Enhanced concentration of COMP (cartilage oligomeric matrix protein) in osteochondral fractures from racing Thoroughbreds

The aim of the present study was to correlate the levels of COMP and aggrecan as indicators of tissue damage, in synovial fluid (sf) from carpal joints of acutely lame racehorses, with macroscopical lesions of articular cartilage (OA), osteochondral fractures and ligament tears found at arthroscopy.


Introduction
Lameness is one of the most common and important causes of training failure in the equine athlete [2]. Often the athletic activity, as a result of repeated overloading of the joint, leads to articular cartilage damage which 'Corresponding author. Tel.: +46-018-671191; fax: +46-018-E-mail addrenes; eva.skioldebrand@telia.com, eva.skioeldebrand.

673532.
es@bayer.se (E. Skioldebrand). may be accompanied by synovitis, ligament tear and a progressive sclerosis of the subchondral bone [29]. Osteochondral fractures ("chip fractures") of the carpal bones are common in young Thoroughbred (TB) and Standardbred (STB) racehorses with a statistically significant difference between the anatomical locations of the fractures in the two breeds [31]. The difference in location is related to the gait at which the horses are required to perform and together with shoeing convention results in a unique load distribution within the joint [31,39]. Typical locations of the chip fractures are sites E. Skiiildeehrund et id. I Journal of Ovrhopuedic Research 23 (2005) 156-163 157 experiencing high mechanical load, most commonly involving the dorsal, proximal joint margins of the carpal joint [39]. The cause of these chip fractures is complex and related to many factors. An important factor is considered to be hyperextension of the joint at exercise, which increases to a maximum with higher speed [12]. The repetitive cyclic load on the subchondral bone leads to bone sclerosis and an accumulation of focal areas of microfractures. The mechanical derangement leads to a collapse of the weakened tissue with fragmentation of cartilage and bone [ 15, 27,33].
Early pathological changes in the metabolism of articular cartilage and subchondral bone cannot be monitored, in vivo, over time, with current diagnostic tools. Hence more effort is now being exerted into biochemical studies in vivo and in vitro, to understand the mechanism behind articular cartilage and bone turnover in normal and diseased joints [25,36,44]. Molecular markers released in synovial fluid (sf) and serums are useful for monitoring metabolic activity (anabolism or catabolism) of cartilage which can be followed over a time period. One such indicator of altered metabolism is the change in glycosaminoglycan structure depicted by the 846 epitope on chondroitin sulphate [8]. This is altered in sf and serum with levels significantly higher in equine carpal joints with osteochondral fractures compared to healthy joints. The concentration of carboxy propeptides of collagen type 11, an indicator of collagen production was elevated in serum from horses with fractured joints [8].
Cartilage oligomeric matrix protein (COMP) is a non-collagenous, homo-pentameric protein belonging to the thrombospondin family with 5 globular domains attached to the central assembly domain by flexible strands [28,30]. COMP is thought to contribute to the organisation of the collagen fibrils into networks [34]. In human adult articular cartilage, COMP is found prominently in the interterritorial region of the middle and superficial layer [40]. COMP is also very prominent in growth cartilage, particularly in the territorial regions of the proliferative zone [6]. The protein is also found in tendon [3,43], and small amounts are present in the synovial membrane [4]. However, the concentration of COMP in intra-articular ligaments and synovial membrane of the equine middle carpal joint is thousand fold and hundred fold lower, respectively than in synovial fluid [42] indicating negligible production in these tissue elements.
A decreased COMP synthesis in the articular cartilage and low concentration of sf-and serum COMP has been correlated to osteoarthritis (OA), and septic joint disease in the equine middle carpal joint [25,26,42]. However, an increase in serum and sf-COMP concentrations is found in early stages of OA in human knee joints [7,32], and the high COMP levels persists several years after injury [19,37,38].
The aim of the present study was to correlate the levels of sf-COMP and aggrecan, from carpal joints of acutely lame racehorses, with macroscopical lesions of articular cartilage, osteochondral fractures and ligament tears found at arthroscopy. We further delineated the expression of COMP mRNA in osteochondral fractures.

