Postnatal Msx1 expression pattern in craniofacial, axial, and appendicular skeleton of transgenic mice from the first week until the second year

Authors

  • Silvana Maria Orestes-Cardoso,

    Corresponding author
    1. Laboratoire de Biologie Odontologie, EA 2380, Université Paris 7, IFR 58, Institut Biomédical des Cordeliers, Esc. E-2è ét., Paris, France
    2. Faculdade de Odontologia de Pernambuco da Universidade Federal de Pernambuco (UFPE), Recife, Brazil
    • Silvana Maria Orestes-Cardoso, Laboratoire de Biologie Odontologie, EA-2380, Université Paris 7, IFR 58, Institut Biomédical des Cordeliers, Esc. E - 2è ét., 15-21 rue de l'Ecole de Médecine, 75270 Paris Cedex 06, France
    Search for more papers by this author
  • Jean Raphael Nefussi,

    1. Laboratoire de Biologie Odontologie, EA 2380, Université Paris 7, IFR 58, Institut Biomédical des Cordeliers, Esc. E-2è ét., Paris, France
    Search for more papers by this author
  • Dominique Hotton,

    1. Laboratoire de Biologie Odontologie, EA 2380, Université Paris 7, IFR 58, Institut Biomédical des Cordeliers, Esc. E-2è ét., Paris, France
    Search for more papers by this author
  • Mohand Mesbah,

    1. Laboratoire de Biologie Odontologie, EA 2380, Université Paris 7, IFR 58, Institut Biomédical des Cordeliers, Esc. E-2è ét., Paris, France
    Search for more papers by this author
  • Maria Do Socorro Orestes-Cardoso,

    1. Faculdade de Odontologia de Pernambuco (FOP), Recife, Brazil
    Search for more papers by this author
  • Benoît Robert,

    1. Unité Génétique Moléculaire de la Morphogenèse, and CNRS URA 1947, Institut Pasteur, Paris, France
    Search for more papers by this author
  • Ariane Berdal

    1. Laboratoire de Biologie Odontologie, EA 2380, Université Paris 7, IFR 58, Institut Biomédical des Cordeliers, Esc. E-2è ét., Paris, France
    Search for more papers by this author

Abstract

Phenotypes associated with Msx1 mutations have established the prominent role of this divergent homeogene in skeletal patterning. Previous studies have been achieved during antenatal development in relation with the early death of null mutant mice. Therefore, the present study is devoted to Msx1 homeogene in the postnatal craniofacial, axial, and appendicular skeleton. A knock-in transgenic mouse line was studied from the first postnatal week until 15 months. Whole-mount β-galactosidase enzymology identified Msx1 protein expression pattern. Maintained expression of Msx1 was observed in growing and adult mice, specifically in the sites where Msx1 plays an early morphogenetic role during initial skeletal patterning. These included the craniofacial sutures, autopodium, mandible, and alveolar bone. Furthermore, active membranous and endochondral bone formation involved Msx1 in the entire skeleton. Histologic sections showed that progenitor as well as differentiating and differentiated cells of all the bone cell lineages could express the Msx1 protein (chondrocytes, osteoblasts, tartrate-resistant acid phosphatase positive osteoclasts and chondroclasts). Recent developments in the genetic and developmental biology of skeletal morphogenesis demonstrate that genes critical for development are jointly expressed in discrete embryonic signalling and growth centers, the enamel knot in teeth, the cranial suture in skull morphogenesis, and the progress zone in the limb buds. The present study suggests that these signalling pathways are jointly important throughout the entire lifetime with an exquisite site-specificity spatially related to early patterning. © 2001 Wiley-Liss, Inc.

INTRODUCTION

Bone is formed through a multistep process involving neural crest-derived cells for most of the craniofacial skeleton (Couly et al., 1993; Köntges and Lumsden, 1996; Le Douarin et al., 1993; Noden, 1988) and mesoderm-derived cells for the axial and appendicular skeleton (Erlebacher et al., 1995; Monsoro-Burq et al., 1994). The molecular pathway of bone formation is presently delineated through the analysis of genetically engineered mice and advances in human genetics (for review, see Ducy and Karsenty, 1998). These studies have provided information essentially on the initial developmental processes: early patterning and cell commitment and differentiation in the chondrocytic, osteoblastic, and osteoclastic cell lineages. Homeobox genes are involved in the patterning. These genes constitute a particular subset of regulatory genes encoding nuclear proteins that contain a highly conserved homeodomain (Boncinelli, 1997). While in the axial and appendicular skeleton, Hox homeogenes are determinant, they are not expressed in the developing head and facial primordia. Several divergent homeogenes have been proposed as alternative candidates for the patterning of craniofacial skeleton. Among these genes, members of Msx (Muscle segment homeobox gene) have recently retained much attention regarding the distinct disruption of early skeletal development in their transgenic null mutant mice (Houzelstein et al., 1997; Liu et al., 1995; Satokata and Maas, 1994; Satokata et al., 2000), and site-specific disorders of the skeleton related to human gene mutations (Francomano et al., 1996; Hollway et al., 1995; Jabs et al., 1993; van den Boogaard et al., 2000; Vastardis et al., 1996) and polymorphism (Hwang et al., 1998). Bone formation occurs through two distinct differentiation pathways. In membranous bone formation (Hall and Myake, 1995), committed cells directly differentiate into osteoblasts. In contrast, endochondral bone formation requires the presence of a temporary cartilage template which is replaced by bone after vascular invasion (Erlebacher et al., 1995; Horton, 1993; Huang et al., 1997). In association with osteoblasts, a third specific cell-type of the skeleton, the osteoclast or bone resorbing cell, is derived from hematopoietic stem cells (Suda et al., 1992). During endochondral bone formation, chondroclasts resorbed cartilage matrix in the growth plate cartilage where longitudinal growth occurs (Vu et al., 1998). The genetic cascade that controls osteoclast differentiation involves several transcription factors and secreted proteins (Hofbauer et al., 2000). The clearest evidence for the activation of skeletal cell differentiation was established in the osteoblast with the convergent identification of a transcription factor NMP2/Osf2/Cbfa1 leading differentiation and targeted osteocalcin expression (Ducy et al., 1997; Merriman et al., 1995; Mundlos et al., 1997; Otto et al., 1997). Interestingly, the osteocalcin gene is also controlled by transcription factors important in skeletal patterning, notably the Msx homeoproteins in osteoblasts in vitro (Newberry et al., 1997; Towler et al., 1994).

