Developmental Dynamics

Ephrin-B stimulation of calvarial bone formation

Authors


Abstract

Introduction: Ephrin-B2 on osteoclasts was reported to promote bone formation as part of homeostasis by activating the EphB4 tyrosine kinase receptor on osteoblasts. Little is known about the role of ephrin-B signaling to EphBs in developmental bone formation. Results: We observed expression of an ephrin-B2 LacZ chimeric allele in the periosteum, sutural bone fronts, and dura mater of embryonic and neonatal mice. Expression in the adult skull was confined to sutures, but was heavily upregulated at sites of bone injury. Culture of embryonic calvariae with soluble recombinant ephrin-B2/Fc doubled their bone content without altering suture width or overall skull morphology. Ephrin-B2/Fc also stimulated osteoblast marker gene expression in cultured MC3T3 preosteoblastic cells without the need for type 1 collagen-induced differentiation. EphB4 was absent in embryonic and adult skulls. However, EphB1 and EphB2, both physiological receptors for ephrin-Bs, were expressed at sites of osteogenesis, and EphB1 knockout mice displayed a reduction in calvarial bone content compared to controls. Conclusions: These data support a role for ephrin-B2 in the development and healing of bone through activation of osteoblast-specific gene expression. EphB1 and EphB2 are likely candidates receptors for the ephrin-B2 in bone. Developmental Dynamics, 2012. © 2012 Wiley Periodicals, Inc.

INTRODUCTION

Bone is produced by osteoblasts (OBs), which develop from mesenchymal precursors in response to a number of growth factors and hormones (Hughes et al., 2006). Their final stage of differentiation into mineralizing OBs requires contact with the type 1 collagen extracellular matrix (ECM) in which the cells embed themselves. The α2β1 integrin recognizes this ECM and activates a signaling cascade that results in osteoblast-specific gene expression (Xiao et al., 1998; Franceschi et al., 2007). Recent studies have identified another contact-mediated mechanism of OB differentiation involving the ephrin family of membrane-bound ligands and their Eph receptors.

The Ephs are the largest family of receptor tyrosine kinases (RTK), and control a wide range of contact-dependent developmental processes, including cell migration, boundary formation, midline septation, and axon guidance (Kullander and Klein, 2002; Klein, 2009). Ephs are classified as A or B based on their affinity for the glycosyl-phosphatidyl inositol-linked ephrin-As or the transmembrane ephrin-Bs. The ephrins are unique among RTK ligands in that they can also function as receptors to transduce signals in ephrin-bearing cells, a process called “reverse signaling” (Murai and Pasquale, 2003). B ephrins signal by way of conserved residues on their cytoplasmic domains that bind signaling intermediates such as src family kinases, SH2 domain adaptor proteins, and PDZ-binding domain containing proteins. A ephrins can signal through their association with membrane-spanning coreceptors, such as the p75 low-affinity neurotrophin receptor (Lim et al., 2008).

Zhao et al. (2006) found that application of ephrin-B2 to OBs to activate forward signaling promoted OB gene marker expression and increased mineralization. At the same time, activation of reverse signaling in osteoclasts (OCs) by application of EphB4 protein suppressed OC differentiation and slowed bone resorption (Zhao et al., 2006). As bone homeostasis, the coordination of bone formation and breakdown, is achieved by coupling of differentiation and activity between bone-forming OBs and bone resorbing OCs, these authors proposed that the ephrin-B2/EphB4 pair form a contact-mediated coupling mechanism. This mechanism would act in concert with the prototypical coupling scheme in which OB-released diffusible RANKL (Receptor Activator of Nf-kB Ligand) binds to OC precursors to increase their differentiation (Rodan and Martin, 2000; Karsenty and Wagner, 2002). The advantage of the novel ephrin mechanism is that it prevents resorption from competing with synthesis at the same physical location on remodeling bone surfaces. These authors' hypothesis is supported by molecular profiling studies by Allan et al. that found ephrin-B2 expression was increased in OBs and osteoblastic cell lines in response to treatment with parathyroid hormone (PTH) or parathyroid hormone–related peptide (PTHrP), both of which are important regulators of bone homeostasis (Silva et al., 2011; McCauley and Martin, 2012).

