Impaired meningeal development in association with apical expansion of calvarial bone osteogenesis in the Foxc1 mutant

Abstract

Loss of function of the mouse forkhead/winged helix transcription factor Foxc1 induces congenital hydrocephalus and impaired skull bone development due to failure of apical expansion of the bone. In this study we investigated meningeal development in the congenital hydrocephalus (ch) mouse with spontaneous loss of function mutant of Foxc1, around the period of initiation of skull bone apical expansion. In situ hybridization of Runx2 revealed active apical expansion of the frontal bone begins between embryonic day 13.5 and embryonic day 14.5 in the wild type, whereas expansion was inhibited in the mutant. Ultrastructural analysis revealed that three layers of the meninges begin to develop at E13.5 in the basolateral site of the head and subsequently progress to the apex in wild type. In ch homozygotes, although three layers were recognized at first at the basolateral site, cell morphology and structure of the layers became abnormal except for the pia mater, and arachnoidal and dural cells never differentiated in the apex. We identified meningeal markers for each layer and found that their expression was down-regulated in the mutant arachnoid and dura maters. These results suggest that there is a close association between meningeal development and the apical growth of the skull bones.

Introduction

The meninges consist of three layers of connective tissue and enclose the brain and spinal cord. The pia mater of the innermost layer is a thin vascular membrane that adheres to the brain and spinal cord. The arachnoid mater is composed of fibrous tissue forming the space for cerebrospinal fluid in the subarachnoid space. The outermost layer is the dura mater, a strong and thick membrane, which is made up of dense fibrous tissue adjacent to the calvarium, and functions as the inner periosteum. The meningeal mesenchyme, derived from mesenchymal cells interposed between the developing neuroepithelium and the surface ectoderm immediately after neural tube closure, initially forms a cellular network in which no specific features of the future meningeal layers are recognizable; subsequently, meningeal layers develop (McLone & Bondareff, 1975; Angelov & Vasilev, 1989).

The calvarium (skull vault) is composed of paired frontal and parietal bones, and an interparietal bone, all formed by intramembranous ossification, which starts with mesenchymal cell condensation, during which osteogenic differentiation takes place. Mesenchymal condensations of the frontal and parietal bones first localize at the basolateral aspect of the head above eye level (Ishii et al. 2003), and extend apically as sheet-like structures between the brain and surface ectoderm towards the top of the head (Iseki et al. 1997; Rice et al. 2000).

The close relationship between the meninges and the overlying calvarial bones in the postnatal stage has already been indicated. It is suggested that the postnatal immature dura mater provides osteogenic growth factors, osteogenic cytokines and extracellular matrix molecules, which are critical for ossification and morphogenesis of the calvaria (Opperman et al. 1995; Greenwald et al. 2000a,b; Spector et al. 2002b). For instance, cell proliferation and cellular maturation of osteoblasts are induced in co-culture with dural cells (Greenwald et al. 2000a; Spector et al. 2002a). Formation of calvarial bones takes place in close contact with the developing meninges during development. We may therefore speculate that tissue interaction with the meninges is important for calvarial development. Transforming growth factor beta type II receptor (TgfbIIr) gene is expressed in the meninges (Lawler et al. 1994), and deletion of this gene in cranial neural crest cells results in the restriction of mineralized bone to the basolateral site in which the bone primordium first forms (Ito et al. 2003).

Mouse Foxc1 gene, one of the forkhead/winged helix transcription factors, is involved in calvarial bone osteogenesis (Kume et al. 1998; Rice et al. 2003). Foxc1 is mainly expressed in the stretching calvarial mesenchyme covering the brain at E13.5 and later in both the periosteum and mature osteoblasts at E15.5 (Rice et al. 2003). Deletion of the Foxc1 gene resulted in the failure of skull development in the apical area, including abnormal histology of the meninges (Kume et al. 1998), which suggests that there is tissue interaction between meninges and overlying skull bones during development. There is a spontaneous mutant mouse strain, congenital hydrocephalus (ch), in which a non-sense mutation in the winged-helix domain of the Foxc1 gene results in a stop codon to generate a truncated protein lacking DNA binding protein (Kume et al. 1998). Apical progression of the bone formation is impaired in this mutant mouse (Rice et al. 2003).

