Sites of collagenase cleavage and denaturation of type II collagen in aging and osteoarthritic articular cartilage and their relationship to the distribution of matrix metalloproteinase 1 and matrix metalloproteinase 13
Shriners Hospitals for Children, Montreal, Quebec, Canada
To determine the sites of cleavage and denaturation of type II collagen (CII) by collagenase(s) in healthy and osteoarthritic (OA) human articular cartilage and their relationship to the distribution of matrix metalloproteinase 1 (MMP-1) and MMP-13.
Single (per subject) full-depth specimens from femoral condylar cartilage were isolated from articulating surfaces at autopsy from 8 subjects without arthritis and during arthroplasty from 10 patients with OA. Fixed frozen sections of cartilage were examined by immunoperoxidase localization, using antibodies to the collagenase-generated cleavage site in CII, to an intrachain epitope recognized only in denatured CII, and to MMP-1 and MMP-13 (proenzyme, activated enzyme, or enzyme/inhibitor complex).
Staining for collagen cleavage, denaturation, and both MMPs was weak to moderate and was frequently observed in pericellular sites in cartilage from younger, nonarthritic subjects. In specimens from older subjects, this staining was often more widespread and of greater intensity. Similar staining was usually, but not always, seen for all antibodies. In OA cartilage, staining was often stronger and more intense than that in normal cartilage from older subjects, and the distribution of staining was often similar for the different antibodies. Pericellular staining in the deep zone was frequently more pronounced in arthritic cartilage and extended to territorial and sometimes interterritorial sites. In very degenerate specimens, staining was distributed throughout most of the cartilage matrix.
These observations provide evidence for the presence of limited cleavage and denaturation of CII restricted to mainly pericellular and superficial sites in cartilage from younger, healthy subjects, where MMP-1 and MMP-13 are also selectively localized. Collagen degradation is more extensive and often more pronounced in cartilage from older, nonarthritic subjects. Characteristic changes in early OA are similar to those seen with aging in cartilage from older, healthy subjects, with collagen damage and collagenases concentrated closer to the articular surface. There was usually a close correspondence between the cleavage and denaturation of CII and the sites at which these collagenases were detected, suggesting that both MMPs are involved in the physiology and pathology. There was no evidence that the damage to CII is ordinarily initiated in sites other than at and near the articular surface and around chondrocytes.
Degradation and loss of articular cartilage are fundamental features of osteoarthritis (OA). The process ordinarily is thought to be very slow, and pathology may not be clinically apparent for up to 20–30 years. Type II collagen (CII), the major component of the extracellular matrix of cartilage and the collagen fibril, provides this tissue with its tensile properties (1, 2). In the setting of OA, denaturation of this molecule in articular cartilage is excessive (3) and is accompanied by increased cleavage by collagenases (4). The net result is a loss of tensile properties and CII (5, 6), associated with damage to this molecule. Recent studies using inhibitors of collagenases have provided evidence that matrix metalloproteinase 1 (MMP-1; collagenase 1) and especially MMP-13 (collagenase 3) are involved in this cleavage (7, 8). Excessive activity of MMP-13 can generate the type of pathology observed in OA (9).
We previously demonstrated that in OA, early damage to articular cartilage is usually characterized by denaturation of CII at and close to the articular surface, extending progressively into the underlying cartilage, first in pericellular sites and then in territorial and more remote interterritorial sites (10). There have also been reports of more pronounced staining for MMPs, such as stromelysin 1 (an activator of proMMPs such as procollagenase) in sites similar to those where some investigators have observed damage (11–13). Other authors have reported the presence of MMP-13 in the deep layer of articular cartilage during early cartilage degeneration in OA (14).
To our knowledge, there have been no human studies in which damage to matrix molecules such as CII has been compared with localization of the proteinases (e.g., collagenases) that may be involved in cleavage of these molecules. Moreover, there is also a lack of definitive studies of collagenase cleavage of CII in human articular cartilage, although recent studies revealed a close association between lesion formation in mice, rats, and guinea pigs and cleavage of this molecule (15–17). The present study examines in detail the sites of cleavage and denaturation of CII by collagenases in OA and nonarthritic human articular cartilage and their relationship to the presence of the collagenases MMP-1 and MMP-13.
MATERIALS AND METHODS
Full-depth specimens of femoral condylar cartilage (0.5–1.0 cm2 surface area) were removed with a sharp scalpel from the anterior articulating regions of the knee joints of 8 adult autopsy subjects of various ages and sexes, within 15 hours of death (Figures 1, 2a, and 3); these subjects had no observable arthritic joint abnormalities and had not recently (within 2–3 months) undergone chemotherapy. Corresponding site–matched full-depth specimens of femoral condylar cartilage from 10 patients (Figures 2b, 4–6) who underwent total knee arthroplasty for OA, diagnosed according to the criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (18), were removed to the laboratory within an hour of surgery. Osteophytic cartilage was excluded from this study.