The horses
In total 63 joints from 49 Standardbred trotters (STB) and 14 Thoroughbreds (TB) were included in the study. These horses were in conventional trainindracing and underwent arthroscopy of their middle carpal or radiocarpal joints. The breed, age and sex of the horses are presented in Table 1.
In total 86 joints were examined but In view of a possible influence between different joints in a given horse, only values from the "primary" joint were used for evaluation. The primary joint was defined as the joint exhibiting the most severe clinical signs.
An additional five TB, euthanized due to non-orthopaedic reasons, with normal articular cartilage, ligament and joint capsule were included as a control group. One TB (3 years old) with severe macroscopical articular cartilage lesions, euthanized due to lameness was included.
Surgery was performed under general anaesthesia with standard arthroscopic techniques [24]. Synovial fluid (sf) samples were taken after anaesthesia, prior to surgery. The skin of the carpal joint was disinfected with 70'1/0 ethanol and the sf was aspirated into sterile tubes without additives. The sf was aspirated as completely as possible from a maximally flexed carpal joint and the volume was recorded. The sf was centrifuged and stored frozen at -70 "C until analysed.

Clinicul esuinination
A routine lameness examination, including local anaesthesia and X-ray, was performed prior to the arthroscopy. The degree of lameness at examination was evaluated on a 0-5 scale including flexion test reaction. The lameness history was documented as days between injury and the date of arthroscopy (days postinjury).

Arthroscopic observurions
The articular cartilage lesions were documented macroscopically and graded as normal, mild, moderate or severe. Mild lesions were defined as superficial fraying of the articular cartilage in a small area. Moderate lesions presented fraying over a larger area including cartilage erosions, with focal, small areas of denuded bone, always in the area of the radial facet. The severe lesions were characterised by fraying, erosions and cartilage loss with denuded bone in a large area. In some joints an osteochondral fracture was present.
Macroscopic evidence of synovitis was considered as hyperemia and hyperplasia of the synovial membrane forming abnormal villi. Mild synovitis was defined as hyperaemia and mild thickening of villi. Moderate synovitis also included focal formation of new villi. Severe synovitis presented these abnormalities of villi diffusely and also with Table 1 Breed, age and sex distribution of ihe 63 horses and additional 5 *horses euthanized due to non-orthopaedic reasons adhesions between villi. The intercarpal ligaments were classified as intact or with partial tears.

Immunoassay for COMP and aggrecan
Concentrations of sf-COMP and sf-aggrecan were measured by an inhibition enzyme-linked immunosorbent assay (ELISA) as described previously [10,36]. Polyclonal antibodies raised in rabbits against bovine COMP and equine aggrecan were used. Microtitre plates (COMP a polystyrene plate, aggrecan: a polyvinyl plate) were coated with purified equine COMP [43] or equine aggrecan and standards of equine COMP and aggrecan were included in each plate. The aggrecan was purified by cesium chloride density gradient centrifugation and by chromatography as described elsewhere [I 11. For COMP analyses SDS was used to dissolve samples and neutralized by a five time excess of triton in the antibody solution [36] and for aggrecan analyses samples were pretreated with hyaluronidase [38]. Enzyme substrate (paranitrophenyl phosphate in diethanolamine) was added and the absorbance at 405 nm was read immediately and after 1 h in a Multiscan Multisoft (Labsystem, Stockholm, Sweden). The increase in absorbance was used for calculations. All samples were analysed in triplicate with the mean used for calculations [9]. The intra-and inter-assay variations of the COMP and aggrecan assays were 5% and lo%, respectively. Samples from the study entry (controls) were always included in every new assay.