We performed this study to gain insights into the relationship between the apical extension of the developing skull bone and meningeal differentiation, by focusing on the frontal bone.

Materials and methods

Animal preparation

The Foxc1^ch (congenital hydrocephalus, ch) heterozygous mice were maintained on the C57BL/6 background and the Foxc1^ch homozygous embryos were obtained by mating heterozygous Foxc1^ch mice. The date when the vaginal plug was found was designated embryonic day (E) 0.5. Genotyping was performed as described previously (Kume et al. 2000). The tissue preparations varied according to the experimental procedures. All animal experiments were performed in accordance with protocols certified by the Institutional Animal Care and Use Committee of University of Helsinki and Tokyo Medical and Dental University.

Transmission electron microscopy

Embryonic heads were fixed in 2.5% glutaraldehyde in 0.1 m phosphate buffer (PB) on ice for 2 h. The embryos were washed with 0.1 m PB, post-fixed in 1% OsO₄ buffered with 0.1 m PB for 2 h, dehydrated in a graded series of ethanol and embedded in Epon 812. Semi-thin sections were cut at 1 µm and stained with toluidine blue. Ultrathin sections, 90 nm, were collected on copper grids, double-stained with uranyl acetate and lead citrate, and then observed using transmission electron microscopy (H-7100, Hitachi, Hitachinaka, Japan).

Immunohistochemistry

For immunohistochemical analysis, embryos were fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) and embedded in Tissue-Tek OCT compound (Sakura, Japan) after sucrose equilibration. The sections were cut at 12 µm and re-fixed again in 4% PFA in PBS on ice briefly at the beginning of the staining procedures. The following primary antibodies were used: rat anti-mouse CD31 (PECAM-1) monoclonal antibody (BD Pharmingen, USA) and rabbit anti-connexin43 (Cx43) antibody (Sigma-Aldrich, USA). Alexa Fluor 555 goat anti-rat IgG and Alexa Fluor 488 goat anti-rabbit IgG (Invitrogen, USA) were applied for visualization of CD31 and Cx43, respectively. Nuclear staining was carried out with Hoechst33342 (Sigma-Aldrich, USA).

In situ hybridization

The following probes were used: (1) a 640-bp fragment of mouse Runx2; (2) a 700-bp fragment of Foxc1; (3) a 900-bp fragment of BMP7; (4) a 1399-bp fragment of mouse Cx43; (5) a 650-bp of Alx4 fragment; and (6) a 400-bp of Msx2. Anti-sense riboprobes were transcribed and reduced to 250 bases by alkaline hydrolysis. In situ hybridization on frozen tissue sections of E12.5-E14.5 heads was carried out as described by Yoshida et al. (2005).

Results

Abnormal growth of the bone begins between E13.5 and E14.5 in ch mutants

We determined when abnormal calvarial bone growth starts in the mutant by the expression pattern of Runx2, the earliest marker of osteoblast differentiation. At E12.5, expression of Runx2 was observed in the mesenchymal condensation of the frontal bone at the basolateral aspect of the head above eye level (Fig. 1A,D, arrows) as previously reported (D'Souza et al. 1999; Ishii et al. 2003; Aberg et al. 2005) in both wild type and ch mutant mice. At E13.5, the frontal bone primordium was growing, represented by the extended expression domain in both apical and basal directions in both wild type and ch mutant mice (Fig. 1B,E, arrowheads), which was consistent with previous observations by Rice et al. (2003). As previously reported by Grunëberg (1943), the shape of the mutant cerebral hemisphere was already abnormal at this stage (Fig. 1E). The expression domain of Runx2 further expanded, and the apical tip of the domain was placed beyond the basal half of the head in E14.5 wild type (Fig. 1C, arrowhead), whereas the mutant frontal bone domain remained at the initiation site (Fig. 1F, arrow). These observations revealed that active apical growth of the frontal bone begins after E13.5, and is inhibited in the mutant.

Fig. 1.

