Bone Morphogenetic Protein 2 Induces Cyclo-oxygenase 2 in Osteoblasts via a Cbfa1 Binding Site: Role in Effects of Bone Morphogenetic Protein 2 In Vitro and In Vivo

JOURNAL OF BONE AND MINERAL RESEARCH Volume 17, Number 8, 2002 © 2002 American Society for Bone and Mineral Research

Bone Morphogenetic Protein 2 Induces Cyclo-oxygenase 2 in Osteoblasts via a Cbfa1 Binding Site: Role in Effects of Bone Morphogenetic Protein 2 In Vitro and In Vivo DAICHI CHIKAZU,1 XIAODONG LI,1 HIROSHI KAWAGUCHI,2 YOKO SAKUMA,1 OLGA S. VOZNESENSKY,1 DOUGLAS J. ADAMS,3 MANSHAN XU,1 KAZUTO HOSHI,2 VEDRAN KATAVIC,1 HARVEY R. HERSCHMAN,4 LAWRENCE G. RAISZ,1 and CAROL C. PILBEAM1

ABSTRACT We tested the hypothesis that induction of cyclo-oxygenase (COX) 2 mediates some effects of bone morphogenetic protein (BMP) 2 on bone. BMP-2 induced COX-2 mRNA and prostaglandin (PG) production in cultured osteoblasts. BMP-2 increased luciferase activity in calvarial osteoblasts from mice transgenic for a COX-2 promoter-luciferase reporter construct (Pluc) and in MC3T3-E1 cells transfected with Pluc. Deletion analysis identified the ⴚ300/ⴚ213-bp region of the COX-2 promoter as necessary for BMP-2 stimulation of luciferase activity. Mutation of core-binding factor activity 1 (muCbfa1) consensus sequence (5ⴕ-AACCACA3ⴕ) at ⴚ267/ⴚ261 bp decreased BMP-2 stimulation of luciferase activity by 82%. Binding of nuclear proteins to an oligonucleotide spanning the Cbfa1 site was inhibited or supershifted by specific antibodies to Cbfa1. In cultured osteoblasts from calvariae of COX-2 knockout (ⴚ/ⴚ) and wild-type (ⴙ/ⴙ) mice, the absence of COX-2 expression reduced the BMP-2 stimulation of both ALP activity and osteocalcin mRNA expression. In cultured marrow cells flushed from long bones, BMP-2 induced osteoclast formation in cells from COX-2ⴙ/ⴙ mice but not in cells from COX-2ⴚ/ⴚ mice. In vivo, BMP-2 (10 ␮g/pellet) induced mineralization in pellets of lyophilized collagen implanted in the flanks of mice. Mineralization of pellets, measured by microcomputed tomography (␮CT), was decreased by 78% in COX-2ⴚ/ⴚ mice compared with COX-2ⴙ/ⴙ mice. We conclude that BMP-2 transcriptionally induces COX-2 in osteoblasts via a Cbfa1 binding site and that the BMP-2 induction of COX-2 can contribute to effects of BMP-2 on osteoblastic differentiation and osteoclast formation in vitro and to the BMP-2 stimulation of ectopic bone formation in vivo. (J Bone Miner Res 2002;17:1430 –1440) Key words: MC3T3-E1 cells, prostaglandin G/H synthase, prostaglandin E2, COX-2 knockout, promoter regulation

INTRODUCTION ROSTAGLANDINS (PGS) ARE produced by osteoblasts and are abundant in bone.(1) They are complex regulators of bone remodeling that can stimulate both bone formation and bone

P

Dr. Raisz receives research funding from Chugai, Colgate, Eisai, Pharmacia, and Procter & Gamble. All other authors have no conflict of interest.

resorption. The major enzyme regulating the conversion of arachidonic acid released from membrane phospholipids to PGs is PGG/H synthase (PGHS), commonly called cyclo-oxygenase (COX). There are two enzymes for COX encoded by separate genes, COX-1 and COX-2.(2,3) COX-1 usually is expressed constitutively, and COX-2 is induced rapidly and transiently. Both COX-1 and COX-2 are expressed in osteoblastic cells, and COX-2 is the main enzyme regulating the production of PGs in response to various hormones and cytokines.(1)

1

Department of Medicine, University of Connecticut Health Center, Farmington, Connecticut, USA. Department of Orthopedic Surgery, University of Tokyo, Bunkyo, Tokyo, Japan. Department of Orthopedic Surgery, University of Connecticut Health Center, Farmington, Connecticut, USA. 4 Molecular Biology Institute, University of California, Los Angeles, Los Angeles, California, USA. 2 3

