Dlx5 drives Runx2 expression and osteogenic differentiation in developing cranial suture mesenchyme

Developmental Biology 304 (2007) 860 – 874 www.elsevier.com/locate/ydbio

Genomes & Developmental Control

Dlx5 drives Runx2 expression and osteogenic differentiation in developing cranial suture mesenchyme Nicolas Holleville c,1 , Stéphanie Matéos a,b,1 , Martine Bontoux c , Karine Bollerot d,e , Anne–Hélène Monsoro-Burq a,b,c,⁎ a

c

Institut Curie, Centre de Recherche, Orsay F-91405, France b CNRS, UMR 146, Orsay F-91405, France CNRS UMR 7128, Institut d’Embryologie Cellulaire et Moléculaire, 94736 Nogent-sur-Marne, France d Université Pierre et Marie Curie, France e CNRS UMR 7622, 75252 Paris Cedex 05, France Received for publication 26 July 2006; revised 7 December 2006; accepted 4 January 2007 Available online 9 January 2007

Abstract Craniofacial bones derive from cephalic neural crest, by endochondral or intramembranous ossification. Here, we address the role of the homeobox transcription factor Dlx5 during the initial steps of calvaria membranous differentiation and we show that Dlx5 elicits Runx2 induction and full osteoblast differentiation in embryonic suture mesenchyme grown “in vitro”. First, we compare Dlx5 expression to bone-related gene expression in the developing skull and mandibular bones. We classify genes into three groups related to consecutive steps of ossification. Secondly, we study Dlx5 activity in osteoblast precursors, by transfecting Dlx5 into skull mesenchyme dissected prior to the onset of either Dlx5 and Runx2 expression or osteogenesis. We find that Dlx5 does not modify the proliferation rate or the expression of suture markers in the immature calvaria cells. Rather, Dlx5 initiates a complete osteogenic differentiation in these early primary cells, by triggering Runx2, osteopontin, alkaline phosphatase, and other gene expression according to the sequential temporal sequence observed during skull osteogenesis “in vivo”. Thirdly, we show that BMP signaling activates Dlx5, Runx2, and alkaline phosphatase in those primary cultures and that a dominant-negative Dlx factor interferes with the ability of the BMP pathway to activate Runx2 expression. Together, these data suggest a pivotal role of Dlx5 and related Dlx factors in the onset of differentiation of chick calvaria osteoblasts. © 2007 Elsevier Inc. All rights reserved. Keywords: Dlx5; Runx2; Chick embryo; Bone differentiation; Primary calvaria cells; Transfection; FGF receptor; BMP; alkaline phosphatase

Introduction During embryogenesis and early post-natal life, brain and skull growth are coordinated by the maintenance of the cranial sutures, e.g. undifferentiated areas located between the ossified skull plates. At the edge of the developing bones, the osteogenic fronts (OFs), the osteoblast precursors proliferate actively then differentiate (Holleville et al., 2003; Iseki et al., 1997, 1999;

⁎ Corresponding author. Institut Curie, Centre de Recherche, Orsay F-91405, France. Fax: +331 69 86 3017. E-mail address: [email protected] (A.–H. Monsoro-Burq). 1 Both authors have contributed equally to this work. 0012-1606/$ - see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2007.01.003

Opperman, 2000; Rice et al., 2003). Mutations in FGF and BMP pathways or transcription factors such as Twist and Msx genes result in skull malformations, due to either delayed or premature ossification of the suture mesenchyme (Lenton et al., 2005). These signals pattern the localized activation of essential regulators of osteogenesis such as Dlx5, Runx2, and Osterix (Osx) (Ryoo et al., 2006). Dlx5 mutant mice present delayed skull ossification as well as severe craniofacial defects (Acampora et al., 1999). Skull ossification does not occur in Dlx5/Dlx6 double mutant mice, reflecting redundant activities of Dlx5 and Dlx6 in this process (Beverdam et al., 2002; Depew et al., 2002; Merlo et al., 2002; Robledo et al., 2002). Runx2 controls osteoblast maturation in all skeletal elements, both in chondroblasts and osteoblasts, whereas Osx controls terminal osteoblast differentiation (Nakashima et al., 2002).

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

In vivo and in vitro approaches have explored the regulation and the role of Dlx5 and Runx2 genes during bone development. In the skull, Dlx5 is expressed in neural crest-derived osteoblasts progenitors, before the onset of skull osteogenesis (Holleville et al., 2003). Dlx5 expression is regulated by BMP2 both in vivo (Holleville et al., 2003) and in cell lines (Lee et al., 2003). In the C2C12 myoblast cell line, which undergoes transdifferentiation towards osteoblasts when triggered by BMP2, Dlx5 is an immediate early target of BMP2 signaling, thus appearing before osteogenic differentiation (Lee et al., 2003). Furthermore, Dlx5 activity is required for BMP2-mediated C2C12 cells transdifferentiation and Runx2 expression (Ryoo et al., 2006). In addition to this early role, e.g. before the initiation of differentiation, the basal levels of Dlx5 expression observed in fetal osteoblast cell lines are further upregulated during the late mineralization stage (Ryoo et al., 1997). Runx2 is regulated negatively by Msx and Twist, and positively by TGFβ1, BMP2, and Dlx5 (Ryoo et al., 2006; Blin-Wakkach et al., 2001; Dodig et al., 1999; Liu et al., 1999; Bialek et al., 2004). Osteogenic mesenchyme development is controlled by surrounding structures such as the ectoderm or the dura mater (Francis-West et al., 1998; Kim et al., 1998; Takahashi et al., 1991). In this study, we first describe similar and unique gene expression in membranous bones located at two distinct craniofacial locations: (i) in the calvaria, where bones form between dura mater and dermis/ectoderm, and (ii) in the mandibula, where membranous bones differentiate in the vicinity of Meckel’s cartilage, under the control of ectoderm and mesenchymal signals. We analyze the patterns of pre-osteoblast and early osteoblast markers (Dlx5, FGFRs, BMPs, Msx1/2, Goosecoid, Twist, that we had previously described in the chick frontal OFs) as well as other genes involved in mandibula patterning and bone formation (Runx2, Prx1, Pitx1, Dlx2) (Ducy et al., 1999; Komori et al., 1997; Lanctot et al., 1999; Martin et al., 1995). Prx1 is a homeobox-containing gene, expressed in the mesenchyme of facial skeletal precursors, involved in the early epithelial–mesenchymal interactions controlling skeletogenesis (Kuratani et al., 1994; Martin et al., 1995). Dlx2 and Pitx1 are involved in early mandibula patterning (Lanctot et al., 1999; Qiu et al., 1995, 1997). Secondly, we address the role of Dlx5 in the initiation of the osteogenic program in primary cells derived from immature suture mesenchyme. It is clear from numerous studies that overexpression of Dlx5 in fetal osteoblasts, or osteoblast cell lines, stimulates bone differentiation (Miyama et al., 1999; Newberry et al., 1998; Tadic et al., 2002). However, these cells are already engaged in osteogenic differentiation and express basal levels of Dlx5 and Runx2 (Miyama et al., 1999). In the C2C12 transdifferentiation assay, forced expression of Dlx5 without additional BMP signals, stimulates Runx2 expression but is not sufficient for Osx or osteocalcin induction (Ryoo et al., 2006). Moreover, in another study carried out using older fetal (E21) rat calvaria and osteosarcoma cells ROS 17/2.8, Dlx5 repressed osteocalcin promoter but did not modify osteopontin and alkaline phosphatase expression (Ryoo et al., 1997). This shows that Dlx5 may display different roles according to the differentiation status of the cell. Altogether, these studies show a role for Dlx5 in the progression of the osteogenic