Apical extension of the frontal bone and light microscopic images of the developing meninges during frontal bone development in wild type and the ch mutant. Coronal sections of E12.5 (A,D,G,J,M,N,Q and R), E13.5 (B,E,H,K,O and S), E14.5 (C,F,I,L,P and T). Squares in (C) indicate the areas of apical site and basolateral site. In situ hybridization of Runx2 in wild type (A–C) and mutant (D–F). Toluidine blue-stained semi-thin sections of wild type (G–I, M–P) and mutant (J–L, Q–T) at the basolateral site of the head. (M,N) and (Q,R) are high magnification images of (G) and (J), respectively. (O,P,S,T) are high magnification of (H,I,K,L), respectively. (A–F) Arrows indicate Runx2 expression domains. Runx2 is first expressed at E12.5 in the frontal bone primordium at the basolateral site of the head above the eye level (A, arrow), and then expands towards the apex in wild-type embryos at E13.5 (B) and E14.5 (C). Arrowheads indicate the apical tip of the expressing domain. Difference in Runx2 expression pattern in the mutant is first found at E13.5 (E), and the expression domain is restricted to the initial mesenchymal condensation sites of mutant embryos at E14.5 (F, arrow). (G–I) Between E12.5 and E14.5, mesenchymal cells interposed between the developing brain (b) and frontal bone mesenchymal condensation (mc) form a well organized structure of the meninges. (J–L) In ch mutants the structure of the mesenchyme is similar to that of wild type. However, after E13.5, meningeal mesenchyme is compacted and layers are not distinguished. Arrowheads in (I,M,P,Q,T) indicate blood vessels with and without haematopoietic cells. Dotted lines in (N,O,R,S) show the boundary of mesenchymal condensation of the frontal bone. a, arachnoid mater; d, dura mater; e, eye; f, frontal bone; i, inner part; mc, mesenchymal condensation; o, outer part; p, pia mater;. Scale bars: 200 µm (A–F), 50 µm (G–L) and (M–T).

Meningeal development is impaired in ch mutants

It has been reported that Foxc1 expression in developing meninges starts at E12.5 and is maintained through the developmental period, whereas the expression in osteogenic cells is detected only in mature osteoblasts (Rice et al. 2003). As Foxc1 expression in the meninges begins before appearance of abnormal bone growth in the mutant, we concentrated our study between E12.5 and E14.5 and investigated the meningeal development of the ch mutant histologically.

We first examined semi-thin sections of the basolateral region of the head where the frontal bone primordium first appears. The innermost layer of the meninges showed an increased number of blood vessels during this period in both wild type and mutant, which suggests that development of the pia mater is little affected in the mutant. At E12.5 there is a blood vessel-forming layer of future pia mater just next to the brain (Fig. 1G,M). The mesenchyme comprised a loose fibroblast meshwork structure with round or polygonal-shaped nuclei, and long cytoplasmic processes covering the pia (Fig. 1M), which was further divided into two parts by cell density, the inner part with a lower cell density and an outer part (Fig. 1G). Densely packed polygonal-shaped preosteoblasts were found outside these layers (Fig. 1G,N). Mutants demonstrated the same structure as wild type at this stage (Fig. 1J,R). At E13.5, cell density of the meshwork was slightly higher in both inner and outer parts, and the condensation of preosteoblasts and osteoblasts became dense in the wild type (Fig. 1H,O). In the mutant, the mesenchymal meshwork structure seemed to be squashed, and preosteoblasts and osteoblasts in the bone domain were more compacted compared to the wild type (Fig. 1K,S). In the E14.5 wild type, three layers were distinguished between the brain and the frontal bone domain – the pia, the arachnoid, and the dura maters – and the blood vessels of the pia mater were aligned and continuously covering the brain (Fig. 1I,P). In contrast, mutant mesenchyme was completely compacted, and only the pia mater layer was distinguished by the presence of blood vessels with haematopoietic cells (Fig. 1L,T).

