Drs Wei and Onyia are employees of Eli Lilly and Co. Dr Häusler is now an employee of Sanofi-Aventis but at the time of conducting the work had no conflict of interest. All other authors state that they have no conflicts of interest.
Published online on March 24, 2008;
With the aim of identifying new pathways and genes regulated by PTH(1–34) and PTH-related protein 1–141 [PTHrP(1–141)] in osteoblasts, this study was carried out using a mouse marrow stromal cell line, Kusa 4b10, that acquires features of the osteoblastic phenotype in long-term culture conditions. After the appearance of functional PTH receptor 1 (PTHR1) in Kusa 4b10 cells, they were treated with either PTH(1–34) or PTHrP(1–141), and RNA was subjected to Affymetrix whole mouse genome array. The microarray data were validated using quantitative real-time RT-PCR on independently prepared RNA samples from differentiated Kusa 4b10, UMR106 osteosarcoma cells, and primary mouse calvarial osteoblasts, as well as in vivo using RNA from metaphyseal bone after a single PTH injection to 3-wk-old and 6-mo-old ovariectomized rats. Of the 45,101 probes used on the microarray, 4675 were differentially expressed by ≥1.5 fold, with a false discovery rate <0.1. Among the regulated genes, ephrinB2 mRNA was upregulated in response to both PTH and PTHrP. This was confirmed by quantitative real-time PCR in vitro and in vivo. Increased ephrinB2 protein was also shown in vitro by Western blotting, and immunostaining of femur sections showed ephrinB2 in both osteoclasts and osteoblasts. Production of ephrinB2, as well as other ephrins or Eph family members, did not change during differentiation of Kusa 4b10 cells. Blockade of ephrinB2/EphB4 interaction resulted in inhibition of mineralization of Kusa 4b10 cells. Together with the shown effect of ephrinB2 promoting osteoblast differentiation and bone formation through action on EphB4, the data raise the possibility that PTH or PTHrP might regulate ephrinB2 to act in a paracrine or autocrine manner on EphB4 or EphB2 in the osteoblast, contributing as a local event to the anabolic action of PTH or PTHrP.
Bone remodeling continues throughout life and is necessary for the structural maintenance and repair of the skeleton and for calcium homeostasis. Circulating hormones exert controlling influences, largely through locally generated cytokines and growth factors that transfer information among osteoblasts, osteoclasts, immune cells, and the bone matrix. Bone remodeling consists of coupled processes of bone resorption and formation. The sequence of events is initiated asynchronously throughout the skeleton at sites that are geographically and chronologically distinct. The process begins with resorption of a quantum of bone followed by new bone formation by osteoblasts, the quantity of new bone matching that of the resorbed bone when under steady-state conditions. These processes occur in the same place in bone multicellular units (BMUs) so that there is no change in the shape of bone. Bone modeling, on the other hand, makes up that process of bone formation taking place on surfaces that have not undergone prior resorption. The importance of local control is amply shown through the remarkable number of skeletal phenotypes that have been recognized and analyzed in genetically manipulated mice engineered to under- or overexpress cytokines and their receptors.
A hormone critical to bone homeostasis is PTH, which is secreted by the parathyroid when serum calcium levels drop and stimulates calcium release from bone by increasing bone resorption. However, transiently high levels of PTH resulting from daily subcutaneous injection has a strongly anabolic effect on bone. This anabolic action may reflect the action of locally produced PTH-related protein (PTHrP), which is critical for maintenance of local bone mass and which can also bind the common PTH/PTHrP receptor (PTHR1) found in osteoblasts. One way that PTH may exert anabolic and catabolic actions is through local bone remodeling pathways. It is now well understood that cells of the osteoblast lineage control the formation and activity of osteoclasts. This is achieved through regulated osteoblastic expression of the TNF ligand family member, RANKL, that is essential for the formation and activation of osteoclasts, and its decoy receptor osteoprotegerin (OPG), which is required for osteoclast inhibition. Much less is known of communication in the reverse direction (i.e., from osteoclast to osteoblast). A local coupling factor linking bone resorption to subsequent formation has long been proposed to explain the regulation and coordination of bone formation in the BMUs. Matrix-derived TGFβ and IGF-I and -II released by osteoclast action have been proposed to contribute, but it is likely that coupling factor activity also emanates from the active osteoclast itself within the BMUs. One recently proposed coupling factor is ephrinB2. Studies using genetically manipulated mice provided evidence that osteoclast-derived ephrinB2 can act through its receptor, EphB4, in osteoblasts to promote osteoblast differentiation and that reverse signaling by osteoblast-derived EphB4 can suppress the formation of osteoclast precursors in a contact-dependent process.