The expression of multiple Ephs and ephrins on neonatal OBs suggests that they might participate in developmental as well as homeostatic bone formation. Indeed, genetic studies have linked a number of mutations in ephrin-B1 to craniofrontonasal syndrome, a disorder characterized by craniosynostosis (Twigg et al., 2004; Wieland et al., 2005). Further, molecular studies by Davy et al. (2006) and Xing et al. (2010) identified deficiencies in OB precursor differentiation caused by cell-autonomous ablation of ephrin-B1 reverse signaling. We asked whether ephrin-B/EphB forward signaling plays a role in developmental bone formation. Here we report the expression patterns of ephrin-B2 and several candidate EphB receptors in developing and adult calvarial bone and the effect of activating Eph forward signaling on bone synthesis and OB gene expression.

RESULTS

Ephrin-B2 Expression in the Mouse Skull

We examined the expression of ephrin-B2 at specific stages of mouse calvarial development by staining for β-galactosidase (β-gal) in ephrin-B2 LacZ mice. These mice harbor a chimeric allele in which the coding region for the cytoplasmic domain of ephrin-B2 is replaced by that of β-gal (Dravis et al., 2004). We observed ephrin-B2/β-gal in the periosteum of mouse calvaria as early as e15.5, when the calvarial bones are growing toward each other to form sutures (Fig. 1B). The potently osteogenic dura mater that underlies these bones also showed high levels of expression.

Figure 1.

Ephrin-B2 expression in embryonic cranial bones. X-gal stain (blue) of ephrin-B2/β-gal mouse skulls. A: Diagram of the bones and sutures in the mouse calvaria, with expanded views of coronal and interfrontal sutures and arrowheads showing directions of developmental bone growth. B: e15.5 sagittal section of parietal bone and developing coronal suture at 200× magnification. C: p0 horizontal section of the coronal suture at 100×. D: p0 horizontal section of the interfrontal suture at 100×. E: Adult sagittal section of lambdoid suture at 40×. pb, parietal bone; fb, frontal bone; sob, supraoccipital bone; cs, coronal suture; ifs, interfrontal suture; ss, sagittal suture; ls, lambdoid suture; d, dura mater.

Expression of ephrin-B2/β-gal in newborn calvariae was still evident on both the periosteal and endosteal surfaces, but at much reduced levels compared to e15.5. Expression was strongest in suture mesenchyme and adjacent bone fronts of both overlapping and abutting sutures (Fig. 1C and D). In adult skulls, we observed only faint expression in the sutural mesenchyme and nowhere else in the calvaria except for a few cells in the bone marrow (Fig. 1E).

Ephrin-B2 Stimulates Bone Growth

Ephs must aggregate in clusters of four or more in the cell membrane to initiate biologically relevant signaling. This clustering can be induced artificially with soluble chimeric proteins made of an ephrin ectodomain fused to human IgG Fc-gamma and clustered by the addition of anti-Fc antibody. These pre-clustered proteins are used to induce Eph signaling. When applied unclustered, Eph or ephrin Fc recombinant proteins still bind their targets, but do not initiate signaling. Thus, the unclustered protein acts as a competitive inhibitor of Eph signaling. We tested the ability of clustered ephrin-B2/Fc to stimulate bone growth in ex vivo cultured embryonic mouse calvariae (Opperman et al., 2006). After 5 days of culture, ephrin-B2 treated skulls had double the bone volume of those treated with IgG-Fc control protein (Fig. 2A). We also incubated skulls with unclustered EphA4/Fc to block signaling. EphA4 is a promiscuous binder of ephrins, including ephrin-B2, and acts as a pan-ephrin inhibitor when applied without clustering (Pasquale, 2004). Surprisingly, this treatment did not reduce bone volume in cultured calvariae. We previously used the same batch of EphA4/Fc successfully to inhibit B ephrin signaling in palatal fusion (San Miguel et al., 2011), and demonstrated that it can reverse ephrin-A5-induced growth cone turning (data not shown). These tests confirmed that the protein is active. Thus, there must be another explanation for its failure to suppress bone growth in this study.

Figure 2.

