Osteoarthritis (OA) is a degenerative condition characterized by loss of articular cartilage and joint remodeling. The degradation of cartilage is characterized by increased expression of matrix metalloproteinases (MMPs), cytokines including interleukin-1 (IL-1) and tumor necrosis factor α (TNFα) and their receptors, and nitric oxide synthase (NOS) (1). Aging mice present valuable opportunities for the study of the pathogenesis of OA. We have previously demonstrated a naturally occurring degeneration of the medial tibial cartilage in the STR/ort mouse as a model of OA (2). Accelerated development of OA in mice can be induced by partial medial meniscectomy (PMM). This results in rapidly developing tibial cartilage degeneration (3). Articular cartilage degeneration in OA is thought to result from an imbalance between anabolic and catabolic pathways involved in the turnover of its extracellular matrix (ECM) (1). Changes in the expression or activity of several cytokines and matrix-degrading enzymes, such as MMPs and aggrecanases, have been implicated in the pathogenesis (1).
Addition of IL-1 to healthy articular cartilage induces the catabolism of cartilage ECM proteoglycans (4) and type II collagen (CII) (5, 6) and suppresses the synthesis of matrix components such as CII and proteoglycan (7, 8). IL-1 is a known inducer of MMPs (9, 10), and its expression is increased at an early stage in the development of OA in the STR/ort mouse (11) and in human OA (12). IL-1β–converting enzyme (ICE) is the physiologic modulator of IL-1β generation and generates the mature 17-kd IL-1β cytokine (13). ICE is expressed in human synovial membrane and cartilage, with significantly more cells staining positive in OA tissue than in normal tissue (14). Overexpression of ICE, also known as caspase 1, can induce apoptosis in transfected cell lines (15).
The MMP stromelysin 1 (SLN-1) is thought to play a major role in cartilage metabolism due to its ability to act on a wide range of protein substrates. It has been shown to be up-regulated in OA cartilage (16). It is an activator of procollagenase (17) and progelatinases (18) and can cleave the nonhelical regions of CII (19) and the aggrecan core protein (20). It can also cleave fibronectin and the minor cartilage collagens, types IX and XI (21). Mice deficient in the gene for SLN-1 fail to exhibit cleavage of aggrecan by MMPs or collagenase cleavage of CII during antigen-induced arthritis (22); nevertheless, cartilage proteoglycans are depleted, presumably by the action of aggrecanases.
The enzyme inducible NOS (iNOS) is responsible for NO production in cartilage. Production of NO is increased in OA, as shown by elevated nitrite concentration in the synovial fluid and serum of OA patients (23). NO can inhibit the synthesis of matrix macromolecules (24–26) and enhance or suppress MMP activity (27, 28). It can also stimulate the synthesis of IL-1 receptors on chondrocytes, which, along with a decrease in IL-1 receptor antagonist production, may lead to increased matrix degradation (29). NO can also induce chondrocyte apoptosis, a recognized feature of OA (30).
Abrogation of the expression of genes encoding such factors may provide information about their role in the disease process. In this study we induced secondary OA by sectioning of the medial collateral ligament and PMM in mice in which genes encoding IL-1β, ICE, SLN-1, and iNOS had been deleted. The results of these deletions in relation to the development of OA lesions are described.
To assess dependency (or lack thereof) of cartilage matrix breakdown on these factors, we graded histologically the severity of tibial plateau cartilage lesions in wild-type (WT) and knockout mice at different times postoperatively. Proteolytic cleavage of CII by collagenase was localized using the Col2-3/4Cshort antibody that specifically recognizes a carboxy-terminal neoepitope on the three-quarter fragment resulting from cleavage of collagen by a collagenase (31). There is evidence of enhanced collagen cleavage and of both aggrecanase- and MMP-mediated turnover of aggrecan in human OA cartilage (31, 32) and in naturally occurring murine OA (34–36), and we investigated cleavage of the matrix proteoglycan aggrecan by aggrecanase 1 or aggrecanase 2, or by MMPs, using the anti–carboxy-terminal NITEGE (32) and VDIPEN (33) neoepitope antibodies, respectively. The occurrence of tibial chondrocyte apoptosis was also assessed, using a DNA fragmentation (TUNEL) assay.
