Quantification and immunolocalisation of porcine articular and growth plate cartilage collagens

Skeletal growth is a highly complex and poorly understood process, which begins with the ossification of foetal cartilaginous tissue and ends with physeal closure at maturity. The structural organization of long bones during skeletal growth exhibits a particular pattern. The proximal and distal ends of the bones (the epiphyses) are covered with articular cartilage, which is initially responsible for epiphyseal growth, while sandwiched between the epiphysis and the metaphysis is the growth plate cartilage, which gives rise to longitudinal bone growth (Brighton, 1984; Farnum and Wilsman, 1989; Hunziker and Schenk, 1989).

As with other hyaline cartilages, growth plate cartilage matrix consists principally of water, collagen and proteoglycans (Muir et al., 1977). The genetically distinct collagen types, being major components, clearly play an important role in the structure and function of the tissue and some have been implicated in the processes of calcification and ossification. The predominant collagen type of the non-calcified matrix, as with other hyaline cartilages, is type II collagen, which is arranged in a fibrillar network (Poole et al., 1989). Other collagens thought to exist in the growth plate are types V, VI, IX, X and XI (Mayne, 1986; Gibson and Flint, 1985) and although extensive studies have been carried out on articular cartilage, little is known about their distribution, quantification or function in growth plate. Studies on articular cartilage have revealed the possible functions of some of the minor collagens. Type IX is found attached to the surface of type II fibrils (Eyre et al., 1987a; Vaughan et al., 1988) and strong evidence exists that it has a role in the regulation of fibril diameter (Wotton et al., 1988). Type X collagen is found exclusively in the hypertrophic region of growth plate cartilage (Gibson and Flint, 1985) and is only rarely detected in the calcifying region of articular cartilage (Gannon et al., 1988). It is found associated with type II collagen (Chen et al., 1990) and its location suggests an important role in the processes of calcification or chondrocyte maturation.

Collagen undergoes numerous post-translational modifications, one of which is the formation of intermolecular cross-links, which confers strength and stability to these cartilaginous tissues (Eyre et al., 1984). Slow protein turnover, as is found in articular cartilage, allows further cross-link maturation, producing a highly stable tissue. The dynamics of growth plate cartilage are, however, very different in that there is a requirement for rapid protein turnover whilst retaining tissue stability. Resorption of collagen may be influenced by the extent of cross-linking but the nature and location of cross-links in growth plate cartilage have not been studied. The turnover of growth plate cartilage in an animal that is growing too quickly may not allow sufficient cross-link maturation giving rise to a weakened, unstable cartilage.

Growth plate defects have been shown to be significant in leg weakness, an economically important disease of pigs. This disease is now generally accepted to be caused by osteochondrosis in both articular and growth plate cartilage (Hill, 1990a,b). Defects are varied and include lack of calcification, fissuring, thickening, areas of necrosis and chipping of the articular surface. Despite extensive research, the aetiology of the disease is still unknown.

The present study is a detailed biochemical investigation of both articular and growth plate cartilage collagens of clinically normal commercial pigs prior to further studies of cartilage from animals with symptoms of osteochondrosis.

Long bones were removed from Yorkshire/Landrace hybrid pigs of average weight 60 kg within 24 h of slaughter and dissected clean of adherent muscle and tendon. Articular cartilage was removed from the distal epiphyses of the femur and humerus taking care to avoid ligament, periosteum or underlying bone. The bones were then cut longitudinally into approximately 0.5 cm slices using a band saw. Both proximal and distal epiphyses of the femur and humerus and the proximal epiphyses of the tibia and ulna were used to provide growth plate cartilage. Bone slices could be cracked along the line of the growth plate and the cartilage removed.

Articular and growth plate collagens were prepared by treatment of cartilage samples with guanidine hydrochloride followed by pepsin extraction at acid pH, as described previously (Rhodes and Miller, 1978).Collagens were partially purified by sequential salt precipitation at acid pH. All samples were dialysed extensively against 0.05 M acetic acid before lyophilisation (Fig. 1).