As the semi-thin sections suggested that meningeal development is impaired in the mutant, we next investigated the ultrastructure of the developing meninges by transmission electron microscopy (TEM). In the basolateral site of E12.5 wild type, a meshwork structure was composed of fibroblasts with a polygonal nucleus and cell density was low in the inner part as observed in the semi-thin section (Fig. 2A). Mutant meninges showed almost the same morphology as wild type at this stage (Fig. 2G). By E13.5 the arachnoid mater next to the pia mater had developed two layers, an inner layer with round nuclear cells and an outer layer with oval nuclear cells, and the dura mater layer was forming next to the arachnoid, showing spindle-shaped cells with collagen deposition (Fig. 2B and data not shown). In ch mutants, although the pia mater appeared to develop normally the shape of the arachnoidal cell was oval and the cells firmly attached each other without making a meshwork structure (Fig. 2H). It was impossible to recognize two layers in the arachnoid as observed in the wild type. At this stage, collagen-secreting dura mater was observed (Fig. 2H). At E14.5, gaps between blood vessels in the pia mater were filled with small supporting cells, the arachnoidal cells of the inner layer established a firm meshwork structure and outer layer arachnoidal cells were more elongated and aligned in a longitudinal direction, which appeared to compose the boundary between the arachnoid and the dura (Fig. 2C). A bundle of dural cells was aligned in a longitudinal direction as observed at E13.5 (Fig. 2C). In the mutant, blood vessels were organized in the pia mater; however, the arachnoid layer was kept as it was at E13.5 and dural cells were now compacted (Fig. 2I). Even the nuclei of osteoblasts of the frontal bone were not polygonal in shape and osteoblasts were packed together by this stage.

Fig. 2.

Ultrastructure of the developing meninges. Transmission electron microscopic (TEM) images of the developing meninges at E12.5 (A,D,G,J), E13.5 (B,E,H,K) and E14.5 (C,F,I,L). Images of wild type (A–F) and the ch mutant (G–L). The brain is to the right. Images of the basolateral site (A–C and G–I) and the apical site (D–F and J–L). Dotted lines indicate boundaries of morphologically different layers. (A) At E12.5, blood vessels (asterisk) are developing adjacent to the developing brain and the meningeal precursors next to the vessel-forming area show typical fibroblast characteristics. (B) Three distinct meningeal layers of the blood vessel-containing pia mater (p), the arachnoid layer (a) containing spherical cells with long cytoplasm protrusions in the inner part (arrowheads) and elongated cells (arrows) in the outer part, and longitudinally arranged fibroblastic dural layer (d) are observed at the basolateral site. (C) By E14.5 each layer is more organized and clearly distinguishable. (D) Only a layer of mesenchymal cells different from the subdermal connective tissue mesenchymal cells is observed next to the brain (b). (E) Inner (arrowheads) and outer (arrow) arachnoid layers are recognizable, whereas the dura layer is not yet detected. (F) The dura mater layer is distinguished by this stage but the cell shape seems to be more fibroblastic. (G) Construction is largely similar to that of wild type. (H) Although three layers are distinguished, they are compacted and do not show reticulated structure as found in wild type. (I) Closely packed meningeal layers. (J) The meningeal mesenchyme can be distinguished from subdermal connective tissue mesenchymal cells. (K) Blood vessels are forming (p) adjacent to the brain (b), however, other mesenchymal layers are not present. (L) Only blood vessel-forming pia mater layer is recognizable. a, arachnoid mater; d, dura mater; f, frontal bone; i, inner part meninges; o, outer part meninges; p, pia mater; asterisks are blood vessels with or without hematopoietic cells. Bars, 10 µm.

In the apical site, the same development of layered structure as observed in the basolateral site was found in the wild type. Although we did not recognize the two parts of the meshwork structure of mesenchyme at E12.5, cells with polygonal-shaped nuclei and blood vessel formation were found just outside of the brain (Fig. 2D). At E13.5 (Fig. 2E) and E14.5 (Fig. 2F), an inner layer with round nuclear cells and an outer layer with elongated nuclear cells were recognizable in the arachnoid, and collagen-secreting dural cells were differentiating. In E12.5 ch mutants (Fig. 2J), we found blood vessels and elongated fibroblasts outside the brain. At E13.5 (Fig. 2K) and E14.5 (Fig. 2L), only the vessel-forming pia mater was present and arachnoidal cells with round or oval nuclei were not seen. There were mesenchymal fibroblasts of subdermal connective tissue next to this layer. Development of the arachnoid or the dura was never found in ch mutants.

