Osteoarthritis (OA) is the most common form of arthritis and a major cause of pain and disability in older adults (1). OA is often referred to as degenerative joint disease. This is a misnomer because OA is not simply a process of wear and tear, but rather, an abnormal remodeling of joint tissues driven by a host of inflammatory mediators within the affected joint. The most common risk factors for OA include age, sex, prior joint injury, obesity, genetic predisposition, and mechanical factors, including malalignment and abnormal joint shape (2, 3). Despite the multifactorial nature of OA, the pathologic changes seen in osteoarthritic joints have common features that affect the entire joint structure, resulting in pain, deformity, and loss of function.
The pathologic changes seen in OA joints (Figures 1 and 2) include degradation of the articular cartilage, thickening of the subchondral bone, formation of osteophytes, variable degrees of inflammation of the synovium, degeneration of ligaments and, in the knee, the menisci, and hypertrophy of the joint capsule. There can also be changes in periarticular muscles, nerves, bursa, and local fat pads that may contribute to OA or to the symptoms of OA. The findings of pathologic changes in all of the joint tissues are the impetus for considering OA as a disease of the joint as an organ that can result in “joint failure.” In this review, we summarize the key features of OA in the various joint tissues affected and provide an overview of the basic mechanisms currently thought to contribute to the pathologic changes seen in these tissues.
The articular cartilage is altered to some degree in all joints with OA. Cartilage provides a smooth surface with a very low coefficient of friction, allowing for an efficient gliding motion during joint movement. This is facilitated by a boundary layer of lubricants on the articular surface provided by lubricin and hyaluronic acid, which are produced by both chondrocytes and synovial cells (4). In OA, the earliest changes in cartilage appear at the joint surface in areas where mechanical forces, in particular shear stress, are greatest (5).
In normal adult cartilage in the resting, nonstressed steady state, chondrocytes are quiescent cells, and there is very little turnover of the cartilage matrix. In OA, the chondrocytes become “activated,” characterized by cell proliferation, cluster formation, and increased production of both matrix proteins and matrix-degrading enzymes (6). Disruption of the normal resting state of chondrocytes may be viewed as an injury response involving the recapitulation of developmental programs, leading to matrix remodeling, inappropriate hypertrophy-like maturation, and cartilage calcification (6).
The matrix-degrading enzymes found in the OA joint include aggrecanases and collagenases, which are members of the matrix metalloproteinase (MMP) family, as well as several serine and cysteine proteinases (7). Matrix degradation in early OA may be due to MMP-3 and ADAMTS-5, which degrade aggrecan, followed by increased activity of collagenases, in particular MMP-13, which is highly efficient at degrading type II collagen. Once the collagen network is degraded, it appears that a state is reached that cannot be reversed.
Chondrocytes have receptors for extracellular matrix components, many of which are responsive to mechanical stimulation. Activation of these receptors stimulates the production of matrix-degrading proteinases and inflammatory cytokines and chemokines, either as initiating or as feedback amplification events. The type II collagen–containing network in the interterritorial region is normally not accessible to degradation by proteinases because it is coated with proteoglycans. The importance of proteoglycan depletion in cartilage erosion was demonstrated in ADAMTS-5–knockout mice, which are protected against progression in the surgical OA model (8). However, aggrecan depletion, by itself, does not drive OA progression, as suggested by recent studies in MMP-13–knockout mice showing that MMP-13 deficiency inhibits cartilage erosion, but not aggrecan depletion (9).
Recent studies suggest that biomechanical stress may initiate the disruption of the pericellular matrix through the serine proteinase HTRA-1 (10). The receptor tyrosine kinase discoidin domain receptor 2 is then exposed to its ligand, native type II collagen (Figure 3), and preferentially induces and activates MMP-13 (11). Syndecan 4, a trans-membrane heparan sulfate proteoglycan involved in the maintenance of homeostasis, is a positive effector of ADAMTS-5 activation through its control of the synthesis of the stromelysin MMP-3 (12).
