Osteoactivin acts as downstream mediator of BMP-2 effects on osteoblast function

Authors

  • Samir M. Abdelmagid,

    1. Department of Anatomy and Cell Biology, School of Medicine, Temple University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Mary F. Barbe,

    1. Department of Anatomy and Cell Biology, School of Medicine, Temple University, Philadelphia, Pennsylvania
    2. Department of Physical Therapy, College of Health Professions, Temple University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Israel Arango-Hisijara,

    1. Department of Anatomy and Cell Biology, School of Medicine, Temple University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Thomas A. Owen,

    1. Department of Cardiovascular and Metabolic Diseases, Pfizer Global Research and Development, Groton, Connecticut
    Search for more papers by this author
  • Steven N. Popoff,

    1. Department of Anatomy and Cell Biology, School of Medicine, Temple University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Fayez F. Safadi

    Corresponding author
    1. Department of Anatomy and Cell Biology, School of Medicine, Temple University, Philadelphia, Pennsylvania
    • Associate Professor, Department of Anatomy and Cell Biology, Temple University School of Medicine, 3400 North Broad Street, Philadelphia, PA 19140.
    Search for more papers by this author

Abstract

Our laboratory previously showed that osteoactivin (OA) is a novel, osteoblast-related glycoprotein that plays a role in osteoblast differentiation and function. The purpose of this study was to examine the regulation of OA expression by BMP-2 and the role OA plays as a downstream mediator of BMP-2 effects in osteoblast function. Using primary osteoblast cultures, we tested different doses of BMP-2 on the regulation of OA expression during osteoblast development. To test whether Smad-1 signaling is responsible for BMP-2 regulation of OA expression, osteoblast cultures were transfected with Smad1 siRNA, treated with 50 ng/ml of BMP-2 and analyzed by Western blot. BMP-2 treatment increased OA mRNA and protein expression in a dose-dependent manner and this upregulation was blocked in Smad1 siRNA transfected cultures. We next examined whether the role of OA as a downstream mediator of BMP-2 effects on osteoblast differentiation and matrix mineralization. Osteoblast cultures were transfected with OA antisense oligonucleotides and treated with 50 ng/ml of BMP-2. Cultures transfected with OA antisense oligonucleotides and treated with BMP-2 showed a reduction of OA expression associated with a significant reduction in early and late differentiation markers induced by BMP-2. Therefore, OA acts, at least in part, as a downstream mediator of BMP-2 effects on osteoblast differentiation and matrix mineralization. Our findings suggest that BMP-2 regulates OA expression through the Smad1 signaling pathway. Our data also emphasize that OA protein acts as a downstream mediator of BMP-2 effects on osteoblast differentiation and function. J. Cell. Physiol. 210: 26–37, 2007. © 2006 Wiley-Liss, Inc.

Osteoactivin (OA) is a novel factor that was initially identified from studies using an animal model of Osteopetrosis (op), the mutation in rats. Using the technique of mRNA differential display, the expression of OA cDNA was highly upregulated in op compared to normal bone (Safadi et al., 2002). OA has high homology to human glycoprotein nmb (gpnmb) (Watermann et al., 1995), and mouse DC-HIL (dendritic cell-associated, heparan sulfate proteoglycan dependent-integrin ligand) (Shikano et al., 2001). It has two isoforms, one is transmembrane type I with a MW of 65 kDa and the other is a secreted glycoprotein with MW of 115 kDa (Safadi et al., 2002). OA was found to be highly expressed in various malignant tumors such as in glioma (Loging et al., 2000), and hepatocellular carcinoma (Onaga et al., 2003). It has been shown that overexpression of OA in glioma cell lines (Rich et al., 2003), as well as in hepatoma cell lines (Onaga et al., 2003), permits tumor invasiveness. The OA protein has been found to modulate osteoblast differentiation and function in vitro by stimulating osteoblast differentiation markers, including alkaline phosphatase (ALP) activity, nodule formation, osteocalcin production, and matrix mineralization, without affecting cell proliferation or viability (Selim et al., 2003).

Bone morphogenetic proteins (BMPs) are secreted growth factors, which form a subgroup of the transforming growth factor (TGF-β) superfamily based on amino acid homology of a highly conserved seven-cysteine domain in the carboxy-terminal region of the proteins (Kingsley, 1994; Eimon and Harland, 1999). BMPs were originally known by their ability to induce ectopic bone and cartilage formation in vivo (Urist, 1965), but recently it became evident that BMPs also act as multifunctional regulators in morphogenesis during development in vertebrates (Wozney, 1998). BMP dimers initiate signaling by binding to both type I and type II serine/threonine kinase receptors and the phosphorylation of type I receptors upon ligand binding (Miyazono et al., 2000). Receptor-regulated Smads (R-Smads) (Smad 1, 5, 8) are activated by type I receptors (BMPR-IA or BMPR-IB) (Kawai et al., 2000), associate with Smad4, and translocate to the nucleus where they interact with transcription factors to regulate the transcription of target genes.

It is known that BMP proteins initiate the cascade of endochondral bone formation where mesenchymal stem cells differentiate into chondrocytes which lay down cartilage that is replaced by bone tissue (Reddi, 1994). BMPs can also act as local factors in the regulation of osteoblast differentiation (Katagiri et al., 1994). Several BMP knockout experiments in mice have contributed to elucidate the role of BMPs in bone formation and development. For example, BMP-2 deficient mice had amnion/chorion malformation and defects in cardiac development, and died during embryonic development (Zhang and Bradley, 1996b). Number of studies have shown that BMP-2, -3, -4,and -7 can upregulate differentiation markers of the mature osteoblast, including short term such as ALP activity, and long-term surrogates such as osteocalcin expression (Zhou et al., 1993). In addition, studies have demonstrated increased expression of osteoblast markers in pluripotent mesenchymal stem cell cultures after stimulation with BMP-2, suggesting that BMPs may regulate specific differentiation pathways in uncommitted cells (Wang et al., 1993).

The similarity in the temporal expression patterns of BMP-2 and OA during osteoblast differentiation and the fact that both of these factors play a role in osteoblast function in vitro led us to examine the relationship between BMP2 and OA in osteoblasts. In this study, we were interested in determining whether BMP-2 regulates the expression of OA through the Smad-1 signaling pathway, and whether OA acts as a downstream mediator of BMP-2 effects on osteoblast development and function.

