[Frontiers in Bioscience 3, d781 -794, August 1, 1998] |
FGF SIGNALING IN SKELETAL DEVELOPMENT
Michael C. Naski and David M. Ornitz
Department of Molecular Biology and Pharmacology, Washington University School of Medicine, Campus Box 8103, 660 S. Euclid Ave., St. Louis, MO 63110
Received12/22/98 Accepted 2/17/98
4. ENDOCHONDRAL OSSIFICATION
4.1. FGFs and cartilage
The pre- and postnatal development of long bones and vertebrae occurs via endochondral ossification. In this process a cartilaginous template is formed by mesenchymal cells which, under the control of regional morphogenetic and proliferative cues, coalesce and secrete an extracellular matrix that is essential for their differentiation. After the cartilaginous templates have been established, bone formation ensues from ossification centers that form in the center of long bones and proceed as a wave extending toward the two ends. This process begins in the primary ossification centers during embryonic development and is recapitulated during post-natal skeletal growth at the epiphyseal growth plates. During endochondral ossification, chondrocytes differentiate through a series of well-defined morphological zones within the epiphyseal growth plate (figure 5). A zone of proliferation provides a renewable source of chondrocytes for longitudinal bone growth. After exiting the cell cycle these maturing chondrocytes secrete a matrix composed of chondroitin-sulfate proteoglycans and type II collagen, as well as other matrix components. Encapsulated in this matrix, the chondrocytes undergo hypertrophy and subsequently express type X collagen and alkaline phosphatase. Hypertrophic chondrocytes undergo an apoptotic death as their surrounding matrix is mineralized and replaced by trabecular bone (100).
Figure 5. Endochondral ossification. Hematoxylin and eosin stained section of the epiphyseal growth plate from the proximal tibia of a two week old mouse.
The effects of FGFs on cartilage physiology have been studied both in vitro and in vivo. The addition of FGF 2 to chondrocytes cultured in soft agar, as well as monolayer culture, resulted in a dramatic increase in cell proliferation (101-103). In fact, when compared to other known mitogens for chondrocytes, FGF-2 was a more potent mitogen than insulin-like growth factor-1, transforming growth factor-beta (TGF-beta) and epidermal growth factor (101, 104). Cultured rabbit growth plate chondrocytes grown as a pelleted mass synthesize an abundant extracellular matrix, undergo cellular hypertrophy and initiate mineralization of the surrounding matrix. In certain respects, the pellet cultures recapitulate events of chondrocyte maturation in the epiphysis during endochondral ossification. The addition of FGF 2 to these cultures only modestly effects cell proliferation, but profoundly effects chondrocyte differentiation (11, 105). FGF 2 inhibited the terminal phase of chondrocyte differentiation as evidenced by a dramatic inhibition of both the rise in alkaline phosphatase activity and the deposition of calcium (11). Interestingly, the cultures became less sensitive to the effects of FGF 2 once the cells differentiated to hypertrophic chondrocytes. This was shown to correlate with the loss of FGF receptor expression in the terminally differentiated cells, as evidenced by the loss of binding of radiolabeled FGF 2 to the cells (105).
4.2. FGF receptors and endochondral ossification
FGF receptors 1 and 3 are expressed in the epiphyseal growth plate. FGF receptor 3 is expressed in proliferating chondrocytes, whereas FGF receptor 1 is expressed in hypertrophic chondrocytes (106, 107). FGF receptor 3 is also expressed in the cartilage of the developing embryo, prior to formation of ossification centers. The results of Iwamoto et al. (105) using a pellet culture system show that terminally differentiated chondrocytes do not bind FGF-2. These data suggest that FGF-2 has a low affinity for FGF receptor 1 or that matrix cofactors, required for receptor binding, are not present in hypertrophic chondrocytes (108). In vitro data indicate that FGF-2 binds avidly to FGF receptor 1 in the presence of heparin (22). Thus, the absence of binding of FGF 2 may reflect differences in FGF receptor expression in the pellet culture model compared to the growth plate, a lack of correlation between receptor expression determined by in-situ hybridization and synthesis of the protein, or modulation of ligand-receptor affinity during chondrocyte hypertrophy by extracellular matrix components. Expression of syndecan-3, a transmembrane glycosaminoglycan capable of binding and presenting FGFs to their receptor, is restricted to proliferating chondrocytes (108). The absence of this co-receptor in hypertrophic chondrocytes could explain the lack of a detectable interaction between FGF 2 and FGF receptor 1.
