The homeoprotein engrailed 1 has pleiotropic functions in calvarial intramembranous bone formation and remodeling

The mammalian skull vault (roof) comprises an amalgamation of skeletal elements that specifically evolved to encase and protect the brain and sensory organs. Consisting of the frontal, parietal, interparietal and occipital bones(in mouse), the skull vault is developmentally complex and receives lineage contributions from both the cranial neural crest (CNC) and the paraxial mesoderm (Jiang et al., 2002). These populations migrate into defined locations overlying the cerebral hemispheres, and subsequently differentiate into condensing osteogenic or chondrogenic mesenchyme (between E7.5-E11.5 in mouse). The osteogenic mesenchyme is characterized by the expression of Runx2 and Osterix (Osx; Sp7 - Mouse Genome Informatics), the earliest molecular determinants of bone formation(Komori et al., 1997; Nakashima et al., 2002; Otto et al., 1997). In contrast to the endochondral bone formation that ensues from a cartilaginous template, most elements of the skull vault develop by intramembranous ossification, characterized by the direct differentiation of osteogenic mesenchyme into osteoblasts. Osteoblasts are specialized cells that produce bone extracellular matrix or osteoid, and that subsequently regulate its mineralization (Aubin,2002).

During the morphogenetic phases of skull development, a pool of highly proliferative osteoprogenitors populates the margins (osteogenic fronts) of each enlarging bone anlagen, thereby maintaining calvarial bone expansion through osteoblast replenishment (Iseki et al., 1999; Opperman,2000). After each bone has acquired its basic form (E15.5), the individual skeletal elements remain separated by fibrous joints, or sutures,composed of skeletogenic mesenchyme and fibroblasts(Opperman, 2000). Continued production of osteoprogenitors in the sutures ensures that calvarial expansion is coordinated with growth of the underlying brain. Thus, sutures serve as major centers for calvarial osteoblast differentiation and new bone formation postnatally. Finally, in addition to osteoblast-mediated bone formation,calvarial bones undergo dynamic modeling and remodeling of their three-dimensional microarchitecture, through the coordinated resorptive activity of haematopoietically derived osteoclasts(Takahashi, 2002).

Genetic and molecular evidence has implicated a number of growth and transcription factors as being important regulators of skull formation. The fibroblast growth factors (FGFs) are a family of secreted polypeptides that act through four related tyrosine kinase receptors (FGFR1-FGFR4) to regulate a plethora of developmental processes, and they are of central significance to intramembranous ossification (Ornitz and Marie, 2002). Human diseases that manifest the precocious osseous obliteration of sutures, known as craniosynostosis, often result from gain-of-function mutations in FGF receptors 1-3 (FGFR1/2/3)(Webster and Donoghue, 1996; Wilkie, 1997). Mouse models of loss- or gain-of-function mutations in Fgfr1 and Fgfr2 have provided further evidence that FGF signaling regulates the proliferation and differentiation of calvarial osteoblasts and osteoprogenitors(Eswarakumar et al., 2004; Eswarakumar et al., 2002; Yu et al., 2003; Zhou et al., 2000). Elucidating the precise mechanisms of osteoblastic FGF signaling, however, has been complicated by the fact that at least four potential ligands (Fgf2,Fgf4, Fgf9 and Fgf18) for Fgfr1-Fgfr3 are expressed in the developing mouse calvarium (Kim et al., 1998; Ohbayashi et al.,2002; Rice et al.,2000). In addition, although a growing number of intracellular antagonists for receptor tyrosine kinases (RTKs) have been identified as being important modifiers of FGF-responsiveness in many developmental contexts,information on how these may regulate calvarial bone formation is currently lacking (Furthauer et al.,2002; Kawakami et al.,2003; Wakioka et al.,2001).

At the transcriptional level, a number of homeodomain proteins have been shown to participate in regulating calvarial bone development. Perturbed calvarial ossification is observed in humans and mice harboring loss-of-function mutations in the homeoproteins Msx2, Dlx5 and Alx4 (Robledo et al.,2002; Satokata et al.,2000; Wilkie et al.,2000; Wuyts et al.,2000). Interestingly, regulation of the osteocalcin gene promoter by Runx2 has been shown to be influenced by FGF signaling and the modifying activities of Dlx5 and Msx2 (Newberry et al., 1998). Furthermore, FGF signaling has been shown to directly induce Msx2 gene expression in calvarial sutures(Ignelzi et al., 2003). These findings are indicative of crucial interactions between growth factors and transciptional regulators of calvarial osteogenesis, and raise the possibility that other such modifiers remain to be identified.

