EQUINE VETERINARY JOURNAL Equine vet. J. (2005) 37 (1) 31-36 31 Analysis of cartilage oligomeric matrix protein (COMP) degradation and synthesis in equine joint disease K. ARAI, K. MISUMI*, S. D. CARTER†, S. SHINBARA, M. FUJIKI and H. SAKAMOTO *Department of Veterinary Medicine, Kagoshima University, 21-24 Korimoto 1-chome, Kagoshima 890-0065, Japan; and †Departments of Veterinary Pathology and Veterinary Clinical Science, University of Liverpool, PO Box 147, Liverpool L69 3BX, UK. Keywords: horse; COMP; fragmentation; monoclonal antibody; osteoarthritis Summary Reasons for performing study: Cartilage oligomeric matrix protein (COMP) is abundant within cartilage; its turnover and/or degradation have been investigated in various equine joint diseases and it has been suggested that COMP fragmentation might be useful for monitoring such conditions. Objectives: To determine whether COMP metabolism is compromised in equine osteoarthritis (OA) and whether COMP degradation is a useful joint marker representing cartilage destruction. Hypothesis: A monoclonal antibody (mAb) with a higher affinity for degraded COMP allows discrimination of diseased joints by quantifying COMP levels and fragmentation. Methods: A mAb (clone14G4) was generated against equine cartilage COMP. The NH2-terminal sequence of enzyme-cut COMP fragments recognised by 14G4 was determined, as was the efficiency of binding to COMP (using a generated COMP peptide). COMP concentration and fragmentation were analysed in synovial fluid (SF) from normal horses and those with OA. Results: The mAb 14G4 had a higher affinity for the smaller fragments of equine COMP, compared with a mAb (clone 12C4) generated against human COMP. The 14G4 epitope was identified as between C134 and F147. The COMP values in OA (mean ± s.d. 205.8 ± 90.9 µg/ml) were significantly higher than in the normal (133.1 ± 31.5 µg/ml) SF. On the immunoblots of OA sample, the proportions of intact COMP were significantly lower, while smaller fragments ranging from 75 to 290 kDa were higher compared with the normal SF. Conclusions and potential relevance: The mAb 14G4 reliably detects COMP degradation as well as synthesis, and fragmentation analysis combined with quantification in SF could be useful to study equine OA. Introduction Cartilage oligomeric matrix protein (COMP), a member of the thrombospondin family (Oldberg et al. 1992), is an abundant noncollagenous extracellular matrix protein in cartilage (Hedbom et al. 1992). COMP is considered important in binding collagen *Author to whom correspondence should be addressed. [Paper received for publication 12.03.04; Accepted 01.06.04] types I and II via its C-terminal globular domain (Rosenberg et al. 1998) and type IX collagen (Holden et al. 2001). Along with data showing binding to fibronectin (Di Cesare et al. 2002), it seems that COMP plays an important role in a well-constructed collagen bundle and network, probably associated with the determination of fibril diameter and/or orientation and interaction with other matrix molecules. In human osteoarthritis (OA), COMP measurements in sera and synovial fluids (SF) have been proposed as a helpful biomarker to predict disease-mediated damage to joint cartilage (Clark et al. 1999; Jordan et al. 2003). Furthermore, the analysis of COMP fragments could be useful for defining the active enzyme pattern in the cartilage matrix in OA associated with the degree of clinical/pathological severity, because some studies have demonstrated that COMP can be cleaved by the proteolytic enzymes associated with OA (Ganu et al. 1998; Stracke et al. 2000; Dickinson et al. 2003). Since Smith et al. (1997) first purified equine COMP from tendon and prepared a specific antibody, its turnover and/or degradation have been investigated in tendonitis (Smith and Heingård 2000), navicular disease (Viitanen et al. 2001) and OA (Skiöldebrand et al. 2001). In a previous report showing the degradation of COMP in SF and sera based on a COMP ELISA and immunoblotting with a monoclonal antibody (mAb, clone12C4) to human cartilage COMP (Misumi et al. 2001), we suggested that COMP fragmentation might be useful for monitoring equine joint disease. The mAb 12C4 was useful because of its cross-reaction