A gain-of-function mutation of Fgfr2c demonstrates the roles of this receptor variant in osteogenesis

Abstract

The b and c variants of fibroblast growth factor receptor 2 (FGFR2) differ in sequence, binding specificity, and localization. Fgfr2b, expressed in epithelia, is required for limb outgrowth and branching morphogenesis, whereas the mesenchymal Fgfr2c variant is required by the osteocyte lineage for normal skeletogenesis. Gain-of-function mutations in human FGFR2c are associated with craniosynostosis syndromes. To confirm and extend this evidence, we introduced a Cys342Tyr replacement into Fgfr2c to create a gain-of-function mutation equivalent to a mutation in human Crouzon and Pfeiffer syndromes. Fgfr2c^C342Y/+ heterozygote mice are viable and fertile with shortened face, protruding eyes, premature fusion of cranial sutures, and enhanced Spp1 expression in the calvaria. Homozygous mutants display multiple joint fusions, cleft palate, and trachea and lung defects, and die shortly after birth. They show enhanced Cbfa1/Runx2 expression without significant change in chondrocyte-specific Ihh, PTHrP, Sox9, Col2a, or Col10a gene expression. Histomorphometric analysis and bone marrow stromal cell culture showed a significant increase of osteoblast progenitors with no change in osteoclastogenic cells. Chondrocyte proliferation was decreased in the skull base at embryonic day 14.5 but not later. These results suggest that long-term aspects of the mutant phenotype, including craniosynostosis, are related to the Fgfr2c regulation of the osteoblast lineage. The effect on early chondrocyte proliferation but not gene expression suggests cooperation of Fgfr2c with Fgfr3 in the formation of the cartilage model for endochondral bone.

Alternative splicing is a major component of the mammalian genome's functional complexity. Three of the four fibroblast growth factor receptor (FGFR) isotypes, encoded by Fgfr1, -2, and -3, use two exons (IIIb and IIIc) alternatively to encode the C-terminal half of the third Ig-like loop of their ligand-binding domain. The resulting b and c variants of the three isotypes bind different groups of ligands (1) and display different localization patterns. Best known are variants of Fgfr2 (2). Fgfr2b is expressed in various epithelia, and its loss-of-function mutation abrogates limb outgrowth and exhibits multiple defects of branching morphogenesis (3). Loss of the mesenchymal Fgfr2c variant causes recessive dwarfism with defective ossification (4). Chimera experiments have revealed that Fgfr2b is required for differentiation of the simple embryonic limb ectoderm into the complex apical ectodermal ridge (AER) epithelium, the signaling center for the proximodistal limb axis (5). The mechanism of action of Fgfr2c is, however, less well understood.

Activating mutations of human FGFR2c resulting in increased ligand binding or ligand-independent signaling cause skull and long-bone growth defects. In certain cases, interchain disulfide bonds form (6) and lead to craniosynostosis with or without mild limb defects (7). More severe changes resulting in greatly increased ligand binding cause the mesenchymal FGFR2c variant to recognize epithelial ligands (8, 9). This change interferes with the control of localization and binding specificity (2), resulting in the severe limb, cranial, visceral, and CNS defects of Apert syndrome (10).

There is considerable interest in using mouse models to clarify the roles of Fgfr2c. Deleting the IIIc exon, together with intronic sequences, caused splicing defects and an Apert syndrome-like dominant-lethal phenotype (11). In contrast, a loss-of-function mutation in the IIIc exon resulted in recessive dwarfism with defective ossification and reduced transcription of the osteoblast-specific Spp1 and Cbfa1/Runx2 genes (4). More recently, a conditional Fgfr2 mutation driven by the mesenchymal Dermo1/Twist2 promoter has been reported (12), which leads to defective ossification and was interpreted as loss of Fgr2c signaling. To further clarify the problem, we created a gain-of-function mutation by using the Cys342Tyr (C342Y) replacement, which is frequently observed in human Crouzon- and Pfeiffer-type craniosynostosis (7). Here we report that the phenotype of this mutation is in most respects opposite to that reported for the loss of Fgfr2c. The heterozygote displays craniosynostosis and enhanced Spp1 expression in the skull vault, whereas perinatal lethal homozygous pups exhibit multiple bony fusions and increased Cbfa1/Runx2 transcription. Studies on cell proliferation in the skull and osteoblast cell number in the femur suggest that the mutation affects both intramembranous and endochondral ossification. These data reinforce and complete the results of the loss-of-function mutation study (4), suggesting that Fgfr2c is required for the normal function of the osteoblast lineage and cooperates with Fgfr3, a regulator of the chondrocyte series, during endochondral osteogenesis (13, 14).

