OSTEOARTHRITIS and CARTILAGE

Transcription

1 Osteoarthritis and Cartilage (1994) 2, Osteoarthritis Research Society /94/ $08.00/0 OSTEOARTHRITIS and CARTILAGE Changes in aggrecan populations in experimental osteoarthritis BY MARK E. ADAMS Department of Medicine, University of Calgary, Alberta, Canada Summary Adul~ articular cartilage contains at least two distinct populations of aggrecan: one is larger and richer in chondroitm sulfate (CS), while the other is smaller with less CS. The smaller form is thought to be derived from the larger. The amount of CS in cartilage decreases with maturation and aging, mainly because of a decrease in the proportion of the larger of these two proteoglycans. In. early osteoarthritis (OA) the amount and concentration of CS in cartilage increases. To test the hypothesis that an increase in the more CS-rich form of aggrecan contributes to the increase in CS in cartilage in OA, experimental OA was induced in dogs by transection of the anterior erucia~e ligament, with sacrifice at various times between 2 days and 64 weeks after the operation. Proteoglycans were extracted from the articular cartilage of eight areas of the joint, fractionated, and CS was quantified by measuring hexuronate. Aggrecan populations were assessed by composite agarose-polyacrylamide gel electrophoresis. The proportion of the more CS-rich form of aggrecan increased with time after operation in all areas of the joint. This increase correlated with increases in tissue mass and hexuronate, showing that the increase in the larger, CS-rich form of aggrecan contributes to the increase in CS in the tissue. It seems paradoxical that this form of aggrecan accumulates in the tissue despite the fact that increased protease activity has been demonstrated in experimental OA. Key words: Osteoarthritis, Proteoglycan, Aggrecan, Metabolism. Introduction PROTEOGLYCANS (PGs) are characterized by the covalent linkage of glycosaminoglycan(s) to protein. Articular cartilage contains several different classes of PGs, the most complex of which is aggrecan. Keratan sulfate (KS), chondroitin sulfate (CS) and both N-linked and O-linked oligosaccharides are covalently attached to aggrecan in specific domains [1,2]. Aggrecan forms large supramolecular aggregates via a specific ternary interaction with hyaluronan and link protein [3,4]. Aggrecan in articular cartilage is polydisperse with respect to molecular weight, with larger molecules generally containing more CS. Aggregating and nonaggregating fragments of aggrecan are also present in cartilage [5 9]. In addition to its polydispersity, aggrecan is heterogeneous or heterodisperse, i.e. it can be separated into two or more distinctly different populations of molecules [10]. This is perhaps best Submitted 30 July 1993; accepted 7 March Supported by The Medical Research Council of Canada and The Arthritis Society. A preliminary report of this work appeared in J Rheumatol 1987; Supplement 14:107-9, and was presented at the 1992 Annual Meeting of the American College of Rheumatology. Address correspondence to: Dr Mark E. Adams, Department of Medicine, University of Calgary, 3330 Hospital Drive NW, Calgary, AB T2N 4NI, Canada. shown with composite agarose-polyacrylamide gel electrophoresis (CAPAGE) [11-13]. Heineg~rd et al. [14] isolated and characterized two aggregating PG populations from bovine nasal cartilage, each of which migrated as a separate band on CAPAGE: a larger, slower migrating CS-rich PG (which they called PG-LA1 and which we call band 1 of aggrecan) and a smaller, faster migrating PG somewhat richer in KS (which they called PG-LA2 and which we call band 2 of aggrecan). These two populations differ not only in composition but also in their tryptic peptide maps [14] and the content of their G3 domain [6]. Though this could be interpreted to suggest that these might be separate products, in vivo radiolabeling studies in rats [15] and dogs (manuscript in preparation) are most consistent with the hypothesis that the band 2 is derived from band 1, presumably by protease cleavage. It is possible that band 2 is heterogeneous but that this heterogeneity is not well resolved with CAPAGE. It is well known that the composition of articular cartilage changes profoundly with maturation and aging (see reviews [16, 17]). In general, the ratios of CS to KS, chondroitin-4 sulfate to chondroitin-6 sulfate, and glycosaminoglycan to protein decrease. Analysis of the aggrecan from tracheal cartilage of steers of different ages showed that the changes in cartilage composition 155