Chondrocytes in OA cartilage, especially those in clonal clusters, express cytokine and chemokine receptors, MMPs, and a number of other genes that enhance or modulate inflammatory and catabolic responses, including cyclooxygenase 2 (COX-2), microsomal prostaglandin E synthase 1, soluble phospholipase A2, and inducible nitric oxide synthase (iNOS; or, NOS-2). Activation of chondrocytes by mechanical and inflammatory stimuli occurs primarily through the NF-κB, stress-induced and MAPK pathways (6). Activation of canonical NF-κB (p65/p50) signaling is required for the chondrocytes to express MMPs, NOS-2, COX-2, and interleukin-1 (IL-1). Upon activation, the ERK, JNK, and p38 MAPK cascades coordinate the induction and activation of transcription factors, such as activator protein 1 (cFos/cJun), Ets, and CCAAT/enhancer binding protein β, that regulate the expression of genes involved in catabolic and inflammatory events.
Another primary response factor for the regulation of cytokine-induced MMP-13 in chondrocytes is hypoxia-inducible factor 2α (HIF-2α) (13), which is strongly induced by NF-κB signaling. Induction of both ADAMTS 4 and 5 requires runt-related transcription factor 2 (14), whereas NF-κB and HIF-2α (13) mediate the up-regulation of ADAMTS-4. Recent studies indicate that epigenetic mechanisms also play a role through modulation of the DNA methylation status on promoters that drive the expression of, for example, IL-1β and MMP-13 genes (15) or through dysregulation of the microRNAs that are important for maintenance of homeostasis (16, 17).
Recent studies have also implicated synovial inflammation (discussed further below) and secreted damage-associated molecular patterns, or alarmins, which act as ligands of Toll-like receptors (TLRs), or receptor for advanced glycation end products (RAGE) in the activation of inflammatory and catabolic events in articular cartilage. TLRs are expressed in chondrocytes activated by inflammatory stimuli (18), and TLR-2 and TLR-4, which are present in lesional areas of OA cartilage, may be activated by specific peptide ligands, leading to increased expression of inflammatory and catabolic genes, including MMP-3, MMP-13, and NOS-2, through the cytosolic adaptor myeloid differentiation factor 88 and subsequent NF-κB signaling (19).
High mobility group box chromosomal protein 1 (HMGB-1) has been implicated as a potentiator and contributor to OA by acting on articular chondrocytes (20) or synoviocytes (21) and enhancing inflammatory insults. The alarmins S100A4, A8, A9, and A11, along with HMGB-1, also signal through RAGE and TLRs to drive inflammation-associated matrix catabolism and increase reactive oxygen species (ROS) through up-regulating cytokines and chemokines (22, 23). Proinflammatory cytokines, prostaglandins, ROS, and nitric oxide may also cause oxidative stress and chondrocyte apoptosis by altering mitochondrial function (24).
Wnt pathway signaling may play a role in cartilage destruction in OA through promotion of chondrocyte hypertrophy (25). Chondrocytes express multiple Wnt family members (26), and activation of canonical Wnt signaling through the Frizzled receptors, leading to increased β-catenin activity, appears to promote matrix destruction, while inhibitors of Wnt activation, such as secreted Frizzled-related proteins, may be protective. Because of the important role of the Wnt pathway in regulating bone formation, alterations in Wnt signaling may be involved in both the cartilage and bone changes seen in OA. However, further studies are needed to better define the components of the Wnt pathways that promote OA and separate them from those that may be protective.
As articular cartilage matrix proteins are degraded, fragments of matrix proteins are produced that can feed back and stimulate further matrix destruction. Fragments found in OA cartilage include fibronectin (27, 28), small leucine-rich proteoglycans (SLRPs) (29), and collagen (30). Fibronectin and collagen fragments can stimulate the production of inflammatory cytokines, chemokines, and MMPs (27, 31, 32). Inflammation also may be driven by cartilage matrix degradation products through activation of innate immune responses. Members of the SLRP family, such as fibromodulin and decorin, may target the classical complement pathway and enhance or inhibit its activation (33). Cartilage oligomeric matrix protein (COMP), on the other hand, is a potent activator of the alternative complement pathway, and complexes of COMP and C3b may be found in OA synovial fluid (34).
Increased age is the strongest risk factor for OA and aging-related changes in cartilage may contribute to the excessive matrix remodeling response. As reviewed recently elsewhere (35), these aging changes include the accumulation of advanced glycation end-products that make the cartilage more “brittle” and the appearance of chondrocytes with features of the senescence-associated secretory phenotype, including increased production of many cytokines, chemokines, and MMPs.