Engrailed 1 (En1), the homolog of Drosophila en, is a homeodomain-containing transcription factor that participates in the regulation of multiple mammalian developmental processes, such as dorsoventral patterning of the distal limb and mid-hindbrain specification(Loomis et al., 1996; Wurst et al., 1994). Interestingly, several ossification and growth abnormalities observed in the sternae and phalanges of En1-null mice, as well as the observed expression of En1 in developing vertebrae, have implicated its involvement in skeletal development(Davidson et al., 1988; Wurst et al., 1994). We show that En1 is expressed during the early and late phases of calvarial osteogenesis. Through the characterization of novel phenotypic features of En1-ablated mice, we demonstrate that EN1 plays a crucial role in regulating intramembranous ossification during craniofacial bone development. In addition, evidence is provided to suggest that EN1 regulates calvarial ossification by influencing FGF responsiveness in osteoblasts. We further show that EN1 has a non cell-autonomous function in regulating osteoclast recruitment and activation, thereby affecting calvarial bone resorption and remodeling.

En1 knockout mice (Hanks et al., 1995; Matise and Joyner,1997; Wurst et al.,1994) were maintained on a wild-type Swiss Webster background. Mice harboring either the En1^hd (homeobox deleted) or En1^lki (lacZ `knock in') null alleles were intercrossed to generate homozygous mutants lacking En1 function.

Visualization of the cartilaginous and mineralized skeleton was facilitated by Alcian Blue and Alizarin Red staining, as previously described(Loomis et al., 1996). For detection of β-galactosidase activity in tissues, specimens were treated for several hours with a gluteraldehyde fixative, washed in phosphate-buffered saline (PBS) and stained with standard X-gal solution for 16 hours at 4°C(Matise and Joyner, 1997).

Micro-computed tomography (microCT) was performed at the Centre for Bone and Periodontal Research located on McGill campus (Montreal, Quebec). Data were acquired on a SkyScan T-1072 microtomograph (Skyscan, Aartselaar,Belgium). Cross-sections along the specimen axis were reconstructed and images quantified using the Cone-Beam Reconstruction Software supplied by SkyScan.

Dissected newborn skulls were fixed overnight at 4°C in 4%paraformaldehyde (PFA), and decalcified for 4 days in 40 mM EDTA (pH 7.3),before dehydrating and embedding in paraffin. X-gal-stained tissues were dehydrated in isopropanol.

Immunolocalization of phosphorylated extracellular signal-regulated kinase(pERK) was performed on 6-μm deparaffinized sections using a rabbit monoclonal antibody specific to phosphorylated ERK1/2, according to the manufacturer's instructions (20G11, Cell Signaling). Signal detection was performed using the ABC-AP Kit (Vector Laboratories) coupled to the BM-Purple substrate (Roche).

Probes for osteopontin (Opn), osteocalcin (Ocn), bone sialoprotein (Bsp) and osterix (Osx), were amplified by reverse transcriptase (RT)-PCR (iScript cDNA Synthesis Kit, BioRad), employing specific primers to each mRNA and total RNA extracted from primary calvarial osteoblasts as a template. Amplified cDNA fragments were cloned into the pGEMT-Easy cloning vector (Promega). Probes for murine Fgfr1 and Fgfr2 were obtained from Dr G. Morriss-Kay (Oxford). The probe for Fgf18 was obtained from Dr B. Hogan (Duke). The probes for Spry1 and Spry2 were obtained from Dr G. Martin (UCSF). Digoxigenin-UTP labeling of RNA riboprobes was performed with the MEGAscript transcription kit (Ambion). In situ hybridization was performed on paraffin-embedded sections or whole-mount calvariae, essentially as described by Wilkinson (Xu, 1999).

Bromodeoxyuridine (BrdU, Sigma; 100 μg/g body weight) was administered intraperitoneally into pregnant mice at the indicated gestational stages;then, 1.5 hours later, mice were sacrificed and embryos collected and fixed overnight in 4% PFA. Immunodetection of BrdU was performed as previously described (Ishii et al.,2003). A comparative analysis of the osteogenic fronts of En1-null and wild-type littermates was performed, and statistical significance analyzed by ANOVA one-way assessment of variance (Graph Prism).

ALP activity was quantitated as previously described(Deckelbaum et al., 2002). The histological detection of TRAP⁺ osteoclasts in paraffin-embedded sections was performed as previously described(Miao et al., 2004). Whole-mount detection of TRAP⁺ osteoclasts was performed as described (Holt et al., 1994). For the detection of mineralized matrix, cultured cells were formalin fixed,ethanol dehydrated, stained with 2% AgNO₃ under ultraviolet light,and then treated with 5% sodium thiosulfate. Detection of mineralized bone matrix in non-decalcified tissue sections was performed as described previously (Valverde-Franco et al.,2004).

