Transforming growth factor-β (TGFβ) is a pleiotropic cytokine that is important in the regulation of joint homeostasis and disease. TGFβ signalling is induced by loading and has an important function in maintaining the differentiated phenotype of articular chondrocytes. Concentrations of active TGFβ differ greatly between healthy and osteoarthritic joints, being low in healthy joints and high in osteoarthritic joints, leading to the activation of different signalling pathways in joint cells. The characteristic pathology of osteoarthritic joints, such as cartilage damage, osteophyte formation and synovial fibrosis, seems to be stimulated or even caused by the high levels of active TGFβ, in combination with altered chondrocyte signalling pathways (which are also observed in ageing joints). In this Review, the changing role of TGFβ in normal joint homeostasis, ageing and osteoarthritis is discussed: TGFβ counteracts pathological changes in a young healthy joint, alters its signalling during ageing and is a driving force of pathology in osteoarthritic joints.
Osteoarthritis (OA) is the most common joint disease, and is increasing in prevalence in the Western world owing to ageing of the population1. The primary characteristics of OA are destruction of articular cartilage, synovial inflammation and fibrosis, osteophyte formation and subchondral bone alterations. OA is progressive and can result in loss of joint function; the ultimate 'therapy' for OA is replacement of the damaged joints by artificial ones. Homeostasis of articular joints is dependent on the proper function and interaction of joint cells. Growth factors, such as the pleiotropic cytokine transforming growth factor-β (TGFβ), are crucial in maintaining joint homeostasis. In this Review, the changing role of TGFβ in normal, ageing and osteoarthritic joints is discussed, along with the possibility of therapeutically targeting TGFβ in OA.
TGFβ and TGFβ signalling pathways
Members of the TGFβ superfamily are involved in most biological processes in mammals. This family consists of more than 35 members and can be divided into two major branches: the TGFβ and activin group, and the bone morphogenetic protein (BMP) and growth and differentiation factor (GDF) group2. TGFβs regulate many of the processes common to cell differentiation, tissue repair and inflammation. In mammals, three TGFβ peptides are present: TGFβ1, TGFβ2 and TGFβ3. The different peptides show a high degree of homology but their expression is differentially regulated at the transcriptional level due to different promoter sequences3. All three peptides are expressed in abundance during development and display overlapping but distinct spatial and temporal expression patterns. TGFβs are synthesized as inactive precursors, consisting of a mature ligand and a latency associated peptide (LAP) that stick together until the mature ligand is liberated. The non-covalent linkage of LAP to the mature ligand prevents receptor binding. Latent TGFβ is bound to the extracellular matrix via latent TGFβ binding proteins (LTBPs)4. Four different LTBPs are known, of which LTBP1 seems to be the most widely expressed. An important mechanism of TGFβ activation is dependent on interactions with cell surface integrins but other mechanisms, such as physical, chemical and enzymatic activation, have also been reported5,6,7.
Advertisement
Active TGFβ is a 25 kD dimer that signals via heteromeric complexes of transmembrane serine/threonine type I and type II receptors, most commonly tetramers composed of two type I and two type II receptors (Fig. 1). The type I receptors, also termed activin receptor-like kinases (ALKs), act downstream of type II receptors and determine receptor specificity8. The canonical signalling route of TGFβ is via the broadly expressed type I receptor ALK5, but in chondrocytes and other cell types additional ALKs, such as ALK1, can be involved9,10,11. Upon type I receptor activation, intracellular signalling is initiated by C-terminal phosphorylation of receptor-regulated (R)-SMAD proteins. Whereas ALK5 stimulates phosphorylation of SMAD2 and SMAD3, ALK1 mediates the activation of SMAD1, SMAD5 and SMAD8 (Refs 11,12). Phosphorylated R-SMADs form heteromeric complexes (trimers) with SMAD4, the common mediator SMAD (co-SMAD); the trimers are made up of two R-SMADs and one SMAD4 (Ref. 13). These heteromeric complexes accumulate in the nucleus where they, together with co-activators and repressors, control transcriptional responses8. Frequently, these two main intracellular SMAD pathways are found to antagonize each other14,15. Both the SMAD2–SMAD3 and SMAD1–SMAD5–SMAD8 signalling cascades have been reported to be involved in chondrogenesis and to regulate chondrocyte differentiation16,17,18. Notably, non-SMAD (non-canonical) pathways can also be activated by ligand-occupied receptors, leading to strengthening or attenuation of downstream cellular responses; non-SMAD pathways have also been described to have a role in chondrocyte biology19,20.
Figure 1
Simplified scheme of TGFβ signalling.
Active transforming growth factor-β (TGFβ) dimers bind to the TGFβ type II receptor (TGFβRII) and recruits activin receptor-like kinase 5 (ALK5) and/or ALK1. This results in phosphorylation of SMAD2 or SMAD3 (via ALK5) or SMAD1, SMAD5 or SMAD8 (via ALK1).The phosphorylated SMADs bind to SMAD4, forming a trimeric complex, and translocate to the nucleus. These complexes accumulate in the nucleus and, together with co-activators and repressors, control transcriptional responses and regulate expression of specific genes.
Articular cartilage is a unique tissue as it not only has a mechanical function but its maintenance actually depends on regular mechanical stimulation. Reduced joint-loading rapidly leads to increased expression of proteolytic enzymes and cartilage degeneration, in both humans and animal models21,22,23,24. Indeed, patients with spinal cord injury show progressive loss of knee cartilage at a faster rate than patients with OA25.
Chondrocytes are mechanosensitive cells that profoundly respond to mechanical loading26. Mechanical stimulation of chondrocytes has been shown to activate TGFβ signalling, as both compressive and tensile stress results in increased SMAD2–SMAD3 signalling27,28. Moreover, TGFβ is stored in a latent form in large amounts in articular cartilage (∼300 ng/ml)29. Not only is TGFβ locally produced and stored in cartilage, but also TGFβ signalling seems to be essential for articular cartilage maintenance. In both humans and mice, an inactivating mutation in SMAD3 results in early-onset cartilage degeneration30,31. In animal models, defective TGFβ signalling, more specifically reduced SMAD3 signalling, has been shown to promote cartilage damage and loss of the differentiated articular chondrocyte phenotype32,33,34.
These data, in combination with the observation that compressive loading rapidly activates TGFβ signalling in cartilage explants35, imply that under normal physiological conditions (regular loading) articular cartilage has a basic level of TGFβ signalling and expression of TGFβ-regulated genes. Adult bovine cartilage stains clearly for phosphorylated SMAD2 shortly after death of the animals (10 min) but a striking decrease in expression is observed by 2 h after death36. Furthermore, expression of TGFβ-regulated genes (PAI1, Smad7 and Alk5) was considerably reduced within 24 h after death of the animals. Reloading of the cartilage samples (after 48 h of being unloaded) rapidly induced staining for both phosphorylated SMAD2 and expression of TGFβ-regulated genes. The effect of reloading could be fully prevented by addition of a specific blocker of ALK4, ALK5 and ALK7 activity. Furthermore, the effects of loading could be fully mimicked by exogenous addition of TGFβ but not activin36.