Ephrin stimulation of bone growth in mouse calvariae. Skulls of e17.5 mice were grown for 5 days in IgG Fc control protein, clustered (bioactive) ephrin-B2/Fc, or unclustered (blocking) EphA4/Fc. A: Tissues were fixed and scanned for bone volume by micro-CT. Values shown are mean ± standard error for a representative experiment. Statistical significance was measured by Student's t-test. Experiments were performed three times (n=7 to 11 for each group per replicate) with identical results each time. B: Calvaria were imaged on each day of culture and interfrontal suture width was determined with Image J software. Suture widths in experimental groups were not significantly different from IgG Fc control (P = 0.4 and 0.12 for ephrin-B2 and EphA4 groups, respectively, as measured by t-test). C: Representative images of Alizarin Red–stained calvariae from each treatment group (top row), hematoxylin-and-eosin-stained coronal sections of frontal bones (middle row), and lateral frontal bone trabeculae (bottom row). Arrows indicate mesenchymal cells enveloping advancing bone fronts in the interfrontal sutures. Arrowheads indicate the leading tips of frontal bone.

Because ephrins control cell migration and suture formation during early embryogenesis (Davy et al., 2004; Merrill et al., 2006), we examined the morphology of ephrin-B2- and EphA4-treated skulls to determine if the anabolic activity observed was accompanied by morphological changes, such as narrowing of sutures. We observed no gross differences in shape of the calvarial bones, and measurement of interfrontal suture width, though appearing to trend lower in the ephrin-B2 and EphA4 treated groups, was not significantly different from the control group when statistical testing was applied (Fig. 2B). Coronal sections of the same calvariae showed normal aggregations of sutural mesenchyme in the interfrontal sutures surrounding the advancing bone fronts, and normal trabeculated bone in the thicker, more lateral areas (Fig. 2C).

Forward Signaling Increases Osteoblast-Specific Gene Expression Independent of the OB-Produced Type 1 Collagen Matrix

Zhao et al. (2006) reported that treatment of primary neonatal osteoblast cultures with clustered ephrin-B2 increased the expression of OB genes such as osteocalcin (Ocn), osterix (Osx), and alkaline phosphatase (Alp). This induction required several days of treatment. These cultures, while enriched as much as possible for osteoblasts by sequential collagenase digestion, are a mixed collection of calvarial cells at various stages of differentiation and lineage commitment (Bellows et al., 1986; Bellows and Aubin, 1989). Thus, it is difficult to determine whether the ephrin effect was the direct consequence of activating OB-specific gene transcriptional elements in OBs or an indirect result of increasing precursor proliferation or commitment earlier in the OB lineage.

We hypothesized that forward signaling from EphBs acts on specific promoter elements of OB genes to increase their transcription. To test this, we observed the action of ephrin-B2 on gene expression in MC3T3-E1 clone MC4 cells. The non-transformed, immortalized MC4 line is a faithful representation of an embryonic preosteoblast that is competent to differentiate into a mineralizing OB upon induction of matrix synthesis. Administration of ascorbate allows secretion of the required type 1 collagen ECM, which then activates the intracellular signaling cascade that causes OB-specific transcription (Benson et al., 1999, 2000; Xiao et al., 2000). Because these cells are uniformly representative of the committed pre-osteoblast, they allow the study of this final stage in a way that primary cultures do not. Our qualitative PCR tests confirmed that MC4 cells express the same panel of ephrins and Ephs as previously published for neonatal mouse calvarial OBs (Zhao et al., 2006; Fig. 3A), thus validating them as a model for OB gene expression in this study. We plated MC4 cells at near confluence (the standard density of 50,000/cm2) and grew them for 6 days with or without ascorbate and either ephrin-B2/Fc or control Fc protein. At this high density, we did not observe consistent, statistically significant elevation of the bone-specific mRNAs for Ocn and Bone sialoprotein (Bsp).

Figure 3.

Ephrin stimulation of osteoblast expression in MC3T3 MC4 cells. A: Cells were cultured for 12 days in the absence (C) or presence (A) of ascorbic acid. RNA was then harvested, reverse-transcribed, and amplified with gene-specific primers for mouse A and B class ephrins and Ephs. Brain RNA (Br) was used as positive control. No-reverse-transcriptase controls were negative. B: Real-time PCR for Ocn and Bsp (normalized to Gapdh) using mRNA from MC4 cells plated at low density and grown for 9 days with preclustered IgGFc or ephrin-B2/Fc. Mean fold stimulation of ephrin-treated samples over control are shown ± SEM for a combination of three experiments. C: Adult CD1 mouse bone marrow stromal cells (BMSC) were plated at 5,000 or 10,000 cells per well and treated with 4 μg/ml clustered ephrin-B2/Fc fusion protein or IgFc. Triplicate wells were harvested at the indicated times for colorimetric MTS assay to quantify relative numbers of viable cells as a measure of proliferation. Shown are mean absorbance at 490 nm ± std. deviation for representative experiments. Ig, IgGFc; EB2, ephrin=B2/Fc; AA, ascorbic acid. EB2-treated MC3T3 cells plated at 5K cells/well were significantly fewer than Ig-treated cells on day 5 of culture (P = 0.0075 by Student's t-test), as were EB2-treated BMSC plated at 10K cells/well (P = 0.04). EB2-treated values were not significantly different from Ig controls at all other points.