This study was undertaken in the course of studies of the anabolic action of PTH and PTHrP. A mouse bone marrow stromal cell line, Kusa 4b10, was differentiated to form pre-osteoblasts that express functional receptor for PTH/PTHrP (PTHR1). The aims of this study were to identify novel genes induced by these treatments under these conditions and to compare the effects of PTH(1–34) and PTHrP(1–141) in the gene responses evoked in these cells. In using a mouse whole genome cDNA microarray to assess the responses to PTH and PTHrP in these cells, we identified members of the ephrin and Eph family as targets of these proteins. We report that, whereas production of ephrinB2, EphB2, and EphB4 is not altered during in vitro osteoblast differentiation in culture, ephrinB2 production is regulated in a dose-dependent manner in osteoblasts through PTHR1. This raises the possibility that, in addition to any coupling action of osteoclast-derived ephrinB2, this protein might act in a paracrine or autocrine manner on the osteoblast itself under the influence of local PTHrP or administered PTH to stimulate osteoblast maturation and/or bone formation.
MATERIALS AND METHODS
Synthetic human PTH(1–34) was purchased from Bachem (Bubendorf, Switzerland), and recombinant human PTHrP(1–141) was prepared as previously described. Affinity purified anti-mouse ephrinB2, EphB2, and EphB4 were obtained from R&D Systems (Minneapolis, MN, USA), and Pan actin was from Laboratory Vision. Ascorbate, β-glycerophosphate, Alizarin, and IBMX were from Sigma Aldrich. The peptide antagonist of ephrinB2 interaction with its receptor, EphB4 (TNYLFSPNGPIARAW), was synthesized by Auspep (Parkville, Australia).
Kusa 4b10 cells, a clonal line derived from Kusa O cells, were maintained as described. In experiments in which differentiation of Kusa 4b10 cells was required, cells were subcultured at a density of 3000 cells/cm2 in αMEM + 10% FBS. After 3 days, the medium was aspirated and replaced with osteoblast differentiating medium (αMEM + 15% heat-inactivated FBS + ascorbate 50 μg/ml) and thereafter changed three times per week. For mineralization studies, 10 mM β-glycerophosphate was added to osteoblast differentiation media. For use in the microarray study or for preparation of independent RNA samples in validation experiments, FBS was reduced to 2% on day 16, and cells were treated with PTH(1–34) (10 nM) or PTHrP(1–141) (20 nM) on day 17, by which time the Kusa 4b10 cells have functional PTHR1 receptors. UMR 106–01 clonal osteogenic sarcoma cells were used as described, and mouse calvarial osteoblasts were prepared by enzymatic digestion of bones from 1- to 3-day-old C57Bl/6 pups. Cells for preparation of RNA for microarray were collected at 1, 6, and 24 h after treatments were applied. Triplicate cultures of Kusa 4b10 and UMR 106–01 cells were used, triplicate real-time PCR estimations were carried out, and single calvarial osteoblast cultures were examined.
Gene expression profiling by DNA microarray analysis
RNA was prepared from Kusa 4b10 cells lysed in Trizol (Invitrogen) according to the manufacturer's instructions. RNA was treated with DNase (Promega) for 30 min at 37°C and purified according to the Qiagen protocol for the RNeasy MiniElute kit (Qiagen). Total RNA was quality tested using the Agilent Bioanalyser 2100 NanoChip protocol. Five micrograms of total RNA was labeled using the Affymetrix One Cycle cDNA synthesis kit (Millenium Sciences), and subsequent cDNA was purified using the Affymetrix GeneChip Sample Cleanup kit (Millenium Sciences). In vitro transcription of cDNA to cRNA was performed using the Affymetrix IVT labeling kit (Millenium Sciences) with biotin-labeled ribonucleotides to incorporate biotin into the cRNA. cRNA was purified using the GeneChip Sample Cleanup kit and quality tested using the Agilent Bioanalyser 2100 NanoChip protocol. Twenty micrograms of labeled cRNA was fragmented (50–200 bp) and quality control checked using the Agilent Bioanalyser 2100 NanoChip protocol before preparing the probe cocktail containing 0.05 μg/ml cRNA in hybridization buffer. Two hundred microliters of each sample were loaded onto Affymetrix Murine 430 version 2.0 GeneChips and hybridized at 45°C for 16 h in an oven with a rotating wheel at 60 rpm. After hybridization, the chip was washed and scanned using the Affymetric GeneChip Scanner 3000 and the chip signal converted into a DAT file by the scanner operating software GCOS.