2 156 Adams: Proteoglycan heterogeneity in experimental OA with maturation and aging correlated with a decrease in the amount of band 1 of aggrecan [18]. The composition of osteoarthritic cartilage, when compared with age-matched normal cartilage, also changes, but generally the changes are opposite to those that occur with age, i.e. the ratios of chondroitin-4 sulfate to chondroitin-6 sulfate, CS to KS, and glycosaminoglycan to protein increase [16, 17, 19, 20]. Furthermore, PG synthesis is increased in osteoarthritis (OA) [20, 21]. Mankin and Lippiello [19] suggested that osteoarthritic cartilage contains PGs characteristic of younger articular cartilage. Because cartilage contains several different extracellular matrix PGs (aggrecan, decorin, biglycan and tlbromodulin, at least), as well as different populations of aggrecan, each with a different composition, a change in the overall composition of cartilage can arise from either a change in the proportion of any of these different species or from a change in the composition of the individual species. To date there has been only preliminary evidence from direct analyses of ex vivo specimens that a change in the proportion of aggrecan populations might contribute to the change in the composition of the osteoarthritic cartilage [22]. Indeed, evidence has also been presented which suggests that the composition of the aggrecan in experimental OA does change [23, 24]. Transection of the anterior (cranial) cruciate ligament (ACL) of the canine knee (stifle) is an excellent model of human OA because the metabolic, biochemical, biomechanical and morphological changes in the articular cartilage resemble those that occur in post-traumatic human OA. With time, full thickness loss of the articular cartilage does develop [25, 26]. However, following ACL transection, PG synthesis is increased both in vivo and in vitro [27], resulting in hypertrophy of the cartilage with an increase in the amount and concentration of PGs in the tissue [28, 29]. The amount of increase differs among the different areas of the joint [30]. The total amount of hexuronate (hexa) in each region of the joint increases progressively with time, essentially reflecting the total mass of cartilage. The concentration of hexa per gram of fresh tissue also increases in the areas with the greatest increase in tissue mass, though the magnitude of the increase in the concentration is not as great as the increase in the tissue mass. In the work presented in this study, experimental OA was induced in canine knees by transection of the anterior cruciate ligament at arthrotomy [31], the PGs were isolated from the articular cartilage of eight areas of the joints and the amount of CS was quantified by measuring hexa. The proportion of the two different aggrecan populations was assessed at CAPAGE. The results of these analyses support the hypothesis that the increase in band 1 of aggrecan contributes to the increase in CS in the cartilage in this experimental model of OA. Materials and methods EXPERIMENTAL MODEL These experiments used 43 male American foxhounds obtained from the Animal Care Center of the University of Oregon Health Sciences Center. All were judged to be skeletally mature by the radiographical appearance of the distal femoral and proximal tibial epiphyses. Experimental OA was induced in 28 dogs by transection of the anterior cruciate ligament in the right knee at lateral arthrotomy [31]; the left knees were nonoperated controls. Seven dogs (sham controls) each had a sham arthrotomy performed on the right knee; the left knees were nonoperated controls. In the sham operation a probe was passed behind the ligament but it was not transected. Eight dogs (nonoperated controls) had no operation on either knee. The number of ACL-transected dogs sacrificed at each time point is listed in Table I. TISSUE HANDLING After sacrifice the joints were opened, photographed using a standard protocol (which included a ruler in the photograph), and samples were taken for histology. The cartilage in the histological samples was not weighed. All the cartilage remaining after sampling for histology was removed with a scalpel as completely as possible from eight separate anatomically-defined regions of the joint with special care taken to exclude subchondral bone or cartilage covering marginal osteophytes, as previously described [20]. The tissue was placed in sealed vials which were weighed immediately. The regions were: A = central medial tibial plateau: B = submeniscal medial tibial plateau; C = central lateral tibial Table I Number of anterior cruciate ligament-transected dogs sacrificed at each time Time (weeks) No. of dogs