Cell death has been observed during the development of OA, and it may also be related to aging. There is evidence of a loss of cells starting in the superficial zone of cartilage that is associated with an age-related decrease in HMGB-2 (36). Increased production of ROS mediated by mechanical injury or in response to cytokines and matrix fragments may also contribute to cell death (37), as well as a decline in autophagy, which serves as a protective mechanism used by cells under stress (38). Proof of concept that cell death contributes to OA was provided by a study using caspase inhibitors to block cell death, which resulted in decreased severity of cartilage lesions in a rabbit model of posttraumatic OA (39).
Calcification of the articular cartilage and meniscus (chondrocalcinosis), which is often accompanied by the presence of crystals (calcium pyrophosphate dihydrate and/or hydroxyapatite) in the joint, is commonly seen in older adults with knee OA (40). A population-based study noted radiographic evidence of both chondrocalcinosis and knee OA in 18.2% of the population over the age of 65 years, while 6.9% had knee OA without chondrocalcinosis (41). Crystals could play a role in the pathogenesis of OA by stimulating TLRs present on chondrocytes and synovial cells to promote the production of inflammatory mediators (42). Hydroxyapatite crystals may stimulate the production of inflammatory mediators, including IL-1 and IL-18, through activation of the NLRP3 inflammasome (43). Given that hydroxyapatite crystals are common in OA, these studies suggest that targeting the inflammasome may be a novel approach for preventing progression in a subset of people with OA.
Meniscus and ligaments
Pathologic changes in menisci and ligaments are common in people with knee OA. It is well accepted that injury to the meniscus and/or joint ligaments predisposes to the development of OA (2, 3), and magnetic resonance imaging (MRI) studies have revealed changes even in individuals without a known history of joint trauma. Meniscal damage occurs in 63% of adults with symptomatic knee OA (44), and in a longitudinal study, symptomatic subjects with significant meniscal damage had an odds ratio of 7.4 for the development of radiographic knee OA 30 months later (45). Likewise, disruption of the anterior cruciate ligament (ACL) is common in older adults with knee OA. In an MRI study, 22.8% of people with symptomatic knee OA had evidence of complete ACL rupture, but less than half of those gave a history of trauma (46).
The pathologic changes in the menisci in both aging and OA have similarities to changes noted in the articular cartilage, including matrix disruption, fibrillation, cell clusters, calcification, and cell death (47, 48). There is significant correlation between gross morphologic changes of OA in the knee cartilage and those in the menisci from the same joints (48). An increase in vascular penetration accompanied by increased sensory nerve densities has been noted in OA menisci, which may relate to the ability of menisci to serve as a source of pain in knee OA (49).
In addition to ligament injury, varus–valgus laxity, possibly related to changes associated with aging in the ligaments, may play a role in the development of knee OA (50). Degenerative changes are common in ligaments from knee joints removed at the time of joint replacement for OA, particularly in the posterolateral bundle of the ACL, which in one study, was severely affected in 78% of the joints (51). Similar to the meniscus, histologic changes include matrix disruption and collagen fiber disruption. ACL pathology, but not posterior cruciate disruption, was noted to correlate with radiographic severity (52). A recent study of ACLs obtained at autopsy from 65 tissue donors ages 23–92 years found similar changes of collagen fiber disorganization and mucoid degeneration, as well as chondroid metaplasia and calcium deposition (53). These changes were more prevalent with increasing age and correlated with the presence of OA-like changes in the articular cartilage. In donors with grade II–IV cartilage lesions (on a scale of 0–IV) all had some abnormality in the ACL and 24.1% had ACL rupture. Further research on the pathogenesis of OA in these and other soft tissues of the joint will be important in order to know if therapies targeted at the articular cartilage will also target the changes in these tissues.
The structural and functional properties of periarticular bone in OA represent the dynamic adaptation to biomechanical factors and the effects of soluble products generated in the adjacent joint tissues. The effects of mechanical load on bone are embodied in Wolff's hypothesis, which states that the distribution and material properties of bone are determined by the magnitude and direction of the applied load (54). In this paradigm, the changes in subchondral bone volume and density that characterize the osteoarthritic process are reflective of the prior loading history. The effects of loading may produce changes in subchondral bone height and contour, termed “attrition.” Bone remodeling in OA also may be initiated at sites of local bone damage resulting from excessive repetitive loading. This form of microdamage is associated with the appearance of microcracks that initiate targeted remodeling, which likely accounts for the bone marrow lesions observed with MRI in patients with OA (Figure 1). Histologic examination of the lesions reveals local fat necrosis and marrow fibrosis at various stages of healing (55). The correspondence of the bone marrow lesions with regions of bone and cartilage damage strongly supports a primary role of a mechanical and traumatic etiology for the subchondral bone marrow changes.