Most of the studies on transcription factors in the skeleton have been achieved during the embryonic period in relation with early lethality of null mutant mice (Chen et al., 1996; Ducy and Karsenty, 1998; Mackenzie et al., 1991a,b; Vainio et al., 1993). Recent studies on Cbfa1 have shown that, in fact, this transcription factor involved in early development may also play a role during postnatal stages (Ducy et al., 1999). Some descriptive data during the postnatal period have documented the expression of a member of Msx homeogene family, notably Msx1. However, the investigation was restricted to the first week (Kim et al., 1998). Therefore, the present study is devoted to Msx1 in the postnatal skeleton. This gene was selected because it seems to play an important role in the patterning of the entire skeleton. Indeed, Msx1-/- null mutant homozygous mice show a craniofacial phenotype including reduced size of the mandible, cleft palate, and tooth agenesis. The mutation is lethal in homozygous Msx1-/- mice. No phenotype is observed in developing limb, presumably in relation with the overlap of expression patterns of Msx1 and Msx2 or other compensatory genes (Houzelstein et al., 1997; Satokata and Maas, 1994; Satokata et al., 2000). However, the limbless mutation in mice is associated with the down-regulation of Msx1 in the progress zone of limb buds (Coelho et al., 1993). Furthermore, a rare allele of the human Msx1 gene is related to congenital malformations of the limbs (Hwang et al., 1998). The variability of the phenotypes related to Msx1 mutation highlight a dosage effect of Msx1 gene product. Therefore, the knowledge of this homeogene expression pattern inside the skeleton actually seems to be a prerequisite in the understanding of diverse skeletal disorders induced by Msx1 mutations.

A transgenic mouse line bearing a mutation of the Msx1 gene was used here. Briefly, a reporter nlacZ gene, which contains a nuclear localization signal (n), was introduced into the second exon of mouse Msx1 gene. This mutated gene encodes for a Msx1-β-galactosidase fusion protein that allows the study of Msx1 gene expression by β-galactosidase histoenzymology (Houzelstein et al., 1997). The investigation of endogenous Msx1 promoter activity and protein synthesis is made possible by the absence of phenotypic abnormalities in Msx1+/- heterozygous mice (Houzelstein et al., 1997; Satokata and Maas, 1994). The present study is devoted to the analysis of the potential postnatal role of Msx1 homeogene and protein during bone modeling in growing mice and homeostasis in adults. This investigation compares the craniofacial complex and the axial and appendicular skeleton. The study period ranged from 14 postnatal (P14) days to 15 months (15M). The obtained data show that Msx1 may contribute to bone tissue growth, involving several osteogenic fronts: craniofacial sutures, growth plate cartilage, and periosteum. Furthermore, postnatal Msx1 overexpression characterized the sites where this homeogene specifically plays a role in skeletal patterning: the distal extensions of the mandible and limbs, sutures, and alveolar bone associate with teeth.

RESULTS

The first stages of postnatal development in mice were investigated at P14 and P21 during the intense growth associated with active area of bone cell proliferation and differentiation. Observations were realized at P14 and P21 in several sites: the craniofacial bone and sutures (Figs. 1A,B, 2A–C), mandible (1G,H), sternum (Fig. 4A), clavicles (Fig. 4A), ribs (Figs. 4A,B, 5E,F), vertebras (Figs. 4E, 5A,B), tail (Figs. 4F, 5C,D), and limbs (Figs. 4A–C, 6A,D–G). During the later period, from 3M to 15M, Msx1 expression was also studied in the same areas (Figs. 1C–F,I–L, 2B–D,G, 3A–G, 4D, 6B,C, 1H–J).

Figure 1.

