BMP Signals Regulate Dlx5 During Early Avian Skull Development

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Available online at www.sciencedirect.com

Developmental Biology 257 (2003) 177–189 www.elsevier.com/locate/ydbio

BMP signals regulate Dlx5 during early avian skull development

Nicolas Holleville,a,1 Alexandra Quilhac,a,b,1 Martine Bontoux,a and Anne-He´le`ne Monsoro-Burqa,*

a Institut d’Embryologie Cellulaire et Mole´culaire, CNRS, UMR 7128, 49 bis, avenue de La Belle Gabrielle, 94736 Nogent-sur-Marne, France b Universite´ Paris VI, Pierre et Marie Curie, e´quipe Formations Squelettiques, CNRS UMR 8570, 2, place Jussieu, 75005 Paris, France Received for publication 22 July 2002, revised 16 December 2002, accepted 14 January 2003

Abstract The vertebrate skull vault forms almost entirely by the direct mineralisation of mesenchyme, without the formation of a cartilaginous template, a mechanism called membranous ossiﬁcation. Dlx5 gene mutation leads to cranial dismorphogenesis which differs from the previously studied craniosynostosis syndromes [Development 126 (1999), 3795; Development 126 (1999), 3831]. In avians, little is known about the genetic regulation of cranial vault development. In this study, we analyze Dlx5 expression and regulation during skull formation in the chick embryo. We compare Dlx5 expression pattern with that of several genes involved in mouse cranial suture regulation. This provides an initial description of the expression in the developing skull of the genes encoding the secreted molecules BMP 2, BMP 4, BMP 7, the transmembrane FGF receptors FGFR 1, FGFR 2, FGFR 4, the transcription factors Msx1, Msx2, and Twist, as well as Goosecoid and the early membranous bone differentiation marker osteopontin. We show that Dlx5 is activated in proliferating osteoblast precursors, before osteoblast differentiation. High levels of Dlx5 transcripts are observed at the osteogenic fronts (OFs) and at the edges of the suture mesenchyme, but not in the suture itself. Dlx5 expression is initiated in areas where Bmp4 and Bmp7 genes become coexpressed. In a calvarial explant culture system, Dlx5 transcription is upregulated by BMPs and inhibited by the BMP-antagonist Noggin. In addition, FGF4 activates Bmp4 but not Bmp7 gene transcription and is not sufﬁcient to induce ectopic Dlx5 expression in the immature calvarial mesenchyme. From these data, we propose a model for the regulatory network implicated in early steps of chick calvarial development. © 2003 Elsevier Science (USA). All rights reserved.

Keywords: Dlx5; Chick embryo; Skull; Intramembranous ossiﬁcation; BMP; Noggin; Msx1; FGFR; FGF4; Goosecoid

Introduction 2000 for a review). During evolution, the huge development of the telencephalic hemispheres coincides with the recruit- The vertebrate skull is formed by the precise assembly of ment of the neural crest cells to form the “new head” (Gans bony and cartilaginous elements which protect the brain, its and Northcutt, 1983; Le Douarin and Kalcheim, 1999). associated sensory organs, and the oral and nasal cavities, While the mesoderm-derived skeleton forms by endochon- while expanding to allow and follow brain growth during dral ossification, the bones of neural crest origin form either embryonic and postnatal life. Close interactions between the cartilage or directly ossify from a mesenchymal state form- developing skull and its underlying tissues, such as the dura ing “dermal” or “membranous” bones. The frontal, parietal, mater, control the formation and differentiation state of the exoccipital, and squamosal bones, which form the major edges of the bony plates, at the level of the cranial sutures part of the avian skull vault, all undergo membranous ossi- (Opperman et al., 1998; Kim et al., 1998; see Opperman, fication. They are surrounded by tissue layers also derived from the neural crest, namely the dura mater and the dermis (Couly et al., 1993). * Corresponding author. Department of Molecular and Cell Biology, Recent studies in the mouse have identified gene activ- 401, Barker Hall, University of California at Berkeley, Berkeley, CA ities implicated in the late stages of calvarial development, 94720, USA. Fax: ϩ1-510-643-6304. E-mail address: [email protected] (A-H. Monsoro-Burq). i.e., suture formation, maintenance, and obliteration (for a 1 These authors have contributed equally to this work. review, see Opperman, 2000). The open suture, formed by

Fig. 1. Dlx5 expression during chick skull development. (A, AЈ,E,F,I)E9;(B,BЈ, G, J) E12; (C, CЈ,CЉ, H, K) E16. (A–C) Dorsal views of in toto staining for bone (red) and cartilage (blue). (AЈ–CЉ) Transverse sections at levels indicated in (A–C). Arrows indicate the OFs, separated by the suture (s). (CЈ,CЉ) At E16, the frontal bones are thicker and separated by a reduced sutural mesenchyme (s, stained in pink at background levels). (D) Scheme of an E7 transverse section. Dlx5 expression is found in the deep mesenchyme (light grey) and the ectoderm (dark grey) and is absent from the dermis. The area enlarged in the following panels is boxed. (E–K) In situ hybridization with Dlx5 (E–H) and osteopontin (I–K) probes. (F/I), (G/J), and (H/K) are adjacent sections; (H) and (K) are stained with Alizarin red after the in situ procedure. (E) At E9, posteriorly to frontal bones, Dlx5 transcripts are found as two bilateral mesenchymal areas (one side is shown). (F–H) As differentiation proceeds, Dlx5 is detected all around the ossiﬁcation areas and is upregulated at the level of the major. (I) At E9, differentiating osteoblasts express the osteopontin gene. (J, K) Osteopontin is expressed in the differentiating osteoblasts but not in the mature osteocytes (embedded into the Alizarin red-stained bone matrix). f, p, n, so: frontal, parietal, nasal, and supraoccipital bones, respectively. dense connective tissue, is an area of growth between two bone development (cleidocranial dysplasia) or by premature osseous plates (Julian et al., 1957; Oudhof, 1982). The differentiation of the sutural mesenchyme (craniosynostosis proliferating cells are located laterally rather than in the syndromes). The genetic analysis of human and murine center of the suture area (Opperman et al., 1998; Iseki et al., syndromes allowed the identiﬁcation of mutations in vari- 1999). Lining the proliferating zone, the edges of the de- ous genes associated with suture development. FGFRs are veloping bones—or osteogenic fronts (OFs)—undergo dif- implicated in the Pfeiffer mutation (FGFR1-2 genes; Schell ferentiation (Decker and Hall, 1985). The suture area is et al., 1995; Bellus et al., 1996), the Crouzon, Jackson- progressively obliterated by the fusion of these adjacent Weiss and Apert syndromes (FGFR2/Bey locus and bones, this process being completed only during adulthood FGF3-4 genes; Jabs et al., 1994; Galvin et al., 1996; Ander- in humans. Alteration in this developmental process leads to son et al., 1998; Carlton et al., 1998), and the Muenke severe malformations of the skull vault, either by defective craniosynostosis (FGFR3; Muenke et al., 1997). TWIST has N. Holleville et al. / Developmental Biology 257 (2003) 177–189 179