An additional mechanism for skeletal adaptation occurs at the joint margins and entheseal sites, where new bone is added by endochondral ossification, recapitulating the cellular mechanisms of skeletal growth and development (56). This process gives rise to the formation of osteophytes. Local production of growth factors, including transforming growth factor β and bone morphogenetic protein 2, has been implicated in this process (57, 58). Although there remains controversy regarding their functional role, osteophytes may serve to stabilize the joint rather than contributing to OA progression (56).
The subchondral bone plate beneath the articular cartilage is organized into cortical bone, whereas the deeper zones transition into a network of cancellous bone. The articular cartilage is separated from the subchondral bone by a zone of calcified cartilage, and the interface between the articular and calcified cartilage can be identified by the so-called tidemark (Figure 2). The calcified cartilage undergoes marked alterations in cellular composition and structure in OA (59, 60). This process involves the penetration of calcified cartilage by vascular elements that extend from the subchondral bone and adjacent marrow space recapitulating the vascular invasion of the growth plate that occurs during development. This results in duplication of the tidemark and advancement of the calcified cartilage into the deep zones of the articular cartilage, leading to local cartilage thinning.
Walsh and coworkers (59–61) observed sensory nerve fibers expressing nerve growth factor in the vascular channels associated with osteochondral angiogenesis and speculated that this could be a potential source of symptomatic pain. The regions of vascular invasion were associated with localized bone marrow replacement by fibrovascular tissue expressing vascular endothelial growth factor (VEGF). VEGF expression was also detected in chondrocytes in proximity to the angiogenesis, where VEGF could provide the signals for recruitment of the vascular elements.
The properties of subchondral bone also are influenced by the organization and composition of the organic bone matrix and mineral content (62, 63). The state of bone mineralization is highly dependent on the rate of bone remodeling. When the rate of bone remodeling is high, the “late” phase of mineral accretion is attenuated, leading to a state of relative hypomineralization, which is associated with a reduction in the elastic modulus. In contrast, under conditions of low bone turnover, the continued deposition of mineral leads to an increase in the elastic modulus, and the bone becomes resistant to deformation and more “brittle,” adversely affecting the overlying articular cartilage (63).
The detection of bone changes in OA prior to the appearance of detectable changes in the articular cartilage can be attributed in part to the marked differential capacity of cartilage and bone to adapt to altered mechanical loads and damage. Bone can rapidly alter its architecture and structure via cell-mediated processes of modeling and remodeling. In contrast, the capacity of chondrocytes to repair and modify their surrounding extracellular matrix is relatively limited in comparison to skeletal tissues (64).
Multiple studies have provided insights into the sequential structural changes in subchondral cortical and trabecular bone in OA. Karvonen et al (65) analyzed the bone mineral density of the subchondral trabecular bone in the knee joints of patients with early OA and observed reduced levels deep to the thickened cortical bone. These findings were confirmed by fractal signature analysis, a computerized textural image analysis method (66, 67). The osteoporotic changes in the subchondral trabeculae were speculated to be related to reduced transmission of load from the thickened cortical plate and to represent a form of so-called “stress shielding.” Recent studies on sclerostin and additional components of the Wnt/β-catenin pathway in osteochondral samples have provided potential mechanistic insights into the molecular signals by which mechanical factors modulate subchondral bone remodeling (68). The sclerostin gene encodes the protein sclerostin, which is a potent inhibitor of the Wnt pathway that contributes to the regulation of bone formation. Sclerostin expression in osteocytes was locally decreased in regions of bone sclerosis. Increased mechanical loading in these regions could be responsible for the down-regulation of sclerostin, with the resultant increase in localized bone formation.
