Explaining the fibroblast growth factor paradox in osteoarthritis: Lessons from conditional knockout mice
Article first published online: 28 NOV 2012
Copyright © 2012 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 64, Issue 12, pages 3835–3838, December 2012
How to Cite
Vincent, T. L. (2012), Explaining the fibroblast growth factor paradox in osteoarthritis: Lessons from conditional knockout mice. Arthritis & Rheumatism, 64: 3835–3838. doi: 10.1002/art.34648
- Issue published online: 28 NOV 2012
- Article first published online: 28 NOV 2012
- Accepted manuscript online: 25 JUL 2012 09:53AM EST
- Manuscript Accepted: 19 JUL 2012
- Manuscript Received: 28 JUN 2012
Fibroblast growth factors (FGFs) have long been regarded as important homeostatic factors within joint tissues, but the literature has been plagued by inconsistent and often contradictory findings in vitro. Recent observations have indicated that different FGF receptors (FGFRs) may mediate opposing effects of FGF-2 in chondrocytes, which provides an attractive and testable explanation for the diverse reported effects of this growth factor in articular cartilage and on the development of osteoarthritis (OA) (1).
FGFs constitute a family of pleiotropic growth factors with diverse developmental and postnatal functions. They bind to 1 of 4 FGFRs that are alternatively spliced, depending on their tissue of origin. In the joint, FGFRs 1, 2, and 3 are considered the most abundant, with FGFR-1 and FGFR-3 being the predominant receptors in human chondrocytes. A recent study by Yan and colleagues demonstrated that FGFR expression in articular chondrocytes is dynamic (1). Those authors found that the ratio of FGFR-3 to FGFR-1 was significantly reduced in OA, and that FGFR-3 can be modulated variably by exogenous growth factors, being down-regulated by FGF-2 and up-regulated by bone morphogenetic protein 7 (BMP-7) (1).
In vitro, stimulation of articular chondrocytes by FGF-2 produces a range of cellular responses. However, when these responses are examined in an FGFR-specific manner, a fairly consistent pattern emerges. FGFR-1 mediates procatabolic and antianabolic effects while, conversely, FGFR-3 mediates proanabolic and anticatabolic effects (1).
In in vivo investigations of FGF-2, mice with knockout of FGF-2 develop accelerated OA spontaneously with aging and following surgical destabilization of the joint, suggesting that this ligand confers an overall protective effect in the joint (2). Examination of the role of individual FGFRs in vivo has been more difficult. Valverde-Franco and colleagues examined spontaneous OA in Fgfr3-knockout mice and found that the mice developed accelerated degenerative changes associated with increased levels of catabolic enzymes, including matrix metalloproteinase 13 (MMP-13) (3). These results provide evidence of a possible protective effect of FGFR-3 in mouse joint tissues postnatally, although it is not possible to exclude an intrinsic weakness in the joint tissue that might have arisen through loss of FGFR-3 during skeletal development. With regard to FGFR-1, the presence of a severe developmental phenotype in animals with constitutive knockout of FGFR-1 precludes examination of the joint with aging or following surgical destabilization.
The study by Weng et al, reported in this issue of Arthritis & Rheumatism (4), circumvents both of these problems by creating an Fgfr1 conditional knockout mouse in which Fgfr1 was deleted in a chondrocyte-specific and inducible manner (in postnatal cartilage). This was achieved by crossing Fgfr1-floxed mice (Fgfr1fl/fl) with a line expressing Cre recombinase driven by the type II collagen promoter and controlled by a tamoxifen-sensitive modified estrogen receptor (Col2a1-CreERT2). When these mice are crossed (Col2a1-CreERT2;Fgfr1fl/fl), treatment with tamoxifen drives the translocation of Cre recombinase to the nucleus, where it is able to delete the floxed gene (4).
The generation of chondrocyte-specific, postnatal gene deletion is challenging. Several mouse lines generated using Cre recombination have been developed to assess the effects of chondrocyte-selective gene deletion. Most of the experience with these lines thus far has been with those driven by the type II collagen promoter. Type II collagen is an attractive promoter to use, as it is generally regarded as chondrocyte specific. This is largely true in mature, postnatal joint tissue (type II collagen is briefly expressed in the endocardial cushion and kidney during development), but its utility in driving Cre expression postnatally is limited by the fact that Col2 expression falls off after skeletal maturity. It has been used successfully to delete floxed genes during development and immediately after birth (5), but the loss of expression of Col2 with age is thought to preclude efficient deletion of floxed genes at later time points (>6 weeks of age) (6). The efficiency of Cre-mediated deletion using this promoter was substantially improved by linking the promoter to the more efficient estrogen receptor (ERT2) (7). Owing to these inherent restrictions, newer chondrocyte-selective Cre recombinase–expressing lines, such as those driven by the aggrecan promotor, have been developed (8).
