Bone morphogenetic proteins (BMPs) have been heretofore implicated in the induction of osteoblast differentiation from uncommitted progenitors during embryonic skeletogenesis and fracture healing. We have tested the hypothesis that BMPs are also involved in the osteoblastogenesis that takes place in the bone marrow in postnatal life. To do this, we took advantage of the properties of noggin, a recently discovered protein that binds BMP-2 and −4 and blocks their action. Addition of human recombinant noggin to bone marrow cell cultures from normal adult mice inhibited both osteoblast and osteoclast formation; these effects were reversed by exogenous BMP-2. Consistent with these findings, BMP-2 and −4 and BMP-2/4 receptor transcripts and proteins were detected in these primary cultures, in a bone marrow–derived stromal/osteoblastic cell line, as well as in murine adult whole bone; noggin expression was also documented in all these preparations. Moreover, addition of antinoggin antibody caused an increase in osteoblast progenitor formation. These findings suggest that BMP-2 and −4 are expressed in the bone marrow in postnatal life and serve to maintain the continuous supply of osteoblasts and osteoclasts; and that, in fact, BMP-2/4-induced commitment to the osteoblastic lineage is a prerequisite for osteoclast development. Hence, BMPs, perhaps in balance with noggin and possibly other antagonists, may provide the tonic baseline control of the rate of bone remodeling on which other inputs (e.g., hormonal, biomechanical, etc.) operate.
Bone remodeling, a process responsible for the renewal of the adult human skeleton approximately every 10 years, is carried out by teams of juxtaposed osteoclasts and osteoblasts, two specialized cell types that originate, respectively, from hematopoietic and mesenchymal progenitors of the bone marrow.(1,2) A continuous and orderly supply of these cells is essential for skeletal homeostasis as increased or decreased production of osteoclasts or osteoblasts and/or changes in the rate of their apoptosis are largely responsible for the imbalance between bone resorption and formation that underlies several systemic or localized bone diseases such as osteoporosis, Paget's, metastatic, and renal bone disease.(3–9) Even now, however, little is known about the factors responsible for sustaining the supply of osteoblasts in postnatal life and how osteoblastogenesis and osteoclastogenesis are coordinated to ensure a balance between formation and resorption during normal bone remodeling.
Bone morphogenetic proteins (BMPs), members of the TGFβ (transforming growth factor β) superfamily of proteins, are unique among growth factors that influence osteoblast differentiation because they can initiate this process from uncommitted progenitors in vitro as well as in vivo.(10–12) In particular, BMP-2 and −4 are expressed during murine embryonal skeletogenesis (days 10–12) and act on cells isolated from murine limb buds to promote their differentiation into osteoblasts. In addition, BMP-2 and −4 are involved in fracture healing, as evidenced by their expression in primitive mesenchymal cells and chondrocytes at the site of callus formation, and the ability of BMPs to accelerate the fracture healing process when supplied exogenously.(10,11) Besides BMP-2 and −4, BMP-5, −6, and −7 may also contribute to osteoblastic cell differentiation and bone formation.(10) BMP-2/4-induced osteoblast commitment is mediated by the type I BMP receptor and involves the phosphorylation of specific transactivators (smad 1, 5, and 8), which then oligomerize with smad 4, and translocate into the nucleus.(13) These events induce an osteoblast-specific transcription factor (CBFA-1 [core binding factor], also known as Osf-2 or PEBP2αA or AML3), which in turn activates osteoblast-specific genes.(14,15)
Several proteins able to antagonize BMP action have recently been discovered. Noggin, chordin, and cerberus were initially found in the Spemann organizer region of the Xenopus embryo and were shown to be essential for neuronal or head development.(16–21) Noggin and chordin inhibit the action of BMPs by binding to them with high affinity and preventing their interaction with their receptors. Of those BMPs tested, noggin displays specificity, in that binding is very tight to BMP-2 and BMP-4 (Kd = 2 × 10−11 M), weak to BMP-7, and undetectable to TGFβ or IGF (insulin-like growth factor)-I.(17)