Overall investigation of craniofacial ossification from 14 postnatal days until 15 months by whole-mount analysis of β-galactosidase. The skull is illustrated during bone growth at 14 (P14; A) and 21 (P21; B) postnatal days as well as in adult lifetime at 3 (3M; C), 6 (6M; D), 9 (9M; E), and 15 months (15M; F). Upper views show the continuous expression of nlacZ-Msx1 fusion protein in the sutures and distal segment of the skull (nasal bones and cartilage). Msx1 expression levels decrease significantly from P21 to 3M and progressively later on. Calvarial bone surfaces are clearly stained at P14 (A) and weakly at P21 (B) but no staining is present afterward (C–F). Similarly, the zygomatic process of the maxilla is only labelled at P14 and P21 (B). The fluctuant Msx1 expression pattern in the mandible is shown at P14 (G,H), 6M (I,J), and 15M (K,L). At P14, a double distoproximal and aboral-oral gradient of Msx1 expression is observed in the mandibular bone. The staining is more significant in the labial (G) than in the lingual (H) aspect. No staining is detected in the alveolar bone supporting the molar and incisor. In contrast, I and J show a shift in Msx1 expression pattern at 6M with a staining of the superficial part of the incisor and molar periodontium. At 15M (K,L), Msx1 expression remains in the same location (K), albeit with a weak labeling on the lingual side (L). Original magnifications: A, ×3.5; B, ×2.9; C, ×2.4; D, ×2.5; E, ×2.4; F, ×2.4; G, ×7; H, ×7; I, ×5; J, ×5; K, ×5.3; L, ×5.3.

Figure 2.

Histologic analysis of membranous bone formation in the sutures and maxilla. High magnification of the skull at postnatal day 14 (P14; A) and at 6 months of age (6M; B) shows the distinct Msx1 pattern in the sagittal (Sa) and parietointerparietal (Pa-Int) sutures. The staining decreases in the sutures and disappears on the bone surface at 6M. C: Histologic frontal section at the level of the sagittal suture (star) and bone periosteum identifies the labelled cells visualized by whole-mount analysis at P21 (Fig 1,B). D: Inferior view of the maxilla and palate at 3M shows the stained nasal cartilages and incisivum bone. Msx1 is expressed on the labial aspect of the incisal alveolar bone, the adjoining diastema zone (arrow D) and the molar alveolar bone (arrow M) of the maxilla. E: High magnification of the molar area clearly evidences the staining of the labial alveolar bone (arrow M) and the absence of staining in the adjoining teeth and palatal mucosa (star). F: On frontal section of maxillary bone, no Msx1 staining is recorded in the dentoalveolar ligament and molar teeth. Only a restricted periosteal surface is labelled (arrow). This specific anatomic site corresponds to a local outgrowth generated by periosteal apposition of lamellar bone (arrow). G: High magnification delineates the positive cells (arrow) inside the periosteum. Original magnifications: A, ×7.5; B, ×5.5; C, ×560; D, ×5; E, ×8; F, ×70; G, ×280.

Figure 3.

Longitudinal sections of the mandible at 3 months of age (3M) and 6M. This figure provides a cellular analysis of the whole-mount observations illustrated on Figure 1 (I,J). A: The distribution of Msx1-positive cells in the dentoalveolar ligament of incisors is organized with an aboral-oral and distoproximal double gradient (arrows). B: The Msx1-positive cells pertain in one hand to the cementum-forming side (Ce) and in the other hand to the alveolar bone-forming side (Bo). The central area of dentoalveolar ligament is devoid of positive cells. C: High magnification of the oral periosteal surface observed on A is presented. Msx1-positive cells appear disseminated in the superficial periodontium including the regions of periosteal bone surface (Pe), dentoalveolar ligament (Dal), and gingival chorion (Gc). D: High magnification of the aboral bone observed on A shows that both the osteoblasts lining bone surfaces (Ob) and osteocytes embedded in bone matrix (Ocy) are expressing Msx1 inside the trabecular bone. E,F: The oral area of the periodontium of the molar area is shown at 3M. The most superficial periodontium shows staining (+). The more internal periodontium does not contain Msx1-positive cells (−). Msx1-positive cells are observed on the periosteal surface (Pe), in the most external dentoalveolar ligament (Dal) including the bone- and cementum-forming sides, and finally in the gingival chorion (Gc). G:The deeper zones of the periodontium do not contain Msx1-positive cells, as illustrated in the apical third of a molar. Original magnifications: A, ×70; B, ×280; C, ×280; D, ×560; E, ×280; F, ×560; G, ×280.

Figure 4.

Overall investigation of the axial and appendicular skeleton at postnatal day 14 (P14), P21, and 9 months (9M) by whole-mount analysis. Frontal (A) and lateral (B) views of the body at P21 show the distinct Msx1 distribution in the clavicles (C), autopodium of upper limbs (star), ribs (R), and sternum (arrow). C: At P14, Msx1 labeling is related to the epiphysis of limb long bones and, more prominently, to the autopodium (A). D: At 9M, only the autopodium (A) contains significant Msx1 labeling. At P21, the vertebras also express the chimeric Msx1-β-galactosidase in association with endochondral bone formation in the areas of thoracic vertebras (E) and tail (F). Original magnifications: A, ×5; B, ×5; C, ×5; D, ×3.5; E, ×9; F ×9.

