three cohorts, an increase in the degree of cartilage erosion of the hind knee joints was observed at termination of the study, after 9 weeks of treatment. The OVX rats had a significantly higher incidence of cartilage surface erosion in the medial tibia and lateral femur than the sham-operated rats. This tendency was also found in the medial femur, but the lateral tibia showed no difference between OVX and sham-operated animals in any of the assessed cohorts. The rats of cohorts B and C showed a more pronounced erosive change to ovariectomy. However, the changes were most pronounced in the femoral condyles in all experiments, suggesting that the relative responses of the different regions of the knee joint are similar at these two ages.
The pathological changes observed in the OVX rats were of a similar nature to the very early changes observed in human OA, where mild erosion and loss of proteglycans are among the earliest changes that have been described [
19,
20]. The histological appearance of the knee articular cartilage in the OVX group differs from the appearance of articular cartilage in models such as ligament transection and meniscal tear [
2,
5]. In these models, more severe erosive changes can often be observed and changes such as fibrillation and vascularization appear markedly increased. The changes in knee cartilage observed after ovariectomy were relatively mild in comparison and may represent features of earlier or less aggressive disease, which are stages of the disease that are difficult to address in many of the currently used models of OA. Thus, the OVX model may be uniquely suitable for the study of early-stage OA.
A significant elevation in CTX-I levels reflecting bone resorption was observed in the OVX rats in comparison with the sham-operated group. This observation is in accord with the expected increase in bone turnover induced by ovariectomy [[
21]; P Qvist and colleagues, unpublished]. The dynamics of the changes in CTX-I levels over the 9-week study period suggests a sustained increase of approximately 100% in OVX rats compared with the sham-operated group (Fig.
6). This increase is similar in magnitude to that seen in bone turnover at the menopause transition [
22]. These observations indicate that ovariectomy in rats induces estrogen deficiency that can evoke the skeletal metabolic changes typically accompanying the menopause. These observations are in accord with findings from other studies [[
12,
15,
21]; P Qvist and colleagues, unpublished]. Also, the observed increase in body weight and decrease in uterus weight observed in all cohorts as a consequence of ovariectomy is in agreement with the known systemic effects of estrogen withdrawal [
15].
Cartilage turnover as assessed by the CTX-II assay was also increased in the OVX rats compared with the sham-operated group. The difference was most pronounced in the first 4 to 6 weeks after ovariectomy, where CTX-II levels were increased by 100%, but at later time points the difference between the OVX and sham-operated animals were diminished. This observation suggests that the increase in cartilage turnover induced by cessation of endogenous estrogen production may be transient in the OVX model, possibly reflecting the activation of mechanisms antedating the actual cartilage damage. The initial increases in the levels of the marker observed immediately after ovariectomy corresponded well with the increase in CTX-II levels observed at the menopause in humans, where a 100% increase has been demonstrated [
18]. We have analyzed animals up to 15 weeks after ovariectomy, in which surface erosions were present to the same extent as seen in the rats maintained for 9 weeks. Whether the surface erosions posses the ability to spontaneously repair after longer times cannot be determined from our studies.
The changes in the cartilage turnover marker (CTX-II) observed after 4 weeks showed close correlation with the histological signs of articular cartilage degradation observed at study termination (Table
2; Fig.
5). Thus, the early changes in the biomarker levels can be considered predictive of the subsequent structural changes in the knee joint. This is in accordance with findings obtained in clinical investigations, where CTX-II levels and changes in this marker are correlated with radiologically assessed damage of articular cartilage in the knee joint [
23‐
25].
The menopause can frequently be accompanied by an increase in body weight, which can partly be ascribed to estrogen deficiency. Increased body weight, especially fat accumulation, may theoretically have an inhibitory effect on articular cartilage degradation through increased production of endogenous estrogens. Increases in body weight may also enhance cartilage degradation evoked by a greater physical challenge of the joints. In the present study, we observed a significant weight gain in OVX rats. However, there was no correlation between body weight and cartilage erosion, suggesting that the observed histological changes of knee articular cartilage in OVX rats is unlikely to be a result of increased body weight and is more likely to be due to estrogen deficiency per se. This observation is also supported by a previous study on healthy humans indicating an apparently minor overall contribution of body weight to cartilage turnover as assessed by the CTX-II assay [
18].
In the present study, we also investigated whether exogenous estrogen and an estrogen-like substance can provide prophylactic effects against the acceleration of cartilage degradation associated with ovariectomy. These hypothesized effects were investigated with reference to the well-known effects of these agents on bone turnover [
12,
26,
27]. Furthermore, it has also been demonstrated that the SERM idoxifene reduces disease severity and bone erosion in adjuvant-induced arthritis, an animal model of RA [
28]. We tested a SERM belonging to the class of
cis-3,4-diaryl-chromanes, which have been demonstrated to provide significant antiresorptive effect in OVX rat studies [
12]. The SERM is structurally very similar to levormeloxifene, which has been tested clinically in postmenopausal women and found to be more potent than hormone replacement therapy in preventing bone loss [
27].
The levels of CTX-II in OVX rats treated with the higher dose of the SERM or with estrogen were similar to levels seen in the sham-operated animals of the same cohort. In contrast, the lower dose of SERM was only partly effective in reducing the elevated CTX-II levels. For CTX-I levels, only estradiol treatment was able to completely suppress bone resorption to levels seen in the sham-operated rats, whereas the two SERM-treated groups showed an intermediate effect. In accord with the effects observed with the biomarkers, the histological examination revealed that whereas the vehicle-treated OVX rats again showed significantly increased erosions of the cartilage surface, the groups treated with estrogen or SERM were indistinguishable from the vehicle-treated sham-operated group. The SERM showed a dose-dependent ability to prevent the erosive changes. There was a high correlation between changes in CTX-II observed in the first 4 weeks of the study period and subsequent erosion of articular knee cartilage.
The three sets of separate experiments described here were all in line with significantly increased cartilage erosion in OVX rats, pointing to an apparent chondroprotective influence of endogenous estrogen on cartilage turnover. Furthermore, administration of exogenous estrogen to OVX rats prevented the erosive changes, thereby further supporting the association between estrogen and cartilage. These observations are in accord with findings from previous studies indicating that the prevalence and incidence of OA is increased among postmenopausal women [
11,
29]. The notion that cartilage metabolism may be influenced by estrogen is conceivable also, because chondrocytes of articular cartilage possess functional estrogen receptors [
15,
30,
31]. Recent publications describing the results of a 3-year follow-up study of ovariectomized cynomolgus monkeys have provided strong evidence that ovariectomy induces OA-like changes in articular cartilage [
9]. In this animal model, administration of exogenous estrogens, but not phytoestrogens, was able to prevent these changes. A similar indication of potential chondroprotective properties of estrogen has been obtained in several epidemiological and case–control studies, where estrogen use in menopausal women has been associated with a decreased incidence of OA [
7,
32]. CTX-II levels are increased twofold after the menopause, and supplemental hormone replacement therapy can suppress this marker to premenopausal levels, further supporting a role of estrogen as a regulator of cartilage metabolism [
18]. Based on these previous observations, it seems reasonable to consider the model of older OVX rats (i.e. 5 months of age or more) as an
in vivo model of postmenopausal OA. However, the ultimate demonstration of the utility of the model awaits the introduction of novel agents with potential chondroprotective effects.
Competing interests
P Høegh-Andersen, TL Andersen, CV Lundberg, JA Mo, A-M Heegaard, J-M Delaissé, and S Christgau are all employees at Nordic Bioscience A/S. LB Tankó is an employee at the Center for Clinical and Basic Research.