Here, we have exploited the BMP-antagonist properties of noggin to test the hypothesis that BMPs are involved in the osteoblastogenesis that takes place in the bone marrow in postnatal life. Osteoclastogenesis and osteoblastogenesis proceed simultaneously in most circumstances(4,8) and the former may not occur without the latter,(15,22) because osteoclast development requires support from stromal/osteoblastic cells.(23) The mechanistic basis of this dependency has been recently explained by the discovery of a membrane-bound cytokine-like molecule, receptor activator of NF-κB ligand (RANKL), which is present in mesenchymal cells and binds to a specific receptor (RANK) on hematopoietic osteoclast progenitors.(24–26) Such binding is essential and, together with M-CSF, sufficient for osteoclastogenesis. Accordingly, we postulated that blockade of BMP action by noggin would interfere not only with osteoblastogenesis, but with osteoclastogenesis as well. We show that noggin does indeed inhibit both osteoblast and osteoclast formation in bone marrow cell cultures from adult mice, whereas inhibition of endogenous noggin production by a neutralizing antibody increases osteoblastogenesis. Consistent with these observations, we also show that the genes for BMP-2 and −4, the BMP-2/4 receptor, as well as noggin are expressed in the adult murine bone marrow, a marrow-derived stromal/osteoblastic cell line, and in murine adult whole bone.
MATERIALS AND METHODS
Human recombinant BMP-2 and −6, and an anti-human BMP-2/4 antibody, which recognizes BMP-2 and −4, were provided by V. Rosen (Genetic Institute, Cambridge, MA, U.S.A.). 1,25(OH)2D3 and murine soluble RANKL were provided by Hoffmann-LaRoche (Nutley, NJ, U.S.A.) and Dr. B. Boyle (Amgen Inc., Thousand Oaks, CA, U.S.A.), respectively. Human PTH (parathyroid hormone) (1–34) and human recombinant M-CSF (macrophage-colony stimulating factor) were purchased from Peninsula Laboratories (Belmont, CA, U.S.A.) and R&D Systems (Minneapolis, MN, U.S.A.), respectively. The doses for BMPs, 1,25(OH)2D3 PTH, soluble RANKL, or M-CSF used in these experiments were determined in preliminary studies. cDNA probes for murine BMP-4 and BMP receptor type IA (BMPR-IA) were provided by Dr. K. Miyazono (Cancer Institute, Tokyo, Japan). A murine RANKL cDNA probe was provided by Immunex Corp. (Seattle, WA, U.S.A.).
Osteoblast progenitor assays
C2C12 cells (2 × 104/cm2) were cultured with 100 ng/ml BMP-2, BMP-6, and/or 10–600 ng/ml noggin for 3 days and alkaline phosphatase (AP) activity was measured by Sigma kit #104 (Fig. 1).(27) For the experiment depicted (later) in Fig. 7, C2C12 cells (2 × 104/cm2) were precultured with 250 ng/ml noggin and 2 or 10 μg/ml of antinoggin antibodies (#06, #16, #18, or #21) overnight. Subsequently, 50 ng/ml BMP-2 was added and the cultures were maintained for 3 days. Bone marrow cells were obtained from femurs of 3-month-old male Swiss Webster mice and cultured at 1 × 106 cells (for CFU-F determination) or 2 × 106 cells (for CFU-OB) per 10 cm2 well in αMEM (α-minimal essential medium; Gibco-BRL, Gaithersburg, MD, U.S.A.) supplemented with 15% preselected FCS (Hyclone, Logan, UT, U.S.A.), 200 μM ascorbic acid, and 10 mM β-glycerophosphate. Cultures were maintained in the absence or presence of different concentrations of recombinant human noggin, without or with 100 ng/ml human recombinant BMP-2. The total number of CFU-F colonies and the number of AP-positive CFU-F colonies were determined after 10 days of culture by staining for AP, and the number of CFU-OB colonies was determined after 28 days of culture by Von Kossa staining, as previously described.(8) UAMS-33, a cell line with preosteoblastic properties, was obtained by limiting dilution subcloning from foci of transformed cells that developed during long-term culture of murine bone marrow cells.(28) For the osteoblast differentiation experiments, UAMS-33 cells were plated at 2 × 104/cm2 well and maintained for up to 8 days in αMEM containing 10% FCS in the presence of the indicated concentrations of recombinant human noggin or a pegylated noggin (PEG-noggin). PEG-noggin was prepared by covalent attachment of a 20-kDa polyethyleneglycol to available lysines of human noggin to form a secondary amine link-age.(29) PEG-noggin was purified by ion-exchange chromatography to isolate noggin cross-linked with a single PEG molecule (unpublished data by K. M. Bailey et al. in Re-generon Pharmaceuticals Inc.). As shown in the case of calcitonin, pegylation of noggin should provide a molecule that is more stable in solution than the unpegylated noggin.(29) Osteoblast differentiation was assessed by measuring AP activity.