Advertisement
TGFβ has been shown to stimulate proteoglycan synthesis in (semi)intact immature bovine cartilage and has been shown to be a very strong blocker of chondrocyte terminal differentiation29,33,37. TGFβ could not block proteoglycan loss in adult bovine cartilage, but both addition of exogenous TGFβ and compressive loading blocked the upregulation of Col10a1 gene expression (a well-accepted marker of early chondrocyte hypertrophy) that occurred by non-loaded culturing36.
These observations indicate that activation of TGFβ signalling seems to be the default mode for articular cartilage when the cartilage is regularly loaded, and unloading results in rapid loss of TGFβ signalling and TGFβ-dependent gene expression. Mechanical loading can result in activation of inactive LAP-bound TGFβ, as has been shown in various systems38,39; the released mature TGFβ ligands will bind to chondrocyte TGFβ receptors whereas non-receptor-bound TGFβ will become rapidly unavailable owing to the presence of a vast quantity of TGFβ binding sites in the extracellular matrix. In this model, mechanical loading results in short-term TGFβ activation confined to the articular cartilage. Activation of this system blocks chondrocyte hypertrophy and stimulates the expression of latent TGFβ1 and ALK5 while downregulating the expression of ALK1. In this context it should be noted that ALK1 signalling (via SMAD1, SMAD5 and SMAD8) stimulates chondrocyte hypertrophy whereas ALK5 signalling (via SMAD2 and SMAD3) blocks chondrocyte hypertrophy40,41. In this model (Fig. 2), loading-induced TGFβ signalling is a self-regulatory system that prevents articular chondrocyte terminal differentiation and cartilage degeneration.
Figure 2
Model of loading-induced TGFβ signalling in articular cartilage.
Compressive loading results in the short-term release of active transforming growth factor-β (TGFβ) from latent TGFβ stored in the matrix. Active TGF1 binds to chondrocyte TGFβ receptors. Non-receptor-bound TGFβ becomes quickly unavailable owing to matrix binding. Binding of active TGFβ to activin receptor-like kinase 5 (ALK5) stimulates the expression of latent TGFβ1 and ALK5 while downregulating the expression of ALK1. Activation of this system blocks chondrocyte hypertrophy (Col10a1 expression).
The observation that responses to TGFβ are age-related has been known for several decades42. The responsiveness of equine cartilage to TGFβ decreases with increasing age in animals between 9 months and 20 years43. Furthermore, TGFβ is able to counteract the inhibitory effect of IL-1 on mouse articular cartilage proteoglycan synthesis in young animals but not in old animals44,45. Comparing TGFβ signalling components in young and old mice revealed that TGFβ receptor expression and the expression of phosphorylated SMAD2 was greatly reduced in old animals46.
Studies in old mice have revealed an age-related increase in the ALK1:ALK5 ratio11, an observation that has since been confirmed in other mouse strains and species47,48. Furthermore, Hui et al. showed a switch in the expression of phosphorylated SMAD2–SMAD3 to SMAD1–SMAD5–SMAD8 in old guinea pigs compared with young ones48. Substantially reduced expression of ALK5 but not of ALK1 has been found in old cows compared with young ones49. Moreover, in old animals the response to TGFβ was decreased whereas the response to BMP-9, an ALK1 ligand, was unaltered49.
ALK5 and ALK1 signalling have opposing functional effects in chondrocytes10,12. Furthermore, both reduced ALK5 expression and increased ALK1 activity increase chondrocyte MMP-13 expression11. Additionally, signalling by SMAD2–SMAD3 stimulates metalloproteinase inhibitor 3 (TIMP-3) expression in chondrocytes50,51 The age-related reduction in expression of ALK5, in combination with increased expression of ALK1, might shift the balance to cartilage degradation and could render articular cartilage prone to development of OA (Fig. 3).
Figure 3
Age-related changes in TGFβ signalling in articular cartilage.
In young, healthy cartilage, transforming growth factor-β (TGFβ) signals mainly via activin receptor-like kinase 5 (ALK5), which stimulates SMAD2–SMAD3 phosphorylation; during ageing, a decrease in expression of ALK5, in combination with stable expression of ALK1 (which mediates SMAD1–SMAD5–SMAD8 phosphorylation), shifts the balance to a higher ALK1:ALK5 ratio. This shift leads to induction of chondrocyte hypertrophy and results in articular cartilage that is prone to development of osteoarthritis (OA).
Unloading of cartilage results in associated loss of TGFβ signalling and loss of the block on chondrocyte hypertrophy (see above). However, old bovine cartilage (from cows 6–13 years old) has a strongly reduced capacity for loading-mediated TGFβ signalling compared with cartilage from young cows (6–36 months)52. This reduced responsiveness might be the result of the observed age-related reduction in expression of ALK5 in old cows or changes in the extracellular matrix that make it harder to deform; indeed, perhaps reduced ALK5 expression in old bovine cartilage is an outcome of diminished TGFβ signalling due to a more rigid matrix. However, reduced ALK5 expression is also observed in old mice46 even though increased matrix stiffness with increasing age is less likely in these short-living animals. Whatever the primary cause, an age-related decrease in TGFβ signalling (via phosphorylated SMAD2–SMAD3) is anticipated to impair cartilage homeostasis. The age-related, loading-associated loss of protective SMAD2–SMAD3 signalling makes articular cartilage more vulnerable to OA development in older than in young individuals.
Loss of circadian rhythm
Ageing is associated with loss of circadian rhythm, resulting in internal desynchronization53,54. Data from humans and animal models indicate that disruption of the circadian clock is associated with development of OA55; expression of the core clock transcription factor BMAL1 is impaired in human osteoarthritic cartilage and cartilage from old mice56. Knockout of BMAL1 in mice diminishes the levels of phosphorylated SMAD2–SMAD3 while increasing levels of phosphorylated SMAD1–SMAD5–SMAD8, indicating a connection with TGFβ signalling56. These observations might link the age-related loss of temporal control to a shift in the ALK5:ALK1 ratio and development of OA.
Role of TGFβ in osteoarthritis
Changes in signalling route
Under physiological conditions, levels of active TGFβ in joint tissues are either very low or load-activated and confined to articular cartilage. However, the pathological changes observed in OA suggest that this situation changes when the OA process is initiated (Fig. 4). Unfortunately, most studies have only reported total synovial fluid TGFβ levels, activated by 1N HCl, which makes it difficult to make a statement about active TGFβ levels in situ57,58. However, Fava et al. measured TGFβ activity in the synovial fluid of patients with OA using a radioreceptor and a bioassay59. The mean level of active TGFβ was ∼4 ng/ml, whereas in a patient with avascular necrosis no TGFβ activity could be detected. Acid activation increased TGFβ activity in all synovial fluids tested.