with equine cartilage COMP, but was limited in its usefulness because of its lack of binding to small equine COMP fragments (Misumi et al. 2001). Such a property of 12C4 could be explained by the evidence that its antigenic epitope localises to the beginning of the COOHterminal globular domain (Vilim et al. 2003). In diseased SF, COMP is a substrate for proteases including different classes of matrix metalloproteinase (MMP) and disintegrin and metalloproteinase with thrombospondin motifs (Dickinson et al. 2003). If the COOH-terminal domain is more susceptible to proteolytic cleavage compared with the NH2-terminal helical domain (Vilim et al. 2003), 12C4 may not recognise degraded COMP fragments lacking the COOH-terminal domain. To accurately define and describe COMP degradation in OA, it is important to have an antibody which can detect the smaller COMP fragments resulting from enzyme activity (in SF and 32 Analysis of COMP degradation in equine joint disease Nonreducing AP ECL Reducing AP ECL kDa - 207 - 119 - 98.5 - 56.7 - 29.5 1 2 1 2 1 2 1 2 Fig 1: Western blot analysis of COMP detected by mAbs 12C4 (1) and 14G4 (2) with alkaline phosphatase reaction (AP) and enhanced chemiluminescence (ECL). Both the nonreduced pentamer and the reduced monomer were bound by 14G4 as well as 12C4. Under the reducing conditions, 14G4 showed higher affinity for the low molecular weight fragments of COMP (Mr<50 kDa), compared with 12C4. the matrix) instead of just detecting the intact or larger molecules generated in any up-regulation of COMP turnover in OA. In this study, we prepared a monoclonal antibody to equine articular COMP which readily detected COMP degradation, and allowed analysis of COMP fragmentation in SF from horses with joint diseases. Materials and methods Samples All samples were from Thoroughbred or Anglo-Arab racehorses. One hundred SF samples were collected from horses with OA, some of which were secondary to osteochondral fracture or osteochondrosis dissecans, either during diagnostic procedures or prior to surgery. Eighty SF (normal) samples were obtained from horses judged to be free of any osteoarthropathies on clinical, radiological or arthroscopic examinations. The samples were centrifuged to remove debris and cells, and the supernatants were stored at -70°C until the assay. Prior to the assay, all SF samples were pretreated with hyaluronidase according to Neidhart et al. (1997). Preparation of equine COMP antigen and monoclonal antibody COMP antigen was prepared from equine articular cartilage as described previously (Misumi et al. 2001). The purity of the preparation was analysed by sodium dodecyl sulphatepolyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting with a mAb12C4, which cross-reacts with equine cartilage COMP (Vilim et al. 1997). The peptide sequences of COMP fragments generated by enzyme digestion were determined to confirm the published sequence for horse COMP (data not shown). Six-week-old female BALB/c mice were inoculated every 2 weeks with a total of 4 injections of equine cartilage COMP (50 µ g COMP/mouse). The first immunisation was by intraperitoneal injection of the antigen mixed with Freund’s complete adjuvant. At the following 2 booster immunisations, the mice were injected intraperitoneally with COMP in incomplete Freund’s adjuvant. Five days after the third injections, rising levels of COMP antibody were confirmed in venous blood samples. At the last immunisation, 4 days before fusion, half of the antigen (25 µg) was injected intraperitoneally and the rest intravenously. Splenocytes from the immunised mice were isolated and fused with the murine myeloma cell line P3U1. Hybridoma supernatants were subsequently screened by a direct enzyme-linked immunosorbent assay (ELISA) for the presence of mouse immunoglobulins and then for the presence of antibodies against COMP. Briefly, culture media from immunoglobulin-positive wells were tested for COMPbinding antibodies by ELISA using microtitre plates coated with equine cartilage COMP (5 µg/ml). Positive colonies of hybridomas were expanded and subcloned at least 6 times by the method of limiting dilution; the culture media were further examined by immunoblotting against reduced and