MATERIALS AND METHODS

Antibodies

An anti-OA antibody was raised against peptide corresponding to amino acids 551–568 of the rat OA protein. This peptide was selected on the basis of its potential antigenicity and screened against the protein sequence databases to assure a lack of cross-reactivity to other proteins. Chickens were immunized with this peptide and the precipitated crude IgY was purified by affinity chromatography on Sepharose 4B derivatized with the immunizing peptide (CRB, Billingham, UK). Rabbit anti-Smad1 was purchased from Upstate (Lake Placid, NY); mouse anti-BMP2 was purchased from R&D Systems (Minneapolis, MN); and HRP-conjugated donkey anti-chicken IgY and HRP-conjugated goat anti-mouse IgG were purchased from Jackson Immunoresearch (West Grove, PA).

Primary osteoblast culture

Primary osteoblasts were isolated as described previously (Safadi et al., 2002) with modifications. Neonatal rat pups (1–4 days) were decapitated and their heads were swabbed with 70% ethanol. After a midline incision, the calvaria were isolated and placed in a Petri dish with 20 ml isolation media [phosphate buffered saline (PBS) +1% penicillin/streptomycin + Hank's media (Sigma-Aldrich, St. Louis, MO)]. After removal of the dura, each calverium was cut along the sagittal and coronal sutures and all pieces transferred to another Petri dish with 20 ml isolation media before being cut into smaller pieces. The pieces were then transferred into a 50 ml siliconized Ehrlenmeyer flask with digest media (PBS + 0.1% collagenase P + 0.25% trypsin). The flask was placed in a shaker bath at 37°C for 5 min. After discarding the supernatant, 10 ml of digestion media was added and the pieces were again cut into smaller fragments. The bone pieces were shaken for 15 min at 37°C. The supernatant was then filtered through a 200 µm mesh metal screen filter (Fisher Scientific-Newark, DE, Millipore filter and screen), split into two tubes, each with 5 ml washing media (Hank's media +1% penicillin/streptomycin +10% fetal bovine serum), and centrifuged for 5 min at 1,200 rpm at 4°C. The supernatant was transferred into two tubes as in the first digestion. The same procedure was repeated again for the third digestion. The cell pellets were re-solubilized into 5 ml of fresh washing media. Fifty microliters were removed and added to another 50 µl of Trypan blue, and the cells counted using a hemacytometer. Cells were then plated in 100 mm cell culture at a density of 500,000 cells with 10 ml initial plating medium (EMEM (Mediatech-Cellgro, Kansas City, MO) +1% penicillin/streptomycin +10% fetal bovine serum) and incubated at 37°C with 5% CO2. To induce osteoblast differentiation, cells were treated with 10 mM β glycerol phosphate +50 µg/ml ascorbic acid on Day 3 and every time the culture media was changed.

Treatment with recombinant BMP-2

Primary osteoblasts were cultured in six-well plates at a density of 50,000 cells/well, rinsed with Hank's medium and treated with different doses (10, 25, 50, 100, and 200 ng/ml) of recombinant BMP-2 (Sigma-Aldrich) depending on the experiment protocol for 24 h in serum free condition before assessment of OA protein expression by Western blot analysis.

RNA isolation

RNA was isolated as described previously (Safadi et al., 2002). Briefly, cell cultures were harvested and frozen at −80°C. Cells were then homogenized in Trizol, separated into organic and aqueous layer by chloroform, and RNA was recovered from the aqueous layer by isopropyl alcohol precipitation. Pellets were washed with 70% ethanol to clean RNA from DNA contamination. Concentration of RNA was calculated using spectrophotometer and RNA integrity was checked on a 1% agarose/paraformaldehyde minigel stained with ethidium bromide.

RT and real-time PCR analysis

RT-PCR analysis for OA and G3PDH was performed as follows. Two micrograms of total RNA isolated from the cell layer were reverse transcribed to cDNA at 42°C for 50 min in a volume of 20 µl containing the following components: 1× first strand buffer (5× = 250 mM Tris, pH = 8.3, 375 mM KCl, and 15 mM MGCl2), 0.5 mM dNTP mix, 10 mM dithiothreitol (DTT), 0.5 µg oligo (dT) and 20 U Superscript II (RNase H free reverse transcriptase) (Invitrogen, Carlsbad, CA). The reaction was stopped at 70°C for 15 min, and 1 U RNase H was added to the mixture followed by incubation at 37°C for another 10 min to degrade the RNA. Two microliters aliquots of the generated cDNA was amplified in 50 µl of PCR reaction mixture containing 1 nM primers, 10 µl 10× advantage buffer, 10 nM dNTP mix, 1 µl DMSO and 1 µl advantage polymerase mix (Clonetech, Mountain View, CA). The primers for OA were sense; 5 CCAGAAGAATGACCGGAACTCG 3 and antisense 5 CAGGCTTCCGTGGTAGTGG 3 and the primers for G3PDH were sense; 5 ACCACAGTCCATGCCATCAC 3 and antisense 5 TCCACCACCCTGTTGCTGTA 3. PCR was performed on ABI PRISM 7700 (Applied Biosystems, Foster City, CA) using the Cyber Green method. Quantification of OA expression level was defined as a ration of OA/G3PDH.

Transfection of OA antisense oligonucleotide

Primary osteoblasts were cultured in six-well plates as described above. Confluent cells (60%, Day 2 in culture) were transfected with different doses (0.25, 0.5, and 1 µM) of OA antisense depending on the experimental protocol or 0.5 µM sense oligonucleotides using Lipofectamine 2000 (Invitrogen). To assess the effect of OA antisense on mineralization (Day 21), cultures were treated on Day 14 with a second dose of OA antisense. The sequence of OA antisense oligonucleotide was 5′-CCCTAGTCCCATCCACCAGG-3′ and the sequence of sense oligonucleotide was 5′-GGGCGTCTCTGAAAGGTAACG-3′. The sequence of the OA antisense oligo was analyzed by Blast search. No homologies other than OA were found in the database. Transection efficiency was determined using BLOCK-iT fluorescent oligonucleotides (Invitrogen). The primary osteoblast cultures were transfected with fluorescent oligo as described above, and then transfection efficiency was evaluated by counting fluorescent labeled cells versus total number of cells. A transfection efficiency of 65% was reproducibly achieved.