3 Osteoarthritis and Cartilage Vol. 2 No plateau; D = submeniscal lateral tibial plateau; G = patellar groove; L = lateral femoral condyle; M = medial femoral condyle; and P = patella. quarter, called ~AiD4', contained no detectable hexa. BIOCHEMICAL ANALYSES GROSS PATHOLOGY GRADING The gross pathological appearance of the articular cartilage was graded for three aspects of the articular cartilage: discoloration, surface integrity, and texture. For each variable a 5-point ordinal scale was used: 0 = normal; 1 = minimal; 2 = mild; 3 = moderate; 4 = severe. The overall grade that is presented in Fig. 1 is the sum of the grades of the three variables. EXTRACTION AND CHARACTERIZATION OF PROTEOGLYCANS The articular cartilage was frozen, cut into 20 pm sections with a cryostat [32], and extracted for 24 h at 4 C in 10 ml of extraction buffer/g of fresh tissue and again in 5 ml of extraction buffer/ g fresh tissue for 3 h. The extraction buffer was 4 M guanidinium chloride, 50 mm sodium acetate, ph 5.8, 10 mm ethylenediaminetetraacetic acid (EDTA), 100mM 6-aminohexanoic acid, 5raM benzamidine HC1, 1 mm phenylmethylsulfonyl fluoride, and 4 mm N-ethylmaleimide. Both extracts were pooled, dialyzed exhaustively against 0.5 M guanidinium chloride, 50 mm sodium acetate, ph 5.8, with the proteinase inhibitors. The extracts were adjusted to a buoyant density of 1.35 g/ml by addition of solid cesium chloride and ultracentrifuged in a Beckman 50 Ti rotor at rpm ( g, ray = 5.91 cm) for 48 h at 10 C. After ultracentrifugation, the tubes were cut with a Beckman tube slicer. The A1 fraction was adjusted to 4 M guanidinium chloride by addition of an equal volume of 7.5M guanidinium chloride containing the protease inhibitors, brought to a buoyant density of 1.35 g/ml by the addition of solid cesium chloride, and ultracentrifuged in a Beckman 50 Ti rotor at rpm for 48 h at 10 C. After ultracentrifugation the tubes were cut with a Beckman tube slicer. From studies of other normal canine stifle joints using this methodology, the extract contained an average of 82% of the total tissue hexa and the residue contained 18~/o; in the associative gradient, the bottom quarter, called ~AI', contained approximately 80% and the top three- quarters, called 'A2-4', contained approximately 2% of the total tissue hexa; in the dissociative gradient, the bottom quarter, called ~A1DI', contained an average of 78%, the middle half, called 'AID2-3', contained 2~/o, and the top HexA was analyzed by the borosulfuric acid carbazole method of Bitter and Muir [33] as automated by Heineg&rd [34]. COMPOSITE AGAROSE-POLYACRYLAMIDE GEL ELECTROPHORESIS (CAPAGE) The procedure of McDevitt and Muir [11] was modified to run in vertical slab gels [35] using a Bio-Rad Protean I electrophoresis cell with frosted glass plates. Gel thickness was 3 mm with 10 lanes per gel. Approximately 25 nmol hexa was applied per lane. A sample of a pooled canine articular cartilage A1D1 fraction prepared separately was used as a standard for mobility and resolution. Electrophoresis was run for 3.5 h at a constant current at 50 ma at 4 C. After electrophoresis, the gels were stained in 0.02% (w/v) toluidine blue in 100 mm acetic acid, destained in five changes of 0.5 M acetic acid, applied to agarose GelBond (FMC Marine Products), and air dried. The dried gels were scanned at 595 nm with a Helena Auto Scanner recording densitometer with a built-in integrator. Although bands migrating faster than band 2 do appear, especially in the OA samples, and some of these may be aggrecan fragments, for the quantitative analyses presented in this study the % band 1 was defined at (100 x band 1)/(band 1 + band 2). As an indication of the precision of the method, the proportion of band 1 in the mobility standard was 62.60% +2.74% (mean_s.e.m. for 53 runs). ASSESSMENT OF THE ABILITY OF THE PROTEOGLYCANS TO AGGREGATE The proportion of the PGs in the AID1 fraction that could aggregate with hyaluronan was tested by the procedure of Heineg~rd et al [36] by mixing the samples with 20% hyaluronan (Healon, Pharmacia) prior to running them in CAPAGE gels. The method yields results similar to those using Sepharose C1-2B chromatography with 2% added hyaluronan [36] (Adams and Grant, unpublished data). STATISTICAL ANALYSES Statistical analyses were performed using Systat for Windows, version 5.02 (Evanston, IL). Analysis of variance (ANOVA) was used to define whether