Materials and Methods

Gene Targeting. The C342Y replacement was introduced into exon 9 of a genomic fragment containing exons 7–10 of Fgfr2 by site-directed mutagenesis and was cloned into the OSdupdel targeting vector as described in ref. 4. Germ line-transmitting chimeras of two independent clones were obtained, and the floxed neo cassette was removed in vivo by mating to the PGK-Cre^m deleter strain (15).

Histology and Skeletal Preparations. Cryostat or paraffin-embedded sections were stained with hematoxylin and eosin. Alizarin-stained skeletal preparations were made as described in ref. 16.

Assessment of Cell Proliferation. BrdUrd uptake and immunohistochemistry were described in ref. 17. Hematoxylin was used as a counterstain. Sections were cut in the transverse plane for analysis of the coronal suture and in the sagittal plane for analysis of the sphenooccipital synchondrosis. An image analysis system (Kontron, Zurich) was used for nuclear area measurements; results were expressed as the percent (by area) of BrdUrd-positive nuclei (brown) vs. total nuclei (blue plus brown). Ten equivalent, nonconsecutive sections were measured from two sets of wild-type and heterozygous embryos in each of the two planes.

Histomorphometry. Femora and tibia were prepared from embryonic day (E) 18.5 cesarean section-derived fetuses and embedded in methylmethacrylate as described in ref. 18. Histomorphometric parameters were analyzed by using a Nikon microscope interfaced with the OsteoMeasure system (OsteoMetrics, Atlanta). All measurements were taken beginning at a standard point in the femur just below the growth plate not including the diaphyseal area. Data are presented in accordance with standard nomenclature.

Alkaline Phosphatase-Positive Colony-Forming Units. Bone marrow cells (10⁷ per well) were cultured in 12-well plates for 3 h in 2 ml of α-MEM with 10% FCS; after the nonadherent cells were removed by washing, 10⁷ mitomycin-C-treated guinea pig bone marrow feeder cells were added. Cultures were fed every 3 days. On day 11, the adherent bone marrow stromal cell colonies were stained for alkaline phosphatase after removal of the nonadherent cells. Colonies of 50 or more cells were counted.

Osteoclast-like cells were generated as described in ref. 19. Briefly, 10⁶ spleen or bone marrow cells per ml were cultured in α-MEM supplemented with 10% FCS and 30 ng/ml of both recombinant murine macrophage colony-stimulating factor (MCSF) (R & D Systems) and recombinant human receptor activator of NF-κB ligand (RANKL) (R & D Systems). Between days 7 and 10, the cells were fixed and stained for tartrate-resistant acid phosphatase by using a kit from Sigma. Tartrate-resistant acid phosphatase-positive, multinucleated (>3) osteoclast-like cells were counted by using light microscopy.

Bone density measurements were made by using peripheral quantitative computed tomography (pQCT) as described in ref. 20. A Stratec XCT960A pQCT machine (Norland, Fort Atkinson, WI) was used to acquire scans, and the total bone density measurements were made at a threshold of 1,300 mg/cm³. For in vivo measurements, dual-energy x-ray absorptiometry with a PIXImus densitometer (Lunar, Madison, WI) was used. Anesthetized mice were scanned with a 1.270-mm-diameter collimator, 0.762-mm line spacing, 0.380-mm point resolution, and an acquisition time of 5 min. Bone mineral density was expressed as mg/cm³.

