Pharmacologic targeting of a stem/progenitor population in vivo is associated with enhanced bone regeneration in mice

Bzb increases serum osteocalcin, trabecular bone, osteoblast number, and bone marrow–derived osteogenic CFUs in mice.

To determine the effect of Bzb on murine osteogenesis, mice were treated with the drug at varying doses and serum osteocalcin, a surrogate marker for bone formation activity, was measured in vivo. At a dose of 0.3 mg/kg (doses typically used for antitumor effect in mice are 0.6–1.3 mg/kg), treatment for 3 weeks resulted in a significant increase in serum osteocalcin compared with vehicle-treated mice (Figure 1A) (the effect was also seen at 0.05 and 0.l mg/kg; data not shown). After 3 weeks of treatment, femurs were analyzed by quantitative micro-CT scanning. Trabecular volume, trabecular number, and trabecular density (Figure 1, B, D, and E) in the distal femur were all increased in drug-treated animals. Histomorphometric analysis on plastic sections revealed an increase in osteoblast number (Figure 1F) and mineral apposition rate (Figure 1G). Tartrate-resistant acid phosphatase (TRAP) staining revealed no significant change in osteoclasts per bone surface perimeter (BSpm, defined as total bone surface length per tissue section) in the secondary spongiosa (Figure 1H).

To determine whether the effect of Bzb was occurring at a stem/progenitor cell level, we performed several in vitro and in vivo studies. MSC activity is typically assessed by the ability of these cells to form stromal colonies (CFU-F) in vitro. In Bzb-treated animals, CFU-F was increased. When the differentiation potential of stromal cells was assessed by plating out CFUs in osteogenic and adipogenic conditions, there was a skewing of colonies toward osteogenic lineage (measured as alkaline phosphatase–positive CFU-Alk) versus adipogenic cells (CFU-Adipo) (Figure 1C). Bzb treatment in vitro (i.e., after the plating of untreated primary cells) had no effect on the CFU-Alk (data not shown). Although alternative hypotheses may explain these findings, these results suggested that Bzb treatment of animals targets a marrow-resident cell population that possesses the capacity for bilineage differentiation and the ability to form MSC stromal colonies in vitro.

Bzb promotes osteoblastogenesis in purified murine MSCs.

Murine bone marrow MSCs can be obtained by culturing plastic adherent bone marrow stromal cells in vitro; these cells have the ability to be serially passaged, retain multipotency, and have been antigenically characterized as CD105⁺, CD45^–, and hematopoietic lineage negative (12–15). To determine whether MSCs respond to Bzb in vitro, plastic-adherent MSC stromal cultures were established from a mouse expressing GFP under the control of an actin promoter (GFP expression was used as a marker in subsequent transplant studies). At 3 to 4 weeks after the establishment of these cultures (i.e., at passages 2–4), CD105⁺ cells were positively selected using magnetized biotinylated beads (the cells were also CD45^– and lineage negative by FACS analysis and represented between 2% and 5% of the original cultures). Under osteogenic conditions, exposure of this cell population to Bzb increased the number of alkaline phosphatase–positive cells (Figure 2B). To further characterize the response of this cell population to Bzb, we measured the transcription of 2 genes: bone sialoprotein (BSP), a marker of differentiated osteoblasts, and Runx-2, which is obligate for osteogenesis. In purified CD105⁺ cells, Bzb treatment significantly increased the expression of BSP and Runx-2 as detected by quantitative PCR (qPCR) analysis (Figure 2C). In contrast, when the same cells were predifferentiated into presumptive osteoblasts using osteogenic medium for 7 days (a point of time when BSP expression increases nearly 80-fold and mineralized nodules begin to appear in vitro), they lost the responsiveness to Bzb. Bzb treatment of MSCs also increased von Kossa staining in in vitro cultures (a measure of calcified extracellular matrix formation), while predifferentiated cells showed no increase in staining (Figure 2D). When MSCs were subjected to adipogenic differentiation in the presence of Bzb, a decreased number of oil red O staining cells was observed, suggesting that osteogenesis and adipogenesis were being reciprocally affected by Bzb.