861

program, but do not address the role of Dlx5 on immature progenitors, and whether Dlx5 can drive a complete osteogenic program from such progenitor cells. Using uncommitted calvaria precursors, we show here that (i) Dlx5 switches on a sequential series of gene expressions, similar to the ones described in vivo, (ii) Dlx5 activates a complete osteogenic program, including Runx2, osteopontin, and alkaline phosphatase expression, and (iii) a dominant-negative Dlx homeodomain construct interferes with BMP-induced Runx2 activation in these primary cells. Material and methods In situ hybridization (ISH) on paraffin sections Fertile Gallus gallus eggs, JA 57 or Haas strains, were incubated for 7 to 12 days (E7 to E12) at 38 °C. Heads were dissected in ice-cold PBS, fixed, embedded in wax, and sectioned (7 mm thick sections, Leica RM2145 microtome). Alternate serial sections were probed with different genes and are presented in Supplemental Fig. 1 for close details of the suture. Antisense RNA probes were synthesized using Riboprobe Combination System SP6/T7 kit (Promega) as described previously for Msx1 and Msx2 (Monsoro-Burq et al., 1995); Bmp2 and Bmp4 (Francis-West et al., 1994); FGF-R1, -R2 (Wilke et al., 1997); Dlx5 (Pera et al., 1999); Bmp7 (Wang et al., 1999); Goosecoid (Goos; Izpisua-Belmonte et al., 1993); Prx1 (Kuratani et al., 1994); Pitx1 (Logan and Tabin, 1999); and Dlx2 (Puelles et al., 1999). The osteopontin, tissue nonspecific alkaline phosphatase (Alk. Phos.) and Twist probes were kind gifts of Drs. P. Castagnola, J-L Millán, and M-C. Delfini. Chick Runx2 gene was cloned from E2–2.5 chick embryo RNA (RNAble kit, Eurobio) by RT-PCR amplification using specific primers: 5′-ATG CGT. ATC CCC GTA GAT RCC AGC-3′ and 5′-TCA GTA KGG CCK CCA NAC VGC-3′ (Enhanced Avian RTPCR Kit, Sigma). Sections were stained either with NBT-BCIP (Roche) or Vector Red (Vector), after ISH procedure according to Etchevers et al. (2001).

Cell culture and transfection E6/6.5 head ectoderm and mesenchyme from the frontal area, or mandibular mesenchyme were dissected out in PBS. The ectoderm and meningis were removed by a mild pancreatin treatment (Gibco BRL). All visible Meckel’s cartilage rudiments were carefully dissected out. The mesenchyme was further dissociated by a mild trypsin digestion, or dissociated through a syringe needle, centrifuged, and collected in a complete culture medium (DMEM supplemented with 10% fetal calf serum, 1% penicillin–streptomycin and 1% glutamine (Gibco BRL)). For low density short-term culture (Fig. 5), the cells were plated in 96-well plates at a density of 1000 cells per square-cm. To improve survival of the cells, other cultures were seeded at a density of 42,000 cells per square-cm (Figs. 6–8). The culture would not survive in the absence of serum, which prevented us from addressing the role of Dlx5 in serum-free medium. After 7– 9 days of culture in DMEM, 10% fetal calf serum, 1% penicillin–streptomycin and 1% glutamine, the medium was supplemented with 50 μg/ml of acid ascorbic and 10 mM of β-GPO4, in order to allow bone differentiation (thus named “differentiation medium”). This technique was modified from previous protocols developed for older primary chick calvaria cells or transdifferentiation assays (Gerstenfeld et al., 1987; Poliard et al., 1995). We have constructed an expression vector encoding the complete ORF of chick Dlx5 (chDlx5, a generous gift of Dr. M. Kessel, GenBank™ accession number U25274). ChDlx5 was subcloned into BamH1/Xho1 of the pCDNA3 vector (Invitrogen). The activation of BMP signaling was achieved using a constitutively activated BMP Type I receptor (AcBMPR, Q233D) cloned in pCS107, a kind gift of Dr. R. Harland. The dominant-negative Dlx construct (DlxHD-EnR) consists of the fusion of zebrafish Dlx3 (homologous to Xenopus and chick Dlx5) homeodomain to the strong Engrailed Repressor domain (EnR). This construct was a generous gift of Dr. K. Artinger (Woda et al., 2003). The cells were transfected 18–24 h after plating, at about 50–60% confluence, either with control pCDNA3, control pCS107, pCS2-GFP, pCDNA3-Dlx5, pCS107-AcBMPR, pCS2-DlxHD-EnR, or a combination of these. We have used Transfectam® (Transfectam® Reagent, Promega) or

862

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

Lipofectamine Plus (Invitrogen), according to the manufacturers protocol. After 3– 6 h of incubation, fresh medium containing 10% serum was added. Medium was changed every 2 days. The short-term cultures were analyzed at 24, 48, or 92 h post-transfection (equivalent to E8, E9, and E12, assuming that the timing of development in vivo and in vitro can be compared). The transfection efficiency by Dlx5 was determined by staining for Dlx5 mRNA (ISH) or protein immunostaining (see below). In most low density experiments, more than 90% of the cells were positive. The higher density cultures were analyzed at day 3 to 17. We have also tried to cultivate early mandibula mesenchyme, dissected out at E4. These cells grow well in vitro when not transfected. Unfortunately, using four alternative transfection methods (Transfectam, Lipofectamin Plus, Effectene (Qiagen), and Exgene (Euromedex)), we did not get a good survival of the transfected cells and could not use this type of cultures for further analysis (not shown).

In situ hybridization and immunocytochemistry on primary chick calvaria cells The cultures were rinsed in PBS, fixed in 4% formaldehyde/PBS, directly prehybridized, and stained according to standard procedures. Some cultures were stained with a polyclonal pan-Dlx antibody, a kind gift of Dr. J. Kohtz, raised as described in (Kohtz et al., 2001; Panganiban et al., 1995). To monitor cell proliferation, BrdU incorporation was performed during 4 h before fixation and analyzed using the Cell proliferation kit (Amersham Pharmacia Biotech). Following short-term culture, about a hundred cells were counted in the center of each well. Within each series of experiments, three replicates were done for each assay (in situ hybridization for each gene or BrdU analysis). A total of five independent series of experiments were done. The percent of positive cells in about 1500 cells is reported in Figs. 5M, N.

Bone sialoprotein promotor reporter assay The transactivating or repressor potential of chDlx5 and DlxHD-EnR constructs were tested using the trimerized Dlx5-binding C-element of the bone sialoprotein promotor, a gift of Dr. Franceschi (Benson et al., 2000). BHK21 cells were transfected using Exgen (Euromedex). Luciferase assays were performed 36–48 h after transfection using Promega Dual Luciferase Reporter Assay System.

the bone, in a domain where osteoprogenitors have not yet differentiated (Fig. 1C). Dlx5 is excluded from the differentiated bone (expressing osteopontin (not shown) and alkaline phosphatase (Fig. 1F)), and from the sutural mesenchyme in the midline (Fig. 1C′). The highest level of expression is found at the OFs (Fig. 1C′). In addition, some Dlx5expressing cells are found located farther from the OFs towards the midline (see Supplemental Fig. 1 for high magnification pictures). Similarly, Runx2 delineates the outer limit of the differentiated bone, the highest levels of expression being found at the OFs (Figs. 1D–D′). Runx2 expression corresponds to the area that expresses the highest levels of Dlx5. Neither the undifferentiated sutural mesenchyme at the midline nor the mineralized structures express Runx2 (Fig. 1D′). We detect Prx1 transcripts in the undifferentiated sutural mesenchyme (Fig. 1E′). Prx1 is thus expressed outside of the Dlx5- and Runx2-positive domains (Fig 1E). These observations suggest that Prx1 labels immature osteoblasts precursors at an earlier stage of their development than Dlx5 and Runx2. Expression of the definitive bone marker alkaline phosphatase is found in cells lining the mineralized bone and within the bone plate (Figs. 1F, F′). We did not detect Pitx1 or Dlx2 expression in the chick frontal area from E9 to E16 (not shown). Comparative gene expression during calvaria and mandibula membranous ossification We then examined several gene expression patterns during membranous bone formation, in E7–E12 chick mandibula. The neural crest-derived mandibular mesenchyme differentiates into Meckel’s cartilage and several membranous bones around it: the articular, angular, supra-angular, dentary splenial, and mentomandibula bones (Couly et al., 1993; Le Lievre, 1978) (Fig. 3B).