In the study by Weng and colleagues (4), mediation of late, postnatal gene deletion was achieved when the Col2-CreERT2 mouse line was utilized. The authors first showed that delivery of tamoxifen to Rosa26R reporter mice (crossed with Col2-CreERT2) at 8 weeks of age was able to activate β-galactosidase expression (by deletion of a floxed stop codon placed upstream of the β-galactosidase gene). Interestingly, based on their results, X-Gal staining appears to be limited to the noncalcified cartilage only (as is also the case with the aggrecan-CreERT2 mouse line). The activation of this reporter construct is not necessarily predictive of efficient deletion of other floxed alleles. Nevertheless, Weng et al showed that FGFR-1 also appears amenable to late deletion, as evidenced by a reduction in FGFR-1 protein staining in the cartilage of tamoxifen-treated Col2-CreERT2;Fgfr1fl/fl mice. The efficiency of this deletion is not quantified, but the demonstration that these mice have protected cartilage with aging and following surgical destabilization clearly indicates a strong functional effect of this reduction in FGFR-1. Moreover, the results of the Weng et al study support the new paradigm that FGFR-1 mediates degradative processes in cartilage and is associated with an up-regulation of MMP-13 and suppression of proteoglycan synthesis, and that inhibition of this pathway in cartilage by reducing receptor expression levels leads to protection of the joint.
The demonstration that deletion of FGFR-1 in chondrocytes modifies disease is important because it clearly indicates a role for chondrocyte-driven FGFR signaling in cartilage degradation and places the chondrocyte centrally in the disease process. It does not, however, address the role of FGFs elsewhere in the joint. A number of genes are strongly regulated in the whole joint in vivo when murine joints are destabilized (9), and many of these are FGF-2 dependent (Chong K, et al: submitted for publication). FGF-2 appears to modulate inflammatory cell responses, as FGF-2–null mice have increased joint inflammation when compared to wild-type mice, immediately following meniscal destabilization (Vincent T, et al: unpublished observations), and increased levels of synovitis and angiogenesis are seen when FGF-2 is injected into wild-type mouse joints 2 weeks following surgical destabilization (10). FGF-2 has also been linked to repair responses in cartilage and could be involved in recruitment and differentiation of progenitor cells in response to injury. It is not currently known whether these responses contribute to the disease phenotype, and whether they are also dependent on specific FGFR signaling.
The concept that specific FGFRs determine different cellular outcomes is not new and has been previously suggested in the context of skeletal development (11). However, in previous studies, investigators failed to show that the cellular response is dependent on the type of receptor expressed. Rather, the results of these studies have supported the notion that the cell, within its specific environment (in this case, the growth plate), determines the outcome irrespective of which FGFR is transporting the signal (11). These studies highlight the important role of the tissue environment, including both insoluble (matrix) and soluble factors that contribute to the response of chondrocytes to FGFR ligation.
Heparan sulfate is one such matrix component that is critical both for controlling the bioavailability of FGF ligands in the matrix and for contributing to receptor–ligand complex formation at the cell surface. Heparan sulfate proteoglycans that are known to be important for FGF signaling include those in the pericellular matrix, such as perlecan, as well as those present on the cell surface, the syndecans and glypicans. FGF-2 is bound to the heparan sulfate chains of perlecan, and it is from this matrix-bound pool that FGF-2 is liberated in response to mechanical injury/loading, thus allowing it to interact with cell surface FGFRs (12, 13). Binding, bioavailability, and signaling are all processes that are dependent on specific sulfation patterns on the heparan sulfate chains. These patterns likely change during development and disease, and are controlled, in part, by sulfation enzymes that are present within the endoplasmic reticulum during proteoglycan synthesis, but also by the activity of pericellular sulfatases that can modify the glycosaminoglycan chains on secreted proteoglycans. These extracellular enzymes are regulated in OA chondrocytes, and their importance in joint disease was recently highlighted in a study by Otsuki and colleagues, in which it was demonstrated that deficiency of either the Sulf1 or Sulf2 sulfatase enzymes resulted in accelerated OA in mice and altered FGF and BMP signaling (14).
The findings presented by Weng and colleagues (4) have high clinical significance. FGFs are ubiquitous growth factors that have many important biologic functions in normal tissue homeostasis, as well as in pathologic processes such as tumor growth and ectopic angiogenesis. Global suppression of FGFR-1 signaling in vivo would be regarded as a high-risk approach, but the possibility that one might be able to alter the balance of FGFRs in the joint is attractive (Figure 1). Since the FGFR-3:FGFR-1 ratio is increased by exogenous stimulation with BMP-7, it would be possible to test whether intraarticular delivery of BMP-7 is able to ameliorate OA by promoting FGFR-3 expression. Assuming that up-regulation of FGFR-3 has beneficial effects on the other tissues of the joint, this could have realistic potential for individuals with OA.
Another approach, which has already been tested in patients, is the delivery of an FGFR-3–specific ligand. FGF-18 is an important FGF that is essential for endochondral ossification. Unlike most of the other FGFs, which have promiscuous interactions with different FGFRs, FGF-18 binds selectively to FGFR-3 and has minimal effects on FGFR-1 (15). Delivery of FGF-18 in animal models of OA has demonstrated the potential chondroprotective effects of this approach, and recent trials in humans, assessing the intraarticular delivery of FGF-18 in patients with established OA and in those with acute cartilage injury, are currently being performed. The outcome of these trials will be highly informative and will add credibility to FGFR-directed approaches to the treatment of OA in the future.
Professor Vincent drafted the article, revised it critically for important intellectual content, and approved the final version to be published, and takes responsibility for the integrity of the data and the accuracy of the data analysis.