Hybridized microarrays were analyzed by Affymetrix microarray analysis software microarray suite 5 (MAS 5). The chip signals were normalized using the default method in Affymetrix MAS5. To identify differentially expressed (DE) genes, an ANOVA model was fitted on each of the probe sets on the chip at each time point. The intensity value, Affymetrix MAS5 signal, of a particular gene was modeled as
where yki is the signal of the ith batch of cells in either control or one of the two treatments k. μk is the group mean for the treatment k, and ϵki is the measurement errors (chip-to-chip variations) assumed to follow an identical normal distribution N(0, σ). Because thousands of hypotheses were tested simultaneously, the issue of multiplicity is a major concern. To control the false-positive rate, false discovery rate (FDR) was used to adjust the p values. Specifically, for each pairwise comparison derived from the above ANOVA model, the FDR procedure is separately applied. Genes were called differentially expressed if their probe(s) passed all three independent filters, namely one of the comparison group average larger than the average chip median signals that is 345, FDR <0.1 and absolute fold change >1.5. Identified DE genes regulated at one or more time points by PTH or PTHrP were analyzed separately using Ingenuity Pathway Analysis tools (Ingenuity Systems). Resultant functional groups with significantly enriched DE genes were compared between PTH and PTHrP.
Real time RT-PCR
RNA was prepared from cells lysed in Trizol (Invitrogen) according to the manufacturers' instructions followed by treatment with DNase (Promega) (30 min at 37°C) and reprecipitation. RNA was quantified by measuring absorbance at 260 nm and quality assessed by the 260/280 nm absorbance ratio. cDNA was prepared using random hexamers (Promega) and Superscript III (Invitrogen) according to the manufacturer's protocol. Primers were as described, and are specified in Table 1. Real-time RT-PCR was performed using the Stratagene MX3000 (Invitrogen) and relative quantities were obtained by generating a standard curve for each primer pair. The expression levels of all genes were normalized to HPRT1. Expression of HPRT1 mRNA was not altered by any of the treatments in the study.
Table Table 1.. Real-Time Rt-Pcr Primers
Western blot analysis and immunoprecipitation
Lysates of differentiated Kusa 4b10 cells were prepared in PLC lysis buffer containing protease inhibitors (Sigma), sonicated briefly, and centrifuged for 10 min at 4°C and 13,000 rpm to remove cell debris. Twenty-five micrograms of protein was heated to 95°C for 5 min in sample buffer and separated through precast polyacrylamide gels (10%; Invitrogen) under reducing conditions and blotted onto PVDF membrane. After blocking in TBS + 5% milk powder, membranes were probed with primary antibody overnight at 4°C, washed, and incubated with secondary antibody for 1 h at room temperature and immunoreactive bands detected by enhanced chemiluminescence (ECL plus kit; Amersham). Filters were stripped and reprobed with Pan actin antibody as a measure of protein loading.
Hind legs from 3-wk-old male rats were fixed in 4% paraformaldehyde/PBS, decalcified, embedded in paraffin, and used for immunohistochemistry as previously described with the following modifications. Endogenous peroxidase was inhibited with 1.7% H2O2 (Merck) in 100% methanol for 30 min, and sections were permeabilized with 1% sodium borohydride (Sigma, St Louis, MO, USA) in PBS followed by 15 min in 0.1% trypsin (JRH). Nonspecific staining was blocked with 1% BSA (Sigma), 0.05% Tween 20 (BDH, Poole, UK), and 5% normal rabbit serum in PBS for 30 min. Sections were incubated overnight with 50 μg/ml goat anti-mouse ephrinB2 followed by 0.5 μg/ml biotinylated swine anti-goat IgG (DakoCytomation, Glostrup, Denmark) in blocking solution for 45 min and 1.6 mg/ml streptavidin-horseradish peroxidase (DakoCytomation) in PBS for 45 min, and positive cells were detected with 0.5 mg/ml diaminobenzidine and 0.006% H2O2 in PBS (Sigma).