Transfection of Smad1 siRNA

Primary osteoblasts were cultured in six-well plate as described above and were transfected on Day 2 with different doses (10–100 nM) of Smad1 siRNA (Santa Cruz, CA) depending on the experimental protocol or 50 nM non-silencing siRNA (Santa Cruz) using Lipofectamine 2000 (Invitrogen) as a vehicle. After 4 hours of transfection, 10% serum + EMEM was added to transfection medium and the cells were incubated at 37°C in CO2 for 72 h until cells were assayed for transgene expression. Transfection efficiency was determined to be 65%.

Protein isolation

After culture termination, cells were rinsed with 10 ml ice cold PBS, then trypsinized with 0.25% trypsin and 2 mM EDTA in Hank's medium for 10 min. Cell layers were harvested and centrifuged at 4°C for 10 min at 1,200 rpm. Cells were lysed in 500 µl of RIPA buffer (50 mM Tris-HCl; pH 7.5; 135 mM NaCl; 1% Triton X-100; 1% sodium deoxycholate; 2 mM EDTA; 50 mM NaF; 2 mM sodium orthovanadate; 10 µg/ml aprotinin; 10 µg/ml leupeptin; and 1 mM PMSF). Samples were centrifuged at 14,000 rpm for 15 min at 4°C and total protein concentration was measured using bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL).

Western blot analysis

Twenty or forty micrograms of total protein isolated from primary osteoblast cultures were mixed with 2× sample buffer and heated at 100°C for 5 min to denature the proteins. Samples were subjected to 10% SDS–PAGE in 1× TGS (0.25 M Tris, 1.92 M glycine, and 1.0% SDS in ddH2O, pH 8.6) (Biorad, Hercules, CA) at 100 mV for 1 hour. Gel was then transferred to PVDF membrane by semi-dry transfer apparatus (Biorad) at 15 mV for 1 hour at room temperature. The blot was incubated in blocking buffer (5% skim milk +1% bovine serum albumin) for 1 hour at room temperature. Primary antibody was added to blocking buffer overnight at 4°C. The next day, the blot was washed 5 times in 1× TTBS (Tris buffered saline +0.1% Tween-20) (Biorad), 5 min each, on a shaker. The blot was then incubated with HRP-conjugated secondary antibody, for 1 hour at room temperature. The blot was washed in TTBS for five times, 5 min each time. Protein was visualized suing ECL kit (Pierce) and signals were detected using XL-exposure films.

Alkaline phosphatase histochemistry

Primary osteoblasts were cultured in 12-well plates. ALP staining was performed on Day 14 using ALP staining kit (Sigma-Aldrich). Briefly, cells were fixed with citrate-acetone-formaldehyde fixative for 1 min, then rinsed with dH2O. Alkaline dye mixture was added to the cells and incubated at room temperature for 15 min with protection from direct light. Cells were then rinsed with dH2O for 2 min before counterstaining with hematoxylin for 2 min. Cells were then rinsed with dH2O and allowed to air dry before evaluating with E600 Nikon inverted microscope.

von Kossa staining of mineralized nodules

von Kossa staining was used to stain mineralized matrix on Day 21 of culture. Cells were rinsed twice with Hank's Balanced Salt solution, and then fixed with 2% paraformaldehyde solution for 10 min at room temperature. The plates were then rinsed with dH2O, stained with 3% silver nitrate solution and were exposed to direct sunlight for 1 hour. Cells were then rinsed with dH2O, fixed by adding 5% sodium thiosulfate for 2 min and counterstained with 1% fast green and air dried. von Kossa staining was then evaluated using E600 Nikon inverted microscope.

MTT-cell viability assay

Primary osteoblasts were plated at 24-well plates at a density of 12,400 cells per well for 5 days and 24 hour prior to termination. MTT (3-[4,5-Dimethylthiazole-2-yl]-2,5-diphenyltetrazolium bromide), (Sigma-Aldrich), (5 mg/ml) substrate was added (100 µl/well) and cells were incubated for 4 hour in 37°C CO2. After the incubation period, solubilizer (20% SDS and 50% DMF) was added (250 µl/well) and plates were rocked at room temperature overnight to solubilize the formazen crystals. The next day, 100 µl aliquots were transferred into 96-well plate and samples were read on a plate reader at 570 nm.

Image analysis

Pictures for stained or non-stained mineralized nodules were taken from different fields using E600 Nikon inverted microscope. Images were analyzed using BIOQUANT 98 (Bioquant Image Analysis Corporation, Nashville, TN) image analysis software. Nodule number was computed using the object count feature. The size of the nodules was computed using the area measurement feature combined with the irregular region of interest (ROI) option of the BIOQUANT program. Nodule mineralization was computed using the video count area array option. Video count area is defined as the number of pixels in a field that meet a user-defined color threshold of staining multiplied by the area of a pixel at the selected magnification. In this case, color thresholds were selected based on mean level of von Kossa staining. Percent area fractions of von Kossa staining were calculated by dividing the video count area containing pixels at or above the defined threshold by the video count area of total number of pixels in the entire field, and multiplying by 100. This determination was made at four different locations per well, three wells per group.

Statistical analysis

For multiple group comparison, analysis of variance (ANOVA) was used to evaluate the effect of one variable on multiple independent groups. In the event of a significant group effect, individual pairs of means were compared using Newman–Keuls post hoc test. A P-value ≤0.05 was considered statistically significant. Group means + standard error of the mean (SEM) are plotted in graphs.

RESULTS

Induction of OA expression by BMP-2

It has been reported previously that osteoblasts in culture undergo three different stages of differentiation, proliferation (1–7 days), matrix maturation (7–14), and matrix mineralization (14–21) (Aronow et al., 1990; Safadi et al., 2003). We first examined the endogenous level of BMP-2 in osteoblasts during different stages of development in culture using Western blot analysis. BMP-2 expression levels were markedly elevated during the third week (the stage of matrix mineralization) compared to first and second weeks of culture (the stages of cell proliferation and matrix maturation, respectively) (Fig. 1A). Densitometric analysis shows a greater than twofold increase in BMP-2 expression level during third week of culture (Fig. 1B). We next examined the Smad1, 5, 8 expression and activity (phosphorylation) during primary osteoblast differentiation. Smad1, 5, 8 expression levels were not dramatically changed during the course of osteoblast differentiation, however, the phospho-Smad 1, 5, 8 levels were markedly increased during the third week compared to the first and the second week in cultures (Fig. 1A). Densitometric analysis shows a greater than twofold increase in phspho-Smad1, 5, 8 levels during the third week in culture (Fig. 1C).

