EphrinB2/EphB4 inhibition in the osteoblast lineage modifies the anabolic response to parathyroid hormone


  • Farzin M Takyar,

    1. St. Vincent's Institute of Medical Research, Fitzroy, Australia
    2. Department of Medicine at St. Vincent's Hospital, The University of Melbourne, Fitzroy, Australia
    Search for more papers by this author
  • Stephen Tonna,

    1. St. Vincent's Institute of Medical Research, Fitzroy, Australia
    2. Department of Medicine at St. Vincent's Hospital, The University of Melbourne, Fitzroy, Australia
    Search for more papers by this author
  • Patricia WM Ho,

    1. St. Vincent's Institute of Medical Research, Fitzroy, Australia
    Search for more papers by this author
  • Blessing Crimeen-Irwin,

    1. St. Vincent's Institute of Medical Research, Fitzroy, Australia
    Search for more papers by this author
  • Emma K Baker,

    1. St. Vincent's Institute of Medical Research, Fitzroy, Australia
    2. Department of Medicine at St. Vincent's Hospital, The University of Melbourne, Fitzroy, Australia
    Search for more papers by this author
  • T John Martin,

    1. St. Vincent's Institute of Medical Research, Fitzroy, Australia
    2. Department of Medicine at St. Vincent's Hospital, The University of Melbourne, Fitzroy, Australia
    Search for more papers by this author
  • Natalie A Sims

    Corresponding author
    1. St. Vincent's Institute of Medical Research, Fitzroy, Australia
    2. Department of Medicine at St. Vincent's Hospital, The University of Melbourne, Fitzroy, Australia
    • St. Vincent's Institute, 9 Princes Street, Fitzroy VIC 3065, Australia.
    Search for more papers by this author


Previous reports indicate that ephrinB2 expression by osteoblasts is stimulated by parathyroid hormone (PTH) and its related protein (PTHrP) and that ephrinB2/EphB4 signaling between osteoblasts and osteoclasts stimulates osteoblast differentiation while inhibiting osteoclast differentiation. To determine the role of the ephrinB2/EphB4 interaction in the skeleton, we used a specific inhibitor, soluble EphB4 (sEphB4), in vitro and in vivo. sEphB4 treatment of cultured osteoblasts specifically inhibited EphB4 and ephrinB2 phosphorylation and reduced mRNA levels of late markers of osteoblast/osteocyte differentiation (osteocalcin, dentin matrix protein-1 [DMP-1], sclerostin, matrix-extracellular phosphoglycoprotein [MEPE]), while substantially increasing RANKL. sEphB4 treatment in vivo in the presence and absence of PTH increased osteoblast formation and mRNA levels of early osteoblast markers (Runx2, alkaline phosphatase, Collagen 1α1, and PTH receptor [PTHR1]), but despite a substantial increase in osteoblast numbers, there was no significant change in bone formation rate or in late markers of osteoblast/osteocyte differentiation. Rather, in the presence of PTH, sEphB4 treatment significantly increased osteoclast formation, an effect that prevented the anabolic effect of PTH, causing instead a decrease in trabecular number. This enhancement of osteoclastogenesis by sEphB4 was reproduced in vitro but only in the presence of osteoblasts. These data indicate that ephrinB2/EphB4 signaling within the osteoblast lineage is required for late stages of osteoblast differentiation and, further, restricts the ability of osteoblasts to support osteoclast formation, at least in part by limiting RANKL production. This indicates a key role for the ephrinB2/EphB4 interaction within the osteoblast lineage in osteoblast differentiation and support of osteoclastogenesis. © 2013 American Society for Bone and Mineral Research.


In its application to anabolic therapy for the skeleton, parathyroid hormone (PTH) is administered by daily subcutaneous injection,1 with PTH(1-34) approved for the treatment of osteoporosis in a number of countries. The pharmacokinetics required for this effect are that a peak of circulating PTH is required, returning to control levels within 3 hours.2 If the elevation is prolonged, osteoclast formation and bone resorption are stimulated.2 This resorption effect is enhanced greatly with infusion of PTH over some hours3 or with the consistently elevated PTH of primary hyperparathyroidism.

Current views of the anabolic action of PTH are that it increases the activation of the basic multicellular units (BMUs) of remodeling, that it acts on committed osteoblast precursors to promote their differentiation, inhibits osteoblast and osteocyte apoptosis,4 and inhibits the production of the bone formation inhibitor, sclerostin.5, 6 There is also much interest in the possibility that PTH treatment results in transient activation of osteoclasts, which in turn produce activity that enhances the osteoblast differentiation effect. The latter may be independent of resorption7 or may result from the release of matrix-bound growth factors, such as TGFβ or insulin-like growth factors in the resorption process.8

Single injection of PTH in mice or rats is accompanied by rapid increases in expression of a number of proteins, including c-fos, IL-6, and RANKL3, 9 and a later decrease in sclerostin.5, 6 In using gene profiling to study the actions of PTH(1-34) and PTHrP(1-141) in murine osteoblasts differentiated to the stage of expressing functional PTH/PTHrP receptor (PTH1R), we found that both treatments resulted in up to 10-fold increases in expression of mRNA and protein for ephrinB2.10 This was confirmed in mouse calvarial cells and UMR106 rat osteogenic sarcoma cells, and PTH treatment in vivo in rats and mice resulted in a 10-fold increase in ephrinB2 mRNA in bone within 1 hour of injection. Further, blockade of signaling between ephrinB2 and its receptor, EphB4, in differentiating mouse stromal cells was found to inhibit the expression of mRNA for several proteins important in late stages of osteoblast differentiation, including osteocalcin and sclerostin.10, 11

