Fgfr2 and osteopontin domains in the developing skull ... - Development

3375

Development 124, 3375-3384 (1997) Printed in Great Britain © The Company of Biologists Limited 1997 DEV4924

Fgfr2 and osteopontin domains in the developing skull vault are mutually exclusive and can be altered by locally applied FGF2 S. Iseki1,2, A. O. M. Wilkie3, J. K. Heath4, T. Ishimaru1,2, K. Eto2 and G. M. Morriss-Kay1,* 1Department 2Department

of Human Anatomy, South Parks Road, Oxford OX1 3QX, UK of Developmental Biology, Graduate School of Dentistry, Tokyo Medical and Dental University, 1-5-45, Yushima, Bunkyo-ku, Tokyo 113, Japan 3Institute of Molecular Medicine, John Radcliffe Hospital, Headington, Oxford OX3 9DS, UK 4School of Biochemistry, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK *Author for correspondence (e-mail: [email protected])

SUMMARY Mutations in the human fibroblast growth factor receptor type 2 (FGFR2) gene cause craniosynostosis, particularly affecting the coronal suture. We show here that, in the fetal mouse skull vault, Fgfr2 transcripts are most abundant at the periphery of the membrane bones; they are mutually exclusive with those of osteopontin (an early marker of osteogenic differentiation) but coincide with sites of rapid cell proliferation. Fibroblast growth factor type 2 (FGF2) protein, which has a high affinity for the FGFR2 splice variant associated with craniosynostosis, is locally abundant; immunohistochemical detection showed it to be present at low levels in Fgfr2 expression domains and at

high levels in differentiated areas. Implantation of FGF2soaked beads onto the fetal coronal suture by ex utero surgery resulted in ectopic osteopontin expression, encircled by Fgfr2 expression, after 48 hours. We suggest that increased FGF/FGFR signalling in the developing skull, whether due to FGFR2 mutation or to ectopic FGF2, shifts the cell proliferation/differentiation balance towards differentiation by enhancing the normal paracrine downregulation of Fgfr2.

INTRODUCTION

of each FGF in cell lines expressing specific FGFR splice variants (Ornitz et al., 1996). The FGFR2 gene is of particular interest in the context of craniofacial development. Dominantly acting missense mutations located mainly in the IgIIIa/IIIc domain or in the IgII/IgIII linker region are associated with a variety of craniosynostosis syndromes (Crouzon, Pfeiffer, Jackson-Weiss, Apert and Beare-Stevenson syndromes) (reviewed by Wilkie et al., 1995; Pryzlepa et al., 1996). The phenotypes have in common that the coronal suture is the most consistent suture to show premature fusion. Some of these syndromes, as well as non-syndromic coronal craniosynostosis, have also been found to be associated with equivalent mutations of FGFR1 or FGFR3 (Bellus et al., 1996 and references therein; Moloney et al., 1997). Abnormal FGFR signalling has also been implicated in the coronal craniosynostosis seen in Saethre-Chotzen syndrome, which is due to mutation of TWIST: the TWIST gene product is a transcription factor that may be a prerequisite for FGFR signalling during mesoderm formation (Howard et al., 1997; El Ghouzzi et al., 1997). Many of these FGFR-related craniosynostosis syndromes additionally show abnormalities of the digits of the hands and feet, but achondroplasia and related syndromes affecting the growth plates of long-bones have to date only been associated with mutation of FGFR3 (reviewed by Webster and Donoghue, 1997). Homozygous disruption of mouse Fgfr genes have been

There are four fibroblast growth factor receptors, FGFRs1-4, three of which (FGFRs1-3) are known to play important roles in skeletal differentiation and growth, as revealed by the identification of specific mutations in previously recognised clinical syndromes (see Muenke and Schell, 1995; Wilkie and Wall, 1996, for reviews). They are transmembrane receptors with three immunoglobulin-like domains (IgI, IgII and IgIII) in the extracellular region and a split tyrosine kinase domain in the intracellular region of the molecule. They show high affinity for fibroblast growth factors (FGFs), of which fourteen mammalian forms are known (Wilkie et al., 1995; Yamasaki et al., 1996; Smallwood et al., 1996). The ligand binding region is thought to include parts of IgII and IgIII, but the precise region of binding is FGF-specific (see Mason, 1994, for review). FGFRs1-3 each exist as two principal isoforms, derived from alternative splicing of a common IgIIIa exon onto either a IIIb or a IIIc exon (Johnson and Williams, 1993). For FGFR2, the variant including the IIIa/IIIb domain is also known as KGFR, and the variant including the IIIa/IIIc domain as BEK. Both isoforms bind with high affinity to FGF1 and FGF4, but differ in their affinity for FGF2 (high affinity for IgIIIa/IgIIIc) and FGF7 (high affinity for IgIIIa/IgIIIb) (Miki et al., 1992; Yayon et al., 1992; Orr-Urtreger et al., 1993). These binding properties are reflected in the mitogenic activity

Key words: FGF2, FGFR2, osteogenesis, cell proliferation, skull, mouse fetus, human