Figure 4
Role of TGFβ in a normal and an osteoarthritic joint.
In a healthy joint, loading-induced transforming growth factor-β (TGFβ) signalling is important in the maintenance of the differentiated chondrocyte phenotype. In an osteoarthritic joint, TGFβ levels are increased and TGFβ receptor expression on chondrocytes is altered. The high TGFβ concentrations stimulate osteophyte formation, synovial fibrosis and subchondral bone sclerosis and remodelling. In addition, the high TGFβ concentrations, in combination with the altered TGFβ receptor expression, push articular chondrocytes towards hypertrophy.
Cells in the joint, such as macrophages and fibroblasts, can produce and activate TGFβ60,61,62. Furthermore, in several organs it has been shown that inflammation results in increased levels of active TGFβ63,64. In mice with experimental knee OA, synovial cells abundantly expressed TGFβ and phosphorylated SMAD2, indicating both TGFβ production and increased TGFβ signalling65. Furthermore, given that articular cartilage contains a reservoir of TGFβ bound to the matrix, cartilage degradation is expected to release TGFβ. The high level of protease activity in osteoarthritic joints further contributes to TGFβ activation66.
Interestingly, the SMAD2–SMAD3 and SMAD1–SMAD5–SMAD8 pathways are differentially activated depending on the active TGFβ concentration present. In human fibroblasts, a relatively high TGFβ concentration (>5 ng/ml) preferentially stimulates SMAD1–SMAD5–SMAD8 phosphorylation whereas low concentrations predominantly stimulate phosphorylated SMAD2–SMAD3 signalling67. A similar pattern has been shown in other cell types such as endothelial cells9. This observation indicates that in a normal joint with low concentrations of active TGFβ, the SMAD2–SMAD3 pathway will preferentially be activated, whereas in a joint with high concentrations the SMAD1–SMAD5–SMAD8 route will be favoured.
Interactions with key signalling pathways
In OA, not only is TGFβ signalling altered but also other signalling pathways are activated as a result of processes such as synovial inflammation. Furthermore, activation of other signalling pathways not only modulates TGFβ signalling, but TGFβ signalling modulates the signalling pathways of other ligands. For example, in chondrocytes from patients with OA, IL-1β has been shown to downregulate TGFβ type II receptor expression and inhibit TGFβ1-induced gene expression and SMAD2–SMAD3 phosphorylation68. Furthermore, in these cells, IL-1 upregulates SMAD7 expression (an intracellular TGFβ signalling inhibitor) via NF-κB activation69. In this way inflammation affects chondrocyte TGFβ signalling.
As described above, the central mediators of intracellular TGFβ signalling are the SMAD proteins. The R-SMADs consist of an MH1 and MH2 domain that are joined by a so-called linker domain. This linker domain can be modified by phosphorylation of specific serine and threonine residues, and these modifications regulate intracellular transport and stability of the SMAD proteins70. Numerous kinases, activated by inflammatory cytokines and growth factors, can phosphorylate the linker region, thus integrating the signals of these ligands with TGFβ signalling71,72,73. However, the role of SMAD-linker modification in OA has been poorly investigated. One study has shown that Wnt signalling modifies TGFβ-induced SMAD signalling in mouse chondrocytes, skewing signalling towards the SMAD1–SMAD5–SMAD8 pathway74. Moreover, in rat chondrocytes ERK1/2 has been shown to activate the SMAD2–SMAD3 signalling pathway, resulting in TGFβ1-induced TIMP-3 upregulation; in human IL-1-stimulated OA cartilage, inhibition of p38 MAPK stimulated TGFβ-induced TIMP1 expression51,75. In the nucleus, interaction of TGFβ signalling with other signalling pathways takes place at the level of transcription factors and chromatin modifications, but these interactions have not been fully investigated in OA yet76,77,78. Undoubtedly TGFβ signalling is integrated with the signalling of other pathways on all levels; however, its role in OA must be further elucidated.
Advertisement
Synovial fibrosis
TGFβ is a well-known profibrotic protein and can even be considered the main driving factor in many fibrotic diseases79. Injection of active TGFβ into mouse knees (three doses of 200 ng TGFβ per knee) results in moderate synovial fibrosis, temporary influx of inflammatory cells, lasting synovitis and extensive osteophyte formation. By contrast, overexpression of active TGFβ using an adenoviral vector leads to profound synovial fibrosis and moderate osteophyte formation80,81. These observations suggest that high peak levels of TGFβ induce osteophyte formation whereas lower but lasting TGFβ levels primarily induce fibrosis; these differences might be related to differential receptor activation by different ligand concentrations.
Inhibition of TGFβ activity in experimental mouse OA using either SMAD7 or LAP substantially inhibited synovial fibrosis82,83. A study investigating the expression of profibrotic genes in human OA synovium, human TGFβ-stimulated fibroblasts and synovium of mice with experimental OA showed that five genes were upregulated in all three conditions (PLOD2, LOX, COL1A1, COL5A1 and TIMP1)84. Moreover, these five genes were also upregulated in vivo in mouse synovium after TGFβ injection. These genes seem to be central to TGFβ-driven matrix deposition in the synovium of osteoarthritic joints. Furthermore, in animal models, elevated PLOD2 expression increases the number of collagen hydroxypyridinoline crosslinks, making the fibrotic tissue more resistant to proteolytic degradation85.
Myofibroblasts can be a source of proteases, and increased expression of MMP-3, MMP-13, ADAMTS-4 and ADAMTS-5 has been found in fibrotic synovial tissue in mice with experimental OA86. Furthermore, it has been suggested that cartilage degradation in OA is attributable to high expression of TGFβ-induced fibrogenic genes (including metalloproteinases) in chondrocytes87. In this perception of events, TGFβ signalling, through ALK5 and SMAD2–SMAD3, drives the transition of chondrocytes and chondroprogenitors to a fibrogenic phenotype, resulting in the destructive processes observed in OA88.
Osteophyte formation
Osteophytes — a typical characteristic of OA — are bony outgrowths at the joint margins of synovial joints. Osteophytes are thought to originate from progenitor cells (residing in the periosteum at the boundary of bone and cartilage) that undergo a process of chondrogenesis and finally endochondral ossification89. Bolus injection of TGFβ in naive mouse knee joints (three doses of 200 ng TGFβ per knee) induced osteophyte formation similar to that observed in experimental OA90. A comparison of osteophyte formation induced by TGFβ to that induced by intra-articular injection of BMP-2 showed that osteophytes induced by BMP-2 were found primarily in the region where the growth plates meet the joint space, whereas those triggered by TGFβ originated from the periosteum at sites distant from the growth plates81. Moreover, chondrocyte-specific, drug-induced overexpression of BMP-2 did not result in osteophyte formation in naive mouse knee joints but boosted experimental OA-dependent osteophyte formation91. These data indicate that, at least in experimental OA, TGFβ and BMPs target different progenitor cell populations92. Moreover, TGFβ is the most likely trigger of osteophyte formation in (experimental) OA but BMPs can greatly enhance oestophyte development when chondrogenesis has been initiated and progenitor cells have entered the chondrogenic differentiation programme.