nonreduced COMP antigen. A hybridoma clone (14G4) was selected to produce ascitic fluid. Hybridoma cells (1 x 107) were injected intraperitoneally into 12-week-old male BALB/c mice previously injected with 0.5 ml Pristane (Sigma T7640). Ascitic fluids were harvested 2–3 weeks later, cells and debris were removed by centrifugation at 10,000 g at 4°C for 10 mins, and the globulin content of the fluid was precipitated at least twice with ammonium sulphate. After dialysis, the globulins were stored at -80°C until use. The immunoglobulin class was determined using a mouse immunoglobulin typing kit (RK008)1. Epitope mapping of enzymatically digested fragments Purified COMP from equine cartilage was treated with TPCKtrypsin2 or lysyl endopeptidase2 under reducing or nonreducing conditions. TPCK-trypsin (at the enzyme/substrate ratio 1:500, w/w) digestion was at 37°C for 6 different times (0, 10, 30, 60, 180 and 300 mins). Lysyl endopeptidase was applied at the varying enzyme/substrate ratios; 0.0022, 0.022, 0.044, 0.22 and 2.2 AU/pg) at 37°C for 24 h. Following SDS-PAGE, immunoblotted COMP fragments were probed with 14G4, as described previously (Misumi et al. 2001). Briefly, following SDS-PAGE on 4–15% gradient gels, the gels were electrotransferred onto polyvinyl difluoride (PVDF) membranes. After blocking, mAb 14G4, diluted 1:20,000 in 2% skimmed milk in Tris-buffered saline containing 0.05% Tween 20 (TBS/Tween), was applied to the membrane. The mAb 12C4 (diluted 1:4000) was used as a positive control to compare the binding of 14G4 with COMP fragments. Positive binding to COMP molecules was detected by alkaline phosphatase (AP) conjugated goat anti-mouse IgG antibody (A3688)2 diluted 1:10,000 in 2% skimmed milk in TBS/Tween, and demonstrated by the development of reaction with a substrate (BCIP/NBT, Sigma Fast/B-5655)2. Based on the preliminary blots, an optimal protocol of COMP digestion was chosen for the NH2-terminal sequencing. After digesting the reduced COMP with lysyl endopeptidase (2.2 AU/pg) at 37°C for 24 h, COMP fragments were separated by SDS-PAGE and transferred onto PVDF membrane in 10 mmol/l CAPS, 10% methanol, pH 11.0, at 70 V, 4°C, for 6 h. Selected bands were cut out of Coomassie Bluestained blots, and sequences obtained by Edman degradation in a protein sequencer (Procise Protein Sequencing System)3. Finally, a predicted COMP peptide was synthesised (Qiagen)4 to determine the affinity of mAb 14G4 for its epitope by direct ELISA. K. Arai et al. 33 12C4 a) 12C4 b) 14G4 14G4 kDa kDa 214 214 kDa kDa 214 214 118 92 118 92 118 92 118 92 52.2 52.2 52.2 52.2 35.7 28.9 20.8 35.7 28.9 20.8 35.7 28.9 35.7 28.9 0 10 30 60 180 300 0 10 30 60 180 300 20.8 20.8 1 2 3 4 5 1 2 3 4 5 Time (mins) Fig 2: Immunoblots of enzymatically cleaved COMP by mAbs 12C4 and 14G4 with alkaline phosphate reaction. On the blots with 14G4, the signal of tryptic COMP faded away in the course of digestion (a), while lysyl endopeptidase produced many smaller fragments as the concentration increased (b). With 12C4, smaller tryptic bands increased over time (a), but the 12C4 epitope in the fragments progressively disappeared with lysyl endopeptidase digestion (b); Enzyme/substrate (Lysyl endopeptidase) ratio (AU/pg values): 1 = 0.0022; 2 = 0.022; 3 = 0.044; 4 = 0.22; 5 = 2.2. Inhibition ELISA An inhibition ELISA to quantify COMP in SF and sera was designed according to previous protocols (Misumi et al. 2001). In brief, 50 µl purified horse COMP antigen in a coating buffer (20 mmol/l sodium carbonate, 20 mmol/l sodium bicarbonate, 0.002% sodium azide, pH 10) was placed into each well at 5 µg/ml and incubated for 2 h at room temperature then overnight at 4°C. Seventy µl of diluted standards (range 13.5–0.01 µg/ml) and SF samples (final dilutions 1/100) were mixed with the same volume of mAb 14G4 (final dilution 1/200,000) in phosphatebuffered saline (PBS) containing 0.05% Tween 20 (PBS/Tween), and then incubated overnight at 4°C. Coated wells were washed with PBS, blocked and incubated with 100 µl/well of the COMPantibody mixture for 1 h at room temperature then for 1 h at 4°C. The primary antibody binding to COMP on the plate was detected by AP reaction. One hour later, the production of chromophore was