Osteoclast development assays
Bone marrow cells were obtained from the femurs of 3- to 4-month-old mice and cultured for 8 days for the determination of osteoclast formation as previously described.(3) Cultures were maintained in the absence or the presence of 10−8 M 1,25(OH)2D3 or 10−8 M human PTH (1–34) without or with 200 ng/ml PEG-noggin, 300 ng/ml BMP-2, or 10 μg/ml antinoggin antibodies #06 or #21. In the experiments shown in Figs. 2C and 2D, osteoclast formation was assessed in 8-day cocultures of nonadherent bone marrow cells (0.5 × 106 cells/cm2) and either UAMS-33 cells (1 × 104/cm2) or murine calvaria cells (1 × 104/cm2). Direct effects of noggin on hematopoietic cell differentiation toward the osteoclastic phenotype were examined using nonadherent murine bone marrow cells (105/well). Nonadherent bone marrow cells obtained by preculturing bone marrow cells for 2 days (to remove stromal/osteoblasts) were cultured for 6 days with 10 ng/ml human M-CSF, 100 ng/ml soluble RANKL, and 10 or 100 ng/ml PEG-noggin. In all these experiments, osteoclastic cells were visualized by staining for tartrate-resistant acid phosphatase (TRAP) using Sigma kit #180.
RNA was prepared from freshly isolated bone marrow cells, or from 2-, 4-, 7-, 14-, and 21-day cultures of these cells as previously described,(30) and analyzed for the expression of BMP-2, BMP-4, osteocalcin, BMPR (BMP receptor)-IA, and BMPR-IB transcripts by RT-PCR. RT-PCR was performed using primers, as detailed previously for GAPDH (glyceraldehyde-3-phosphate dehydrogenase).(30) For murine BMP-2, the primer set was: 5′-CTAGTGTTGCTGCTTCCCCA (forward), 5′-GAGTTC-AGGTGGTCAGCAAG (reverse); for murine BMP-4, the primer set was: 5′-GCGCCGTCATTCCGGATTAC (forward), 5′-CATTGTGATGGACTAGTCTG (reverse); for murine BMPR-IA, the primer set was: 5′-GGCAGAA-TCTAGATAGTATGCTCC (forward), 5′-GAAGTTAA-CGTGGTTTCTCCCTG (reverse); for murine BMPR-IB, the primer set was: 5′-CACCAAGAAGGAGGATGGAG-AGA (forward), 5′-CTACAGACAGTCACAGATAAGC (reverse); for murine osteocalcin, the primer set was: 5′-TCTGACAAAGCCTTCATGTCC (forward), 5′-AAA-TAGTGATACCGTAGATGCG (reverse); and for murine noggin, the primer set was: 5′-TGGACCTCATCGAA-CATCCAGAC (forward), 5′-ACTTGGATGGCTTAC-ACACCATGC (reverse). Using these primers, the expected sizes of the PCR products are exactly as depicted later in Fig. 6A.
In situ RT-PCR
Bone marrow cultures were established and maintained for 2 weeks. After fixation with 4% paraformaldehyde, BMP-4 and osteocalcin transcripts were detected using the BMP-4 and osteocalcin primers described earlier and a method previously detailed elsewhere.(30) As a negative control, parallel cultures were processed for BMP-4 detection without reverse-transcriptase.