In the mouse or rat embryo, Dlx5 and Dlx6 transcripts are present in immature chondrocytes, developing skull bones, and periosteum of more mature bones (Acampora et al., 1999). Both genes display closely related patterns of expression up to 12.5 dpc in mouse. Later on, Dlx6 expression decreases, whereas Dlx5 remains strongly in the periosteum and the skull (Acampora et al., 1999). In the Dlx5Ϫ/Ϫ mutant embryos, almost every bone of the skull and craniofacial structures is affected, either by primary or by second- ary defects. In particular, parietal and interparietal ossification is almost absent, suggesting that, in these structures, Dlx5 lack of activity is not compensated by another Dlx family member such as Dlx6 (Depew et al., 1999). Redun- dancy between these two genes has been demonstrated recently in the double knock out of both Dlx5 and Dlx6 that results into mouse embryos without calvaria, in addition to other severe skeletal defects (Robledo et al., 2002). How- ever, in vitro studies using murine cell lines have shown a series of contradictory results about Dlx5 expression and activity. Briefly, Dlx5 has been found in the differentiating calvaria osteoblasts. Both Dlx5 and osteocalcin gene expression are upregulated during the mineralization phase, but according to the cell line studied, Dlx5 overexpression either represses osteocalcin gene activity (Ryoo et al., 1997) or induces it (Miyama et al., 1999). DLX5 protein binds osteoblast-specific elements present in the Bone SialoPro- tein (BSP) promotor (Benson et al., 2000) and interacts with transcription factors of other families. In particular, Dlx5, Msx1, and Msx2 can bind one to another and antagonise their respective activities (Zhang et al., 1997). For example, Dlx5 suppresses Msx2 repression of osteocalcin gene transcription (Newberry et al., 1998) in accordance to their differential expression in proliferating (Msx2) and differentiating (Dlx5) osteoblasts (Ryoo et al., 1997). Finally, BMP2 upregulates Dlx5 expression in mouse whole embryos and in osteoblast in culture (Miyama et al., 1999). Fig. 2. Dlx5 is found in a subdomain of FGFR2 and Twist areas. (A–C) While the formation of the facial and hypobranchial Adjacent transverse E12 sections corresponding to the area boxed in Fig. skeleton has been extensively studied in avians, gene activ- 1D. Dlx5 overlaps with both Twist and FGFR2 around the bone (A, thick ities during skull vault development remain to be docu- arrow). FGFR2 and Twist extend further in the mesenchyme (B, thin arrow), and FGFR2 stains the broadest area, comprising the sutural mes- mented. In this study, we detail Dlx5 gene expression and enchyme (C, arrowhead). regulation during calvarial formation, in relationship to the BMP and FGF signalling pathways in the chick embryo. We focus on chick anterior frontal area development from E6 to been involved in the Saethre-Chotzen syndrome (el Ghouzzi E16, i.e., at early mesenchymal stage, during calvarial ini- et al., 1997) and MSX2 in Boston-type craniosynostosis tiation, bone differentiation, and suture formation. Dlx5 (Jabs et al., 1993). In vivo and in vitro studies conducted in expression profile was compared with the patterns of genes the mouse have defined a model for the genetic network that previously implicated in mouse cranial suture development, implicates FGF2/4-FGFR1-3, BMP4-Msx2, TGF␤ 2/3, and namely Bmp2, Bmp4, Bmp7, FGFRs, Msx1, Msx2, and Twist genes in the regulation of cell proliferation, apoptosis, Twist. Osteoblast differentiation was assessed by osteopon- and differentiation in the sutures and the adjacent OFs tin gene expression and bone matrix deposition. We show development (Iseki et al., 1997, 1999; Kim et al., 1998; Rice here that Dlx5 expression is initiated and reinforced in et al., 2000; Opperman et al., 2000). proliferating preosteogenic cells but not in the differentiated Dlx5 is a new candidate gene implicated in skull forma- osteoblasts. Using an explant culture system, we find that tion (Depew et al., 1999; Acampora et al., 1999). In the Dlx BMP2/4 signalling specifically regulates Dlx5 gene expres- family of homeobox genes, Dlx5 and Dlx6 present the sion. FGF4 upregulates Bmp4 expression level in the sub- unique property of being expressed in all skeletal elements. dermal mesenchyme but is not sufficient to induce Bmp7 or 180 N. Holleville et al. / Developmental Biology 257 (2003) 177–189

Dlx5. Finally, we show that Dlx5 and Msx1 partially over- (v/v) DMEM, 10% (v/v) foetal calf serum, 2% (v/v) chicken lap, and that a putative Dlx5 target gene, Goosecoid,is serum. A total of 76 explants were included in this study. expressed within the Dlx5-positive area. We integrate these Recombinant human BMP2 was secreted by a quail cell data in a model of the genetic network controlling early line containing the defective retrovirus pCRNCM-hBmp2 avian calvarial development. (Duprez et al., 1996). This vector was shown to direct production of hBMP2, which is biologically active in the chick embryo and mimicks both chBMP2 and chBMP4 Materials and methods activities (Monsoro-Burq et al., 1996). Xenopus Noggin protein was produced by a stable CHO line kindly provided Chick (Gallus gallus, JA 57 strain) E6–E16 embryonic by R. Harland (Lamb et al., 1993) and shown to block heads were dissected in ice-cold PBS and used either for BMP2/4 signalling in the chick (Monsoro-Burq and Le explant culture or ﬁxed for in toto and in situ hybridization Douarin, 2001). The other sources of recombinant factors or histological staining for mineralized tissues. In toto skel- were chick ﬁbroblast cells (O-line) transiently transfected eton staining was realized according to standard procedures, by RCAS retrovirus encoding mBmp4 (Duprez et al., 1996) cartilage being stained with Alcian blue and bone with or mFGF4 (Edom-Vovard et al., 2001) genes. Alizarine Red.