The antibodies used in this study have been previously described and include a mouse monoclonal antibody to an intrachain epitope in CII, which is used to detect denaturation of CII (i.e., the antibody does not react with triple helical CII unless the CII has been denatured [proteolytically cleaved by a collagenase]) (3); a rabbit polyclonal antibody to a neoepitope generated by cleavage of CII by a collagenase that resides at the carboxy terminus of this primary cleavage site (named Col2-3/4Cshort) (4); a rabbit antipeptide antibody to a human MMP-13 peptide sequence (9, 19) that recognizes the sequence PNPKHPKTPEK corresponding to amino acids 275–285; and a rabbit antipeptide antibody to human MMP-1 (19). These antibodies, recognizing the peptide sequence AEKAFQLWSNV corresponding to amino acids 133–143, exhibited no cross-reactivity or activity with any other known protein.
The methods used were generally similar to those previously described (10). Briefly, frozen sections were cut and fixed in paraformaldehyde. After rinsing with phosphate buffered saline, sections were always treated with chondroitinase ABC (10) to deplete chondroitin sulfate. Depletion was confirmed by a lack of Safranin O staining, which was performed as previously described (10). Localization by immunoperoxidase staining involved use of primary antibody binding of diluted antiserum or monoclonal antibody, which is detected by secondary anti-mouse or anti-rabbit antibodies labeled with biotin. Primary antibody binding is finally detected with streptavidin–peroxidase. The reagents were obtained from Dako Cedarlane (Hornby, Ontario, Canada). Binding of these antibodies in 6-μm–thick frozen sections was examined in triplicate on at least 2 different occasions to ensure reproducibility. Antibody dilutions and color development were arranged to provide maximum staining intensity. Duplicate controls were prepared by prior absorption of the primary antibody, using the peptides listed above and in the manner described (10).
Figures 1 and 3 show specimens obtained from 8 autopsy subjects. In specimens from the 4 younger subjects (ages 20–35 years), staining patterns for CII denaturation, cleavage, MMP-1, and MMP-13 were similar (Figure 1). Staining was generally weak to moderate and was restricted mainly to pericellular sites throughout the full depth of the articular cartilage (Figures 1 and 2a). Sometimes pericellular staining was more pronounced in the mid to deep zones. In cartilage from 2 subjects, staining was weak to moderately diffuse immediately under the articular surface in more superficial sites. Staining was also observed at the articular surface (Figure 1).
In cartilage from older subjects (≥52 years), the presence of pericellular staining and more intense staining was often also observed more remote from the cell in territorial sites, particularly in the lower mid and deep zones (Figures 3b and d). A band of strong, diffuse staining for all epitopes throughout the matrix in the superficial and mainly mid zones (Figure 3b) and sometimes extending further into the deep zone, where pericellular staining was less pronounced than in younger individuals (compare Figure 3b with Figure 1), was often observed in cartilage obtained from older subjects. Frequently, this strong staining ceased abruptly in the mid-deep zone, as shown in Figure 3b.
In cartilage from older subjects, evidence of collagen denaturation was observed throughout the cartilage matrix and the depth of the cartilage and was most pronounced in pericellular sites (Figure 3d). In some cases staining for both denaturation and cleavage of collagen was similar (Figures 1–3), but in other cases staining for cleavage was either more or less pronounced than that for denaturation (Figures 3c and d).
In some samples, the intensity and distribution of staining patterns for MMP-1 and MMP-13 were similar (Figures 1, 2a, and 3c), but in others the intensity for MMP-13 was clearly more pronounced (Figures 3a, b, and d). In most specimens, the distribution and relative intensity for cleavage of collagen were comparable to that for MMP-13. In one specimen, however (Figure 3c), staining for cleavage was selectively observed in the superficial zone, while staining for MMP-13 was weak in this site and elsewhere.
Specificity of staining.
The specificity of staining for all antibodies was clearly demonstrated by an almost complete inhibition of antibody/Fab binding following prior absorption with the specific collagen or MMP peptide against which the antibody had been prepared. Figure 4 shows an example of OA cartilage in which staining for denaturation and cleavage of CII, as well as MMP-1 and MMP-13, was largely prevented following this pretreatment. The only staining observed was weak, diffuse, and of an intensity that was generally much lower than that observed with antibody alone. Because some superficial staining persisted in these OA cartilage specimens, the specificity is not proven. Control samples with intact articular surfaces (prepared from cartilage obtained from nonarthritic subjects) did not reveal any surface staining after absorption of the peptide, pointing to its specificity (data not shown).