3376 S. Iseki and others uninformative with respect to skull development (Deng et al., 1994, 1996; Yamaguchi et al., 1994; Muenke and Schell, 1995; Colvin et al., 1996). In the absence of mouse mutants relevant to understanding the biological basis of FGFR-related craniosynostosis, epigenetic manipulations affecting FGFR signalling can be carried out; however, before the results of these experiments can be interpreted, we first need to define the normal patterns of expression of the mouse Fgfr2 gene. The expression patterns of mouse Fgfr1 and Fgfr2 were published before the insights from clinical molecular genetics were available, and did not include developmental stages late enough to show development of the skull vault (Orr-Urtreger et al., 1991, 1993; Peters et al., 1992). In this study, we show that Fgfr2 is expressed mainly at the growing margins of the flat bones of the skull vault of mouse fetuses, including an area of overlap in the coronal suture. Its expression coincides with areas of rapid cell proliferation but is mutually exclusive with domains of osteogenic differentiation. Craniosynostosis resulting from IgIIIa/IIIc FGFR2 mutation has been interpreted as due to constitutive activation of the receptor (Wilkie et al., 1995; Neilson and Friesel, 1995, 1996; Galvin et al., 1996). In order to test the hypothesis that the increased differentiation seen in coronal craniosynostosis is due to increased FGFR signalling, we implanted beads soaked in FGF2 onto the coronal suture of E15 fetuses, which were then allowed to continue development to E17. FGF2 was chosen for this experimental approach because it has a high affinity for the IgIIIa/IIIc isoform of FGFR2 (see above). Furthermore, in a preliminary study using dot-blot analysis and immunohistochemistry, we found FGF2 to be present in great abundance compared to other FGFs in the day 16 fetal skull (data not shown). Implantation of the FGF2 beads resulted in ectopic (intrasutural) expression of the bone differentiation marker osteopontin and displacement of Fgfr2 expression. This result indicates that the technique is a valid experimental model for investigating mechanisms of FGF/FGFR signalling in normal and pathological growth and differentiation at the margins of the developing skull bones.

MATERIALS AND METHODS Animals and double-stained skeletal preparation C57Bl/6 mouse fetuses (plug day = E0) were used for all experiments. Preparation of double-stained fetal skulls was carried out as described by Wood et al. (1996). Probes for in situ hybridization The following probes were used: (1) a 1.96 kb fragment of mouse Fgfr2 (IgIIIa/IIIc splice variant, including the whole extracellular and transmembrane domains) (Raz et al., 1991) cloned into pBluescript KS+ (because of its length, this probe detects transcripts of both IgIIIa/IIIc and IgIIIa/IIIb variants of Fgfr2); (2) a 984 bp HindIII fragment of mouse osteopontin cloned into pGEM1 and pGEM2 (Nomura et al., 1988). To generate antisense and sense transcripts, the plasmids were linearized and transcribed as described by Wilkinson (1992) using T3, T7 or Sp6 RNA polymerase; the transcripts were reduced to an average size of 200 bases by limited alkaline hydrolysis. In situ hybridization Single whole-mount in situ hybridization was carried out as described

by Wilkinson (1992) with some modifications. For the skull preparations, the skin was removed beforehand. For double-labelled in situ hybridization digoxygenin-labelled Fgfr2 probe and fluoresceinlabelled osteopontin probe were used. BCIP (Boeringer Mannheim) and Magenta-Phos (Biosynth) were used for immunodetection of digoxygenin and fluorescein, respectively. For sectioning after wholemount in situ hybridization, specimens were cut frozen. In situ hybridization was also carried out on unfixed frozen sections with digoxygenin-labelled probe. BrdU administration and immunohistochemical detection Pregnant mice were given 100 mg/kg of bromodeoxyuridine (BrdU) (Boehringer Mannheim) in PBS solution by intraperitoneal injection 1 or 2 hours before killing. BrdU immunohistochemical detection was carried out on whole heads following fixation in 4% paraformaldehyde or following whole-mount in situ hybridization for Fgfr2, using antiBrdU antibody (Boehringer), anti-mouse IgG antibody conjugated with alkaline phosphatase (Vector) and NBT/BCIP or Magenta-Phos as substrate. BrdU immunohistochemical detection was also carried out on paraffin sections using diaminobenzidine/peroxide detection, counterstained with haematoxylin. Immmunohistochemical detection of FGF2 Frozen sections were mounted on epoxysilane-coated slides, air-dried for 30 minutes and fixed in acetone. Endogenous peroxidase was blocked with H2O2/methanol. Anti-human bFGF polyclonal antibody (R & D systems) diluted in PBT was applied overnight; the secondary antibody was biotinylated anti-rabbit horse IgG (Vector) in PBT (30 minutes) followed by avidin-biotin complex/PBS. Detection was by diaminobenzidene/peroxide, counterstained with methyl green. For comparison between FGF2 immunostaining, sites of bone matrix deposition, and expression of Fgfr2 and osteopontin, parallel sections were processed for FGF2 immunohistochemistry, histochemical staining with Alcian blue at pH 2.5 and in situ hybridization for each of the two genes. Subcutaneous insertion of FGF2 beads in fetal heads Heparin-coated acrylic beads (Sigma), 125-150 µm diameter, were soaked in either PBS or in 400 mg/ml human FGF2 for at least 1 hour on ice and rinsed twice with PBS prior to implantation. Surgery was carried out on 48 E15 fetuses as described by Muneoka et al. (1990), leaving the placenta undisturbed, attached to the opened uterus. The yolk sac and amnion were opened, a small incision was made with an iridectomy knife in the head skin, just behind the eye, and four beads were inserted into the slit. After surgery, the membranes were closed with a fine suture, the uterus was left open and the abdomen was closed. The fetuses were allowed to develop until E17, when they were collected from the dam, fixed in paraformaldehyde and processed for in situ hybridization to detect transcripts of either osteopontin or Fgfr2, or both. For each pregnant female, two fetuses were implanted with beads soaked in FGF2 and two with PBS beads; the other fetuses were removed.