4 158 Adams: Proteoglycan heterogeneity in experimental OA or not laterality was significant. A combination of cluster analysis using the join command with graphical tree output, means testing and ANOVA was used to subgroup the time points into phases of the natural history of this experimental posttraumatic OA. Post-hoc testing of the ANOVA for multiple means comparisons was performed using the Tukey HSD method. For the Pearson correlation coefficients, Bonferroni's correction was applied. The Pearson correlation coefficients were calculated pairwise. O 12 iel NONLINEAR REGRESSION ANALYSIS AND CURVE FITTING The curve superimposed on the data points in Fig. 2 was derived using the nonlinear regression module of Systat for Windows, version 5.02, with the model equation of the form: Tissue mass = C+ A/(1 + e -b(~-time)) where Time is the independent variable, Tissue mass is the dependent variable, and A, b, C and # are parameters derived by the nonlinear regression. This logistic equation models a proliferation of the tissue mass in the form of a binomial distribution during the growth phase, with the maximum growth centered at time -- #. For the graph in Fig. 2, which is presented for illustrative purposes only, the parameters A, b, C and # were further adjusted arbitrarily to increase the aesthetics of the fit. Results GROSS PATHOLOGY OF THE ARTICULAR CARTILAGE Gross pathological changes occurred in the articular cartilage rapidly and progressively (Fig. 1). It is notable that these changes preceded changes in the increase in tissue mass (Fig. 2). No full thickness losses occurred, even in the 64-week samples. COMPARISON OF RIGHT AND LEFT SIDES AND OF SHAM-OPERATED DOGS Analysis of variance showed that there was no difference between any of the variables (tissue mass, hexuronate and band 1/band 2 ratio) between the right and between the left sides of the nonoperated dogs, i.e. there was no innate laterality. Therefore, in order to simplify the data presentation and to minimize individual vari: Time (weeks) Fro. 1. Gross pathology grading of the articular cartilage. The grading scale used is described in Materials and Methods. ~2 3 1i I I I t I ] I Time (weeks) FIG. 2. Tissue mass vs time (weeks) for the articular cartilage of the femoral groove. This graph shows the relative increase in tissue mass (the right, operated side mass divided by. the left, nonoperated side mass) vs time (weeks) after transection of the anterior cruciate ligament. It is presented to illustrate the degree, the phases, and the kinetics of the cartilage hypertrophy in this model. It is notable that the cartilage mass increases in all the areas of the femoral condyles, while those of the tibia change little, if at all (see Table II). The error bars show the S.E.M. The number of dogs sacrificed at each time is listed in Table I. The curve superimposed on the data points results from nonlinear regression modeling and curve fitting (as explained in Materials and Methods), and is modeled upon the assumption of a binomial growth of the cartilage mass with the maximum growth at 14 weeks after anterior cruciate ligament transection.

5 Osteoarthritis and Cartilage Vol. 2 No ability, subsequent analyses were performed on the ratios of the right side variable to the left side variable. Furthermore, when compared pairwise, there was no difference between the right and the left sides of the sham-operated dogs. (Because they were older, the 32- and 64-week sham-operated dogs did have a lower proportion of band 1 of aggrecan than did the nonoperated dogs or the sham-operated dogs from the earlier time points.) Therefore, no further analyses were done for this presentation using the sham-operated animals. STAGING OF EARLY EXPERIMENTAL OA The number of ACL-transected dogs which were sacrificed at each time point is listed in Table I. On visual inspection of the graphs of tissue mass vs time by area (previously presented by Adams [30]), several distinct phases were apparent (see Fig. 2 for an example). Clearly the nonoperated animals are an exclusive group (control) and, of course, should be classed as such irrespective of the results of any statistical analyses. It was also clear that there was ~plateau' in the response at times equal to or greater than 32 weeks. Finally, there appeared to be a lag in the increase in tissue mass after ligament transection. Thus, the natural history of the experimental post-traumatic OA can be divided into at least five logical phases: control (pre-injury) (time = 0), early, growth, plateau, and the decompensation with loss of cartilage to subchondral bone, as shown by Brandt et al. [37]. (As previously mentioned, none of the animals in this experimental set was in the last phase.) On a few of