Expression of meningeal markers is down-regulated in the ch mouse

Histological observation showed that abnormal meningeal development in the homozygote begins around E13.5. We investigated the expression pattern of Foxc1 in the basolateral site of wild type in which the meningeal layers are easily defined (Fig. 3). At E12.5, Foxc1 is clearly expressed in the prechondrogenic mesenchyme and endothelial cells, which is consistent with previous observations (Fig. 3A, Rice et al. 2003; Zarbalis et al. 2007). In addition, we found some of the mesenchymal cells of both inner and outer parts which would form the meninges in future expressing Foxc1 (Fig. 3A). At E13.5, strong expression of Foxc1 was present in the mesenchymal layer outside the blood vessel-forming layer of the pia, suggesting that Foxc1 is strongly expressed in the arachnoid layer, and the dura present between mesenchymal condensation and the arachnoid hardly expresses Foxc1 (Fig. 3B). At E14.5, Foxc1 is expressed in the arachnoid layer whereas the dura mater did not seem to express Foxc1 (Fig. 3C). Strong expression of Foxc1 was not seen in the osteoblastic cells (Fig. 3C).

Fig. 3.

Expression of Foxc1 in the basolateral site. Foxc1 expression at E12.5 (A), E13.5 (B) and E14.5 (C) in wild type. Arrows represent borders between the brain and the meninges. Dotted lines indicate the area of mesenchymal condensation (mc) or frontal bone primordium (f). (A) Foxc1 is strongly expressed in the prechondrogenic mesenchyme (pc) that will eventually form the skull base. Blood vessels (arrowhead) and some of the mesenchymal cells in both the inner and outer parts (double arrows) also express Foxc1. (B) Foxc1 is transcribed in some of the arachnoidal cells (double arrows); however, the expression level in the dura is low. (C) Arachnoidal cells express Foxc1. b, brain. Scale bar: 50 µm.

Previous reports showed that some signaling molecules and transcription factors are involved in calvarial formation; among them, Bmp4, Bmp7, Msx2 and Alx4 are expected to function in apical growth of the developing skull bones (Kim et al. 1998; Rice et al. 2000; Satokata et al. 2000). We compared expression of these genes in the meninges between wild type and the ch mutant at E13.5 (Fig. 4) and E14.5 (Fig. 5) when the meningeal layers were recognized.

Fig. 4.

Expression of meningeal markers in E13.5 ch mice. Corresponding merged images of immunohistochemical staining (A–D) of CD31 (red) and Cx43 (green) with Hoechst stained nuclei (blue). In situ hybridization of Cx43 (E–H), Bmp7 (I–L), Alx4 (M–P) and Msx2 (Q–T). Arrows show borders between the brain and the meninges. Positive staining is indicated by arrowheads. Dotted lines demarcate the frontal bone domain defined by Runx2 expression. (A–D) CD31 staining in developing blood vessels is found in the pia mater (arrowheads, red) in both basolateral and apical sites of wild type and mutant. Cx43 is dispersed in the outer layer of CD31 positive layer (green). In the mutant, expression level of Cx43 is decreased (C,D). (E–H) Cx43 mRNA expression (arrowheads) is restricted to the arachnoid mater, which is consistent with protein localization. In situ hybridization of Cx43 shows slight down-regulation in the ch mutant (G,H, arrowheads). Proteins and transcripts of Cx43 are present in the surface ectoderm of the mutant (D,H). Expression of Bmp7 is found in the arachnoid layer of the basolateral site in wild type, whereas the expression is dramatically decreased in the ch mutant (I,K, arrowheads). At the apical site, Bmp7 transcripts are not found in the meninges of either wild type or mutant (J,L). (M–P) Alx4 is broadly expressed in the meninges and in a certain population of osteoblasts of the frontal bone in wild type (M,N, arrowheads). In the ch, Alx4 expression is found in the meninges of the basolateral site (O, arrowheads), but not in the apical site (P, asterisk). (Q–T) Expression of Msx2 is present in the dura mater of developing meninges (Q,R, arrowheads) of wild type, whereas it is detected only at the basolateral site (S, arrowheads) but not at the apical site (T, asterisk) of the mutant meninges. b, brain; ec, ectoderm; f, frontal bone. Bar, 50 µm.