Dissected calvarial bones were collected in Hank's balanced salt solution(HBSS, Gibco). To extract osteoblasts, bone fragments were treated with collagenase type IA (1 mg/ml in Hank's balanced salt solution; Sigma) for 30 minutes at 37°C, and then with EDTA (4 mM in PBS), and released cells were plated at a density of 10⁶ cells/35 mm in complete culture media[cCM: 10% FCS (Gemini), α-MEM (Gibco), 100 U/ml penicillin, 50 μg/ml streptomycin]. At confluence, cCM was supplemented with 100 μg/ml ascorbate and 5 mM β-glycerophosphate to promote osteogenic differentiation and mineralization (Bakker,2003).

Northern blot analysis was performed as previously described using total RNA (20 μg) extracted from primary osteoblasts, separated on a denaturing formaldehyde gel, and blotted onto supported nitrocellulose(Deckelbaum et al., 2002). Quantitative assessment of autoradiograms was performed using ScionImage software.

In addition to its well-described activation in the embryonic mid-hindbrain, we noted En1 expression in the developing murine skull. To address a possible role for En1 in craniofacial development, we first analyzed its expression at various embryonic stages using a mouse line that harbors a functional β-galactosidase reporter cassette inserted into the first exon of the En1 locus(Hanks et al., 1995; Matise and Joyner, 1997). The En1-lacZ allele (En1^lki), from which the expression of functional EN1 protein is abrogated, precisely recapitulates the endogenous En1 expression. Importantly, En1^lki/+heterozygotes display no discernable phenotype, and are used interchangeably with wild-type mice in this study. Furthermore, the En1 homolog En2 is not expressed by any calvarial tissues (data not shown),excluding the possibility that it compensates for En1 function in these tissues.

Calvarial expression of En1, as detected by whole-mount X-gal staining, initiates at embryonic day (E) 11.5 within lateral aspects of the head (Fig. 1A, part a; arrow),overlying the diencephalic-telencephalic border in the forebrain. Between E12.5 and E13.5, En1 expression expands rostrocaudally and anteriorly, encompassing the frontonasal and mandibular prominences(Fig. 1A, parts b,c; arrows). During phases of overt intramembranous ossification [E14.5 to postnatal day(P) 1], En1 is detected in all developing calvarial bones and sutures(Fig. 1A, parts d-f; arrows,asterisk). Coronal sections of E11.5 heads revealed En1 expression within the presumptive calvarial bone mesenchyme, as confirmed by its colocalization with the early osteogenic marker ALP(Fig. 1B, part a; data not shown). By E13.5, this domain had expanded further toward the base and apex of the skull (Fig. 1B, part b). Histological analysis of the neonatal skull showed that En1 is predominantly expressed by ectoperiosteal osteoblasts lining the bone surfaces(Fig. 1B, parts c,d; arrow), as well as by osteoblasts and osteocytes populating the endosteal surfaces and the matrix of the cranial bone trabeculae(Fig. 1B, part d; arrowheads). Curiously, we found En1 expression to differ between the types of calvarial sutures. Within the abutting interfrontal suture, En1expression remains restricted to the sutural mesenchyme and is excluded from the osteoprogenitors populating the bone margins(Fig. 1B, parts c,e). By comparison, the osteoprogenitors outlining the frontal and parietal bones exhibit En1 expression (Fig. 1B, part f). Taken together, our expression analyses suggest potential roles for En1 in regulating both the primordial and later stages of calvarial ossification.

To determine whether En1 function is required for calvarial bone formation, we first compared the skulls of En1^-/- and wild-type mice at P1 by staining with Alcian Blue for cartilage and Alizarin Red for mineralized bone (Fig. 2A). Although displaying no overt patterning defects, the cranial membranous bones of En1 mutants exhibit profound hypoplasia (reduced size). Notably, the expansion of unossified areas result in pronounced frontal foramina, and gaping coronal and lambdoid sutures(Fig. 2A, parts b,c). Although most En1^-/- mice die during early postnatal life,secondary to mid-hindbrain defects, analysis of the calvariae of rare surviving mutants revealed persistent coronal and lambdoid suture widening. Moreover, the interfrontal suture, which is normally obliterated by P35 through osseous fusion of the abutting frontal bones, undergoes incomplete closure in En1^-/- mice(Fig. 2B, parts a,b). In addition, adult En1 mutants display a severe hypoplasia of the frontonasal and maxillary prominences that results in malocclusion and misalignment of the upper and lower incisors(Fig. 2B, parts c,d). Although hypoplasia of the skull base is often the primary cause for malocclusion in other animal models (Chen et al.,2003), we found no differences between wild-type and En1^-/- mice in this regard.