the graphs, it appeared there might be a slight loss in the tissue mass immediately after ligament transection. Statistical analyses were used to define these phases more precisely. By using a combination of cluster analysis and ANOVA, it was seen that the times equal to or greater than 24 weeks belonged in the plateau phase of the increase in tissue mass vs time, while those at 12 and 16 weeks were clearly in the intermediate (growth) phase. No sorting of the remaining data was apparent with cluster analysis. ANOVA was used to assess whether or not the times from 0.3 weeks (2 days) to 8 weeks differed among each other. The times differed significantly (P < 0.05) only in the lateral fermoral condyle (area L), so they were not subgrouped further. Thus the times 0, weeks (15 dogs), weeks (4 dogs) and weeks (9 dogs) were grouped, respectively, and called control, early phase, growth phase and plateau phase. COMPOSITE ARAGOSE-POLYACRYLAMIDE GEL ELECTROPHORESIS CAPAGE gels showing the electrophoresis results from the left and right knees of a 64-week post-operative animal are shown in Fig 3. The amount of band 1 is clearly greater in the samples from the operated (right) side. The greater increase in areas of the femoral condyles [areas G, L, M and P in Fig. 3] is obvious when compared with those of the tibial plateau [areas A, B, C and D in Fig. 3]. Densitometric analysis corroborated these visual impressions. The results from the analyses of all the cartilage AID1 fractions from all the areas are shown in Table II. The extent of the change varied among all the regions of the STD LA RA LB RB LC RC LD RD STD LG RG LL RL LM RM LP RP FIG. 3. CAPAGE of 64-week femoral condylar A1D1 samples showing increase in the proportion of the slower migrating band in the operated samples. STD is the standard (pooled sample of canine articular cartilage A1D1); L is left; R is right; A = central medial tibial plateau; B = submeniscal medial tibial plateau; C = central lateral tibial plateau; D--submeniscal lateral tibial plateau; G = patellar groove; L = lateral femoral condyle; M = medial femoral condyle; P = patella.

6 160 Adams: Proteoglycan heterogeneity in experimental OA Table II Means and standard deviations of relative increases in the tissue mass, hexa in the AID1 fraction, and /o CS-rich form of aggrecan (band 1). All data are presented as ratios of the right (operated) side divided by the left (nonoperated) side Tissue mass HexA in AIDA Percent band 1 Area Phase N Mean S.D. Mean S.D. Mean S.D A B C D G L M P control early growth plateau * * control early : growth plateau * * control early growth plateau t NS control early growth plateau ~ * control early growth plateau t ~ control early growth plateau t ~ control early growth plateau ~ t control early growth plateau t ~ *Plateau differs significantly from control and early; tplateau differs significantly from control, early or growth; Splateau differs signficantly from early; NS: plateau does not differ significantly from control, early or growth phase. A = central medial tibial plateau; B = submeniscal lateral tibial plateau; C = central lateral tibial plateau; D = submeniscal lateral tibial plateau; G = patellar groove; L = lateral femoral condyle; M = medial femoral condyle; P = patella. joint in the same way as the tissue masses and hexa. It can be seen that, except for the change in the hexa in the A1D1 fraction of the articulating portion of the lateral tibial plateau (area C), all the changes in the plateau phase values differed significantly from some of the other values. In some cases the changes in the growth phase values were of sufficient magnitude that the plateau phase values did not differ from them. Furthermore, the changes were proven to be most profound and significant (corrected for multiple comparisons) in the areas of the femoral condyles, where the hypertrophy is greater [29, 30]. CORRELATIONS BETWEEN TISSUE MASS, HEXA IN THE A1D1 FRACTION, AND BAND 1 OF AGGRECAN Table III shows the Pearson correlation coefficients and the probabilities (with the Bonferroni corrections for multiple comparisons) between the