Data showing that FGF 2 inhibits the terminal differentiation of chondrocytes (105) in conjunction with the finding that FGF receptor 3 is expressed in proliferating but not hypertrophic chondrocytes suggests that FGF receptor 3 mediates the inhibitory effects of FGF 2 on chondrocyte differentiation. This model predicts that in the absence of FGF receptor 3 signaling, accelerated chondrocyte differentiation would occur. However, in FGF receptor 3 null mice the primary ossification centers form normally (109) and skeletal overgrowth ensues. These data suggest that the primary role for FGF receptor 3 may be to inhibit chondrocyte proliferation (106, 109). This is supported by recent data showing that FGF receptor 3 can induce the expression of cell-cycle inhibitors (110). Additionally, transgenic mice overexpressing FGF-2 are dwarfed (111), consistent with a role for FGF receptor 3 as an inhibitor of chondrocyte proliferation. The growth plate in FGF receptor 3 null mice showed an expansion of the hypertrophic zone, while in transgenic mice overexpressing FGF 2, a reduction of the hypertrophic zone was observed. This implies that FGF receptor 3 may also directly or indirectly regulate chondrocyte differentiation by altering the rate at which cells enter the hypertrophic phase. An interesting and as of yet unresolved issue is the identification of the endogenous ligand(s) for FGF receptor 3 in the epiphyseal growth plate. FGF 2 is a likely candidate given that it is present in the growth plate (112) and is a known ligand for FGF receptor 3 (22). Surprisingly, however, targeted disruption of the mouse Fgf 2 gene produces no gross or histological skeletal phenotype (113). Therefore, other FGFs in addition to FGF 2 must be present within the growth plate and may be functionally redundant with FGF 2.
Mutations in FGF receptor 3 cause the human dwarfing conditions achondroplasia, thanatophoric dysplasia and hypochondroplasia (figure 3). Point mutations in the coding sequence for the receptor cause amino acid substitutions in the extracellular, transmembrane, and kinase domain of the receptor. Additional mutations in the stop codon of FGF receptor 3, presumably resulting in a protein of extended length, also have been identified. These observations emphasize the requirement of tightly regulated FGF receptor 3 activity to maintain normal skeletal growth. We and others have found that the mutations causing achondroplasia and thanatophoric dysplasia are gain of function mutations resulting in increased receptor tyrosine kinase activity (114-117). The G380R mutation in the transmembrane domain of FGF receptor 3, that is responsible for most cases of achondroplasia partially activates FGF receptor 3 (114). By measuring the mitogenic activity of a chimeric receptor consisting of the extracellular domain of FGF receptor 3 fused to the tyrosine kinase domain of receptor 1, we found that the G380R receptor increased the basal activity to approximately 18 percent of its maximal activity. The basal activity of this receptor could be augmented by the addition of ligand and the dose response curve suggested that this receptor has a similar ligand binding affinity to that of the wild type receptor. Studies of receptor tyrosine phosphorylation showed ligand independent receptor autophosphorylation. The K650E and R248C mutations of thanatophoric dysplasia are also activating mutations. These mutations result in ligand independent receptor activation as evidenced by ligand independent cell proliferation and receptor tyrosine phosphorylation. Significantly, these mutations were more strongly activating than the mutation causing achondroplasia. This implies a correlation between the degree of receptor activation and the severity of the dwarfing condition. The mutations causing thanatophoric dysplasia strongly activate the receptor and lead to a severe phenotype. The R248C mutation constitutively activates FGF receptor 3 by forming a disulfide linked receptor homodimer (114). This mutation introduces an unpaired cysteine residue into the extracellular domain of the receptor that forms an intermolecular disulfide linkage. The K650E mutation occurs in a highly conserved lysine residue of the activation loop of the receptor (118, 119). This mutation results in a constitutively active tyrosine kinase, presumably by altering the structure of this loop to that of an active conformation. Unlike the R248C mutation which shows constitutive activation matching that of maximally stimulated wild-type receptor, the K650E mutant receptor can be further activated by ligand to a level greater that that of the wild-type receptor. These observations are consistent with the role that the R248C mutation regulates dimerization whereas the K650E mutation affects the regulation of the kinase activity.