Changes in bone volume represent only one of the factors that determine the mechanical properties of bone. Day and coworkers (63) constructed finite element models based on micro–computed tomography scans of subchondral trabecular bone derived from the proximal tibiae of cadaver specimens with early cartilage damage. They found that the volume fraction of trabecular bone was increased, but observed that the tissue modulus of the bone was reduced in the condyles in which there was damage in the overlying articular cartilage. They attributed the reduction in modulus to a decrease in mineral density, which they speculated was related to incomplete mineralization due to an increase in the rate of bone remodeling. These observations indicate that the properties of the subchondral bone in certain stages of OA may be associated with decreased, rather than increased, bone tissue modulus and have significant implications with respect to treatment strategies for targeting subchondral and periarticular bone remodeling in OA. The lack of efficacy of a recent trial of risedronate in reducing progression of the cartilage changes in OA highlights the complexity of the issues surrounding the influences of bone adaptation and its effects on the natural history of OA (69).
Synovial inflammatory infiltrates are identified in many OA patients, although they are generally of lower grade than those observed in RA patients (70) (Figure 2). Recent histologic surveys have demonstrated that synovitis occurs even in early stages of disease (71) and after joint injuries (70, 72), which increase the risk of OA. Specific aspects of synovial inflammation, such as the numbers of infiltrating macrophages, may be higher during early disease (71), but the prevalence of synovitis increases with advancing disease stage (72, 73). The “synovitis” observed in OA and posttraumatic joint disease encompasses a variety of histologic patterns, including infiltration by macrophages and lymphocytes, either diffusely or in perivascular aggregates, which are detected in over 50% of patients with knee OA (74). Lining or villous hyperplasia is common, and fibrosis and cartilage/bone detritus are more typical of advanced-stage disease. Increased vascularity also is seen and may be a target for therapy (49).
Despite lower severity and greater variability in OA-associated synovitis as compared to RA, many groups have reported that low-grade synovitis is associated with disease manifestations. For example, in a prospective study of 422 patients (75), synovitis observed during arthroscopy was associated with progression of cartilage lesions, and those exhibiting synovial inflammation had both more severe baseline chondropathy and more severe progression of cartilage pathology. Although an earlier MRI study in patients with established knee OA (76) failed to confirm these findings, a more recent study of 514 patients with knee pain without radiographic evidence of knee OA provided evidence that effusion and synovitis were associated with the subsequent development of cartilage erosion (77). The presence of synovial effusion on ultrasound imaging also predicted progression to joint replacement (78).
A relationship between synovitis and symptoms was first noted by Torres and colleagues (79), who showed that synovitis, meniscal tears, and bone marrow lesions detected by MRI all correlated with symptoms. Other investigators (76) reported that changes in pain scores over time varied with changes in synovitis, suggesting a causal relationship. We recently observed a relationship between synovitis and knee symptoms, even in patients without radiographic evidence of OA (72). In addition to subjectively measured symptoms, synovitis was recently associated with inferior knee joint function, as measured objectively by walking and stair-climbing times (80).
Soluble mediators of inflammation, including cytokines and chemokines, which can promote synovitis, are increased in the synovial fluid in osteoarthritic and post–joint injury tissues. The most extensively studied are IL-1β and tumor necrosis factor α (TNFα), which can suppress matrix synthesis and promote cartilage catabolism (6). Attempts to block their activity in patients, however, have demonstrated only minimal symptomatic efficacy (81, 82). Therefore, many other mediators of inflammation that can affect synovitis deserve investigation. The perivascular inflammatory infiltrates in OA synovium are largely composed of lymphocytes (71), and the common γ-chain family of cytokines play important roles in the activation, survival, and function of T lymphocyte populations. Of these family members, serum levels of IL-15 have been associated with the incidence and progression of radiographic OA (83), and synovial fluid levels are elevated in early-stage OA (84). IL-17, which is predominantly produced by T lymphocytes, is an additional cytokine implicated in OA pathogenesis. In vitro studies have shown that IL-17 can induce chemokine production by both synovial fibroblasts and chondrocytes, particularly in synergy with IL-1 or TNFα (85), and IL-17 blockade has been demonstrated to decrease synovial thickening and IL-6 levels in a murine meniscectomy model, although no difference in cartilage appearance was noted (86).
Many chemokines are produced in the joint tissues of patients with OA and after joint injury (87). In patients with early-stage OA, synovial chemokine expression was found to be associated with the presence of synovial inflammation (72), and expression of CCL19 and its receptor CCR7 was associated with greater symptoms. Other chemokines, including monocyte chemotactic protein 1 and monocyte inflammatory protein, have also been associated with knee pain levels (88). Certain chemokines, however, may play a positive role. For example, the chemokine stromal cell–derived factor 1 recruits mesenchymal progenitors during tissue repair (89).