In contrast to MSCs, osteoprogenitors (Ops) and osteoclasts also did not respond to Bzb treatment. To investigate the effect of Bzb on Ops, we used a mouse strain in which Ops are marked by the expression of a GFP-Cre fusion protein under the control of the osterix (Osx) promoter (16, 17). Osx is expressed during early osteogenic commitment (after Runx-2 expression) and downregulated as cells progress into mature osteoblasts (17). We isolated CD45^–GFP⁺ cells from Osx-GFP compact bone to obtain a population of GFP⁺ Op-rich fraction (10,000 cells/mouse) and a GFP^– Op-depleted fraction. The Op-enriched fraction did not show a response to Bzb as measured by collagen I staining while the Op-depleted (CD45^–GFP^–) fraction, containing the undifferentiated (i.e., uncommitted) MSCs, showed a response (Figure 3). We do note, however, that Osx-positive Ops, which are downstream of Runx-2 activation (17), appear to have a relatively poor proliferative capacity in vitro compared with the highly proliferative MSCs. Since osteogenesis depends on the proliferation and density of cells, the absence of responsiveness may thus reflect a general ability of this subpopulation of cells to form appropriately dense colonies in vitro.

Osteoclasts were also insensitive to Bzb at these doses in vitro. Osteoclasts were derived from bone marrow cells in the presence of M-CSF and RANKL. TGF-β served as a positive control and caused a significant increase in TRAP activity, while the addition of Bzb did not affect TRAP activity of these cultures (Supplemental Figure 1A; supplemental material available online with this article; doi:10.1172/JCI33102DS1). Bzb also had no effect on the size of pits formed by osteoclasts in vitro, suggesting that the resorptive activity of osteoclasts was unaffected at these doses (Supplemental Figure 1B).

We next determined whether the effects of Bzb on murine MSCs could be recapitulated in human bone marrow MSCs. These cells have been extensively characterized by other groups and are available for research use as a standardized reagent (18–21). By immunophenotype, these cells were homogenously CD105⁺CD73⁺CD11b^–CD31^–CD45^– and were capable of multilineage differentiation (data not shown). When primary human MSCs (hMSCs) were exposed to Bzb, there was an increase in alkaline phosphatase index (API) (Figure 4A) and in alizarin red staining (Figure 4B) and a decrease in adipocytes (Figure 4C). These data suggest that hMSCs and mouse MSCs respond comparably to Bzb.

Bzb stabilizes Runx-2 degradation and depends on the presence of Runx-2 to increase osteoblastogenic genes.

Next, we investigated a potential mechanism for the effect of Bzb on osteoblastic differentiation. The transcription factor Runx-2 (CBFA-1) is known to be the earliest expressed of the essential regulators of osteogenic differentiation (22, 23). Runx-2 is degraded by the proteasome and reciprocally affects osteogenic versus adipogenic differentiation (24–26). To investigate whether Bzb affects Runx-2 degradation, MSC cultures were treated with Bzb in vitro under baseline, nonosteogenic conditions (in the absence of dexamethasone, ascorbate, and β-glycerol phosphate) for 6 hours. Bzb increased the steady-state levels of Runx-2, particularly of the p62 isoform, associated with osteogenic commitment (26). Treatment with lactacystin (Figure 5A), a structurally unrelated proteasome inhibitor, demonstrated a similar effect on Runx-2 protein stability (Figure 5A). The transcriptional level of Runx-2 under these conditions was unaffected by Bzb treatment as measured by qPCR (data not shown). Runx-2 is also known to be a potent activator of the osteocalcin gene (26, 27), and we next asked whether Bzb-mediated Runx-2 stabilization might increase the expression of a Runx-2 target gene. To test this, we used a reporter system consisting of a vector that expresses firefly luciferase under the control of 6 Runx-2 binding elements derived from the osteocalcin gene (27). When a Runx-2–expressing vector was cotransfected with this reporter construct into C3HT1/2 fibroblasts (that have negligible endogenous Runx-2 activity), an increase in luciferase activity was noted. Notably, treatment with both Bzb and the proteasome inhibitor lactacystin led to a significant further increase in luciferase activity, suggesting that proteasome inhibitor–mediated Runx-2 stabilization can increase the transcriptional activity of Runx-2 (Figure 5B).