RESULTS Osteogenesis of the skull vault The mammalian skull vault consists mainly of five bones that mineralise directly in membrane: a pair of frontal bones, a pair of parietal bones and an interparietal bone; the interparietal bone later fuses with the endochondral component of the occipital bone to become the superior part of that bone. Double-stained skeletal preparations of mouse fetuses (Fig. 1, right-hand column) show the mineralizing patterns of these skull bones within the skeletogenic membrane, which itself

FGF/FGFR2 signalling in skull development 3377

Fig. 1. Fgfr2 and osteopontin expression during skull development. The left column shows the Fgfr2 transcript localization pattern and the middle column shows the osteopontin transcript localization pattern from E15 to E18; the right column shows double-stained skulls from specimens one day older (E16 to E18). The diagram (bottom right) shows the position of the developing bones and sutures illustrated: c, coronal suture; e, eye; fr, frontal bone; ip, interparietal bone; l, lambdoid suture; m, metopic suture; pa, parietal bone; n, nasal bone; nc, nasal cartilage; s, sagittal suture. Scale bar, 1 mm.

forms the outermost of the meningeal layers that surround the brain. Mineralisation of the skull vault is initiated later than that of the facial bones (premaxilla, maxilla and mandible) and of the endochondral bones of the skull base, occiput and the vertebral column. Frontal bone mineralisation starts around E16 at the position of the superciliary arch; parietal bone mineralisation starts at the same time, in the lateral region just caudal to the mineralised frontal bone area. Mineralisation of the interparietal bone is not observed at this stage. By E17, the frontal and parietal bones are mineralised in the lateral part of their future domains; interparietal mineralisation can be observed as a very narrow transverse line at some distance caudal to the parietal bones. By E18, mineralisation of the bones has extended to the top of the head, separated by undif-

ferentiated skeletogenic membrane in the midline. The future coronal suture can be observed as a very narrow mineralisation-free space at this stage. The unmineralised membrane between the frontal bones (future metopic suture) has an irregular edge; in contrast, borders of the sagittal suture (between the parietal bones) and the lambdoid suture (between the parietal and interparietal bones) are smooth. Osteopontin expression is first detected at E14.5 as two lateral domains (frontal and parietal anlagen); an interparietal domain cannot be detected until E15, by which stage the frontal and parietal domains have extended medially (Fig. 1, middle column). The coronal suture is a narrow expression-free region between the frontal and parietal domains, but the future metopic, sagittal and lambdoid sutures are represented by extensive areas

3378 S. Iseki and others of expression-free skeletogenic membrane. By E16, the frontal and parietal domains extend almost to the midline at the top of the skull; as observed in the skeletal preparations, the medial edges of the frontal osteopontin expression domains are irregular, in contrast to the smooth edges of the parietal domains. At all stages, the caudal edge of each parietal osteopontin domain coincides with that of the cerebral hemisphere beneath it and the caudal edge of the interparietal domain coincides with the transverse intercollicular sulcus; there are no morphological brain landmarks coinciding with the coronal suture. Fgfr2 gene expression during skull development The Fgfr2 mRNA localization pattern in the future skull vault (Fig. 1, left-hand column) is closely related to that of osteopontin. Fgfr2 transcripts are first observed on E14.5 in the same sites as those in which osteopontin transcription is initiated. Weak expression domains in a reticular pattern are observed in the parietal and interparietal regions; from E16 onwards, they are clearly outlined by a border showing a high level of transcription. In contrast, in the region of developing frontal bone, there is faint and ubiquitous transcription, with a higher level of transcripts in the medial part of the domain; this pattern is observed from E16 to E18. Correlation between early osteogenesis and the Fgfr2 gene expression pattern Double-labelled in situ hybridization was carried out on day 16 fetal heads to clarify the relationship between the localization patterns of osteopontin and Fgfr2 gene transcripts. In the parietal and interparietal domains, the osteopontin transcripts are entirely within the Fgfr2 outlines (Fig. 2A,B). In the frontal bone domain, which is not outlined, Fgfr2 transcripts spread further medially than those of osteopontin (Fig. 3A,B). In the region of the coronal suture, a single linear Fgfr2 domain separates the frontal and parietal osteopontin domains (Fig. 2B). In general, transcripts of the two genes localize in a mutually exclusive fashion: both at the borders and in the interior of the domain of each future bone, areas can be seen in which one gene is transcribed and the other is not (Fig. 2B). This pattern of mutual exclusivity is confirmed by sections of the whole-mount specimens, which additionally confirm that both genes are expressed exclusively in the outer (skeletogenic) membrane and not in the underlying dura mater or inner meningeal layers (Fig. 2C). Correlation between cell proliferation and Fgfr2 gene expression Cell proliferation is one of the features of the immature, predifferentiated condition; it is essential for growth of the skull vault bones, which takes place mainly at the margins but is also associated with thickening and remodelling of the differentiated regions. Cell proliferation was investigated in the developing skull vault in order to compare its pattern, by means of BrdU uptake and subsequent immunohistochemical detection, with that of gene expression (Fig. 3). On E16, whole-mount immunohistochemical detection of BrdU shows that dividing cells are scattered throughout the skull vault but regions in which a high proportion of cells show BrdU incorporation show a distinct pattern. This pattern resembles that of Fgfr2 transcription; conversely, regions in which low numbers of cells show BrdU incorporation appear to coincide with regions

Fig. 2. Double-labelled in situ hybridization of Fgfr2 and osteopontin. (A) Double-labelled whole-mount in situ hybridization specimen of E16 skull: blue shows Fgfr2 transcripts and red shows osteopontin transcripts. (B) Enlargement of the parietal region: the parietal osteopontin domain is entirely within the Fgfr2 outline; the mutually exclusive transcription patterns of Fgfr2 and osteopontin are clearly observed in the medial region of the future parietal bone (arrowheads). (C) Frozen section of double-labelled whole-mount in situ hybridization specimen at the caudal edge of the parietal domain (rostral is left): blue (Fgfr2) and red (osteopontin) signals are adjacent within the outer (skeletogenic) membrane. Scale bars: A, 1 mm; B, 100 µm; C, 10 µm.

in which osteopontin is expressed (compare Fig. 3A with Fig. 2A,B). The apparent correlation between Fgfr2 transcripts and BrdU uptake was confirmed by revealing the two localization patterns on the same specimen (Fig. 3B). In order to discover whether BrdU was localized to the same tissue layer as Fgfr2 transcripts, BrdU detection was carried out on frozen sections of specimens previously processed for Fgfr2 in situ hybridization. On the sections, it is clear that the cells that strongly transcribe the Fgfr2 gene are those that are most actively proliferating, and that both Fgfr2 transcription and the most active cell proliferation are localized to the outer layer of the developing meninges, i.e. the skeletogenic membrane (not illustrated).