In situ hybridization of whole fetal heads was described by Iseki et al. in ref. 17. Radioactive in situ hybridization and the Fgfr2IIIb and -IIIc probes were described by Orr-Urtreger et al. in ref. 2. For Cbfa1/Runx2, the probe was from G. Karsenty (M. D. Anderson Cancer Center, Houston); for Ihh, from A. McMahon (Harvard University, Cambridge, MA); for PTHrP, from M. Kronenberg (Massachusetts General Hospital, Boston); for Spp1, from B. Hogan (Vanderbilt University, Nashville, TN); for Sox9, from R. Behringer (M. D. Anderson Cancer Center); for Col2a, from Y. Yamada (National Institutes of Health, Bethesda); and for Col10a, from B. Olsen (Harvard Medical School, Boston).

Results

The Fgfr2c-C342Y Gain-of-Function Mutation in Mice. The TGC codon encoding cysteine at position 342 of the IIIc exon of Fgfr2 was mutated to tyrosine (TAC) by site-directed mutagenesis combined with gene targeting (Fig. 1 A–C). Fgfr2c^C342Y/+ heterozygotes are viable and fertile, characterized by shortened face, domed skull, and widely spaced protruding eyes (Fig. 1D). A minority of heterozygotes had a lateral deviation of the nasal area, causing malocclusion leading to feeding difficulties. Homozygous newborns (25% of F₁ crosses) died shortly after birth with signs of respiratory failure, displaying extremely shortened facial area and open eyelids (Fig. 1E).

Fig. 1.

Targeted activation of the Fgfr2c transcriptional alternative. (A) Genomic structure and targeting events; exons are shaded, with the exon number above and the protein domain name underneath. X and Y, 3′ and internal probe, respectively. (B) DNA sequence of the region used for site-directed mutagenesis, showing the C342Y mutation and the newly formed RsaI site. (C) Southern blot analysis of the homologous recombination in embryonic stem cells, probed with the 3′ external probe after EcoRI digestion. (D) Head of wild-type (left) and viable Fgfr2c^C342Y/+ heterozygote (right). Shortened face and slightly bulging, widely spaced eyes are clearly visible. (E) Wild-type (left) and perinatal lethal homozygous Fgfr2c^C342Y/C342Y (right) littermate pups: mutants show smaller size, greatly shortened face, cyanosis, lack of milk spot, and open eyelids. B, BamHI; H, HindIII; RI, EcoRI; S, SacI; TM, transmembrane exon; #, site of point mutation.

Heterozygotes Exhibit Premature Fusion of Calvarial Sutures. The skulls of 4-week-old Fgfr2c^C342Y/+ mice were rounded because of shortening of the rostrocaudal axis (Fig. 2A). The coronal sutures were fused, and the lambdoid sutures were partially fused, whereas the sagittal sutures, which do not close in the mouse, were only partially separable. In contrast to the loss-of-function mutation, where the basioccipital–exoccipital–basisphenoid synchondroses were precociously fused (4), in Fgfr2c^C342Y/+ mice, these sutures remained open as in the wild type (Fig. 2B). However, the skull-base shortening was mainly in the sphenoid region, suggesting compromised growth of this bone. The maxilla was also shortened, and one molar was missing. Histology of the coronal suture showed increased overlap of the frontal and parietal bones by E16.5 (Fig. 2C), and by 4 weeks postnatal, the site of the obliterated coronal suture could be detected in sections only by the contrast in thickness between the frontal and parietal bones. The axial and appendicular skeleton of the heterozygote appeared normal.

Fig. 2.

Skull phenotype of Fgfr2^C342Y/+ heterozygotes. (A and B) Littermates at postnatal day 29 (P29). (A) Arrows point to the obliterated coronal and fused lambdoid sutures, with multiple bone inserts. (B) Arrowheads indicate the normal basioccipital–basisphenoid–exoccipital synchondroses. (C) Histological sections of the coronal suture showing increased overlap of frontal (f) and parietal (p) bones at E16.5 and obliterated suture at P29 (arrows). (Scale bars, 200 μm.)