1430

BMP-2 INDUCTION OF COX-2

Bone morphogenetic proteins (BMPs) belong to the transforming growth factor (TGF) ␤ superfamily.(4) Originally, BMPs were identified by their ability to form ectopic bone.(5) BMP-2 is expressed during murine embryonal skeletogenesis and can stimulate osteoblast differentiation from uncommitted progenitors both in vitro and in vivo.(4,6) Exogenous BMP-2 can accelerate fracture healing, and BMP-2 is expressed in primitive mesenchymal cells and chondrocytes at the site of callus formation.(4) BMP-2 also can stimulate or enhance the process of osteoclastogenesis by its action on osteoblastic cells.(7–9) BMP-2 signaling is mediated by the type I BMP receptor and activation of Smads, which translocate into the nucleus, where they interact with DNA-binding partners and activate gene transcription.(10,11) Core-binding factor activity 1 (Cbfa1), also known as Runx2, is an essential transcriptional factor for osteoblast differentiation.(12,13) Cbfa1-deficient mice lacked both intramembranous and endochondral ossification because of the failure of osteoblasts to differentiate.(12–14) Cultured Cbfa1⫺/⫺ calvarial cells failed to generate bone nodules even in the presence of BMP-2.(15) Osteoclastogenesis also was found to be retarded in Cbfa1-deficient mice, presumably because of a requirement of Cbfa1 for the expression of RANKL, which mediates the interaction between osteoblast and osteoclast lineage required for osteoclast differentiation.(5,13,16) Hence, Cbfa1 appears essential not only for osteoblasts to become bone-forming cells but also for osteoblasts to become support cells for osteoclast formation. Several studies have suggested that Cbfa1 is a downstream factor in the BMP-2 signaling pathway. BMP-2 induced Cbfa1 mRNA expression in human bone marrow stromal cells(17) and 2T3 cells.(18) In addition, recently, Cbfa1 has been shown to be a common target of BMP-2 and TGF-␤ signaling pathways, cooperating with Smads to induce osteoblast-specific gene expression in C2C12 cells.(19) Because both BMP-2 and PGs can stimulate bone formation and resorption, it is possible that PGs may mediate some actions of BMP-2 on bone. BMP-2 has been shown to induce COX-2 expression in osteoblastic cells.(8) In this study, we examined transcriptional regulation of COX-2 by BMP-2, identified a Cbfa1-binding element mediating the BMP-2 induction of COX-2 promoter activity, and showed the relevance of the BMP-2 induction of COX-2 expression for osteoblastic and osteoclastic differentiation in vitro.

MATERIALS AND METHODS Materials Murine COX-2 cDNA and DNA constructs (⫺371/⫹70 bp of the COX-2 gene or 5⬘ deletions of this region fused to a luciferase reporter gene in pXp-2 vector) have been described previously.(20 –22) Murine COX-1 cDNA was the gift of Dr. David DeWitt and Dr. William Smith (Michigan State University, East Lansing, MI, USA). Cbfa1 cDNA was the gift of Dr. Gerard Karsenty (University of Texas Dental Branch, Houston, TX, USA). Murine osteocalcin cDNA was the gift of Dr. John Wozney (Genetics Institute, Cambridge, MA, USA). cDNA for GAPDH was amplified by using an amplimer kit from Clontech (Palo Alto, CA,

1431

USA). Recombinant human BMP-2 (rhBMP-2), kindly supplied by Yamanouchi Pharmaceutical Co. (Tokyo, Japan), was used for all in vitro studies except for the study in Table 1. For that study and for the in vivo studies, rhBMP-2 was purchased from R & D Systems (Minneapolis, MN, USA). NS-398 was purchased from Cayman Chemical Co. (Ann Arbor, MI, USA). Phorbol 13-myristate 12-acetate (PMA) and other chemicals were purchased from Sigma Chemical Co. (St. Louis, MO, USA).

Animals Mice carrying ⫺371/⫹70 bp of the 5⬘-flanking murine COX-2 DNA fused to a luciferase reporter (Pluc) were developed in a CD-1 background by the Transgenic Animal Facility at the University of Connecticut Health Center (UCHC). COX-2 knockout (COX-2⫺/⫺) mice were developed at the University of North Carolina.(23) COX-2⫺/⫺ and wild-type (COX-2⫹/⫹) mice were bred by mating ⫹/⫺ offspring of C57Bl/6⫹/⫺ and 129/sv⫹/⫹ crosses for reasons discussed previously.(24)