Results Nested patterns of Dlx5, Runx2, and Prx1 during chick frontal bone ossification We have previously described the coordinate expression of BMP2/4/7, Msx1/2, Dlx5, FGFR1/2/4, and Twist genes in the frontal suture and bones of chick embryo calvaria. We have also shown that Dlx5 is activated in proliferating osteoblast precursors at 7/8 days of development (E7–8), i.e. before any osteoblast differentiate in this area, and is maintained during the early stages of differentiation (Holleville et al., 2003). At E12, the frontal bones are still separated by an open suture. Thus, on the same transverse section, both undifferentiated mesenchyme and differentiated osteoblasts can be observed (Figs. 1A–B). The suture's mesenchyme and the outer part of the OFs comprise immature progenitors, whereas the inner part of the OFs and the bone are formed by more mature and differentiated osteoblasts respectively. Here, we compare the E12 expression patterns of Dlx5 (Figs. 1C, C′), Runx2 (Figs. 1D, D′), Prx1 (Figs. 1E, E′), alkaline phosphatase (Figs. 1F, F′), Pitx1, and Dlx2 (not shown) on adjacent sections. Dlx5 is expressed in the mesenchyme located around

Dlx5, Bmp7, and Bmp4 expression partially overlap in E7 mandibular arch In early skull mesenchymal condensations, Dlx5 activation is correlated with areas of Bmp4 and Bmp7 coexpression (Holleville et al., 2003). In the mandibula at E7, Meckel’s cartilage is surrounded by undifferentiated mesenchymal condensations. Membranous bone elements start to differentiate around E8, as detected by alizarin red staining (not shown). The undifferentiated mesenchyme located around Meckel’s cartilage strongly expresses Dlx5: at E7, there is an enhanced Dlx5 expression ventral to Meckel’s cartilage (Fig. 2A, bracket 1) and at the junction between the maxillary process and mandibular arch (Fig. 2A, bracket 2) (Fig. 2D, pink shade). Bmp7 is expressed in the mandibular mesenchyme on both sides of Meckel’s cartilage, coincindent with the strongest levels of Dlx5 (Fig. 2B, brackets 1 and 2). Bmp7 expression is also observed in Dlx5-negative domains, such as the mesenchyme and epithelium of the tongue primordium and the maxillary mesenchyme (Figs. 2B, D, blue dots). A gradient of Bmp4 expression from mandibular epithelium to the underlying mesenchyme has been reported between stages 24 and 28 (E6) (Barlow and Francis-West, 1997; Francis-West et al.,

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

863

Fig. 1. Expression of Dlx5, Runx2, Prx1, and alkaline phosphatase in frontal bone osteogenic fronts at E12. (A) A cleared skeletal preparation of an E12 chick embryo head is shown in dorsal view, anterior to the left. The white bar indicates the level of the transverse sections shown in panels B–F. The cartilage is stained with alcian blue, and the bone with alizarin red. (B) Schematic representation of the sections shown in panels C–F, with dorsal to the top. The nasal septum cartilage (ns) is shown in blue, bilateral frontal bone elements are in red. The red square indicates the frames of panels C–F sections. (C, C′) In situ hybridization (ISH) for Dlx5 shows a strong staining around the ossified element, enlarged area in panel C′ is indicated by the box in panel C. (D, D′) ISH for Runx2 on an adjacent section shows that Runx2 domain overlaps with Dlx5 domain. (E, E′) Prx1 is expressed in the suture mesenchyme, outside Dlx5 domain. (F, F′) Alkaline phosphatase is expressed in the cells lining the ossified element, inside Dlx5 domain. The black arrows in panels C′, D′, E′, and F′ indicate the most medial extent of the mineralized bone. Additional adjacent sections shown at high magnification are shown in Supplemental Fig. 1.

1994). At E7, Bmp4 is expressed in the mesenchyme underlying the oral epithelium in the mandibula, the tongue and the maxillary process (Figs. 2C, D, green). Thus, Dlx5 activation around Meckel’s cartilage surrounds the areas of high Bmp7 expression but Dlx5 expression is not as closely correlated to Bmp4 expression as observed in frontal bone progenitors. Nested gene expression patterns in E12 mandibula We analyzed markers for mesenchyme, pre-osteoblasts, and osteoblasts, as well as members of BMP and FGF signaling

pathways (a total of 15 genes) on alternate sections at E12 (Figs. 3 and 4). In each panel, the lateral part of mandibula is oriented towards the left and the medial part is on the right. The genes are grouped according to their expression in the immature mesenchyme (Fig. 3, group 1) and in progressively more mature cells, closer to the differentiated bone: pre-osteoblasts, early osteoblasts, and embedded osteoblasts (Fig. 4, groups 2 and 3). At E12, Dlx5 expression outlines each membranous bone (Figs. 3A and B) and is not expressed in Meckel’s cartilage. Dlx5 is excluded from the differentiated osteoblasts but is

864

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

Fig. 2. Expression of Bmp2, Bmp4, and Bmp7 compared to Dlx5 in the mandibula of E7 chick embryos. (A) In transverse sections through the head of E7 embryos, Dlx5 is expressed in the mesenchyme of the mandibular arch (bracket 1) and at the junction between the mandibula and the maxillary processes (bracket 2). Meckel's cartilage (Mc) does not express Dlx5. (B) Bmp7 is expressed within Dlx5 domain (brackets 1 and 2), in the tongue and maxillary processes and in the facial ectoderm. (C) Bmp4 is found farther from the Dlx5-positive areas, under the oral and tongue ectoderm and in the maxillary process. (D) schematic illustrates the overlap of the expression of Dlx5 (pink), Bmp7 (blue), and Bmp4 (green) in the mandibula of E7 chick embryos.

expressed close to the mineralized structure, in the preosteoblasts (Fig. 3C). The expression patterns of the other genes are compared to the Dlx5-positive area, and described either as located “farther” or “closer” to the differentiated cells (the bone is delimited by a dotted line). None of these genes is detected in Meckel’s hypertrophic cartilage (Figs. 3 and 4). Prx1 and Pitx1 expression is observed the farthest from the bone (Figs. 3D, E). Prx1 is strongly expressed in undifferentiated mesenchyme (Fig. 3D′). As observed around the frontal bones, a Prx1-negative area surrounds the bones (Fig. 3D′, compare to Dlx5 in Fig. 3C). Pitx1 is weakly expressed as a fine line around the pre-osteoblast area except at the tip of the bone element (Figs. 3E and E′). Twist, Msx1, and FgfR2 label both the immature mesenchyme located distal to the bone elements and the cells within the Dlx5-positive area (Figs. 3F–H′). In contrast, Msx2 appears as a weak and scattered pattern around the OFs of the membranous bones, hardly above background levels (Figs. 3I and I′). Goosecoid, Dlx2, and Runx2 are detected in a similar pattern to that of Dlx5, although Dlx2 is weakly expressed at this stage

(Figs. 4A–C′). As noted earlier in the skull, Runx2 and Dlx5 domains again overlap (compare Fig. 3C and Fig. 4C′). Bmp2 and Bmp4 are expressed in the pre-osteoblasts lining bones (Figs. 4D–E). Moreover, Bmp2 is found in some early osteoblasts not yet embedded in the bone matrix (Fig. 4D′). In contrast, Bmp7 is located mainly around the tips of the bones (Figs. 4F and F′). As in the calvaria, FgfR1 and osteopontin (osp) genes are expressed in early and differentiated osteoblasts, surrounded by bone matrix (Figs. 4G–G′, H–H′). From this comparative analysis, we conclude that the general gene expression pattern during mandibula membranous ossification is similar to what is reported in the calvaria. Prx1 labels the immature mesenchyme. Runx2 and Dlx5 domains mainly overlap. However, a few differences between the two locations are observed. At E12, Twist and Goosecoid are hardly detected in the mandibula whereas their expression is prominent in the frontal area. Secondly, Pitx1 and Dlx2 are found around mandibula bone primordia but not in the frontal area at any of the stages analyzed.