Figure 1.

BMP-2 Induces OA expression in osteoblast culture. (A, B) Primary osteoblasts were cultured and terminated at the end of 1, 2, or 3 weeks of culture. A: Immunoblot shows endogenous level of BMP-2, Smad 1, 5, 8 and phospho-Smad1, 5, 8 expression reaching a maximum (for BMP-2 and Phospho-Smad1, 5 8) at 3 weeks of culture. β-actin was used as a loading control. B and C: Densitometry of three immunoblots quantifying percent of BMP-2 and phospho-Smad1, 5, 8, expression as a ratio of β-actin. Data presented as mean + SEM. **P < 0.01 compared to the first week in culture. D–F: Primary osteoblasts were cultured for 5 days before switching to serum-free media containing different doses of BMP-2 (0–50 ng/ml) for 24 hour before termination. D: immunoblot shows that OA expression reached maximum in cultures treated with 50 ng/ml BMP-2. β-actin was used as a loading control. E and F: Densitometry of three immunoblots quantifying percent of the glycosylated (E) and transmembrane (F) isoforms of OA protein expression over β-actin. Data presented as mean + SEM. *P < 0.05 and **P < 0.01 compared to untreated, control cultures.

Endogenous OA protein expression was also examined during osteoblast development and also showed maximal expression levels during osteoblast matrix mineralization stage at 3 weeks (Selim et al., 2003) and Figure 2.

Figure 2.

Regulation of OA expression by BMP-2 during osteoblast differentiation. Primary osteoblasts were cultured and switched to serum free media with (+) or without (−) 50 ng/ml BMP-2 for 24 hour before termination at the end of 1, 2, or 3 weeks of culture. A: Quantitative real-time PCR of RNA colleted at different time points (1, 2, 3 weeks) in cultures with (+) and without (−) BMP-2 treatment. Data are presented as the ratio of OA to G3PDH. BMP-2 significantly increased OA mRNA levels at all time points examined. Data presented as mean + SEM. **P < 0.05, &&P < 0.01 and ΨΨP < 0.01 when compared to untreated controls of each time points. B: immunoblot shows upregulation of two OA isoforms (115 and 65 kDa) in BMP-2 treated cultures compared to untreated controls. β-Actin was used as a loading control. C and D: Densitometry of three immunoblots quantifying percent of the glycosylated (glyco) (115 kDa) (C), and the transmembrane isoforms (65 kDa) (D) of OA protein as a ratio of β-actin. Data presented as mean + SEM. *P < 0.05 and **P < 0.01 when compared to untreated 1 week cultures. &P < 0.05 and &&P < 0.01 when compared to untreated 2-week cultures. ΨΨP < 0.01 when compared to untreated 3-week cultures.

To examine a possible relationship between BMP-2 and OA proteins and whether BMP-2 may induce OA expression in vitro, primary osteoblast cultures were treated with different doses (0–200 ng/ml) of BMP-2 for 24 hour in serum-free media and OA expression was assessed by real-time PCR (data not shown) and Western blot analyses. Levels of the transmembrane (65 kDa) and the glycosylated (115 kDa) OA isoforms were gradually increased in a dose-dependent manner, reaching maximum at a dose of 50 ng/ml and above (data not shown) of BMP-2 (Fig. 1D). Densitometric analysis shows four- to five fold increase in glycosylated and transmembrane OA isoforms, compared to untreated control (Fig. 1E,F). These data indicate that 50 ng/ml of BMP-2 can induce OA expression in vitro. Similar results were obtained using quantitative PCR (data not shown).

BMP-2 regulates OA expression during osteoblast differentiation

To examine the regulation of OA mRNA and protein expression by BMP-2 during osteoblast differentiation in vitro, osteoblasts were treated with 50 ng/ml of BMP-2 for 24 hour before termination of the culture at three time points during their course of development (1, 2, and 3 weeks). Quantitative real-time PCR analysis was performed showing that expression of OA mRNA levels was increased following BMP-2 treatment at 1, 2, and 3 weeks compared to untreated control (Fig. 2A). Using Western blot analysis, expression of OA glycosylated and transmembrane isoforms (MW 115 and 65 kDa, respectively) was increased after BMP-2 treatment during each stage of development compared to untreated controls (Fig. 2B). Densitometric analysis showed that expression of both the glycosylated (115 kDa) and transmembrane (65 kDa) OA isoforms were significantly increased after BMP-2 treatment (Fig. 2C,D). Collectively, these data demonstrate that BMP-2 upregulates OA mRNA and protein expression during different stages of osteoblast differentiation in culture. These data also show that endogenous levels of the glycosylated OA isoform increased while the transmembrane OA isoform decreased as the cultures terminally differentiated (Fig. 2B–D).

BMP-2 regulates OA expression through the Smad1 signaling pathway

To examine the mechanism by which BMP-2 regulates OA expression in osteoblast cultures, Smad1 siRNA oligonucleotides were used to downregulate Smad1 expression in culture. To select the appropriate dose of Smad1 siRNA, osteoblasts were transfected (Day 2 in culture) with different doses of Smad1 siRNA. Smad1 expression (Day 5 in culture) (Fig. 3A) was not inhibited at 25 nM of Smad1 siRNA, while 50 and 100 nM doses significantly inhibited Smad1 level expression, compared to untreated controls (data not shown) or controls transfected with 50 nM of non-silencing siRNA; both controls showed similar results. Densitometric analysis showed a greater than 50% reduction of Smad1 expression in cultures transfected with 50 nM, and a greater than 90% reduction in cultures transfected with 100 nM of Smad1 siRNA, compared to controls (Fig. 3B). To select the appropriate non-toxic dose of Smad1 siRNA, the same doses were used in cell viability MTT assay (Day 5 in culture). Smad1 siRNA at doses of 25 and 50 nM showed no inhibition in cell viability, while a dose of 100 nM siRNA had significant reduction of cell viability (Fig. 3C). These data suggest that Smad1 siRNA at a dose of 50 nM is the most appropriate for downregulating Smad-1 expression without affecting cell viability.

Figure 3.