C342Y Homozygotes Show Cleft Palate, Multiple Synostoses, and Lung Defects. The skull of Fgfr2c^C342Y/C342Y newborn pups was shortened in the nasomaxillary region with domed and round cranium (Fig. 3 A–D). The skull base and face were shortened, and the secondary bony palate failed to close (Fig. 3 D and F). Overt cleft palate (Fig. 3 E and F) was observed in 3 of 4 homozygotes, in 1 of 15 heterozygotes, and in none of 7 homozygous wild-type P1 littermates. Sagittal sections of the head revealed robustly thickened nasal and palatal bones and a thickened and curved skull base (see Fig. 5 C and D).

Fig. 3.

Shortening of the nasomaxillary area and cleft palate in the P1 homozygous mutant skull. (A–D) Lateral and ventral views of alizarin-stained preparations. Note the shortened nasomaxillary and sphenoid regions. (E and F) Fixed heads and palatal views showing overt cleft palate in the mutant. wt, wild type.

Fig. 5.

Ossification-related gene expression in wild type (A, C, E, G, I, K, M, and O) and heterozygote (B) or Fgfr2^C342Y/C342Y homozygote (D, F, H, J, L, N, and P). (A and B) Spp1 expression is enhanced in the heterozygote skull vault at E18.5. (C–L) Cbfa1/Runx2 expression in sagittal sections of the P1 skull base (C and D), with higher-power views of the basioccipital–basisphenoid synchondrosis (E–H) and in longitudinal sections of the E13.5 humerus (I–L), showing enhanced signal in the perichondrium and periosteum of the mutant. (M–P) There is a slight increase in Ihh expression in the E13.5 humerus. pit, pituitary.

Alizarin-stained skeletal preparations revealed excessive bone formation. The axial skeletons of newborn mutants were shorter and thicker than those of the wild type, and the knee joint was fused in most fetuses by a posterior bony bridge (Fig. 4 A and B). There was synostosis of the sternebrae (Fig. 4C) and fusion of the lower cervical and upper thoracic (Fig. 4D) as well as the sacral and caudal vertebrae. Other primary ossification centers, e.g., those of the pelvic girdle (Fig. 4 A and B), were closer together but not fused. An unexpected observation was the defective development of the tracheal cartilage rings, which were fused into a thin cartilaginous sleeve (Fig. 4E) and associated with underdeveloped lungs (Fig. 4F). These tracheal and lung defects, together with the frequently observed cleft palate, may explain the lethality of this phenotype.

Fig. 4.

Synarthroses and trachea/lung defects in the homozygous mutant neonate. (A and B) Alizarin red staining. (C–E) Alizarin red and alcian blue staining. (A and B) Knee-joint synarthrosis; arrowhead shows mineralized bridge between the femur and tibia of the mutant. (C) Fused sternebrae and asymmetric rib cage in the mutant (arrows). (D) Fusion of cervical vertebral arches (arrowheads) but lack of ossification of vertebral bodies. (E) Tracheal rings replaced by a cartilaginous sheath. (F) Reduced lungs in the homozygote (right) but not heterozygote (center), compared with wild type (left).

Gain of Fgfr2c Function Results in Increased Osteogenesis. We analyzed the functional basis for the skeletal phenotype of this mutation. Evidence that the C342Y mutation of Fgfr2c affects osteoblasts was obtained by investigating osteopontin (Spp1) expression. Osteopontin is one of the major noncollagenous bone matrix proteins expressed by early osteoblasts. Wholemount in situ hybridization of E18.5 fetal heads revealed a significant increase in Spp1 expression in the nasal, frontal, parietal, and occipital bones in Fgfr2c^C342Y/+, compared with those of wild type (Fig. 5 A and B). As expected, this effect was opposite to that observed with the loss-of-function mutation of Fgfr2c (4). Increased Spp1 expression also was observed in the E18.5 femur (not shown).