In vivo experiments
All animal experiments were approved by the St Vincent's Health Animal Experimentation Committee. Three-week-old female Sprague-Dawley rats (45–50 g) and 6-mo-old female Sprague-Dawley rats that had been ovariectomized at 5 mo of age were randomly allocated to control or PTH treatment groups. Rats were given a single injection of vehicle or PTH(1–34) (30 μg/kg) in 2% heat-inactivated rat serum in 0.9% saline. At the indicated times, animals were anesthetized, and tibias were removed. The distal epiphysis including the growth plate was removed from each femur, and a subjacent 5-mm band of the metaphyseal primary spongiosa was resected and snap-frozen in liquid nitrogen. RNA was prepared from individual samples using a QIAGEN RNeasy Lipid Midi Kit after homogenization in Ultraspec-II reagent (Biotecx, Houston, TX, USA) using an LS 10–35 Polytron homogenizer (Brinkmann Instruments, Westbury, NY, USA) as recommended by the manufacturer. RNA was quantified by measuring absorbance at 260 nm, and the quality was assessed by agarose gel electrophoresis for the integrity of 18S and 28S ribosomal RNA bands and by the 260/280 nm absorbance ratio. cDNA was prepared as above and amplified by real-time RT-PCR.
Of the 45,101 probes used on the microarray carried out on the differentiating Kusa 4b10 cells, 4675 were differentially expressed by ∼1.5 fold on treatment with either PTH or PTHrP. When the gene expression profiles were analyzed, data for each time point were analyzed to identify those genes undergoing increased or decreased expression common to both treatments, as well as identifying changes in gene expression at each time that were unique to one or other treatment. The data are summarized in Fig. 1A, showing how many genes were significantly changed at each time point in the overlapped regions of the Venn diagram, and the total of all time points, by both PTH and PTHrP, and separately the number of changes specific for either treatment. The largest changes were found at the 6-h time point, where 2382 and 1546 genes were significantly changed by PTH and PTHrP, respectively, among which 1179 genes were significantly changed by both treatments, 1203 genes changed specifically by PTH, and 367 genes specifically by PTHrP. Using the bioinformatic criteria described, significant treatment differences are recognized at any of three time points. Thus, any differences between the two treatments could reflect either effects on different genes or, importantly, different time courses of responses to the two agonists.
When the differentially expressed genes regulated at each time point by PTH and PTHrP were examined by ingenuity pathway analysis, most of the significantly enriched functional groups were similarly affected by PTH and PTHrP, with the exception of two groups: protein synthesis and molecular transport. The top 10 functional groups with significantly enriched differentially expressed genes by both PTH and PTHrP are shown in Fig. 1B, with the results for protein synthesis and molecular transport shown at the right of Fig. 1B. Detailed data for these and the remaining functional groups are provided as supplementary data online (Table S1).
When the data were examined for novel genes influenced by action on the PTHR1, and genes were functionally categorized according to their gene ontology, changes were noted in expression of members of the ephrin/Eph gene family. Figure 1C shows the array data from probe sets for several of the ephrins and their receptors, Ephs, in which significant expression levels were reached, according to the bioinformatic criteria outlined in the Materials and Methods section. Only PTH(1–34) data are shown because PTHrP(1–141) responses did not differ significantly from those of PTH(1–34). The most obvious effect noted is stimulation of expression of mRNA for ephrinB2, evident in all of five probesets, and the few other effects on ephrins or Ephs were minor. Small inhibitory effects on expression of ephrinB1, ephrinA5, EphA2, and EphB6 (Fig. 1C) were noted with some probe sets, but no other member of the ephrin/Eph family reached significance, including EphB2 and EphB4. Supplementary information online (Table S2) shows the array data from all the probe sets in the array that represented members of the Eph and ephrin families, including those with signals below the cut-off limit.