RNase protection assay
RNA was prepared from 2-week cultures of murine bone marrow cells and from 3-day cultures of UAMS-33 cells, and assayed by RNase protection for BMP-4 and BMPR-IA transcripts. For the preparation of the cDNA probes, plasmids containing the coding regions of murine BMP-4 and BMPR-IA were subcloned in Bluescript KS(+) plasmids and were linearized. The respective riboprobes were synthesized in the presence of 50–100 μCi of [32P]UTP (3000 Ci/mmol, Amersham Corp., Arlington Heights, IL, U.S.A.), and T7 or SP6 RNA polymerase, as appropriate (Promega, Madison, WI, U.S.A.). Total RNA (30 μg) was extracted from 3-day cultures of UAMS-33 cells and 2-week cultures of murine bone marrow cells. RNA and 32P-labeled riboprobes in hybridization buffer (80% formaldehyde; 40 mM PIPES, pH 6.4; 400 mM NaCl; 1 mM EDTA) were annealed at 45°C overnight after heating at 85°C for 5 minutes. Subsequently, annealed RNAs were treated with RNase A (40 μg/ml) at 30°C for 60 minutes, and the enzyme was inactivated by proteinase K (100 μg) and 10% SDS. Finally, the samples were loaded onto 4.5% polyacrylamide gels with 7 M urea after extraction by phenol and ethanol.
Two-week cultures of murine bone marrow cells and 3-day cultures of UAMS-33 cells were immunohistostained with an antibody against BMP-2 and −4. Cells were fixed with 10% formalin for 10 minutes, treated with 0.1% H2O2 for 30 minutes to remove endogenous peroxidase activity, and blocked with 5% normal goat serum for 1 h. They were then incubated with mouse anti-human BMP-2/4 antibody for 1 h and subsequently for 30 minutes with biotinylated second antibody (Vector, Burlingame, CA, U.S.A.), followed by incubation with peroxidase-conjugated streptavidin, and 3,3′-diaminobenzidine tetrachloride (Santa Cruz Biotechnology Inc., Santa Cruz, CA, U.S.A.). As a negative control, cells were processed without primary antibody.
Northern blot analysis
Total RNA (30 μg) or polyA+ RNA (8 μg) were prepared from 3- or 6-day cultures of UAMS-33 cells, and were electrophoresed on 1% agarose gels. Northern blotting for RANKL, M-CSF, and noggin expression was performed as previously described.(30)
Western blot analysis
Cells were lysed in 10 mM phosphate buffer (pH 7.4), 10% glycerol, 1%NP-40, 0.1% SDS, 4 mMEDTA, 0.15 M NaCl, 0.01 M NaF, 0.1% sodium orthovanadate, 1 mM PMSF, 5 μg/ml trypsin inhibitor, and 5 μg/ml protease inhibitors, and centrifuged at 14,000g for 10 minutes. For preparation of bone homogenates, femurs were frozen in liquid N2 and homogenized in the above-mentioned buffer using a Polytron homogenizer, and then centrifuged. Supernatant proteins (50–200 μg) from the cell lysates or the bone homogenates were subjected to 12% SDS-PAGE (polyacrylamide gel electrophoresis) and electroblotted onto a PVDF membrane (Millipore, Bedford, MA, U.S.A.). Membranes were incubated for 2 h at room temperature in 5% dry milk in 20 mM Tris, (pH 7.2), 0.15 M NaCl containing 0.05% Tween 20, and subsequently with anti-BMP-2/4 antibody, or the antinoggin antibody, and then appropriate second antibody (HRPO [horseradish peroxidase]-goat anti-mouse IgG antibody for BMP-2/4 or HRPO-goat anti-rat IgG antibody for noggin). The bound antibody on the membrane was detected by enzyme reaction using an enhanced chemiluminescence kit (Dupont NEN, Boston, MA, U.S.A.).(30)
Inhibition of osteoblastogenesis by noggin
The ability of human recombinant noggin to antagonize BMP-mediated commitment to the osteoblast lineage, as well as the specificity of this effect, was established using C2C12 cells. Human recombinant noggin (100–600 ng/ml) inhibited BMP-2-induced alkaline phosphatase (AP) activity, a phenotypic marker of osteoblastic cells, in a dose-dependent manner (Fig. 1). At the highest concentration of noggin, the effect of BMP was completely blocked. Like BMP-2, BMP-6 stimulated AP activity in C2C12 cells; however, the effect of BMP-6 was not affected by noggin, indicating that noggin is not an antagonist of BMP-6. As shown previously,(31) BMP-12 or −13 added alone did not induce AP activity in C2C12 cells, nor did they interfere with the stimulatory effect of BMP-2 or −6 (data not shown).