To gain a more precise three-dimensional perspective of the skull, we performed quantitative microCT on P1 and P5 calvariae of wild-type and En1^-/- mice (Fig. 3A). Our analysis confirmed the calvarial ossification defects in En1^-/- mice and indicated that these do not resolve by P5(Fig. 3A, part b). Quantitative morphometric analyses reveal a significant reduction in bone volume (BV)compared with tissue volume (TV) in En1 mutants (BV/TV: wild type,3.460±0.4033, n=8; En1^-/-,2.205±0.3031, n=6; P=0.0301). Although trabecular thickness is unaltered, a significant increase in the distance separating the trabecular spicules was observed in the mutants (mean distance: wild type,0.7007±0.05129, n=8; En1^-/-,0.8362±0.02945, n=6; P=0.0448), which is suggestive of marrow space expansion (Fig. 3B). Histological analysis of non-decalcified calvariae revealed that, by comparison with the wild type, the membranous bones of En1mutants are composed of reduced and under-mineralized osteoid that is heavily perforated by fibrous tissue (Fig. 3C). These observations suggest that the hypoplasia observed in En1^-/- calvariae results from decreased bone matrix deposition (osteopenia) and mineralization.

To evaluate the cellular basis for the defects in calvarial bone mineralization, we evaluated the capacity of En1 mutant osteoblasts to induce mineralization in culture. Osteoblasts released from newborn wild-type and En1^-/- calvariae were cultured for 21 days under conditions that promote extracellular matrix mineralization. We observed that, although wild-type osteoblasts formed extensive mineralized three-dimensional nodules, En1^-/- osteoblasts exhibited a poor capacity to induce mineralization over the culture period(Fig. 3D). Moreover, ALP activity, the osteoblastic expression of which is indispensable for bone mineralization (Fedde et al.,1999; Murshed et al.,2005), was significantly reduced in En1^-/-osteoblasts in comparison to the wild-type population(Fig. 3E). Collectively, these results imply that En1 is required for calvarial osteoblast differentiation and bone mineralization.

We therefore investigated the possibility that the calvarial defects in En1 mutants result from impaired osteoblast differentiation in vivo. The expression of osteopontin (Opn), a secreted phosphoglycoprotein that is normally activated in differentiating calvarial osteoblasts(Iseki et al., 1997), was examined by whole-mount ISH. We first detected calvarial expression of Opn in wild-type mice at E14.5, within the ossification centers of the presumptive parietal bones (Fig. 4A). By comparison, Opn was nearly absent in En1^-/- skulls. By E16.5, Opn expression extended into all calvarial membranous bones of wild-type and En1^-/- skulls but its domain was considerably reduced in the mutants. This deficiency in Opn expression persisted at E18.5 in En1^-/- rudiments, indicating that En1 plays a role in promoting osteoblast differentiation during the prenatal stages of calvarial bone development.

To evaluate whether the delay in Opn initiation in En1mutants arises from impaired commitment and early differentiation of the cranial skeletogenic mesenchyme, we examined the expression of the osteogenic determinant Osx in E13.5 skulls. Interestingly, Osx and En1 expression overlapped in the calvarial skeletogenic mesenchyme of wild-type embryos (Fig. 4B;compare with Fig. 1B, part b). By contrast, virtually no expression of Osx was observed in the calvarial mesenchyme of En1^-/- littermates. As comparable levels of Osx occurred within the mutant and wild-type mandibular mesenchyme (not shown), this result suggests that En1 has a selective role in enhancing the osteogenic potential of the skull vault mesenchyme. Furthermore, Osx expression remained reduced in En1^-/- calvarial osteoblasts both in vivo and in vitro(Fig. 4B,D). These findings suggest that En1 function is specific to and indispensable for inducing early phases of calvarial osteogenesis, possibly through the activation of Osx.

To gain a more detailed perspective on calvarial osteogenesis, we examined the coronal suture morphology and the expression of osteoblastic genes on sagittal sections of wild-type and En1^-/- skulls at P1. We made three unanticipated observations. First, the thin layer of mesenchyme,which typically separates the overlapping parietal and frontal bone fronts in the wild-type coronal suture, was considerably thickened in En1mutants (Fig. 4C, arrow). Moreover, the margins of the mutant frontal and parietal bones failed to overlap within the suture proper. Second, although En1 mutants at this developmental stage exhibited normal Opn expression in ectoperiosteal and endosteal osteoblasts, terminally differentiated osteocytes continued to express this gene inappropriately (4C, insets, arrowheads). Third, the calvarial expression of osteocalcin (Ocn), a specific marker of late osteoblast differentiation, was almost abolished in En1 mutants. In corroboration, the expression of Ocn and bone sialoprotein (Bsp), an additional marker of the mature osteoblast, was impaired in cultures of differentiating En1^-/- calvarial osteoblasts(Fig. 4D). These findings suggest that En1 functions beyond early osteogenesis to regulate the expression of the late osteoblastic genes commonly associated with matrix mineralization. In addition, the morphological changes within the suture suggest that EN1 may regulate osteoprogenitor proliferation and/or differentiation.