Advertisement
TGFβ is expressed in developing osteophytes, in both human and experimental OA65,93,94. In experimental OA, specific inhibition of TGFβ activity reduces osteophyte formation82,95. These studies indicate that TGFβ is essential in osteophyte formation in vivo in mouse OA models. To date, it is not clear what the crucial factors in human osteophyte formation are, although TGFβ is a very plausible candidate. Further supporting a role for TGFβ in osteophyte formation, at least in mice, is the observation that the periosteal niche involved in osteophyte formation harbours a unique mesenchymal progenitor population of SCX-positive and SOX9-positive progenitors cells and that TGFβ signalling in these cells is required for the development of bone eminences during development96,97.
Species-specific differences are well known with respect to osteophyte formation. For instance, mice develop osteophytes not only during OA but also during inflammatory arthritis. This observation indicates that either the factors present in human and mouse joints with arthritis are dissimilar or that human mesenchymal stem cells react differently to these factors compared with mouse mesenchymal stem cells. The fact that human OA is characterized by osteophyte formation shows that adult human mesenchymal stem cells retain the ability to undergo chondrogenesis, and that differences in the growth factor and cytokine milieu probably determines whether osteophytes are formed or not.
Subchondral bone changes
Subchondral bone shows pathological alterations in OA, such as sclerosis, cyst formation and accelerated remodelling98,99; some investigators even believe that changes in subchondral bone precede or cause articular cartilage changes100,101. Osteoblasts isolated from subchondral bone in OA have increased TGFβ expression, and increased activity of TGFβ is observed in subchondral bone in OA94,102. This enhanced TGFβ activity in subchondral bone is thought to increase the severity of OA. Zhen et al. showed that inhibition of TGFβ activity in subchondral bone attenuated the degeneration of articular cartilage in experimental OA, and that knockout of the TGFβ type II receptor in nestin-positive mesenchymal stem cells resulted in less-severe OA in mice102. Furthermore, transgenic mice that overexpress TGFβ specifically in osteoblasts, and that have high levels of active TGFβ in bone marrow, develop mandibular condyle cartilage degradation103. In addition, systemic neutralization of excessive TGFβ prevented subchondral bone microarchitecture changes and OA progression in experimental OA (cruciate ligament transection model in mice)104. In Dunkin-Hartley guinea pigs, which develop OA spontaneously at a relatively young age, changes in subchondral bone occurred in parallel with increasing levels of active TGFβ, which implies that excessive bone formation is linked to TGFβ activity47. These findings indicate that OA is characterized by enhanced subchondral bone TGFβ activity, resulting in excessive bone remodelling.
TGFβ as a therapy
In a healthy joint, extra TGFβ can help maintain preservation. However, this action will only be effective when TGFβ is targeted to the cartilage and when its active levels are relatively low. In cartilage, low levels of TGFβ help to maintain the differentiated chondrocyte phenotype but this mechanism is only effective when the actual chondrocyte TGFβ receptor expression pattern supports inhibition of hypertrophy (via ALK5 and phosphorylated SMAD2–SMAD3). If this is not the case, or if the whole joint is exposed to relatively high TGFβ concentrations, this will lead to synovial fibrosis, osteophyte formation and can even push articular chondrocytes on the track of hypertrophy.
Systemic inhibition of TGFβ could be an option to block pathology in osteoarthritic joints, especially in joints in which subchondral bone changes are the leading characteristic. However, systemic inhibition will also result in inhibition of TGFβ activity in healthy cartilage in non-osteoarthritic joints (if the inhibitor is able to penetrate cartilage), and might lead to loss of the hypertrophy block in non-osteoarthritic chondrocytes (Table 1). Moreover, TGFβ has a role in inhibition of inflammation, and general blockade of TGFβ could result in a progressive inflammatory reaction, as observed in TGFβ1-deficient mice105. Joint-specific inhibition of TGFβ activity could be an alternative therapy to block OA pathology. However, in most joints with OA, cartilage damage starts as a focal process and inhibition of TGFβ might have a harmful effect on the still relatively normal cartilage.
Table 1
Expected effects of TGFβ modulation in a young, old and osteoarthritic joint
This table shows that the effect of TGFβ modulation depends on the state of the joint (young or old, healthy or diseased) and the specific target tissue. The differential effects mean TGFβ modulation is not a simple therapeutic option, and modulation of OA in experimental models can have variable results depending on the type and course of the model and the approach of TGFβ modulation. OA, osteoarthritis; TGFβ, transforming growth factor-β.
Cartilage defects have been treated by TGFβ supplementation in in vivo experimental models. Injection of TGFβ-overexpressing chondrocytes, fibroblasts and mesenchymal stem cells accelerated repair of cartilage defects in rabbits; in sheep, genetic modification of blood clots with an adenoviral vector encoding TGFβ1 had a similar outcome106,107,108,109. Experimental OA has been treated by administration of extra TGFβ as well as by TGFβ inhibition. Overexpression of TGFβ1 in the joints of rabbits with experimental OA by intra-articular TGFβ1 transfection led to substantial reduction of cartilage matrix degradation110. By contrast, systemic application of a TGFβ type I receptor kinase inhibitor in a mouse model of post-traumatic OA attenuated cartilage damage102. The inhibitor used in this study was not ALK5-specific, rendering interpretation of the findings more complicated, but the results were confirmed by application of an anti-TGFβ antibody102. Furthermore, systemic treatment of mice with the TGFβ type II receptor inhibitor losartan or a TGFβ-neutralizing antibody delayed the progression of articular cartilage degeneration in experimental post-traumatic OA104,111.
In light of the different roles TGFβ can have in healthy or diseased joints, it is not surprising that modulation of TGFβ activity can have disparate effects on OA severity and progression. Depending on the main target tissue, residual TGFβ activity and disease stage, altering TGFβ activity will have variable effects on disease outcome. To be able to target TGFβ-dependent pathways more predictably, it is of utmost importance to elucidate in detail the different intracellular signalling pathways and to discriminate the deleterious and protective effects of TGFβ on joint tissues and the body as a whole. Only when this information is available can TGFβ-based OA therapy be developed as an effective and safe treatment.