How synovitis is triggered in OA, in which evidence of a systemic immune response or infection is lacking, is an area of current investigation. Disruption of the articular extracellular matrix is a hallmark of OA, and molecular products of extracellular matrix catabolism have been linked to inflammation through at least two mechanisms: stimulation of TLRs and activation of the complement cascade. TLRs are implicated in the triggering of cellular inflammation and repair responses to both pathogenic and endogenous “danger” signals produced during infection or during noninfectious tissue injury (90). There are many putative endogenous TLR ligands, some of which are modified in form or concentration in OA. These include matrix components such as tenascin C (91), fibronectin isoforms (27, 28), and fragments of hyaluronic acid (92). TLR activation results in the production of many chemokines (e.g., IL-8 and CCL5) and cytokines (e.g., IL-1, IL-6, and TNF) (90). TLR-4 deficiency was shown to reduce disease severity in a model of inflammatory arthritis (93), but efficacy of specific TLR deficiency or blockade has not yet been reported in models of OA. Recent data implicate TLR-2 and TLR-4 in promoting the catabolism of murine cartilage explants in vitro (19).
Certain matrix components can activate the complement cascade. Fibromodulin (94) and COMP (34) activate the classical and alternative pathways of complement, respectively, while other matrix components act as inhibitors (95). Synovial complement deposition in patients with cartilage degeneration has been reported and may be increased during acute flares (96). Complement components have been identified in OA synovial fluid (97) and in vesicles released from OA cartilage in vitro (98). A recent study in murine models of OA demonstrated that mice deficient in C5 or C6 are partially protected from the development of OA (99).
The modern definition of OA must encompass patient-reported symptoms as well as structural changes within the joint, including not only the remodeling of articular cartilage and neighboring bone, but also the synovial inflammation and damage to ligaments and menisci. Driven by mechanical factors, OA is an active response to injury, rather than a degenerative process. Now that we have started to gain a better understanding of the processes that affect the individual tissues in the OA joint, there is a need to determine the mechanisms of cross-talk and feedback among the tissues that are relevant to disease progression. Which factors released from bone and synovium are promoting cartilage degradation, and which factors released from cartilage drive synovitis and bone remodeling? Is the meniscus caught in the crossfire, or is it also an active contributor to the altered milieu in the joint fluid? Should OA be classified based on the site within the joint where the earliest change is noted (100)? Will targeting one process in one tissue be sufficient to slow or reverse the changes that have occurred in the other tissues?
Answering these and other questions relevant to OA will most likely require a systems approach that can integrate data from the molecular level, including genomics, epigenetics, proteomics, and metabolomics, with data obtained at the structural level, including joint tissue remodeling, as well as biomechanical measures. Advances in imaging and biochemical markers will also be needed in order to address many of these questions, as will a structure-modifying therapy with proven and unequivocal benefit. Although we are not there yet, we are getting much closer to realizing the dream, with a growing list of targets (Table 1) being tested in preclinical and early-phase human studies.
Table 1. Biologic processes and mediators responsible for joint tissue destruction in osteoarthritis and potential therapeutic interventions*
MMPs = matrix metalloproteinases; ILs = interleukins; OSM = oncostatin M; GROα = growth-related oncogene α; MCP-1 = monocyte chemotactic protein 1; TGFα = transforming growth factor α; TIMPs = tissue inhibitor of metalloproteinases; TLR = Toll-like receptor; IGF-1 = insulin-like growth factor 1; BMP-7 = bone morphogenic protein 7; OP-1 = osteogenic protein 1; FGF-18 = fibroblast growth factor 18; IA = intraarticular; HMGB-2 = high mobility group box protein 2; iNOS = inducible nitric oxide synthetase; RUNX-2 = runt-related transcription factor 2; HIF-2α = hypoxia-inducible factor 2α; PTH = parathyroid hormone; VEGF = vascular endothelial growth factor; TNFα = tumor necrosis factor α; MIP-1β = macrophage inflammatory protein 1β.
MMPs 1, 3, 9, and 13, ADAMTS 4 and 5, cathepsin K, serine proteases (HTRA-1) driven by cytokines (ILs 1, 6, 7, 8, 17, and 18, OSM), chemokines (IL-8, GROα, GROγ, RANTES, MCP-1) and others (S100 proteins, TGFα, matrix fragments, leukotrienes, and prostaglandins)