In situ and in toto hybridization Results For in toto analysis, the dorsal part of the skull was dissected from its ectoderm and treated as indicated in Dlx5 is expressed by immature pre-osteoblasts in vivo Henrique et al. (1995) with a 45-min proteinase K step. For nonradioactive in situ hybridisation, 7-␮m-thick paraffin The progress of differentiation of the skull was analysed sections were treated as described in Edom-Vovard et al. in entire heads (Fig. 1A–C) and transverse sections in the Ј Љ (2001). Double in situ hybridization was revealed by using nasal–frontal area (Fig. 1A –C ). Bone matrix deposition INT-BCIP. Some sections were counterstained with Aliza- began at E8.5–E9 in the lateral parts of the frontal and rin red. 35S-UTP-labelled in situ hybridization was de- squamosal bones (Fig. 1A) and progressed medially at E12 scribed in Monsoro-Burq et al. (1995). Antisense RNA (Fig. 1B). The parietal areas began to be ossified at E13 (not probes were synthesized as described previously for chMsx1 shown). The mineralization process was almost complete by and chMsx2 (Monsoro-Burq et al., 1995), chBmp2 and E16 (Fig. 1C). This oriented and progressive differentiation chBmp4 (Francis et al., 1994); chFGF-R1, -R2 (Wilke et al. was also observed in the frontal–nasal area. At E9, the first 1997); chFGF-R4 (Marcelle et al. 1994); chDlx5 (Pera et Alizarin red-stained structures were seen laterally to the al., 1999); chBmp7 (Wang et al., 1999); mFgf4 (Mahmood nasal septum, adjacent to the optic vesicles (Fig. 1AЈ), the et al., 1995); xnoggin, (Lamb et al. 1993); hBmp2 (Wozney bony plates progressed medially towards the sagittal plane et al. 1988); and chGoosecoid (Izpisua-Belmonte et al., and, to a lesser extent, laterally above the optic cups (Fig. 1993). The osteopontin and chTwist probes were kindly 1BЈ,CЈ, and CЉ). This area was chosen for further study on provided by P. Castagnola and M-C. Delfini. The osteocal- transverse sections. In this domain, the frontal bones formed cin fragment was amplified by PCR from published se- rather close to one another and differentiated rapidly (Fig. quences. 1AЈ–CЉ). The olfactory structures, the nasal septum cartilage, and the optic vesicle formed convenient anatomical Cell proliferation detection cues along the anterior–posterior (AP) axis, from early stages, before any bone formation can be seen (Fig. 1D). E8–E16 embryos were subjected to intracardiac injec- Bone development proceeded according to an anterior-to- tion of 1 mg/ml bromo-deoxy-uridine (BrdU) solution and posterior gradient, allowing to follow the sequential steps of incubated further for 2 h. Proliferating cells were detected calvarial development by analysing sections taken at differ- by using an HRP-anti-BrdU monoclonal antibody detection ent AP levels. Finally, the frontal bones partially overlapped kit (Amersham Life Sciences). with the posterior part of the nasal bones, thus allowing the study of several kinds of bone fusion domains (sutures). Cell and bead implantation into calvaria explants By in toto hybridisation on E9–E12 heads, the nasal and frontal bones were strongly Dlx5-positive. Posteriorly, only E9 heads were collected into cold PBS. The lower jaw the edges of frontal, parietal, and squamosal developing and the brain were removed but not the dura mater and the areas, i.e., around the coronal suture, were labelled (not ectoderm. Cell aggregates were implanted into the mesen- shown). At E9, the medial part of this posterior area was chyme of the presumptive calvaria as drawn in Fig. 5A. devoid of Dlx5 expression; it was thus chosen for the ec- Explants were cultivated for 36 h, either in serum-free topic expression assays (see below). Although the egg tooth Dulbecco’s modified Eagle’s medium (DMEM) or in 80% or nasal bones were strongly stained, the levels of expres- N. Holleville et al. / Developmental Biology 257 (2003) 177–189 181 sion in the other skull bones were low and not consistently tissue, presented lower BrdU labelling. At E12, the areas of obtained. most active proliferation in the skull forming area also We next analysed transverse sections from the E6–E16 coincided with the high Dlx5 expression (Fig. 3C and D). nasal–frontal area (Fig. 1AЈ–K). Adjacent sections were We counted dividing cells and Dlx5-expressing cells in each stained for Dlx5 (Fig. 1E–H) or the early osteoblast marker area. on several consecutive sections. At E9, the density of osteopontin (Fig. 1I–K) (Gotoh et al., 1990). We found that BrdU-positive cells was about three times higher in the Dlx5 osteocalcin gene expression appeared later than that of domain than in the saggital mesenchyme. In the Dlx5 area, osteopontin (not shown). At E6, we did not detect Dlx5 at E9, 70% of the BrdU-positive cells simultaneously ex- expression in the loose mesenchyme overlying the brain, pressed Dlx5 (n ϭ 315, at high magnification, overlapping even in the most anterior areas above the olfactory bulbs, Dlx5 and BrdU staining resulted in black-labelled cells, Fig whereas Dlx5 was strongly expressed in the mantle layer of 3B, D, and E; compared with light brown Dlx5-negative the developing brain (not shown). Dlx5 appeared at E7, as a proliferating cells in the dermis, Fig. 3F). We thus con- weak signal in the deep parts of dorsal mesenchyme, under cluded that Dlx5 was expressed by actively dividing preos- the future dermis, rather broadly distributed (Fig. 1D). At teoblasts. E8 in anterior areas, and later in more posterior parts, Dlx5 expression was localised as two strong and symmetrical Relationships between Dlx5 expression and BMP and spots, located on each side of the sagittal plane, in the place FGF signalling pathways where the bone blastema would form (Fig. 1D and E); this area was devoid of osteopontin expression (not shown). At In order to understand Dlx5 gene regulation during early E9, in anterior areas where bone formation had begun (Fig. skull development, we compared its expression with that of 1AЈ and I), Dlx5 was excluded from the differentiated os- a series of molecules known to act in calvarial development teoblasts (compare Fig. 1F with the adjacent section in Fig. in other vertebrate species. To our knowledge, this is the 1I). Dlx5 was expressed around the mineralized structure, in first description of these gene patterns, during calvaria de- the undifferentiated mesenchyme. The highest levels of velopment in the chick. In particular, we were looking for expression were found where the bone followed its major factor(s) acting upstream of Dlx5, whose pattern would directions of growth (Fig. 1F). This expression pattern was account for Dlx5 initial restriction as two distinct bilateral maintained from E12 to E16: Dlx5 was found in the mes- spots. The analysis of serial alternate sections allowed us to enchymal tissues around the bony pieces, at the main OFs compare precisely the expression of the different genes in (Fig. 1G and H). Osteopontin labelled the cells which were spite of their dynamic expression profiles (Fig. 4A, sche- not yet embedded in the bony matrix (Fig. 1J and K). We matic of a transverse section). did not detect Dlx5 staining in the midsutural mesenchyme We focused first on BMP signalling and analysed the at any of these stages. expression of the Bmp2, Bmp4, and Bmp7 genes from E6 to We then compared Dlx5 expression with that of Twist E16 in the frontal and nasal areas. Bmp4 transcripts were and FGFR2, two genes recorded to be expressed in the already detected in the neural crest-derived mesenchyme mouse sutural mesenchyme. In adjacent E12 sections, Dlx5 from E3 to E5 (not shown). At E6, Bmp4 was found in the was localised in the Twistϩ, FGFR2ϩ areas close to the entire dermis area and the meninges but not the deeper differenciated bone (Fig. 2A–C). However, both Twist and bone-forming area (not shown). Bmp7 was not expressed in FGFR2 were also expressed in the mesenchymal cells lo- the dorsal mesenchyme at that stage, but was present in the cated further apart from the bone elements. Notably, both meninges around the brain. Bmp2 staining was faint and are found in the mesenchyme between the nasal and frontal rather uniformly distributed in the neural crest-derived mes- bones, and FGFR2 was found in the medial part of the enchyme. At E7, Bmp4 was largely expressed in the dermis sutural mesenchyme (Fig. 2). and Bmp7 in the ectoderm. The deeper mesenchymal areas around the nasal cartilage also expressed strong levels of Dlx5 is expressed by proliferating preosteoblasts Bmp4 and began to express Bmp7 transcripts (Fig. 4B). This area corresponded to the earliest domain expressing Dlx5 in The observation that Dlx5 was upregulated before osteo- the future skull (Fig. 1D), thus suggesting that BMP7, or a blast differentiation and mostly in growing areas suggested combination of BMP4 plus BMP7 signals, could play a role that it might correlate with proliferating osteoblasts precur- in the initiation of Dlx5 transcription. Supporting this idea, sors. We tested this hypothesis by simultaneously staining Dlx5 was also activated in a subset of the ectodermal areas for Dlx5 and BrdU incorporation. At E9 and E12, the dermis positive for Bmp7 gene, whereas no Dlx5 expression was and ectoderm proliferated rapidly (Fig. 3A, C, and F). In the found in the Bmp4ϩ but Bmp7Ϫ dermis. As development E9 deep mesenchyme, the immature posterior frontal blast- proceeded, gene expression was reinforced and organized emas were formed as two bilateral areas of highly dividing around the differentiating osseous elements. At E9, in the cells, which overlapped with the Dlx5-positive areas (Fig. posterior parts of the still mesenchymal frontal progenitors, 3A and B). The Dlx5-negative mesenchyme located at the Dlx5 pattern overlapped again with the intersection of Bmp4 midline (suture), as well as the other parts of this deep and Bmp7 domains, all the expression levels being rein- 182 N. Holleville et al. / Developmental Biology 257 (2003) 177–189