7 Osteoarthritis and Cartilage Vol. 2 No Table III Pairwise Pearson correlation coefficients between tissue mass, hexa in the AID1 fraction (HxAD1) and the percentage of the more CS-rich form of aggrecan. R is the Pearson correlation coefficient; P is the probability that the differences were due to chance. Data from all the analyses were used in generating these pairwise correlation coefficients Tissue mass-hxad1 Tissue mass-~/o band 1 HxADl-%Band 1 Area R P R P R P A B C D G L M P relative increases in tissue mass, the amount of hexa in the A1D1 fraction and percentage of band 1. The correlation coefficients were generally lower for the areas of the tibial plateau where there was little, if any, cartilage hypertrophy. However, for all the areas of the femoral condyle and for the patella the correlation coefficients were high and the probabilities that these were due to chance were very low. Thus, where there is substantial cartilage hypertrophy, this hypertrophy correlates with the increase in the hexa, as previously shown [29, 30], and with the increase in the percentage of band 1. Furthermore, the very good correlations between the concentration of the hexa and the percentage of band 1 support the hypothesis that the increase in the proportion of band 1 of aggrecan contributes to the increase of CS in the tissue. ASSESSMENT OF THE POTENTIAL OF THE PROTEOGLYCANS TO AGGREGATE The proportion of aggrecan that was capable of aggregating with hyaluronan showed neither a significant change between the right operated joints nor the right nonoperated controls. Likewise there was no change with time between the right operated joints and the contralateral nonoperated joints from the same animals (data not shown). Thus, band 1 represents the more CS-rich form of aggrecan, rather than a fragment, and it increased in proportion progressively after the operation. Discussion This study demonstrates that the increased proportion of the more CS-rich form of aggrecan in the articular cartilage in vivo correlates with the increased hexa and tissue mass in this model of experimental OA, confirming our preliminary results [22]. The changes vary among the areas and with time after induction of the experimental OA. It is notable that in the study by Carney et al. [24], no variability was noted either among the areas or with time after the operation. In their experiments they cultured explants for 48 h after excision and then extracted the proteoglycans. We have performed in vivo radiolabeling experiments (manuscript in preparation) which showed that the proportion of radioactive label in the proteoglycans in band 1 and band 2 fully achieved the resident (nonradioactive) level at 1 week, but had not by 24 h. The pioneering work by Collins and McElligott [21] and by Mankin and co-workers [20] suggested an increase in the rate of synthesis of cartilage PGs in human OA. Mankin, Lipiello and coworkers [19], observing that the ratio of galactosamine to glucosamine was increased, suggested that the cartilage in OA was synthesizing PGs characteristic of younger cartilage. In 197i, McDevitt and Muir [11] demonstrated heterogeneity of aggrecan in CAPAGE. Inerot and Heineg~rd [18] used CAPAGE to demonstrate that the" increase in the g!ucosamine content of cartilage that occurs with maturation of bovine tracheal cartilage was paralleled by an increase in the ~KS-rich', faster migrating PG (band 2). We have observed that such an increase in band 2 also occurs with maturation and aging in canine articular cartilage (Adams, Grant and Ho, unpublished data). The molecular basis for the separation of band 1 and band 2 is not well understood [6], but would be worth knowing because it changes with disease and aging. Furthermore, it is entirely possible that band 2 is itself composed of a

8 162 Adams: Proteoglycan heterogeneity in experimental OA heterogeneous population of aggrecan molecules or fragments, but that these are not well resolved with CAPAGE. That a reversal of this increase in the proportion of band 1 could contribute to the increase in CS in the articular cartilage in OA is an attractive hypothesis. However, Cox et al. [23], using the canine anterior cruciate ligament transection model of OA, demonstrated that the CS-rich region (obtained by hydroxylamine treatment of the articular cartilage aggrecan) contained longer CS chains and a more highly glycosylated protein core than controls. The finding of new CS epitopes on aggrecan in experimental OA [24] further supports the hypothesis that the glycosylation pattern has changed. In spite of these data suggesting increased and altered glycosylation of aggrecan in OA, the data presented in this study strongly support the hypothesis that an increase in the proportion of band 1 of aggrecan also contributes significantly to the increase in CS concentration in the cartilage. That at least some of the in vivo change in the composition of articular cartilage in OA arises from a change in the proportion of the two populations of aggrecan does not preclude a concomitant change in the glycosylation