Conclusions
TGFβ is a peptide with many different and varied actions. This peptide is thought to provide a selective advantage during evolution, but it can also function as a factor that drives pathology. For example, during oncogenesis112,TGFβ changes from a factor that inhibits tumour growth to a factor that promotes tumour progression. A similar pattern can be seen in OA, as TGFβ alters from a protective, homeostasis-maintaining factor in a healthy young joint to a pathogenic factor in an (old) osteoarthritic joint. In a young, healthy joint TGFβ dampens inflammation and maintains the differentiated chondrocyte phenotype, whereas in an osteoarthritic joint it seems to add to loss of the differentiated chondrocyte phenotype and accelerates cartilage damage, and induces synovial fibrosis and osteophyte formation. Furthermore, data suggest that TGFβ might even contribute to joint pain by inducing β-nerve growth factor (β-NGF) expression, although TGFβ itself has also been shown to be able to act as a suppressor of pain113,114,115,116. This shifting role of TGFβ has the consequence that targeting TGFβ as a therapy for OA is not a simple option; blocking TGFβ or TGFβ supplementation could have either deleterious or positive effects depending on the target tissue and disease state.
Competing interests
The author declares no competing financial interests.
Johnson, V. L. & Hunter, D. J. The epidemiology of osteoarthritis. Best Pract. Res. Clin. Rheumatol.28, 5–15 (2014).PubMed
2.
de Caestecker, M. The transforming growth factor-β superfamily of receptors. Cytokine Growth Factor Rev.15, 1–11 (2004).PubMed
3.
Govinden, R. & Bhoola, K. D. Genealogy, expression, and cellular function of transforming growth factor-β. Pharmacol. Ther.98, 257–265 (2003).PubMed
4.
Unsold, C., Hyytiainen, M., Bruckner-Tuderman, L. & Keski-Oja, J. Latent TGF-β binding protein LTBP-1 contains three potential extracellular matrix interacting domains. J. Cell Sci.114, 187–197 (2001).PubMed
5.
Sheppard, D. Integrin-mediated activation of latent transforming growth factor β. Cancer Metastasis Rev.24, 395–402 (2005).PubMed
6.
Munger, J. S. et al. Latent transforming growth factor-β: structural features and mechanisms of activation. Kidney Int.51, 1376–1382 (1997).PubMed
7.
Wipff, P. J. & Hinz, B. Integrins and the activation of latent transforming growth factor β1 — an intimate relationship. Eur. J. Cell Biol.87, 601–615 (2008).PubMed
8.
Heldin, C. H., Miyazono, K. & ten Dijke, P. TGF-β signalling from cell membrane to nucleus through SMAD proteins. Nature390, 465–471 (1997).PubMed
9.
Goumans, M. J. et al. Balancing the activation state of the endothelium via two distinct TGF-β type I receptors. EMBO J.21, 1743–1753 (2002).PubMedPubMedCentral
10.
Finnson, K. W., Parker, W. L., ten Dijke, P., Thorikay, M. & Philip, A. ALK1 opposes ALK5/Smad3 signaling and expression of extracellular matrix components in human chondrocytes. J. Bone Miner. Res.23, 896–906 (2008).PubMed
11.
Blaney Davidson, E. N. et al. Increase in ALK1/ALK5 ratio as a cause for elevated MMP-13 expression in osteoarthritis in humans and mice. J. Immunol.182, 7937–7945 (2009).PubMed
12.
Finnson, K. W. et al. Endoglin differentially regulates TGF-β-induced Smad2/3 and Smad1/5 signalling and its expression correlates with extracellular matrix production and cellular differentiation state in human chondrocytes. Osteoarthritis Cartilage18, 1518–1527 (2010).PubMed
13.
Chacko, B. M. et al. The L3 loop and C-terminal phosphorylation jointly define Smad protein trimerization. Nat. Struct. Biol.8, 248–253 (2001).PubMed
14.
Goumans, M. J. et al. Activin receptor-like kinase (ALK)1 is an antagonistic mediator of lateral TGFβ/ALK5 signaling. Mol. Cell12, 817–828 (2003).PubMed
15.
Daly, A. C., Randall, R. A. & Hill, C. S. Transforming growth factor β-induced Smad1/5 phosphorylation in epithelial cells is mediated by novel receptor complexes and is essential for anchorage-independent growth. Mol. Cell. Biol.28, 6889–6902 (2008).PubMedPubMedCentral
16.
Hellingman, C. A. et al. Smad signaling determines chondrogenic differentiation of bone-marrow-derived mesenchymal stem cells: inhibition of Smad1/5/8P prevents terminal differentiation and calcification. Tissue Eng. Part A17, 1157–1167 (2011).PubMed
17.
Keller, B. et al. Interaction of TGFβ and BMP signaling pathways during chondrogenesis. PLoS ONE6, e16421 (2011).PubMedPubMedCentral
18.
de Kroon, L. M. et al. Activin receptor-like kinase receptors ALK5 and ALK1 are both required for TGFβ-induced chondrogenic differentiation of human bone marrow-derived mesenchymal stem cells. PLoS ONE10, e0146124 (2015).PubMedPubMedCentral
19.
Zhang, Y. E. Non-Smad pathways in TGF-β signaling. Cell Res.19, 128–139 (2009).PubMed
20.
Gao, L. et al. TAK1 regulates SOX9 expression in chondrocytes and is essential for postnatal development of the growth plate and articular cartilages. J. Cell Sci.126, 5704–5713 (2013).PubMedPubMedCentral
21.
Jurvelin, J., Kiviranta, I., Tammi, M. & Helminen, J. H. Softening of canine articular cartilage after immobilization of the knee joint. Clin. Orthop. Relat. Res.207, 246–252 (1986).
22.
Palmoski, M. J., Colyer, R. A. & Brandt, K. D. Joint motion in the absence of normal loading does not maintain normal articular cartilage. Arthritis Rheum.23, 325–334 (1980).PubMed
23.
Hinterwimmer, S. et al. Cartilage atrophy in the knees of patients after seven weeks of partial load bearing. Arthritis Rheum.50, 2516–2520 (2004).PubMed
24.
Ando, A. et al. Increased expression of metalloproteinase-8 and -13 on articular cartilage in a rat immobilized knee model. Tohoku J. Exp. Med.217, 271–278 (2009).PubMed
25.
Vanwanseele, B., Eckstein, F., Knecht, H., Stussi, E. & Spaepen, A. Knee cartilage of spinal cord-injured patients displays progressive thinning in the absence of normal joint loading and movement. Arthritis Rheum.46, 2073–2078 (2002).PubMed
26.
Sanchez-Adams, J., Leddy, H. A., McNulty, A. L., O'Conor, C. J. & Guilak, F. The mechanobiology of articular cartilage: bearing the burden of osteoarthritis. Curr. Rheumatol. Rep.16, 451 (2014).PubMedPubMedCentral
27.
Bougault, C. et al. Dynamic compression of chondrocyte-agarose constructs reveals new candidate mechanosensitive genes. PLoS ONE7, e36964 (2012).PubMedPubMedCentral
28.
Furumatsu, T. et al. Tensile strain increases expression of CCN2 and COL2A1 by activating TGF-β-Smad2/3 pathway in chondrocytic cells. J. Biomech.46, 1508–1515 (2013).PubMed
29.