Characterization of COMP in synovialJluids
Synovial fluid from one TB with an osteochondral fracture (2 years old) and from one TB with severe articular cartilage lesions (3 years old) were selected based on high sf-COMP concentrations from ELISA analysis (80 respectively 29 pglml). Aliquots (100 pl) of each synovial fluid were dialysed with 500 ml Tris-NaCl buffer (25 mM Tris, 25 mM NaC1, pH 7.5) for 24 h at 4 "C, to lower the salt concentration. Thereafter the samples were run through a SwellGel@ Blue Albumin Removal disks (Boule Nordic AB, Sweden) to remove excess of albumin. A wash buffer of 25 mM Tris pH 7.2 was used in all steps. All washing steps were pooled. The samples were precipitated in IO-volume ethanol containing 50 mM sodium acetate for 18 h and precipitates were recovered after centrifugation, 20 min, at 16,060g and then dried under vacuum. Precipitates were dissolved in 60 pl electrophoresis buffer (0.125 M tris-HC1, 2% SDS, pH 6.8, 0.002% Bromphenol blue and 20% Glycerol) with and without 2-mercaptoethanol, and boiled at 100 "C for 4 min. The samples were applied to SDS-PAGE 4-15% gradient gel according to the protocol of Laemmli [16]. After the electrophoresis the proteins were electrotransferred onto nitrocellulose membranes in Tris-glycine buffer (25 mM Tris, 192 mM glycine, pH 8.3,20% methanol using a BioRad system at 120 V for 2 h). Following transfer, the membranes were blocked with 5% skimmed milk in Tris buffered saline pH 7.4, including 0.05% Tween 20 (TBSEween). Antigenic COMP was detected with the two different polyclonal antibodies (dilutions; anti-bovine 1:2000, anti-equine 1:2000) in 1% skimmed milk in TBSTTween. Swine anti rabbit horseradish peroxidase (HRP) (Daco AIS, Denmark), 1:20.000 in 1% skimmed milk in TBS/ Tween was used as secondary antibody. To avoid cross-reactions with immunoglobulins from the horse, 10% of horse serum was added to the antibody. Super Signal@ West Pic0 Chemiluminescent substrate was used for detecting HRP on the immunoblot. The film was developed manually with GBX developer and fixer (Eastman Kodak Company, USA).

Immunohistochemistry and in situ hybridization
Osteochondral fragments from two TB joints and an articular cartilage biopsy from the opposite side of the fracture in one of the horses, and articular cartilage from the TB with severe OA (all specimens obtained at arthroscopy) were immediately fixed in a mixture of paraformaldehyde (3%) and glutaraldehyde (0.1Yn) in 0.1 M phosphate buffer, pH 7.4, at room temperature. The tissue was cut in half, and either frozen or dehydrated and embedded in paraffin and cut into 6 pm thick sections. The frozen sections were then treated with H202 and non-specific staining was blocked with normal swine serum. Immunostaining for COMP was done at room temperature for 60 min with the anti-equine COMP antibody (diluted 1:500). Specificity con-trol was achieved by incubating with excess of antigen or by substituting the primary antibody with non-immune rabbit serum (diluted 1:2000). The visualisation of the antibody-antigen complex was done after the sections were rinsed in PBS, incubated with an excess of the second antibody (swine anti-rabbit IgG) and a final incubation with PAP-complex (rabbit anti-peroxidase) and the colour developer 3,3diaminobenzidine. Sections were decalcified in 10% formic acid and stained with Toluidine Blue.
For in situ hybridization, a cocktail of three 32-to 35-mer oligonucleotide probes complementary to equine COMP mRNA (GenBank accession number AF325902) with the following 5' to 3'-sequences was used; 32-mer: CAG TTA TGC TGC CCG GTC TCA CAC TCG TCA AT, 35-mer: TGT CCT GAG TTG GGT ACC GTC ACG CAG TTA TCC TT, 33-mer: GTC ACA GAC ATC CCC TAT ACC ATC GCC ATC ACT. A mixture of two sense probes (a 33-mer and a 35-mer) was used as a negative control (Cybergene AB, Huddinge, Sweden). The probes were labelled at the 3' end with digoxigenin-1 1-dUTP using a DIG Oligonucleotide Tailing Kit (Roche, Mannheim, Germany). Paraffin sections were dewaxed, rehydrated, treated with Proteinase K (5 pglrnl; Roche, Mannheim, Germany) for 30 min at 37 "C, and postfixed in 4% paraformaldehyde for 5 min at 4 "C. The sections were then acetylated twice for 5 rnin each, with 0.25% acetic anhydride (Sigma) in 0.1 M triethanolamine buffer (Sigma). Thereafter the sections were prehybridized in a commercial hybridization solution containing 60% formamide and 5x SSC (Dako, Glostrup, Denmark) without probe for 2 h at 37 "C. After rinsing in 2x SSC, the sections were overlaid with hybridization solution containing 5 pmol/ml of each anti-sense probe or control sense-probe, and incubated overnight in a humidified chamber at 39 "C. The slides were immersed in 2x SSC to remove the cover slips and then washed at 39 "C as follows: 2 x 15 min in 2x SSC, l x SSC and finally in 0 . 2 5~ SSC.
Immunological detection of the bound probe was performed according to suppliers instructions (Boehringer Mannheim) with slight modifications. Briefly, the sections were blocked for 30 min in TBS buffer (100 mM Tris-HCI, pH 7.5, 150 mM NaC1) containing 0.1% Tween20 (Merck) and 2% normal sheep serum. (National Veterinary Institute, Uppsala, Sweden.) Subsequently, the sections were overlaid with alkaline phosphatase-conjugated anti-digoxigenin antibody (Roche) diluted 1/1000 in TBS containing 1% normal sheep serum and incubated for 2 h at RT. For colour development the section were incubated with NBT/BCIP (Roche) substrate solution containing 1 mM levamisole for 18 h in the dark at RT.