Eph receptors and their Eph receptor interacting (Ephrin) ligands together form an important cell communication system with widespread roles in normal physiology and disease.12 Ephrin/Eph family members are recognized as local mediators of cell function through contact-dependent processes in development and in maturity.13, 14 A particular feature is their capacity for bidirectional signaling: When an ephrin ligand and its corresponding Eph receptor tyrosine kinase interact, signals are transduced both through the receptor (forward signal) and the ligand (reverse signal).15 Eph/Ephrin signaling mediates cell attraction and adhesion but often also provides signals that cause cell repulsion; these events have demonstrated roles in tissue remodeling, including cell migration, axon guidance, and synapse plasticity. The main ligand for EphB4 is ephrinB2, with ephrinB1 interacting with much less affinity.16 Some evidence suggests that osteoclast-derived ephrinB2 acts through a contact-dependent mechanism on EphB4 forward signaling in osteoblasts to promote osteoblast differentiation and bone formation, and that through reverse signaling, osteoblast-derived EphB4 acts upon ephrinB2 in the hemopoietic lineage to suppress osteoclast formation.17 To achieve these effects, the forward signaling would require contact between mature osteoclasts and cells differentiating in the osteoblast lineage, and in that model the reverse signaling mechanism requires contact between osteoblastic lineage cells and hemopoietic precursors of osteoclasts. The present work was focused on ephrinB2/EphB4 interactions within the osteoblast lineage because both are expressed by osteoblasts and aimed to determine the effect of blockade of ephrinB2/EphB4 interaction and whether it might influence the anabolic action of PTH. It arises from our findings that PTH and PTHrP rapidly enhanced ephrinB2 production by osteoblasts and that blockade of ephrinB2/EphB4 signaling impaired mineralization by osteoblasts in vitro.10, 11

Materials and Methods

Peptides, antibodies

Soluble protein comprising the extracellular domain of EphB4 (sEphB4) was prepared by expression in mammalian cells as previously described and was kindly provided by Dr V Krasnoperov (Vasgene Therapeutics, Los Angeles, CA, USA).18 Synthetic human PTH(1-34) was purchased from Bachem (Bubendorf, Switzerland). EphB receptor peptide antagonists previously identified by phage display19 were also used: TNYLFSPNGPIARAW (TNYL-RAW), the peptide antagonist of ephrin-B2 interaction with its receptor, EphB4, and SNEWIQPRLPQH (SNEW), the selective inhibitor of EphB2 receptor were synthesized by Auspep (Parkville, Australia). Affinity-purified anti-mouse antibodies to EphrinB2, EphB2, and EphB4 and the recombinant anti-mouse Ephrin-B2/Fc chimera were obtained from R&D Systems (Minneapolis, MN, USA).

Cell culture

Primary calvarial osteoblasts were obtained from 1- to 3-day-old C57Bl/6 mice, as previously described.20 Briefly, calvariae were resected from the animal and placed into ice-cold PBS to remove surrounding tissue and pooled. Sequential digestion of the calvariae was carried out using collagenase type II (Worthington, Lakewood, NJ, USA), Dispase (Gibco, Grand Island, NY, USA) in PBS (Sigma, St. Louis, MO, USA) for 4 × 10 minute incubations. The supernatant from each digestion was pooled and plated in complete medium (αMEM + 10% FBS) and grown at 37°C with 5% CO2. In experiments studying mRNA expression during differentiation, cells were grown in osteoblast differentiation media (αMEM + 15% heat-inactivated FBS + ascorbate [50 µg/mL; Sigma]) for the times indicated. NFκB activation was assessed using RAW264.7 cells stably expressing NFκB-luciferase (kindly provided by Dr Jiake Xu, University of Western Australia), co-cultured with Kusa4b10 cells, and treated with 1,25D for 24 hours as recently described.21

Western blotting and immunofluorescence

Calvarial osteoblasts were grown to confluence and starved of serum for 3 hours in αMEM. Cells were treated with 2 µg/mL ephrinB2-Fc and recombinant human IgG-Fc (clustered Ephrin-B2 Fc) or 2 µg/mL clustered EphB4-Fc for 30 minutes in the presence or absence of TNYL-RAW (30 µM), SNEW (50 µM), or sEphB4 (5 µg/mL), after which cells were lysed in RIPA lysis buffer containing protease and phosphatase inhibitors and frozen at −20°C. After defrosting, samples were divided into two aliquots and immunoprecipitated with 30 µL of washed protein G sepharose beads (GE Lifesciences, Buckinghamshire, UK) and either anti-EphB2 or anti-EphB4 antibody (2 µg) overnight at 4°C on rotation. The immunoprecipitates were washed with RIPA lysis buffer, then PBS. Samples were heated for 10 minutes at 60°C under reducing conditions, resolved on 4–12% gradient SDS-PAGE, and transferred to a PVDF membrane. Membranes were blocked in 3% BSA/TBS/0.1% Tween 20 and probed with either mouse-anti-phosphotyrosine (4G10) (Millipore, Billerica, MA, USA) or EphB4 or EphB2 antibodies, followed by anti-mouse or anti-goat-coupled HRP, respectively (Dako, Glostrup, Denmark), and bands were detected by chemiluminescence. To detect ephrin phosphorylation, cells were fixed with 4% paraformaldehyde and permeabilized with acetone. Samples were blocked in 5% BSA/TBST and immunoprobed for phosphorylated ephrinB receptors (Phospho-ephrinB2 Y324/329, Cell Signaling, Danvers, MA, USA), followed by an Alexa-594 fluorescent secondary conjugate (Invitrogen, Carlsbad, CA, USA) for detection by scanning confocal microscopy (Olympus, Tokyo, Japan).

To detect RANKL protein, primary calvarial osteoblasts were cultured in osteoblast differentiation media for 14 days and treated with sEphB4 (5 µg/mL) and 1,25-dihydroxyvitamin-D3 (10−8 M) for 6 hours. Western blotting was carried out with anti-RANKL22 followed by anti-rabbit coupled HRP (Dako) and bands detected by chemiluminescence. After this, membranes were stripped in 0.2 M sodium hydroxide, blocked in 5% BSA/TBST, and immunoprobed for pan-actin (Cell Signaling) followed by anti-mouse HRP (Dako), and bands detected by chemiluminescence.

In vivo experiments

Male C57BL/6 mice were purchased from the Animal Resources Centre (Canning Vale, Australia). Animals were housed in a 12-hour light and dark cycle with food and water provided ad libitum. All animal experiments were approved by the St. Vincent's Health Animal Ethics Committee.