FGF/FGFR2 signalling in skull development 3379 extend from the margins of each of the osteoid plates, the parietal and frontal components overlapping in the coronal suture (Fig. 3C). There are also some BrdU-positive cells on the outer surface of the osteoid plates, above the blue-stained osteoblasts which are attached to the matrix, which they secrete. Preosteoblasts, not yet associated with matrix, separate the BrdU-positive cells from the osteoid plates. Immunohistochemical localisation of FGF2 Serial frozen coronal sections of day 16 fetal mouse heads were mounted sequentially on four slides, so that one in four sections were processed for each of the following: (1) Alcian blue histochemistry at pH 2.5 to reveal extracellular proteoglycans that might be available for binding FGF2 protein, (2) immunohistochemical detection of FGF2, (3) in situ hybridization to reveal osteopontin and (4) in situ hybridization for Fgfr2 transcripts. Comparisons between the sections within each group of four were made in the region of the coronal suture, including the adjacent frontal and parietal areas. One group of four sections is illustrated (Fig. 4). Alcian blue staining revealed high levels of proteoglycans and other polyanions in the skin, connective tissue and skeletogenic membrane; the premineralised bone matrix (osteoid) of the future frontal and parietal bones showed up as particularly dense sites of staining. A strong FGF2-positive immunohistochemical reaction was observed to coincide with the osteoidassociated Alcian blue staining; lower levels of extracellular staining were observed in the mesenchyme of the coronal suture, i.e. between the frontal and parietal osteoid plates. The osteopontin domains extended only slightly further into the sutural mesenchyme than the bone matrix. Cells positive for Fgfr2 transcripts were scattered along the surfaces of the frontal and parietal osteoid plates, but were most abundant as dense groups extending from the margin of each plate of matrix into the adjacent mesenchyme of the coronal suture as two overlapping lines, as shown by the BrdU-positive cells (Fig. 3C). This overlap of the marginal Fgfr2 expression domains of the frontal and parietal bones explains why there appears to be only a single line of Fgfr2 expression within the coronal suture in the whole-mount preparations (Fig. 2A,B).

Fig. 3. Fgfr2 expression and cell proliferation patterns in the E16 skull. (A) Whole-mount immunohistochemistry of BrdU (one hour uptake), skin removed. (B) Fgfr2 in situ hybridization and BrdU immunohistochemistry combined on a single specimen; blue shows Fgfr2 transcripts and red shows BrdU staining; the two staining patterns largely coincide, giving a purple colour. (C) Paraffin section (skin retained) showing BrdU immunohistochemistry (DAB detection) counterstained with Cason’s trichrome: large arrows mark the sutural borders of the frontal (f) and parietal (p) osteoid plates; blue-stained osteoblasts are attached to the osteoid; preosteoblasts are present between the margins of the matrix plates and the brownstained BrdU-positive osteogenic stem cells, which extend from each of the two future bone domains as indicated by the brackets (some are also present above the outer layer of osteoblasts: small arrows). Scale bars: A,B, 1 mm; C, 100 µm.

BrdU uptake was also observed on sections counterstained to show unmineralized bone matrix and osteoblasts/preosteoblasts. Osteogenic stem cells showing BrdU incorporation

Effect of FGF2-soaked beads on expression of Fgfr2 and osteopontin To test the effects of an FGF-induced increase in the level of FGFR signalling, we implanted beads soaked in human FGF2 onto the coronal suture of E15 mouse fetal skulls by ex utero surgery (Fig. 5A). The fetuses were left to develop for 48 hours, then removed and processed for in situ hybridization to detect transcription of Fgfr2 or osteopontin, or the two combined (Fig. 5B-D). All of the implanted FGF2-soaked beads were associated with induction of ectopic osteopontin expression; Fgfr2 expression was absent from the area of induced osteopontin expression, but encircled it. Beads soaked in PBS had no detectable effect on the expression patterns of either of the two genes (data not shown). The induced osteopontin expression was more strongly stained than the indigenous domains (Fig. 5D), and Fgfr2 expression was totally eliminated (Fig. 5C), suggesting that the applied FGF2 was present at a higher concentration than that of the endogenously available FGF2 protein. The effect of FGF2 bead implantation on BrdU uptake was

3380 S. Iseki and others Fig. 4. Serial transverse sections of the coronal suture region of the E16 head, with frontal (f) and parietal (p) bone domains (actual order of sections as cut: B, D, C, A). The sutural limit of each bone domain-associated staining reaction is indicated by arrows. (A) Alcian blue staining at pH 2.5 to show the distribution of polyanionic material, including proteoglycans; strongest staining is associated with the frontal and parietal plates of unmineralised bone matrix (osteoid). (B) Immunohistochemical detection of FGF2 protein, for which strongest immunoreactivity coincides with the bone matrix staining, with lower levels within the extracellular matrix of the sutural mesenchyme. (C) In situ hybridization for osteopontin transcripts, which are within osteoblasts where matrix is present, and preosteoblasts prior to their secretion of matrix, on the sutural side of the osteoid plate. (D) In situ hybridization for Fgfr2 expression, which is within osteogenic stem cells extending from the osteopontin domain into the suture as overlapping frontal and parietal plates, as well as in dispersed cells on the outer (and to a lesser extent the inner) surfaces of the plates of osteoid. b, brain; f, frontal bone; p, parietal bone; s, skin. Scale bar, 100 µm.