Collagen types were separated by SDS-polyacrylamide gel electrophoresis (Laemmli, 1970) and either stained with Coomassie Brilliant Blue R or subjected to western blotting (Towbin et al., 1979). Blots were probed using monoclonal or polyclonal antibodies, mostly prepared in this laboratory and which were raised against pepsin-treated molecules purified from cartilage, tendon or placenta. Antibodies were screened for reactivity against porcine collagens prior to this study, and selected for specificity and strength of signal by enzyme-linked immunosorbent assay (ELISA) using native collagens, or by western blotting using SDS denatured collagens. Antibodies used were goat anti-bovine type I, rabbit anti-bovine type II, goat anti-human type III, goat anti-porcine type V, mouse anti-bovine type VI, rabbit anti-porcine type IX, mouse anti-porcine type X and rabbit anti-porcine type XI. A commercial polyclonal antibody raised against human type V collagen (Southern Biotechnology Associates Ltd.) was also used. All antibodies were polyclonal with the exception of a monoclonal against type VI collagen. Non-immune sera from rabbit, goat and mouse were used as negative controls for the polyclonal sera and an unrelated mouse monoclonal antibody as a negative control for the type VI antibody. Polyclonal antibodies raised against bovine and chicken type X were generous gifts from Dr Anne Marriott and Dr Alvin Kwan (University of Manchester), respectively, and were used as positive controls for our previously uncharacterized type X antibodies. An alkaline phosphatase-conjugated second antibody detected with 5-bromo-4-chloro-3-indolyl phosphate, p-toluidine salt (BCIP) and nitro blue tetrazolium (NBT) as substrates was used with all antibodies on western blots. Where weak or negative signals were produced, as in the case of antibodies against collagen types III, V and VI, an alternative, more sensitive method employing a peroxidase-conjugated second antibody detected with a chemiluminescent substrate (ECL kit, Amersham) was used.

Lyophilised whole cartilage or salt-fractionated samples were hydrolysed in 6 M HCl for 20 h at 115°C and the total collagenIcontent estimated by automated hydroxyproline analysis (Woessner, 1976). Fractions derived from guanidine hydrochloride were first treated with hyaluronidase and pepsin in order to release collagens aggregated by high molecular mass proteoglycan and allow their resolution by SDS-PAGE. Ratios of whole alpha chains were estimated from SDS-polyacrylamide gels stained with Coomassie Brilliant Blue R and analyzed by an LKB Ultroscan XL scanning laser densitometer. Staining of types I and II collagen was found to be linear in a range of at least 1 µg-15 µg per track and previous work in this laboratory has shown that peptides of pure types I and II collagen bind Coomassie blue to the same extent (Barr, 1991). All tracks from each collagen preparation were scanned and bands that had been identified as specific collagen types by western blotting were integrated. Peak areas for each collagen band were expressed as a percentage of the total collagen peak areas in that sample and then expressed as a weight of collagen in the sample using the previously calculated value from the hydroxyproline assay. A total weight for each collagen type in the whole preparation could be calculated and expressed as a percentage of the whole collagen content of the tissue, although a number of assumptions have been made in order to avoid errors in the calculation. Aggregates of high molecular mass cross-linked collagen molecules (beta and gamma chains) were solubilised and were evident on the gels but their molecular mass made separation and subsequent quantification impossible and thus they were not included in the calculation. However, it is likely that they exist in proportion with their alpha chains and their exclusion, therefore, should not alter the relative amount of each collagen type. Co-migration of the alpha 1(I) chain and the alpha 3(XI) chain with the alpha 1(II) chain is also taken into account as these values can be estimated from the values of the visible, and therefore quantifiable, alpha 2(I) chain and the alpha 1(XI) and alpha 2(XI) chains.

Analysis of collagen cross-links was carried out as described previously (Sims and Bailey, 1992). Lyophilised whole cartilage was reduced using potassium borohydride prior to hydrolysis as above. Hydrolysed samples were subjected to column chromatography on a cellulose CF1 column. Analysis for the intermediary, reducible collagen cross-links hydroxylysyl-ketonorleucine (HLKNL) and dehydro-hydroxylysinonorleucine (deH-HLNL), and the mature non-reducible collagen cross-links hydroxylysyl-pyridinoline (HYL-PYR) and lysyl-pyridinoline (LYS-PYR) was carried out on a LKB 4400 amino acid analyser.