Fig. 3. Dlx5 expressing cells are proliferating. The BrdU staining (brown) overlaps with the Dlx5-positive areas (blue) in the prospective skull mesenchyme at E9 (A, enlargement in B, E) and E12 (C, enlargement in D). (A, B, E) The E9 section is taken at a posterior level of the frontal area, where no bone or osteopontin expression is seen yet: the whole bone blastemas express Dlx5. Most of the cells that have just divided also express Dlx5 (arrows in E). A high rate of proliferation is also present in the dermis (d), which is negative for Dlx5 expression (A, F). (C, D) At E12, Dlx5 strongest staining is found in the highly proliferative areas, medial to the OFs (arrows). Note the absence of Dlx5 staining in the midsutural mesenchyme(s). Scale bars: (A, C) 416 ␮m; (B, D) 104 ␮m; and (E, F) 2 ␮m. forced as compared with E7 patterns (Fig. 4C and E is Vovard et al., 2001), suggesting either no or very low adjacent to Fig. 1E). More anteriorly, at E9, E12, or E16, the expression levels in the skull-forming mesenchyme. Thus, same general features were observed around the differenti- we analysed the expression of chick FGF receptors and ating osseous elements, in spite of bone growth and varia- compared it with that of Dlx5 or Bmp genes using serial tions between the very flat nasal bones and the thicker alternate sections. At E7, in undifferentiated areas, FGFR-1 frontal bones (Fig. 4D and F is adjacent to Fig. 1H and K). gene expression was found in the dermis and the deep Bmp4 was found almost all around the bone, in the undif- mesenchyme, but excluded from the ectoderm. FGFR-1 was ferentiated cells lining it, with increased staining at the weaker but overlapping to Bmp4 signals (Fig. 4B). At E9, growing Dlx5-positive ends (Fig. 4D). Bmp7, in contrast, FGFR-1 expression was localized adjacent to the Bmp4- was essentially restricted to these ends, with positive cells positive area (Fig. 4G). In contrast, at E9, FGFR-2 was not found in the mesenchyme close to the bone (Fig. 4F). present in the dermis or in the FGFR-1-expressing areas. Around the rest of the bone, only a very faint Bmp7 staining Instead, it was present at low levels all around these do- was observed. Thus, at the tip of the bony plate, Bmp7 had mains (Fig. 4I). From E12 onwards, FGFR-1 and FGFR-2 a very similar pattern to that of Dlx5, but Dlx5 also extended staining were enhanced: FGFR-1 was present all around the farther and at higher levels in the mesenchyme located bone and in the recently differentiated cells already sur- around the bone. All over these stages, Bmp2 was ubiqui- rounded by bone matrix (Fig. 4H), overlapping with os- tously expressed at low levels (although slightly above teopontin expression domain (not shown). FGFR-2 expres- background levels). In summary, from E6 to E16, Bmp4 and sion was found surrounding the bones, external to the Bmp7 delineated the forming bones, with enhanced expres- FGFR-1-positive area, in mesenchymal cells negative for sion at the OFs. Dlx5 profile comprised these domains and osteopontin (Fig. 2C). From E7 to E16, FGFR-4 was not extended further into the surrounding mesenchyme. expressed in the skull-forming mesenchyme. However, it Looking for FGF expression during skull development, strongly stained cartilage elements and muscles (Fig. 4J). In we did not detect Fgf4 and Fgf8 gene expression by in situ summary, at early stages, FGF receptor transcripts were not hybridization, although the experiments were sucessfully found in the Dlx5-positive areas at significant levels; later made in parallel with positive chick limb sections (Edom- on, they were located in a subset of the mesenchymal cells Fig. 4. Bmp4, Bmp7, FGFR-1, and FGFR-2 expression in developing frontal bone. (A) Schematic of a transverse section in the frontonasal area. The area shown in (C–J) is indicated by the box. (B) Schematic of Bmp gene expression at E7. (C, E, G, I are adjacent to Fig. 1E). (C) At E9, Bmp4 transcripts are found in the dermis (d) and mesenchymal bone progenitors (arrows). (D) At E16, Bmp4 expression is maintained all around the developing bone, with more intense areas such as the dorsal side of the bones (asterisks). (E) At E9, Bmp7 expression is strong in the skeletogenic areas of the deep mesenchyme, overlapping with Bmp4 domain (arrows). (F) At E16, Bmp7 is expressed in the mesenchymal cells surrounding OFs (arrowheads). (G) In E9 areas where bone formation began, a weak FGFR-1 expression appears, adjacent to the Bmp4- and Bmp7-positive domains (arrows). FGFR-1 is also found in the overlying dermis. (H) At E16, FGFR-1 is strongly expressed in the differentiating cells, both in a thin layer of mesenchyme lining the bone and in cells embedded into the Alizarin red-stained matrix. (I) At E9, FGFR-2 is low in the areas where the bone appeared (red), which correspond to the FGFR-1-positive areas (arrows). A widely distributed signal is present in the rest of the mesenchyme. (J, E16) At none of these stages are FGFR-4 transcripts found in the calvaria (b, bone). They are present in muscles and cartilage elements. n.s., nasal septum (F, H, L are counterstained with Alizarin red). 184 N. Holleville et al. / Developmental Biology 257 (2003) 177–189 around the bone, in the Dlx5-positive area, without en- was the same as these with BMP sources. Around the hanced expression in the sites of highest Dlx5 staining. Dlx5 implant, a strong Bmp4 gene expression was induced (Fig. is thus found in a large domain around the bone, comprising 5F, n ϭ 5/8) as compared with the normal expression level the FGFR-1(ϩ)/osteopontin(Ϫ) area and included in the in this area (Fig. 5H). However neither Dlx5 (Fig. 5G) nor FGFR-2(ϩ) domain, thus Dlx5 is likely to be expressed Bmp2 or Bmp7 (not shown) were induced by mFGF4 (n ϭ over several determination/differentiation steps in the osteo- 0/8). Together, these data show that Dlx5 induction is spe- blast lineage. cifically regulated by BMP signals. This was achieved ex- perimentally by providing high amounts of BMP2 in an Dlx5 is regulated by BMP but not by FGF4 signals immature area.