- Top of page
- MATERIALS AND METHODS
Unexpectedly, ICE-, iNOS-, and IL-1β–null mice that underwent surgery in one limb developed OA lesions in the lateral tibial plateau of the contralateral unoperated limb as well as in the operated limb. Since such lesions occurred only rarely in unoperated limbs of WT and SLN-1–null mice, it is unlikely that they are due solely to mechanical changes following contralateral surgery. Moreover, whereas both WT and all gene-deficient mouse groups developed OA lesions in the medial tibial plateau of meniscectomized joints, these appeared more rapidly and were of greater overall severity in the knockout mice. These two observations suggest that each of the deleted genes, though encoding factors favoring catabolism, is important physiologically in maintaining the balance between anabolism and catabolism of cartilage matrix. If one of these genes is knocked out, other less well-regulated catabolic factors/pathways may take over (see below), leading to changes in cartilage metabolism and an abnormal matrix being laid down.
Changes in immunostaining for the VDIPEN, NITEGE, and Col2-3/4short neoepitopes provided direct evidence of enhanced CII and aggrecan catabolism in lesional areas. Although VDIPEN and NITEGE epitopes are present in normal murine cartilage (36), they, together with the Col2-3/4Cshort neoepitope, which is not seen in normal cartilage (35), are evident in increased amounts in focal areas around and within even the earliest histopathologic OA lesions. There is also associated loss of Alcian blue staining, denoting a depletion of proteoglycan which is mainly aggrecan.
Recently it was reported that AOAMTS-4 (aggrecanase 1), while cleaving aggrecan core protein primarily at the Glu373/Ala374 site, is also capable of cleaving it slowly and secondarily at the Asn341/Phe342 site, potentially creating VDIPEN neoepitopes, which were thought previously to be produced exclusively by MMPs (42). However, it is noteworthy that strong immunostaining for the aggrecanase neoepitope NITEGE was observed at the surface of normal tibial cartilage in this study (Figure 4A) and a previous investigation (36), in contrast with the absence of staining for VDIPEN neoepitope at this site (Figure 3A). This suggests that such secondary cleavages of aggrecan by aggrecanase do not necessarily occur in vivo, or if they do, only very low levels of VDIPEN neoepitope are produced.
The distribution patterns of aggrecan and collagen neoepitopes associated with osteoarthritic lesions were the same for all mice, irrespective of whether the limbs were unoperated (e.g., IL-1β–knockout mice) or operated, and independent of the gene deletion or WT mouse group investigated. These results are consistent with previous findings in the STR/ort mouse (35, 36) and in C57BL/6 and BALB/c mice (34) and indicate the focal nature of the biochemical as well as the histopathologic changes in murine OA, especially in the early stages of the disease. These focal changes are very different from the widespread changes in neoepitope levels seen in inflammatory arthritides (34, 35). OA, irrespective of its initiating cause, may be a disorder, at least initially, of localized groups of chondrocytes, whereas inflammatory arthritis involves widespread chondrocyte responses to signals originating outside the cartilage. Similar neoepitope expression is seen in human OA (31, 32).
Another emerging feature of OA is its association with chondrocyte death, irrespective of the initiating cause. This occurs in human OA (30, 43, 44), in idiopathic OA in the STR/ort mouse (2), and, in the present study, in the tibial cartilage of operated limbs of WT and knockout mice and in contralateral unoperated limbs of iNOS-, ICE-, and IL-1β–knockout animals. One consequence must be that no new cartilage matrix would be synthesized in the absence of living chondrocytes in areas already subjected to enhanced degradation, leading to increased compromise of the mechanical function of cartilage and the accelerated development of lesions.
Why should deletion of genes for catabolic factors cause increased susceptibility to lesion formation? IL-1β–deficient mice undergo normal growth (45), implying normal cartilage matrix synthesis in the absence of IL-1β signaling. However, compensatory changes in the expression of other genes involved in cartilage catabolism, such as TNFα or IL-1α, may occur. Also, since binding of IL-1β to its receptor activates several kinases in the cell (46), total deficiency of the cytokine may change the overall kinase activity profile in chondrocytes, with downstream effects on the cartilage matrix.
It is noteworthy that IL-1β plays a crucial role in the propagation of joint inflammation (47) and IL-1β–knockout mice are resistant to collagen-induced arthritis (48) and to zymosan-induced arthritis (49). However, this may not be important in OA, in which, at least in the early stages, inflammation is absent. It is interesting that overall, more severe OA lesions occurred in the lateral tibial plateau of unoperated limbs of IL-1β–null mice than in other knockout groups. Mechanical load on the lateral cartilage is likely to differ from that on the medial side, but the intracellular signaling pathways in chondrocytes which respond to load are not understood. It is possible that there is cross-talk between them and other pathways, including that triggered by IL-1β binding to its receptor. Absence of IL-1β would change the pattern of cross-talk, leading to downstream effects. This concept may also help to explain why development of OA in the medial cartilage after surgery is accelerated in IL-1β–null mice compared with WT mice, since the surgery must induce marked changes in loading on that cartilage, in addition to the existing cytokine deficiency.