To determine whether sEphB4 modifies the anabolic response to PTH(1-34), 8-week-old male C57Bl/6 mice were randomly allocated to control, PTH, sEphB4, or sEphB4 + PTH treatment groups with 8 mice per group. For PTH treatment, mice were administered 30 µg/kg PTH(1-34) by intraperitoneal injection, 5 days a week for 4 weeks. sEphB4 (10 mg/kg) was injected intraperitoneally 3 days a week. Vehicle (20 mM Tris-HCl, 150 mM NaCl at pH = 8 with 2% heat-inactivated mouse serum) was administered 5 days a week. A fourth group (sEphB4 + PTH) was injected with PTH(1-34) and sEphB4 at the same doses and timings as above. The PTH dose of 30 µg/kg was aimed at achieving an anabolic effect that could be measured with confidence, but was not the maximum that could be achieved with higher doses, and was based on previous experiments in similar aged animals.23 The sEphB4 dose was based on the effective use of sEphB4 in inhibiting the growth of human cancer cells in immune-deficient mice.18 Mice were weighed daily, and injection volumes were adjusted according to weight changes over the 4-week period. Double calcein labeling was performed by intraperitoneal injection of calcein 7 and 2 days before tissue collection.24 One hour after the last injection, a terminal blood sample was collected by cardiac puncture exsanguination and stored at 4°C. The femoral distal epiphysis including the growth plate was removed and the remaining bone was flushed of marrow and snap-frozen in liquid nitrogen for RNA preparation as previously described.25 Tibias were fixed in 4% paraformaldehyde and embedded in methylmethacrylate, and 5-µm sections were stained with toluidine blue for static histomorphometry or xylenol orange for dynamic histomorphometry.24 Histomorphometry was carried out in the secondary spongiosa of the proximal tibia (Osteomeasure, Osteometrics, Atlanta, GA, USA).24 One femur and the lumbar vertebrae were fixed in 4% paraformaldehyde overnight followed by storage in 70% ethanol for micro-CT analyses.

Micro-computed tomography (micro-CT)

Micro-CT analyses were performed on femoral and vertebral specimens with a Skyscan 1174 micro-CT system (Skyscan, Aartselaar, Belgium). During scanning, femora were enclosed in a plastic tube filled with 70% ethanol. The X-ray source was set at 50 kV and 800 µA and sharpening was set to 40%. Projections were acquired over 180° (step of 0.6°), pixel size of 6.48 µm, and exposure of 2000 ms. Image slices were reconstructed by NRecon (Skyscan) with the following settings: beam-hardening correction 30%, ring artifact correction 12, no smoothing, and no defect pixel masking. Reconstructed images were straightened with Dataviewer (Skyscan).

Micro-CT analysis of the primary spongiosa in the distal end of the femur was performed to assess the anabolic effects of intermittent PTH. Regions of interest were defined manually in the coronal plane as an inner trabecular area covering 20% of the width of the femur (∼400 µm); the length of the region was 15% of the total femoral length, beginning immediately below the growth plate. The threshold used for this analysis was from 42 to 255 for all samples.

RNA isolation and RT-PCR

RNA was isolated from bone specimens after homogenization in QIAzol Lysis Reagent (Qiagen Sciences, Gaithersburg, MD, USA) with an LS-10-35 Polytron homogenizer (Brinkmann Instruments, Westbury, NY, USA) using Qiagen RNeasy Lipid Tissue Mini Kit (Qiagen Sciences) and concentration determined by spectrophotometer (Nanodrop ND1000). RNA samples were DNase treated with Ambion Turbo DNA-free Kit (Ambion Inc, Austin, TX, USA) and cDNA was synthesized (Random primers, 10 mM dNTP, 5× First Strand buffer, 0.1 M DTT, RNaseOUT, SuperscriptTM III RT [200 U/mL]) with 1 mg material as follows: 5 minutes at 65°C, 1 minute at 4°C; 5 minutes at 25°C, 60 minutes at 50°C, 15 minutes at 70°C, held at 4°C (Biometra T3000 Thermocycler, Biometra GmbH, Göttingen, Germany). Amplification was carried out with AmpliTaq Gold (Perkin-Elmer, Norwalk, CT, USA), SYBR Green (Invitrogen), and specific oligonucleotide primers with a Stratagene Mx3000P (Invitrogen). Multiplex PCR was used to detect mRNA for osteocalcin (Bglap1), Osterix (Osx), IL-6 (Il6), RANKL (Rankl), and OPG (Opg), MEPE (Mepe), sclerostin (Sost), DMP1 (Dmp1), ephrinB2 (Efnb2), EphB2 (Ephb2), EphB4 (Ephb4), and EphA4 (Epha4). Alkaline phosphatase (Alpl), DC-STAMP (Tm7sf4), Cathepsin K (Ctsk), Collagen 1-α1 (Col1a1), EphA4, EphrinB1, EphB1, EphB3, and Fgf23 were determined using SYBR Green. SYBR Green primers for Hprt1,10 Col1a1,26 EphrinB1,27 and EphB328 have been described previously. Hypoxanthine guanine phosphoribosyl transferase 1 (Hprt1) was used as a housekeeping gene by both methods and was not significantly different between treatment groups. All novel primers are listed in Table 1.

Table 1. Primers Used for Real-Time PCR
Gene symbolAccession numberForwardProbeReverse
Multiplex primers  
SYBR Green primers 

Statistical analysis

Statistically significant differences were determined by one-tailed Mann-Whitney test or, where only one comparison was being made, by Student's t test using Prism 5. All data are presented as means ± SEM. A p value <0.05 was considered statistically significant.


EphrinB2 mRNA and its receptors, EphB4, EphB2, and EphA4, were expressed in primary calvarial osteoblasts throughout differentiation as previously described for Kusa 4b10 cells (Fig. 1A). Treatment of primary mouse calvarial cells with clustered ephrinB2-Fc promoted rapid tyrosine phosphorylation of EphB2 and EphB4 (shown at 30 minutes in Fig. 1B). Similar results were obtained in Kusa 4b10 and UMR106-01 cells (data not shown). Phosphorylation of EphB4 was inhibited by sEphB4 and by TNYL-RAW, a specific peptide antagonist of ephrinB2/EphB4 interaction, whereas neither TNYL-RAW nor sEphB4 blocked EphB2 phosphorylation by ephrinB2-Fc. The peptide antagonist of ephrinB2/EphB2 interaction, SNEW, inhibited phosphorylation of EphB2 induced by ephrinB2-Fc but did not inhibit EphB4 phosphorylation. This established the specificity of these inhibitors in osteoblasts.