In the light of these findings, and because the filtering parameters for the array data were set at fold differences of 1.5, it was important that validation experiments be carried out on all those members of the ephrin/Eph family represented on the microarray that showed significant responses. In gene validation studies, we used real-time RT-PCR to determine changes in expression of ephrins and Ephs, in addition to a selection of other genes shown in the array to respond to both treatments. Gene validation was carried out on RNA samples prepared from independent experiments in which differentiated Kusa 4b10 cells, primary mouse calvarial cells, and UMR 106–01 rat osteogenic sarcoma cells were treated with maximally effective concentrations of PTH(1–34) or PTHrP(1–141). Only PTH(1–34) data are shown, but very similar responses were obtained with PTHrP(1–141) in all cases. Figure 2 shows the responses to PTH at 1, 6, and 24 h obtained by microarray analysis of differentiated Kusa 4b10 cells (A), RNA from independently treated differentiated Kusa 4b10 cells (B), primary mouse calvarial osteoblast-like cells (C), and UMR 106–01 cells (D). The responses of a range of differentially regulated genes identified in the Kusa 4b10 array analysis (RGS2, vitamin D receptor [VDR], RANKL, interleukin 6 [IL-6], and leukemia inhibitory factor) were confirmed using independent RNA samples, as well as RNA from UMR106–01 and mouse calvarial osteoblasts. IL-18, production of which was noted as increased by PTH in a previous microarray study in UMR 106–01 cells, was not affected by treatment in the Kusa 4b10 array, the independent RNA samples, or in RNA from mouse calvarial cells, but was confirmed as an increase in UMR 106–01 cells. Similarly, PTH enhanced amphiregulin mRNA levels in UMR 106–01 cells, and we confirmed this in our experiments when Kusa 4b10, primary mouse calvarial osteoblast-like cells, or UMR 106.01 cells were treated with PTH (Figs. 2B–2D); however, the one probe set for amphiregulin in the gene array showed no response (Fig. 2A).
The substantial ephrinB2 response was confirmed in each of the cell types used for validation studies. The small effects noted in the microarray on ephrinB1, ephrinA5, EphA2, and EphA4 were not confirmed in any of the subsequent validation experiments (data not shown). Although neither EphB4 nor EphB2 showed any significant response in the microarray data (Table S2; Fig. 1C), small stimulatory effects on EphB2 were noted consistently in validation experiments in Kusa 4b10 cells. Western blotting with antibodies against ephrinB2, EphB2, and EphB4 showed that treatment resulted in an increase in ephrinB2 protein evident at 6 h and sustained up to 36 h, with no detectable effect on either EphB2 or EphB4 protein (Fig. 3).
Dose–response studies were carried out after preliminary experiments to determine the time course of responses. The increase in production of mRNA for ephrinB2 is dose-dependent in Kusa 4b10 cells (Fig. 4A), UMR106–01 cells (Fig. 4C), and mouse calvarial osteoblasts (Fig. 4D), and the dose response to PTH is shifted to the left when treatment is carried out in the presence of a low dose of the phosphodiesterase inhibitor, IBMX (Figs. 4A and 4C). The plant diterpene stimulator of adenylyl cyclase, forskolin, also stimulated ephrinB2 production in a dose-dependent manner, with its response curve shifted to the left by IBMX in a similar manner (Fig. 4B). Because the gene validation studies suggested that there are small effects on expression of EphB2, this was examined also in these experiments, and small but reproducible increases in EphB2 mRNA were noted (data not shown). Repeated validation experiments in each of the cell types did not establish any consistent effect of treatment on mRNA for EphB4. It should also be noted that no effect on ephrinB2 expression was obtained in response to 1,25-dihydroxyvitamin D3 or with IL-11 (data not shown). These were examined because they have in common with PTH and PTHrP the ability to promote production of RANKL and hence osteoclast formation. The effect of PTH and of PTHrP promoting production of mRNA for ephrinB2 is a transcriptional one, based on its inhibition by actinomycin D. In mouse calvarial osteoblasts, the 5.5 ± 0.14–fold increase in mRNA for ephrinB2 obtained after 1-h PTH treatment was completely prevented by pretreatment of the cells with actinomycin D (data not shown).
We next examined the expression of mRNA for the Ephs and ephrins during osteoblast differentiation. Although ephrinB2 was clearly regulated by PTH, its expression as well as that of EphB2 and EphB4 was not altered during osteoblast differentiation in Kusa 4b10 cells from days 0 to 21 (Fig. 5A). As controls, we examined the expression of osteocalcin, alkaline phosphatase ALP), and PTHR1. Expression of each was enhanced during the differentiation process (Fig. 5A), whereas SDF-1 showed a near-maximal mRNA expression by day 4 of osteoblast differentiation (Fig. 5A). Data from Fig. 5A were obtained by duplicate estimations on individual culture dishes. The pattern of change in gene expression was maintained in several repeated experiments.