In addition to impaired osteoblast differentiation, another contributing factor to the observed frontal foramina of En1^-/-calvariae could be a reduction in the proliferative capacity of cells populating the osteogenic fronts of the frontal bones. We therefore examined the proliferation of cells within the rudimentary frontal bones and osteogenic fronts of the interfrontal suture by BrdU labeling wild-type and En1^-/- embryos between E12.5 and E18.5. Although no significant difference in proliferation was observed between E12.5 and E16.5(Fig. 5A,B; data not shown), a significant decrease in BrdU incorporation was observed in the interfrontal sutures of En1^-/- calvariae at E18.5. These results suggest that En1 function is important for osteoprogenitor proliferation following, but not prior to, interfrontal suture formation. The lack of En1 expression in the interfrontal osteoprogenitors, however,implies that En1 affects their proliferation through indirect mechanisms.

We next evaluated BrdU incorporation within the coronal sutures of wild-type and En1^-/- calvariae at E18.5. The wild-type coronal suture characteristically comprised a thin layer of mesenchyme containing relatively few proliferating cells at the osteogenic fronts of the overlapping parietal and frontal bones(Fig. 5C). By contrast, the En1^-/- suture was considerably thickened and hypercellular, and contained increased numbers of BrdU-positive cells. These findings indicate that En1 plays differential roles in regulating osteoprogenitor proliferation within distinct suture types.

In order to assess En1 function within the context of known regulators of cranial osteogenesis, we analyzed components of the fibroblast growth factor (FGF) signaling pathway in wild-type and En1-null calvariae. Like EN1, this pathway impacts on both cellular proliferation and differentiation during calvarial bone formation. We first compared the expression pattern of Fgfr1 and Fgfr2 between wild-type and En1 mutants on sagittal sections of newborn calvariae. In the wild type, Fgfr1 expression was observed in ectoperiosteal osteoblasts lining the calvarial bones and the endosteal osteoblasts of the frontal bone trabeculae, but was excluded from the osteogenic fronts and mesenchyme of the coronal suture (Fig. 6A). By comparison, Fgfr2 was co-expressed with Fgfr1 in wild-type osteoblasts in most locations; however, it was also detected within sutural osteoprogenitors. En1 mutants displayed a similar pattern of receptor expression in bone-lining osteoblasts with one exception: Fgfr1 and,more significantly, Fgfr2 were strongly upregulated within the coronal sutural mesenchyme. Examination of Fgfr1 and Fgfr2in wild-type and En1^-/- cultured calvarial osteoblasts showed that both receptors were upregulated over the course of a 21-day differentiation period (Fig. 6B); however, no differences in relative expression between the two populations were observed by semi-quantitative assessment (data not shown). Sharing phenotypic similarities with En1 mutants, Fgf18^-/- mice display cranial ossification defects as a result of impaired osteoprogenitor proliferation and osteoblast differentiation (Ohbayashi et al.,2002). We found comparable expression of Fgf18 in sutural osteoprogenitors and mature calvarial periosteal osteoblasts between wild-type and En1^-/- mice. These findings suggest that En1is crucial for attenuating Fgfr1 and Fgfr2 expression within the coronal suture mesenchyme, and raises the possibility that En1regulates osteoprogenitor proliferation through FGFR signaling.

In addition, we examined the possibility that the impact of En1 on osteoblast differentiation may be mediated through regulating events downstream of FGFR signaling. Members of the Sprouty(Spry1-Spry4) gene family encode intracellular antagonists of receptor tyrosine kinases (RTKs) and are also transcriptional targets of FGF signaling in vertebrates (Minowada et al.,1999). As Spry1 and Spry2 are commonly expressed in FGF-responsive tissues (Minowada et al., 1999; Zhang et al.,2001), we examined their expression in the calvarial bones of wild-type and En1^-/- mice. Although Spry1 was not detected in wild-type or En1^-/- calvariae (data not shown), Spry2 was expressed by wild-type ectoperiosteal osteoblasts,osteoprogenitors of the coronal suture, and, to a lesser extent, endosteal osteoblasts of the frontal bone trabeculae(Fig. 6C). In contrast to the expression in wild-type calvariae, we found almost no expression of Spry2 in parallel sections of En1 mutants. This finding,together with the FGF ligand/receptor expression data, suggests that EN1 is required for regulating the signal transduction events downstream of FGFR that are necessary for activating Spry2 transcription (see Fig. 6E for proposed model).