Figure 1.

Effects of ephrinB2/EphB4 inhibition in primary calvarial osteoblasts. (A) mRNA levels for ephrinB2 (Efnb2) and its receptors (EphB2, EphB4, and EphA4) in primary calvarial osteoblasts throughout 21 days of culture in osteoblast differentiation media. Values are mean ± SEM, from triplicate wells, representative of three biological replicates. EphB4 and EphB2 (B) phosphorylation in primary calvarial osteoblasts are both stimulated with ephrinB2-Fc, and ephrinB phosphorylation (C) is stimulated by EphB4-Fc treatment, respectively. Induced EphB4 and ephrinB phosphorylation were specifically inhibited by 30 µM TNYL-RAW (TNYL) and 5 µg/mL soluble EphB4 (sEphB4), whereas EphB2 phosphorylation was inhibited by 50 µM SNEW. Blots and immunofluorescence are representative of three repeated experiments.

In addition, both sEphB4 and TNYL-RAW inhibited ephrin (reverse) signaling (Fig. 1C). This was detected by a pan-phospho-ephrinB antibody.29 Of all ephrinBs, only ephrinB1 and ephrinB2 are expressed in osteoblasts,10, 17 and EphB4 binds ephrinB2 with a 100- to 1000-fold higher affinity than ephrinB1.16 EphB4-Fc induced ephrinB phosphorylation in osteoblasts (Fig. 1C), and this was blocked by both sEphB4 and TNYL-RAW (Fig. 1C).

In previous work, we showed that blocking ephrinB2/EphB4 receptor interaction by treating differentiating mouse stromal cells with TNYL-RAW or treating Kusa 4b10 cells with sEphB4 reduced expression of several mRNAs that feature in later stages of osteoblast differentiation.10, 11 This was examined further in calvarial osteoblasts, using receptor inhibition with sEphB4 to prevent interaction between ephrinB2 and EphB4. In calvarial osteoblasts differentiated for 7 days, both early and late osteoblast markers are readily detectable, and treatment with sEphB4 for 24 hours resulted in decreased expression of late osteoblast markers, including osteocalcin, Dmp1, and Mepe (Fig. 2A). Because Sost mRNA levels after 7 days' differentiation are only barely detectable, we assessed the effect of sEphB4 treatment on Sost expression at 14 days' differentiation, where we observed a significant decrease at this time point. In contrast to the inhibition of these genes with sEphB4 treatment, the expression of RANKL mRNA was increased several-fold in primary calvarial osteoblasts (Fig. 2A). We also assessed mRNA levels of EphrinB2 and EphrinB1, as well as each receptor that these ephrins act through. None of the receptors were significantly modified, and of the two ligands, only EphrinB2 mRNA levels were downregulated by the addition of sEphB4 (Fig. 2B).

Figure 2.

Effects of 24 hours of sEphB4 treatment on osteoblast markers and ephrin family members in primary calvarial osteoblasts and Kusa 4b10 stromal cells. (A, B) sEphB4 effects on osteoblast markers and Ephrin family members in primary calvarial osteoblasts differentiated for 7 days or for 14 days in the case of sclerostin. Genes shown are Runx2, collagen type 1-α1 (Col1a1), alkaline phosphatase (Alpl), parathyroid hormone receptor type 1 (Pthr1), osteocalcin (Blgap1), dentin matrix protein 1 (Dmp1), matrix-extracellular phosphoglycoprotein (Mepe), sclerostin (Sost), and Rankl. (C) The Kusa 4b10 stromal cell line grown in osteoblast differentiation media was used as an alternative model. Data are shown from the days of osteoblastic differentiation where the maximal effect was detected. Values are mean ± SEM from triplicate wells; representative of three experiments.

Because calvarial osteoblast preparations are an impure population, we assessed the effect of sEphB4 on Kusa 4b10 cells. These were differentiated for 14 to 21 days before treatment to allow ready detection of late osteoblast/osteocyte markers. As observed in primary osteoblasts, Dmp1, Bglap, and Sost were significantly reduced in Kusa4b10 cells (Fig. 2B). The increase in RANKL mRNA induced by sEphB4 was only observed in Kusa 4b10 cells at day 2, suggesting that the increase in RANKL induced by sEphB4 may be restricted to less differentiated osteoblasts. Treatment with sEphB4 did not change mRNAs associated with earlier stages of osteoblast differentiation, for example, alkaline phosphatase, runx2, osx, coll1a1, PTHR1, in either cell type or at any time point (Fig. 2A and data not shown).

In view of the evidence from these and our earlier experiments10, 11 that ephrinB2/EphB4 receptor blockade impairs late stages of osteoblast differentiation and mineralized nodule formation in vitro, and that PTH treatment greatly enhances ephrinB2 production by osteoblasts in vitro and by bone in vivo,10 we next carried out an in vivo experiment aimed at determining whether receptor blockade influences the anabolic action of PTH, by studying effects of treatment of mice with PTH and sEphB4, each alone and in combination.

Micro-CT analysis of distal femoral samples confirmed a significant 13% increase in trabecular bone volume (BV/TV) with PTH treatment, as well as a significant increase in trabecular thickness (Tb.Th.) (Fig. 3A). Treatment with sEphB4 alone also significantly but more mildly increased both BV/TV and Tb.Th. Surprisingly, however, although combined treatment with PTH and sEphB4 still increased Tb.Th. significantly, trabecular number (Tb.N.) was significantly reduced and trabecular separation (Tb.Sp.) increased. The increase in BV/TV observed with PTH treatment alone was not observed with sEphB4 treatment. Thus, sEphB4 treatment impaired the anabolic action of PTH.

Figure 3.