Fig. 5.

Expression of meningeal markers in E14.5 ch mice. The corresponding merged images of immunohistochemical staining (A–D) of CD31 (red) and Cx43 (green) with Hoechst stained nuclei (blue). In situ hybridization of Cx43 (E–H), Bmp7 (I–L), Alx4 (M–P) and Msx2 (Q–T). Arrows represent borders between the brain and the meninges. Positive staining is indicated by arrowheads. Dotted lines demarcate the frontal bone domain defined by Runx2 expression. (A–D) CD31 staining in developing blood vessels is found in the pia mater (arrowheads, red) in both basolateral and apical sites of wild type and mutant. Cx43 is localized in the outer layer of CD31 positive layer (green). In the mutant high levels of Cx43 expression are present in the surface ectoderm (ec). (E–H) Cx43 mRNA expression (arrowheads) is restricted to the arachnoid mater, which is consistent with protein localization. In situ hybridization of Cx43 shows slight down-regulation in the ch mutant (G,H, arrowheads). The transcription level in the surface ectoderm does not differ greatly between wild type and mutant. (I–L) Expression of Bmp7 is present in osteoblasts of the frontal bone primordia and the arachnoid layer of the meninges in wild type (I,J, arrowheads), whereas the expression is dramatically decreased in both the basolateral site (K, arrowheads) and apical site (L, asterisk) of the mutant. (M–P) Alx4 is broadly expressed in the meninges and a certain population of osteoblasts of the frontal bone in wild type (M,N, arrowheads). In the ch mouse, Alx4 expression is found in the meninges of the basolateral site (O, arrowheads), but it is not detected at the apical site (P, asterisk). (Q–T) Expression of Msx2 is present in the dura mater of developing meninges (Q,R, arrowheads) of the wild type, whereas it is detected only at the basolateral site (S, arrowheads) but not at the apical site (T, asterisk) of the mutant meninges. b, brain; ec, ectoderm; f, frontal bone. Bar, 50 µm.

We investigated the localization of CD31 (also known as PECAM1), an endothelial cell marker (Baldwin et al. 1994), which represents blood vessel formation in the pia mater. At E13.5, distinct staining of CD31 was seen just next to the brain, which can be recognized as blood vessels by the staining pattern in the wild type (Fig. 4A,B, red). At E14.5, CD31-positive cells almost continuously outlined the brain in both basolateral and apical sites (Fig. 5A,B, red). This CD31 localization pattern was also observed in the mutant (Figs 4C,D, 5C,D, red).

Expression of Connexin43 (Cx43), a gap junction-forming protein that is prevalent during development and shows strong expression in the meninges (Yancey et al. 1992), was examined. Superimposition of the images of Cx43 and CD31 immunohistochemical staining showed that dotted staining of Cx43 constitutes a strip which covers CD31 positive layer (Fig. 4A, green), which suggested it is strongly expressed in the arachnoid layer, and the staining pattern was sparse in the apical site (Fig. 4B, green) in E13.5 wild type. By E14.5, the Cx43 expression strip was clear in the apical site as well as in the basolateral site (Fig. 5A,B, green). In ch mutants, localization of Cx43 was consistent with that of wild type (Figs 4C,D, 5C,D); however, there were fewer Cx43 staining dots in arachnoidal mesenchymal cells compared to wild type, especially in the apical site. At E13.5 there were few Cx43 positive cells in the apical site (Fig. 4D). By E14.5 some positive cells were visible (Fig. 5D). Interestingly, Cx43 was strongly localized in the surface ectoderm of the apical site in ch mutants at E14.5 (compare Fig. 5D with 5B). In situ hybridization of Cx43 confirmed the results of immunohistochemistry (Figs 4E–H, 5E–H).