The induction of Spry2 expression is mediated in part by the ERK-signaling cascade downstream of FGFR(Ozaki et al., 2001). More importantly, ERK phosphorylation is crucial for promoting FGFR-mediated osteogenesis in calvarial cell and organ cultures(Kim et al., 2003; Xiao et al., 2002). To evaluate the role of this signal transduction pathway in vivo, we used an antibody specific to the phosphorylated form of ERK (pERK) to probe sections of calvarial bone. As shown in Fig. 6D, we observed weak pERK expression in the osteogenic fronts of the parietal and frontal bones of wild-type and En1-null calvariae,whereas no activity was detected in periosteal osteoblasts of either genotype. By contrast, strong pERK activity was observed in the endosteal osteoblasts of wild-type frontal bone trabeculae at E16.5 (not shown) and P1. Strikingly, the number of endosteal osteoblasts exhibiting pERK activity in En1^-/- calvariae was significantly reduced. These in vivo findings show that Spry2 and pERK lie within distinct spatial domains of calvarial bone, which strongly suggests that signal transduction pathways other than the ERK/MAPK cascade mediate osteoblastic induction of Spry2 downstream of FGFs. In addition, ERK activity correlates with advanced osteoblast maturation, and its deficiency in En1^-/- calvariae suggests that impairment in FGF signaling might be a contributing factor to the perturbed osteogenic differentiation that characterizes En1 mutants.

In addition to impaired osteogenesis, increases in osteoclast number or activity could potentially contribute to the calvarial osteopenia in En1 mutants. Our observations of multiple perforations within the mature frontal and parietal bones of En1^-/- skulls suggested an osteolytic process (Fig. 6A). This prompted us to evaluate osteoclast activity in wild-type and En1^-/- calvariae by using whole-mount staining for TRAP activity. TRAPs are produced by both mono- and multi-nuclear activated osteoclasts, and are localized primarily to the osteoclast ruffled border and the extracellular resorptive space(Minkin, 1982). We observed that TRAP-staining in the parietal bones of wild-type mice at P5 was stronger near the bone margins, whereas areas distant from the sutures stained less intensely (Fig. 6B). In contrast to wild-type bones, the analogous bones of En1^-/-mice showed uniform TRAP activity throughout. Histological analysis of coronal sections of wild-type calvariae at P1 confirmed that osteoclasts were restricted to areas of mature trabecularized bone, and predominantly occupied the endocranial surfaces (Fig. 6C, arrows). By comparison, En1^-/- cranial bones displayed significantly more TRAP⁺-osteoclasts, which were often larger and located along trabeculae distant from the endocranial surface(Fig. 6C, arrowheads). Substantiating these observations, quantitative histomorphometry performed on E18.5 and P1 calvariae showed increased osteoclast numbers in En1mutants, consistent with the hypothesis that dysregulated osteoclastogenesis underlies the osteolytic phenotype of En1 null calvariae (E18.5: wild type, 17.11±1.559, n=9; En1 null, 47.11±2.796, n=9; P1: wild type, 46.83±5.974, n=6; En1null, 76.50±2.766, n=6). These results suggest that increased osteoclast numbers may be a contributing factor to the osteopenic phenotype of En1 mutants.

To determine whether increased osteoclast recruitment in En1mutants results from alterations in known osteoclastogenic regulators, we examined the expression of receptor activator of NFκB ligand (RANKL;TNFSF11 - Mouse Genome Informatics), a TNF-related cytokine that promotes osteoclast differentiation (Takahashi,2002). Primary osteoblasts were released from wild-type and En1 mutant calvariae at P2 and cultured for 7 days under conditions promoting mineralization. Under these conditions, Rankl was strongly upregulated in En1^-/- cells(Fig. 6E,F). By comparison,expression of osteoprotegerin (Opg), a decoy receptor and inhibitor of Rankl, was comparable between wild-type and En1^-/- cells. These findings point toward an additional and non cell-autonomous role for En1 in regulating osteoclast differentiation and/or recruitment.

The present study demonstrates a novel and important requirement for En1 in regulating calvarial osteogenesis. Although not exhibiting gross patterning defects of the skull vault, En1^-/- mice develop calvarial bone hypoplasia and osteopenia, increased suture patency,and postnatal malocclusion. As both CNC- and mesodermally-derived bones are similarly affected in En1 mutants, it is unlikely that the phenotypes observed resulted from an impairment in CNC cell migration or determination(Jiang et al., 2002). Furthermore, En1 expression initiates in the calvarial skeletogenic mesenchyme at E11.5, several days following the cessation of CNC migration(Jiang et al., 2002). Interestingly, En1 expression temporally precedes that of the osteogenic determinant Osx, and, in the absence of En1, the onset of Osx expression is delayed. As Osx is necessary for potentiating the osteogenic fate of the skeletogenic mesenchyme(Nakashima et al., 2002), its perturbed expression provides a mechanistic basis for the delayed calvarial ossification in En1^-/- mice. Furthermore, that Osx expression remains impaired in En1-null osteoblasts,suggests that En1 also lies upstream of Osx during later phases calvarial osteogenesis, and thus mediates distinct functions in osteoblast differentiation.