The recently published study by Xing and colleagues (2010) suggested that endogenously produced EphBs in bone marrow stromal cells would cause paracrine activation of ephrin-B1 reverse signaling, and so to reduce this effect the authors plated their cultures at low density to minimize cell–cell contact (Xing et al., 2010). We applied this strategy to our MC4 cultures. When we examined ephrin-B2/Fc action on cells plated at low density (5,000/cm2; roughly 10% confluence), we observed a dramatic induction of both Ocn and Bsp mRNAs (Fig. 3B), even without ascorbate addition. Our experiments showed that Eph/ephrin forward signaling in osteoblasts stimulates expression of genes associated with OB differentiation, and we were able to separate this stimulation from the induction caused by collagen ECM-induced OB differentiation.

Eph forward signaling controls cell proliferation in craniofacial structures (Bush and Soriano, 2010; Genander and Frisén, 2010), and we considered that ephrin-B2 stimulation of OB gene expression might be partly a result of increased proliferation. We tested this in primary mouse bone marrow stromal cells (BMSCs) and in MC4 cells by measuring the number of viable cells in culture over a time course of ephrin-B2/Fc treatment. Although plated at subconfluence to minimize paracrine Eph activation, neither cell type showed an ephrin-induced increase in cell numbers (Fig. 3C). Thus, it is unlikely that ephrin stimulation of bone production acts through stimulation of preosteoblast proliferation. On the contrary, both BMSCs and MC4 cells showed a slight but statistically significant reduction in numbers with ephrin treatment at the end of the time course (P = 0.04 and 0.0075, respectively). This result may indicate the influence of ephrin signaling away from proliferation and toward a differentiation program consistent with our gene expression results.

Ephrin-B2 Is Upregulated at Sites of Bone Injury

Because ephrin-B2/β-gal expression in adult skulls was expressed in sutures, and the suture mesenchyme is a source of osteoblastic cells in the adult mouse, we explored whether the healing response evoked by bone injury would change ephrin-B2 expression in the calvaria. We administered a circular defect to adult ephrin-B2 LacZ mice and examined β-gal expression by X-gal stain. By 1 week post-injury, the defect largely filled with tissue attached to the old bone on the edges of the hole. This new tissue contained high levels of ephrin-B2/β-gal (Fig. 4A and Bi, ii). Suture expression was maintained, as well as that in the vasculature, but none was observed in the bone tissue away from the injury site (Fig. 4A and Biii, iv). This showed that the upregulation was specific to the injury site, and not a global reactivation of ephrin expression throughout the periosteum.

Figure 4.

Ephrin-B2 stimulation at site of bone injury. A: Whole mount image of an adult ephrin-B2 LacZ mouse stained with X-gal 1 week after a 3-mm hole was drilled into the parietal bone. Right: Higher magnification of left panel. Arrow indicates blue suture tissue. Arrowhead indicates blue blood vessels. B: Sagittal sections of the skull in A stained with X-gal and counterstained with Nuclear Fast Red. Frame (i), anterior margin of defect at 40× magnification; (ii) edge of defect at 100×; (iii) coronal suture anterior to defect at 100×; (iv) lambdoid bone posterior to defect, 100×.

Expression of EphB Candidate Receptors for the Ephrin-B2 in Bone

Primary calvarial cultures and MC4 cells both express a number of EphBs that might serve as receptors for ephrin-B2 in bone. We examined the expression in mouse calvariae of EphB1, EphB2, and EphB4, all of which have been implicated in OB differentiation and for which we have β-gal marker mouse lines. EphB2 LacZ and EphB1 LacZ mice harbor cytodomain β-gal fusions in Eph B1 and B2, respectively, similar to the ephrin-B2 LacZ mouse (Henkemeyer et al., 1996; Chenaux and Henkemeyer, 2011). EphB4 LacZ mice have a tau-β-gal coding sequence under control of the EphB4 start codon (Gerety et al., 1999).