Normally, Runx-2 is recruited for ubiquitination and subsequent proteasomal destruction by the adaptor protein Schnurri-3 (Shn3) (27). In Shn3^–/– MSCs, Runx-2 levels are endogenously high (Figure 5A), leading to increased osteogenesis in vivo. If Bzb is acting through the Runx-2 pathway, then endogenous upregulation of Runx-2, resulting in accelerated differentiation of mature osteoblasts, should abrogate the responsiveness of MSCs to the drug. To test this hypothesis, we treated WT and Shn3^–/– MSCs with Bzb. In WT cells, Bzb treatment led to an increase in von Kossa–positive bone nodules (Figure 5C). In contrast, von Kossa–staining ECM nodules in Shn3^–/– cells were markedly elevated at baseline but remained unchanged with Bzb treatment.

Conversely, deletion of Runx-2 may be expected to decrease osteogenic differentiation and the responsiveness to Bzb. Runx-2^–/– MSCs could not be obtained because Runx-2^–/– mice are unable to form mineralized bone and die in utero (22, 23, 27, 28). However, embryonic mesodermal fibroblasts can be derived from these mice. The WT cells express Runx-2 and can differentiate into adipocytes and osteoblasts capable of giving rise to mineralized nodules in vitro (29). Embryonic mesodermal fibroblasts from Runx-2^–/– mice (versus WT controls) were treated with Bzb in vitro and assessed for osteogenic differentiation using a panel of 3 genes. For 2 genes (BSP and alkaline phosphatase [Alp]), Runx-2^–/– cells did not show responsiveness to Bzb compared with WT cells (Figure 5D). For the third gene, collagen I, both WT and Runx-2^–/– cells showed increased expression with Bzb treatment. Exposure to bone morphogenetic protein–2 (BMP-2), a bone-inducing signaling molecule, accelerated the conversion of the WT (i.e., Runx-2 intact) cells into Ops (reflected by the upregulation of osteogenic genes such as osteocalcin and BSP) and rendered them unresponsive to Bzb. These results suggest that although some genes involved in osteoblastogenesis, such as collagen I, may be increased by Bzb even in the absence of Runx-2, Bzb depends on the presence of Runx-2 to increase the expression of other genes obligate for osteoblastogenesis. Thus, Runx-2–dependent osteogenesis is a pathway critical to the Bzb response. However, Bzb must also target other pathways in MSCs, since the effect of Bzb on collagen I was still maintained despite Runx-2 ablation.

Implanted hMSCs and mouse MSCs respond to Bzb in vivo.

We next sought to determine whether MSCs respond to Bzb in vivo. MSCs can be implanted into mice, and they engraft and differentiate and can thus be assessed in vivo (30, 31). We performed 2 studies. First, undifferentiated hMSCs or MSCs preinduced to osteoblastic differentiation for 1 week were loaded into collagen sponges and implanted subcutaneously in immunodeficient mice. Under these conditions, MSCs typically form ectopic bone while osteoblasts show limited mineralization capability (this model has been extensively used to determine the differentiation activity of MSCs) (30). After implantation, animals were treated either with Bzb or saline for 10 doses (3 times a week for 27 days). In MSC-loaded sponges, Bzb treatment enhanced the formation of ectopic bone (Figure 6A) and increased the presence of mineralizing osteoblasts, assessed by endosteal cells associated with the matrix-incorporated dye xylenol orange (Figure 6A). Sponges containing cells that had already been exposed to osteogenic differentiation medium in vitro (i.e., before implantation) showed no response to Bzb. Undecalcified sponges containing MSCs were sectioned and stained with alizarin red (Figure 6A) to stain calcified matrix, and the fraction of tissue staining positive for alizarin was found to be significantly increased for Bzb-treated mice.