Consistent with a later role for En1 in osteoblast differentiation and function, our quantitative morphometric analysis showing reduced bone volume in En1 mutants is indicative of generalized calvarial osteopenia. In correlation with this, En1^-/- osteoblasts were deficient in mediating osteoid mineralization and exhibited reduced ALP activity, an enzyme that is essential for this process(Fedde et al., 1999; Murshed et al., 2005). Corroborating its role in osteoblast function, En1 is expressed postnatally by ectoperiosteal and endosteal osteoblasts, as well as by terminally differentiated osteocytes. Moreover, ablation of En1results in impaired Ocn and Bsp expression, genes that are normally associated with advanced osteoblast differentiation(Aubin, 2002). Ocnexpression has also been shown to be dependent on Osx(Nakashima et al., 2002). However, the fact that Opn expression in En1^-/-calvariae is restored to wild-type levels postnatally, suggests that early phases of osteoblast differentiation can eventually occur in the absence of En1 (Fig. 4C,D). Taken together, our results strongly indicate that, in addition to its role in early osteogenic commitment, En1 is directly required for mediating late calvarial osteoblast differentiation and bone matrix mineralization.

Our study demonstrates that the En1^-/- calvarial phenotype arises in part from alterations in FGF signaling. The FGF signaling pathway is intimately involved in regulating calvarial osteogenesis, where it plays a major function in promoting osteoblast differentiation(Ornitz and Marie, 2002). Activating mutations in FGFR1 or FGFR2 that cause craniosynostosis in humans are recapitulated by knock-in mouse models that exhibit enhanced osteoblast differentiation(Eswarakumar et al., 2004; Zhou et al., 2000). Conversely, selective ablation of Fgfr2IIIc, the mesenchymal splice variant of the Fgfr2, results in impaired calvarial ossification and osteoblast differentiation (Eswarakumar et al., 2002). Mice lacking FGF18 ligand share phenotypic similarities with En1 mutants: they display defective calvarial ossification and delayed terminal osteoblast differentiation(Ohbayashi et al., 2002). Thus, the associated alterations in osteoblastic differentiation affecting En1^-/- calvariae are consistent with an underlying perturbation in FGF signaling. Importantly, our data demonstrate that these effects are not due to changes in the osteoblastic expression of Fgfr1,Fgfr2 and Fgf18, but are rather attributed to hampered events downstream of FGF receptor activation.

Two lines of evidence indicate that En1 regulates signaling mediated by FGFRs. First, the activation ERK, normally restricted to the mature endosteal osteoblasts of wild-type calvarial bone, is severely impeded in En1 mutants (Fig. 7D,E). Second, En1 ablation results in loss of the FGF target gene Spry2 in ectoperiosteal osteoblasts. The significance of limiting pERK to the most mature osteoblasts in bone is not entirely clear;however, it correlates with previous studies ascribing inductive functions for pERK during calvarial bone formation and advanced osteoblast differentiation(Kim et al., 2003; Xiao et al., 2002). Therefore, En1 might mediate the terminal differentiation of endosteal osteoblasts by potentiating ERK activity(Fig. 6E). The ERK/MAPK cascade has been demonstrated as an important intracellular mediator of FGF-signaling in multiple developmental contexts. Interestingly, recent studies have shown that ERK activity is frequently limited to the sub-domains of FGF-responsive regions (Corson et al., 2003). Reciprocally, a number of FGFR inhibitors (Sprouty, Sef, Spred, Mkp3)are induced by FGF signaling, and are expressed in patterns consistent with their role in restricting ERK activity(Kawakami et al., 2003; Lin et al., 2002; Minowada et al., 1999; Wakioka et al., 2001; Zhang et al., 2001). Here, we present novel evidence demonstrating the existence of select domains for ERK activation in calvarial bone. Accordingly, we found that Spry2, a biochemical antagonist of the ERK/MAPK pathway(Hanafusa et al., 2002), is preferentially expressed by ectocranial osteoblasts and sutural osteoprogenitors, indicating that it may play a role in spatially modulating FGF responsiveness. However, the fact that loss of Spry2 expression in En1^-/- calvariae did not result in enhanced ERK activity in ectoperiosteal osteoblasts suggests that other antagonists may also modulate ERK. Indeed, in the developing limb bud, Fgf8 signal responsiveness is attenuated in the mesenchyme by the cooperative activities of Spry1 and Mkp3, limiting ERK activation to the overlying ectoderm (Corson et al., 2003; Kawakami et al., 2003; Minowada et al., 1999). We postulate that EN1 regulates the establishment of a negative-feedback loop within the calvarial skeletal rudiments by inducing the expression of Spry2 in ectocranial osteoblasts, while potentially repressing the expression of other FGFR-signaling attenuators (e.g. Mkp3, Sef) in endosteal osteoblasts (Fig. 6E). Consequently, loss of En1 function would result in the observed reduction in endosteal pERK(Fig. 6E). Furthermore, En1 may regulate alternative FGF-signaling effectors known to affect osteoblast differentiation, such as p38 MAPK or PKC(Kozawa et al., 1999; Lemonnier et al., 2000; Lomri et al., 2001). A precise temporal and spatial delineation of these intracellular pathways will enable a better understanding of how osteoblastic differentiation is coordinated by EN1 and FGFs.