Unfixed, full-thickness cartilage blocks were snap-frozen in isopentane cooled in liquid nitrogen and 5-10 µm cryostat sections prepared. Sections were treated with testicular hyaluronidase (1 mg/ml) in phospate buffered saline (PBS), pH 7.4, for 16 h at 20°C, washed extensively with PBS pH 7.4 and primary, collagen type-specific antibodies applied for 4 h. Antibodies were as described earlier. Non-immune sera of the appropriate species or unrelated monoclonal supernatants were used as negative controls. After further washes a secondary, species-specific antibody conjugated with fluoroscein isothiocyanate (FITC) was applied for 1 h. Sections were washed exhaustively with PBS, pH 7.4, before mounting in Citifluor AF1 and viewing by epifluorescence on a Leitz Dialux microscope.

Samples taken from the pepsin-insoluble residue, guanidine extracts and salt-precipitated fractions appeared to be completely soluble in the SDS sample buffer used for electrophoresis. Collagen types I, II, IX, X and XI were present in sufficient amounts for the alpha chains to be visible by staining with Coomassie blue (Fig. 2). The identity of the collagen bands was confirmed by western blotting using collagen type-specific antibodies (Fig. 3A-F). Type VI was shown to be present at very low levels, detectable only by western blotting, and appeared in the guanidine extracts and also in the pepsin-insoluble and 0.7 M salt fractions of both cartilages, although a considerably stronger signal was produced by samples from articular cartilage (Fig. 3C). Type V collagen could not be detected even using a chemiluminescent substrate, which greatly enhances the sensitivity of the immunoblotting technique. All the collagens present in growth plate cartilage were also present in articular cartilage with the exception of type X, which could not be detected in articular cartilage even by western blotting (Fig. 3E). Type X from growth plate is completely extracted by the sequential guanidine hydrochloride and pepsin treatment, with the whole chain M_r 59,000 band appearing in the precipitated guanidine extract (Fig. 3E, Gup) and the pepsinised M_r 47,000 band precipitating at 2 M-3 M salt. Analysis of reduced samples indicated that, unlike bovine type X (Marriott et al., 1991), and canine type X (Gannon et al., 1991), there is no evidence that porcine type X contains intermolecular disulphide bonds (data not shown).

Analysis of hydroxyproline content of lyophilised growth plate cartilage produced a figure of approximately 32.4% ± 3.1% collagen (dry weight). Articular cartilage treated by the same method comprised approximately 52.4% ± 2.7% collagen.

Analysis of Coomassie-stained gels by scanning laser densitometry in combination with total collagen quantification by hydroxyproline assay allowed the proportions of the major collagen types in each fraction, and thus in each cartilage, to be calculated (Table 1). As would be expected for hyaline cartilages, type II collagen was the major collagen type present in both tissues although the percentages are lower than those normally quoted for other mammalian cartilages. Both porcine cartilages contained a significant proportion (>10%) of type I collagen with growth plate containing almost 20%. Values for type IX collagen were also proportionately lower than expected and growth plate contained three times less than the amount found in articular cartilage. The percentage of type XI collagen present is similar to published values for other species, but again, differences were found between growth plate and articular cartilage, with a significantly greater proportion present in the growth plate. However, polyacrylamide gels stained with Coomassie blue failed to show the presence of any type XI collagen in the 1.2 M salt fraction of growth plate (Fig. 2, track 1.2G) although a small quantity is present, detectable only by western blotting (Fig. 3F). Despite its relative abundance in growth plate compared to articular cartilage, this result suggested that growth plate type XI collagen is more difficult to extract by pepsin treatment. The percentages of collagen types extracted compared with those remaining in the pepsin-insoluble tissue were therefore calculated (Table 2).

The major collagen types (I, II, and XI) were shown to be more easily extracted from articular cartilage than from growth plate. Types IX and X were present in such low amounts in the pepsin-insoluble pellet that quantification was not possible and thus it was assumed that 100% of each collagen was extracted with pepsin.

Variable extraction of collagen by pepsin could reflect differences in the degree of collagen cross-linking present in these tissues. Whole articular and growth plate cartilage were therefore analyzed for mature (HYL-PYR and LYS-PYR) and intermediary (HLKNL and deH-HLNL) crosslinks (Table 3). Articular cartilage contained a slightly higher overall level of mature collagen cross-links than growth plate cartilage although the latter contained significantly higher levels of intermediary cross-links.