pattern of the individual aggrecan populations. Considering all the evidence, both changes probably have occurred, but defining this will require comparison of the composition of band 1 of aggrecan from osteoarthritic articular cartilage with that from normal cartilage. It is possible that the exaggerated hypertrophic phase in this study results from more effective retention of newly synthesized aggrecan in younger tissue. Mort and co-workers [38] showed that human articular cartilage link protein becomes progressively fragmented with age. This also occurs with canine articular cartilage link protein (Adams, Mort and Roughley, unpublished data). Melching and Roughley [39] showed that link protein facilitates the interaction of newly synthesized PG monomer with hyaluronan. Plaas and Sandy [40] found that the proportion of link protein synthesized relative to PG decreased with age in rabbits. Such changes could lead to increased retention of the newly synthesized aggrecan in younger animals with OA. Other agerelated changes, such as a decrease in IGF-1 or other growth factors, could result in a decreased synthesis of aggrecan in response to trauma with age. These changes are of potential importance in the pathogenesis of OA. An increase in the thickness of the articular cartilage has been noted in a previous study by Vignon et al. [28] using this model of OA. They also noted increased fixed charge density, suggesting an increase in the PGs in the matrix. We have presented data that corroborates and expands on this information [30]. Early human OA can also apparently have a hypertrophic phase [41], but most studies of human OA have used surgical specimens which would be 'end stage' and more likely from older individuals. Such a hypertrophic response with synthesis of the larger more CS-rich form of aggrecan might be considered a form of remodeling or repair response that could help preserve cartilage function. The experience of Jackson and Holtby [42t that older individuals develop cartilage damage sooner after spontaneous rupture of the anterior cruciate ligament is consistent with this hypothesis. This emphasizes the need to elucidate the cause for such change with age in the response of cartilage to trauma. However, one cannot be certain that this accumulation of aggrecan is beneficial for the long-term health of the tissue. For example, in acromegaly the initial increase in the synthesis of PGs and eventual accumulation of excess PGs in the tissue is followed by cartilage degeneration. However, it seems reasonable that in the context of OA such a putative 'repair' might help compensate for the increased PG destruction. Further support of this hypothesis is given by the acceleration of cartilage damage by treatment with high doses of salicylates [43], which suppress aggrecan synthesis [44]. It is noteworthy that the increase in the amount of cartilage is lowest in the areas of the tibial plateau, the sites where lesions form soonest. Clearly, further studies are needed to delineate the nature of the PGs in these tissues and the factors controlling their synthesis, deposition, metabolism and degradation, because clarification of these mechanisms may help define sites for possible therapeutic intervention. Finally, it is important to define whether or not similar differences in aggrecan heterogeneity occur in human OA. Acknowledgments The author gratefully acknowledges the capable technical assistance of Amy Ho and Michael Grant and helpful and stimulating discussions with Dick HeinegArd, Cahir McDevitt and Eric Vignon. References 1. Doege K, Sasaki M, Yamada Y. Rat and human cartilage proteoglycan (aggrecan) gene structure. Biochem Soc Trans 1990;18:200-2.

9 Osteoarthritis and Cartilage Vol. 2 No Antonsson P, Heineg~rd D, Oldberg A. The keratan sulfate-enriched region of bovine cartilage proteoglycan consists of a consecutively repeated hexapeptide motif. J Biol Chem 1980;264: : Heineg~rd D, Oldberg A. Structure and biology of cartilage and bone matrix noncollagenous macromolecules. FASEB J 1989;3: Carney SL, Muir H. The structure and function of cartilage proteoglycans. Physiol Rev 1988;68: Flannery CR, Lark MW, Sandy JD. Identification of a stromelysin cleavage site within the interglobular domain of human aggrecan. Evidence for proteolysis at this site in vivo in human articular cartilage. J Biol Chem 1992;267: Flannery C, Stanescu V, MSrgelin M, Boynton R, Gordy J, Sandy J. Variability in the G3 domain content of bovine aggrecan from cartilage extracts and chondrocyte culture. Arch Biochem Biophys 1992;297: Sandy JD, Flannery CR, Neame