We analysed the regulation of Dlx5 gene expression Goosecoid and Msx1 are also present in the early using explant culture of head dorsal tissues. We used either calvarial structures cell lines secreting hBMP2, xNoggin, or cells transiently transfected with RCAS viruses and secreting mBMP4 or The Dlx5 mutant mouse presents reduced Goosecoid mFGF4. The effect of overexpressing BMPs was tested by gene expression in frontonasal and branchial areas and their implanting the cells posterior to the presumptive frontal facial skeleton has similarities to that of the Goosecoid bones, in an immature area devoid of Dlx5 expression (Fig. defective embryos, suggesting that Dlx5 might regulate 5A). BMP2 overexpression was followed by the induction Goosecoid gene transcription (Depew et al., 1999). Looking of ectopic Dlx5 expression around the graft (n ϭ 20/22), for potential targets of Dlx5 in the skull progenitors, we whereas control cells never caused Dlx5 overexpression found that Goosecoid was expressed at E12, in the undif- (Fig. 5B). Sections showed induction both in the deep mes- ferentiated mesenchymal cells around bone, in a pattern enchyme, under the dermis and in the ectoderm (Fig. 5C). similar to that of Dlx5, although at lower levels (Fig. 6A and This result was obtained with the BMP2 cell line, which B). Goosecoid was also present in the dermis (not shown) secretes high amounts of BMP2 protein, whether or not and in the mesenchyme expressing Twist and FGFR2 (Fig. serum was added in the culture medium (n ϭ 12/13 cases of 2). At E9, we did not detect any significant levels of induction without serum). In two cases, the BMP2 implant Goosecoid transcripts in the early mesenchymal proliferat- also induced an osseous-like structure expressing low levels ing cells expressing Dlx5. We also tested whether BMP2- of the osteopontin gene (not shown). When the implants induced Dlx5 ectopic expression was sufficient to induce were inserted in an even more immature area, such as the Goosecoid ectopically, but did not observe any induction in center of the coronal suture, no induction was observed (n ϭ these conditions (not shown), suggesting the need for addi- 4/4, not shown). If the mBMP4-RCAS-transfected cells tional factors and/or acting on more mature cells. were grafted, we did not observe effective Dlx5 induction (n DLX5 protein was shown to interact with MSX1 and ϭ 1/6). This might be due to the lower levels of protein MSX2 factors and to modulate target gene activity. We produced under the control of the RCAS promotor, com- analysed Msx gene expression in the chick nasal–frontal pared with the CMV promotor, as shown in other circum- area from E7 to E16. Msx1 was continuously expressed in stances. Alternatively, exogenous BMP2 might be able to the neural crest-derived mesenchyme from early E3 stages mimick the effects of both BMP7 and BMP4, whereas (not shown). At E6 and E7, it was found in the deep BMP4 alone could not do so. mesenchymal domain expressing Bmp4 (not shown, as in The requirement for BMP signalling for Dlx5 expression Fig. 4B). At E9, Msx1 was not found in the posterior was tested by implanting Noggin-secreting cells into a more proliferating Dlx5–Bmp-positive mesenchyme. At anterior mature area, nearby where the frontal bones were undergo- levels, however, where bone had started to form, Msx1 was ing differentiation, but still into undifferentiated tissues found expressed in a pattern very similar to that of Bmp4, (Fig. 5A). The blocking of BMP signalling was followed by i.e., in a domain included in the larger Dlx5-positive area. a strong downregulation of Dlx5 expression, especially in Msx1 pattern presented subtle differences to that of Dlx5. the most immature areas, towards the saggital plane (n ϭ For example, Msx1 was not detected at the tips of the bony 5/7, Fig. 5D and E, compare the left grafted side with the plates (Fig. 6A and C), but rather in the areas of lower Dlx5 nonoperated right side in D). Control cells did not prevent activity, along the dorsal face of the osseous plate, as was the onset of Dlx5 expression. These two results show that Bmp4. These overlapping but distinct patterns suggest that BMP signals are necessary and sufficient for Dlx5 induc- differential interactions between DLX5 and MSX1 proteins tion. When the Noggin cells were implanted into areas occur locally, around the bony elements. This might lead to already expressing Dlx5, no downregulation was noticed: a fine control of the skull morphogenesis. Finally, we tried BMPs are not likely to take part in the maintenance of Dlx5 to analyse whether MSX1 could directly control Dlx5 gene expression (not shown). expression by implanting cells transfected with a competent Since several FGF receptors were found in the areas chMsx1-RCAS retrovirus (D Duprez, unpublished observa- where Dlx5 was activated, we overexpressed mFGF4 in the tions). Unfortunately, the Dlx5-negative implantation area immature posterior area (Fig. 5A). The position of the graft also presented low levels of proliferation; this did not allow N. Holleville et al. / Developmental Biology 257 (2003) 177–189 185

the spread of the virus into the host tissues and thus pre- vented us from doing this analysis. By nonradioactive in situ hybridization, in spite of high expression levels in other tissues, we did not detect Msx2 expression in the presumptive skull before E16: it was then found in the remodeling bone (not shown). It was only by increasing sensitivity by using the radioactive in situ hybridization technique that we found a faint Msx2 expression in the E10–E12 midsaggital mesenchyme, and lining some skeletal elements (Fig. 6D). Msx2 was not detected at the OFs or in other Dlx5-positive areas.

Discussion

Dlx5 is expressed at several stages of osteoblasts development

Our results provide a scheme of the gene expression changes during progressive osteoblast development (Fig. 7A). In the mouse embryo, FGFRs are differentially expressed at different stages of osteoblast development. FGFR-2 and Twist are present on proliferating precursors, whereas early differentiating cells are associated with FGFR-1 transcripts (Anderson et al., 1998; Iseki et al., 1999; Fragale et al., 1999). We show here that Dlx5 is expressed by at least three distinct populations of the osteoblast lineage. FGFR-2 and Twist expression are found in the nondifferentiated sutural mesenchyme and surrounding the bones, thus in a larger domain than that of Dlx5 (Figs. 1, 2, and 6A). This suggests that the early population of Dlx5ϩ cells corresponds to the more mature FGFR-2ϩTwistϩ cells. Closer to the bone, Dlx5 overlaps with Bmp7 expression, deﬁning a second Dlx5ϩ population. Finally, FGFR-1 is found in a thin layer external to the mineralised bone as well as in the recently embedded osteopontin-positive cells (Figs. 1 and 4). The most external part of the FGFR-1-

Fig. 5. Regulation of Dlx5 by BMP2, Noggin, and FGF4. (A) Schematic of the grafts in the head explants. Control immature mesenchyme (orange) does not express Dlx5, whereas the more mature area (green) would normally express Dlx5 (blue domains) by the time of ﬁxation. (B) Dlx5 is induced (orange arrow) after implantation of BMP2-secreting cells but not around control cells (black arrow), both grafted in the posterior immature area (asterisk indicate the cell aggregates). (C) On transverse sections through an immature area grafted with BMP2 (orange arrow, Bmp2 mRNA in red), Dlx5 is induced in the ectoderm (arrowhead) as well as in the deep mesenchyme (arrows). (D) After the implantation of Noggin-producing cells in the more mature area (green arrow, Noggin mRNA in red), Dlx5 expression is strongly downregulated around the graft, especially in the medial area (underline in “a”). On the contralateral side, a robust Dlx5 expression is detected (underline “b”, the saggital plane is indicated). (E) General view of (D). The saggital plane is positioned according to the brain hemispheres. (F) After the graft of FGF4-secreting cells (orange arrow, FGF4 mRNA in red), a strong induction of Bmp4 expression is observed (arrows, compare with the low or lack of staining in control areas (H). (G) In contrast, similar FGF4 implants (orange arrow) do not induce ectopic expression of Dlx5. 186 N. Holleville et al. / Developmental Biology 257 (2003) 177–189

proliferating FGFR2ϩ/FGFR1Ϫ mesenchymal population. Moreover, in these cell lines, DLX5 function might be modiﬁed by the presence of interacting proteins, such as MSX1 or MSX2, which are negative modulators of DLX5 activity (Zhang et al., 1997; Newberry et al., 1998). We showed here that Msx1 gene was expressed in a subdomain of the Dlx5-positive area, supporting the idea that local and mutual modulation of these gene activities occurred around the developing bone.