Noggin had no effect on the number of colony forming units–fibroblast (CFU-F) that were formed during a 10-day primary culture of murine bone marrow cells obtained from 3-month-old mice (Fig. 2A). Noggin, however, led to a dose-dependent decrease in the number of AP-positive CFU-F colonies (Fig. 2B). Noggin also inhibited bone nodule formation by subsets of CFU-F colonies, designated CFU-osteoblast (CFU-OB), in 28-day primary cultures of bone marrow cells (Fig. 2C). Both effects were detectable with as little as 6–10 ng/ml of noggin. Practically complete suppression of AP expression in CFU-F or bone nodule formation by noggin could be observed with 150–600 ng/ml. The inhibitory effect of noggin on AP expression in CFU-F colonies could be reversed by addition of 100 ng/ml BMP-2, except at the highest concentration of noggin (600 ng/ml), confirming competitive antagonism between noggin and BMPs (Fig. 2B). Similarly, noggin and PEG-noggin dose-dependently reduced AP activity in UAMS-33 cultures maintained under basal conditions (Fig. 2D), or in the presence of exogenous BMP-2 (data not shown). Noggin also prevented calcium deposition by UAMS-33 cells cultured in the presence of ascorbic acid and β-glycerophosphate for 5 days (data not shown). In line with the expectation that pegylation would increase the stability of noggin, we found in all these experiments that PEG-noggin was 10-fold more potent than unpegylated noggin. Withdrawal of PEG-noggin after 4 days restored AP levels in UAMS-33 cells, suggesting that the effects of the protein observed herein were not the result of cytotoxic actions (Fig. 2D). Consistent with this, noggin had no effect on cell viability measured by trypan blue exclusion.
Inhibition of osteoclastogenesis by noggin
Osteoclast formation induced by either 1,25(OH)2D3 or PTH was also inhibited by the addition of PEG-noggin in bone marrow cell cultures (Figs. 3A and 3B), or cocultures of nonadherent bone marrow cells, a source of osteoclast progenitors, and either UAMS-33 cells (Fig. 3C) or neonatal murine calvaria osteoblastic cells (Fig. 3D). UAMS-33 cells can support osteoclast differentiation induced by 1,25(OH)2D3, but unlike calvaria cells, they do not support PTH-induced osteoclast development. The effect of noggin was dose dependent and identical in experiments using cells from 3- or 6-month-old mice. In contrast, exogenous human recombinant BMP-2 increased osteoclast formation as much as 4- to 6-fold over baseline in PTH-stimulated primary bone marrow cell cultures (Fig. 3B), and in cocultures of hematopoietic progenitors and calvaria cells (Fig. 3D). An increase, albeit smaller, was also observed in 1,25(OH)2D3-stimulated cultures (Figs. 3A and 3C). However, in different experiments, in which cocultures of UAMS-33 cells and nonadherent hematopoietic precursors were pretreated with BMP-2 before exposure to 1,25(OH)2D3, BMP-2 stimulated osteoclast formation by 3- to 4-fold (data not shown).
To investigate the possibility that noggin could have inhibited osteoclast formation in part as a result of direct actions on hematopoietic cells, we employed preparations of nonadherent murine bone marrow cells (devoid of stromal/osteoblastic cells) and cultured them for 6 days in the presence of human recombinant M-CSF and soluble murine RANK ligand (sRANKL)(Fig. 4). As shown previously by Lacey et al.,(24) sRANKL along with M-CSF was sufficient for the induction of osteoclast formation in the absence of stromal/osteoblastic support cells or stimuli like 1,25(OH)2D3 or PTH. In this system of stromal/osteoblastic cell-independent osteoclastogenesis, noggin had no effect on osteoclast development. Consistent with the conclusion that noggin does not exert its anti-osteoclastogenic effects through direct actions on hematopoietic osteoclast precursors, PEG-noggin had no effect on M-CSF-stimulated proliferation of nonadherent hematopoietic progenitors isolated from murine bone marrow (data not shown).