In the present study, Msx1 was distinctly and permanently associated with most of the craniofacial sutures from P14 until 15M, as shown previously during early development by in situ hybridization (Kim et al., 1998) and confirmed in Msx1-nlacZ transgenic mice (data not shown). In the sutures, the same anatomic distribution was observed here during allometric growth (P14 and P21) and in adult skeleton (3, 6, 9, 12, and 15M). However, the signal intensity significantly decreased between P14 and 3M. Furthermore, at P14, Msx1 was also expressed at the level of the periosteal surface of the developing skull bones. This finding was clearly evidenced at P14 (Figs. 1A, 2A) and, albeit with a lower intensity, at P21 (Fig. 1B). At 3M and later on, the transgene was not expressed on the periosteal surface of the calvarial bones, except on the distal segments of the skull, i.e., nasal bones and cartilage (Figs. 1C–F, 2B).

In contrast to the continuous pattern described here in the sutures, Msx1 distribution in the maxilla and mandible was fluctuating, following the three major steps of bone formation: patterning in embryonic life, allometric growth in postnatal period, and homeostasis in adult life. First, during early development, the distribution of Msx1 fusion protein (data not shown; Houzelstein et al., 1997) showed that the protein was expressed in the corresponding areas where Msx1 mRNA has been previously reported (Mackenzie et al., 1991a,b). These Msx1-positive cells included the progenitor neural-crest derived cells of alveolar bone for the entire dentition and basal bone of maxilla and mandible. Second, during the period of bone growth (birth−1M), Msx1 expression stopped in alveolar bone of all molars and the upper incisors. Alveolar bone cells of the mandibular incisor and the adjoining mandibular basal bone maintained a high expression level of the transgene, as shown here at P14 (Fig. 1G,H). This pattern was similar to the one observed during early development, i.e., with a clear distoproximal and aboral-oral double gradient. Third, in adult mice, the pattern differed, as illustrated at 6M (Fig. 1I,J) and 15M (Fig. 1K,L). Msx1 appeared to be clearly re-expressed in restricted areas of the periosteal surface of maxillary and mandibular bone. Distinct patterns were discerned in these sites. Whole-mount studies showed that Msx1 was distributed in the labial bone adjoining the upper molars, diastema, and incisor areas (Fig. 2D,E). Histologic serial sections demonstrated that Msx1-positive cells were restricted to the periosteum of a large labial bone outgrowth (Fig. 2F,G). In the mandible, whole-mount studies showed that Msx1 expression was present in the entire oral side of the alveolar bone associated with incisors as described in early stages, but also with molars at 6M (Fig. 1I,J) and 15M (Fig. 1K,L). During late process, Msx1 expression significantly decreased (15M; Fig. 1K,L), as previously described in the sutures.

In the axial and appendicular skeleton (P14 and P21), whole-mount staining illustrates the generalized Msx1 presence in the sites of endochondral bone formation, i.e., limbs (Fig. 4A– C), vertebras (Fig. 4E) including the ones located in the tail (Fig. 4F), sternum, and ribs (Fig. 4A,B). Specific sites (distal region of the limbs) maintained an high expression level throughout mice lifetime (Fig. 4A–C). Histologic sections allowed the identification of Msx1-expressing cells. Numerous Msx1-positive cells were present in the ossification centers of the epiphyseal growth plate cartilage (Fig. 5A–F). The same distribution patterns were found in the various sites of endochondral bone formation presently investigated: cervical (Fig. 5A,B) and caudal (Fig. 5C,D) vertebras and ribs (Fig. 5E,F). Msx1-positive cells were identified as chondroclasts and osteoclasts, based on their size, numerous nuclei, and insertion in cartilage (Fig. 5A) and bone lacunae (Fig. 5B). Colocalization of TRAP and Msx1 confirmed the cell identity (Fig. 5B). On the other hand, the site-specific continuity of Msx1 expression in the autopodium is illustrated in the upper limbs at P21 (Fig. 6A), 6M (Fig. 6B) and 9M (Fig. 6C). In bone diaphyses, Msx1-positive cells were identified in the periosteum (Fig. 6E–I). And, the articular cartilage was also characterized by a striking continuous Msx1 expression in the most external chondrocytes, specifically in the autopodium (Fig. 6G–J). In contrast, in all bones, neither osteoblasts nor osteoclasts were labeled in the epiphysis when the activity of endochondral ossification centers stopped (shown here in the autopodium at 6M, Fig. 6H).

Figure 5.

Histologic analysis of the axial and appendicular skeleton. A: Transverse section of the ribs at postnatal day 21 (P21) shows that the staining is limited to the areas of endochondral bone formation (arrow). B: At higher magnification, Msx1 protein is clearly localized in the osteoclasts and chondroclasts. These cells are large and contain numerous nuclei (arrows) inserted in calcified cartilage and bone lacunae (L). Localization of tartrate-resistant acid phosphatase (TRAP) allows the demonstration of the coexpression of TRAP and Msx1 in these same cells (insert). C: The same association of Msx1 expression and growth plate cartilage is observed in the vertebras at P21. Msx1-positive cell location corresponds to the osteogenic front. D: Higher magnification evidences the large Msx1-positive cells inserted in the calcified cartilage (arrow) and bone (star) lacunae. E: Finally, the rib growth plate cartilage at P21 illustrates the same pattern. F: Msx1-positive cells may be identified on a higher magnification as large cells in contact with the calcified cartilage (arrow). Original magnifications: A, ×280; B, ×1.400; (Insert, ×1,400); C, ×70; D, ×560; E, ×140; F, ×560.