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

865

Fig. 3. Expression of immature osteoblast progenitor markers in E12 chick mandibula. The expression of a series of genes present around the developing bones was compared to expression of Dlx5, in the mandibular membrane bones that form around Meckel's cartilage. (A) Dlx5 is expressed in the immature osteoblasts lining the membrane bones (schematized in red in panel B). (C) Enlargement at the position indicated by the star in panel A shows the respective position of Dlx5-positive cells and mineralized structure (dotted line). Adjacent sections show expression of Prx1 (D, D′), Pitx1 (E, E′), FgfR2 (F, F′), Twist (G, G′), Msx1 (H, H′), and Msx2 (I, I′) in general view (D-I) and enlargements at the level indicated by the star (D′-I′). See text for the comparative analysis of these expression patterns. All these markers are expressed outside Dlx5-positive domain compared to the mineralized bone (indicated by the dotted line).

866

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

Fig. 4. Expression of committed and mature osteoblast markers in E12 chick embryo mandibula. The expression of Dlx5 was further compared to that of its putative target Goosecoid (A, A′), Dlx2 (B, B′), the committed osteoblast marker Runx2 (C, C′), Bmp2 (D, D′), Bmp4 (E, E′), and Bmp7 (F, F′). These genes have overlapping expression domains. In contrast FgfR1 (G, G′) and osteopontin (H, H′) are expressed by the embedded osteoblasts. See text for details.

Overexpression of Dlx5 in early primary calvaria drives the osteogenic program In order to address the role of Dlx5 during the onset of osteogenesis in the calvarial progenitors, we have overexpressed Dlx5 in suture mesenchyme. Due to the low proliferation rate in the immature suture areas, RCAS retrovirus-mediated infection “in vivo” and in explants failed. Therefore, we have transfected Dlx5 expression plasmid into primary cultures of calvaria mesenchyme. Previous studies have used differentiated Dlx5-positive fetal osteoblasts. Here we have dissected E6–6.5 immature mesenchyme located in the

frontal area and prospective coronal suture domain (Fig. 5A), in order to collect calvaria progenitors before they activate endogenous Dlx5 gene expression (Holleville et al., 2003). We address three questions: (1) does Dlx5 act on cell proliferation, (2) does Dlx5 overexpression activate a subset or all of the genes detected during the different steps of early osteogenesis in Figs. 3 and 4, (3) does Dlx5 trigger terminal osteogenic differentiation in these cells grown in vitro? Cells transfected with the control plasmid do not express Dlx5 after 1– 9 days in culture (as detected by ISH (B, B′) or immunostaining (not shown)), showing that the cells have not received sufficient, if any, Dlx5-inducing signals before they are isolated, and that

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

867

Fig. 5. Dlx5 induces the expression of early osteoblast markers in short-term culture of immature calvaria mesenchyme. Dlx5 was transfected into primary cultures of immature, non committed calvaria mesenchyme, taken at E6 in the frontal area of chick embryos (A, indicated by red dash). The control cells do not express either Dlx5 or Runx2 at the time of explantation or later in vitro. B: in situ hybridization for Dlx5, B′: phase contrast of the same culture. The percent of positive cells is quantified in panels M and N. Dlx5 was efficiently transfected into these primary cells (C: in situ hybridization, D, Dll antibody staining, M). Proliferation was analyzed (E, M). The expression of the immature mesenchyme markers FgfR2 (F), Twist (G, control; G′ transfected by Dlx5), of the committed osteoblasts Runx2 (H), Msx1 (I), Goosecoid (J) and differentiated osteoblasts FgfR1 (K) and osteopontin (L) were analyzed and quantified (M, N; blue bar indicates percent of positive cells in control pCDNA-transfected culture, red bar indicate percent of positive cells in Dlx5-transfected culture).

the culture medium alone is insufficient to elicit Dlx5 expression (Figs. 5B, B′). In the short-term culture condition, an average of 94% of the cells transfected with pCDNA3-Dlx5 express Dlx5 after 48 h in culture (detected by in situ hybridization or immunocytochemistry using an anti-Dll polyclonal antibody) (Figs. 5C, D, M). We have detected Dlx5 expression up to 4 days after transfection.

Dlx5 does not alter pre-osteoblast proliferation or mesenchymal marker expression (Prx1, Twist), but it does enhance FgfR2 expression The expression of Dlx5 by proliferating pre-osteoblasts from E7–8 onwards suggests that Dlx5 could participate in either control of their proliferation or in the onset of their differentiation program (Holleville et al., 2003). Here, we have analyzed

868

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

BrdU incorporation in Dlx5-transfected and control cultures. As observed in vivo, the proliferation rate of these primary cells is low. After Dlx5 transfection, we do not observe a significant increase in cell proliferation (Figs. 5E, M). In the frontal area, the suture mesenchyme expresses both Twist, Prx1, and FgfR2 whereas cells around the OFs express Dlx5, Twist, and FgfR2. In control cultures, 28% of the cells expressed FgfR2, 41% Prx1, and 62% Twist. Dlx5 expression results in the significant upregulation of FgfR2 (88% of positive cells) but not of Twist (77% of positive cells) or Prx1 (36% of positive cells), at 24–36 h after transfection in the primary calvaria mesenchyme (Figs. 5F–G). In conclusion, Dlx5 does not significantly modify suture mesenchyme markers Twist and Prx1 expression, nor cell proliferation, but is able to markedly increase FgfR2 expression. Dlx5 induces committed and mature osteoblast markers Subsequently, we have analyzed the expression of genes present in pre-osteoblasts both in skull and mandibula progenitors (Runx2, Msx1, Goosecoid), and in mandibula only (Dlx2, Pitx1). Additionally we have examined genes expressed by the mature osteoblasts in both locations (FgfR1 and osteopontin). We have also looked at Msx2 expression, although it is present only at background levels in both skull and mandibula at E12 (Fig. 3). Results are shown in Figs. 5H–N. Runx2, Msx1, Pitx1, Dlx2, and osteopontin are virtually not expressed in control calvaria mesenchyme (less than 10% of the cells, Fig. 5N). As observed on sections, Msx2 expression is comparable to background, no differences are observed between control cells and Dlx5-transfected cultures (not shown). Goosecoid and FgfR1 are present at basal level of expression in 25% and 18% of the cells, respectively. As early as 24 h after Dlx5 transfection, the calvaria precursors exhibit a marked induction of Runx2 and Msx1 expression (64% and 78% of the cells, respectively, Figs. 5H–I) as well as a strong increase in Goosecoid expression (75%, Fig. 5J). Similarly, Dlx5 induces a strong expression of Dlx2 and Pitx1 after 24 or 36 h (respectively 77% and 73% of positive cells, Fig. 5N). FgfR1 expression is upregulated by Dlx5 overexpression 48-h posttransfection (Figs. 5K, N). A few Osteopontin-expressing cells are observed 92 h after transfection of pCDNA3-Dlx5 (Figs. 5L, N). This experiment shows that transfection of Dlx5 in the progenitors of skull bones elicits induction/overexpression of the genes described in vivo, in a temporal manner reflecting the maturation steps of the osteoblasts. Surprisingly, Dlx2 and Pitx2 are also activated in skull progenitors in vitro, although their in vivo expression is restricted to the mandibula. This suggests that local regulation, potentially provided by surrounding tissue, may be missing in our culture conditions. Long-term cultures reveal that Dlx5 drives terminal bone differentiation from immature mesenchyme To fully assay bone differentiation, we have developed longterm culture conditions for the immature E6 calvaria progenitors, which are based on established conditions used for in vitro differentiation of E16 chick calvaria or C1 mesoderm cell line