Faint expression of Bmp7 was seen in the developing arachnoid area of the basolateral site but was hardly detected in the apical site at E13.5 (Fig. 4I,J). Bmp7 expression became intense in the arachnoid and in some of the pia mater cells in the basolateral site (Fig. 5I), and it was present in the apical site although it was not clear whether the expression is present only in the arachnoid mater or in both the pia and the arachnoid maters (Fig. 5J) in E14.5 wild type. In the mutant, Bmp7 expression was not detected except at the basolateral site, where some Bmp7 expressing cells were observed (Figs 4K,L and 5K,L).

At E13.5, Alx4, a homeodomain-containing transcription factor, was transcribed in mesenchymal cells located between the brain and the surface ectoderm, except for those underneath the surface ectoderm that would eventually constitute subdermal connective tissue (Fig. 4M,N). By E14.5 the expression was seen in all layers of the meninges and in some osteoblastic cells (Fig. 5M), and it was weaker in the apex (Fig. 5N). In mutants, the expression pattern was the same as wild type (Fig. 4O) in the basolateral site, whereas we did not detect the significant level of Alx4 expression in the apical site (Fig. 4P) at E13.5. This expression pattern was maintained till E14.5 (Fig. 5O,P). Compared to Alx4, Msx2 was expressed in a more restricted area. It was expressed outside the Bmp7 expressing arachnoid layer and within the developing bone domain, the dura mater, in both basolateral and apical sites at E13.5 and E14.5 (Figs 4Q,R, 5Q,R). In the mutant, Msx2 expression was present in the dura mater layer of the basolateral site but almost no expression was found in the apical site at E13.5 (Fig. 4S,T) and E14.5 (Fig. 5S,T).

Discussion

Impaired growth of calvarial bones and the defective meningeal development are associated in ch mutants

In situ hybridization of Runx2 (Fig. 1) and histological observations (Figs 1, 2) indicate that initiation of apical growth of calvarial bones and meningeal differentiation begin almost concomitantly at the basal region and progress to the apex in normal development. In the ch mutant, although skull bone osteogenesis starts as it does in wild type, active apical expansion of the frontal bone domain fails, and establishment of meningeal layers is inhibited. The cause of the skull bone defects in ch mutants has been a matter of controversy and two possibilities have been proposed – as a mechanical consequence of hydrocephalus and/or as a result of defective meningeal formation (Grunëberg 1943; Grunëberg & Wickramaratne 1974). Using skull organ culture, Rice et al. (2003) suggested that hydrocephalus is not directly involved in skull bone defects in the mutant. Supporting this report, a mouse mutant line, Foxc1^hith/hith, in which Foxc1 protein is destabilized, shows a milder phenotype than Foxc1 null or ch mutants; the calvarial bones almost grow normally except for the apex area, at which defects of osteogenesis and meninges are present, and hydrocephalus develops only postnatally (Zarbalis et al. 2007). Kume et al. (1998) suggested that Foxc1 is involved in chondrogenesis and meningeal development. Combining these reports with our Foxc1 expression study it is suggested that meningeal development begins but further establishment of mature layered meninges is impaired due to lack of Foxc1 expression in the meninges, which subsequently affects overlying developing skull bone primordium in the ch mutant.

Previous studies have already suggested an association between skull bones and the underlying dura mater, mainly after the stage of apical growth of the skull bone had been completed. The dura mater influences the cranial (coronal) suture obliteration in both fetal and neonatal rats (Opperman et al. 1995) and provides overlying calvarial bones with growth factors and extracellular matrix molecules (Opperman et al. 1995; Greenwald et al. 2000a,b; Spector et al. 2002a,b). The dura mater also plays a role in reossification of the calvarial defects (Greenwald et al. 2000a,b; Gosain et al. 2003).