Second, we implanted GFP⁺ murine MSCs into the bones of sublethally irradiated mice by direct intrafemoral injection (31). The mice were then treated with Bzb or saline. MSCs and MSC-derived cells were identified by GFP expression (Figure 6B), and osteoblasts were marked by alkaline phosphatase staining. In saline-treated animals, GFP⁺ MSCs gave rise to some ectopic mineralization. A few GFP⁺ osteoblasts were observed occupying their normal positions at the endosteal surface, but most GFP⁺ donor cells remained alkaline phosphatase negative (Figure 6B). In contrast, in Bzb-treated mice, abundant ectopic mineralization was observed in the intrafemoral space. Notably, donor-derived ectopic osteoblasts (alkaline phosphatase–positive, GFP-positive cells) appeared several cell layers beyond the endosteal surface (Figure 6B). In saline-treated mice, implanted MSCs typically retained the expression of the MSC marker CD105, while in Bzb-treated animals, GFP⁺ cells were typically found embedded inside mineral and had lost CD105 expression (Figure 6B). Taken together, these results suggest that, in this transplantation system, bone regeneration is induced from implanted MSCs upon treatment with Bzb.

Bzb increases trabecular bone volume and bone formation in ovariectomized mice.

To determine whether drug-induced MSC differentiation might rescue bone loss in a disease state, ovariectomized adult 9-week-old female mice were treated with saline or with Bzb 2.5 weeks after ovariectomy. Nonovariectomized mice treated with Bzb showed the previously observed increase in bone mineral density (BMD), trabecular bone volume compared with saline-treated controls. As expected, ovariectomized mice experienced a decrease in BMD and trabecular volume. However, treatment with Bzb for 6 weeks (18 doses) partially rescued the osteoporotic phenotype: BMD, trabecular number, and trabecular density increased, and the mineralized architecture appeared similar to normal bone (Figure 7, A and B), demonstrating that Bzb-induced MSC differentiation may be therapeutically useful in this mouse model of tissue degeneration. Dynamic histomorphometry was performed to evaluate bone formation under these conditions. In sham-treated (i.e., nonovariectomized) mice, Bzb treatment caused an increase in bone formation rate (BFR) as previously observed. In ovariectomized mice, there was also a relative increase in BFR above baseline (consistent with prior studies that suggest increased osteoblast activity in the face of estrogen depletion; ref. 32). Notably, Bzb treatment further increased the BFR (per unit BSpm) in ovariectomized mice (Figure 7C). Similar trends were observed in mineral apposition rates, but the increase in mineral apposition in ovariectomized mice with Bzb was not statistically significant. Thus, Bzb increases bone formation and trabecular bone volume in the setting of osteoporosis.

In vivo treatment with Bzb.

C57BL/6 mice (female, 7 weeks; Charles River Laboratories) were treated with 0, 0.05, 0.125, or 0.3 mg/kg of Bzb (or saline control) 3 times a week — Mondays, Wednesdays, and Fridays — by i.p. injection. Peripheral blood was collected weekly in the morning and osteocalcin levels assessed by ELISA (Biomedical Technologies Inc.) according to the manufacturer’s instructions. Calcein double labeling was performed as previously described (40). For statistical analysis, values for serum osteocalcin at each point were normalized for every individual animal to its own osteocalcin level at week 0 (i.e., before treatment) and plotted as fold change. Differences in the groups were then analyzed by a 2-tailed Student’s t test. Animal care, experiments, and protocols were approved by the Institutional Animal Care and Use Committee, Massachusetts General Hospital.

Colony formation assay.

Spines from mice treated with saline or 0.3 mg/kg Bzb for 3 weeks (as above) were dissected and cleaned. A fixed mid-spine segment (approximately 4 cm) was crushed by mortar and pestle. After lysis of rbc, cells from 5 control and 5 treated mice were pooled and 1 × 10⁶ primary bone marrow cells from each group (isolated as described below) were plated in 24-well plates in a medium composed of αMEM, 20% fetal bovine serum (HyClone), and penicillin and streptomycin solution (CellGro) — henceforth referred to as α20%. Medium was changed at 24 hours to eliminate nonadherent cells. For CFU-F assay, cells were left in α20%; for CFU-Alk, cells were moved to osteogenic medium (see below) on day 3; for CFU-Adipo, cells were left in α20% for 6 days to allow initial colony formation and then switched to adipogenic medium (see below). After 16 days, colonies were assessed by methylene blue staining for the CFU-F assay, BCIP staining (alkaline phosphatase) for CFU-Alk, or oil red O staining for CFU-Adipo (see below). Only colonies containing a majority of histologically stained cells (>50%) were scored as positive.