Dlx5 gene regulation in the avian calvarial mesenchyme

Several signalling cascades have been described at the level of the mouse osteogenic fronts. We show here that Dlx5 is part of the BMP signalling pathway regulating calvarial development. Dlx5 is specifically induced by BMP signalling but not by FGF signals. These results when high levels of BMP2 were provided, suggesting that the endogenous BMP signalling, might be achieved differently. Het- erodimers of BMPs act more potently than homodimers in several instances. BMP2/BMP7 or BMP4/BMP7 heterodimers have been implicated in several developmental mechanisms (Suzuki et al., 1997; Nishimatsu and Thomsen, 1998; Schmid et al., 2000). Notably, BMP4/7 heterodimers potently induce intramembranous bone formation in vitro and in ectopic locations in adult rats (Aono et al., 1995). Consistent with these data, we see high Dlx5 activation where Bmp4 and Bmp7 genes are coexpressed. Depriving the tissues of BMP2/BMP4 activity locally blocked Dlx5 activation in the osteogenic mesenchyme. It would be in- teresting to do the converse experiment, blocking BMP7 activity by using a potent source of follistatin, a specific BMP7 antagonist, to analyse the requirement for this factor on Dlx5 expression. BMP and FGF pathways activate a common target, Fig. 6. Expression of Dlx5, Goosecoid, Msx1, and Msx2 at E12. Adjacent Msx1, in several neural crest-derived tissues: murine cal- E12 transverse sections are stained for Dlx5 (A), Goosecoid (B), or Msx1 varia and tooth mesenchyme (Kim et al., 1998; Bei and (C) gene expression. (B) Goosecoid is detected in a pattern similar to that of Dlx5 (A), although more diffuse. (C) In this area, Msx1 is present at the Maas, 1998). Although we did not detect Fgf4 or Fgf8 gene dorsal and ventral edges of the bony plate, but absent from the OFs expression, a recent study localised endogenous FGF2 pro- (arrows). (D) Radiolabeled in situ hybridization evidences Msx2 transcripts tein in the chick calvarial mesenchyme, showing elevated in the ventral part of the bony plate (arrows). None of these genes is levels in the nondifferentiated mesenchyme, around the expressed in the midsutural mesenchyme (s). bones and in the sutural mesenchyme (Moore et al., 2002). Since we found that FGFR-1 and FGFR-2 were expressed expressing area overlaps with the Dlx5 domain, whereas from E6–E7 onwards, in all the neural crest-derived mes- Osteopontinϩ/FGFR-1ϩ cells surrounded with bone matrix enchyme then at the OFs, FGF signalling is also likely to be do not (Fig. 7A). Thus, Dlx5 is not present in the undiffer- active at various steps of calvarial formation. However, our entiated sutural mesenchyme and is present in proliferating experimental overexpression of FGF4 did not activate Dlx5 but not differentiating preosteoblasts, then in early osteo- in the early sutural mesenchyme. In mice embryos, FGF2 blasts. When osteoblast cell lines were derived from various activates Twist gene expression in the midsutural mesen- tissues, they were likely to be at one of these stages. The chyme and at the OFs (Rice et al., 2000). We show here upregulation of the osteocalcin gene by Dlx5 observed by that, in the chick, Dlx5 and Twist patterns overlap in the area Miyama et al. (1999) could reflect DLX5 activity in the of most intensive cell proliferation, near the OFs. However, FGFR-1-positive subpopulation, while the repression of os- Twist is more largely distributed and included in the larger teocalcin by DLX5 (Ryoo et al., 1997) might occur in the FGFR-2ϩ domain. N. Holleville et al. / Developmental Biology 257 (2003) 177–189 187

Fig. 7. Gene expression and regulation during chick calvaria development. (A) We have summarised the overlapping expression patterns analysed here and integrated them with the different steps of the differentiation of the osteoblasts. (B) A model of the gene interactions implicated in Dlx5 regulation.