stopped by addition of 2 mol/l H2CO3 (30 µl/well), and the absorbance at 405 nm (OD405) was read. To determine the intraand interassay variability, fresh aliquots of an SF sample were thawed and measured 10 repeats on one plate for 10 consecutive working days (10 plates). expressed as mean ± 1 s.d. As data were normally distributed, we selected the parametric means from statistical analysis. The differences in COMP concentrations between the groups (normal vs. OA) were analysed by factorial ANOVA, and Scheffé’s method was used for simultaneous multiple comparisons. A P value <0.01 was considered statistically significant. Results Characterisation and epitope mapping of monoclonal antibody 14G4 As shown on the immunoblots using AP reaction or ECL (Fig 1), mAb 14G4 bound to both the nonreduced oligomeric and reduced monomeric equine COMP molecules, as did mAb 12C4. The 14G4 also identified some smaller fragment bands, which were not well detected by antibody 12C4. On the immunoblots with 14G4, the detectable COMP signal faded away as the tryptic digestion time progressed, while the cleavage with lysyl endopeptidase produced a lot of smaller antigenic fragments as the enzyme concentration increased (Fig 2). With 12C4, detectable smaller fragments in tryptic digest increased over time, but the positive bands progressively disappeared in lysyl endopeptidase digestion (Fig 2). The NH2-terminal sequence of a Electrophoresis and immunoblotting of synovial fluids As described above, following SDS-PAGE on 4–20% gradient gels, SF proteins were electrotransferred onto PVDF membranes. After blocking, mAb 14G4 diluted 1:10,000 in 2% skimmed milk in TBS/Tween was applied to the membrane. Positive binding of 14G4 to COMP fragments was detected by 1:20,000 dilution of a goat anti-mouse IgG horseradish peroxidase (HRP) conjugated antibody (sc-2005)5 in 2% skimmed milk in TBS/Tween. Signal of HRP reaction was detected by enhanced chemiluminescence (ECL plus)6. The membrane images were captured by digital camera and the relative intensity of the bands was determined by densitometry using relevant software7. Statistics All quantitative group data obtained in ELISA and immunoblotting were analysed for the normality by box plot, and kDa 214 118 92 52.2 35.7 28.9 20.8 NTVMECDACGMQ Fig 3: NH2-terminal sequence of a 14G4-positive fragment (approximately 20 kDa) generated by lysyl endopeptidase digestion under reducing conditions. The sequence was determined as N63TVMECDACGMQ74. 34 Analysis of COMP degradation in equine joint disease 0.8 a) Purified COMP (µg/ml) SF Serum Urine 0.7 OD405 0.6 0.5 0.4 kDa 214 118 92 CF1 CF2 CF3 CF4 CF5 52.2 35.7 CF6 20.8 0.3 Normal OA 0.2 c) b) 0.1 0.01 0.1 0 10 100 1000 10000 Log [µg purified COMP/ml] 0.00001 0.0001 0.001 0.01 0.1 1 10 Log [dilution ratio of SF, serum and urine] Fig 4: Inhibition ELISA standard curves. The linear portion using purified COMP was considered to be between 0.8 and 13.5 µg/ml and parallel to the inhibition curves obtained by diluting serum and urine as well as synovial fluid (SF). Mean + 2.58 s.d. COMP (µg/ml) 700 Mean + 1.96 s.d. 600 Mean + s.d. 500 400 OA Fig 6: Immunoblots of synovial fluid (SF) from normal joints and those with osteoarthritis (OA) by 14G4 with enhanced chemiluminescence (ECL). COMP in SF was detected by monoclonal antibody 14G4 under nonreducing conditions. The fragmentation of COMP in the normal SF (b) was relatively less severe than in OA SF (c). On the diseased immunoblots, there were 6 consistently detected fragment bands (a), which were over 290 kDa (CF1), approximately 290 kDa (CF2), 220–170 kDa (CF3), 140–110 kDa (CF4), 95–75 kDa (CF5) and 45–35 kDa (CF6). Mean 300 200 Mean - s.d. 100 Mean - 1.96 s.d. 0 Normal Mean - 2.58 s.d. Normal SF (n = 80) OA SF (n = 100) Fig 5: COMP measurements in equine synovial fluids (SF). The mean COMP value in osteoarthritis (OA SF) (205.8 ± 90.9 µg/ml) was significantly (P<0.0001) higher than in the normal (133.1 ± 31.5 µg/ml) sample. o and x = outliers. 