Histomorphometric analysis (static and dynamic).

For histomorphometry, mice were treated with Bzb as above for the period specified. For experiments depicted in Figure 1, 7-week-old C57BL/6 female mice were used and treated for 3 weeks (9 doses). For experiments depicted in Figure 7, 9-week-old FVB/N female mice were either sham treated or ovariectomized and then housed for an additional 2.5 weeks (to allow for osteoporotic phenotype to develop) before starting Bzb treatment for 6 weeks. In both cases, at 10 days and 3 days before sacrifice, calcein was injected i.p. as previously described (40). Bones were fixed in 4% paraformaldehyde, and undecalcified sections embedded in plastic were sectioned at 8 μm. For static analysis depicted in Figure 1, sections were stained with toluidine blue. Images of the secondary spongiosa were obtained with a ×20 objective, and 2 images were obtained per section to cover the entire spongiosa. Osteoblasts were identified as mononuclear cells directly abutting either mineralized bone or osteoid and restricted within 1–3 cell diameters from the endosteal surface. For mineral apposition rates and BFRs, images of the secondary spongiosa were obtained as above, with simultaneous images in bright-field and under green fluorescence using a ×10 objective. Bright-field images were used to outline the BSpm, and then BSpm was calculated by ImageJ analysis (http://rsb.info.nih.gov/ij). Double-labeled areas (i.e., areas where both day –10 and day –3 calcein label was present) and single-labeled areas were also marked on the images. The average distance between double-labeled areas was calculated by ImageJ analysis using 3 measurements per double label. Mineral apposition rate (MAR) was calculated by dividing the average double-label distance (in μm) for each sample and dividing by the interlabel time of 7 days. To calculate BFR, the total length of double label and single label was calculated for each sample. BFR (with BSpm referent) was calculated as MAR × (0.5 single-label length + double-label length)/BSpm.

TRAP staining for osteoclasts.

Femurs fixed in 4% paraformaldehyde were decalcified in EDTA and then sectioned after paraffin embedding. After deparaffinization, TRAP staining was carried out per manufacturer’s protocol (387A-1KT; Sigma-Aldrich). Images of the secondary spongiosa were obtained as above, and the number of TRAP-positive cells abutting the endosteal surface was calculated for every section. BSpm was calculated by drawing lines manually on images and quantifying the length as above.

Isolation of primary bone marrow cells for MSCs.

C57BL/6 actin-GFP mice (003291; Jackson Laboratories) were sacrificed; tibiae, femurs, and spine were removed and excess soft tissue was eliminated. Using a pestle and mortar, the bones were crushed and washed in PBS with 0.5% FBS and passed through a 70-μm filter into a collection tube. The slurry was spun at 470 g for 5 minutes; the supernatant was removed, and cells were resuspended in a minimal volume of ACK lysing buffer (Cambrex) for 4 minutes on ice and washed once with PBS. After pelleting once again, the cells were resuspended and plated in α20% and incubated at 33°C with 5% CO₂. After 4 weeks of culture and expansion, CD105 isolation was performed by magnetic isolation (Dynabeads M-280 Streptavidin; Invitrogen) or MACS beads (Miltenyi Biotech) using an anti-mouse CD105 biotin antibody (clone MJ7/18; eBioscience) at 10 μg/ml. The CD105-positive cells were then maintained in α20% as before. Passage number for experiments has been indicated below.

Alkaline phosphatase staining.

Purified CD105⁺ cells (passages 3–5) were isolated by MACS beads using a biotinylated CD105 antibody and plated at 2 × 10³ cells/well in a 96-well plate (BD Biosciences) at 33° in osteogenic induction medium: α20% modified with glycerol 2-phosphate (2.16 mg/ml), 2-phospho-l-ascorbic acid (0.05 mg/ml), and dexamethasone (10 nM) (Sigma-Aldrich). After 18 hours of differentiation, alkaline phosphatase staining was carried out with BCIP/NBT solution (Sigma-Aldrich) per the manufacturer’s instructions.

BSP and Runx-2 expression analysis.

