Osteoarthritis (OA) is the most common of the rheumatic disorders and affects the quality of life of millions of people each year. Approximately 10% of individuals older than age 60 years have some form of OA, but the disease also occurs frequently in younger persons. OA affects the joints and ultimately leads to erosion of cartilage and sclerosis of subchondral bone (1). Other features of OA include the formation of osteophytes (bony outgrowth at the joint margins) and fibrosis of the synovium, as well as changes in the composition of synovial fluid. Although the etiology of OA still must be elucidated, it is likely that there is not just one cause. In some cases, the primary cause may be a biomechanical problem (e.g., trauma). In other cases, OA might be caused by a derailment of chondrocytes in the cartilage. Increasing evidence suggests that synovial activation may also be involved in the etiology of OA, because an accumulation of macrophages and even inflammation are demonstrated in the synovium of many patients with early OA (2).
Recently, we demonstrated the involvement of macrophages in OA-related pathology (3, 4). In a study in which synovial lining macrophages in the knee joint were selectively depleted prior to elicitation of collagenase-induced experimental OA, we observed that macrophages mediate osteophyte formation and fibrosis in the early stages of experimental OA (5). Collagenase-induced experimental OA is a model that is based on the induction of joint instability by intraarticular injection of collagenase. This causes weakening of ligaments that normally help to stabilize the joint, which subsequently leads to OA pathology within 6 weeks after induction. No direct damage of cartilage by the injected collagenase has been observed in that model. Another model for OA is a spontaneous aging model, because the incidence and severity of OA, in both mice and humans, increase with age.
Eventually, OA leads to the erosion of cartilage. These erosions are thought to be mediated by matrix-degrading enzymes, such as ADAMTS and matrix metalloproteinases (MMPs). Macrophages are capable of expressing and producing a broad spectrum of MMPs. They are also capable of producing mediators that can induce MMP expression by chondrocytes, such as interleukin-1 (IL-1) and tumor necrosis factor α. Several MMPs have been suggested to be involved in OA. There is strong evidence, based on both clinical and experimental data, that MMP-13 plays an important role in cartilage damage in OA (6, 7). The involvement of other MMPs in OA is less clear. For instance, MMP-2 is known to be able to cleave several collagens and is reported to activate other MMPs produced by OA chondrocytes in vitro (8). In patients with OA, plasma MMP-3 levels correlate closely with joint space narrowing (9). Also, during experimental OA, the expression of MMP-3 is increased in synovium and cartilage (10). However, it remains unclear whether these enzymes are involved in the actual OA process. Regulation of the natural inhibitors of MMPs, tissue inhibitors of metalloproteinases (TIMPs), can also be responsible for changes in MMP-mediated cartilage damage during OA. TIMP-1, TIMP-2, and TIMP-3 have been described to be able to inhibit a wide variety of MMPs, including MMP-3 (11).
The objective of the present study was to investigate the involvement of synovial macrophages in MMP-mediated cartilage damage in murine OA, and to explore whether MMP-3 is involved in this process.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
Earlier studies at our laboratory showed that synovial macrophages are involved in OA-related pathology, such as synovial fibrosis and osteophyte formation (3, 4). Based on the results of these aforementioned studies, it was concluded that macrophages produce a (growth) factor that induces this OA-related pathology. In the present study, we clearly demonstrated that macrophages are indeed involved in early cartilage damage, as determined by the occurrence of neoepitopes in the cartilage within 2 weeks after induction of experimental OA. This appearance of neoepitopes is an important first step in the generation of irreversible cartilage damage (14). The macrophage depletion studies demonstrated clear involvement of macrophages in this process. Although F4/80 staining indicated that macrophage depletion was successful, macrophages started to repopulate the lining 3 weeks after the injection of clodronate-laden liposomes (12). Therefore, the effects of macrophages on cartilage damage could not be studied for more than 2 weeks following induction of OA.
In control mice, early cartilage damage markedly increased between week 1 and week 2, whereas no increment in macrophage-depleted joints was observed during this period. Apart from the important observation that macrophages are involved in OA-related cartilage pathology, these results indicate that the MMP-mediated damage is not directly caused by intraarticular injection of collagenase, because the damage in control joints occurred after the first week, at which time the injected collagenase has already been cleared from the joint for several days due to the physical properties of collagenase (5). The damage that develops is caused by the structural damage provoked by collagenase in the ligaments, not by direct effects of collagenase on cartilage.
In several experimental arthritis models, the significant role of macrophages in joint pathology has been demonstrated (15, 16). However, an important role for macrophages in OA was never clearly established, although in recent years more suggestive evidence for the involvement of the synovium and synovial macrophages in OA pathology has emerged (2). An association between the number of macrophages in the synovium and progression of damage has been described in human RA (17, 18), and in early OA an increased presence of macrophages has been shown (2), which suggests that macrophages add to joint damage in OA as well. Macrophages are potent producers of MMPs and have been shown to produce several MMPs (19).
Most MMPs are capable of generating the VDIPEN neoepitopes in aggrecan, which we have now shown to be mediated by synovial macrophages. In the current study, the presence or absence of macrophages in the synovium had a large effect on the expression of MMPs in the synovium, especially MMP-2, MMP-3, and MMP-9. Surprisingly, MMP-12, which is also known as macrophage elastase, was not differentially expressed. In fact, MMP-12 was not expressed at all on day 7. No strong effect on the inhibitors of MMPs in the synovium was observed.