The direct effects of impaired meningeal differentiation on overlying calvarial bone formation in the mutant have yet to be clarified. There is less proliferative activity of preosteoblasts at the apical tip of the growing skull bone of the ch mutant (Rice et al. 2003). We analysed the immunohistochemistry of caspase-3 to detect cell death, and we hardly observed any caspase-3 positive cells in either wild type or mutant (data not shown). Another possibility is that the meningeal layer is required for induction of preosteoblast differentiation. However, we recently found that the calvarial bone extends apically by intrinsic growth without detectable recruitment of adjacent mesenchymal cells (manuscript in preparation). This process could involve cell proliferation and migration. Based on these results, we suggest that abnormal meninges of ch mutants have an influence on preosteoblast proliferation and/or preosteoblast migration. Further studies are required to clarify whether preosteoblasts within the bone primordium at the basolateral site migrate along the dura mater to the apex.

Loss of Foxc1 gene effect on meningeal development

Bmp7 expression was mostly abolished in the arachnoid layer of the mutant. Bmps, members of the transforming growth factor beta (TGF-β) family, are potent stimulators of osteogenesis and are involved in other cellular functions (reviewed by Chen et al. 2004).

Compacted arachnoid in the mutant suggests excess cell adhesion or alteration of cytoskeleton patterning. As recent findings suggest involvement of Bmps in the control of cell adhesion (Wallingford & Harland, 2007), changes in the expression of cell adhesion molecules and cytoskeleton patterning will be investigated.

It was shown that Bmps mediate the Foxc1-dependent activation of Alx4 and Msx2 in the head mesenchyme in mice (Rice et al. 2003). The loss of Alx4 and Msx2 expression in the apical site of the ch mutant supports this observation and it is possible to speculate that induction of Alx4 and Msx2 transcription in the basolateral site is not completely dependent on the presence of Foxc1. Alx-4 null mutants have abnormalities in skeleton, including a decrease in the size of the skull bones, but the phenotype is milder than in the ch mutant (Qu et al. 1997; Antonopoulou et al. 2004). A highly conserved homeobox gene Msx2, transcribing in various structures, plays a part in mediating the tissue interaction of tooth, limb and cranium development (MacKenzie et al. 1992; Liu et al. 1995; Rice et al. 2003; Antonopoulou et al. 2004). Loss of function of Msx2 results in calvarial foramina that is a similar phenotype to Alx4 null mutants (Wilkie et al. 2000; Ishii et al. 2003; Antonopoulou et al. 2004). An incremental exacerbation of the calvarial defect in double Alx4/Msx2 mutants is related to the increase in the number of lost Msx2 and Alx4 alleles, and the double homozygous mutant demonstrates the calvarial phenotype close to the ch mutant (Antonopoulou et al. 2004). These results suggest that Alx4 and Msx2 partially play redundant roles in the skull development and study on meningeal development of these mutants could provide some insights into the molecular mechanism of meningeal development and apical expansion of the bone primordium.

In vitro studies demonstrated the pivotal roles of Cx43 in proliferation and differentiation of osteoblasts (Li et al. 1999; Furlan et al. 2001; Gramsch et al. 2001) and targeted disruption of the Cx43 gene in mice affects skeletal shape and mineralization with a greater delay of osteogenesis in intramembranous bones than in endochondral bones (Lecanda et al. 2000). Our in situ hybridization and immunohistochemical analysis revealed the strong expression of Cx43 in the meninges, especially in the arachnoid mater (Figs 4,5) compared to other tissues such as osteoblasts or endothelium that were previously shown to predominantly express Cx43 (Li et al. 1999; Furlan et al. 2001). This was an unexpected result, and although coexpression of multiple connexin family members within the same cell type suggests some functional redundancy (reviewed in Laird, 2006), Cx43 might play an important role in the development of the arachnoid mater. In addition, the mechanism and consequence of strong Cx43 expression in mutant surface ectoderm in the apical site has yet to be elucidated.

Concluding remarks

Our study demonstrated that development of the arachnoid and dura mater was impaired in the ch mutant, which was consistent with down-regulation of some of the meningeal markers. Abnormal development of the meninges and apical extension of the bone development begin concomitantly in the mutant. These results suggest that there is a close association between development of the meninges and skull.