Statistical methods
The mean and standard deviation (SD) for sf-COMP, sf-aggrecan values and ratio between aggrecan and COMP for the different study groups is reported in Table 2. The concentrations of sf-COMP and sfaggrecan were compared between STB and TB with normal, moderate and severe articular lesions as well as osteochondral chip fracture using a Student's t-test. Pearson's correlation coefficient for linear regression was calculated. A p-value of <0.05 was interpreted as statistically significant.

Re s u 1 t s
All TB [13] had osteochondral fractures and one had moderate articular cartilage lesions. Four of the STB had osteochondral fractures, 2 had severe, 23 moderate, 12 mild articular cartilage lesions and 8 had a normal appearing articular cartilage. An additional five TB euthanized due to non-orthopaedic reasons, with normal articular cartilage, ligament and joint capsule were included as a control group. Hence the data evaluated were derived from 68 joints.
All 13 of the TB with osteochondral fractures (mean age 4 years) showed synovitis ( 3 severe, 8 moderate, 2 mild), no intercarpal ligament tear and a duration of Five TB with normal articular cartilage euthanized due to non-orthopaedic reasons lameness of less than one month. The four STB with osteochondral fractures (mean age 5.5 years) had synovitis (3 severe, 1 moderate), no intercarpal ligament tear and a duration of lameness of 1-2 months. All the osteochondral fractures of the STB were located in the intercarpal joint. The osteochondral fractures of the TB were located in the intercarpal [6] or radiocarpal [7] joints. The concentration of sf-COMP from the TB with osteochondral fractures was statistically significantly higher compared to normal joints (TB and STB) and joints from STB with articular cartilage lesions extending down to bone (classified as moderate and severe) including osteochondral fractures (Table 2). In contrast, there was no statistical difference in the concentration of sf-aggrecan in any of the joints examined ( Table 2). The ratio of aggrecan/COMP was statistically significantly lower in TB with osteochondral fractures compared to STB with osteochondral fractures (p = 0.04) and STB with articular cartilage lesions extending down to bone (p = 0.05).
The concentration of sf-COMP from the TB with osteochondral fractures increased with time after injury (I. = 0.57, p = 0.04) (Fig. l), whereas no significant alterations of sf-aggrecan were seen postinjury. SF-COMP decreased with age in the STB ( r = 0.34, p = 0.00). No such changes were seen in the TB (I. = 0.10, A positive correlation between articular cartilage lesions down to bone (in TB and STB) and synovitis was found (I. = 0.43, p = 0.00). STB horses with a more severe articular cartilage lesion showed a higher degree of synovitis (I. = 0.34, p = 0.00). However STB with intercarpal ligament involvement had less severe articular cartilage lesions ( r = -0.5, p = 0.00).