EphB4 was reported to play an important functional role in bone homeostasis as the ephrin-B2 receptor. We examined β-gal expression in EphB4 LacZ mice, and found none in e19.5 embryonic (Fig. 5A), or adult (Fig. 5B) skulls, although intense expression was observed in blood vessels, as expected from the literature (Fig. 5C) (Gerety et al., 1999). When we applied our calvarial defect to these mice, we detected a few β-gal-positive cells in the sutures after a week post-op, but none in the defect itself (Fig. 5D and E).

Figure 5.

EphB4 LacZ expression in cranial bones. Xgal stains (blue) of tissues from EphB4 LacZ mice. A: Sagittal section of coronal suture in an e19.5 embryo at 100×. B: Sagittal section of lambdoid suture in an adult skull at 40×. C: Section of skin at 100× in the same embryo shown in A confirming expression in blood vessels. D: Whole-mount view of a skull harvested 1 week after calvarial defect surgery (marked by bracket) at 40×. E: Sagittal section of the coronal suture of the skull shown in at 100×. F: Tissue from center of defect in the same section as shown in E at 100×.

By contrast, we observed high levels of β-gal in the skulls of EphB2 LacZ mice at e18.5. Expression was highest at the bone fronts of both overlapping and abutting sutures, and in the dura mater (Fig. 6A and B). In adult mice, we no longer observed staining in these sites, but in osteocytes instead (Fig. 6B). These data implicate EphB2 as an ephrin receptor during developmental, but not adult, bone growth, and suggested to us that EphB2 knockout mice would show defects in calvarial bone and/or suture formation. However, when we screened EphB2 knockout skeletons by x-ray, we saw no obvious deformities or defects (M.D.B., unpublished observations). This indicates that EphB2 is not on its own responsible for embryonic bone growth. Understanding its role in bone will require a more detailed analysis of EphB2 skeletons.

Figure 6.

EphB2 and EphB1 expression in postnatal cranial bones. A: X-gal stain of coronal suture in an e18.5 EphB2 LacZ embryo. B: Coronal section of the interfrontal suture of the same embryo shown in A. C: Sagittal section of x-gal stained adult parietal bone from an EphB2 LacZ mouse. D: X-gal stain of EphB1 LacZ mouse skulls at postnatal days 3, 5, and10; bottom panels are higher magnifications of the images above. E: Coronal section of p10 skull from D counterstained with Nuclear Fast Red. F: Average mineralized bone volume of p5 EphB1 calvariae from wild type (wt) and knockout (ko) littermates. G: Average bone volume of adult EphB1 knockout calvariae and wild type littermates. N is 4 to 7 for each group. Error bars are ± SEM; P = 0.04 for knockout group compared to wild type by Student's t-test. wt, wild type; ko, EphB1 homozygous.

We observed β-gal expression in postnatal EphB1 LacZ mice, not in the sutures, but in OBs on the periosteal surface (Fig. 6D and E). Expression was faint at postnatal day 3 but became much stronger by p10. This suggested that EphB1 is not involved in embryonic bone formation, but mediates ongoing bone growth as the skull expands and thickens. As a first step toward addressing this possibility, we analyzed bone volume in the calvariae of 5-day-old and adult EphB1 knockout mice, in which the EphB1 coding sequence is interrupted by a neo cassette (Williams et al., 2003). There was no difference in calvarial bone volume between homozygotes and their wild type littermates at day 5, but we observed 20 to 25% less mineralized tissue in the skulls of adult knockouts compared to wild type (Fig. 6F and G). We do not yet know if this is due to a specific deficiency in OB differentiation or function or is indicative of a non-skeletal-specific widespread growth deficiency that manifests later in life. However, it is consistent with the pattern of EphB1 expression and our hypothesis for EphB1 as a mediator of postnatal bone growth.