To determine whether there might be any functional consequence of interfering with ephrinB2 interaction with its receptor EphB4, within a population of osteoblastic cells, a specific peptide inhibitor of the ephrinB2/EphB4 interaction was used. Treatment of Kusa 4b10 cells with this peptide (TNYLFSPNGPIARAW, abbreviated as TNYL) resulted in dose-dependent inhibition of mineralization (Fig. 5B) and inhibition of expression of mRNA for some genes associated with osteoblast differentiation (Fig. 5C). Those whose expression was inhibited tended to be genes associated with relatively late stages of differentiation in Kusa 4b10 cells, including dentine matrix protein-1 (DMP-1), sclerostin, and interferon-induced transmembrane protein 5 (Ifitm5), which has been noted to decrease in associated with inhibited mineralization of osteoblasts in vitro.
In view of the confirmed effects of PTH and PTHrP on ephrinB2 expression in vitro, in vivo experiments were carried out, in which femoral metaphyseal bone mRNA was prepared from young rats and from 6-mo-old ovariectomized rats after subcutaneous injection of PTH(1–34) (30 μg/kg). Figures 6A and 6B show that ephrinB2 mRNA expression in metaphyses was increased 14- and 10.5-fold within 1 h of PTH treatment in the young and the ovariectomized rats, respectively. In each case, by 4 h after injection, ephrinB2 mRNA levels were the same as those of controls. No significant changes were observed in either EphB4 or EphB2 mRNA in metaphyseal bone in either study. As a control, IL-6 mRNA was measured, and as expected, IL-6 mRNA levels were significantly increased within 1 h of PTH treatment in both young and ovariectomized rats, with a 25-fold increase in the latter.
Finally, immunoperoxidase staining for ephrinB2 was carried out on sections of femur from 3-wk-old male rats. Osteoclasts throughout the sections stained positively for ephrinB2. Osteoblasts were also positively stained, but not uniformly. Positive staining for ephrinB2 occurred in groups of cells, particularly in more mature bone (Fig. 6A), whereas in less mature bone, osteoblasts were generally unstained (Fig. 6B). Some osteocytes in both regions stained positively for ephrinB2. In careful examination throughout sections, there were no examples found of positively stained osteoblasts located closely to osteoclasts.
PTHR1 is expressed by mature osteoblasts, and because the osteoblastic differentiation of the Kusa 4b10 mouse marrow stromal cell line has been clearly defined, we were able to select a differentiation stage very soon after appearance of functional PTHR1 to assess the effects of PTH and PTHrP. We compared PTH(1–34) and PTHrP(1–141), because mouse genetic experiments indicate that locally generated PTHrP in bone is a crucial regulator of bone remodeling. Thus, locally produced PTHrP may be the more likely local effector on osteoblasts and precursors in the bone formation process. This raises the possibility that biological activities within the PTHrP molecule, other than those acting through the PTHR1, might modify gene expression in ways that distinguish PTHrP from PTH action, either through uniquely induced genes or by changing the time course of gene activation. In the analysis of differentially expressed genes, summarized in Fig. 1, there was substantial concordance between major effects of PTH(1–34) and PTHrP(1–141). Differences between the two might result from effects on uniquely expressed genes, but more commonly on the magnitude and, importantly, the time course of responses. The pursuit of gene expression effects that might be ascribed to either PTHrP or PTH specifically will require extensive and carefully executed validation studies, based on the array data. That work continues and will be the subject of a separate report.
Gene validation studies were carried out on independent RNA samples from differentiated Kusa 4b10 cells, as well as from mouse calvarial osteoblasts and UMR 106–01 rat osteogenic sarcoma cells. The different cell sources were chosen to ensure that appropriate changes were observed in those genes frequently reported to respond to PTH or PTHrP and to compare with published arrays, showing some responses to PTH in vitro and in vivo. There was consistency among these results, with the only notable exception being expression of mRNA for IL-18, which was substantially increased by PTH in UMR 106–01 cells and in rat calvarial osteoblasts in a previous study. Whereas we confirmed this reported effect in UMR 106–01 cells, we could identify no response of IL-18 transcript levels to PTH or PTHrP either in the array data or in independent Kusa 4b10 or mouse calvarial osteoblast samples. This result suggests a species difference in regulation of this gene, but consistency in the other genes chosen for the validation studies.