Early relationship between BMP and FGF pathways expression results in the downregulation of Bmp4 expression in the growth plate cartilage and block endochondral BMP and FGF signalling have been implicated sepa- bone growth (Naski et al., 1998). This suggests that differ- rately in suture development in the mouse. Here, we have ent regulatory responses can occur in the different types of shown that at earlier stages, in the yet undifferentiated head skeletogenic differentiation and/or at different stages of mesoderm, FGF4 was able to activate Bmp4 but not Bmp7 these differentiation processes. gene expression. FGFR-1 and FGFR-2 expression patterns This study has focused on Dlx5 regulation from the early from E6 to E7 are consistent with a role of this pathway in steps of the osteogenic mesenchyme formation up to suture Bmp4 activation and/or maintenance from these early formation in the avian embryo. We propose a model inte- stages. In another system, it was shown that FGF signalling grating our results (Fig. 7B, our data are shown in black) modulated negatively Bmp4 gene expression: FGFR-3 over- together with data obtained in the mouse embryo which are 188 N. Holleville et al. / Developmental Biology 257 (2003) 177–189 likely to be modulate the BMP-Dlx5 pathway in avians as fibroblast growth factor receptor genes in autosomal dominant cranio- well (grey, Kim et al., 1998 and Opperman, 2000; hypo- synostosis syndromes. Nat. Genet. 14, 174–176. thetical interactions are italicized, or indicated in green). Benson, M.D., Bargeon, J.L., Xiao, G., Thomas, P.E., Kim, A., Cui, Y., Franceschi, R.T., 2000. Identification of a homeodomain binding ele- The factor(s) responsible for Bmp7-localised activation in ment in the bone sialoprotein gene promoter that is required for its the deep mesenchyme is still unknown, but our results osteoblast-selective expression. J. Biol. Chem. 275, 13907–13917. showed that high levels of FGF4 were not sufficient to do Carlton, M.B., Colledge, W.H., Evans, M.J., 1998. Crouzon-like craniofa- so; we hypothesize that non-FGF signal might be involved. cial dysmorphology in the mouse is caused by an insertional mutation In vivo, the combination of BMP7 and BMP4 (with a at the Fgf3/Fgf4 locus. Dev. Dyn. 212, 242–249. Couly, G.F., Coltey, P.M., Le Douarin, N.M., 1993. The triple origin of potential contribution of ubiquitously expressed BMP2) is skull in higher vertebrates: a study in quail-chick chimeras. Develop- likely to induce Dlx5 in localised areas of the head mesen- ment 117, 409–429. chyme expressing FGFR2 and Twist. These restricted areas Decker, J.D., Hall, S.H., 1985. Light and electron microscopy of the new present a high rate of proliferation and are fated to become born sagittal suture. Anat. Rec. 212, 81–89. bone. In parallel, FGF2/4 and BMP4 have been shown to Depew, M.J., Liu, J.K., Long, J.E., Presley, R., Meneses, J.J., Pedersen, induce Msx1, at later stages of calvarial development (Kim R.A., Rubenstein, J.L., 1999. Dlx5 regulates regional development of the branchial arches and sensory capsules. Development 126, 3831– et al., 1998). Although understanding the role of the DLX5 3846. protein in the DLX5–MSX1 interactions in this system Duprez, D., Bell, E.J., Richardson, M.K., Archer, C.W., Wolpert, L., awaits further studies, we propose that these interactions Brickell, P.M., Francis-West, P.H., 1996. Overexpression of BMP-2 would establish a regulatory loop that modulate the prolif- and BMP-4 alters the size and shape of developing skeletal elements in eration level versus the differentiation status of the osteo- the chick limb. Mech. Dev. 57, 145–157. blasts, ultimately leading to the proper morphogenesis of Edom-Vovard, F., Bonnin, M.A., Duprez, D., 2001. Misexpression of Fgf-4 in the chick limb inhibits myogenesis by down-regulating Frek the skull bones. expression. Dev. Biol. 233, 56–71. el Ghouzzi, V., Le Merrer, M., Perrin-Schmitt, F., Lajeunie, E., Benit, P., Renier, D., Bourgeois, P., Bolcato-Bellemin, A.L., Munnich, A., Acknowledgments Bonaventure, J., 1997. Mutations of the TWIST gene in the Saethre- Chotzen syndrome. Nat. Genet. 15, 42–46. We thank T. Grammer, K. Liu, and J. Wallingford for Fragale, A., Tartaglia, M., Bernardini, S., Di Stasi, A.M., Di Rocco, C., Velardi, F., Teti, A., Battaglia, P.A., Migliaccio, S., 1999. Decreased their critical reading of the manuscript. We also thank Drs. proliferation and altered differentiation in osteoblasts from genetically Brickell, Castagnola, Delphini, Duprez, Harland, Houston, and clinically distinct craniosynostotic disorders. Am. J. Pathol. 154, Kessel, Marcelle, Mahmood, and Richman for the generous 1465–1477. gift of probes and other materials. We thank also C. Vincent Francis, P.H., Richardson, M.K., Brickell, P.M., Tickle, C., 1994. Bone for sharing Figs. 1A, and B, S. Gournet, M. Fromaget, and morphogenetic proteins and a signalling pathway that controls patterning in the developing chick limb. Development 120, 209–218. F. Beaujean for help in the illustration work. This study was Galvin, B.D., Hart, K.C., Meyer, A.N., Webster, M.K., Donoghue, D.J., supported by Association de Recherche contre le Cancer 1996. Constitutive receptor activation by Crouzon syndrome mutations (grant #9319) and Hoescht-Marion-Roussel funding to (A- in fibroblast growth factor receptor (FGFR)2 and FGFR2/Neu chime- H.M-B.) and by general support from Centre National de la ras. Proc. Natl. Acad. Sci. USA 93, 7894–7899. Recherche Scientifique (CNRS) and Colle`ge de France to Gans, C. Northcutt, R.G., 1983. Neural crest and the evolution of verte- the UMR 7128. N.H. is a CNRS/Hoescht-Marion-Roussel brates a new head. Science 220, 268–274. Gotoh, Y., Gerstenfeld, L.C., Glimcher, M.J., 1990. Identification and and FRM fellow. characterization of the major chicken bone phosphoprotein. Analysis of its synthesis by cultured embryonic chick osteoblasts. Eur. J. Biochem. 187, 49–58. References Henrique, D., Adam, J., Myat, A., Chitnis, A., Lewis, J., Ish-Horowicz, D., 1995. Expression of a Delta homologue in prospective neurons in the Acampora, D., Merlo, G.R., Paleari, L., Zerega, B., Postiglione, M.P., chick. Nature 375, 787–790. Mantero, S., Bober, E., Barbieri, O., Simeone, A., Levi, G., 1999. Izpisua-Belmonte, J.C., De Robertis, E.M., Storey, K.G., Stern, C.D., 1993. Craniofacial, vestibular and bone defects in mice lacking the Distal- The homeobox gene goosecoid and the origin of organizer cells in the less-related gene Dlx5. Development 126, 3795–3809. early chick blastoderm. Cell 74, 645–659. Anderson, J., Burns, H.D., Enriquez-Harris, P., Wilkie, A.O., Heath, J.K., Iseki, S., Wilkie, A.O., Health, J.K., Ishimaru, T., Eto, K., Morriss-Kay, 1998. Apert syndrome mutations in fibroblast growth factor receptor 2 G.M., 1997. Fgfr2 and osteopontin domains in the developing skull exhibit increased affinity for FGF ligand. Hum. Mol. Genet. 7, 1475– vault are mutually exclusive and can be altered by locally applied 1483. FGF2. Development 124, 3375–3384. Aono, A., Hazama, M., Notoya, K., Taketomi, S., Yamasaki, H., Tsukuda, Iseki, S., Wilkie, A.O., Morriss-Kay, G.M., 1999. Fgfr1 and Fgfr2 have R., Sasaki, S., Fujisawa, Y., 1995. Potent ectopic bone-inducing activ- distinct differentiation- and proliferation-related roles in the developing ity of bone morphogenetic protein-4/7 heterodimer. Biochem. Biophys. mouse skull vault. Development 126, 5611–5620. Res. Commun. 210, 670–677. Jabs, E.W., Li, X., Scott, A.F., Meyers, G., Chen, W., Eccles, M., Mao, Bei, M., Maas, R., 1998. FGFs and BMP4 induce both Msx1-independent J.L., Charnas, L.R., Jackson, C.E., Jaye, M., 1994. Jackson-Weiss and and Msx1-dependent signaling pathways in early tooth development. Crouzon syndromes are allelic with mutations in fibroblast growth Development 125, 4325–4333. factor receptor 2. Nat. Genet. 8, 275–279. Bellus, G.A., Gaudenz, K., Zackai, E.H., Clarke, L.A., Szabo, J., Franco- Jabs, E.W., Muller, U., Li, X., Ma, L., Luo, W., Haworth, I.S., Klisak, I., mano, C.A., Muenke, M., 1996. Identical mutations in three different Sparkes, R., Warman, M.L., Mulliken, J.B., et al., 1993. A mutation in N. Holleville et al. / Developmental Biology 257 (2003) 177–189 189