DISCUSSION

The data we report here indicate a central role for ephrin-B2 in bone growth during development, expansion, and healing of the skull. During embryogenesis and the early postnatal period, when bone grows at its fastest rate, ephrin-B2 expression was highest at the primary sites of calvarial bone synthesis, and application of soluble, active ephrin-B2/Fc dramatically increased mineral deposition in this tissue. The intensity and pattern of expression suggests that the main influence of ephrin-B2 in growing bone is through paracrine signaling by cells in the periosteal and osteoblast layers and the adjacent dura mater. Interestingly, Opperman and colleagues reported years ago that the dura mater produces a highly osteo-inductive activity that is not diffusible (Opperman et al., 1995). Our data suggest that this activity may be due to ephrin-B2.

Different Roles for Ephrins in Developing and Adult Bone

We did not observe ephrin-B2 expression on periosteal, endosteal, or trabecular bone surfaces in adult skull or long bones, and thus cannot confirm a role for this ephrin in bone homeostasis, although we did not attempt to colocalize β-gal staining in Ephrin-B2 LacZ bones with TRAP stain for osteoclasts. Our histological observations would seem on the surface to be at odds with Zhao's report of robust ephrin-B2 expression in cultured OBs and thus a presumed role in homeostasis of adult bone (Zhao et al., 2006). However, the cultured OBs in Zhao's study were taken from neonatal mouse calvariae, at a stage when we too found robust ephrin-B2 expression in OB layers in the skull, and so our data support that of those authors. This highlights the need to take into account the developmental stage of the bone when examining the mechanisms of ephrin action.

In our studies, the sole location of ephrin-B2 in undamaged adult skull was in the suture mesenchyme. The cranial sutures are a form of stem cell niche in that while they remain patent they maintain a population of potentially osteogenic cells (Opperman et al., 1995; Rawlins and Opperman, 2008). These cells can be induced to generate OBs that produce bone for calvarial bone expansion. They can also ossify the sutures and cause synostosis in the presence of altered Wnt and FGF signaling (Maruyama et al., 2010; Mirando et al., 2010). Since ephrin-B2 is associated with these suture stem cells, and then appears at adjacent sites of bone injury, we hypothesize that the tissue growing within the injury is produced by cells that have migrated from the suture. Further, we hypothesize that ephrin-B2 controls these cells' migration. This scenario is analogous to a recently described role for ephrins in migration of tooth pulp stem cells to areas of damaged dentin (Arthur et al., 2009). However, we recognize that the injury-generated ephrin-B2 cells might also come from surrounding periosteal cells in which ephrin-B2 expression is reactivated, or even from the adjacent vascular tissue, the endothelium of which Olsen and coworkers have shown is capable of generating osteoblasts (Medici et al., 2010; Medici and Olsen, 2011, 2012). To test our hypothesis, future experiments will need to trace the lineage of ephrin-B2-positive cells at sites of calvarial injury, examine the role of ephrin-B2 in their migration to those sites, and test the requirement for ephrin-B2 in bone healing.

We were surprised to find that unclustered EphA4/Fc protein was not able to reduce bone volume in cultured calvariae. One possible explanation for this result lies in a report by Irie et al. (2009) that EphA2 forward signaling in osteoblasts inhibits differentiation of cultured neonatal osteoblasts. In this scenario, unclustered EphA4/Fc might in fact counter both osteogenic EphB signals and inhibitory EphA2 signals, thus canceling each other out. We have typically used EphA4/Fc as a pan-ephrin blocker because of its ability to promiscuously bind and inhibit both A and B pathways. However, we may find it more useful in future experiments to block osteogenesis with EphB Fc proteins that do not bind to ephrin-As and impinge on their receptors.

Our MC4 culture experiments demonstrated that ephrin-B2 induces OB-specific gene expression independent of signaling from the collagen ECM because our growth conditions did not allow for secretion of collagen. Nevertheless, we do not yet know if the Eph-mediated signal inside OBs is conducted through the known transcription factor intermediates described for OB differentiation (e.g., Runx2, Osx, Atf4, Dlx5, etc.). Xing et al. (2010) described a function for ephrin-B1 reverse signaing in activation of Osx, demonstrating the plausibility of a direct link between Eph/ephrin signaling cascades and transcriptional activation in OBs. Further studies will identify the cis-acting promoter elements in the Ocn and Bsp genes that respond to the ephrin-B2-induced forward signal in our experiments. The effect of ephrin treatment we observed on MC4 cells took 9 days under our conditions, by which time the cells had proliferated to near-monolayer density. Thus, it may be that OBs must be in contact to express differentiation markers even in the presence of elevated levels of exogenous ephrin. Davy et al. (2006) reported that ephrin-B1 is important for gap junctional communication between OBs, and that its lack leads to impaired OB differentiation. It is possible that a contact-mediated signal must be in place before the forward signaling activated by ephrin-B2 addition can take effect.