Treatment with sEphB4 (10 mg/kg/3 × week) increases bone volume in femoral primary spongiosa and blunts anabolic effects of 4 weeks of PTH (30 µg/kg/5 × week) treatment in 8-week-old male C57Bl/6 mice. (A) Micro-CT analysis of distal femoral primary spongiosa. Trabecular bone volume (BV/TV) and trabecular thickness (Tb.Th.) were significantly greater in PTH and in sEphB4-treated mice. The combination of PTH and sEphB4 (sEphB4 + PTH) did not increase BV/TV but reduced trabecular number (Tb.N.) and increased trabecular separation (Tb.Sp). (B) Histomorphometry of the distal tibial metaphyseal secondary spongiosa confirmed the micro-CT result. Tb.Th was increased by PTH treatment, and sEphB4 + PTH combined treatment reduced BV/TV, Tb.Th, and Tb.N, while increasing Tb.Sp. Data are mean ± SEM (n = 8 per group). *p < 0.05, **p < 0.01, ***p < 0.001 compared with vehicle-treated animals. +p < 0.05, ++p < 0.01 compared with PTH-treated animals.

There was no significant effect of PTH treatment, sEphB4 treatment, or their combination on femoral cortical thickness or volume or marrow volume (data not shown). In vertebral samples, although PTH did not significantly increase BV/TV, sEphB4 + PTH significantly reduced BV/TV and Tb.N compared with the PTH-treated group (data not shown).

Analysis of a third site, the secondary spongiosa of the proximal tibia, by a second method, histomorphometry, revealed a similar effect of sEphB4 treatment. The anabolic effect of this PTH treatment protocol was less profound in this region, leading only to a significant increase in Tb.Th (Fig. 3B). There was no significant effect of sEphB4 treatment alone on trabecular structure in this region, but combined treatment of PTH and sEphB4 significantly reduced BV/TV and Tb.N and increased Tb.Sp compared with vehicle, and prevented the increase in Tb.Th associated with PTH treatment, thus confirming the suppression of PTH anabolic effect by ephrinB2/EphB4 blockade.

Histomorphometric indices of osteoblasts and osteoid production were significantly increased with PTH treatment alone, as previously observed with this dose in mice of a similar age.23 Number of osteoblasts (NOb.BPm), osteoblast surface (ObS/BS), osteoid surface (OS/BS), and osteoid volume (OV/BV) were all significantly elevated, with no modification in osteoid thickness (O.Th) (Fig. 4A). Dynamic markers of bone formation including mineral appositional rate (MAR), mineralizing surface (MS/BS), and bone formation rate (BFR/BS) were also significantly elevated by PTH treatment (Fig. 4A).

Figure 4.

Treatment with sEphB4 increases osteoblast numbers in the presence and absence of anabolic PTH but does not influence bone formation. (A) Osteoblast numbers (N.Ob/B.Pm), osteoblast surface (Ob.S/BS), osteoid surface (OS/BS), and osteoid volume (OV/BV) were all significantly increased by sEphB4 treatment and more strongly by PTH treatment. In mice treated with PTH + sEphB4, N.Ob/B.Pm was increased still further. No change was observed in osteoid thickness (O.Th). Despite the increase in N.Ob/B.Pm with sEphB4 treatment, dynamic markers of bone formation including mineral apposition rate (MAR), bone formation rate (BFR/BS), and mineralizing surface (dLS/BS) were not affected by sEphB4 treatment, whether in the presence or absence of PTH. Von Kossa/toluidine blue stained sections from representative samples, showing the secondary spongiosa for each treatment group; note the increased osteoid and osteoblast surface (white arrow) in the sEphB4-treated section and the increased osteoblast size in the sEphB4 + PTH-treated section (scale bar = 50 µm). (B) mRNA levels of mid-stage osteoblast markers were increased in femora collected 1 hour after the final PTH and sEphB4 injections confirm the increase in osteoblasts with sEphB4 and PTH treatment. Shown are runx2, alkalkine phosphatase (ALP), collagen 1-α1 (Col1a1) and parathyroid hormone receptor (PTHR1). Values are mean ± SEM. Treatment groups are as described in Fig. 2 (n = 8 per group). *p < 0.05, **p < 0.01, ***p < 0.001, compared with vehicle-treated animals. +p < 0.05, ++p < 0.01 compared with PTH-treated animals.

Treatment with sEphB4 alone significantly increased NOb.BPm, ObS/BS, and OS/BS (Fig. 4A). However, although sEphB4 treatment doubled the number of osteoblasts and extent of osteoblast surface, there was no significant increase in osteoblast activity as indicated by MAR, MS/BS, and BFR/BS (Fig. 4A), suggesting that the mineralization activity per osteoblast was reduced.

Although PTH treatment increased NOb.BPm to more than three times greater than baseline levels, combined treatment of PTH and sEphB4 increased the number of osteoblasts still further (Fig. 4A). In addition, the osteoblasts formed in mice treated with both PTH and sEphB4 appeared to be larger, a phenotype often associated with increased osteoid production (see images, Fig. 4A). However, as observed with sEphB4 treatment alone, sEphB4 did not augment the effect of PTH on mineral appositional or bone formation rates. This suggests that although sEphB4 can increase formation of cells recognizable by morphology and location as osteoblasts, it does not increase the mineralizing activity of these cells in the presence or absence of PTH.

Analysis of mRNA from femurs of the mice was consistent with the measured increase in osteoblast numbers; mid-stage markers of osteoblast differentiation were increased by both sEphB4 and PTH treatment. Thus, mRNA levels for runx2, Alp, and Pthr1 were significantly increased by each of PTH and sEphB4 treatment (Fig. 4B), and the effects of the two treatments were additive, except in the case of PTHR1. On the other hand, although sEphB4 treatment increased osteoblast numbers, it did not significantly modify late-stage osteoblast/osteocyte markers; there were no significant changes in mRNA for Dmp1, Mepe, Fgf23, or Sost in response to sEphB4 or PTH, either alone or in combination (Table 2). These findings suggest that with ephrinB2/EphB4 receptor blockade, osteoblast differentiation proceeds to a stage of increased numbers of cells with morphological features of osteoblasts and increased expression of mRNA levels for these cells, but late stages of differentiation are impaired.

Table 2. mRNA Levels of Osteoblast and Osteocyte Markers and Ephrin Family Members From Femora Collected 1 Hour After the Last PTH and sEphB4 Injections and Flushed of Marrow
Gene symbolVehiclePTHsEphB4PTH + sEphB4
  • Dmp1 = dentin matrix acidic phosphoprotein 1; Fgf23 = fibroblast growth factor 23; Mepe = matrix extracellular phosphoglycoprotein.

  • All data are shown relative to the housekeeping gene HPRT1 ± SEM.