PJ, Lohmander LS. The structure of aggrecan fragments in human synovial fluid. Evidence for the involvement in osteoarthritis of a novel proteinase which cleaves the Glu 373-Ala 374 bond of the integlobular domain. J Clin Invest 1992;89: Loulakis P, Shrikhande A, Davis G, Maniglia CA. N-terminal sequence of proteoglycan fragments isolated from medium of interleukin-l-treated articular-cartilage cultures. Putative site(s) of enzymic cleavage. Biochem J 1992;284: Inerot S, Heineg~rd D, Olsson SE, Telhag H, Audell L. Proteoglycan alterations during developing experimental osteoarthritis in a novel hip joint model. J Orthop Res 1991;9: Gibbons RA. Physicochemical methods for the determination of the purity, molecular size and shape of glycoproteins. In: Gottshalk A, Ed. Glycoproteins, Vol 5A. Amsterdam: Elsevier 1972: McDevitt CA, Muir H. Gel electrophoresis ofproteoglycans and glcosaminoglycans on large-pore composite polyacrylamide-agarose gels. Anal Biochem 1971;44: R0ughley PJ, Mason RM. The electrophoretic heterogeneity of bovine nasal cartilage proteoglycans. Biochem J 1976;157: Stanescu V, Maroteaux P, Sobczak E. Proteoglycan populations of baboon (Papio papio) articular cartilage. Biochem J 1977;163: Heineg~rd D, Wieslander J, Sheehan J, Paulsson M, Sommarin Y. Separation and characterization of two populations of aggregating proteoglycans from cartilage. Biochem J 1985;225: Heineg~rd D, Franzen A, Hedbom E, Sommarin Y. Common structures of the core proteins of interstitial proteoglycans. In: Evered D, Whelan J, Eds. Functions of the proteoglycans. CIBA Foundation Symposium 124. Chichester: John Wiley 1986: McDevitt CA. The proteoglycans of cartilage and the intervertebral disc in ageing and osteoarthritis. In: Glynn LE, Ed. Handbook of inflammation, Vol 3: Tissue repair and regeneration. Amsterdam: Elsevier/North Holland 1981: Inerot S, Heineg~rd D. Articular cartilage proteoglycans in aging and osteoarthritis. In: Horowitz MI, Ed. The Glycoconjugates. 4th Ed. New York: Academic Press 1982: Inerot S, Heineghrd D. Bovine tracheal cartilage proteoglycans. Variations in structure and composition with age. Coll Relat Res 1983;3: Mankin HJ, Lippiello L. The glycosaminoglycans of normal and arthritic cartilage. J Clin Invest 1971;50: Mankin HJ, Dorfman H, Lippiello L, Zarins A. Biochemical and metabolic abnormalities in articular cartilage from osteoarthritic human hips. J Bone Joint Surg [Am] 1971;53: Collins DH, McElligott TF. Sulphate (SOt) uptake by chondrocytes in relation to histological changes in osteoarthritic human articular cartilage. Ann Rheum Dis 1960;19: Adams ME, Grant MD, Ho YA. Cartilage proteoglycan changes in experimental canine osteoarthritis. J Rheumatol 1987;14:(Suppl): Cox MJ, McDevitt CA, Arnoczky SP, Warren RF. Changes in the chondroitin sulfate-rich region of articular cartilage proteoglycans in experimental osteoarthritis. Biochim Biophys Acta 1985;840: Carney SL, Billingham MEJ, Caterson Bet al. Changes in proteoglycan turnover in experimental canine osteoarthritic cartilage. Matrix 1992;12: Brandt K, Braunstein E, Visco D et al. Anterior cruciate ligament transection: a bona fide model of canine osteoarthritis, not merely of cartilage injury and repair. Arthritis Rheum 1990;33:$31 (Abstract). 26. Brandt KD, Myers SL, Burr D, Albrecht M. Osteoarthritic changes in canine articular cartilage, subchondral bone, and synovium fifty-four months after transection of the anterior cruciate ligament. Arthritis Rheum 1991;34: Sandy JD, Adams ME, Billingham ME J, Plaas A, Muir H. In vivo and in vitro stimulation of chondrocyte biosynthetic activity in early experimental osteoarthritis. Arthritis Rheum 1984;27: Vignon E, Arlot M, Hartmann DJ, Moyen B, Ville G. Hypertrophic repair of articular cartilage in experimental osteoarthrosis. Ann Rheum Dis 1983;42: Adams ME, Brandt KD. Hypertrophic repair of canine articular cartilage in osteoarthritis after anterior cruciate ligament transection. J Rheumatol 1991;18: Adams ME. Cartilage hypertrophy following canine anterior cruciate ligament transection differs ~:among different areas of the joint. J Rheumatol 1989;16: Adams ME, Pelletier J-P. The canine anterior cruciate ligament transection model of osteoarthritis. In: Greenwald RA, Diamond HS, Eds. Animal models for the rheumatic diseases. Boca Raton: CRC Press 1988: Bayliss MT, Venn M, Maroudas A, Ali SY. Structure of proteoglycans f~om different layers of human articular cartilage. Biochem J 1983:209: Bitter T, Muir HM. A modified uronic acid carbazole reaction. Anal Biochem 1962;4: HeinegArd D. Automated procedures for the deter-