Specific polyclonal antibodies were used to localise the collagen types identified in both tissues (Figs 4 and 5). Articular cartilage sections showed an area of strong staining for type I collagen at the articular surface, which was composed of both a matrix and a predominantly pericellular signal. (Fig. 4A). Pericellular staining was also predominant in the calcifying region of articular cartilage (Fig. 4B). Staining for type I collagen was seen in the resting zone of growth plate as a halo surrounding individual chondrocytes or groups of chondrons (Fig. 4C) and as a more diffuse area in the matrix of the hypertrophic region (Fig. 4D). A strong signal was seen where cartilage had ossified at the metaphysis and epiphysis indicating that type I had been laid down, presumably by osteoblasts.

Antibodies against type II collagen produced an overall staining of the matrix in both types of cartilage similar to that previously reported. In the hypertrophic zone of growth plate cartilage and the calcifying region of articular cartilage, the signal became more pericellular (data not shown).

Type VI collagen was mainly localised at the articular surface giving a rather discontinuous matrix stain with some pericellular signal (Fig. 4E) but was not detected in the calcifying region. Type VI collagen was not detected in either the resting zone of growth plate cartilage (Fig. 4F) or the hypertrophic region. Type IX collagen had a predominantly pericellular location in both cartilages, weak in the upper zones (not shown) and becoming progressively stronger towards the calcifying region of articular cartilage (Fig. 5A) and the hypertrophic region of the growth plate (Fig. 5B). Type X was absent from articular cartilage (Fig. 5C) and was only found in the hypertrophic region of the growth plate (Fig. 5D). Type XI collagen showed a weak overall matrix stain in both tissues, similar to that of type II, with a slightly more intense signal around the chondrocytes of the lower regions (Fig. 5E and F).

Cartilage collagens are difficult to quantify accurately owing to the extensive intermolecular cross-linking that rapidly occurs in the extracellular matrix rendering them resistant to extraction. Also, the similar physicochemical properties of many of the collagen types make this family of proteins difficult to separate. SDS-polyacrylamide gel electrophoresis provides one of the best methods for rapid and sensitive separation of most collagen types into their constitutive chains, which can then be quantified and related to the total present in the tissue. Quantification is often carried out on cyanogen bromide (CNBr) digests, a method which has the advantage of totally solubilising most cartilage collagens. However, CNBr digests do not always produce well-defined gel bands and patterns are complex, especially when derived from mixtures of collagen types. Collagens from immature animals are relatively soluble and it was established that the material remaining after treatment with pepsin was totally soluble in SDS sample buffer. Thus densitometry of Coomassie-stained alpha chains was chosen as being the most straightforward method of accurate quantification, taking into account the assumptions mentioned previously.

The presence of a relatively large amount of type I collagen in porcine cartilage is surprising as it had been thought that mammalian cartilage does not contain this collagen type. Type I collagen expression does occur during cartilage repair, as in osteoarthritis (Adam and Deyl, 1983), and type I collagen mRNA exists in the developing long bones of the human foetus but this may originate from the perichondrium rather than cartilage (Sandberg and Vuorio, 1987). Recently, however, type I collagen has been detected at the surface of adult human cartilage in very low amounts (Aigner et al., 1993). Previous research using immunofluorescence has shown that type I collagen exists at the surface of porcine articular cartilage (Duance, 1983), and has been described in both baboon (Stanescu et al., 1976) and avian cartilage (Eyre et al., 1978). It was assumed that type I was, however, only a minor component but our results indicate a significant amount of type I collagen is present. It is possible that a breakdown product from alpha 1(II) may co-migrate with alpha 1(I) giving rise to an incorrectly high value for type I collagen when subjected to densitometry, but there was no evidence for this from western blotting. The quantity and location of this collagen throughout both porcine articular and growth plate cartilages plus the concomitant decrease in type II collagen must alter the overall mechanical characteristics of these tissues when compared with other species.