Morales, T. I., Joyce, M. E., Sobel, M. E., Danielpour, D. & Roberts, A. B. Transforming growth factor-β in calf articular cartilage organ cultures: synthesis and distribution. Arch. Biochem. Biophys.288, 397–405 (1991).PubMed
30.
Yao, J. Y. et al. Mutation analysis of the Smad3 gene in human osteoarthritis. Eur. J. Hum. Genet.11, 714–717 (2003).PubMed
31.
van de Laar, I. M. et al. Mutations in SMAD3 cause a syndromic form of aortic aneurysms and dissections with early-onset osteoarthritis. Nat. Genet.43, 121–126 (2011).PubMed
32.
Wu, Q. et al. Induction of an osteoarthritis-like phenotype and degradation of phosphorylated Smad3 by Smurf2 in transgenic mice. Arthritis Rheum.58, 3132–3144 (2008).PubMedPubMedCentral
33.
Yang, X. et al. TGF-β/Smad3 signals repress chondrocyte hypertrophic differentiation and are required for maintaining articular cartilage. J. Cell Biol.153, 35–46 (2001).PubMedPubMedCentral
34.
Serra, R. et al. Expression of a truncated, kinase-defective TGF-β type II receptor in mouse skeletal tissue promotes terminal chondrocyte differentiation and osteoarthritis. J. Cell Biol.139, 541–552 (1997).PubMedPubMedCentral
35.
Madej, W., van Caam, A., Blaney Davidson, E. N., van der Kraan, P. M. & Buma, P. Physiological and excessive mechanical compression of articular cartilage activates Smad2/3P signaling. Osteoarthritis Cartilage22, 1018–1025 (2014).PubMed
36.
Madej, W., van Caam, A., Blaney Davidson, E., Buma, P. & van der Kraan, P. M. Unloading results in rapid loss of TGFβ signaling in articular cartilage: role of loading-induced TGFβ signaling in maintenance of articular chondrocyte phenotype? Osteoarthritis Cartilage24, 1807–1815 (2016).PubMed
37.
Ballock, R. T. et al. TGF-β1 prevents hypertrophy of epiphyseal chondrocytes: regulation of gene expression for cartilage matrix proteins and metalloproteases. Dev. Biol.158, 414–429 (1993).PubMed
38.
Albro, M. B. et al. Shearing of synovial fluid activates latent TGF-β. Osteoarthritis Cartilage20, 1374–1382 (2012).PubMedPubMedCentral
39.
Hinz, B. The extracellular matrix and transforming growth factor-β1: tale of a strained relationship. Matrix Biol.47, 54–65 (2015).PubMed
40.
Ferguson, C. M. et al. Smad2 and 3 mediate transforming growth factor-β1-induced inhibition of chondrocyte maturation. Endocrinology141, 4728–4735 (2000).PubMed
41.
Li, T. F. et al. Smad3-deficient chondrocytes have enhanced BMP signaling and accelerated differentiation. J. Bone Miner. Res.21, 4–16 (2006).PubMed
42.
Deie, M. et al. The effects of age on rabbit MCL fibroblast matrix synthesis in response to TGF-β1 or EGF. Mech. Ageing Dev.97, 121–130 (1997).PubMed
43.
Iqbal, J., Dudhia, J., Bird, J. L. & Bayliss, M. T. Age-related effects of TGF-β on proteoglycan synthesis in equine articular cartilage. Biochem. Biophys. Res. Commun.274, 467–471 (2000).PubMed
44.
van Beuningen, H. M., van der Kraan, P. M., Arntz, O. J. & van den Berg, W. B. In vivo protection against interleukin-1-induced articular cartilage damage by transforming growth factor-β1: age-related differences. Ann. Rheum. Dis.53, 593–600 (1994).PubMedPubMedCentral
45.
Scharstuhl, A., van Beuningen, H. M., Vitters, E. L., van der Kraan, P. M. & van den Berg, W. B. Loss of transforming growth factor counteraction on interleukin 1 mediated effects in cartilage of old mice. Ann. Rheum. Dis.61, 1095–1098 (2002).PubMedPubMedCentral
46.
Blaney Davidson, E. N., Scharstuhl, A., Vitters, E. L., van der Kraan, P. M. & van den Berg, W. B. Reduced transforming growth factor-β signaling in cartilage of old mice: role in impaired repair capacity. Arthritis Res. Ther.7, R1338–R1347 (2005).PubMedPubMedCentral
47.
Zhao, W. et al. Cartilage degeneration and excessive subchondral bone formation in spontaneous osteoarthritis involves altered TGF-β signaling. J. Orthop. Res.34, 763–770 (2016).PubMed
48.
Hui, W. et al. Oxidative changes and signalling pathways are pivotal in initiating age-related changes in articular cartilage. Ann. Rheum. Dis.75, 449–458 (2016).PubMed
49.
van Caam, A. et al. Expression of TGFβ-family signalling components in ageing cartilage: age-related loss of TGFβ and BMP receptors. Osteoarthritis Cartilage24, 1235–1245 (2016).PubMed
50.
Qureshi, H. Y., Ricci, G. & Zafarullah, M. Smad signaling pathway is a pivotal component of tissue inhibitor of metalloproteinases-3 regulation by transforming growth factor β in human chondrocytes. Biochim. Biophys. Acta1783, 1605–1612 (2008).PubMed
51.
Wang, X. et al. Interaction of ERK1/2 and Smad2/3 signaling pathways in TGF-β1-induced TIMP-3 expression in rat chondrocytes. Arch. Biochem. Biophys.564, 229–236 (2014).PubMed
52.
Madej, W. et al. Ageing is associated with reduction of mechanically-induced activation of Smad2/3P signaling in articular cartilage. Osteoarthritis Cartilage24, 146–157 (2016).PubMed
53.
Gubin, D. G., Weinert, D. & Bolotnova, T. V. Age-dependent changes of the temporal order — causes and treatment. Curr. Aging Sci.9, 14–25 (2016).PubMed
54.
Weinert, D. Age-dependent changes of the circadian system. Chronobiol. Int.17, 261–283 (2000).PubMed
55.
Gossan, N., Boot-Handford, R. & Meng, Q. J. Ageing and osteoarthritis: a circadian rhythm connection. Biogerontology16, 209–219 (2015).PubMed
56.
Dudek, M. et al. The chondrocyte clock gene Bmal1 controls cartilage homeostasis and integrity. J. Clin. Invest.126, 365–376 (2016).PubMed
57.
Jiang, Q., Qiu, Y. T., Chen, M. J., Zhang, Z. Y. & Yang, C. Synovial TGF-β1 and MMP-3 levels and their correlation with the progression of temporomandibular joint osteoarthritis combined with disc displacement: a preliminary study. Biomed. Rep.1, 218–222 (2013).PubMed
58.
Wei, X. & Messner, K. Age- and injury-dependent concentrations of transforming growth factor-β1 and proteoglycan fragments in rabbit knee joint fluid. Osteoarthritis Cartilage6, 10–18 (1998).PubMed
59.