Figure 6.

Continuous Msx1 expression in the upper limb autopodium. Whole-mount staining in the autopodium at postnatal day 21 (P21; A), 6 months (6M; B), and 9M (C) shows the maintained Msx1 expression associated with the autopodium, albeit with a progressive slight decrease. D: In transverse sections of the phalanges at P21, Msx1 is expressed on the periosteal bone surfaces, in articular cartilage, and also in the active growth plate cartilage. E: High magnification of the periosteal bone surface indicates the strongly stained osteoblasts. F: High magnification of the growth plate cartilage illustrates the constant expression in chondroclasts (arrow). G Finally, the most external part of articular cartilage was significantly expressing Msx1. H: Transverse section of the autopodium at 6M confirms the significant decrease in Msx1 labeling intensity, observed by whole-mount analysis (comparing A, B, and C). No labeling is recorded in the epiphysis, in relation with the achievement of growth plate cartilage activity. The distinct Msx1-positive cells were the cells of the periosteum (I) and the most external articular chondrocytes (J). Original magnifications: A, ×7; B, ×5; C, ×5; D, ×140; E, ×560; F, ×560; G, ×1.400; H, ×140; I, ×280; J, ×1.400.

DISCUSSION

Previous data (Hill et al., 1989; Houzelstein et al., 1997; Mackenzie et al., 1991a,b; Shashikant et al., 1991) obtained in the early patterning provide, along with the present study, an overall knowledge of Msx1 expression pattern in the skeleton from Theiler stage E10.5 to the second postnatal year.

Msx1 Expression in the Craniofacial Sutures and Skull Bones

Previous investigations on Msx1 in prenatal and postnatal development (Kim et al., 1998) and the present study suggest that Msx1 plays a role in skull suture formation. Interestingly, a human disease, which results from a premature fusion of sutures, the craniosynostosis, has been related to mutation in several genes encoding FGFRs, TWIST, and also notably MSX homeoproteins (El Ghouzzi et al., 1997; Hollway et al., 1995; Howard et al., 1997; Jabs et al., 1993). Approximately 100 syndromes associated with craniosynostosis are known, and more than 50 have a genetic basis. These syndromes are usually monogenic and inherited as autosomal dominant traits (Wilkie, 1997). For instance, Adelaide-type craniosynostosis is a rare autosomal dominant syndrome associated with digital abnormalities. The gene responsible for Adelaide-type craniosynostosis is localized to 4p16 (Hollway et al., 1995). This region contains two potential candidate genes, the MSX1 homeobox gene and the FGFR3 fibroblast growth factor receptor gene (Hollway et al., 1995). The present data on the murine Msx1 gene further support the possibility that the human MSX1 gene is a relevant candidate in the disorder of skull sutures.

The vertebrate skull is a complex structure that must function as a rigid, protective barrier and, at the same time, must adjust for the expansion of a developing brain. These dual requirements are met by the separation of the bony plates of skull with Msx1-positive sutures that allow the calvarial bones to expand as the brain enlarges (Liu et al., 1995). During the growth period (P14 and P21), cranial sutures, which are mesenchymal fibrous joints, may be considered as growth centers implicated in the coordination of the skull growing bones in relation with the slowly expanding brain. Calvarial skull flat bones physiologically grow in two steps. First, these bones grow outward from osteogenesis centers and unify in fibrous joints. Interestingly, Msx1 was present in the periosteum during this allometric bone growth (P14 and P21). Then, in a second step and in relation with brain expansion, the growth of calvarial bones takes place in sutures where Msx1 was continuously expressed. These cranial sutures are characterized initially by a wedge-shaped proliferation of cells at the periphery of the expanding bone field, called the osteogenic fronts (Ofs), whose cells undergo osteogenic differentiation to become mature osteoblasts, which then lay down bone matrix (Hall, 1971; Kim et al., 1998). After the achievement of brain growth, the calvarial bones merge. Interestingly, the present study documented that Msx1 expression was down-regulated during this process (Fig. 2A–C). Cross-talks between BMP and FGF growth and differentiation factors secreted by the dura mater and mesenchymal Msx homeogenes have been identified experimentally during the interactions leading to formation and maintenance of the sagittal suture during embryonic calvarial development (Opperman et al., 1995). Msx1 and Msx2 homeogene expression in the mesenchymal cells were induced by BMP4 (Kim et al., 1998), as shown in developing tooth germs where BMP4 is secreted by the dental epithelium (Chen et al., 1996; Vainio et al., 1993). In the present study, only Msx1 was involved postnatally in the sagittal suture during the first postnatal week (see also Kim et al., 1998) as Msx2 appeared to be locally down-regulated. The present data support the initial proposal by Kim et al. (1998) that Msx1 may play a pivotal role in the membranous bone formation of the calvaria and morphogenesis of mouse sagittal suture. The presently described Msx1 pattern extends this proposal to the entire postnatal growth and other craniofacial sutures. In conclusion, the data on human craniosynostosis, experimental developmental biology, and the present study support the existence of several cross-talks in the Msx1 signalling pathway, which would regulate the rate and geometry of cranial bone growth and the timing of suture closure and, therefore, the shape of the skull (Kim et al., 1998).