Fgfr2 in the normal mouse fetal skull vault (calvaria). The patterns observed from E15 to E18 were compared with those of indicators of cell proliferation and of skeletogenic differentiation, and with the localisation of FGF2 protein, a ligand with high affinity for FGFR2 which is locally abundant. The three major elements of the calvaria are the paired frontal and parietal bones and the unpaired interparietal (supraoccipital) bone. Since human mutations of FGFR2 primarily affect the coronal suture, the discussion will focus on this area.

also studied. The results showed increased BrdU uptake by osteoblasts, not only in the vicinity of the beads but over a much greater area of the skull surface. This observation (not illustrated) together with our preliminary observations on Fgfr1 expression suggests a complex response involving both receptors; it is the subject of a more detailed time course series of experiments, which will be reported in due course. DISCUSSION In this study, we have described the pattern of expression of

Patterns of development of the bones and sutures of the skull vault The coronal suture is established at the outset of osteogenesis as a narrow gap between the frontal and parietal osteopontin domains, in which there are proliferating cells that express Fgfr2. In contrast, all other presutural regions consist of broad areas of skeletogenic membrane expressing neither of these two genes, up to E18. These patterns within the normal fetal skull, and the predominance of the coronal suture in human FGFR2 mutation-associated craniosynostosis, suggest that normal FGFR2 signalling is essential for maintaining separate growth of the frontal and parietal bones from very early differentiation stages, whereas other factors and greater distances separate the skeletal elements bordering the midline and lambdoid sutures. Tissue interactions between the calvarial bones and the underlying brain may play important roles in their initial formation and pattern of growth. The frontal and parietal bones overlie the developing cerebral hemispheres, and the interparietal bone overlies the caudal part of the superior colliculus. As the caudal edges of the cerebral hemispheres grow caudally over the mesencephalon, the parietal and interparietal bones move closer together, thereby creating the lambdoid suture. In contrast, the two parts of the cerebral hemispheres underlying the frontal and parietal bones do not change their position relative to each other. The initial position of the coronal suture may coincide with the boundaries of gene expression domains in the expanding cerebral hemispheres, such as the caudal boundary of Brn2 expression (Alvarez-Bolado et al., 1995), or the rostral boundary of Otx1 (Shimamura et al., 1995). There is experimental evidence for a differentiation-inhibiting interaction between the dura mater and the skeletogenic membrane

FGF/FGFR2 signalling in skull development 3381

Fig. 5. Implantation of FGF2soaked beads onto the coronal suture by ex utero surgery. (A) Appearance of the E15 head immediately after implantation of the beads (arrowheads). (B-D) Whole-mount in situ hybridization for osteopontin (B), Fgfr2 (C), and both genes (D), 48 hours after bead implantation. Bead implantation on the sutural membrane induces osteopontin expression and displaces Fgfr2 expression to a ‘halo’ area (arrows on D) around the ectopic osteopontin domain.

in the developing sutures at late fetal and neonatal stages in the rat (Opperman et al., 1993), but interactions with the brain are not known. Fgfr2 expression, cell proliferation and differentiation: correlations with endogenous and ectopic FGF2 protein The relationship between Fgfr2 expression, BrdU incorporation and preosteoblast differentiation is the same in all five calvarial bones examined in this study: Fgfr2 expression is correlated with a high level of cell proliferation adjacent to, and mutually exclusive with, areas in which bone differentiation is being initiated. These observations suggest that FGF/FGFR2 signal transduction in the skeletogenic membrane has a mitosis-related function specifically within those cells that are becoming committed to undergo preosteoblastic differentiation. The identification of smaller numbers of cells expressing Fgfr2 on the outer (and to a lesser extent, the inner) surface of the developing bone matrix is consistent with a function in thickening and remodelling the expanding skull. These cells, whether at the margins or at the surface of the plates of bone matrix, are considered to be osteogenic stem cells. FGF2 protein is localized in a pattern that is largely reciprocal of Fgfr2 transcripts, being high in areas in which bone differentiation has been initiated and low within the sutural mesenchyme. In both sites, the signal is extracellular, being associated with the developing bone matrix and the mesenchymal extracellular matrix. It is likely that this staining represents available ligand, since exogenous FGF2 attached to heparin beads is clearly able to stimulate FGFR signalling. The bone matrix secreted by preosteoblasts includes a heparan sulphate proteoglycan (HSPG) (Fedarko et al., 1992). FGF2 binds avidly to HSPG (Folkman et al., 1988), forming a complex that is required for FGF-FGFR binding (Yayon et al., 1991). Biologically active FGF2-heparin and FGF2-heparan sulphate complexes are released from their sources both in vivo

and in vitro, and are able to diffuse freely through extracellular matrices to their sites of activity (Flaumenhaft et al., 1989, 1990). In the developing skull vault, it appears that FGF2 diffuses away from the osteoid plates to which it is bound, so that cell proliferation in the undifferentiated stem cells at the margins of the plates is associated with low levels of FGF2 protein and high levels of Fgfr2 transcripts, whereas osteogenic differentiation is associated with high levels of FGF2 and low levels of Fgfr2. Interestingly, the same is true of the skin, in which low FGF2/high Fgfr2 is seen in the basal layer of the epidermis, and the reciprocal pattern in the differentiated layers. The bead implantation experiments challenged this pattern by presenting high levels of FGF2 to the proliferating mesenchyme of the coronal suture. The effects indicate that increasing the ligand:receptor ratio alters the outcome of signal transduction in favour of osteogenic differentiation. A similar effect has been reported during teratocarcinoma cell differentiation, leading to the suggestion that cell surface FGF receptors are down-regulated as an autocrine response to FGF ligand binding (Moscatelli, 1994). The outcome of the FGF2 bead experiments mimics, on a local basis, the more generalised effects on the coronal suture of increased FGFR2 signalling resulting from gain-of-function mutations such as that of Crouzon syndrome. A particularly interesting feature of the bead implantation results was that the suture, as defined by high Fgfr2 expression and the absence of osteopontin expression, was displaced rather than obliterated. Induction of a high level of FGF2 within the suture appeared to inhibit extension of the differentiated edges of the frontal and parietal bones. This effect is most likely to be due to maintenance of the original stem cell population rather than dedifferentiation, since FGF2 beads placed within the differentiated areas did not show an Fgfr2 ‘halo’ (data not shown). It will be interesting to discover whether FGFR2 signalling in the skull sutures is analogous to that of FGFR3 in the growth