CD105⁺ cells were isolated from MSC cultures as described above and plated at 5000/cm² in 6-well plates. On day 1 after plating, they were switched to osteogenic medium containing Bzb at 1.5 nM versus carrier (0 nm) alone. After 30 hours at 33°C, cells were immediately lysed into TRIzol and were frozen for further analysis. In parallel experiments, the same cells were differentiated in osteogenic medium for 8 days and then switched into osteogenic medium containing 0 or 1.5 nM Bzb. At 30 hours, cells were again lysed into TRIzol as described above. RNA isolation, qPCR, normalization, and statistical analysis were as described below for embryonic fibroblasts except that transcript levels were normalized to GAPDH (primers were from Applied Biosystems; see below). For statistical analysis, the average dCt value (difference in amplification cycles required to detect the amplification product) and its standard deviation (sdCt) was used in a standard 2-tailed Student’s t test.

Von Kossa assay and staining.

CD105 cells isolated from MSC cultures (passage 2) were plated at 5000 cells/well in 96-well plates and differentiated in osteogenic medium as above at 33°C. At 8 days, cells were fixed and washed in water, and a 5% silver nitrate solution was added to the well under incandescent light for 20–45 minutes. After granules developed, the silver nitrate was removed and wells were washed with water to stop the reaction. Two images with a ×10 objective were obtained per well and the results quantified by ImageJ analysis.

In situ collagen I staining.

For in situ collagen I staining, cells treated for 5 days with Bzb or DMSO were washed with PBS/0.5% FBS and exposed for 15 minutes to a rabbit anti-mouse collagen I antibody (21286; Abcam) diluted 1:200 in PBS. After incubation, the wells were washed 3 times and the secondary antibody, Alexa Fluor–conjugated anti-rabbit IgG, was added at 1:200. Wells were again washed 3 times and finally fixed with 2% PFA for 10 minutes prior to visualization by microscopy. Brightly positive collagen I clusters were quantified for each well.

Adipogenesis.

For adipogenic differentiation, CD105⁺ cells (passages 3–4) were plated at 1 × 10⁴ cells/well in a 48-well plate (BD Biosciences) in an adipogenic induction medium: DMEM–low glucose (CellGro) with l-glutamine and penicillin and streptomycin solution (CellGro), 10% fetal bovine solution (HyClone), dexamethasone (50 μM), 3-isobutyl-1-methylxanthine (0.5 mM), insulin (10 μg/ml), and indomethacin (100 μM; Sigma-Aldrich) at 33°C. After 4 days of induction, cells were switched to DMEM with 10% FCS containing 5 μM rosiglitazone and insulin (10 μg/ml) and dexamethasone (50 μM), again maintaining Bzb concentration indicated. Oil red O staining was carried out on day 5 per the manufacturer’s instructions (Electron Microscopy Sciences).

Implantation of MSCs in collagen sponges.

Three-dimensional collagen scaffolds (BD Biosciences) were soaked overnight in α20%. Cultured hMSCs were resuspended at 2 × 10⁶ in 30 μl PBS and pipetted directly on top of each scaffold and cultured for 7 days in MSC basal medium to generate MSC-implanted sponges. Another set of sponges was switched to osteogenic medium 1 day after implantation of cells and cultured for 7 additional days (osteoblast-implanted sponges). Beige nude XID mice (female, 6–7 weeks; Harlan Sprague) were anesthetized, and a small pocket was formed subcutaneously. A single scaffold was gently inserted into the pocket (30). The mice were then treated with Bzb or control saline, as described above, for 10 doses (3 doses/week, given on alternate days — Mondays, Wednesdays and Fridays — at 0.3 mg/kg i.p.).

Direct intrafemoral injection of MSCs.

C57BL/6 mice (Jackson Laboratories) were anesthetized, and a small incision was made over the right knee to gain access to the kneecap. A 27-gauge needle was used to drill a hole in the femur, and 20 μl of cultured GFP-positive MSCs at 5 × 10⁷ cells/ml was slowly injected into the cavity with a 26-gauge Hamilton Microliter Syringe (31). The mice were then treated with Bzb or control saline, as described above, for 6 doses (3 doses/week, given on alternate days — Mondays, Wednesdays and Fridays — at 0.3 mg/kg i.p.).