the homeodomain of the human MSX2 gene in a family affected with Nishimatsu, S., Thomsen, G.H., 1998. Ventral mesoderm induction and autosomal dominant craniosynostosis. Cell 75, 443–450. patterning by bone morphogenetic protein heterodimers in Xenopus Julian, L.M., Tyler, W.S., Hage, T.J., Gregory, P.W., 1957. Premature embryos. Mech. Dev. 74, 75–88. closure of the spheno-occipital synchrondrosis in the horned Hereford Opperman, L.A., 2000. Cranial sutures as intramembranous bone growth dwarf of the “short-headed” variety. Am. J. Anat. 100, 269–287. sites. Dev. Dyn. 219, 472–485. Kim, H.J., Rice, D.P., Kettunen, P.J., Thesleff, I., 1998. FGF-, BMP- and Opperman, L.A., Adab, K., Gakunga, P.T., 2000. Transforming growth Shh-mediated signalling pathways in the regulation of cranial suture factor-beta 2 and TGF-beta 3 regulate fetal rat cranial suture morpho- morphogenesis and calvarial bone development. Development 125, genesis by regulating rates of cell proliferation and apoptosis. Dev. 1241–1251. Dyn. 219, 237–247. Lamb, T.M., Knecht, A.K., Smith, W.C., Stachel, S.E., Economides, A.N., Opperman, L.A., Chhabra, A., Nolen, A.A., Bao, Y., Ogle, R.C., 1998. Stahl, N., Yancopolous, G.D., Harland, R.M., 1993. Neural induction Dura mater maintains rat cranial sutures in vitro by regulating suture by the secreted polypeptide noggin. Science 262, 713–718. cell proliferation and collagen production. J. Craniofac. Genet. Dev. Le Douarin, M.L., Kalcheim, C., 1999. The Neural Crest. Cambridge Biol. 18, 150–158. University press, Cambridge. Oudhof, H.A., 1982. Sutural growth. Acta Anat. 112, 58–68. Mahmood, R., Bresnick, J., Hornbruch, A., Mahony, C., Morton, N., Pera, E., Stein, S., Kessel, M., 1999. Ectodermal patterning in the avian Colquhoun, K., Martin, P., Lumsden, A., Dickson, C., Mason, I., 1995. A role for FGF-8 in the initiation and maintenance of vertebrate limb embryo: epidermis versus neural plate. Development 126, 63–73. bud outgrowth. Curr. Biol. 5, 797–806. Rice, D.P., Aberg, T., Chan, Y., Tang, Z., Kettunen, P.J., Pakarinen, L., Marcelle, C., Eichmann, A., Halevy, O., Breant, C., Le Douarin, N.M., Maxson, R.E., Thesleff, I., 2000. Integration of FGF and TWIST in 1994. Distinct developmental expression of a new avian fibroblast calvarial bone and suture development. Development 127, 1845–1855. growth factor receptor. Development 120, 683–694. Robledo, R.F., Rajan, L., Li, X., Lufkin, T., 2002. The Dlx5 and Dlx6 Miyama, K., Yamada, G., Yamamoto, T.S., Takagi, C., Miyado, K., Sakai, homeobox genes are essential for craniofacial, axial, and appendicular M., Ueno, N., Shibuya, H., 1999. A BMP-inducible gene, dlx5, regu- skeletal development. Genes Dev. 16, 1089–1101. lates osteoblast differentiation and mesoderm induction. Dev. Biol. Ryoo, H.M., Hoffmann, H.M., Beumer, T., Frenkel, B., Towler, D.A., 208, 123–133. Stein, G.S., Stein, J.L., van Wijnen, A.J., Lian, J.B., 1997. Stage- Monsoro-Burq, A., Le Douarin, N.M., 2001. BMP4 plays a key role in specific expression of Dlx-5 during osteoblast differentiation: involve- left–right patterning in chick embryos by maintaining Sonic Hedgehog ment in regulation of osteocalcin gene expression. Mol. Endocrinol. 11, asymmetry. Mol. Cell 7, 789–799. 1681–1694. Monsoro-Burq, A.H., Bontoux, M., Vincent, C., Le Douarin, N.M., 1995. Schell, U., Hehr, A., Feldman, G.J., Robin, N.H., Zackai, E.H., de Die- The developmental relationships of the neural tube and the notochord: Smulders, C., Viskochil, D.H., Stewart, J.M., Wolff, G., Ohashi, H., et short and long term effects of the notochord on the dorsal spinal cord. al., 1995. Mutations in FGFR1 and FGFR2 cause familial and sporadic Mech. Dev. 53, 157–170. Pfeiffer syndrome. Hum. Mol. Genet. 4, 323–328. Monsoro-Burq, A.H., Duprez, D., Watanabe, Y., Bontoux, M., Vincent, C., Schmid, B., Furthauer, M., Connors, S.A., Trout, J., Thisse, B., Thisse, C., Brickell, P., Le Douarin, N., 1996. The role of bone morphogenetic Mullins, M.C., 2000. Equivalent genetic roles for bmp7/snailhouse and proteins in vertebral development. Development 122, 3607–3616. bmp2b/swirl in dorsoventral pattern formation. Development 127, 957– Moore, R., Ferretti, P., Copp, A., Thorogood, P., 2002. Blocking endoge- 967. nous FGF-2 activity prevents cranial osteogenesis. Dev. Biol. 243, Suzuki, A., Kaneko, E., Maeda, J., Ueno, N., 1997. Mesoderm induction by 99–114. BMP-4 and -7 heterodimers. Biochem. Biophys. Res. Commun. 232, Muenke, M., Gripp, K.W., McDonald-McGinn, D.M., Gaudenz, K., Whi- 153–156. taker, L.A., Bartlett, S.P., Markowitz, R.I., Robin, N.H., Nwokoro, N., Wang, Y.H., Rutherford, B., Upholt, W.B., Mina, M., 1999. Effects of Mulvihill, J.J., Losken, H.W., Mulliken, J.B., Guttmacher, A.E., Wil- BMP-7 on mouse tooth mesenchyme and chick mandibular mesen- roy, R.S., Clarke, L.A., Hollway, G., Ades, L.C., Haan, E.A., Mulley, J.C., Cohen Jr., M.M., Bellus, G.A., Francomano, C.A., Moloney, chyme. Dev. Dyn. 216, 320–335. D.M., Wall, S.A., Wilkie, A.O., et al., 1997. A unique point mutation Wilke, T.A., Gubbels, S., Schwartz, J., Richman, J.M., 1997. Expression of in the fibroblast growth factor receptor 3 gene (FGFR3) defines a new fibroblast growth factor receptors (FGFR1, FGFR2, FGFR3) in the craniosynostosis syndrome. Am. J. Hum. Genet. 60, 555–564. developing head and face. Dev. Dyn. 210, 41–52. Naski, M.C., Colvin, J.S., Coffin, J.D., Ornitz, D.M., 1998. Repression of Wozney, J.M., Rosen, V., Celeste, A.J., Mitsock, L.M., Whitters, M.J., hedgehog signaling and BMP4 expression in growth plate cartilage by Kriz, R.W., Hewick, R.M., Wang, E.A., 1988. Novel regulators of bone fibroblast growth factor receptor 3. Development 125, 4977–4988. formation: molecular clones and activities. Science 242, 1528–1534. Newberry, E.P., Latifi, T., Towler, D.A., 1998. Reciprocal regulation of Zhang, H., Hu, G., Wang, H., Sciavolino, P., Iler, N., Shen, M.M., Abate- osteocalcin transcription by the homeodomain proteins Msx2 and Dlx5. Shen, C., 1997. Heterodimerization of Msx and Dlx homeoproteins Biochemistry 37, 16360–16368. results in functional antagonism. Mol. Cell. Biol. 17, 2920–2932.