3382 S. Iseki and others

Fig. 6. Model for the relationship between FGF2, FGFR2, growth and differentiation in the early fetal coronal suture. (A) Normal development of the coronal suture: FGF2 bound to osteoid is released to diffuse into the surrounding extracellular matrix. Preosteoblasts differentiate where cells expressing FGFR2 are in a high FGF2 environment (adjacent to the deposited osteoid); down-regulation of FGFR2 expression prior to differentiation occurs through a paracrine feedback loop when high FGF2 levels increase FGFR2 activity. Cells in a lower FGF2 environment at a greater distance from the FGF2 source maintain FGFR2 expression and proliferative activity. Most of these cells are within the suture, resulting in appositional growth, but some are on the outer surface, resulting in increase of bone thickness. (B) Addition of ectopic FGF2 via a heparincoated bead mimics osteoid-attached FGF2, resulting in down-regulation of FGFR2 and differentiation to preosteoblasts; a domain of FGFR2 expression is maintained around the newly differentiated areas of skeletogenic membrane.

plates of long bones, in which FGFR3 appears to maintain a balance between proliferation (increased in Fgfr3 null mutants) and differentiation (increased in FGFR3 mutation) (Rousseau et al., 1994; Shiang et al., 1994; Tavormina et al., 1995; Colvin et al., 1996; Deng et al., 1996; Naski et al., 1996; Webster and Donoghue, 1996). A comparable mechanism is consistent as an explanation for the pattern of craniosynostosis observed in Crouzon syndrome, in which most or all sutures of the skull vault are fused by 30 months after birth (Cinalli et al., 1995). A model for the mechanism of calvarial growth and sutural maintenance A hypothesis concerning the relationship between FGF2, FGFR2, osteogenic differentiation and cell proliferation in the developing skull vault is presented diagrammatically in Fig. 6A. It is based on the following features, beginning in the sutural mesenchyme and progressing laterally from it. (1) Fgfr2 is expressed in osteogenic stem cells within the skeletogenic membrane, which are actively proliferating; their local extracellular environment has low but immunohistochemically detectable FGF2 levels. (2) Immediately adjacent to the sutural margins of the FGF2-rich osteoid plates, some cells have differentiated to preosteoblasts expressing osteopontin, in which Fgfr2 has been down-regulated. (3) Preosteoblasts become osteoblasts when they begin to secrete osteoid; FGF2 adheres to osteoid (presumably its heparan sulphate proteoglycan component) and is available at high levels for binding to

FGFR2 receptors on the surface of adjacent stem cells, resulting in a paracrine loop that leads to the down-regulation of Fgfr2, exit from the cell cycle and the initiation of osteoblastic differentiation. This arrangement ensures that new osteogenesis accrues to sites in which matrix production has already been initiated, at the margins of the plates for increase in area of the bone and on their surfaces for increase in thickness. Fig. 6B summarises the results of the bead implantation experiments, which show that high Fgfr2 signalling activity resulting in differentiation of stem cells to osteoblasts can be induced by the addition of ectopic FGF2 to the proliferating osteoblasts. Maintenance of a population of osteogenic stem cells (as indicated by Fgfr2 expression) around the island of osteoblasts (as indicated by osteopontin expression) induced by the FGF2 beads mimics the natural state in which a population of stem cells is maintained at the periphery of the differentiating bony plates and suggests an FGF2 concentration threshhold effect in the proliferation/differentiation switch. A similar intrasutural site of high signalling activity is suggested as the basis for formation of Wormian bones (small supernumerary calvarial bones) as a variant in normal skull development. An association between high levels of FGFR2 signalling and down-regulation of FGFR2 is also suggested by the finding that, in Crouzon syndrome, the coronal sutures show a lower than normal proportion of cells positive for FGFR2 antibody (Bresnick and Schendel, 1995). Our observations on the differential effects of locally different levels of FGF2 are

FGF/FGFR2 signalling in skull development 3383 also reflected in studies on type II thanatophoric dysplasia, which indicate that different levels of FGFR3 activity have qualitatively different biological consequences (Su et al., 1997). It is clear from the proposed model that a minor shift in the timing of the proliferation to differentiation switch in favour of differentiation will result in premature loss of the stem cell population. This shift can be induced by increased signalling, either by adding ligand, as here, or by gain-of-function mutation of the receptor. The key feature of these observations is that a high level of FGFR2 signalling results in negative feedback on Fgfr2/FGFR2 expression. The fact that mutations of FGFR1 (Muenke et al., 1994) and FGFR3 (Bellus et al., 1996) are also able to induce coronal craniosynostosis means that we must now investigate the localisation and functions of these proteins and their ligands during skull development. This study was funded by a Human Frontier Science Program postdoctoral fellowship to S. I., by a grant from Action Research to G. M. M.-K., and by Grants-in-Aid for Scientific Research from the Japanese Ministry of Education, Science and Culture 08557079 to K. E.; A. O. M. W. and J. K. H. are supported by the Wellcome Trust. We thank Yvonne Jones for helpful discussions throughout the course of this study.