  • Values for Osx, Dmp1, Fgf23, and Mepe are all ×103.

  • *

    p < 0.05;

  • **

    p < 0.01;

  • ***

    p < 0.001 compared with vehicle-treated animals;

  • +

    p < 0.05;

  • ++

    p < 0.01 compared with PTH-treated animals.

Bglap132.1 ± 5.347.1 ± 10.933.4 ± 5.651.6 ± 10.8
Sost0.171 ± 0.010.155 ± 0.0360.166 ± 0.020.195 ± 0.042
Osx16.23 ± 1.2718.96 ± 7.7027.30 ± 5.49*19.66 ± 3.23
Dmp111.95 ± 2.1230.06 ± 4.95**25.83 ± 11.6628.91 ± 3.75***
Fgf237.79 ± 0.956.72 ± 0.777.87 ± 2.688.54 ± 1.52
Mepe70.56 ± 10.1173.43 ± 8.4847.67 ± 10.0796.33 ± 20.75
ephrinB10.37 ± 0.050.31 ± 0.070.81 ± 0.19*0.64 ± 0.17
EphB10.016 ± 0.0020.016 ± 0.0010.027 ± 0.0060.026 ± 0.003** ++
EphB20.016 ± 0.0020.022 ± 0.0020.031 ± 0.006*0.017 ± 0.002
EphB30.05 ± 0.010.06 ± 0.020.24 ± 0.10**0.18 ± 0.06** +
EphB40.68 ± 0.050.97 ± 0.140.89 ± 0.100.72 ± 0.06

Although the accumulation of partially differentiated osteoblasts could contribute to the impaired anabolic action of PTH, the reduction in Tb.N suggested that changes in osteoclast number might provide an explanation. PTH treatment at this dose, as previously reported,23 did not significantly alter osteoclast surface (OcS/BS) or number (NOc/BPm). sEphB4 treatment alone caused a slight reduction in both parameters, but this was not statistically significant (p = 0.09). However, the combination of PTH and sEphB4 treatment significantly increased both OcS/BS and NOc/BPm (Fig. 5A). This could provide an explanation for both the reduction in Tb.N and the lack of anabolic action of PTH in the presence of sEphB4.

Figure 5.

Increased osteoclast formation in mice treated with a combination of sEphB4 and PTH. (A) Histomorphometric analysis shows increased number of osteoclasts (N.Oc/BPm) and osteoclast surface (OcS/BS) in mice treated with a combination of sEphB4 and PTH, although treatment with either agent alone did not significantly modify either of these parameters. Representative images of TRAP staining in the secondary spongiosa for each group (scale bar = 50 µm). (B) mRNA levels of osteoclast marker genes cathepsin K (Ctsk), DC-STAMP (Tm7sf4), osteoclast regulatory factors RANKL (Rankl), and OPG (Opg) and their ligand RANK (Rank) in whole femora collected 1 hour after the final PTH and sEphB4 injections. Values are mean ± SEM. Treatment groups are as described in Fig. 2 (n = 8 per group). *p < 0.05, **p < 0.01, ***p < 0.001 compared with vehicle-treated animals. +p < 0.05, ++p < 0.01 compared with PTH-treated animals.

To investigate the increase in osteoclast numbers with combined treatment of PTH and sEphB4 further, we assessed mRNA levels of osteoclast markers and factors that contribute to osteoclastogenesis in femora (Fig. 4B). DC-STAMP (Tm7sf4), a marker of osteoclast fusion, and Cathepsin K (Ctsk), an osteoclast marker, were both significantly increased in bones treated with sEphB4 alone. These were also increased by PTH treatment, but the combination of sEphB4 and PTH did not cause a further elevation. As previously reported in rats3, 25 and mice,23 mRNA levels for RANKL and OPG were both dramatically influenced by PTH treatment. In contrast, at the time point we assessed, treatment with sEphB4 alone did not modify either of these factors, either in the presence or absence of PTH (Fig. 5B). RANK mRNA levels were significantly elevated by combined treatment of sEphB4 and RANKL (Fig. 5B).

To determine the effects of sEphB4 on osteoclast formation, cell culture systems were used. Since osteoclasts and their precursors express ephrinB2, and RANK levels were increased in vivo,10, 17 we first determined whether sEphB4 might influence osteoclast formation from bone marrow macrophages treated with RANKL and M-CSF, but sEphB4 did not stimulate osteoclast formation in these cultures (Fig. 5A). This suggests that the stimulation of osteoclast formation by sEphB4 is not mediated by inhibition of ephrinB2 signaling within the osteoclast lineage and consistent with the finding that EphB4 is not produced by osteoclasts.10, 17

In a co-culture system of osteoclast precursors with calvarial osteoblastic cells, sEphB4 treatment significantly increased the formation of osteoclasts stimulated by 1,25-dihydroxyvitamin-D3 (1,25D) alone or in combination with prostaglandin E2 (PGE2) (Fig. 6A), indicating the dependence of the pro-osteoclastic effect of ephrinB2 on the presence of osteoblasts/stromal cells. A similar effect was observed with PTH (number of TRAP + ve multinucleated cells/well: sEphB4 alone: 0; PTH alone: 1 ± 0.3; PTH + sEphB4: 52 ± 8) and when an alternative inhibitor of EphrinB2/EphB4 interaction was used (TNYL-RAW) in combination with 1,25D/PGE2 (number of TRAP + ve multinucleated cells/well: TNYL-RAW: 0; 1,25D/PGE2: 88 ± 6; 1,25D/PGE2 + TNYL-RAW: 138 ± 28, p < 0.05).