Although IL-1β is a known inducer of MMPs (7, 10) and of aggrecanase activity (50), the absence of IL-1β did not affect the pattern or intensity of VDIPEN and NITEGE staining in either unoperated or operated knockout mice. Collagenase cleavage of CII was also evident in IL-1β–null mice. Thus, factors other than IL-1β must be involved in stimulating the activity of all of these enzymes in murine cartilage.
SLN-1–knockout mice have a phenotype that is apparently normal (38). SLN-1 cleaves aggrecan to produce the VDIPEN epitope (33), but since the pattern and intensity of VDIPEN immunostaining were the same in SLN-1–null mice as in WT mice, other enzymes must also do this. VDIPEN staining was unchanged in SLN-1–knockout mice with collagen-induced arthritis (38), but was eliminated in the antigen-induced arthritis model, where it was absent even in control mice of another strain (22). Col2-3/4Cshort staining was unaffected in our knockout mice despite the fact that SLN-1 can activate procollagenase (17). It seems likely that SLN-1 may be replaced by SLN-2 (MMP-10) or another MMP. SLN-2 has proteoglycanase activities that are indistinguishable from those of SLN-1 and can activate zymogen forms of MMP-1 and MMP-8 (51). Little is known about SLN-2 in articular cartilage, but expression in human neonatal rib chondrocytes has been reported (52).
ICE-knockout mice also have an overtly normal phenotype (39). Several other proteases can convert the IL-1β precursor to its active form and appear to do so in synovial fluid of patients with inflammatory arthritis (53). Fas ligand also induces IL-1β processing independently of ICE (54). Presumably, the reason advanced lesions did not form in the unoperated limbs of ICE-knockout mice as frequently as in IL-1β–knockout mice is related to IL-1β activation by other proteases. ICE is clearly not essential for stimulating MMP and aggrecanase activity in the surgical model as indicated by VDIPEN, NITEGE, and Col2-3/4Cshort immunostaining, but, as discussed above, IL-1β is also not obligatory for these activities. Neither does ICE play an essential role in chondrocyte apoptosis in advanced OA, since TUNEL-positive cells are seen in the damaged cartilage of ICE-null mice. Accelerated OA in these animals probably occurs for the same reasons as in IL-1β–null mice.
The bones and joints of iNOS-knockout mice have been reported to be normal (40). However, iNOS is up-regulated in chondrocytes after sectioning of the canine anterior cruciate ligament (55). Nitric oxide is produced by chondrocytes stimulated with IL-1β, possibly mediating the suppressive effect of the cytokine on matrix synthesis (24–26, 56). Thus, we anticipated that deletion of iNOS may retard development of OA in the PMM model. However, development of OA was accelerated. One speculation is that the other two isoforms of NOS, neuronal NOS and endothelial cell NOS, contribute to cartilage degradation. Cohen et al reported that reduced NO accumulation, and subsequent matrix preservation, in arthritic human cartilage explants following treatment with methylene blue, an NOS inhibitor, is due to the down-regulation of all 3 isoforms (57). Thus, iNOS deficiency in chondrocytes of gene-deleted mice may be compensated for by increased NO synthesis by the other 2 NOS isoforms. However, it has also been shown that NOS inhibitors can enhance proteoglycan catabolism occurring in response to IL-1 (25, 28). Thus NO may also normally regulate MMP expression.
In conclusion, each deleted gene in this study encodes a factor which, in theory, has an important role in the development of OA. However, our results show accelerated development of OA in the medial tibial plateau cartilage after surgery in all knockout groups and, in the iNOS, ICE, and IL-1β knockouts, a higher level of OA lesions in the lateral cartilage of unoperated mice than is found in WT mice. One possible explanation for this is that these genes encode molecules that are essential for homeostasis in healthy cartilage, balancing anabolism and catabolism of the cartilage matrix. Another is that overcompensation of like genes may have taken place during embryonic development, resulting in the knockouts becoming more susceptible to cartilage degradation.