Smad1 inhibition downregulates OA expression. A: Primary osteoblasts were cultured and then transfected with different doses of Smad1 siRNA (0–100 nM). A: Immunoblot (Day 5 in culture) shows Smad1 expression was inhibited by 50 and 100 nM doses of Smad1 siRNA compared to control untreated (not shown) or cultures treated with non-silencing siRNA. β-Actin was used as a loading control. B: Densitometry of three independent immunoblots quantifying percent of Smad1 expression as a ratio of β-actin. C: MTT assay for Smad1 siRNA at different doses. Primary osteoblasts were cultured and transfected with different doses of Smad1 siRNA and MTT assay for cell viability was performed (Day 5 in culture). Non-silencing siRNA (control) and Smad1 siRNA (50 nM dose) had no significant effect on osteoblast viability. Data presented as mean + SEM. *P <  0.05 and **P <  0.01 when compared to control. D–F: Primary osteoblasts were cultured then transfected with 50 nM Smad1 siRNA before switching to serum-free media with 50 ng/ml BMP-2 for 24 hour prior to termination (Day 5 in culture). D: BMP-2 treatment significantly increased Smad-1 and OA expression when compared to non-silencing siRNA transfected cultures. Smad-1 siRNA transfected cultures showed inhibition of both Smad-1 and OA protein expression levels when compared to non-silencing siRNA transfected control cultures. The Smad1 and OA expression levels induced by BMP-2 were blocked by Smad1 siRNA compared to BMP-2 treated cultures and to levels comparable to non-silencing siRNA transfected control cultures. β-Actin was used as a loading control. E and F: Densitometry of three independent immunoblots quantifying percent of Smad1 (E) or glyco-OA (F) protein expression as a ratio of β-actin. Data presented as mean + SEM. *P < 0.05 and **P < 0.01 when compared to non-silencing siRNA transfected controls; &P < 0.05 when compared to BMP-2 treated cultures.

To assess whether the regulation of OA expression by BMP-2 is mediated through Smad1 in cultured osteoblasts, cultures at 60% confluence were transfected with 50 nM of Smad1 siRNA and then received BMP-2 treatment at 50 ng/ml in the last 24 hour of a 5-day culture period. Smad1 and OA protein expression (Fig. 3D) were upregulated in BMP-2 treated cultures compared to untreated controls (Fig. 3D) or cultures transfected with 50 nM non-silencing siRNA. Inhibition of Smad1 expression by Smad1 siRNA showed an inhibition of OA expression in primary osteoblast cultures. BMP-2 treatment of Smad1 siRNA transfected cultures resulted in levels of Smad1 and OA protein expression that were similar to control levels (Fig. 3D). Quantitative analysis showed that BMP-2 treatment alone caused a significant increase in Smad-1 and OA protein expression (Fig. 3E,F). There was a significant reduction in Smad1 and OA expression, in Smad1 siRNA transfected cultures compared to controls (Fig. 3E,F), while BMP-2 treatment of these cultures resulted in Smad1and OA expression levels similar to the control cultures. Collectively, these data suggest that BMP-2 regulates OA expression, in part, through the Smad1 signaling pathway.

Downregulation of OA expression by OA antisense oligonucleotides

To examine the role of the OA protein in regulation of osteoblast differentiation with BMP-2, OA antisense oligonucleotides were used to downregulate OA expression in osteoblast cultures. For assessment of transfection efficiency in primary osteoblasts, cells were cultured for 2 days, transfected with 50 nM fluorescent-tagged scrambled oligonucleotides using Lipofectamine 2000. Cells with a green fluorescent signal (Fig. 4B) were counted and it was determined that ∼65% of osteoblasts were transfected (Fig. 4C).

Figure 4.

Downregulation of OA expression with OA antisense oligonucleotides. A–C: Primary osteoblasts were cultured and transfected with 50 nM fluorescent oligonucleotides for 4 hour. Oligo uptake was assessed and transfection efficiency determined to be ∼65%. Magnification: 300×. D and E: Primary osteoblasts were cultured and transfected with 0.25, 0.5, or 1 µM of OA antisense. D: Immunoblot shows both isoforms of OA expression were greatly inhibited by 0.5 and 1 µM doses of OA antisense when compared to OA sense transfected control cultures. β-Actin was used as a loading control. E: Densitometry of three immunoblots quantifying percent of the glycosylated isoform of OA expression as a ratio of β-actin. Data presented as mean + SEM (**P < 0.01 when compared to OA sense transfected control). F: An MTT assay for cell viability was performed after 5 days of culture. Control represents cultures transfected with 0.5 µM OA sense. 0.25 and 0.5 µM of OA antisense had no significant effect on osteoblast viability. Data presented as mean + SEM. *P < 0.05 when compared to sense transfected control. G, H: Primary osteoblasts were cultured, transfected with 0.5 µM OA antisense, treated at Day 14 with a second dose of OA antisense and then terminated at Day 21. G: Immunoblot shows inhibition of both isoforms of OA, compared to OA sense transfected controls. β-Actin was used as loading control. H: Densitometry of three immunoblots quantifying percent of glycosylated OA expression as a ratio of β-actin. Data presented as mean + SEM. **P < 0.01 compared to control.

To select the appropriate dose of the OA antisense oligonucleotides, primary osteoblasts were transfected with different doses of OA antisense and assessed by Western blot analysis (Day 5 in culture). OA expression was not inhibited by the 0.25 µM dose, while 0.5 and 1 µM dose resulted in a robust inhibition of OA expression, compared to untreated controls (data not shown) or controls transfected with 0.5 µM OA sense oligonucleotides (Fig. 4D,E). The OA expression levels were similar in untransfected cultures and cultures transfected with 0.5 µM OA sense oligonucleotides. To select the appropriate non-toxic dose of OA antisense oligonucleotides, MTT assay (Day 5 in culture) for cell viability was performed using 0.25, 0.5, and 1 µM of OA antisense (Fig. 4F) and both 0.25 and 0.5 µM doses of OA antisense had no effect, but 1 µM dose of OA antisense showed a significant reduction in cell viability. Therefore, a dose of 0.5 µM OA anitsense was used for all subsequent experiments.