Figure 6.

sEphB4 stimulates production of RANKL and other osteoclastic factors and in the presence of 1,25D3, stimulates NFκB activation and osteoclast formation in co-culture systems. (A) Formation of tartrate-resistant acid phosphatase-positive multinucleated osteoclast-like cells (TRAP + MNCs) from C57Bl/6 bone marrow macrophages by RANKL/M-CSF treatment was not modified by sEphB4 treatment. However, in the presence of osteoblasts (O'blasts) and 1,25-dihydroxyvitamin-D3 (1,25D3), osteoclast formation was further enhanced by 7.5 µg/mL sEphB4. Formation of TRAP + MNCs was not observed with sEphB4 treatment alone in either culture system. Data are mean + SEM, representative of two independent experiments performed with six replicates. ***p < 0.001 versus nontreated sEphB4 control. (B) RANKL mRNA (Rankl) levels were induced after 24 hours of sEphB4 treatment in primary calvarial osteoblasts differentiated for 7 days and enhanced the PTH- and 1,25-dihydroxyvitamin-D3 (1,25D) effects on Rankl expression. (C) Long-term induction of Rankl, IL-6 (Il6), and OSMR (Osmr) but not OPG (Opg) in Kusa 4b10 stromal cells and primary calvarial osteoblasts treated with 7.5 µg/mL sEphB4 for 2, 7, 14, and 21 days. (D) NFκB induction by 1,25D3 in the presence and absence of sEphB4 using NFκB-luciferase expressing RAW264.7 cells co-cultured with Kusa4b10 cells. Values are mean ± SEM; all data shown are representative of at least three independent experiments. Western blotting for RANKL protein in primary calvarial osteoblasts was differentiated for 14 days and treated with sEphB4 in the presence and absence of 1,25D3 for 6 hours and pan-actin loading control; blot shown is representative of two experiments.

Since sEphB4 treatment induced RANKL expression in primary calvarial osteoblasts (Fig. 1), we then determined whether sEphB4 would have this effect in the presence of other osteoclastic stimuli. RANKL mRNA levels were increased by sEphB4 over and above the effects of 1,25D and PTH (Fig. 6B), again demonstrating a role for the ephrinB2/EphB4 interaction within the osteoblast lineage. OPG mRNA levels were not significantly modified by sEphB4 treatment (data not shown).

In addition to the increase in RANKL in both Kusa4b10 cells and in primary calvarial osteoblasts, IL-6 (Il6) and oncostatin M receptor (Osmr) mRNA levels were also stimulated by sEphB4 receptor blockade over the time course of osteoblast differentiation (Fig. 6C). Furthermore, we assessed NFκB activation using RAW264.7 cells stably expressing NFκB-luciferase and co-cultured with Kusa4b10 cells. This activity was significantly enhanced by 1,25D treatment and increased further when sEphB4 was added to the cultures (Fig. 6D). Similar results were obtained when calvarial osteoblasts were used in the co-cultures (not shown). These data indicate a general increase in the osteoclastogenic profile of osteoblasts when ephrinB2/EphB4 signaling is blocked by sEphB4 in the presence of 1,25D.

Western blotting of primary calvarial osteoblasts treated with sEphB4 and 1,25D detected a mild increase in RANKL protein expression (Fig. 6D), but this was not detected when cells were treated with sEphB4 alone. In fact, the reverse was seen, suggesting that although sEphB4 treatment on its own increases the production of RANKL mRNA, the protein is either quickly degraded or is not translated effectively, providing a possible explanation for the lack of effect of sEphB4 alone on osteoclast differentiation.


The decision to undertake the in vivo study reported here was based on our earlier findings that PTH treatment substantially increased ephrinB2 production by osteoblasts in vitro and in bone in vivo, and that blockade of the interaction of ephrinB2 with EphB4 resulted in delayed late stages of differentiation of osteoblasts.10, 11 Phosphorylation of EphB2 and EphB4 in response to ephrinB2-Fc treatment and ephrinB2 phosphorylation in response to EphB4-Fc in cultured osteoblasts confirms that the interactions of these family members reported in other cell types16 can take place within the osteoblast lineage. This is of note because osteoblasts work in teams and are known to depend on cell-cell contact for their function both during bone formation and when embedded as osteocytes within the bone matrix.

The in vivo findings we report here show that blockade of EphrinB2/EphB4 signaling results in an increased number of osteoblasts and increased expression of early osteoblast marker genes. Despite this enhanced osteoblast formation, there was no increase in mRNA levels associated with late-stage osteoblasts and osteocytes, nor any increase in the extent or rate of mineralization. Furthermore, although sEphB4 doubled the number of osteoblasts present on trabecular bone surfaces, BV/TV was only mildly elevated, and this was only significant at the femoral site. This result suggests sEphB4 impairs osteoblast differentiation near the stage of alkaline phosphatase expression. sEphB4 treatment increased osteoblast numbers just as PTH did, but unlike PTH did not increase bone formation rate. Thus, relative to the number of osteoblasts, the level of mineralization was reduced with sEphB4 treatment. Similarly, in vitro, where cell numbers are maintained at a comparable level, sEphB4 treatment reduced late osteoblast markers and inhibited mineralized nodule formation10 without influencing osteoblast proliferation or affecting early markers of osteoblast differentiation. Together this provides evidence that ephrinB2/EphB4 interaction within the osteoblast lineage is a control mechanism required for continuation of late stages of osteoblast differentiation. The data suggest an accumulation of partially differentiated osteoblasts on bone surfaces that have reached a “checkpoint” in differentiation where interaction of ephrinB2 with EphB4 is required to progress to the next stage of late osteoblast/osteocyte differentiation.

EphrinB2/EphB4 signaling has been implicated in the processes of cell-cell adhesion,30 cell attachment,31 and cell migration.32 The impairment in late osteoblast differentiation might relate to changes in these processes in the osteoblast lineage. Cell-cell adhesion is an important component of osteoblast action and is required for full osteoblastic activity, for example, through gap junctions.33 We have reported that ephrinB2/EphB4 interactions play a role in adhesion of human mesenchymal stem cells to culture dishes;31 this influence may also be required, not only for effective integration of orthopedic implants but also for efficient bone formation during the process of remodeling. Migration of osteoblasts is regulated by a wide range of factors and is required for bone remodeling, fracture healing, implant fixation, and is likely to play a role in the pathogenesis of osteosarcoma; ephrinB2/EphB4 signaling may also play a role in these actions.