(Gerstenfeld et al., 1987; Poliard et al., 1995). In contrast to late stage osteoblasts, the E6 immature cells are difficult to grow for long periods (2–3 weeks). When several culture wells are inoculated in parallel from the same cell suspension, only a small subset of the wells grow efficiently as confluent monolayers, necessitating multiple replicates in each experiment. Higher cell density somewhat improves culture survival. Using these novel long-term conditions, we have analyzed the expression of alkaline phosphatase, a widely accepted marker of definitive bone, in cultured E6 calvaria progenitors. Three independent experiments have been performed. Cultures are analyzed at day 3/4 post-transfection for Runx2 and osteopontin expression, and later switched to differentiation medium on day 9. Alkaline phosphatase expression is analyzed at day 9 and day 16–17 post-transfection. Control cultures do not show either Dlx5, Runx2, osteopontin, or alkaline phosphatase expression (in the case of alk. phos., 7/9 control cultures do not exhibit any positive cells, 2/9 cultures had a few sparse positive cells, not shown). As in lower density culture, Dlx5 is efficiently transfected, as detected 3–4 days after transfection (not shown). Runx2 is detected in most of the cells at D3–4 (Figs. 6A, B) but not at D9–11 and D14–17 (not shown). Osteopontin is found in a few cells at D3–4 (Figs. 6C, D) and shows similar weak expression at D9 and D11 (not shown). Alkaline phosphatase expression is found in sparse cells at E9 in the absence of differentiation medium (not shown). In contrast, when cells are grown in differentiation medium for 3 to 8 days, strong and widespread alkaline phosphatase staining is observed in the Dlx5-transfected cultures (10/10 cultures analyzed), most often in dense areas of the culture (Fig. 6E). We conclude that when Dlx5 is expressed in E6 calvaria progenitors grown in vitro for short and long periods, the cells recapitulate the sequential steps of bone differentiation observed in vivo. Specifically, we demonstrate that Dlx5 overexpression does not affect cell proliferation but instead activates or enhances the expression of several genes expressed at the osteogenic fronts in vivo, including a sustained Runx2 expression during the first days. Subsequently, our analysis of long-term E6 calvaria cells reveals that Dlx5 can promote full bone differentiation in culture. Dlx activity is essential for BMP-mediated Runx2 activation in primary calvaria cells Our previous work has shown that Dlx5 is activated by BMP signals in chick calvaria. Moreover, BMP signals activate both Dlx5 and Runx2 expression as well as bone differentiation in various cell contexts. Here, we have activated BMP signaling in a cell-autonomous manner, by transfecting the E6 calvaria cells with a constitutively active type I BMP receptor (AcBMPR, Fig. 7). Three to four days after transfection, both Dlx5 (Figs. 6A, B) and Runx2 (Figs. 6C, D) are activated in the AcBMPR-transfected cultures whereas the control cultures remain negative. Under long-term culture conditions using differentiation medium, alkaline phosphatase-positive cells can differentiate (Figs. 6E, F).

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

869

Fig. 6. Dlx5 induces Runx2, osteopontin, and alkaline phosphatase in long-term culture of immature calvaria mesenchyme. High density culture of immature E6 chick calvaria was assayed for induction of the definitive bone marker alkaline phosphatase. In these condition, Dlx5 induces Runx2 and osteopontin expression at 3–4 days in culture as observed in low density short-term cultures (A: Runx2 expression at D4 in Dlx5-transfected culture, B: ISH for Runx2 at D4 in control pCDNA3transfected culture; C, D: osteopontin expression at D4 in Dlx5 and control cultures). Alkaline phosphatase expression is induced at D16 in Dlx5-transfected but not in control cultures (E, F).

In order to test if Dlx5 or related proteins (such as Dlx6) are critical for the initiation of Runx2 in immature calvaria mesenchyme triggered by BMP signals, we use here a dominant-negative construct encoding zebrafish Dlx3 homeodomain fused to the repressor domain of Engrailed (DlxHDEnR, Woda et al., 2003). First, we demonstrate that this construct interferes with chick Dlx5 transactivating activity. For this, we used the Dlx5-binding element of the bone sialoprotein (BSP) promotor driving luciferase as a reporter construct (Benson et al., 2000) and performed luciferase reporter assays in BHK21 cells. We observe that the DlxHD-EnR specifically represses the endogenous activity of the promotor in BHK21 cells (Fig. 8A). Moreover, chick Dlx5 activates the promoter 3–4-fold above endogenous levels (Fig. 8B). Increasing concentrations of DlxHD-EnR co-transfected with chick Dlx5 decrease this activation (Fig. 8B). Secondly, we tested if DlxHD-EnR interferes with Runx2 activation by BMP signals. DlxHD-EnR prevents the induction of Runx2 by AcBMPR in the E6–6.5 calvaria cells (Fig. 8C). This shows that a Dlx-type activity is essential for the induction of Runx2 by BMPs in this system. Finally, in order to see if Dlx-EnR can also interfere with endogenous Runx2 expression, we transfected E6–6.5

mandibular mesenchyme cells, that already express Runx2 in control transfections. We observe a general decreased (but not abolition) of Runx2 expression in those cultures (Fig. 8C), showing that the activity of Dlx factors plays a role in the maintenance of Runx2 expression. Discussion Dlx5 and Runx2 are major regulators of bone differentiation. However, the precise role of the Dlx5 homeoprotein during the development of embryonic skull progenitors, specifically at the osteogenic front, remains to be addressed. Here, we ask three questions: is Dlx5 activity important for proliferation of progenitors pools, is Dlx5 activity important for initiating the osteogenic program (i.e. Runx2 activation), and is Dlx5 sufficient to drive a complete osteogenesis from immature calvaria progenitors (i.e. promoting Runx2 induction and definitive bone markers expression)? The analysis of these questions by a loss of function approach in mouse embryos is hampered by the redundant activities of Dlx6 and Dlx5. In the Dlx5 mutant mice, Runx2 is activated normally in the ribs but expression in the skull

870

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

Fig. 7. BMP signals activate Dlx5, Runx2, and alkaline phosphatase expression in chick immature calvaria cells. E6–E6.5 calvaria primary cells were transfected with an activated type I BMPR or a control expression vector. While control transfections remain negative (A, C, E), AcBMPR transfection results in the activation of Dlx5 (B), Runx2 (D) at day 4, and Alkaline Phosphatase (F) at day 17. Scale bar = 200 μm.