REFERENCES Alvarez-Bolado, G., Rosenfeld, M. G. and Swanson, L. W. (1995). Model of forebrain regionalisation based on spatio-temporal patterns of POU-III homeobox gene expression, birthdates, and morphological features. J. Comp. Neurol. 355, 237-295. Bellus, G. A., Gaudenz, K., Zackai, E. H., Clarke, L. A., Szabo, J., Francomano, C. A. and Muenke, M. (1996). Identical mutations in three different fibroblast growth factor receptor genes in autosomal dominant craniosynostosis syndromes. Nature Genetics 14, 174-176. Bresnick, S. and Schendel, S. (1995). Crouzon’s disease correlates with low fibroblastic growth factor receptor activity in stenosed cranial sutures. J. Craniofac. Surg. 6, 245-248. Cinalli, G., Renier, D., Sebag, G., Sainte-Rose, C., Arnaud, E. and PierreKahn, A. (1995). Cronic tonsillar herniation in Crouzon’s and Apert’s syndromes: the role of premature synostosis of the lambdoid suture. J. Neurosurg. 83, 575-582. Colvin, J. S., Bohne, B. A., Harding, G. W., McEwen, D. G. and Ornitz, D. M. (1996). Skeletal overgrowth and deafness in mice lacking fibroblast growth factor receptor 3. Nat. Genet. 12, 390-397. Deng, C., Wynshaw-Boris, A., Shen, M., Daugherty, C., Ornitz, D. M. and Leder, P. (1994). Murine FGFR-1 is required for early postimplantation growth and axial organization. Genes Dev. 8, 3045-3058. Deng, C., Wynshaw-Boris, A., Zhou F., Kuo, A. and Leder P. (1996). Fibroblast growth factor receptor 3 is a negative regulator of bone growth. Cell 84, 911-921. El Ghouzzi, V. E., Le Merrer, M. L., Perrin-Schmitt, F., Lajeunie, E., Benit, P., Renier, D., Bourgeois, P., Bolcato-Bellemin, A-L., Munnich, A. and Bonaventure, J. (1997). Mutations of the TWIST gene in the SaethreChotzen syndrome. Nature Genet. 15, 42-46. Fedarko, N. S., Vetter, U. K., Weinstein, S. and Robey, P. G. (1992). Agerelated changes in hyaluronan, proteoglycan, collagen and osteonectin synthesis by human bone cells. J. Cell Physiol. 151, 215-227. Flaumenhaft, R., Moscatelli, D., Saksela, O. and Rifkin, D. B. (1989). Role of extracellular matrix in the action of basic fibroblast growth factor: matrix as a source of growth factor for long-term stimulation of plasminogen activator and DNA synthesis. J. Cell Physiol. 140, 75-81. Flaumenhaft, R., Moscatelli, D. and Rifkin, D.B. (1990). Heparin and heparan sulphate increase the radius of diffusion and action of basic fibroblast growth factor. J. Cell Biol. 111, 1651-1659. Folkman, J., Klagsbrun, M., Sasse, J., Wadzinski, M., Ingber, D. and Vlodavsky, I. (1988). A heparin-binding angiogenic protein – basic

fibroblast growth factor – is stored within basement membrane. Am. J. Pathol. 130, 393-400. Galvin, B. D., Hart, K. C., Meyer, A. N. and Webster, M. K. (1996). Constitutive receptor activation by Crouzon syndrome mutations in fibroblast growth factor receptor (FGFR)2 and FGFR2/Neu chimaeras. Proc. Natl. Acad. Sci. USA 93, 7894-9. Howard, T. D., Paznekas, W. A., Green, E. D., Chiang, L. C., Ma, N., Ortiz De Luna, R. I., Delgado, C. G., Gonzalez-Ramos, M., Kline, A. D. and Jabs, E. W. (1997). Mutations in TWIST, a basic helix-loop-helix transcription factor, in Saethre-Chotzen syndrome. Nature Genet. 15, 36-41. Johnson, D. E. and Williams, L. T. (1993). Structural and functional diversity in the FGF receptor multigene family. Adv. Cancer. Res. 60, 1-41. Mason, I. J. (1994). The ins and outs of fibroblast growth factors. Cell 78, 547552. Miki, T., Bottaro, D. P., Fleming, T. P., Smith, C. L., Burgess, W. H., Chan, A-M. L. and Aaronson, S. A. (1992). Determination of ligand-binding specificity by alternative splicing: two distinct growth factor receptors encoded by a single gene. Proc. Natl. Acad. Sci. USA 89, 246-250. Moloney, D. M., Wall, S. A., Ashworth, G. J., Oldridge, M., Glass, I. A., Francomano, C. A., Menke, M. and Wilkie, A. O. M. (1997). Prevalence of Pro250Arg mutation of fibroblast growth factor receptor 3 in coronal craniosynostosis. Lancet 349, 1059-1062. Moscatelli, D. (1994). Autocrine downregulation of fibroblast growth factor receptors in F9 teratocarcinoma cells. J. Cell Physiol. 160, 555-62. Muenke, M., Schell, U., Hehr, A., Robin, N. H., Losken, H. W., Schinzel, A., Pulleyn, L. J., Rutland, P., Reardon, W., Malcolm, S. and Winter, R. M. (1994). A common mutation in the fibroblast growth factor receptor 1 gene in Pfeiffer syndrome. Nature Genet. 8, 269-274. Muenke, M. and Schell, U. (1995). Fibroblast-growth-factor receptor mutations in human skeletal disorders. Trends Genet. 11, 308-313. Muneoka, K., Wanek, N., Trevino, C. and Bryant, S. V. (1990). Exo utero surgery. In Postimplantation Mammalian Embryos: A practical Approach (ed. A. J. Copp and D. L. Cockroft), pp41-59. IRL Press at Oxford University Press, Oxford. Naski, M. C., Wang, Q., Xu, J. and Ornitz, D. M. (1996). Graded activation of fibroblast growth factor receptor 3 by mutations causing achondroplasia and thanatophoric dysplasia. Nature Genet. 13, 233-237. Neilson, K. M. and Friesel, R. E. (1995). Constitutive activation of fibroblast growth factor receptor-2 by a point mutation associated with Crouzon syndrome. J. Biol. Chem. 270, 26037-26040. Neilson, K. M. and Friesel, R. E. (1996). Ligand-independent activation of fibroblast growth factor receptors by point mutations in the extracellular, transmembrane and kinase domains. J. Biol. Chem. 271, 25049-57. Nomura, S., Wills, A. J., Edwards, D. R., Heath, J. K. and Hogan, B. (1988). Developmental expression of 2ar (osteopontin) and SPARC (osteonectin) RNA as revealed by in situ hybridization. J. Cell Biol. 106, 441-450. Opperman, L. A., Sweeney, T. M., Redmon, J., Persing, J. A. and Ogle, R. C. (1993). Tissue interactions with underlying dura mater inhibit osseous obliteration of developing cranial sutures. Dev. Dynam. 198, 312-322. Ornitz, D. M., Xu, J., Colvin, J. S., McEwan, D. G., MacArthur, C. A., Coulier, F., Gao, G. and Goldfarb, M. (1996). Receptor specificity of the fibroblast growth factor family. J. Biol. Chem. 271, 15292-15297. Orr-Urtreger, A., Givol, D., Yayon, A., Yarden, Y. and Lonai, P. (1991). Developmental expression of two murine fibroblast growth factor receptors, flg and bek. Development 113, 1419-1434. Orr-Urtreger, A., Bedford, M. T., Burakova, T., Arman, E., Zimmer, Y., Yayon, A., Givol, D. and Lonai, P. (1993). Developmental localization of the splicing alternatives of fibroblast growth factor receptor-2 (FGFR2). Dev. Biol. 158, 475-486. Peters, K. G., Werner, G., Chen, G. and Williams, L. T. (1992). Two FGF receptor genes are differentially expressed in epithelial and mesenchymal tissues during limb formation and organogenesis in the mouse. Development 114, 233-243. Pryzlepa, K. A., Paznekas, W., Zhang, M., Golabi, M., Bias, W., Bamshad, M. J., Carey, J. C., Hall, B. D., Stevenson, R., Orlow, S., Cohen, M. M. Jr. and Jabs, E. W. (1996). Fibroblast growth factor receptor 2 mutations in Beare-Stevenson cutis gyrata syndrome. Nature Genet. 13, 492-494. Raz, V., Kelman, Z., Avivi, A., Neufield, G., Givol, D. and Yarden, Y. (1991). PCR-based identification of new receptors: molecular cloning of a receptor for fibroblast growth factors. Oncogene 6, 753-760. Rousseau, F., Bonaventure, J., Legeai-Mallet, L., Pelet, A., Rozet, J.M., Maroteaux, P., LeMerrer, M. and Munnich, A. (1994). Mutations in the gene encoding fibroblast growth factor receptor-3 in achondroplasia. Nature 371, 252-254.