Msx1 Expression in the Maxilla and Mandible

In the first branchial arch, Msx1 is expressed in neural crest-derived cells before (Hill et al., 1989; Houzelstein et al., 1997), during, and after (Shashikant et al., 1991) their migration and, as shown here, transformation into mesenchyme cells and even after their differentiation. Numerous human congenital defects result from an abnormal development of the first branchial arch derived from neural crests. These pathologic conditions include 1.) oligodontia, cleft lip and palate related to two distinct mutations of MSX1 gene (Hu et al., 1998; van den Boogaard et al., 2000; Vastardis et al., 1996); and 2.) a series of craniofacial syndromes related to the mutation of other genes, such as Waardenburg's syndrome (Waardenburg, 1951), Treacher Collins' syndrome (Sulik et al., 1987), Pierre-Robin's sequence (Dennison, 1965), velo-cardio-facial syndrome (Shprintzen et al., 1978), and DiGeorge's syndrome (Van Mierop and Kutsch, 1986). Consistently, transgenic mice bearing mutations of several genes, i.e., Edn1 (Kurihara et al., 1995), dHand (Thomas et al., 1998b), and also Msx1 (Houzelstein et al., 1997; Satokata and Maas, 1994) harbor similar craniofacial phenotypes. Analysis of their respective disruption allowed the proposal (Thomas et al., 1998b) of the existence of a molecular pathway between endothelin1, dHand, and Msx1.

The Msx1 expression in postnatal stages, presently evidenced, supports a role for this homeogene in the later process of bone modeling during craniofacial growth and bone homeostasis in adult mice. The observed discontinuous expression in molars and continuous expression in incisors suggests that late Msx1 function may be specifically related to the continuous eruption of the rodent incisors (Warshawsky and Moore, 1997). However, the continuously erupting incisors in the maxilla showed the same arrest of Msx1 expression than the one observed in molars. Therefore, the opposition of discontinuous and continuous Msx1 expression in the molars and incisors may not be simply related to a limited and continuous eruption, respectively. Divergent homeogenes pertaining to Msx family, Dlx, and also Hox homeogenes determine the positional identity of specific skeletal units during the initial patterning. For instance, Dlx1/Dlx2 double null mutants do not form maxillary molars (Thomas et al., 2000). Incisor and molar development is impaired in Msx1-/- mice (Houzelstein et al., 1997; Satokata and Maas, 1994). Such a site-specific role for homeogenes is suggested to also occur in bone modeling and homeostasis by the present developmental pattern of Msx1 homeogene in the postnatal stages.

The specific function of Msx1 in bone modeling may be discussed, based on previous in vivo and in vitro functional studies in other cell systems and developmental stages. Msx1 may determine local pools of bone cells in the osteoprogenitor compartment by inhibition of their terminal differentiation and enhancement of cell proliferation. Indeed, forced expression of Msx1 in myoblast inhibits the expression of Myo-D muscle master gene and enhances their proliferative activity (Song et al., 1992; Woloshin et al., 1995). Such a role in bone cells is further supported by the observed inhibition of the expression of a characteristic marker of terminal osteoblast differentiation, the osteocalcin, by Msx1 overexpression (Hoffman et al., 1994; Towler et al., 1994).

The observed distinct patterns of Msx1 expression in the maxilla and mandible might reflect biomechanical factors applied by mastication onto the alveolar bone in relation with the different shape of these two bones and their spatial relationships with the skull, face, and their muscle complexes. Indeed, biomechanical forces have been proposed to control cell activity and, therefore, local bone shape (Erlebacher et al., 1995; Hall, 1971). For instance, the requirement for mechanical tension in the formation of bone outgrowth was established in mice lacking both Myf-5 and Myo-D genes (Rudnicki et al., 1993). These mice lack long tubercles presumably as a result of impaired muscle development and, therefore, reduced mechanical tension at tendon insertion sites. During mastication, an equilibrated occlusal function is required for a balanced apposition and resorption of the alveolar bone (Moss, 1997). Indeed, the specific steps of activation and deactivation of bone cell promoters associated with the trio of possible responses to mechanical loading (deposition, resorption, and maintenance of bone tissues) are further examples of epigenetic mechanisms that control cell activity. Recent studies on transcription indicate (for review, see Bidwell et al., 1998) that promoter geometry and activity may be modulated by architectural nuclear proteins, which induce bends and twists in the DNA. Msx1 transcription factor may also be involved in these mechanotransductions.