To examine if OA antisense maintains inhibition of OA expression during terminal osteoblast differentiation (3 weeks in culture), cells were transfected on Day 2 of culture with 0.5 µM OA antisense and treated at Day 14 with a second dose of OA antisense. OA expression was assessed on Day 21 by Western blot analysis and was found to be dramatically downregulated compared to control cultures transfected with OA sense oligonucleotides (Fig. 4G). Quantification showed a greater than 70% reduction of OA expression level in OA antisense transfected osteoblasts compared to sense transfected controls (Fig. 4H). These data suggest that 0.5 µM of OA antisense is the most appropriate dose for oligo-transfection resulting in significant downregulation of OA expression without affecting cell viability.

Downregulation of OA expression inhibits BMP-2-induced early osteoblast differentiation

To examine whether OA is a downstream mediator of BMP-2 effects on early osteoblast differentiation in culture, primary osteoblasts were treated with either BMP-2 alone (data not shown) or BMP-2 transfected with sense, cells transfected with 0.5 µM of OA antisense alone OA, and cell transfected with 0.5 µM of OA antisense then treated with 50 ng/ml of BMP-2 for 24 h before culture termination at days 5 (data not shown) and 14 (Fig. 5). Cultures transfected with sense only served as control. By Western blot analysis, OA expression was highly upregulated in BMP-2 treated cultures, transfected with sense, compared to sense transfected control cultures. Cultures transfected with OA antisense showed dramatic decrease in OA expression when compared to sense transfected control cultures. The increase in OA expression induced by BMP-2 was blocked in OA antisense transfected cultures (Fig. 5A). BMP-2 transfected with sense control showed similar results to BMP-2 alone treated cultures for all subsequent experiments (data not shown). Densitometric analysis showed that BMP-2 treatment with sense transfection significantly increased OA expression, however, OA antisense alone significantly decreased OA expression, while BMP-2 treatment of OA antisense transfected cultures demonstrated OA levels that were comparable to sense transfected, control cultures (Fig. 5B).

Figure 5.

Downregulation of OA expression inhibits BMP-2-induced alkaline phosphatase activity. A and B: Primary osteoblasts were cultured, transfected with OA sense and treated with BMP-2 or transfected with OA antisense (0.5 µM) alone or with 50 ng/ml of BMP-2 for 24 hour prior to termination at Day 5 (data not shown) and Day 14. A: Immunoblot shows BMP-2 treatment increased OA expression and this increase was blocked with OA anti-sense to levels comparable to sense transfected controls. β-Actin was used as a loading control. B: Densitometry of three independent immunoblots quantifying percent of glyco-OA expression over β-actin. Data presented as mean + SEM. *P < 0.05 when compared to OA sense transfected controls; &P < 0.05 when compared to BMP-2 treated cultures. C–F: Primary osteoblasts were cultured, treated as above and terminated for alkaline phosphatase (ALP) staining at Day 14. Photomicrographs of alkaline phosphatase staining (purple). BMP-2 treatment (D) shows intense staining for ALP compared to sense transfected controls (C). OA anitsense transfected cells shows less ALP staining compared to sense transfected cells (E), while OA antisense transfected cells treated with BMP-2 show less alkaline phosphatase staining compared to BMP-2 treated cultures and to levels comparable to OA sense transfected cultures (F). Low power photomicrograph magnification: 60×; inset magnification: 300×. (G) Bioquant analysis of three independent experiments quantifying percent area fraction of the field occupied by ALP staining. Data presented as mean + SEM (*P < 0.05, **P < 0.01 when compared to sense transfected controls, &P < 0.05 when compared to BMP-2 treated cultures).

To determine whether the effects of BMP-2 on alkaline phosphates (ALP) activity is OA-dependent, primary osteoblasts were cultured, transfected with 0.5 µM of OA antisense and treated with 50 ng/ml of BMP-2 for 24 hour before termination on Day 14. BMP-2 treatment in cells transfected with sense greatly enhanced ALP staining (Fig. 5D) compared to OA sense trasnfected control cultures (Fig. 5C). OA antisense transfected cultures showed dramatic decrease in ALP staining (Fig. 5E). In contrast, ALP staining was reduced to control levels in OA antisense transfected cultures treated with BMP-2 (Fig. 5F).

The percent of area fraction of ALP staining was measured and calculated for each condition using computerized bioquantification (BIOQUANT) software. BMP-2 treated cultures showed a significant increase in percent ALP area fraction staining (Fig. 5G) compared to controls. OA antisense trasnfected cultures showed as significant decrease in percent ALP area fraction staining. In contrast, the increase in area fraction of ALP staining induced by BMP-2 was decreased in OA antisense transfected cultures resulted in a percent ALP area fraction similar to control levels (Fig. 5G). These data suggest that OA can act, at least in part, as a downstream mediator of BMP-2 effects on early osteoblast differentiation and that the induction of ALP activity by BMP-2 is OA-dependent.

Downregulation of OA expression inhibits BMP-2-induced nodule formation and mineralization

To examine whether downregulation of OA expression by OA antisense has any effect on BMP-2-induced ostoeblast nodule formation and matrix mineralization, primary osteoblasts were either treated with BMP-2 alone (data not shown), transfected with sense and then treated with BMP-2 or transfected with 0.5 µM of OA antisense at Day 2 in culture, treated with second dose of OA antisense at Day 14 only or treated with BMP-2 and terminated at Day 21 for the measurement of OA expression, nodule formation and matrix mineralization. By Western blot analysis, OA expression was upregulated by BMP-2 treatment compared to cultures transfected with OA sense oligonucleotides (Fig. 6A). This increase in OA expression induced by BMP-2 was inhibited in OA antisense transfected cultures (Fig. 6A,B).

Figure 6.