Next, we investigated Cbfa1 expression in Fgfr2c^C342Y/C342Y newborn mice. Cbfa1/Runx2 is a master gene of the osteocyte lineage; when it is inactivated, only the chondroskeleton forms (21, 22). In situ hybridization of sagittal sections of the skull revealed increased Cbfa1 expression in bones of the skull base, skull vault, and nasomaxillary area (Fig. 5 C and D). Higher magnification of the basioccipital–basisphenoid synchondrosis revealed that Cbfa1 transcripts accumulate in the bone collar, periosteum, and perichondrium, presenting heavier signals in the mutant than in the wild type (Fig. 5 E–H). As an example of early osteogenesis in the appendicular skeleton, the E13.5 humerus was investigated. Cbfa1 expression was stronger in the periosteum and proliferating chondrocyte layer of the mutant than in the wild type (Fig. 5 I–L).

Chondrocyte differentiation is orchestrated by a regulatory loop between the Indian hedgehog (IHH) and the parathyroid hormone-related peptide (PTHrP) (23). Recent data emphasize that Ihh, along with its involvement with chondrocyte differentiation, also is a mediator of chondrocyte proliferation (24). We therefore investigated the effect of the Fgfr2c^C342Y/C342Y mutation on Ihh (Fig. 5 M–P) and PTHrP expression (not shown) in E13.5 embryos. Ihh transcription was detected in the perichondrium as well as in the prehypertrophic center of the late cartilage model of the E13.5 humerus; the expression domain of the mutant was somewhat increased, compared with that of the wild type. PTHrP was expressed in the perichondrium and prehypertrophic chondrocytes of the prospective epiphysis; its expression was not detectably altered in the mutant (not shown).

Although no significant increase was observed in the levels of Ihh and PTHrP expression, it remained possible that Fgfr2c acts at an earlier stage of chondrogenesis. Therefore, we examined genes that are involved in chondrogenesis. Sox9, which regulates expression of the chondrocyte-specific Col2a gene, is expressed in all cartilage primordia (25). In situ hybridization with Sox9, Col2a, and Col10a, did not, however, reveal a significant difference between wild-type and newborn Fgfr2c^C342Y/C342Y mice (data not shown). This gene-expression analysis suggests that Fgfr2c affects the osteoblast rather than the chondrocyte lineage.

The Fgfr2c^C342Y Mutation Affects Osteoprogenitor Cell Numbers in Long Bones. Histomorphometric analysis was carried out on the femora of E18.5 fetuses. Both heterozygous and homozygous mutants displayed a significantly higher number of osteoblasts than did the wild type, whereas the number of osteoclasts was unaffected (Table 1). Consistent with gene expression data, this analysis also suggests that this mutation affects the osteoblast lineage.

Table 1. Histomorphometric data of femora from E18.5 embryo.

Measurement	Fgfr2c+/+	Fgfr2cC342Y/+	Fgfr2cC342Y/C342Y
No. of osteoblasts
Per tissue area, mm2	4,909.3 ± 352.5	6,615.3 ± 527.1*	8,504.2 ± 740.0*
Per bone perimeter, mm	75.3 ± 5.3	88.7 ± 3.4	128.7 ± 10.0*
No. of osteoclasts
Per tissue area, mm2	78.3 ± 26.1	104.5 ± 20.1	87.0 ± 10.1
Per bone perimeter, mm	1.2 ± 0.4	1.5 ± 0.4	1.6 ± 0.2

^*, P < 0.05 compared with wild type.

We analyzed the functional basis of this phenotype. Pluripotent mesenchymal bone marrow-derived stem cells differentiate into osteoblast progenitors in vitro, forming colony-forming units with osteoblastic features, such as alkaline phosphatase expression. We examined the effect of the C342Y mutation on osteoblast progenitor cells and osteoclast-like cells in bone marrow stromal cell cultures of 6-week-old mice. The Fgfr2c^C342Y/+ heterozygotes showed a significant increase in the number of alkaline-phosphatase-positive colonies, suggesting an increased number of osteoblast progenitor cells (Table 2), whereas there was no significant difference in the number of osteoclast-like cells (Table 2).