Isolation of Ops from Osx-GFP mice.

Femurs and spine from 6-week-old Osx-GFP mice were isolated and crushed with mortar and pestle. Supernatant was pooled and kept aside (to generate the first fraction). The bone fragments were digested with collagenase I at 37°C and 1 hour in a shaking water bath. Collagenase was inactivated with excess PBS containing 1% FCS and washing with medium containing 20% serum. The bone slurry was then triturated extensively to dislodge all remaining cells and filtered through a 70-μm filter; the material was pooled with cells from the first fraction. The cells were then treated as described for crushed bone marrow samples.

Osteoclast differentiation.

Bone marrow from 8-week-old C57BL/6 mice was harvested. 1 × 10⁵ mononuclear cells/well were plated in flat-bottom 96-well plates in α-MEM (CellGro) containing 10% FCS (HyClone). Osteoclastogenesis was induced by adding M-CSF (20 ng/ml, R+D) and RANKL (200 ng/ml, gift of Yongwon Choi, The Leonard and Marilyn Abramson Family Cancer Research Institute, University of Pennsylvania, Philadelphia, Pennsylvania, USA) in the presence of 1 nM or 2 nM Bzb. After 5 days, culture supernatants were analyzed for TRAP activity and cells were fixed and stained for TRAP cytochemistry (Sigma-Aldrich).

Immunoblotting for Runx-2.

Primary MSCs cultured with Bzb (10 nM) or lactacystin (20 μM) for 6 hours were harvested, washed, and lysed using RIPA buffer: 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 5 mM EDTA, 5 mM NaF, 1 mM Na₃VO₄, 1 mM PMSF, 5 μg/ml leupeptine, and 5 μg/ml aprotinin. Whole-cell lysates were subjected to SDS-PAGE, transferred to a PVDF membrane (Bio-Rad Laboratories), and immunoblotted with anti–Runx-2 (Calbiochem) and GAPDH (Santa Cruz Biotechnology Inc.) antibodies.

Micro-CT analysis of femurs.

Micro-CT analysis was carried out as described in Pierroz et al. (41). In brief, femurs were dissected, cleaned, fixed in 4% paraformaldehyde, loaded onto CT scanning devices, and imaged according to the protocol previously described (41).

hMSC culture.

For in vitro osteoblast and adipocyte differentiation, hMSCs (Cambrex) were maintained and differentiated following the manufacturer’s protocol. Cambrex cells are typically obtained at passage 3. For the quantitative alkaline phosphatase assay and osteoblast differentiation, hMSCs (passages 4–5) were plated in Optilux 96-well plates (BD Biosciences) at a concentration of 3.1 × 10³ cells/cm² in MSC growth medium (MSGM). Following an overnight incubation, the growth medium was replaced with osteogenic induction medium that contained Bzb or vehicle. Cells were cultured in the presence of Bzb or vehicle for 6 days, at which point osteoblast differentiation was assayed by alkaline phosphatase expression (described below). For adipogenesis, hMSCs were plated in MSGM at a concentration of 2 × 10⁵ cells/well of a 6-well plate. MSCs were maintained in MSGM until the cells reached confluency. Cells were then cultured for 3 days in adipogenic induction medium (Chemicon) in the presence of Bzb or vehicle. The cells were then cultured for 1 day in adipogenic maintenance medium per manufacturer’s protocol (Chemicon) followed by another 3-day culture period in adipogenic induction medium. At each medium change, Bzb or vehicle was added fresh to the cultures. Following the last culture period in adipogenic induction medium, adipogenesis was assessed by fixing cells with 10% neutral-buffered formalin (VWR) and staining with oil red O to visualize lipid droplets.

Luciferase reporter assay for osteocalcin.

C3H10T1/2 cells (ATCC) were grown in DMEM with 10% FCS. Cells were plated into 12-well plates at 60,000 cells/well. One day after plating, cells were transfected with the indicated combinations of 6xOSE2-luciferase, pTK-RL, and Runx-2 expression constructs using Effectene transfection reagent (QIAGEN). 18 hours later, the medium was changed and the indicated concentration of Bzb was added. 24 hours later, cells were harvested and firefly/renilla luciferase activity was determined per the manufacturer’s instructions (Promega). Each experiment was performed in triplicate and repeated at least 3 times. 6xOSE2-luciferase and Runx-2 expression constructs and additional methods are described in ref. 27.

qPCR analysis of mouse embryonic cells.