Downregulation of OA expression inhibits BMP-2-induced matrix mineralization. Primary osteoblasts were cultured, transfected with sense and treated with BMP-2 or transfected with OA antisense (0.5 µM) and treated with a second dose of OA antisense at Day 14 in culture with and without 50 ng/ml BMP-2 for 24 hour prior termination and von Kossa staining at Day 21. A: Immunoblot shows upregulation of OA expression by BMP-2 treatment. Treatment with BMP-2 + OA antisense resulted in inhibition of this upregulation to levels comparable to OA sense transfected controls. β-Actin was used as a loading control. B: Densitometry of three independent immunoblots quantifying percent of the glycosylated OA isoform expression as a ratio of β-actin. Data presented as mean + SEM (*P < 0.05 when compared to OA sense transfected controls, &P < 0.05 when compared to BMP-2 treated cultures). C–F: Photomicrographs of von Kossa staining (black). BMP-2 treatment (D) shows intense staining for mineral compared to OA sense transfected controls (C). OA antisense transfected shows dramatic decrease in von Kossa staining (E), while OA antisense transfected cells treated with BMP-2 shows less von Kossa staining levels compared to BMP-2 treated cultures and similar to OA sense transfected cultures (F). Low power photomicrograph magnification: 60×. G: Bioquant analysis of three independent experiments quantifying percent area fraction of the field occupied by von Kossa staining, an indicator of mineralization. Data presented as mean + SEM (*P < 0.05, **P < 0.01 when compared to sense transfected controls, &P < 0.01 when compared to BMP-2 treated cultures).

von Kossa staining was used to stain minerals. Cultures treated with BMP-2 (Fig. 6D) showed larger areas of mineral staining compared to OA sense transfected control cultures (Fig. 6C), while cultures transfected with OA antisense alone showed a dramatic decrease in mineral staining (Fig. 6E). In contrast cultures transfected with OA antisense then treated with BMP-2 showed a reduction in mineralization compared to BMP-2 treatment alone (compare Fig. 6F with D).

Mineralization (percent of area fraction of von Kossa), nodule size (area) and nodule number were measured and calculated for each condition using BIOQUANT software. BMP-2 treated cultures showed a threefold increase in percent von Kossa area fraction staining (Fig. 6G), as well as significant increases in nodule size and number (Table 1) compared to OA sense transfected controls cultures. The treatment of OA antisense transfected cultures with BMP-2 resulted in a percent von Kossa area fraction staining comparable to OA sense transfected control cultures (Fig. 6G). These data clearly demonstrate that OA acts, at least in part, as a downstream mediator of BMP-2-induced nodule formation and matrix mineralization in primary osteoblasts culture.

Table 1. Nodule count and average nodule size in each condition tested
ConditionNodule count mean ± SEM (% control)Average nodule size (µm) mean ± SEM (% control)
  • Numbers represent mean ± SEM of three independent experiments.

  • *

    P < 0.05 when compared to BMP-2 treated cultures.

  • **

    P < 0.01 when compared to OA sense transfected controls. Control = 100%.

Control34 ± 2.3 (100)3.4 ± 0.25 (100)
BMP-245 ± 3.7** (134)5.4 ± 0.14** (152)
OA-antisense20 ± 2.6** (62)1.4 ± 0.2** (44)
BMP-2 + OA antisense28 ± 2.4* (84)3.1 ± 0.3* (93)

DISCUSSION

As previously described, OA is homologous to other family members of trans-membrane proteins such as GPNMB (Wetermann et al., 1995), DC-HIL (Shikano et al., 2001), PMEL17, and human growth factor inducible neurokinin (HGFIN) (Metz et al., 2005). These family members play a role in differentiation of multiple cell types, such as DC-HIL in dendritic cells (Shikano et al., 2001), PMEL17 in melanocytes (Berson et al., 2001), and HGFIN in differentiation of lymphohematopoietic stem cells (Bandari et al., 2003). In this report, we examined the regulation of OA expression by BMP-2 during osteoblast differentiation and whether OA modulates BMP-2 effects on early and terminal osteoblast differentiation. The stages of osteoblast development have been well characterized in numerous previous studies. Primary osteoblasts in vitro undergo three distinct stages beginning with cell proliferation (Days 0–7), followed by nodule formation, collagen deposition and matrix maturation (Days 7–14), and ending with osteoblast differentiation and matrix mineralization (Days 14–21) (Aronow et al., 1990; Safadi et al., 2003). The fact that OA and BMP-2 have similar pattern of expression during osteoblast development in culture (see Figs. 1 and 2) and both of these factors have been shown to exhibit an overlapping effect in regulating osteoblast differentiation and function. (Urist, 1965; Selim et al., 2003). Data presented in this report suggest a relationship between OA and BMP-2 in regulating osteoblast function.

The OA protein has two isoforms, one is secreted (glycosylated at 115 kDa) and one is trans-membrane (native at 65 kDa) (Safadi et al., 2002). As primary osteoblasts develop in culture, the secreted isoform of OA reaches its highest level during the terminal differentiation of osteoblasts, while the trans-membrane isoform reaches its lowest levels during terminal differentiation (3 weeks in culture). In support of our findings, another group reported similar findings for DC-HIL, the mouse ortholog of OA (Shikano et al., 2001). They showed that in SX52, a long-term mouse dentritic cell line, DC-HIL is detected in both the cytosolic fraction that represents the secreted isoform and the membranous fraction that represents the transmembrane isoform. They also showed that the transmembrane isoform of DC-HIL mediates the adhesion of SVEC, mouse vascular endothelial cell line. Our group has previously shown that neutralizing the secreted isoform of OA using an anti-osteoactivin antibody inhibited osteoblast differentiation as evidenced by decreasing ALP activity, nodule formation, osteocalcin production, and matrix mineralization (Selim et al., 2003). Thus, these findings indicate possible dual roles for both isoforms of OA during osteoblast differentiation, an adhesion role for the transmembrane isoform and a differentiation role for the secreted isoform. However, more experiments are warranted to explore these possibilities.

The results in this study also showed that expression of both isoforms of OA are upregulated by BMP-2 in a dose-dependent manner, reaching maximum levels at 50 ng/ml of BMP-2. The fact that BMP-2 upregulates the expression of OA isoforms during the different stages of osteoblast development in culture suggests that OA may play a role as a downstream mediator of BMP-2 during osteoblast development in vitro. Other factors have been reported to be regulated by BMP-2 and are key regulators of osteoblast differentiation including, BIG-3 (Gori and Demay, 2005), Runx-2 (Karsenty and Wagner, 2002), and Osterix (Nakashima et al., 2002).