Despite the decrease in transcripts associated with late osteoblast/osteocyte differentiation when ephrinB2/EphB4 interaction was blocked, this same treatment increased production of mRNA for RANKL, most dramatically in the least differentiated stromal cells. The significance of this observation became clearer when the in vivo results were obtained. The most striking finding was that the combined treatment of PTH and sEphB4 resulted in trabecular bone loss, not necessarily because of the impaired osteoblast differentiation but because of increased osteoclast formation. Thus, ephrinB2/EphB4 receptor blockade converted the PTH anabolic response to a resorptive one in vivo. This is, in some respects, similar to an effect we observed in oncostatin M receptor (OSMR) knockout mice, in which the anabolic response to PTH was converted to a catabolic one by a sustained elevation in osteoblastic RANKL expression.23 In this situation, where sEphB4 is administered, the mechanism is somewhat different: the restriction on RANKL production appears to be lifted, leading to a more substantial increase in osteoblastic production of RANKL and of RANKL/RANK-dependent NFκB reporter activation, and elevated osteoclast formation. Although the in vitro data clearly indicated an increase in RANKL production and activity, this was not detectable in vivo, which may be limited by the time point we analyzed (1 hour after last injection).

Increased osteoclast formation and RANKL protein levels were observed only when sEphB4 was combined with an osteoclastic stimulus such as PTH or 1,25D. Interestingly, when sEphB4 was administered alone, while mRNA levels of RANKL were high, RANKL protein was reduced, a possible explanation for the lack of effect of sEphB4 treatment alone on osteoclast formation.

RANKL was not the only osteoclastogenic factor induced by sEphB4 treatment of osteoblasts: increased transcripts of IL-6 and OSMR were also observed, suggesting that a number of factors could contribute to the enhanced osteoclast formation induced by sEphB4 in combination with PTH or 1,25D. This increase in osteoclastogenic factors with sEphB4 treatment contrasts with the downregulation of late osteoblast markers in the osteoblast cultures. It is possible that when osteoblasts are prevented from maturing to their late stages, their production of osteoclastogenic factors is stimulated. Alternatively, the impairment in osteoblast differentiation may lead to their accumulation at a stage of differentiation that is particularly supportive of osteoclast formation, with high expression of RANKL, IL-6, and OSMR. Notably, all of these osteoclastogenic stimuli were increased most dramatically by sEphB4 in undifferentiated Kusa 4b10 stromal cells. This concept contrasts with recent reports suggesting that it is the osteocytes that predominantly support osteoclast formation,34, 35 but is quite consistent with earlier work indicating that early osteoblasts can provide key support for osteoclast formation.36, 37

In addition to the increased support of osteoclastogenesis by osteoblasts in response to sEphB4, increased RANK expression was detected in the bones collected from the in vivo study. This may relate to increased presence of osteoclast precursors and/or blockade of the reverse signaling action of ephrinB2 to inhibit osteoclast formation17 and this question continues to be investigated. Recent work has suggested that reverse signaling through ephrinB1 within the myeloid lineage also limits osteoclast differentiation,38 but inhibition of ephrinB1 is unlikely to play a role in this study because of the specificity of sEphB4 and TNYL-RAW actions. Binding of EphB4 and ephrinB1 has been either not detectable39 or, when detected, occurs at an affinity more than 100 times less than that between EphB4 and ephrinB2.40 In addition, although ephrinB1 mRNA levels were elevated in the in vivo experiment, this was not observed in cultured calvarial osteoblasts. Given the recapitulation of the influence of sEphB4 on osteoclast formation and osteoblast differentiation in vitro in the absence of any change in ephrinB1 transcript levels, this seems unlikely to be of mechanistic importance to our observations. Indeed, if the increased osteoclast formation observed in vivo related to an influence on ephrinB1, it is a reduction, not an increase, that might be expected.

The use of EphB4 inhibition to reduce tumor growth and angiogenesis has been shown to be effective in a number of animal models, including melanoma41 and breast cancer.18 In addition, sEphB4 impaired angiogenesis in models of retinopathic vascularization42 and choroidal neovascularization.43 Inhibition of vascularization by sEphB4 may also contribute to the impaired anabolic action of PTH. However, because reduced mineralization and late-stage osteoblast mRNA expression10 and increased support of osteoclastogenesis have both been detected in isolated osteoblasts treated with sEphB4, an effect on vascularization is not required. The enhanced osteoclast formation induced by sEphB4 in response to several stimuli of osteoclast formation in this study reveals an action of sEphB4 that warrants investigation in the case of its use in treatment of tumors that might be prone to bone metastasis,18 a process that involves the stimulation of osteoclast formation by PTHrP and other cytokines.44

In conclusion, we report that the interaction between ephrinB2 and EphB4 within the osteoblast lineage plays a key role in regulating osteoblast maturation: Inhibition of this pathway appears to block late stages of osteoblast differentiation. In addition, inhibition of ephrinB2/EphB4 interactions within the osteoblast lineage stimulates production of RANKL and other osteoclastic factors, leading, in the presence of PTH or 1,25D, to a further enhancement in osteoclast formation. Such effects are consistent with an important role for ephrinB2/EphB4 interaction within the osteoblast lineage in the process of bone remodeling.


All authors state that they have no conflicts of interest.


The authors acknowledge the excellent technical assistance of Narelle McGregor, Ingrid Poulton, and Joshua Johnson, and thank the St. Vincent's Bioresources Centre staff for excellent animal care. We thank V Krasnoperov for helpful advice and supply of sEphB4, and JM Quinn for advice on the NFκB-luciferase assay. The work was supported by NHMRC (Australia) Project Grant 620600 and supported in part by the Victorian Government OIS program; NAS is supported by an NHMRC (Australia) Senior Research Fellowship. FT is supported by a University of Melbourne Melbourne Research Scholarship, a Melbourne International Fee Remission Scholarship, and a St. Vincent's Institute Foundation Scholarship. ST is supported by an NHMRC (Australia) Peter Doherty Fellowship. EKB is supported by a Cure Cancer Australia Fellowship. The Osteomeasure system was purchased with the generous support of the Jack Brockhoff Foundation and purchase of the microCT system was partially supported by the Potter Foundation.

Authors' roles: Study design: NAS and TJM. Study conduct: FT, ST, BCI, PWMH, and EKB. Data collection: FT, ST, BCI, and PWMH. Data analysis: FT, ST, BCI, PWMH, TJM, and NAS. Drafting manuscript: NAS and TJM. Revising manuscript: FT, ST, BCI, EKB, TJM, and NAS. Approving final version of manuscript: all authors. NAS takes responsibility for the integrity of the data analysis.