In contrast to the findings in synovial tissue, the enhanced expression of MMPs by chondrocytes during early experimental OA was not under the control of macrophages. Therefore, synovial macrophages most likely contribute directly to cartilage damage during experimental OA via production of MMPs by the synovial cells and their subsequent diffusion to the cartilage, rather then via induction of MMP production by chondrocytes. There is no conclusive evidence proving whether this diffusion of MMPs leads to direct cleavage of matrix proteins in the cartilage, or whether the diffusing MMPs activate the proMMPs expressed by chondrocytes. The mechanism of cartilage matrix breakdown probably is a combination of both direct cleavage of matrix proteins and activation of other proMMPs, and macrophages are key players in both scenarios. The pericellular staining of VDIPEN neoepitopes suggests that in experimental OA, chondrocytes do actively participate in cartilage degradation when macrophages are present.
Apart from degradation of matrix proteins, MMP-2, MMP-3, and MMP-9 have all been described to be potent activators of latent MMPs. Dreier et al (20) reported that MMP-9 is an important mediator in human OA that is produced mainly in its pro form by OA synovial macrophages, not by OA chondrocytes. Those investigators reported activation of MMP-9 either by MMP-3 or by the membrane type 1–MMP/MMP-13 cascade, both of which are produced by OA chondrocytes.
Depletion of macrophages results in a strong reduction of MMP-3 expression in the synovium during experimental OA, as was evident in the present study. Because MMP-3 is a key mediator of irreversible cartilage damage in experimental arthritis (10), we examined cartilage damage in MMP-3–deficient mice during experimental OA. Although the pathology was very mild, MMP-3 was shown to be involved in OA-related cartilage damage in collagenase-induced experimental OA, because cartilage damage was significantly decreased in MMP-3–deficient mice. In addition, in spontaneously occurring OA, the involvement of MMP-3 was even greater, especially in terms of severe cartilage damage. Probably due to the relative resistance of this murine strain to OA, the severe damage did not occur in 1-year-old mice, but in 2-year-old mice severe cartilage damage was clearly mediated by MMP-3. In 1-year-old mice with a B10.RIII background, only minor OA-like changes were observed, mainly proteoglycan depletion and some superficial fibrillations, whereas mice with other genetic backgrounds, such as C57BL/6 mice, already show marked OA-like pathology at this age (21). The notion that genetic background is important in the sensitivity to OA has been described previously (22, 23) and reminds us to be careful with the interpretation of and comparison between studies of knockout animals, especially when animals with different genetic backgrounds are used. This importance of genetic background may also explain the conflict between our findings and those described by Clements et al (24).
The joint compartment in which damage preferentially occurs differs between these 2 models of OA (induced and spontaneous). In both models, damage most often occurs laterally, but the tibia shows the most damage in collagenase-induced OA, and the femur shows the most damage in spontaneous OA. This disparity may be attributable to the differing mechanisms involved in the 2 models. During collagenase-induced OA, a sudden change in joint loading is generated, whereas the process is very slow in spontaneous OA, and differences in composition of the cartilage matrix may be very important.
Although cartilage damage was significantly mediated by MMP-3, part of the cartilage damage remained MMP-3 independent in both models. As noted above, MMP-3 was not the only MMP in the synovium that was regulated by macrophages; strong effects on the expression of MMP-2 and MMP-9 were also demonstrated. For this reason, we tested MMP-9–deficient mice by initiating collagenase-induced OA. However, MMP-9–deficient mice already show spontaneous cartilage pathology at a very young age (Blom AB, et al: unpublished observations), which makes any results difficult to interpret. MMP-9 and many other MMPs are involved in development of the cartilage matrix (25) and other tissue. It would therefore be preferable to use conditional gene targeting to specifically target cartilage, e.g., using tetracycline-controlled transcriptional regulation combined with the Cre/lox system (26, 27).
Interestingly, in MMP-3–knockout mice, the significantly reduced level of cartilage damage that was observed coincided with a significantly decreased presence of macrophages compared with that in control mice. This finding is compatible with the involvement of synovial macrophages in OA pathology. Whether the increased amount of macrophages in control mice results in increased cartilage damage, or whether the presence of macrophages is a result of the enhanced cartilage damage and the subsequent presence of cartilage debris and matrix fragments is not yet clear. In the early stages of experimental OA, macrophages are activated, as was measured by an increase in the expression of myeloid-related protein 14 by these cells (4).
Which factors activate these cells in the synovium still needs clarification, but it is very likely that these stimuli leak from the cartilage, implying that macrophage involvement is a secondary process in OA. However, the possibility that the contribution of macrophages is substantial or that macrophages become involved early in the disease cannot be ruled out. Once stimulated, macrophages can produce high amounts of interleukin-1 (IL-1), which may further activate the synovium in a paracrine or autocrine manner. IL-1 itself is a potent inducer of MMP expression, and blockade of IL-1 during experimental arthritis results in a strong decrease in VDIPEN neoepitope expression in cartilage (28, 29). In several animal models of OA, inhibition of IL-1 has proven successful (30). However, the questions of if and to what extent IL-1 is involved in OA pathology are still being explored by many groups of investigators worldwide.
Among the possible candidate receptors for macrophage activation is Toll-like receptor 4. The involvement of this receptor is suggested because various cartilage matrix fragments that are already implicated in OA, such as fibronectin fragments, are able to activate Toll-like receptor 4 (31, 32). Another possible candidate is the receptor for advanced glycation end products (RAGE), because these advanced glycation end products can be found in (particularly) older cartilage. RAGE and several of its ligands, such as calgranulins (S100 proteins), have been implicated in OA and are expressed on OA macrophages (33, 34).
These findings indicate that the macrophage is a key player in OA pathology, both in the generation of several MMPs in the synovium and in the generation of (possibly MMP-mediated) neoepitopes in the cartilage during an early phase of disease. This may have strong implications for future therapies, because targeting synovial macrophages is possible, as has been described previously (13, 15), and is also a feasible approach in OA, because usually only a limited number of joints are involved.