Table 2. In vitro culture of bone marrow stromal cells and splenocytes.

		No. of osteoblast progenitors in bone marrow	No. of osteoclast-like cells
Exp.	Genotype		In bone marrow	In spleen
1	Wild type	20 ± 1.7	177 ± 12.8	128 ± 14
	Fgfr2cC342Y/+	36 ± 3.4*	205 ± 4	127 ± 7.4
2	Wild type	11 ± 1.6	148 ± 6	91 ± 0.7
	Fgfr2cC342Y/+	32 ± 1.4*	150 ± 10.1	102 ± 9.5

Two independent experiments were conducted. *, P < 0.05 compared with wild type.

Next, we investigated the impact of this mutation on bone mineral density. For in vivo measurements, 6-week-old mice were anesthetized and scanned by dual-energy x-ray absorptiometry using the PIXImus scanner. The head and shaft of the right femur showed no significant difference between wild type and heterozygote. For ex vivo bone density measurements, we used peripheral computed tomography. Consistent with the dual-energy x-ray absorptiometry results, at 6 weeks of age, the mean total density of wild-type and mutant tibiae, humeri, and third lumbar vertebrae did not differ significantly.

Cell Proliferation Is Affected Differently in Membranous Skull Vault and Endochondral Skull Base. After the observation of premature growth and fusion of the frontal and parietal bones in the coronal suture (Fig. 2 A and C), we carried out a cell proliferation analysis (BrdUrd uptake) of wild-type and heterozygous pups at three time points (Table 3). At E14.5, the proportion of BrdUrd-positive osteoprogenitor cells in the frontal and parietal bone fronts of the coronal suture was significantly (P < 0.05) increased in heterozygous, compared with that of wild-type fetuses; however, at E16.5, the proportion was equal, and at P1, the proportion was slightly (but not significantly) decreased (P = 0.1). This result (Table 3) suggests that more osteoprogenitors are formed at the start of coronal suture formation, but that at later stages, formation is stabilized or possibly decreased. These data may explain the more rapid growth of the overlapping frontal and parietal bones observed at E16.5 (Fig. 2C). In contrast, the basisphenoid–basioccipital synchondrosis showed greatly decreased chondrocyte proliferation at E14.5, especially on the basisphenoid side (P < 0.0001, compared with P < 0.005 for the basioccipital side). At E16.5 and P1, the numbers were equal or slightly but not significantly increased. These results suggest that the growth of the intramembranous skull vault and the growth of endochondral skull base are affected at an early stage of bone formation, but through different processes.

Table 3. Cell proliferation assessed by BrdUrd uptake in the coronal suture and in each side of the basisphenoid—basioccipital synchondrosis.

		BrdUrd uptake, %
Suture	Genotype	E14.5	E16.5	P1
Coronal	Wild type	35 ± 1	20.4 ± 1.61	55.6 ± 3.1
	Fgfr2cC342Y/+	39.6 ± 1.02*	20.1 ± 1.7	48.76 ± 2.7
Basisphenoid	Wild type	14.08 ± 1	2.44 ± 0.04	4.34 ± 0.53
	Fgfr2cC342Y/+	3.79 ± 0.3†	3.58 ± 3.58	5.98 ± 0.6
Basioccipital	Wild type	9.07 ± 0.8	2.96 ± 0.3	4.44 ± 0.77
	Fgfr2cC342Y/+	5.56 ± 0.7‡	2.7 ± 0.28	5.75 ± 0.6

Numbers represent BrdUrd-positive nuclear area measurements expressed as percentage of the total nuclear area. *, P < 0.05; †, P < 0.0001; ‡, P < 0.005 (Student's t test) compared with wild type.