14G4-positive fragment (approximately 20 kDa), which was generated by lysyl endopeptidase digestion under reducing condition, was determined as N63TVMECDACGMQ74 (Fig 3). A synthetic peptide of C134FPRVRCINTSPGF147 was prepared, which is included in the 20 kDa fragment subsequent to N63, rich in arginine (R; digestible for trypsin) but not lysine (K; resistible to lysyl endopeptidase) and well conserved across several different species (mAb 14G4 cross-reacts to COMP from several different animals, data not shown). The peptide was recognised by 14G4. ELISA measurement of COMP levels in synovial fluids using 14G4 In the standard curve of purified COMP, the linear portion was considered to be between 0.8 and 13.5 µg equine cartilage COMP/ml, and parallel to the inhibition curves obtained by diluting serum, urine and SF sample (Fig 4). The mean COMP value of repeat measurements of one SF was 70.2 µg/ml, and the intra- and interassay variability were calculated to be 7.1 and 6.7%, respectively. Using the standard curve, the COMP values in OA SF (205.8 ± 90.9 µg/ml) were significantly (P<0.0001) higher than in normal SF (133.1 ± 31.5 µg/ml) (Fig 5). Analysis of COMP fragmentation in synovial fluids using 14G4 The SF immunoblots probed with 14G4 are shown in Figure 6. The fragmentation of COMP molecules in the normal SF (Fig 6b) was relatively much less pronounced than in OA samples (Fig 6c). On OA immunoblots, there were 6 consistently detected fragment bands (Fig 6a), which were one fragment over 290 kDa (CF1); approximately 290 kDa (CF2); 220–170 kDa (CF3); 140–110 kDa (CF4); 95–75 kDa (CF5) and 45–35 kDa (CF6) fragments. Relative percentage of each fragment per total antigenic COMP (measure by densitometric analysis) showed that, in OA SF, CF1 was lower (P<0.0001) while CF2, CF3, CF4 and CF5 fragments were significantly higher, compared with the normal SF samples (Table 1). TABLE 1: Relative percentage of COMP fragments in synovial fluids (SF) COMP fragment (CF): kDa CF CF CF CF CF CF 1: 2: 3: 4: 5: 6: >290 290 220–170 140–110 95–75 45–35 Normal SF (%) 92.5 4.1 2.8 0.5 0.1 0 ± ± ± ± ± ± 8.4 7.0 5.6 1.2 0.2 0 OA SF(%) 71.2 14.4 11.6 2.5 0.3 0.1 ± ± ± ± ± ± 15.0 125 11.6 6.1 0.6 0.7 P value <0.0001 <0.0001 <0.0001 0.0049 0.0036 0.07 (NS) Fragmentation is shown as relative percentage (± 1 s.d.) of total detectable COMP in each sample and derived from image density analysis of SF Western blots. OA = osteoarthritis; NS = not statistically significant. K. Arai et al. 35 Discussion lysine (K) because the cleavage sites of trypsin are the R or K; 2) being resistible to cleavage by lysyl endopeptidase, meaning that the epitope does not have lysine because the cleavage site of the enzyme is K; and 3) being conserved across several different species, as mAb14G4 cross-reacts with human, canine, rat and mouse COMP molecules (data not shown). To that end, we prepared a synthetic peptide of C134FPRVRCINTSPGF147 at the EGF-like repeats, which includes 2 arginines and is a consensus sequence among the species analysed. A direct ELISA showed that the peptide was recognised by 14G4. This evidence may explain the different affinities of the 2 antibodies for smaller COMP fragments, with 14G4 recognising more degraded COMP in SF compared to 12C4. COMP fragments in osteoarthritic SF would be monomeric or oligomeric. Since the epitope of mAb 12C4 is near the COOHterminal globular domain, this antibody can recognise large-sized COMP oligomers which are less degraded in osteoarthritic SF, while the consequent monomeric units including the 12C4 epitope might be so small as to not be retained in the joint fluid. On the other hand, the 14G4 epitope, which locates nearer to the NH2-terminal, might be conserved well in smaller, more degraded COMP oligomers. Furthermore, the monomeric fragments, including the 14G4 epitope, would be less likely to be eliminated due to their larger size compared with fragments containing the 12C4 epitope. If the NH2-terminal helical domain is more resistant to proteolysis than the COOH-terminal globular domain (Vilim et al. 2003), larger oligomeric COMP fragments would be expected to be retained more in diseased SF rather than the smaller fragments. In the present study, the increased COMP measurements with 14G4 in diseased SF could be explained as being due to 14G4 recognising more degraded small-sized COMP oligomers (as well as intact COMP). In addition, it seems that the 14G4 epitope is relatively unaffected