In addition to defects in calvarial osteoblast differentiation, altered osteoprogenitor proliferation is likely to contribute to the frontal foramina and gaping of the coronal sutures in En1 mutants. The interfrontal suture forms late in development (E18.5-P1) through the gradual approximation of the frontal bone margins. Prior to suture closure at the cranial apex, the bone margins are interposed by extensive mesenchyme that would preclude the effectiveness of regulatory signals between the opposing osteogenic fronts. By contrast, the coronal suture is established early (E12-E14) along the CNC-mesodermal lineage boundary, where a thin layer of mesenchyme maintains a consistent separation between the closely juxtaposed, but non-fusing frontal and parietal bones (Jiang et al.,2002). Together with recent studies showing distinct FGF-responsiveness between the sutures, it is reasonable to infer that the interfrontal and coronal sutures represent unique sites for intramembranous bone formation (Ignelzi et al.,2003).

Interestingly, En1 is selectively expressed by the interfrontal sutural mesenchyme, but is excluded from the osteoprogenitors(Fig. 1). We postulate that the proliferative defect in the interfrontal suture, which becomes apparent only by E18.5, stems from deficient commitment of the mesenchyme rather than from a direct requirement for En1 in promoting osteoprogenitor mitosis. By comparison, in the coronal suture En1 is expressed by osteoprogenitors along the opposing bone margins, and its absence results in increased proliferation and mesenchymal thickening. This suggests that EN1 is a direct negative regulator of osteoprogenitor proliferation at this location. Previous studies indicated that Fgfr1 and Fgfr2 elicit a differential mitogenic response to FGFs in coronal suture osteoprogenitors(Ignelzi et al., 2003; Iseki et al., 1997; Iseki et al., 1999). Moreover,specific activating mutations in Fgfr2 are known to enhance proliferation within this population(Eswarakumar et al., 2004). It is therefore possible that upregulation of Fgfr2 and Fgfr1in the coronal sutures of En1^-/- calvariae results in the enhancement of an FGF-mediated mitotic response.

Following the completion of calvarial morphogenesis, prenatal and postnatal cranial bone expansion is modulated by the resorptive activity of osteoclasts. Calvarial osteoclast activity has been shown to initiate as early as E16.5 in the mouse and is important for the modeling and remodeling of the skull vault during brain growth (Rice et al.,1997). Here, we show that loss of En1 function results in focal calvarial osteolytic lesions that correlate with a significant increase in osteoclast number and activation. Consistent with resorptive bone loss, En1-ablated mice exhibit reduced calvarial bone volume and increased marrow space. In agreement with this, osteoclast numbers within the calvarial bone rudiments of En1^-/- mice are significantly increased.

Osteoclast differentiation and activation occurs in response to specific cytokines and growth factors secreted by osteoblasts and their progenitors within the bone marrow microenvironment(Takahashi, 2002). By binding to its cellular receptor RANK, RANKL mediates signal transduction pathways that result in overt osteoclast differentiation. Osteoclastogenesis, in turn,is balanced by the osteoblast-specific expression and secretion of osteoprotegrin (OPG), a decoy receptor capable of binding and inhibiting RANKL. Curiously, osteoprogenitors and less differentiated osteoblasts have been shown to express higher levels of RANKL and to support osteoclastogenesis to a greater extent than differentiated osteoblasts(Atkins et al., 2003; Gori et al., 2000). Accordingly, we have demonstrated that En1-null osteoblasts, arrested in an early stage of differentiation, display a specific increase in RANKL expression. Furthermore, the aberrant expression of Opn, a known osteoclastic chemoattractant, by terminally differentiated osteocytes has been associated with increased bone remodeling and osteoclast recruitment(Terai et al., 1999; Yamazaki et al., 1999). Interestingly, we observed abnormal Opn expression in En1^-/- calvarial osteocytes, suggesting an additional mechanism for increased osteoclast recruitment in these animals. In addition to its direct role in regulating calvarial osteogenesis, these findings demonstrate a novel role for En1 in inhibiting osteoclastogenesis by osteoblasts. The En1-null mouse thus provides a valuable tool for studying the interactions between osteogenesis and osteoclastogenesis during intramembranous ossification.

The authors thank Drs Gillian Morriss-Kay and Lynn Opperman for valuable discussion and suggestions. We thank Jean-Sebastien Binette and Ailian Li for their expert advice, and Dr Branka Dabovic for technical assistance and reviewing this manuscript. We thank Dr Alexandra Joyner for the ongoing evaluation of this work and the critical review of this manuscript. This work was supported by NIH grant R01AR47194.