Fava, R., Olsen, N., Keski-Oja, J., Moses, H. & Pincus, T. Active and latent forms of transforming growth factor β activity in synovial effusions. J. Exp. Med.169, 291–296 (1989).PubMed
60.
Metukuri, M. R. et al. Activation of latent transforming growth factor-β1 by nitric oxide in macrophages: role of soluble guanylate cyclase and MAP kinases. Wound Repair Regen.17, 578–588 (2009).PubMed
61.
Lee, J. K., Lee, Y. R., Lee, Y. H., Kim, K. & Lee, C. K. Production of TGF-β1 as a mechanism for defective antigen-presenting cell function of macrophages generated in vitro with M-CSF. Immune Netw.9, 27–33 (2009).PubMedPubMedCentral
62.
McGowan, S. E. Influences of endogenous and exogenous TGF-β on elastin in rat lung fibroblasts and aortic smooth muscle cells. Am. J. Physiol.263, L257–L263 (1992).PubMed
63.
Torrego, A., Hew, M., Oates, T., Sukkar, M. & Fan Chung, K. Expression and activation of TGF-β isoforms in acute allergen-induced remodelling in asthma. Thorax62, 307–313 (2007).PubMedPubMedCentral
64.
Travis, M. A. & Sheppard, D. TGF-β activation and function in immunity. Annu. Rev. Immunol.32, 51–82 (2014).PubMed
65.
Blaney Davidson, E. N., Vitters, E. L., van der Kraan, P. M. & van den Berg, W. B. Expression of transforming growth factor-β (TGFβ) and the TGFβ signalling molecule SMAD-2P in spontaneous and instability-induced osteoarthritis: role in cartilage degradation, chondrogenesis and osteophyte formation. Ann. Rheum. Dis.65, 1414–1421 (2006).PubMed
66.
Koli, K., Saharinen, J., Hyytiainen, M., Penttinen, C. & Keski-Oja, J. Latency, activation, and binding proteins of TGF-β. Microsc. Res. Tech.52, 354–362 (2001).PubMed
67.
Remst, D. F. et al. TGF-β induces Lysyl hydroxylase 2b in human synovial osteoarthritic fibroblasts through ALK5 signaling. Cell Tissue Res.355, 163–171 (2014).PubMed
68.
Bauge, C. et al. Interleukin-1β impairment of transforming growth factor β1 signaling by down-regulation of transforming growth factor β receptor type II and up-regulation of Smad7 in human articular chondrocytes. Arthritis Rheum.56, 3020–3032 (2007).PubMed
69.
Bauge, C. et al. Interleukin-1β up-regulation of Smad7 via NF-κB activation in human chondrocytes. Arthritis Rheum.58, 221–226 (2008).PubMed
70.
Matsuzaki, K. Smad phospho-isoforms direct context-dependent TGF-β signaling. Cytokine Growth Factor Rev.24, 385–399 (2013).PubMed
71.
Yu, J. S. et al. PI3K/mTORC2 regulates TGF-β/Activin signalling by modulating Smad2/3 activity via linker phosphorylation. Nat. Commun.6, 7212 (2015).PubMed
72.
Yumoto, K. et al. TGF-β-activated kinase 1 (Tak1) mediates agonist-induced Smad activation and linker region phosphorylation in embryonic craniofacial neural crest-derived cells. J. Biol. Chem.288, 13467–13480 (2013).PubMedPubMedCentral
73.
Kamato, D. et al. Transforming growth factor-β signalling: role and consequences of Smad linker region phosphorylation. Cell Signal.25, 2017–2024 (2013).PubMed
74.
van den Bosch, M. H. et al. Canonical Wnt signaling skews TGF-β signaling in chondrocytes towards signaling via ALK1 and Smad 1/5/8. Cell Signal.26, 951–958 (2014).PubMed
75.
Studer, R. K. & Chu, C. R. p38 MAPK and COX2 inhibition modulate human chondrocyte response to TGF-β. J. Orthop. Res.23, 454–461 (2005).PubMed
76.
Guo, P., Zhao, K. W., Dong, X. Y., Sun, X. & Dong, J. T. Acetylation of KLF5 alters the assembly of p15 transcription factors in transforming growth factor-β-mediated induction in epithelial cells. J. Biol. Chem.284, 18184–18193 (2009).PubMedPubMedCentral
77.
Wilkinson, D. S. et al. A direct intersection between p53 and transforming growth factor β pathways targets chromatin modification and transcription repression of the α-fetoprotein gene. Mol. Cell. Biol.25, 1200–1212 (2005).PubMedPubMedCentral
78.
Costamagna, E., Garcia, B. & Santisteban, P. The functional interaction between the paired domain transcription factor Pax8 and Smad3 is involved in transforming growth factor-β repression of the sodium/iodide symporter gene. J. Biol. Chem.279, 3439–3446 (2004).PubMed
79.
Meng, X. M., Nikolic-Paterson, D. J. & Lan, H. Y. TGF-β: the master regulator of fibrosis. Nat. Rev. Nephrol.12, 325–338 (2016).PubMed
80.
Bakker, A. C. et al. Overexpression of active TGF-β-1 in the murine knee joint: evidence for synovial-layer-dependent chondro-osteophyte formation. Osteoarthritis Cartilage9, 128–136 (2001).PubMed
81.
van Beuningen, H. M., Glansbeek, H. L., van der Kraan, P. M. & van den Berg, W. B. Differential effects of local application of BMP-2 or TGF-β 1 on both articular cartilage composition and osteophyte formation. Osteoarthritis Cartilage6, 306–317 (1998).PubMed
82.
Scharstuhl, A., Vitters, E. L., van der Kraan, P. M. & van den Berg, W. B. Reduction of osteophyte formation and synovial thickening by adenoviral overexpression of transforming growth factor β/bone morphogenetic protein inhibitors during experimental osteoarthritis. Arthritis Rheum.48, 3442–3451 (2003).PubMed
83.
Blaney Davidson, E. N., Vitters, E. L., van den Berg, W. B. & van der Kraan, P. M. TGF β-induced cartilage repair is maintained but fibrosis is blocked in the presence of Smad7. Arthritis Res. Ther.8, R65 (2006).PubMedPubMedCentral
84.
Remst, D. F. et al. Gene expression analysis of murine and human osteoarthritis synovium reveals elevation of transforming growth factor β-responsive genes in osteoarthritis-related fibrosis. Arthritis Rheumatol.66, 647–656 (2014).PubMed
85.
Remst, D. F. et al. Osteoarthritis-related fibrosis is associated with both elevated pyridinoline cross-link formation and lysyl hydroxylase 2b expression. Osteoarthritis Cartilage21, 157–164 (2013).PubMed
86.
Blaney Davidson, E. N. et al. Connective tissue growth factor/CCN2 overexpression in mouse synovial lining results in transient fibrosis and cartilage damage. Arthritis Rheum.54, 1653–1661 (2006).PubMed
87.