in the pathological degradation of COMP in SF, as compared with the epitope of 12C4. Consequently, mAb 12C4 would probably be more useful for an ELISA system to monitor the up-regulated COMP turnover (i.e. COMP synthesis) rather than degradation within the equine diseased joints, while mAb14G4 would be valuable for identifying and quantifying COMP fragmentation. Interestingly, the 14G4 epitope can be detected in urine from dogs, man, rats, mice and horses (K. Arai et al., unpublished data). There have been no previous reports of COMP measurements in urine, and the 14G4 epitope may be a promising tool to monitor COMP turnover in human OA and rheumatoid arthritis. The fragmentation pattern of COMP seen in SF may represent increased protease activity in OA joints (Ganu et al. 1998; Stracke et al. 2000). A recent study (Fang et al. 2001) on western blot analysis of COMP in human SF from the injured and contralateral (normal) joints showed 3 low molecular weight degradation fragments in injured joints. These corresponded to bands at >100, 70<Mr<90 and Mr<60 kDa, while only intact COMP and the >100 kDa fragment were found in normal joints. These data are similar to our data in which 6 molecular weight COMP fragments ranging from 35 to 290 kDa were more readily identified in osteoarthritic SF than in normal SF. It is difficult to be sure that our mAb 14G4 detected all fragments of COMP or whether the fragments identified were entirely due to enzymatic degradation. However, fragment analysis by the immunoblotting of synovial fluids combined with COMP measurements by ELISA may be a more reliable tool to understand the mechanisms responsible for degradation of the matrix molecules in osteoarthritic joints. The In previous studies, we analysed equine COMP using mAb 12C4 generated against human cartilage COMP, and reported that the COMP levels in synovial fluids and sera from osteoarthritic horses were significantly decreased compared with those for normal horses (Misumi et al. 2001, 2002a). That result was at variance with some other data from man and animals, where COMP levels were higher in synovial fluids from patients with joint diseases than in controls (Clark et al. 1999; Skiöldebrand et al. 2001; Misumi et al. 2002b). The discrepancy of the results could be due to several factors, although two are most likely; and the first would be the different stages of OA which may be investigated in different studies. The second would be the different affinity or specificity of the primary antibodies used for COMP detection. In human studies, the COMP assay has used either rabbit polyclonal antisera or monoclonal antibodies against COMP. One report suggested that a polyclonal antibody showed higher affinity for the lower molecular weight fragments of COMP degraded in OA, while monoclonal antibodies preferentially bound the intact or less degraded COMP molecules (Vilim et al. 1997). Similarly, the monoclonal anti-human COMP antibody 12C4 used in our studies showed a higher affinity for the less degraded equine COMP molecules, as demonstrated by immunoblotting. Consequently, we speculated that the lower COMP values seen in OA sera and joint fluids represent the disappearance or depletion of the epitopes detected by 12C4 in accordance with cartilage destruction and associated COMP degradation. If so, 12C4 would be less useful for defining COMP degradation in the progressive stages of cartilage pathology when the cartilage matrix is largely intact. Monoclonal antibodies have an advantage in that their antigenic specificity is easier to select and define than that of polyclonal antibodies; this property is important to obtain reproducible data. If the monoclonal antibody was prepared against an antigenic component of horse COMP and was able to identify and map the process of the COMP degradation, it would be a much more useful tool than those used previously. To our knowledge, this is the first report describing a monoclonal antibody against equine cartilage COMP. This new mAb 14G4 recognised the same nonreduced oligomeric and reduced monomeric COMP molecules purified from equine articular cartilage as mAb 12C4 produced against human COMP. Furthermore, 12C4 showed higher affinity for the high