A variety of staining patterns were observed in OA cartilage, several of which were similar to those observed in nonarthritic cartilage. Figures 5 and 6 show results in cartilage specimens obtained from 10 different patients during knee arthroplasty.
Pericellular staining similar to that seen in cartilage from younger (ages 20–30 years), nonarthritic persons (Figure 5a) was observed occasionally, although superficial fibrillation was clearly detectable. Pericellular cleavage was again seen in the absence of significant denaturation (see Figure 3d) and corresponded more to the presence and intensity of staining for MMP-13 than to that for MMP-1, as observed in nonarthritic cartilage (Figures 3a and d).
What was most striking in these OA cartilage specimens was the frequent similarity of staining for all antibodies in articular cartilage that exhibited early superficial fibrillation. In 6 of the 8 specimens in this category (early degeneration) that showed staining other than pericellular (Figures 5b–e and Figures 6b and c), staining was generally present throughout the matrix in the superficial and mid zones, and was usually most intense in pericellular sites. It was in these pericellular sites, in the deep zone, that weaker staining in nonarthritic cartilage was most commonly seen. In all 6 of these cases, staining for collagen denaturation and cleavage was generally seen in the same sites as was staining for MMP-1 and MMP-13. As was sometimes observed in nonarthritic cartilage (Figure 3b), OA cartilage often displayed a sharp boundary between the zone of almost total cartilage matrix involvement and the zone where staining was more confined to pericellular sites (Figures 5b–d and Figure 6b). OA cartilage with an intact articular surface often demonstrated more evidence of collagen damage, associated with the presence of MMPs, in the deep zone. In 3 OA specimens showing intact articular surfaces, collagen degradation was seen primarily in pericellular sites (Figures 5e and 6b and c). A similar observation was made in 1 specimen of normal cartilage (Figure 3d).
Only occasionally did we observe discontinuous distribution of staining, where collagen damage was seen in more than 1 separate site. In 1 specimen (Figure 6a), bands of cleavage were shown in the superficial-mid and deep zones, with an isolated pocket of activity in the mid zone that closely corresponded to the distribution of staining for both MMP-1 and MMP-13. This cleavage was not accompanied by collagen denaturation in this sample or in the sample shown in Figure 4a, unlike in the majority of other specimens (8 of 10). In the setting of advanced degeneration (Figure 6e), collagen denaturation and cleavage were observed throughout the cartilage in sites similar to those where MMP-13 and MMP-1 were localized. The specimen shown in Figure 6e was typical of very degenerate, advanced lesions (data not shown).
In essentially all specimens from patients with OA, we observed a codistribution of MMP-1 and MMP-13 in sites where cleavage by collagenase was observed, as was seen in the majority of specimens from nonarthritic subjects. In both normal and OA cartilage, collagen denaturation was sometimes seen in interterritorial sites in the deep zone, remote from chondrocytes, in the absence of staining for collagen cleavage and for MMP-1 and MMP-13 (Figures 3c, 5b–e, and 6c).
Immunoanalyses of the cleavage (4) and denaturation (3) (which follows cleavage) of the triple helix of CII have clearly revealed a significant increase in these parameters in human OA cartilage recovered at arthroplasty. It is apparent that immunolocalization can be used to distinguish the excessive damage to CII in OA cartilage. Immunoperoxidase localization shows the increased damage to CII and the increased presence of MMP-1 and MMP-13 in pericellular and territorial sites in the deep zone of less degenerate OA cartilage. It demonstrates the widespread distribution of collagen damage and the presence of these collagenases throughout very degenerate OA cartilage. It also shows where such damage occurs in both aging and OA cartilage, regardless of whether sites of denaturation correspond to sites of cleavage (and vice versa), and whether these events are related to the sites in which collagenases are localized in these cartilages. However, because primary antibody dilution and color development were arranged to provide maximum intensity of staining, it is not possible in this study to compare the relative concentrations of the epitopes against which the antibodies were directed. Therefore, we make no attempt to compare relative amounts of collagenases using these methods.
What is very clear from the present study is that the cleavage of CII by collagenases, as revealed by antibodies to this primary cleavage site in the triple helix of CII, frequently (but not always) corresponds to the sites where MMP-1 and MMP-13 are mainly detected, as observed by others (17). Such an association was recently observed for collagen cleavage and MMP-13 in a transgenic mouse overexpressing active human MMP-13 in articular cartilage, resulting in development of degenerative focal lesions resembling OA (9). At our laboratory, recent studies of human articular cartilage have provided evidence for a role for increased activities of these collagenases in the cleavage of both resident molecules (MMP-13) and those that are newly synthesized (MMP-1) (7).