Type V collagen has been implicated in the control of type I fibril formation (Birk et al., 1990) and is thought to be present in cartilage (Eyre et al., 1987b; Furuto et al., 1991). However, using an antibody that gave a strong signal with purified type V from porcine placenta and a very sensitive chemiluminescent immunoblotting technique, this collagen type was not detected in either cartilage. It must therefore be assumed that porcine cartilage does not contain type V collagen or that the levels are below the detection limits of the methods used here. Type VI collagen is thought to have a role in cell attachment as the molecule is known to contain RGD sequences. However, in this study it appears as a minor component mainly in the matrix at the articular surface, which suggests a role other than cell adhesion. Type VI collagen has been shown to interact strongly with type I collagen (Bonaldo et al., 1990) and this function may be reflected in the co-localization of these collagen types at the articular surface. The presence of both types I and VI may indicate that the articular surface in pig has differentiated unusually for a mammalian cartilage. If the articular surface is examined progressively closer to the periochondrial ring, the staining for types I and VI becomes gradually thicker and more fibrous and there is no definite change from cartilage surface staining to periosteal staining (data not shown). Both tissues develop from the cartilaginous anlage and it is possible that the surface region of porcine articular cartilage is closely related to periosteal tissue. This has also been suggested to be the case in human articular cartilage, although a much weaker signal was obtained (Aigner et al., 1992).

Type IX collagen is known to have a role in the regulation of type II fibril diameter (Wotton et al., 1988) and hypothetically as a molecule that binds together the type II collagen meshwork (Smith and Brandt, 1992). It is possible that the low amount of type IX found in porcine articular and growth plate cartilages compared to other species partly reflects the reduction in type II collagen. However, the ratio of type IX to type II in articular cartilage is considerably different to that of growth plate cartilage so there are probably structural differences between the two tissues. Type XI collagen, like types V and IX, has also been proposed as a candidate for fibril diameter regulation (Mendler et al., 1989). There is a considerable difference in the ratio of type XI to type II between porcine articular and growth plate cartilage and therefore it would seem likely that the two types of cartilage would show comparatively large differences in fibril diameter. This will be the subject of future studies.

The hypothesis that type XI lies within the type II fibril (Mendler et al., 1989) may be supported by the extraction data (Table 2). At least half of the total type II collagen in growth plate can be removed without any apparent loss of type XI collagen and the majority of type II in articular cartilage is removed with a comparatively small loss of type XI. It is possible that the type XI molecule is protected against extraction by the presence of the type II collagen that surrounds it. Subsequent treatment of the pepsin-insoluble pellet with additional pepsin or with leech hyaluronidase will extract more type II and a small proportion of type XI from growth plate and slightly more of both types from articular cartilage, but the majority of growth plate type XI still cannot be extracted (data not shown). The resistance to extraction cannot be explained by cross-linking as type XI appears unable to form mature cross-links (Wu and Eyre, 1984). The immature, divalent keto-imine cross-links form between the N-telopeptide and the helical region and would thus be susceptible to pepsin digestion. The observed total solubility in SDS of the pepsin-insoluble material and the ability of various high ionic strength salt buffers to extract some type XI collagen suggests ionic interactions with other components of the matrix. It should also be noted that despite the presence of nearly twice the amount of type XI collagen in growth plate when compared to articular cartilage, a weaker signal is obtained from growth plate using immunofluorescence. This suggests that epitopes on the type XI molecule in growth plate may be less accessible to the antibody and thus may be less readily digested by the enzymes used for the extraction procedures. This data, coupled with the overall differences in quantity of type XI present in the two tissues, again points to considerable structural differences between the two types of cartilage and the apparent resistance to extraction of type XI may have implications in the turnover of growth plate cartilage.

The postulated structural differences between the two tissues are supported by the collagen cross-link data. Articular cartilage is generally considered to be a stable tissue with a very slow turnover, even in rapidly growing immature animals. Growth plate cartilage is more dynamic and thus would be expected to have fewer mature collagen cross-links than articular cartilage in the same animal. This is borne out by the data presented in this study. The differences in the levels of the mature cross-link HYL-PYR between the two tissues is sufficient to postulate that growth plate cartilage is a structurally weaker tissue than articular cartilage in immature pigs, although the higher levels of immature cross-links in growth plate may well redress the balance. However, it should be noted that, without exception, the major collagens (types I, II, IX, and XI) are all less easily extracted by pepsin treatment from growth plate than articular cartilage (Table 2). This suggests some form of matrix interaction other than cross-linking which may confer additional stability on growth plate cartilage which would enable rapid turnover without loss of structural integrity.

This study has evaluated the collagenous components of porcine growth plate and shown that there are several significant structural and compositional differences between growth plate and articular cartilage. The data presented here will provide a useful basis for further studies on disorders of joint cartilages in commercially produced pigs.

This work was supported by a grant from the Agricultural and Food Research Council.