progenitors has not been reported (Acampora et al., 1999; Depew et al., 1999). In addition, skull ossification is delayed but not absent, suggesting that the main bone differentiation signals are activated to a certain extent in these mutants. In contrast, Dlx5/Dlx6 double mutants exhibit a complete absence of calvaria, among other severe skeletal and non-skeletal defects (Robledo et al., 2002). In long bones of the double mutants, Runx2 is expressed normally but the analysis of Runx2 activation in early skull progenitors, before exencephaly occurs, has not been reported. Hence, the dependence upon Dlx5/6 activity for Runx2 initiation in skull progenitors remains to be analyzed. Previous reports have analyzed the effects of Dlx5 overexpression either in established cell lines, or at late osteogenesis stages, which already display high endogenous Dlx5 and Runx2 expression levels (see for example Ryoo et al., 2006). Here, we use primary chick calvaria dissected at E6 to analyze Dlx5 activity at the earliest steps of intramembranous bone differentiation. To do so, we first analyze multiple gene expression patterns in the skull and in the mandibula of chick embryos; we describe common and specific characteristics for different stages of osteoblast development for each of the skull

or mandibula primordium. We then cultivate early primary calvaria cells to identify the genes regulated by Dlx5 activation and the phenotype of these cells. By the detailed analysis of an array of fifteen genes on serial sections, we propose that three main domains can be defined, which are similar in the developing membrane bones of skull and mandibula. We further use this classification to define the state of commitment of the cells in culture. First, the most immature areas, which are located most distal to bone, express a combination of Prx1, FgfR2, and Twist genes. This area corresponds essentially to the sutural mesenchyme in the skull (Fig. 1 and Holleville et al., 2003) and is located around the future membranous bones in the mandibula (Fig. 3). This pattern of gene expression is related to areas where the mesenchyme is maintained in an undifferentiated state, to prevent premature fusion of the bones. It was shown that Prx1 is critical for the formation of the pre-skeletal mesenchymal condensations (Martin et al., 1995) and that Prx1 is a negative regulator of osteoblast differentiation by binding to the collagen I alpha-I and the Osteocalcin promoters (Hu et al., 1998). Additionally, Twist and FgfR2 have been implicated in the maintenance of the proliferative state in mouse calvaria (Anderson et al., 1998;

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

871

Fig. 8. A Dlx dominant-negative construct interferes with Dlx5 activity and prevents Runx2 activation by BMP signals. A Dlx homeodomain fused to Engrailed Repressor domain (DlxHD-EnR) is used to block Dlx-like activity in two systems. First, reporter assays using the Dlx5-binding element of the Bone Sialoprotein showed that DlxHD-EnR inhibits endogenous levels of activity of the promotor in BHK21 cells, compared to control pCDNA3 or EnR alone (A) and that increased doses of DlxHD-EnR counteract the transactivation of the promotor by Dlx5 (B). Second, the transfection of DlxHD-EnR prevents Runx2 activation by AcBMPR in E6–6.5 calvaria cells in vitro and decreases the endogenous levels of expression of Runx2 in E6–6.5 mandibular mesenchyme in vitro (C).

Fragale et al., 1999; Iseki et al., 1999). In particular, Twist1/2 factors block Runx2 DNA binding activity and thus prevent the precocious activation of the osteogenic program (Bialek et al., 2004).

The primary calvaria mesenchyme put in culture maintains Twist, Prx1, and FgfR2 expression but does not express Dlx5 or Runx2 above background (Fig. 5). Overexpression of Dlx5 does not alter the proliferation rate of the cells or repress the

872

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

expression of these genes. The number of Prx1 or Twistpositive cells is not significantly modified while FgfR2 expression is stimulated in the first few days in culture (Fig. 5). The proliferation of the neural crest-derived frontal calvaria was recently shown to depend upon a TGFβ-FGF signaling pathway (Sasaki et al., 2006), whereas Dlx5 is activated by BMPs but not by FGF signals (Holleville et al., 2003). However, Dlx5 could participate indirectly in response to proliferative signals, by enhancing FgfR2 expression of the progenitors located at the OFs (Fig. 5M), in agreement with the expression of Dlx5 in proliferating cells (Holleville et al., 2003). The second domain defines committed osteoblast progenitors. This area corresponds to Dlx5 and Runx2 expression domains, and includes Goosecoid, Msx1, and FgfR2-positive cells (Figs. 3 and 4). As soon as 24–48 h post-transfection, Dlx5 induces the osteogenic program, defined by Runx2 induction, whereas the control cultures remain negative (Figs. 5 and 6). This indicates that, at the time, they are harvested, the neural crest-derived calvaria progenitors are not committed towards osteoblast development, and that supplying Dlx5 to those cells is sufficient to activate this program. Runx2 expression is sustained for at least 4 days post-transfection (Fig. 6). This does not exclude that a parallel pathway could cooperate to activate Runx2 in vivo (e.g. Dlx6 or other genes). In addition, the fact that the expression of Dlx5 itself is lost in the Runx2 mutant at day 15.5 dpc (Choi et al., 2005), e.g. at a time when craniofacial defects are already obvious in the Dlx5 mutant (Acampora et al., 1999), indicates that the relationship between the two genes may involve complex feedback loops for maintenance of gene expression in vivo, especially during later stages of bone development. Moreover, additional Dlx5 presumptive target genes are activated. Goosecoid, whose expression is decreased in Dlx5 mutant mice (Acampora et al., 1999), is activated by Dlx5 in the cultures; Msx1, which is expressed around the developing bones, is strongly activated in the cultures, whereas Msx2, whose expression is comparable to background around the membrane bones, is not increased by Dlx5 activation. This lack of effect on Msx2 differs from the regulation of Msx2 reported in the limb of Dlx5/Dlx6 mutants, wherein Msx2 expression is lost in the mutant and restored upon Dlx5 re-activation (Robledo et al., 2002). Taken together, these observations suggest that, in this in vitro system, Dlx5 activates several genes characteristic of committed osteoblasts, and that the calvaria cells in culture retain some tissue-specific regulation. However, some patterning information is lost, since Dlx5 overexpression induced Dlx2 or Pitx2 in the calvaria cells, which do not normally express these genes in vivo whereas mandibular osteoblasts do. Future studies will explore which environmental cues control these aspects of Dlx5 activity. In the third domain, we define the differentiated osteoblasts by their expression of FgfR1, osteopontin, and alkaline phosphatase. Interestingly, overexpression of Dlx5 sequentially activated these genes later than the genes defining committed osteoblasts. FgfR1 expression appears at 48 h post-transfection, osteopontin is weakly activated after 4 days, and alkaline phosphatase is strongly activated after 16 days (Figs. 5 and 6).