Development and Growth of the Axial and Appendicular Skeleton

In the axial and appendicular skeleton, the present data show that specific Msx1 distribution in chondroclasts and osteoclasts was related to growth plate cartilages during active endochondral bone formation and growth. The same data were observed in the chondrocranium (data not shown). When all growth plates were turned over in bone, long bone growth ended and no more Msx1-positive chondroclasts and osteoclasts were noted in the epiphyses. On the other hand, a continuous site-specific expression was evidenced in the distal part of limbs, which involved the periosteum and articular cartilage. This temporospatial continuity involves the sites where Msx1 shows a developmental gradient in early prenatal stages, i.e., limb buds, craniofacial sutures, mandible, and alveolar bone (beginning at Theiler stage E10.5; Hill et al., 1989; Houzelstein et al., 1997; Robert et al., 1989; data not shown). Furthermore, one common characteristic of these sites is that Msx1 plays a morphogenetic role during their early patterning (Mackenzie et al., 1991a,b; Robert et al., 1989; Satokata and Maas, 1994; Vainio et al., 1993).

As evidenced in the sagittal suture (Kim et al., 1998), tooth germ (Chen et al., 1996; Vainio et al., 1993), and mandible (for review, see Francis-West et al., 1998), Msx1 expression in the mesenchyme of the limb bud is dependant on epitheliomesenchymal interactions (for review, see Tickle, 2000). FGF may act by inducing Shh that then, together with FGF, activates Msx1 expression in the progress zone (Cohn and Tickle, 1996; Hogan, 1996). From all these studies related to the development of structures resulting from epithelial-mesenchymal interactions where limb bud was taken as an example, it seems that Msx1 is expressed in undifferentiated progress zone cells even if it is not clear how Msx1 expression is related to progress zone function. Therefore, the involvement of Msx1 in the set up of a growth process may extend to tissue growth during adult life, namely 1.) normal growth of intramembranous bones at osteogenic fronts in the suture tissue, 2.) longitudinal bone growth in growth plate cartilage at the epiphyses, and 3.) radial bone growth in periosteal osteoblastic cells. In comparison to previous reports, it was unusual to find Msx1 expression associated to nondividing and terminally differentiated cells such as osteocytes and osteoclasts during embryonic and postnatal period when structures are still in a growing state. It is interesting to note that one cell expressing Msx1 in growth plate was the chondroclast/osteoclast, as the initiation of longitudinal bone growth goes through the turnover of the cartilage into bone; therefore, the first step of this growth is the resorption of this tissue. In addition, regeneration studies have proposed that limb regeneration might depend on expression of Msx1. The maintained expression in distal structures may be related to regeneration processes where distinct Msx1-positive osteocytes dedifferentiate (Kostakopoulou et al., 1996). These data show for the first time that Msx1 is expressed in the progenitor, differentiating and/or differentiated cells of the osteoblastic, chondroblastic, osteoclastic, and chondroclastic lineages, which are important for bone growth and homeostasis.

Recent developments in the genetic and developmental biology of skeletal morphogenesis demonstrate that genes critical for development are jointly expressed in discrete embryonic signalling or growth centers, the enamel knot in teeth, the cranial sutures in skull morphogenesis, and the progress zone in the limb buds (Ferguson, 2000). The present study on Msx1 suggests that these signalling pathways may be jointly important throughout the entire lifetime, with an exquisite site-specificity spatially related to the initial pattern of early development. Nowadays, different bone cell lineages have been reasonably characterized regarding their specific phenotypes (for review, see Ducy and Karsenty, 1998). All the osteoblasts/cytes are described to share a unique phenotype, even when they originate from divergent developmental cascades. This phenomenon is called, the ontogenic convergency (Noden, 1988). However, distinct modulations of osteoblast behaviour and gene expression, depending on their location inside the skeleton, is emerging. The obtained data on Msx1 contributes to the identification of molecular determinants of bone cell biodiversity related to their anatomic distribution, strongly suggested here to be dependant on genetic control and initial patterning.

EXPERIMENTAL PROCEDURES

Generation and Genotyping of Embryos Bearing a nlacZ Reporter Gene in the Msx1 Locus

Construction of the transgenic mice carrying the bacterial nlacZ gene driven by the Msx1 promoter has been described previously (Houzelstein et al., 1997). Briefly, a lacZ gene, which contains a nuclear localization signal (n), was introduced into the second exon of the mouse Msx1 gene by homologous recombination, such that it interrupts the third helix of the homeodomain. This method provides an Msx1-β-galactosidase fusion protein to study Msx1 gene expression in heterozygous Msx1+/- mice, and an Msx1 null mutation related to protein functional invalidation in homozygous Msx1-/- mice.

The Msx1 mutant allele was kept on a C57BL/6J background. Heterozygous mice were produced by mating normal C57BL/6 mice (C. River, France) with heterozygous Msx1/nLacZ transgenic mice. Mice from 14 days old postnatally (P14) until 15 months of age (15M) were killed by cervical dislocation. The different bones were dissected out. Of a total of 247 mice, 137 were wild-type and 110 were Msx1 +/-. The transgenic animals were identified for the Msx1-nlacZ allele by polymerase chain reaction on tail DNA by using two sets of primers, as previously described (Houzelstein et al., 1997).