It has been well documented that BMP stimulation of osteoblast cell differentiation is mediated by heterotetrameric serine/threonine kinase receptors and the downstream transcription factors Smad1, 5, 8 (Sykaras and Opperman, 2003). We showed here that OA is regulated by BMP-2 and this regulation is mediated through the Smad-1 signaling pathway. Smad1 is an essential intracellular component that is specifically phosphorylated by BMP receptors and translocated into the nucleus upon ligand stimulation (Yang et al., 2000). Phosphorylation of Smad1 involves serines in the carboxy-terminal motif. These residues are phosphorylated directly by a BMP type I receptor in vitro. Mutation of these carboxy-terminal serines prevents Smad1 association with the related protein, accumulation in the nucleus, and gain of transcriptional activity (Kretzschmar et al., 1997). Transgenic mice expressing the Smad1 domain, termed Smad1C, show increased skeletal bone mineral density compared to their littermates. Bone histomorphometric analysis of transgenic mouse tibiae showed that Smad1C significantly increases trabecular bone area and length of trabecular surface covered with osteoid, and upregulates several osteoblast-related genes in cultured osteoblasts derived from Smad1C transgenic mouse (Liu et al., 2004). Targeted deletion of the Smad1 gene results in early embryonic lethality due to failure of the allantois to fuse to the chorion (Lechleider et al., 2001; Tremblay et al., 2001). In conclusion, our results demonstrate that BMP-2 signaling plays an important role in the regulation of OA expression. By close analysis of the OA promoter, multiple Smad1 binding motifs (CAGAC) (Lopez-Rovira et al., 2002) have been identified. These motifs are located in tandem, 1,442 base pairs upstream from the ATG starting codon (data not shown). Further analysis of the OA promoter by generating deletion constructs of the Smad1 binding motifs will clearly demonstrate the regulation of OA expression by BMP-2. Our study indicates that Smad1 inhibition in osteoblast cultures by Smad1 siRNA at a dose of 50 nM inhibited OA expression, a result that suggests regulation of OA expression by BMP-2 is mediated by (Smad 1) signaling.

We were also interested to examine whether the effects of BMP-2 on early and late osteoblast differentiation are OA dependent. In order to test this possibility, we used an antisense approach that has been shown previously to be effective in blocking different factors in osteoblast and other cell types (Bonnelye et al., 2001; Galindo et al., 2005). Using OA antisense oligonucleotides, we were able to inhibit OA expression significantly in cultures terminated at Day 14 and Day 21. We have also previously shown that neutralizing the constitutively secreted OA protein in primary osteoblast cultures with anti-OA antibody inhibited osteoblast differentiation (Selim et al., 2003). In this study, we demonstrated that short-term treatment of osteoblast cultures with BMP-2 under serum-free condition increased osteoblast ALP production (Day 14), nodule formation and mineralization (Day 21). Similar results were reported by Hay et al. (1999), where short-term treatment with BMP-2 in human neonatal primary osteoblasts stimulated cell differentiation markers. However, when OA expression was blocked in our cultures, BMP-2-induced early and late markers of differentiation were decreased to levels comparable to control. These data suggest that BMP-2-induced osteoblast function is, at least in part, OA dependent.

The mechanism whereby OA acts downstream of BMP-2 effects on osteoblast function is not fully understood. Our findings suggest that OA production is required for BMP-2-mediated osteoblast maturation and mineralization in primary osteoblast cultures. One possible mechanism is that a downregulation of OA expression inhibits osteoblast differentiation markers indirectly. The inhibition of OA expression could activate some BMP-2 antagonistic/inhibitory regulatory pathways that influence osteoblast differentiation. For example, the soluble BMP-2 antagonists such as, noggin, chordin, chordin-like, cerebrus, and gremlin bind BMP-s in the extracellular space and mask receptor binding interfaces for BMP type I and type II receptors (Balemans and Van Hul, 2002; Groppe et al., 2002). Another alternative mechanism whereby OA acts downstream mediator of BMP-2 effects on osteoblast maturation and terminal differentiation could be explained by the fact that the inhibitory effects of OA antisense oligonucleotides could decrease the expression/phosphorylation of regulatory Smads (1, 5, and 8) or increase the expression/phosphorylation of inhibitory Smads (Smad 6 and 7). The latter Smads inhibit signaling by either interacting with phosphorylated BMP type I receptors to prevent activation of receptor-activated Smads (Imamura et al., 1997; Nakao et al., 1997; Souchelnytskyi et al., 1998), or through competition to prevent formation of the receptor-activated Smad/co-Smad complex (Hata et al., 1998). Data from our laboratory showed that transfection of primary osteoblasts with OA antisense oligonucleotides resulted in a dramatic reduction in the amount and the phosphorylation levels of Smad1, 5, 8 and an increase in the amount and phosphorylation levels of Smad-7 (un-published observations), suggesting that OA acts, at least in part, as downstream mediator of BMP-2 actions on osteoblasts through modulating regulatory (Smad1, 5, 8) and inhibitory (Smad7) signaling molecules (Fig. 7).

Figure 7.

Schematic diagram of the relationship between BMP-2 and osteoactivin in osteoblasts. BMP-2 dimers bind to serine/threonine receptors and induce phosphorylation (P) of Smad1. P-Smad1 binds to Smad4 in multimeric complex then translocates into nucleus. P-Smad1 might activate the transcription of OA gene through binding to Smad1 response elements in the OA promoter. The secreted OA protein induces alkaline phosphatase activity, nodule formation and matrix mineralization. Smad1 siRNA inhibits the expression of OA through inhibition of Smad1 expression. OA antisense inhibits OA expression that resulting in blocking BMP-2-induced alkaline phosphatase activity, nodule formation, and matrix mineralization.

Another possibility is that downregulation of OA expression might stimulate other intracellular molecules, such as Smurf1 and Smurf2 (Smad ubiquinitation regulatory factors), which selectively target activated type I receptors and Smad proteins for degradation (Zhu et al., 1999; Kavsak et al., 2000; Zhang et al., 2001). Several transcription factors, such as Runx-2 (Gersbach et al., 2004), and growth factors, such as Wnt3a (Rawadi et al., 2003) modulate at least partially, the effects of BMP-2 on osteoblast differentiation and function. Similar results were presented where connective tissue growth factor (CTGF), a factor that plays a role in osteoblast differentiation in vitro and in vivo (Safadi et al., 2003). CTGF expression is regulated by TGF-β and acts as a downstream mediator of TGF-β-induced effects such as matrix production and differentiation of osteoblasts and other cell types (Qi et al., 2005; Arnott et al., 2006).

In this study, we have examined OA expression and its regulation by BMP-2 and have explored regulatory interaction between these two proteins. We also showed that the effects of BMP-2 on ALP activity, nodule formation and matrix mineralization in osteoblasts are partially mediated through OA protein. Further dissection of the relationship between BMP-2 and OA in osteoblasts will lead to a better understanding of the role of OA in osteoblast differentiation in vitro and bone formation in vivo.

Ancillary