In support of the notion that the effect of noggin on osteoclastogenesis was the result of attenuation of stromal/osteoblastic cell differentiation toward a state capable of supporting osteoclast development, PEG-noggin at 200 ng/ml decreased the expression of the mRNA for RANK ligand by 40% in 1,25(OH)2D3-treated UAMS-33 cells (Fig. 5); however, at the same concentration, PEG-noggin had no effect on M-CSF mRNA. Besides affecting the differentiated phenotype, the inhibitory effects of PEG-noggin on osteoclast development could be caused by a reduction in the total number of stromal/osteoblastic cells. This possibility was excluded by showing that PEG-noggin did not affect UAMS-33 or calvaria cell proliferation under the conditions used in the cocultures of Fig. 3 (data not shown).
Expression of BMP-2/4, their receptors, and noggin in bone marrow and bone
The results presented earlier strongly suggested that the genes encoding BMPs and their receptors are expressed in bone marrow cultures from normal adult mice and that their products are required for osteoblastogenesis as well as osteoclastogenesis. Therefore, we searched for BMP-2, −4, and −7 transcripts and proteins in bone marrow cells and homogenates of bone from femurs of 3-month-old mice (Fig. 6). BMP-4 transcripts could be detected by RT-PCR in adherent bone marrow cells maintained in culture as early as 4 days and BMP-2 transcripts in cells maintained for 2 weeks, but not in freshly isolated bone marrow cells (Fig. 6A). In contrast to BMP-2 and −4, BMP-7 transcripts were not detected in either freshly isolated or cultured bone marrow cells (data not shown). BMP receptor type IA was detected by RT-PCR in freshly isolated cells and throughout the culture; whereas the type IB receptor was first visualized at 4 days. Osteocalcin mRNAs was first detected at 1 week of culture. We previously showed that IL-6, gp130, and osteopontin transcripts are absent from freshly isolated bone marrow cells but appear during culture.(30,32) Hence, the expression of BMPs and BMP receptor genes on culture of bone marrow cells is most likely the result of an expansion of the osteogenic cell population. Using in situ RT-PCR analysis, the BMP-4 and osteocalcin transcripts were localized in a subset of cells present within the CFU-F colonies (Fig. 6B). Independent confirmation of the RT-PCR results for BMP-4 and BMPR-IA expression in bone marrow cultures was obtained using RNase protection assay of RNA isolated from 2-week cultures; identical transcripts were also observed in the UAMS-33 cells (Fig. 6C). In agreement with the mRNA studies, the BMP-2/4 proteins were detected in bone marrow cells cultured for 2 weeks, as well as in UAMS-33 cells, by immunostaining with an antibody that recognizes both BMP-2 and −4 (Fig. 6D). Moreover, BMP-4 transcripts and BMP-2/4 proteins as well as noggin transcripts and proteins could be shown by RT-PCR or Northern blot analysis (Fig. 6E) and Western blot analysis (Fig. 6F) in bone marrow cultures, homogenates of whole femurs from 3-month-old mice, and also in UAMS-33 cells. In UAMS-33 cells, noggin expression was highest in post-confluent cultures (6 days), at which time the activity of the osteoblast phenotypic marker alkaline phosphatase was maximal.