When the genes whose expression was modified in the array were functionally categorized using pathway analysis, we noted that expression of genes belonging to the family of ephrins and their receptors were upregulated by PTH and PTHrP. This was of interest because ephrin/Eph family members have been recognized for some time as local mediators of cell function through largely contact-dependent processes in development and in maturity. They mediate cell attraction and adhesion, but often also provide signals that separate the cells, and have shown roles in tissue remodeling, including cell migration, axon guidance, and synapse plasticity. Based on the ephrin ligand structure the ephrins are in two classes, with the ephrin A class glycophosphatidylinositol (GPI)-tethered to the membrane and the B class consisting of type II membrane proteins. Although first considered to bind with class specificity (i.e., ephrinA to EphA and ephrinB to EphB), there are exceptions, with EphA4 being activated through ephrinB1 and ephrinA5 binding to and signaling through EphB2. A particular feature of ephrin/Eph biology is their capacity for bidirectional signaling, in that when an ephrinB acts on its EphB receptor tyrosine kinase, the latter can signal in the reverse direction, acting through the ligand by promoting rapid phosphorylation on highly conserved tyrosine residues within the cytoplasmic tail.
Possible involvement of the ephrin/Eph family in cell communication processes of bone remodeling has been raised by evidence that ephrinB2 produced by osteoclasts acts on EphB4 in osteoblasts to promote osteoblast differentiation and bone formation, an effect achieved without requiring the cytoplasmic domain of ephrinB2, and thus consistent with a contact-dependent process. As an example of bidirectional signaling, EphB4 in turn interacted with ephrinB2 to inhibit osteoclast differentiation. This effect did require participation of the ephrinB2 cytoplasmic domain. Thus, a contact-dependent interaction between osteoclasts and osteoblasts was proposed, with the capacity to regulate differentiation in both cell types through their intimate location and exchange of ephrin/Eph signals moving in either direction.
In this study, we provided data establishing that production of mRNA for ephrinB2 in differentiating mouse marrow stromal cells, in murine calvarial osteoblasts, and in UMR106 rat osteogenic sarcoma cells is increased by PTH and PTHrP in a dose-dependent manner. This response was only manifest in the mouse stromal cells once the cells expressed functional PTHR1 in the course of their differentiation (data not shown), and the effect seems to be a transcriptional one, the mechanism of which will be studied further. Treatment with forskolin as an adenylyl cyclase activator also increased ephrinB2 mRNA production in each of these cell types, with that effect also enhanced by phosphodiesterase inhibition. Increased mRNA for ephrinB2 was also found in metaphyseal bone of young and of aged ovariectomized rats treated with PTH. It should be noted that mRNA levels for ephrinB2, EphB2, and EphB4 in the Kusa 4b10 stromal osteoblastic cells remained virtually unchanged throughout 3 wk of growth of the cells under differentiating conditions compared with the changes taking place in expression of mRNA for osteocalcin, ALP, PTHR1, and SDF-1. Similarly, EphB4, EphB2, and ephrinB2 mRNA levels were relatively constant during differentiating cultures of mixed mouse calvarial osteoblasts. Furthermore, only ephrinB2 protein production was enhanced by PTH/PTHrP treatment in vitro and in vivo, without any effect on EphB2 or EphB4 protein. Thus, the regulated production of ephrinB2 in the face of the otherwise constant expression throughout differentiation of ephrinB2 and of the receptors, EphB2 and EphB4, provides for a paracrine or autocrine control mechanism within the osteoblast lineage regulated by PTH and/or PTHrP.
EphB2 mRNA was increased to some extent by PTH/PTHrP treatment in each of the osteoblastic cells in our experiments, but the amplitude of the response was much less than with ephrinB2, and there was no change in EphB2 mRNA in bone in the in vivo experiments. mRNA for the other receptor predominantly expressed in osteoblasts, EphB4, was not changed by PTH or PTHrP treatment in the array data or in any of the gene validation experiments using Kusa 4b10, mouse calvarial osteoblasts, and UMR 106–01 rat osteogenic sarcoma cells, nor was there any significant change in EphB4 mRNA after PTH treatment in rat metaphyseal bone. The ephrinB2 response to PTH and PTHrP in osteoblastic cells once they gain the capacity to respond through the receptor raises the possibility that such regulated production of ephrinB2 might exert a paracrine or autocrine action in committed pre-osteoblasts, favoring osteoblast differentiation and bone formation, and perhaps thereby contributing to the manner in which PTHrP, generated locally, can function physiologically in bone. Figure 7 depicts this, with ephrinB2 produced by responsive cells in the osteoblast lineage under the influence of PTH or PTHrP and acting on adjacent cells of the same lineage to promote osteoblast differentiation and bone formation, in the course of either remodeling or modeling of bone. The experiments carried out with the ephrinB2/EphB4 receptor antagonist showed functional outcomes of receptor blockade, with significant inhibition of mineralization and expression of mRNA for some genes that are expressed relatively late in differentiation in the Kusa 4b10 cells. These observations, together with the finding of ephrinB2 immunoreactivity particularly in groups of osteoblasts in more mature bone, are consistent with the idea that ephrinB2 is produced by osteoblasts and acts through its receptor, EphB4, within that lineage. It is also possible that reverse signaling could take place, with constitutively produced EphB2 or EphB4 activating the regulated ligand, ephrinB2, by promoting phosphorylation of its cytoplasmic domain.