Western blot analysis of synoviui fluid for COMP
In synovial fluid from a TB with an osteochondral fragment, only high-molecular-weight anti-COMP reactive molecules were observed under non-reducing conditions. These most likely represent intact COMP, which under reducing conditions had an apparent molecular mass of N 100 kDa which corresponds to the predicted size of the intact subunit. Smaller sized bands were not detected under reducing or non-reducing Reducing Non-reducing OA OF OA OF were detected by the antiserum in reducing conditions (Fig. 2). conditions. In contrast, in the synovial fluid from the TB with severe OA lesions, smaller size fragments of COMP Microscopical analysis of stained sections showed a major loss of Toluidine Blue staining in the superficial and intermediate zones of the articular cartilage of the osteochondral fragments (Fig. 3a). The immunohistochemical staining with antibodies against COMP was most prominent in the pericellular matrix of the deep zone (Fig. 3b), but was also present to a moderate degree in the interterritorial matrix of the intermediate zone in some parts (data not shown). COMP mRNA was expressed in the intermediate and deeper zones of the articular cartilage of the osteochondral fragments (Fig. 3c). The in situ hybridisation signal was localised within the cytoplasm of most, but not all, chondrocytes whereas no staining was seen with the control sense probes. The articular cartilage from the opposite side of the fracture and articular cartilage with severe OA did not show detectable expression of COMP mRNA in the cells with the antisense probes.