Increased osteoblastogenesis by a noggin-neutralizing antibody
Finally, based on the observations that noggin was expressed in the primary murine bone marrow cell cultures, we attempted to establish the biological significance of this phenomena using a noggin-neutralizing antibody. For this purpose, we employed a rat-derived monoclonal antinoggin antibody (#06), prepared by standard hybridoma technology, which recognizes both human and murine noggin. The ability of this antibody to neutralize the activity of noggin was shown in the results shown in Fig. 7. Indeed, the inhibitory effect of noggin on BMP-2-induced differentiation of C2C12 cells toward osteoblasts could be reversed completely by 2–10 μg/ml of the #06 antibody. In contrast to antibody #06, a different antinoggin antibody (#21) did not exhibit neutralizing activity at the same doses as #06. Therefore, the latter was used in these studies as a negative control. As shown in Fig. 8, addition of 10 μg/ml of the noggin-neutralizing antibody #06, but not antibody #21, to murine bone marrow cell cultures caused an increase in the number of both AP-positive CFU-F and CFU-OB colonies (Fig. 8). However, the noggin-neutralizing antibody did not affect osteoclast formation in bone marrow cell cultures (osteoclast number 51.5 ± 12.1 in group with #21 vs. 63.5 ± 13.1 in group with #06).
Nature recapitulates evolutionary successful mechanisms to accomplish similar tasks at different stages of life. The results of this study reinforce this truism by showing that BMP-2 and −4 the same proteins that have been implicated in the induction of osteoblast formation in the embryo and during fracture repair may be required for osteoblastogenesis in the murine bone marrow in postnatal life. In addition our data reveal that induction of mesenchymal cell differentiation toward the osteoblast phenotype by BMP-2/4 is a prerequisite for osteoclastogenesis. Thus the requirement of BMPs for osteoclastogenesis may offer an entirely new perspective of how bone homeostasis is maintained during physiologic remodeling and may explain the in vivo observations that osteoclastogenesis(4,8,15) and loss of bone(22) cannot occur without osteoblastogenesis.
In our studies, BMP-2 alone did not induce osteoclast formation and, consistent with this finding, did not induce the expression of RANKL in stromal/osteoblastic cells. However, it did stimulate osteoclast formation (and probably RANKL expression, as evidenced by the decrease of RANKL levels by noggin) in the presence of 1,25(OH)2D3. This finding is somewhat different from the results of Kanataki et al.(33) and Hentunen et al.,(34) who have reported that BMP-2 or −7 alone has a stimulatory effect on osteoclast formation in vitro, although in those other studies, 1,25(OH)2D3 dramatically enhanced the level of osteoclastogenesis produced by BMPs alone. This apparent discrepancy may result from the difference in the culture systems used in those studies versus the present one. Even so, although both BMPs and 1,25(OH)2D3 seem to influence the process of osteoclastogenesis through effects on stromal/osteoblastic cells, the mechanism(s) by which they influence the process and the stage of cellular differentiation at which they seem to work are quite distinct. Indeed, whereas BMP-2/4 induce uncommitted mesenchymal progenitors to differentiate toward the osteoblastic lineage (perhaps by inducing CBFA-1 expression), in studies not shown here, we have found that 1,25(OH)2D3 inhibits BMP-4 expression by 80% and CBFA-1 expression to undetectable levels in bone marrow–derived UAMS-33 stromal/osteoblastic cells. Therefore, unlike BMPs, which act to induce the differentiation of uncommitted mesenchymal progenitors toward the stromal/osteoblastic lineage, 1,25(OH)2D3, at least in vitro, appears to act at a later stage to retain committed cells to a state of preterminal differentiation under which they are capable of supporting osteoclast development. This notion is consistent with our other studies, indicating that the osteoclast support cells in the bone marrow are not fully differentiated osteoblasts, but rather cells expressing a preterminally differentiated phenotype that exhibits mixed osteoblastic/adipocytic characteristics.(35)
The evidence that BMPs are required for osteoblast and osteoclast development implies that in addition to their previously known roles in skeletal development and repair, these proteins may also be involved in bone remodeling. All these processes require osteoblasts and osteoclasts, but it appears that the production of these cells is governed by the same proteins throughout life; consequently, biomechanical and local signals must determine how they are deployed for different purposes. The evidence that the genes for BMP-2 and −4 as well as noggin are expressed in the adult murine bone marrow raises the possibility that the balance between BMPs and noggin provides a tonic baseline control of osteoblastogenesis and osteoclastogenesis on which other inputs (e.g., biomechanical, hormonal, etc.) operate, either by influencing the balance between BMPs and noggin or by providing independent signals that alter the pro-differentiating effects of BMPs. We are quick to point out, however, that noggin may be just one of several BMP antagonists with a role in the regulation of osteoblastogenesis and osteoclastogenesis, because other proteins such as chordin have similar BMP antagonist properties.(16–21) Because BMP-6 has been implicated in osteoblast differentiation,(10) in studies as shown in Fig. 1, we searched for and found that noggin did not antagonize BMP-6 activity. Therefore, it is unlikely that the effects of noggin described herein are the result of interference with BMP-6.