This result indicates that Dlx5 can initiate and drive a complete bone differentiation program in immature, non committed calvaria precursors. Previous studies have suggested a difference between embryonic bone development, involving Runx2 as a primary osteogenic switch on one hand, and BMP2-induced bone differentiation in cell lines where Dlx5 mediates BMP2 signals to activate Runx2 on the other hand (reviewed in Ryoo et al., 2006). Here, we reconcile the two models of bone differentiation, showing that (i) Dlx5, which is induced by BMP2 in prospective skull bone condensations as early as E8 in vivo (Holleville et al., 2003), is also activated by BMPs in vitro, (ii) Dlx5 can activate Runx2 and osteogenic differentiation in uncommitted embryonic calvarial mesenchyme, and (iii) BMP signals also activate Runx2 in the chick primary calvaria cells grown in vitro and this induction depends on the activity of Dlx factors. Acknowledgments We thank Drs. G. Levi, C. Burns, and D. Roche for helpful comments on the manuscript. We are grateful to Drs. Khotz, Kessel, Millan, Harland, Franceschi, Artinger, Castagnola, and Delfini for their generous gift of several reagents, to S. Gournet, M. Fromaget, and F. Beaujean for help in the illustration work. N. H. was supported by Hoescht-Marion-Roussel-CNRS and Fondation pour la Recherche Médicale fellowships. This study was supported by grants from Association pour la Recherche contre le Cancer (ARC), Hoescht-Marion-Roussel, Ligue contre le Cancer, Fondation Pour la recherche Médicale, Fondation NRJ-Institut de France and Centre National de la Recherche Scientifique to A-H. M-B. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ydbio.2007.01.003. References Acampora, D., Merlo, G.R., Paleari, L., Zerega, B., Postiglione, M.P., Mantero, S., Bober, E., Barbieri, O., Simeone, A., Levi, G., 1999. Craniofacial, vestibular and bone defects in mice lacking the Distal-less-related gene Dlx5. Development 126, 3795–3809. Anderson, J., Burns, H.D., Enriquez-Harris, P., Wilkie, A.O., Heath, J.K., 1998. Apert syndrome mutations in fibroblast growth factor receptor 2 exhibit increased affinity for FGF ligand. Hum. Mol. Genet. 7, 1475–1483. Barlow, A.J., Francis-West, P.H., 1997. Ectopic application of recombinant BMP-2 and BMP-4 can change patterning of developing chick facial primordia. Development 124, 391–398. Benson, M.D., Bargeon, J.L., Xiao, G., Thomas, P.E., Kim, A., Cui, Y., Franceschi, R.T., 2000. Identification of a homeodomain binding element in the bone sialoprotein gene promoter that is required for its osteoblastselective expression. J. Biol. Chem. 275, 13907–13917. Beverdam, A., Merlo, G.R., Paleari, L., Mantero, S., Genova, F., Barbieri, O., Janvier, P., Levi, G., 2002. Jaw transformation with gain of symmetry after Dlx5/Dlx6 inactivation: mirror of the past? Genesis 34, 221–227. Bialek, P., Kern, B., Yang, X., Schrock, M., Sosic, D., Hong, N., Wu, H., Yu, K., Ornitz, D.M., Olson, E.N., Justice, M.J., Karsenty, G., 2004. A twist code determines the onset of osteoblast differentiation. Dev. Cell 6, 423–435.

N. Holleville et al. / Developmental Biology 304 (2007) 860–874 Blin-Wakkach, C., Lezot, F., Ghoul-Mazgar, S., Hotton, D., Monteiro, S., Teillaud, C., Pibouin, L., Orestes-Cardoso, S., Papagerakis, P., Macdougall, M., Robert, B., Berdal, A., 2001. Endogenous Msx1 antisense transcript: in vivo and in vitro evidences, structure, and potential involvement in skeleton development in mammals. Proc. Natl. Acad. Sci. U. S. A. 98, 7336–7341. Choi, K.Y., Kim, H.J., Lee, M.H., Kwon, T.G., Nah, H.D., Furuichi, T., Komori, T., Nam, S.H., Kim, Y.J., Kim, H.J., Ryoo, H.M., 2005. Runx2 regulates FGF2-induced Bmp2 expression during cranial bone development. Dev. Dyn. 233, 115–121. Couly, G.F., Coltey, P.M., Le Douarin, N.M., 1993. The triple origin of skull in higher vertebrates: a study in quail-chick chimeras. Development 117, 409–429. Depew, M.J., Liu, J.K., Long, J.E., Presley, R., Meneses, J.J., Pedersen, R.A., Rubenstein, J.L., 1999. Dlx5 regulates regional development of the branchial arches and sensory capsules. Development 126, 3831–3846. Depew, M.J., Lufkin, T., Rubenstein, J.L., 2002. Specification of jaw subdivisions by Dlx genes. Science 298, 381–385. Dodig, M., Tadic, T., Kronenberg, M.S., Dacic, S., Liu, Y.H., Maxson, R., Rowe, D.W., Lichtler, A.C., 1999. Ectopic Msx2 overexpression inhibits and Msx2 antisense stimulates calvarial osteoblast differentiation. Dev. Biol. 209, 298–307. Ducy, P., Starbuck, M., Priemel, M., Shen, J., Pinero, G., Geoffroy, V., Amling, M., Karsenty, G., 1999. A Cbfa1-dependent genetic pathway controls bone formation beyond embryonic development. Genes Dev. 13, 1025–1036. Etchevers, H.C., Vincent, C., Le Douarin, N.M., Couly, G.F., 2001. The cephalic neural crest provides pericytes and smooth muscle cells to all blood vessels of the face and forebrain. Development 128, 1059–1068. Fragale, A., Tartaglia, M., Bernardini, S., Di Stasi, A.M., Di Rocco, C., Velardi, F., Teti, A., Battaglia, P.A., Migliaccio, S., 1999. Decreased proliferation and altered differentiation in osteoblasts from genetically and clinically distinct craniosynostotic disorders. Am. J. Pathol. 154, 1465–1477. Francis-West, P.H., Tatla, T., Brickell, P.M., 1994. Expression patterns of the bone morphogenetic protein genes Bmp-4 and Bmp-2 in the developing chick face suggest a role in outgrowth of the primordia. Dev. Dyn. 201, 168–178. Francis-West, P., Ladher, R., Barlow, A., Graveson, A., 1998. Signalling interactions during facial development. Mech. Dev. 75, 3–28. Gerstenfeld, L.C., Chipman, S.D., Glowacki, J., Lian, J.B., 1987. Expression of differentiated function by mineralizing cultures of chicken osteoblasts. Dev. Biol. 122, 49–60. Holleville, N., Quilhac, A., Bontoux, M., Monsoro-Burq, A.-H., 2003. BMP signals regulate Dlx5 during early avian skull development. Dev. Biol. 257, 177–189. Hu, Y.S., Zhou, H., Kartsogiannis, V., Eisman, J.A., Martin, T.J., Ng, K.W., 1998. Expression of rat homeobox gene, rHOX, in developing and adult tissues in mice and regulation of its mRNA expression in osteoblasts by bone morphogenetic protein 2 and parathyroid hormone-related protein. Mol. Endocrinol. 12, 1721–1732. Iseki, S., Wilkie, A.O., Heath, J.K., Ishimaru, T., Eto, K., Morriss-Kay, G.M., 1997. Fgfr2 and osteopontin domains in the developing skull vault are mutually exclusive and can be altered by locally applied FGF2. Development 124, 3375–3384. Iseki, S., Wilkie, A.O., Morriss-Kay, G.M., 1999. Fgfr1 and Fgfr2 have distinct differentiation- and proliferation-related roles in the developing mouse skull vault. Development 126, 5611–5620. Izpisua-Belmonte, J.C., De Robertis, E.M., Storey, K.G., Stern, C.D., 1993. The homeobox gene goosecoid and the origin of organizer cells in the early chick blastoderm. Cell 74, 645–659. Kim, H.J., Rice, D.P., Kettunen, P.J., Thesleff, I., 1998. FGF-, BMP- and Shhmediated signalling pathways in the regulation of cranial suture morphogenesis and calvarial bone development. Development 125, 1241–1251. Kohtz, J.D., Lee, H.Y., Gaiano, N., Segal, J., Ng, E., Larson, T., Baker, D.P., Garber, E.A., Williams, K.P., Fishell, G., 2001. N-terminal fatty-acylation of sonic hedgehog enhances the induction of rodent ventral forebrain neurons. Development 128, 2351–2363. Komori, T., Yagi, H., Nomura, S., Yamaguchi, A., Sasaki, K., Deguchi, K., Shimizu, Y., Bronson, R.T., Gao, Y.H., Inada, M., Sato, M., Okamoto, R.,