The concepts arising from this study do not necessarily exclude the mechanism that has been proposed for coupling of bone formation to resorption in the BMUs, which invokes osteoclast-derived ephrinB2 acting in a contact-dependent manner on EphB4 in osteoblasts. In that work, it was found that mice with conditional ephrinB2 knockout in the myeloid lineage had no bone phenotype. The findings reported here of such regulation in the osteoblast might help to explain that and illustrate the plasticity of local regulatory mechanisms in bone, where so much depends on the locations of participating cells and whether mechanisms are autocrine or paracrine with a requirement for cell contact. The forward and reverse signaling interactions between Eph receptors and ligands require cell–cell contact because both molecule classes are anchored to the plasma membrane; therefore, spatiotemporal regulation of expression patterns must be achieved. Such controls could be achieved within the BMUs, where osteoclast-derived ephrinB2 might act on adjacent osteoblasts, whereas PTHrP-enhanced ephrinB2 could make that ligand available for paracrine actions on committed preosteoblasts in other parts of the BMU, as well as in the modeling process (Fig. 7).
Actions of ephrinB2 on osteoblasts, in addition to the forward signaling from the osteoclast that has been proposed, will require further study on the nature of ephrinB2 signaling. This includes the possibility of reverse signaling and signaling between classes of ephrins and receptors. A novel cytoplasmic protein, PDZ-RGS3, was discovered to bind to a PDZ (PSD-95/Dlg/ZO-1) domain in the cytoplasmic tail of B class ephrins, and its mRNA was significantly downregulated by PTH and PTHrP in the array as well as in independent RNA preparations (unpublished data). Src family kinases and a PDZ domain–containing phosphatase have been identified as regulators of ephrinB phosphorylation. Only the last 33 amino acids are needed for the PDZ domain interaction, and ephrinB2 and ephrinB1 sequences are identical in that region. This highlights the importance of redundant actions in this signaling system, which seems to be an important mediator of the action of transmembrane ephrinB ligands as receptor-like signaling molecules. Treatment with soluble EphB4 (EphB4-Fc) inhibited the chemoattractant effect of SDF-1 on cerebellar granule cells, thus showing a way in which ephrin reverse signaling can provide a mechanism for selective regulation of responsiveness through a G protein–coupled receptor. In fact, a PDZ domain interaction has been shown in the binding of PTHR1 to Na+/H+exchanger regulatory factors (NHERFs), with the PTHR1 containing an atypical, carboxy-terminal PDZ consensus motif, EVTM. In that work, it was shown that changes in the amount of NHERF in some cells might alter the balance between adenylyl cyclase and phospholipase C responses in those cells.
These data suggest that the effect through the PTHR1 on ephrinB2 production is a transcriptional one. The time courses of response differed in the cells studied in vitro, in that the peak response in differentiating Kusa 4b10 and UMR106–01 cells occurred at 4–6 h, whereas the peak response in mouse calvarial osteoblasts occurred at ∼1 h. The peak response in metaphyseal bone of young and aged ovariectomized rats after PTH injection was also at about 1 h. Such a rapid induction of formation of ephrinB2, confirmed by Western blotting showing rapid and sustained increase in protein, provides a means by which ephrinB2 can contribute to local remodeling and/or remodeling events by interacting in a paracrine or autocrine way with either or both of the two receptors, EphB2 and EphB4 (Fig. 7). Such a mechanism would provide one of several local mechanisms contributing to the physiological role of PTHrP in bone and the pharmacological effects of administered PTH as an anabolic therapy for bone. These might include ephrin-mediated modification of PTH/PTHrP signaling.
This work was supported by a Program Grant from the National Health and Medical Research Foundation of Australia. MTG and NAS are Principal and Senior Research Fellows, respectively, of the NHMRC. The Australian Genome Research Foundation (Parkville, Australia) is acknowledged for hybridization and probing of the arrays.