Discussion

The mutant mice described carry the C342Y replacement, equivalent to a mutation associated with Crouzon and Pfeiffer syndromes in humans (7). The skull-vault craniosynostosis, shallow orbits, and ocular proptosis with no obvious limb defects observed in our Fgfr2c^C342Y/+ mice also are hallmarks of Crouzon syndrome. Although homozygotes for the C342Y mutation are unknown in humans, some features displayed in the homozygous pups resemble those reported as part of the Crouzon spectrum. These include cleft lip and palate, maxillary hypoplasia with crowded teeth, absence of tracheal rings, and vertebral and rib anomalies (26).

We investigated the mechanism of action of this C342Y mutation on osteoblasts and chondrocytes by analyzing bone differentiation, localized gene expression, cell proliferation, osteoblast cell number, and bone density. The observed increase of Spp1 and Cbfa1/Runx2 expression in the gain-of-function mutant contrasts with their decreased expression in the Fgfr2c loss-of-function mutants (4). These genes are expressed in preosteoblasts and early osteoblasts (17, 27), suggesting that FGFR2c signaling is upstream of both and therefore regulates early osteoblast development. This result is consistent with in vivo and in vitro studies demonstrating elevated levels of Spp1 and Runx2 in cells treated with FGF ligands or in cells expressing activating FGFR mutations (28).

Osteoprogenitor and osteoblast cell numbers were largely consistent with the gene-expression observations. Mutant mice showed a significant increase in osteoblast number in the femur (Table 1) and number of osteoprogenitor cells in the bone-marrow stem cells (Table 2). There also was an increase in the number of proliferating (osteoprogenitor) cells in the coronal suture at E14.5 (Table 3), suggesting that increased production of osteoblasts at this stage, but not later, may lead to the more rapid growth of the frontal and parietal bones observed at E16.5 and P1. Together with the increased Spp1 and Runx2 expression, this early increase in cell number is likely to explain the eventual coronal synostosis. The increased osteoblast number was translated into thickened bones and synarthroses but not increased mineralization. Mineralization is a late event in osteogenesis, and increased fibroblast growth factor signaling can actually inhibit this process. In vitro studies also indicate that FGFR2c increases the proliferation of osteoblasts while inhibiting mineralization (29, 30). Our in vivo data confirm the proliferation aspects but showed no loss of mineralization in the entire Fgfr2c^C342Y/+ skeleton.

In contrast to the increased number of osteoprogenitor cells and osteoblasts, proliferation of chondrocytes in the skull base was decreased (Table 3). The effect was greater on the basisphenoid side of the basisphenoid–basioccipital synchondrosis, reflecting the greater shortening of the preoccipital part of the skull base. Reduced chondrocyte proliferation and/or differentiation also was suggested by the failure of cartilaginous rings to form in the trachea of homozygous mutant pups. This finding suggests that FGFR2c signaling has different effects on the osteoblast and chondrocyte lineages and hence on intramembranous and endochondral ossification. However, some of the effects are region-specific; e.g., in the cervical vertebrae of homozygous pups, there is overgrowth leading to synarthrosis of the vertebral arches but failure of formation of primary ossification centers in the vertebral bodies.

In general, both the phenotype and the gene expression pattern of gain- and loss-of-function mutants of Fgfr2c were opposite, as expected. A notable exception was that both mutants showed coronal synostosis. This observation supports the interpretation proposed by Iseki et al. (17), that increased fibroblast growth factor signaling down-regulates Fgfr2 and inhibits cell proliferation while up-regulating Fgfr1 and enhancing osteogenic gene expression. Hence, both gain and loss of Fgfr2c function would result in decreased FGFR2c signaling in the coronal suture, leading to premature fusion. Taken together, our investigations of the gain- and loss-of-function mutants of Fgfr2c suggest that this splice variant is required for normal function of the osteoblast lineage during both intramembranous and endochondral osteogenesis, and that it cooperates with Fgfr3, a regulator of the chondrocyte series, during endochondral osteogenesis (13, 14)