873

Kitamura, Y., Yoshiki, S., Kishimoto, T., 1997. Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell 89, 755–764. Kuratani, S., Martin, J.F., Wawersik, S., Lilly, B., Eichele, G., Olson, E.N., 1994. The expression pattern of the chick homeobox gene gMHox suggests a role in patterning of the limbs and face and in compartmentalization of somites. Dev. Biol. 161, 357–369. Lanctot, C., Moreau, A., Chamberland, M., Tremblay, M.L., Drouin, J., 1999. Hindlimb patterning and mandible development require the Ptx1 gene. Development 126, 1805–1810. Le Lievre, C.S., 1978. Participation of neural crest-derived cells in the genesis of the skull in birds. J. Embryol. Exp. Morphol. 47, 17–37. Lee, M.H., Kim, Y.J., Kim, H.J., Park, H.D., Kang, A.R., Kyung, H.M., Sung, J.H., Wozney, J.M., Kim, H.J., Ryoo, H.M., 2003. BMP-2-induced Runx2 expression is mediated by Dlx5, and TGF-beta 1 opposes the BMP-2induced osteoblast differentiation by suppression of Dlx5 expression. J. Biol. Chem. 278, 34387–34394. Lenton, K.A., Nacamuli, R.P., Wan, D.C., Helms, J.A., Longaker, M.T., 2005. Cranial suture biology. Curr. Top. Dev. Biol. 66, 287–328. Liu, Y.H., Tang, Z., Kundu, R.K., Wu, L., Luo, W., Zhu, D., Sangiorgi, F., Snead, M.L., Maxson, R.E., 1999. Msx2 gene dosage influences the number of proliferative osteogenic cells in growth centers of the developing murine skull: a possible mechanism for MSX2-mediated craniosynostosis in humans. Dev. Biol. 205, 260–274. Logan, M., Tabin, C.J., 1999. Role of Pitx1 upstream of Tbx4 in specification of hindlimb identity. Science 283, 1736–1739. Martin, J.F., Bradley, A., Olson, E.N., 1995. The paired-like homeo box gene MHox is required for early events of skeletogenesis in multiple lineages. Genes Dev. 9, 1237–1249. Merlo, G.R., Paleari, L., Mantero, S., Zerega, B., Adamska, M., Rinkwitz, S., Bober, E., Levi, G., 2002. The Dlx5 homeobox gene is essential for vestibular morphogenesis in the mouse embryo through a BMP4-mediated pathway. Dev. Biol. 248, 157–169. Miyama, K., Yamada, G., Yamamoto, T.S., Takagi, C., Miyado, K., Sakai, M., Ueno, N., Shibuya, H., 1999. A BMP-inducible gene, dlx5, regulates osteoblast differentiation and mesoderm induction. Dev. Biol. 208, 123–133. Monsoro-Burq, A.H., Bontoux, M., Vincent, C., Le Douarin, N.M., 1995. The developmental relationships of the neural tube and the notochord: short and long term effects of the notochord on the dorsal spinal cord. Mech. Dev. 53, 157–170. Nakashima, K., Zhou, X., Kunkel, G., Zhang, Z., Deng, J.M., Behringer, R.R., de Crombrugghe, B., 2002. The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell 108, 17–29. Newberry, E.P., Latifi, T., Towler, D.A., 1998. Reciprocal regulation of osteocalcin transcription by the homeodomain proteins Msx2 and Dlx5. Biochemistry 37, 16360–16368. Opperman, L.A., 2000. Cranial sutures as intramembranous bone growth sites. Dev. Dyn. 219, 472–485. Panganiban, G., Sebring, A., Nagy, L., Carroll, S., 1995. The development of crustacean limbs and the evolution of arthropods. Science 270, 1363–1366. Pera, E., Stein, S., Kessel, M., 1999. Ectodermal patterning in the avian embryo: epidermis versus neural plate. Development 126, 63–73. Poliard, A., Nifuji, A., Lamblin, D., Plee, E., Forest, C., Kellermann, O., 1995. Controlled conversion of an immortalized mesodermal progenitor cell towards osteogenic, chondrogenic, or adipogenic pathways. J. Cell Biol. 130, 1461–1472. Puelles, L., Kuwana, E., Puelles, E., Rubenstein, J.L., 1999. Comparison of the mammalian and avian telencephalon from the perspective of gene expression data. Eur. J. Morphol. 37, 139–150. Qiu, M., Bulfone, A., Martinez, S., Meneses, J.J., Shimamura, K., Pedersen, R.A., Rubenstein, J.L., 1995. Null mutation of Dlx-2 results in abnormal morphogenesis of proximal first and second branchial arch derivatives and abnormal differentiation in the forebrain. Genes Dev. 9, 2523–2538. Qiu, M., Bulfone, A., Ghattas, I., Meneses, J.J., Christensen, L., Sharpe, P.T., Presley, R., Pedersen, R.A., Rubenstein, J.L., 1997. Role of the Dlx

874

N. Holleville et al. / Developmental Biology 304 (2007) 860–874

homeobox genes in proximodistal patterning of the branchial arches: mutations of Dlx-1, Dlx-2, and Dlx-1 and -2 alter morphogenesis of proximal skeletal and soft tissue structures derived from the first and second arches. Dev. Biol. 185, 165–184. Rice, D.P., Rice, R., Thesleff, I., 2003. Molecular mechanisms in calvarial bone and suture development, and their relation to craniosynostosis. Eur. J. Orthod. 25, 139–148. Robledo, R.F., Rajan, L., et al., 2002. The Dlx5 and Dlx6 homeobox genes are essential for craniofacial, axial, and appendicular skeletal development. Genes Dev. 16, 1089–1101. Ryoo, H.M., Hoffmann, H.M., Beumer, T., Frenkel, B., Towler, D.A., Stein, G.S., Stein, J.L., van Wijnen, A.J., Lian, J.B., 1997. Stage-specific expression of Dlx-5 during osteoblast differentiation: involvement in regulation of osteocalcin gene expression. Mol. Endocrinol. 11, 1681–1694. Ryoo, H.M., Lee, M.H., Kim, Y.J., 2006. Critical molecular switches involved in BMP-2-induced osteogenic differentiation of mesenchymal cells. Gene 366, 51–57. Sasaki, T., Ito, Y., Bringas Jr., P., Chou, S., Urata, M.M., Slavkin, H., Chai, Y., 2006. TGFbeta-mediated FGF signaling is crucial for regulating cranial

neural crest cell proliferation during frontal bone development. Development 133, 371–381. Tadic, T., Dodig, M., Erceg, I., Marijanovic, I., Mina, M., Kalajzic, Z., Velonis, D., Kronenberg, M.S., Kosher, R.A., Ferrari, D., Lichtler, A.C., 2002. Overexpression of Dlx5 in chicken calvarial cells accelerates osteoblastic differentiation. J. Bone Miner. Res. 17, 1008–1014. Takahashi, Y., Bontoux, M., Le Douarin, N.M., 1991. Epithelio–mesenchymal interactions are critical for Quox 7 expression and membrane bone differentiation in the neural crest derived mandibular mesenchyme. EMBO J. 10, 2387–2393. Wang, Y.H., Rutherford, B., Upholt, W.B., Mina, M., 1999. Effects of BMP-7 on mouse tooth mesenchyme and chick mandibular mesenchyme. Dev. Dyn. 216, 320–335. Wilke, T.A., Gubbels, S., Schwartz, J., Richman, J.M., 1997. Expression of fibroblast growth factor receptors (FGFR1, FGFR2, FGFR3) in the developing head and face. Dev. Dyn. 210, 41–52. Woda, J.M., Pastagia, J., Mercola, M., Artinger, K.B., 2003. Dlx proteins position the neural plate border and determine adjacent cell fates. Development 130, 331–342.