Whole-Mount β-Galactosidase Staining

For analysis, the mice were immediately fixed by immersion in 4% freshly prepared paraformaldehyde (Sigma, France) in 0.1 M phosphate buffered saline (PBS), pH 7.4, for 30 min at 4°C, then rinsed in PBS. The craniofacial complex, mandible, upper limb, and trunk were excised and dissected, and then fixed in the same way. The tails were treated separately for polymerase chain reaction genotyping as previously described (Houzelstein et al., 1997). After fixation, specimens were washed three times in PBS. Detection of β-galactosidase activity was performed with the chromogenic substrate containing 4 mM potassium ferricyanide, 4 mM potassium ferrocyanide, 2 mM MgCl2, 400 μg/ml X-gal and 0.02% Nonidet P-40 in PBS. Samples were immersed overnight in the dark at 37°C, and washed three times in PBS for 20 min each at room temperature. The samples were then post-fixed with the PFA fixative solution overnight at 4°C.

The presently used parameters were established in preliminary experiments on paired Msx1 +/− and Msx1 +/+ 14-day-old mice by comparing the labeling intensity and distribution in the autopodium, craniofacial sutures, and mandible. The duration of primary fixation (5, 15, 30, and 1 hr), histoenzymatic labeling (6, 16, 24, and 40 hr) and the type of decalcifying agent (disodium-ethylene-diamine-tetraacetic acid at pH 7.3, acetic and formic acids) were tested. Overfixation (>30 min) induced false negative data in Msx1 +/− mice. Overexposure for β-galactosidase labeling (>24 hr) induced a superficial uniform background associated with skin in both Msx1 +/− and Msx1 +/+ mice, a pattern that distinctly diverged from the presented data. EDTA seemed to optimally preserve the β-galactosidase staining. Paired Msx1 +/− and Msx1 +/+ mice were systematically used for each studied stage (data not shown). Control Msx1 +/+ mice did not shown any labeling (data not shown). Furthermore, to check the possibility for false-negative data secondary to penetration difficulties related to inadequate permeabilization before whole-mount labeling, sections of Msx1 +/− and Msx1 +/+ mice were labelled a second time for β-galactosidase activity. No difference in the staining was observed before and after this second labeling of sections (data not shown).

Animal and Tissue Processing for Histologic Analysis

After fixation and rinsing in PBS, tissues were demineralized with 10% disodium ethylenediamine-tetraacetic acid (EDTA, Sigma) in 0.2% paraformaldehyde, at pH 7.5, and 4°C (with daily changes of the solution) for 60 days or less depending on the extent of biomineralization. After extensive washing in PBS, the tissues were dehydrated in ascending series of ethanol, treated with toluene and paraffin overnight at 56°C, and finally embedded in paraffin (Paraplast Plus, Sigma). Serial sagittal sections of the craniofacial, axial, and appendicular skeleton were realized by using a microtome RM 2145 (Leica, France). Tissue sections (8–10 μm thick) were mounted on glass slides, precoated with a 0.1% w/v solution of poly-L-lysine (Sigma) in water, and dried overnight at 37°C. Immediately before use, paraffin sections were deparaffinized in toluene, rehydrated in a descending series of ethanol, and finally rinsed in water.

Histologic Analysis

The 8- to 10-μm sections were stained with Van Gieson method (Bancroft et al., 1994). After being deparaffinized, sections were stained with Groat's haematoxylin, rinsed with water, and stained in picrofuchsine 1%. The slides were then dehydrated in a graded series of ethanol solutions, immersed in toluene, and mounted with DePex (Gürr, Germany) before observation by light microscopy. Tissues were photographed on a DMRB (Leica, France) microscope.

Immunohistochemical Analysis

Polyclonal antibodies raised to recombinant rat TRAP enzyme were used (Ek-Rylander et al., 1997). These antibody generously provided by Göran Andersson (Karolinska Institutet, Huddinge, Sweden) recognize specifically the 39-kDa TRAP enzyme purified from bone. The sections were dewaxed and rehydrated as previously described for histology, then subjected to immunostaining procedures. To inhibit endogenous peroxidase activity, the sections were treated with 3% H2O2 vol/vol in 100 mM Tris-HCl, pH 7.6, for 30 min at room temperature and rinsed three times for 9 min in the same buffer containing 1% bovine serum albumin (BSA). The nonspecific binding sites were blocked for 30 min with normal goat serum (Nordic, The Netherlands) diluted at 1/10. The sections were incubated in a moisture chamber overnight at 4°C with the primary antibodies by using their optimal dilution (1/50 for TRAP). After incubation with primary antibodies, sections were rinsed in the same buffer, incubated for 30 min at room temperature with secondary biotinylated anti-rabbit IgG diluted at 1/800, and rinsed. They were finally incubated for 30 min with horseradish peroxidase-streptavidin diluted at 1/300. The peroxidase distribution was visualized by incubating the sections for 4–10 min, at room temperature, with 3,3′-diaminobenzidine tetrahydrochloride (DAB, Sigma) in 100 mM Tris, pH 7.6, 30% H2O2. The sections were finally rinsed with water, counterstained with haematoxylin, dehydrated, and mounted in DePex (Gürr, Germany). They were observed and photographed with a Zeiss Orthoplan microscope. In control sections, the primary antibodies were replaced by nonrelevant rabbit antibodies at 1/50 dilution.

Acknowledgements

Silvana Maria Orestes-Cardoso was supported by Coordenadoria de Aperfeiçoamento de Pessoal de Nível Superior (CAPES–Brezil).

Ancillary