In the present study, we cannot discriminate between these collagen molecules in terms of when they were synthesized, nor can we ascribe activity to these collagenases, because our antibodies detect proenzyme, activated enzyme, and tissue inhibitor of metalloproteinases–enzyme complex (Wu W, Poole AR: unpublished observations). However, there was often close correspondence in terms of staining location between cleavage of CII by collagenases and the presence of MMP-1 and MMP-13. The concentration of collagenase-cleaved CII mainly in pericellular sites in cartilage obtained from the 4 younger subjects (20–35 years of age) no doubt reflects a normal turnover process that is a feature of healthy cartilage in the young adult. The fact that this concentration is seen predominantly in pericellular sites indicates that chondrocytes release the collagenases that produce these cleavages. In such cases, we usually found the MMPs in these same pericellular sites, suggesting that they also originated from chondrocytes. These proteinases are probably present in extracellular sites because collagenases can bind to collagen fibrils by their hemopexin domains (12), providing a means of retaining them in the extracellular matrix, bound to collagen fibrils at sites of activity. Involvement in this degradative process of collagen molecules more remote from the chondrocytes in territorial and interterritorial sites, which was seen in older individuals and patients with OA, indicates cleavage of collagen fibrils in sites where this is not usually observed in younger adults. Thus, some of the changes observed in nonarthritic cartilage may represent preclinical changes that could lead to OA.
The predominance of collagen damage and MMPs in sites closer to the articular surface (with less pronounced evidence of this in the deep zone) suggests, as in earlier studies of denaturation alone (10), that this degradative process begins in the more superficial cartilage and proceeds progressively into the underlying articular cartilage of the mid and deep zones. The often-present line of demarcation between extensive matrix staining and the adjacent deeper, predominantly pericellular staining suggests that the deeper cells may be activated by cytokines acting in a paracrine manner, which are generated by the adjacent, more superficial cells. This suggests a creeping substitution of good tissue with damaged cartilage as part of the pathologic process, which we believe takes place over many years.
This observation is in contrast to a report that MMP-13 is found predominantly in the deep zone in developing human OA (14). In studies of the natural development of OA in mice (16), guinea pigs (17), and rats (15), CII cleavage was usually observed in more superficial sites and progressed deeper into the articular cartilage as lesions developed. In a study of cartilage-specific expression of the transgene for active MMP-13 in the mouse, which causes development of a lesion closely resembling human OA (9), MMP-13 and CII cleavage were also detected in pericellular sites in the more superficial cartilage prior to fibrillation, although in the present study, the opposite was sometimes seen in nonarthritic cartilage. This suggests that in OA it is ordinarily these more superficial cells from which these collagenases are initially up-regulated, leading to the excessive cleavage and denaturation of CII seen in experimental OA and in the present investigation.
The striking demarcation between sites in cartilage where CII was cleaved by collagenase and denatured serves to demonstrate the strict localization of this cleavage with respect to the cell and, on a larger scale, with respect to the articular surfaces. Remarkably, these collagenases usually exhibited the same distribution of staining, suggesting that the activity of at least some of the MMP localized by the antibodies accounted for this cleavage.
In specimens in which denaturation was detected in the absence of detectable cleavage, it is likely that the intrachain epitope reflecting denaturation may have persisted from earlier cleavage or damage to the collagen molecule. A cleavage epitope may no longer be observed, because it can be subsequently removed from cartilage, as we have shown in culture (7, 8).
Together, these studies represent a comprehensive investigation of the changes and interrelationships between CII cleavage and denaturation in human OA and nonarthritic cartilage and show how these events relate to the presence of collagenases that have the potential to produce these events. They provide evidence for the often close relationship of these collagenases to development of damage to CII in aging and OA. The results demonstrate that collagen cleavage and denaturation, matrix degeneration, and the presence of collagenases in human articular cartilage in healthy and OA cartilage are usually closely interrelated. They do not provide further evidence supporting the activity of one collagenase over another, because only the localization of these MMPs, not the concentrations, was examined (conditions of staining did not permit this comparison). Neither do these studies provide direct evidence to indicate that these proteases are in fact involved in the degradative process. The activation of these collagenases is as important as their localization.
These reagents and methods, when used in the manner we have described, do not offer a definitive means of distinguishing between OA cartilage and normal cartilage from older persons. The staining patterns in cartilage from older subjects are often similar to those seen in OA. Using immunoassay, however, we previously showed that cleavage and denaturation of CII are increased in OA cartilage at arthroplasty (3, 4). The present results do reveal that changes that occur with aging are often similar to those seen in OA specimens recovered at arthroplasty. Whether the age-related changes are fully representative of earlier degenerative changes that occur in patients with OA remains to be established. Therefore, these immunohistochemical studies indicate that development of OA may sometimes be part of the natural aging process.