3384 S. Iseki and others Shiang, R., Thompson, L. M., Zhou, Y. Z., Church, D. M., Fielder, T. J., Bocian, M., Winokur, S. T. and Wasmuth, J. J. (1994). Mutations in the transmembrane domain of FGFR3 cause the most common genetic form of dwarfism. Cell 78, 335-342. Shimamura, K., Hartigan, D. J., Martinez, S., Puelles, L. and Rubenstein, J. L. R. (1995). Longitudinal organization of the anterior neural plate and neural tube. Development 121, 3923-3933. Smallwood, P. M., Munoz-Sanjuan, I., Tang, P., Macke, J. P., Hendry, S. H. C., Gilbert, D. J., Copeland, N. G., Jenkins, N. A. and Nathans, J. (1996). Fibroblast growth factor (FGF) homologous factors: new members of the FGF family implicated in nervous system development. Proc. Natl. Acad. Sci. USA 93, 9850-9857. Su, W-C. S., Kitagawa, M., Xue, N., Xie, B., Garofalo, S., Cho, J., Deng, C., Horton, W. A. and Fu, X-Y. (1997). Activation of Stat1 by mutant fibroblast growth factor receptor activity in thanatophoric dysplasia type II dwarfism. Nature 386, 288-292. Tavormina, P. L., Rimoin, D. L., Cohn, D. H., Zhu, Y-Z., Shiang, R. and Wasmuth, J. J. (1995). Another mutation that results in the substitution of an unpaired cysteine residue in the extracellular domain of FGFR3 in thanatophoric dysplasia type 1. Hum. Mol. Genet. 4, 2175-2177. Webster, M. K. and Donoghue, D. J. (1996). Constitutive activation of fibroblast growth factor receptor 3 by the transmembrane domain point mutation found in achondroplasia. EMBO J. 15, 520-527. Webster, M. K. and Donoghue, D. J. (1997). FGFR activation in skeletal disorders: too much of a good thing. Trends Genet 13, 178-182. Wilkie, A. O. M., Morriss-Kay, G. M., Jones, E. Y. and Heath, J. K. (1995)

Functions of fibroblast growth factors and their receptors. Curr. Biol. 5, 500507. Wilkie, A. O. M. and Wall, S. A. (1996). Craniosynostosis: novel insights into pathogenesis and treatment. Curr. Opin. Neurobiol. 9, 146-152. Wilkinson, D. G. (1992). Whole mount in situ hybridization of vertebrate embryos. In In Situ Hybridization: A Practical Approach (ed. D. G. Wilkinson), pp75-83. IRL Press at Oxford University Press, Oxford. Wood, H. B., Ward, S. J. and Morriss-Kay, G. M. (1996). Effects of all-transretinoic acid on skeletal pattern, 5’HoxD gene expression, and RAR 2/ 4 promoter activity in embryonic mouse limbs. Dev. Genet. 18, 74-84. Yamaguchi, T. P. (1994). fgfr-1 is required for embryonic growth and mesodermal patterning during mouse gastrulation. Genes. Dev. 8, 30323044. Yamasaki, M., Miyake, A., Tagashira, S. and Itoh, N. (1996). Structure and expression of the rat mRNA encoding a novel member of the fibroblast growth factor family. J. Biol. Chem. 271, 15918-15921. Yayon, A., Klagsbrun, M., Esko, J. D., Leder, P. and Ornitz, D. M. (1991). Cell surface, heparin-like molecules are required for binding of basic fibroblast growth factor to its high affinity receptor. Cell 64, 841-848. Yayon, A., Zimmer,Y., Guo-Hong, S., Avivi, A., Yarden, Y. and Givol, D. (1992). A confined variable region confers ligand binding specificity on fibroblast growth factor receptors: Implications for the origin of the immunoglobin fold. EMBO J. 11, 1885-1890. (Accepted 11 July 1997)