We have not definitively identified the cells expressing BMP-2 and −4 and noggin in bone marrow and intact bone. Nonetheless, the in situ RT-PCR analysis of the bone marrow cultures suggests that they are of the mesenchymal/osteoblastic lineage. Consistent with this, we detected BMP-4, BMPR-IA, as well as noggin transcripts and proteins in UAMS-33 cells, a bone marrow–derived preosteoblastic line.(28) The level of expression of these genes in UAMS-33 cultures, a homogeneous preparation, was considerably higher when compared with the heterogeneous primary bone marrow cell cultures. This observation adds support to the contention that most of the cells expressing these genes in the bone marrow are of the mesenchymal/osteoblastic lineage. The observation that expression of noggin increased in the UAMS-33 cells during culture suggests that noggin expression by osteoblastic cells may increase as they advance to more differentiated stages (Fig. 6F). More important, the evidence that neutralization of endogenous noggin increased the number of CFU-F and CFU-OB colonies in our murine bone marrow cell cultures (Fig. 8) strongly suggests that the biological role of BMPs is counterbalanced by noggin. In fact, if noggin is produced by terminally differentiated osteoblasts, lining cells, or osteocytes, it could serve to restrict inappropriate osteoblastogenesis, and thereby target remodeling to appropriate sites. In support of the contention that the balance between BMPs and noggin has important biologic implications, it was recently shown that absence of regulated BMP activity in mice lacking noggin leads to failure of joint development; (36) and that BMPs as well as TGFβ induce the expression of noggin in osteoblastic cells.(37) Moreover, in line with the view that noggin may serve to counteract the effects of BMPs in vivo, PEG-noggin administration to mice inhibits bone formation in subcutaneous implants of BMP-2-impregnated matrigel, a solubilized basement-membrane preparation.(38) Definitive evidence, however, for or against the importance of these proteins in adult bone remodeling must, of course, await the results of future studies that examine the effects of administration of BMP-2 or −4 or noggin to adult animals.
In agreement with the evidence that osteoblastogenesis is a prerequisite for osteoclastogenesis—highlighted herein by the evidence that the effects of noggin on osteoclastogenesis are the result of antagonism of BMP actions on mesenchymal cells, as opposed to direct actions on hematopoietic progenitors—we have found and reported elsewhere that the promoters of both the murine and human RANK ligand gene contain two functional CBFA-1 binding sites; and that mutation of these sites abrogates the transcriptional activity of the RANKL promoter.(39) We, therefore, propose that a BMP → CBFA-1 → RANK ligand gene expression cascade in cells of the stromal/osteoblastic lineage constitutes the molecular basis of the linkage between osteoblastogenesis and osteoclastogenesis, with the last component probably requiring additional stimuli such as 1,25(OH)2D3, PTH, or gp130-activating cytokines.
In conclusion, the evidence presented in this study suggests that the balance between BMPs and their antagonists is a critical determinant of osteoblastogenesis as well as osteoclastogenesis, and thereby the rate of bone remodeling. It is therefore possible that novel therapies for bone diseases may be developed, based on the manipulation of the balance between these proteins.
The authors thank Drs. A. Mchael Parfitt, George Yancopoulos, and David Roodman for critical review of the manuscript before its submission; Catherine Smith for technical assistance; and Tonya Smith for help with the preparation of the manuscript. This work was supported by the National Institutes of Health (grants P01 AG/AR13918 and R01 AR43003) and the Department of Veterans Affairs.