molecular sizes of COMP, rather than the more degraded small fragments. Conversely, 14G4 identified not only the less degraded large molecular bands, but also the smaller fragment bands (MW <50 kDa) which were not identified by 12C4. This property of 14G4 could also explain the ELISA data for COMP measurements in SF from normal and OA horses which were obtained using 14G4; in this assay, the COMP levels in OA samples were significantly higher compared with those of the normal samples. The different affinity for COMP fragments between the 2 antibodies could be dependent on the location of the epitopes. A previous study demonstrated that the 12C4 epitope was localised to the beginning of the COOH-terminal globular domain (Glu525–Ala757) (Vilim et al. 2003). The epitope mapping data in the present study showed that 14G4 might recognise a specific peptide included in an approximately 20 kDa fragment which starts at N63TVMECDACGMQ74 of the NH2-terminal sequence, with the following 3 properties: 1) disappearance after tryptic digestion, meaning that the epitope could be rich in arginine (R) or 36 Analysis of COMP degradation in equine joint disease identification of smaller fragments on the blots could well be indicative of the presence of any arthropathy rather than being specific for the osteoarthritic processes of cartilage matrix degradation. Further studies to identify cleavage sites of fragment bands seen on the immunoblots and analyse the relationship between the fragment pattern and proteolytic activity would be useful. Holden, P., Meadows, R.S., Chapman, K.L., Grant, M.E., Kadler, K.E. and Briggs, M.D. (2001) Cartilage oligomeric matrix protein interacts with type IX collagen and disruptions to these interactions identify a pathogenetic mechanism in a bone dysplasia family. J. Biol. Chem. 276, 6046-6055. Jordan, J.M., Luta, G., Stabler, T., Renner, J.B., Dragomir, A.D., Vilim, V., Hochberg, M.C., Helmick, C.G. and Kraus, V.B. (2003) Ethnic and sex differences in serum levels of cartilage oligomeric matrix protein. The Johnston County Osteoarthritis project. Arthritis Rheum. 48, 675-681. Misumi, K., Vilim, V., Clegg, P.D., Thompson, C.M. and Carter, S.D. (2001) Measurement of cartilage oligomeric matrix protein (COMP) in normal and diseased equine synovial fluids. Osteoarthritis Cartilage 9, 119-127. Acknowledgements This study was supported by The Home of Rest for Horses. The authors would like to thank Dr V. Vilim, Institute of Rheumatology, Praha, for supplying a monoclonal antibody (12C4). Misumi, K., Vilim, V., Hatazoe, T., Murata, T., Fujiki, M., Oka, T., Sakamoto, H. and Carter, S.D. (2002a) Serum level of cartilage oligomeric matrix protein (COMP) in equine osteoarthritis. Equine vet. J. 34, 602-608. Misumi, K., Vilim, V., Carter, S.D., Ichihashi, K., Oka, T. and Sakamoto, H. (2002b) Concentrations of cartilage oligomeric matrix protein in dogs with naturally developing and experimentally induced arthropathy. Am. J. vet. Res. 63, 598-603. Manufacturers’ addresses Neidhart, M., Hauser, N., Paulsson, M., DiCesare, P.E., Michel, B.A. and Häuselmann, H.J. (1997) Small fragments of cartilage oligomeric matrix protein in synovial fluid and serum as markers for cartilage degeneration. Br. J. Rheum. 36, 1151-1160. 1The Binding Site Ltd., Birmingham, West Midlands, UK. Chemical Co., St. louis, Missouri, USA. 3PE Applied Biosystems Inc., Foster City, California, USA. 4Qiagen KK., Tokyo, Japan. 5Santa Cruz Biotechnology Inc., Santa Cruz, California, USA. 6Amersham-Pharmacia Biotech., Piscataway, New Jersey, USA. 7PDI Inc., New York, USA. 2Sigma Oldberg, Å., Antonsson, P., Lindblom, K. and Heinegård, D. (1992) COMP (cartilage oligomeric matrix protein) is structurally related to the thrombospondins. J. Biol. Chem. 267, 22346-22350. References Rosenberg, K., Olsson, H., Mörgelin, M. and Heinegård, D. (1998) Cartilage oligomeric matrix protein shows high affinity zinc-dependent interaction with triple helical collagen. J. Biol. Chem. 273, 20397-20403. Clark, A.G., Jordan, J.M., Vilim, V., Renner, J.B., Dragomir, A.D., Luta, G. and Kraus, V.B. (1999) Serum cartilage oligomeric matrix protein reflects osteoarthritis presence and severity. Arthritis Rheum. 