Different Ephs for Developmental and Postnatal Bone Growth

Despite the attention given to EphB4 as a receptor for ephrin-B2 in bone (Zhao et al., 2006; Martin et al., 2010; Matsuo, 2010), we did not detect β-gal in the skulls of the EphB4 LacZ mouse except for a smattering of cells in the suture after calvarial injury. This is consistent with the previous report by Xing et al. (2010) that EphB4 was undetectable in bone marrow stromal–derived OBs, but inconsistent with our and Zhao et al.'s (2006) PCR data showing EphB4 expression in MC4 cells and primary OB cultures. Instead we observed expression of EphB2 and EphB1. EphB2/β-gal was expressed in the same domains as ephrin-B2 during development, but disappeared from these areas in the adult, while EphB1/β-gal was found in the osteoblast layer of postnatal calvaria, but not in bone fronts near the sutures. Thus, both of these Ephs are in a position to function as mediators of bone formation, either as receptors for ephrin-B2 in forward signaling, or as ligands for B ephrins in reverse signaling. Their differing patterns of expression imply that they are not simply redundant, but perform different functions.

Given the intense EphB2 expression in developing bone fronts, we expected that mice lacking EphB2 would show gross deficiencies in calvarial bone growth or morphology. The fact that they did not by our initial x-ray screening would seem to argue against an important role for EphB2 in bone. However, EphBs are known to collaborate in other developmental systems such that only compound EphB knockouts reveal significant phenotypes. For example, EphB2 and EphB3 are both expressed in mouse secondary palate. Neither EphB2 LacZ homozygotes nor EphB3 knockouts have abnormal palate development, but double mutants have cleft palate (Risley et al., 2009). And, proliferation defects are evident in the intestinal crypts of EphB2;EphB3 double knockout mice, while each single mutant shows no such defect (Holmberg et al., 2006). Thus, the relative importance of EphB2, and other Ephs, will likely require more detailed analysis of bone phenotypes in both single and compound knockout mouse lines. The presence of EphB2 in osteocytes is intriguing in light of the emerging importance of osteocytes in phosphate homeostasis (Kramer et al., 2010; Nakashima et al., 2011).

Our initial finding of reduced bone in the skulls of adult, but not neonatal, EphB1 knockout mice is also intriguing in that it suggests that different Ephs are required for different stages in bone growth. EphB1 was observed in the calvaria only after birth, and its loss had no effect on bone volume until later, indicating a sort of “handoff” of bone promoting responsibility among the Ephs. We will need detailed analysis of these animals' skeletons to determine the relative importance of EphB1 on bone growth. Equally important will be learning the identity of the ephrin ligand for EphB1 in adults, as ephrin-B2 largely disappears in the periosteum just as EphB1 expression ramps up.

EXPERIMENTAL PROCEDURES

Fc Protein Production and Treatment

Soluble forms of ephrin-B2 and EphA4 were made by fusing the ectodomains of these proteins to human Fc-gamma. These fusion proteins were purified from CHO cell–conditioned media as previously described (San Miguel et al., 2011). IgG Fc for use as a negative control was purchased from Calbiochem (San Diego, CA). Ephs and ephrins must be clustered to induce biologically relevant signaling (Knöll et al., 2007). To activate ephrin-B2/Fc, protein was incubated with anti-Fc antibody before adding to culture media. EphA4/Fc was used unclustered as a blocker of ephrin action. Ephrin-B2 was used at 4 μg/ml. EphA4/Fc was used at 20 μg/ml.

Mouse Lines, Surgeries, and Beta-Galactosidase Stain

Construction of ephrin-B2 LacZ (Dravis et al., 2004), EphB1 LacZ (Chenaux and Henkemeyer, 2011), EphB2 LacZ (Henkemeyer et al., 1996), EphB4 LacZ (Gerety et al., 1999), and EphB1 knockout (Williams et al., 2003) mouse lines have all been previously described All mice were on either a CD-1 or a CD-1/129SvJ mixed genetic background. Adult mice used in experiments were between 8 and 12 weeks old.