Li, J. et al. Hyaluronan injection in murine osteoarthritis prevents TGFβ1-induced synovial neovascularization and fibrosis and maintains articular cartilage integrity by a CD44-dependent mechanism. Arthritis Res. Ther.14, R151 (2012).PubMedPubMedCentral
88.
Plaas, A. et al. The relationship between fibrogenic TGFβ1 signaling in the joint and cartilage degradation in post-injury osteoarthritis. Osteoarthritis Cartilage19, 1081–1090 (2011).PubMed
89.
van der Kraan, P. M. & van den Berg, W. B. Osteophytes: relevance and biology. Osteoarthritis Cartilage15, 237–244 (2007).PubMed
90.
van Beuningen, H. M., van der Kraan, P. M., Arntz, O. J. & van den Berg, W. B. Transforming growth factor-β1 stimulates articular chondrocyte proteoglycan synthesis and induces osteophyte formation in the murine knee joint. Lab. Invest.71, 279–290 (1994).PubMed
91.
Blaney Davidson, E. N. et al. Inducible chondrocyte-specific overexpression of BMP2 in young mice results in severe aggravation of osteophyte formation in experimental OA without altering cartilage damage. Ann. Rheum. Dis.74, 1257–1264 (2015).PubMed
92.
Blaney Davidson, E. N. et al. Resemblance of osteophytes in experimental osteoarthritis to transforming growth factor β-induced osteophytes: limited role of bone morphogenetic protein in early osteoarthritic osteophyte formation. Arthritis Rheum.56, 4065–4073 (2007).PubMed
93.
Uchino, M. et al. Growth factor expression in the osteophytes of the human femoral head in osteoarthritis. Clin. Orthop. Relat. Res.377, 119–125 (2000).
94.
Sakao, K. et al. Asporin and transforming growth factor-β gene expression in osteoblasts from subchondral bone and osteophytes in osteoarthritis. J. Orthop. Sci.14, 738–747 (2009).PubMed
95.
Scharstuhl, A. et al. Inhibition of endogenous TGF-β during experimental osteoarthritis prevents osteophyte formation and impairs cartilage repair. J. Immunol.169, 507–514 (2002).PubMed
96.
Blitz, E., Sharir, A., Akiyama, H. & Zelzer, E. Tendon-bone attachment unit is formed modularly by a distinct pool of Scx- and Sox9-positive progenitors. Development140, 2680–2690 (2013).PubMed
97.
Li, T. et al. Joint TGF-β type II receptor-expressing cells: ontogeny and characterization as joint progenitors. Stem Cells Dev.22, 1342–1359 (2013).PubMed
98.
Moskowitz, R. W. Bone remodeling in osteoarthritis: subchondral and osteophytic responses. Osteoarthritis Cartilage7, 323–324 (1999).PubMed
99.
Yusup, A. et al. Bone marrow lesions, subchondral bone cysts and subchondral bone attrition are associated with histological synovitis in patients with end-stage knee osteoarthritis: a cross-sectional study. Osteoarthritis Cartilage23, 1858–1864 (2015).PubMed
100.
Quasnichka, H. L., Anderson-MacKenzie, J. M. & Bailey, A. J. Subchondral bone and ligament changes precede cartilage degradation in guinea pig osteoarthritis. Biorheology43, 389–397 (2006).PubMed
101.
Lajeunesse, D., Massicotte, F., Pelletier, J. P. & Martel-Pelletier, J. Subchondral bone sclerosis in osteoarthritis: not just an innocent bystander. Mod. Rheumatol.13, 7–14 (2003).PubMed
102.
Zhen, G. et al. Inhibition of TGF-β signaling in mesenchymal stem cells of subchondral bone attenuates osteoarthritis. Nat. Med.19, 704–712 (2013).PubMedPubMedCentral
103.
Jiao, K. et al. Overexpressed TGF-β in subchondral bone leads to mandibular condyle degradation. J. Dent. Res.93, 140–147 (2014).PubMed
104.
Xie, L. et al. Systemic neutralization of TGF-β attenuates osteoarthritis. Ann. NY Acad. Sci.1376, 53–64 (2016).PubMed
105.
Letterio, J. J. et al. Autoimmunity associated with TGF-β1-deficiency in mice is dependent on MHC class II antigen expression. J. Clin. Invest.98, 2109–2119 (1996).PubMedPubMedCentral
106.
Lee, K. H. et al. Regeneration of hyaline cartilage by cell-mediated gene therapy using transforming growth factor β1-producing fibroblasts. Hum. Gene Ther.12, 1805–1813 (2001).PubMed
107.
Song, S. U. et al. Hyaline cartilage regeneration using mixed human chondrocytes and transforming growth factor-β1- producing chondrocytes. Tissue Eng.11, 1516–1526 (2005).PubMed
108.
Guo, X. et al. Repair of full-thickness articular cartilage defects by cultured mesenchymal stem cells transfected with the transforming growth factor β1 gene. Biomed. Mater.1, 206–215 (2006).PubMed
109.
Ivkovic, A. et al. Articular cartilage repair by genetically modified bone marrow aspirate in sheep. Gene Ther.17, 779–789 (2010).PubMed
110.
Zhang, P., Zhong, Z. H., Yu, H. T. & Liu, B. Exogenous expression of IL-1Ra and TGF-β1 promotes in vivo repair in experimental rabbit osteoarthritis. Scand. J. Rheumatol.44, 404–411 (2015).PubMed
111.
Chen, R. et al. Attenuation of the progression of articular cartilage degeneration by inhibition of TGF-β1 signaling in a mouse model of osteoarthritis. Am. J. Pathol.185, 2875–2885 (2015).PubMedPubMedCentral
112.
Principe, D. R. et al. TGF-β: duality of function between tumor prevention and carcinogenesis. J. Natl Cancer Inst.106, djt369 (2014).PubMedPubMedCentral
113.
Blaney Davidson, E. N. et al. TGF-β is a potent inducer of Nerve Growth Factor in articular cartilage via the ALK5-Smad2/3 pathway. Potential role in OA related pain? Osteoarthritis Cartilage23, 478–486 (2015).PubMed
114.
Chen, G., Park, C. K., Xie, R. G. & Ji, R. R. Intrathecal bone marrow stromal cells inhibit neuropathic pain via TGF-β secretion. J. Clin. Invest.125, 3226–3240 (2015).PubMedPubMedCentral
115.
Lantero, A. et al. TGF-β and opioid receptor signaling crosstalk results in improvement of endogenous and exogenous opioid analgesia under pathological pain conditions. J. Neurosci.34, 5385–5395 (2014).PubMedPubMedCentral
116.
Chen, N. F. et al. TGF-β1 attenuates spinal neuroinflammation and the excitatory amino acid system in rats with neuropathic pain. J. Pain14, 1671–1685 (2013).PubMed