42, 2356-2364. Skiöldebrand, E., Lorenzo, P., Zunino, L., Rucklidge, G.J., Sandgren, B., Carlsten, J. and Ekman, S. (2001) Concentration of collagen, aggrecan and cartilage oligomeric matrix protein (COMP) in synovial fluid from equine middle carpal joints. Equine vet. J. 33, 394-402. Di Cesare, P.E., Chen, F.S., Moergelin, M., Carlson, C.S., Leslie, M.P., Perris, R. and Fang, C. (2002) Matrix-matrix interaction of cartilage oligomeric matrix protein and fibronectin. Matrix Biol. 21, 461-470. Smith, R.K.W. and Heinegård, D. (2000) Cartilage oligomeric matrix protein (COMP) levels in digital sheath synovial fluid and serum with tendon injury. Equine vet. J. 28, 52-56. Dickinson, S.C., Vankemmelbeke, M.N., Buttle, D.J., Rosenberg, K., Heinegard, D. and Hollander, A.P. (2003) Cleavage of cartilage oligomeric matrix protein (thrombospondin-5) by matrix metalloproteinases and a disintegrin and metalloproteinase with thrombospondin motifs. Matrix Biol. 22, 267-278. Smith, R.K.W., Zunino, L., Webbon, P.M. and Heinegård, D. (1997) The distribution of cartilage oligomeric matrix protein (COMP) in tendon and its variation with tendon site, age and load. Matrix Biol. 16, 255-271. Fang, C., Johnson, D., Leslie, M.P., Carlson, C.S., Robbins, M. and Di Cesare, P.E. (2001) Tissue distribution and measurement of cartilage oligomeric matrix protein in patients with magnetic resonance imaging detected bone bruises after acute anterior cruciate ligament tears. J. Orthop. Res. 19, 634-641. Stracke, J.O., Fosang, A.J., Last, K., Mercuri, F.A., Pendas, A.M., Llano, E., Perris, R., Di Cesare, P.E., Murphy, G. and Knauper, V. (2000) Matrix metalloproteinases 19 and 20 cleave aggrecan and cartilage oligomeric matrix protein (COMP). FEBS Lett. 28, 52-56. Ganu, V., Goldberg, R., Peppard, J., Rediske, J., Melton, R., Hu, S.I., Wang, W., Duvander, C. and Heinegard, D. (1998) Inhibition of interleukin-1alpha-induced cartilage oligomeric matrix protein degradation in bovine articular cartilage by matrix metalloproteinase inhibitors: potential role for matrix metalloproteinases in the generation of cartilage oligomeric matrix protein fragments in arthritic synovial fluid. Arthritis Rheum. 41, 2143-2151. Viitanen, M., Bird, J., Makela, O., Schramme, M., Smith, R., Tulamo, R.-M. and May, S. (2001) Synovial fluid studies in navicular disease. Res. vet. Sci. 71, 201-206. Vilim, V., Lenz, M.E., Vytasek, R., Masuda, K., Pavelka, K., Kuettner, K.E. and Thonar, E.J-M.A. (1997) Characterization of monoclonal antibodies recognising different fragments of cartilage oligomeric matrix protein in human body fluids. Arch. Biochem. Biophys. 341, 8-16. Hedbom, E., Antonsson, P., Hjerpe, A., Aeschlimann, D., Paulsson, M., RosaPimente, E., Sommarin, Y., Wendel, M., Oldberg, Å. and Heinegård, D. (1992) Cartilage matrix proteins. An acidic oligomeric protein (COMP) detected only in cartilage. J. Biol. Chem. 267, 6132-6136. Vilim, V., Voburka, Z., Vytasek, R., Senolt, L., Tchetverikov, I., Kraus, V.B. and Pavelka, K. (2003) Monoclonal antibodies to human cartilage oligomeric matrix protein: epitope mapping and characterization of sandwich ELISA. Clin. Chim. Acta 328, 59-69. Errata 1) The article Efficacy of oral and intravenous dexamethasone in horses with recurrent airway obstruction by C. J. Cornelisse, N. E. Robinson, C. E. A. Berney, C. A. Kobe, D. T. Boruta and F. J. Derksen was published in Equine Veterinary Journal Volume 36, pp 426-430. The key for Figure 2 (p 429) was printed with incorrect labels, for which we apologise. The correctly labelled key appears below. Saline i.v. Dex i.v. (0.1 mg/kg bwt) HDF (0.164 mg/kg bwt, fed) HDNF (0.164 mg/kg bwt, not fed) LDNF (0.082 mg/kg bwt, not fed) 2) The article Effects of manipulating intrauterine growth on post natal adrenocortical development and other parameters of maturity in neonatal foals by J. C. Ousey, P. D. Rossdale, A. L. Fowden, L. Palmer, C. Turnbull and W. R. Allen was published in Equine Veterinary Journal Volume 36, pp 616-621. The address of the authors C. Turnbull and W. R. Allen was listed incorrectly, for which we apologise. The correct affiliation for these authors is: University of Cambridge, Equine Fertility Unit, Mertoun Paddocks, Woodditton Road, Newmarket, Suffolk CB8 9BH, UK.