To quantify bone volume in 5-day-old or adult EphB1 mice, calvariae were dissected from homozygotes and their wild type littermates and scanned on a Scanco35 instrument. Bone volume was determined for 4 to 7 mice per group in each instance. Bone volumes were reported as mean ± standard error of the mean, with a Student's t-test used to determine statistical significance.

To introduce a calvarial defect into ephrin-B2 LacZ and EphB4 LacZ mice, a 3-mm hole was drilled in the left parietal bone of each 2-week-old adult mouse under anesthesia. The skin was then sutured over the skull and the mice were allowed to survive for 1 week before sacrifice.

To detect expression of β-gal in whole mouse tissues, animals were perfused lightly with 4% formaldehyde, and their calvariae or long bones were placed in X-gal solution (5 mM each potassium ferrocyanide and potassium ferricyanide, 2 mM magnesium chloride, 0.02% NP-40, 1 mg/ml X-gal) overnight at 37°C. For more detailed analysis, tissues were decalcified in 0.5M EDTA, pH 8.0, at 4°C, then cryoprotected in 30% sucrose/PBS, and crysectioned at 12 μm onto slides. Sections were then stained in X-gal solution and counterstained in Nuclear Fast Red. We performed extensive controls to confirm that activity of the β-gal enzyme in these tissues was not diminished by decalcification at 4°C (data not shown).

Calvarial Organ Culture

Calvariae (containing the frontal, parietal, and part of the occipital bones) were harvested from day-17.5 CD-1 mouse embryos and cultured as previously described (Opperman et al., 2006), except that the tissues were cultured for 5 days in alpha-MEM with 10% fetal bovine serum. The medium was changed on day 3. Bright-field pictures were taken each day with a stereomicroscope and digital camera. Suture width was measured from these pictures by an observer blinded to the samples' identities using ImageJ software (NIH). On day 5, the calvariae were fixed in 4% formaldehyde. Bone mineral volume and density were measured by micro-computed tomography on a Scanco35 instrument. These experiments were performed three separate times, with 7 to 11 samples per group in each experiment. Bone volumes were reported as mean ± standard error of the mean, with a Student's t-test used to determine statistical significance. Calvariae were then stained in 0.01% alizarin red for photography. Then, they were cryosectioned at 10 μm and stained with hemotaxylin and eosin in the coronal orientation to examine suture and trabecular morphology.

MC3T3-E1 and Bone Marrow Stromal Cell Culture

We used subclone MC4, a derivative of the MC3T3-E1 mouse calvarial cell line. These cells are propagated in a pre-OB state until given ascorbic acid, at which point they secrete a type 1 collagen matrix that activates integrin signaling and thus OB-specific gene expression. The MC4 line is a more highly differentiating subclone of the parent line (Wang et al., 1999). For each experiment, cells were plated at 5,000/cm2 in 35-mm dishes and grown for 9 days in alpha-MEM supplemented with 10% FBS, 1% penn/strep antibiotics, 50 μg/ml ascorbate (where specified), and 4 μg/ml of clustered ephrin-B2/Fc or IgG Fc. Medium and treatment were changed every other day.

Total RNA was harvested from cells with Trizol and treated with DNaseI to remove genomic DNA contamination. We performed real-time PCR analysis of Ocn and Bsp expression on cDNA synthesized from this RNA using a BioRad (Hercules, CA) CFX96 real-time machine. Eph and ephrin qualitative PCRs used Taq polymerase from Denville with primers specific for known ephrins and Ephs. Primer sequences are available on request.

Bone marrow stromal cells (BMSCs) were isolated from the bones of 8-week-old adult CD-1 mice as previously described and expanded in culture through one passage.

For proliferation assays, cells were plated at 5,000 or 10,000 cells per well in 96-well plates and grown for up to 4 days. Wells were harvested in triplicate on each day, and the number of viable cells in each well was detected by absorbance at 490 nm using a colorimetric MTS assay from Roche Applied Sciences (Indianapolis, IN).

Acknowledgements

The authors thank Dr. Paul Dechow for guidance in performing micro-CT analysis, Dr. Kathy Svoboda for critical reading of the manucript and for helpful discussions, and Dr. Priyam Yani for assistance with histology. This work was supported by R03DE020119 from NIDCR to M.D.B.

Ancillary