Mouse embryonic mesodermal fibroblasts were isolated 12.5 dpc from WT and Runx-2^–/– mice as previously described (42). In brief, individual embryos were homogenized using an 18-gauge syringe in the presence of 0.25% trypsin and 1 mM EDTA (Invitrogen). The trypsin was inactivated after 15 minutes by addition of DMEM supplemented with 15% fetal bovine serum (Hyclone), 2 mM l-glutamine, 20 U/ml penicillin, and 20 μg/ml streptomycin. Cells were plated and allowed to adhere for 24 hours. For these experiments, WT and Runx-2^–/– cells were replated in 6-well plates at densities of 8 × 10⁴ and 5 × 10⁴ cells/well, respectively. Because Runx-2^–/– osteoblasts proliferate more quickly than WT osteoblasts, these seeding densities allowed cells to reach confluency at the same time (39). After overnight adherence, cells were treated with 1.5 nM Bzb. DMSO (1.5 μl/ml) was used for control cells. When cells were near confluency (day 4), the medium was replaced with an osteogenic medium (αMEM supplemented with 15% FBS [SH30070.03; HyClone], 2 mM l-glutamine, 20 U/ml penicillin, 20 μg/ml streptomycin, 25 μg/ml ascorbic acid, and 10 mM β-glycerophosphate). Cells were harvested 5 days after plating, and RNA was isolated using TRIzol (Invitrogen) according to the manufacturer’s protocol. Oligo-dT primers were used in conjunction with the SuperScript First-Strand Synthesis System (Invitrogen) to synthesize cDNAs. mRNA levels of osteoblast-related genes were analyzed by real-time qPCR using the Power SYBR Green Master Mix (Applied Biosystems) on an ABI PRISM 7000 sequence detection system (Applied Biosystems). Primer sequences are as follows: Alp: forward, TTGTGCGAGAGAAAGAGAGAGA, reverse, GTTTCAGGGCATTTTTCAAGGT; BSP: forward, GCACTCCAACTGCCCAAGA, reverse, TTTTGGAGCCCTGCTTTCTG; Col1a1: forward, CCCAAGGAAAAGAAGCACGTC, reverse, AGGTCAGCTGGATAGCGACATC; mCox: forward, ACGAAATCAACAACCCCGTA, reverse, GGCAGAACGACTCGGTTATC; and Runx-2: forward, CGGCCCTCCCTGAACTCT, reverse, TGCCTGCCTGGGATCTGTA. Transcript levels were normalized to mitochondrial cytochrome c oxidase subunit II (mCox) or GAPDH and are expressed relative to WT control cells.

API.

To determine the API, cell numbers were first established by culturing cells in medium containing Alamar Blue (Biosource) for 4 hours at 37°C. Plates were read on a fluorimeter at 570 nm. Medium containing Alamar Blue was removed, and cells were washed once with sterile PBS. Cells were then incubated with alkaline phosphatase substrate (Sigma-Aldrich) for 1 hour at room temperature. The plate was read at 405 nm. Alkaline phosphatase levels were then normalized to cell number to establish API (API = alkaline phosphatase fluorimeter reading/Alamar Blue fluorimeter reading × 1000).

Ovariectomy and Bzb treatment.

Nine-week-old FVB/N female mice were ovariectomized by a commercial vendor (Charles River Laboratories). Bzb treatment was initiated an additional 2.5 weeks later and given i.p. at 0.3 mg/kg Mondays, Wednesdays, and Fridays for a total of 6 weeks.

Osteoclast resorption assays.

Osteoclast resorption was carried out as described (43). Resorption pits were stained and imaged with a ×20 objective and average pit size calculated by ImageJ analysis of the stained sections.

Statistics.

In all cases, analysis was performed by a standard 2-tailed Student’s t test. The maximum number of independently compared variables was limited to 2 for any experiment. All data have been plotted as average ± SEM.