Discussion
The aim of this study was to evaluate whether altered concentrations of COMP and aggrecan and/or their fragments released into the synovial fluid may assist in the clinical classification of articular cartilage lesions, synovitis and intercarpal ligament tears in the carpal joints of young racehorses with acute lameness. Our hypothesis was that the pathological changes within the equine joint from acutely lame horses will result in increased synthesis of extracellular matrix components, similar to what have been reported in joints from humans with OA [1,18,20,21]. The present study shows that TB suffering from acute osteochondral fractures has higher concentrations of sf-COMP compared to TB and STB with normal joints and STB with articular cartilage lesions and osteochondral fractures. A number of possible mechanisms could result in the observed increase in concentration of COMP in the TB. There may be an increase in protein synthesis with a proportion of the newly synthesized macromolecules released into the synovial fluid either by escaping incorporation into the extracellular matrix [22], or by a decrease or inhibition of protein degradation; alternatively, there may be a lowered rate of clearance of the protein from the joint cavity or a decrease in synovial fluid volume [17] thus raising the protein apparent concentration. An abnormal clearance rate from the joint cavity appears less likely since all TB with osteochondral fractures exhibited macroscopical synovitis suggesting there may be increased capillary filtration and lymph flow [45]. The synovial fluid volume did not differ between joints, which is in agreement with a previous study [42].
It is possible that degradative mechanisms are less active in the articular cartilage of the osteochondral fragments or that the architecture of the ECM of the cartilage is sufficiently altered to facilitate the loss of intact COMP into the synovial fluid. However, the increase in expression of COMP mRNA in the cells of the articular cartilage of the osteochondral fragments favours the argument for an increase in protein synthesis leading to COMP molecules released directly into the synovial fluid.
Furthermore, the presence of COMP predominantly in the form of intact protein in the synovial fluid from TB with osteochondral fragments suggests that degradative processes are limited in activity. In contrast, the majority of the sf-COMP from TB with OA was detected as low molecular weight fragments, which are in agreement with findings in equine and human OA studies [7,25].
Although in the present study the source of the increased concentration of COMP was the articular cartilage, however, it is possible that the subchondral bone plate, the intercarpal ligament and/or the synovial membrane could also contribute to this. The lack of COMP mRNA in the subchondral bone of the fractures and the negligible levels of COMP in the intercarpal ligament and the synovial membrane of the equine middle carpal joint [42] indicates against these tissues as the source for COMP.
The concentration of sf-COMP from the TB with osteochondral fractures increased significantly with the duration of lameness which may be related to ongoing inflammation and repair processes. However, previous studies have failed to correlate elevated COMP levels with joint inflammation, and in both human RA and animal models of RA inflammation and tissue destruction appear to be uncoupled processes [13,14,35]. Taken together, the data suggests the increase in COMP concentration may be an indicator of repair.
The dislocation of the osteochondral fragment results in a lower loading of the cartilage which could influence the synthetic activity of the chondrocytes. The chondrocytes in the cartilage specimen from the part of the fractured bone where load continues to be exerted did not express COMP mRNA, which is in agreement with our other study in which a decrease in COMP synthesis correlated with increased load [41]. This suggests that the reduced mechanical compression of the dislocated osteochondral fragment stimulates chondrocyte synthetic activity, thus contributing to the up-regulation of COMP observed in the articular cartilage in this region. In the STB with osteochondral fractures sf-COMP remained unchanged. A low concentration of sf-COMP has been correlated to strenuous training in the STB [41], suggesting that the difference in cartilage metabolic activity found in the two breeds could be related to the amount of training they have been exposed to. An agedependent decrease in sf-COMP concentration was noted in the STB but this was not the case in the TB, which is consistent with data from previous studies [42]. The strenuous training during maturation in the STB appears to inhibit COMP synthesis. The TB and STB breeds of race horses start to train at an early age (2 years old), but the loading experienced within the carpal joint is dependent on the gait at which the horses perform [31]. Galloping in the TB generates a greater force compared to trotting in the STB, however during trotting the third carpal bone is subjected to more repetitive cyclic load with forces of lower magnitude [39]. These conditions reflected the joint pathology observed in the two breeds, with most of the TB exhibiting osteochondral fractures whilst the most prominent feature of the STB joints was OA. Arthroscopic surgery with removal of the osteochondral fragments gives immediate relief of pain and is thought to prevent the development of chronic OA [23]. Studies of performance after surgery in 445 horses with carpal fractures show that 68% of the horses raced at a level equal to or better than preinjury levels [24]. One indicator of improved prognosis of these fractures could be the use of 162 E. Skioldehrand t't al. I Journal of Orthopaedic Research 23 (2005) [156][157][158][159][160][161][162][163] sf-COMP measurements to determine synthetic or degradative activities.
There was no statistical difference in the concentration of sf-aggrecan between the two breeds and between different pathological changes in the joint. The clear difference in synovial fluid concentrations between COMP and aggrecan is likely to be a result of the different mechanisms involved in the metabolism of these proteins in the equine joint. This is further supporter by findings that different cartilage matrix components are released at distinct stages of development in joint diseases [37], and in explant cultures of articular cartilage stimulated with IL-1 where there is a time-dependent release of aggrecan, COMP and collagen [5]. More recent studies have shown that patients with psoriatic arthritis have a high concentration of sf-COMP and a low concentration of sf-aggrecan compared to patients with rheumatoid arthritis [22]. Although we can only speculate that the increased sf-COMP in psoriatic arthritis may be representative of newly synthesised molecules following a repair response, the results are in agreement with the low aggrecan/COMP ratio in the TB with osteochondral fractures. Measurements of aggrecan/COMP ratios may also have applications in monitoring prognosis after intra-articular fractures in human joints.
The data from the current study shows an enhanced concentration of sf-COMP in joints from TB with osteochondral fractures. This indicates that elevated sf-COMP concentration can be a useful marker for diagnosis of and monitoring equine intercarpal joint fractures in the TB. The release of COMP predominantly as an intact protein coupled with the elevated expression of COMP mRNA in the chondrocytes of dislocated fractures suggests that protein synthesis is increased. Furthermore, the increase in sf-COMP concentration correlated to the length of time postinjury, suggesting that COMP may be a suitable marker for longitudinal studies to evaluate its role in joint healing.