Osteoblastic cell cultures MC3T3-E1 cells were the gift of Dr. Yoshiyuki Hakeda (Meikai University School of Dentistry, Sakado, Saitama, Japan). To make primary osteoblastic cultures, whole calvariae were excised from neonatal Pluc transgenic mice or from 6- to 8-week-old COX-2⫹/⫹ and COX-2⫺/⫺ mice. Calvariae were digested sequentially with 0.5 mg/ml of crude collagenase P (Boehringer Mannheim, Indianapolis, IN, USA) and 20% trypsin-EDTA. Five digests were performed, all for 10 minutes except the last one, which was for 30 minutes. Released cells were removed, the reaction was stopped with 10% FCS (Gibco BRL, Grand Island, NY, USA), and the solution was filtered through a Nitex membrane (Millipore Corp., Bedford, MA, USA) to ensure a single cell suspension. Digests 2–5 were pooled and plated, grown to confluence, and passaged once before use. For experiments, cells were plated in 6-well dishes at 5000/cm2 and grown in DMEM without phenol red (Sigma Chemical Co.) containing 10% FCS, penicillin (100 U/ml), and streptomycin (50 ␮g/ml) in a humidified atmosphere of 5% CO2 at 37°C. To study BMP-2 induction of COX-2 mRNA or luciferase, cells were grown to confluence for 6 days and changed to serum-free medium with 0.1% bovine serum albumin (BSA; Sigma Chemical Co.) 24 h before BMP-2 treatment. For measurement of ALP activity, cells were cultured with 50 ␮g/ml of L-ascorbic acid phosphate (Wako, Ltd., Osaka, Japan) and 10 mM of ␤-glycerophosphate. To measure ALP activity, cultures were washed with PBS and sonicated in 10 mM of Tris-HCl buffer (pH 7.5) containing 0.1% Triton X-100. ALP activity in the lysate was measured by the hydrolysis of p-nitrophenyl phosphate to p-nitrophenol. Absorbance was determined at 405 nm and compared with a p-nitrophenol (Sigma Chemical Co.) standard titration curve. ALP activity was normalized to total proteins measured with a BCA protein assay kit (Pierce Chemical Co., Rockford, IL, USA).

1432

Bone marrow cell cultures Bone marrow cells (2 ⫻ 106 cells/well) were flushed from tibias and femurs of 6- to 8-week-old COX-2⫹/⫹ and COX2⫺/⫺ mice. Cells were cultured in 24-multiwell dishes with ␣-modified minimum essential medium (␣-MEM; Gibco BRL) containing 10% FCS, penicillin (100 U/ml), and streptomycin (50 ␮g/ml). After 7 days of culture, cells were washed with PBS, fixed with 2.5% glutaraldehyde for 30 minutes, and stained using a leukocyte acid phosphatase A kit (Sigma Chemical Co.). TRAP⫹ multinucleated cells (MNCs) containing more than three nuclei were counted as osteoclast-like cells.

Stable and transient transfections Stable and transient transfections of MC3T3-E1 cells were performed as previously described.(25,26) For stably transfected cultures, colonies (⬎200) were pooled to minimize effects secondary to variable integration sites. Luciferase activity was measured in soluble cell extracts prepared with a kit from Promega (Madison, WI, USA) using an automatic injection luminometer (Berthold Lumat; Wallac, Inc., Gaithersburg, MD, USA). For each experiment, three wells of a 6-well dish were analyzed per treatment group. Activity was normalized to total protein measured with a BSA protein assay kit (Pierce Chemical Co.).

Site-directed mutation The putative Cbfa1-binding sequence 5⬘-AACCACA-3⬘ at ⫺267/⫺261 bp in the COX-2 5⬘-flanking region was changed to 5⬘-AATAACA-3⬘ (mutation of Cbfa1 [muCbfa1]) using the QuickChange Site-Directed Mutagenesis Kit from Stratagene (La Jolla, CA, USA). Mutations of the AP-1 and cAMP response element (CRE) sites have been described previously.(26,27) After introduction of mutations, the COX-2 5⬘-flanking region was sequenced (Automated DNA Sequence Facility, UCHC). To avoid the accidental introduction of mutations into regions outside the COX-2 5⬘-flanking region, the 5⬘-flanking region was released from the pXp2 plasmid and subcloned into a fresh pXp2 vector.

EMSA Nuclear extracts were prepared as previously described.(26) For gel electrophoresis, single-stranded oligonucleotides were end-labeled with [␥ 32P]-adenosine triphosphate (ATP; New England Nuclear) using T4 kinase (Gibco BRL). The complementary oligonucleotide was annealed, and the resulting double-stranded DNA was purified. Six micrograms of nuclear extract was incubated in 20 ␮l of binding reaction mixture (10 mM of Tris-HCl, pH 7.5, 1 mM of dithiothreitol [DTT], 1 mM of EDTA, 5% glycerol, and 2 ␮g of poly dI-dC [Pharmacia and Upjohn, Bridgewater, NJ, USA]) with 50,000 cpm of labeled probe. Five percent nondenaturing acrylamide gel electrophoresis was performed for 2 h and gels were fixed for 30 minutes in 10% methanol and 10% acetic acid. Dried gels were exposed to X-ray film. Competitor (100 M excess) or supershifting

CHIKAZU ET AL.

antibody (4 ␮g) was added to the binding mixture 30 minutes before addition of the probe, and incubation continued for 30 minutes. Antibody designed to supershift Cbfa1-binding complexes was obtained from Oncogene Research Products (Cambridge, MA, USA).

mRNA extraction and Northern blot analysis Three wells of cells in 6-well dishes were pooled for RNA extraction using TRI-REAGENT (Molecular Research Center, Inc., Cincinnati, OH, USA). After quantitation at 260 nm, 10 ␮g of total RNA was run on a 1% agarose–2.2 M of formaldehyde gel and transferred to a nylon membrane (Genescreen; New England Nuclear Research, Boston, MA, USA) by capillary pressure using Turboblotter (Schleicher & Schuell, Keene, NH, USA). After 2 h of prehybridization in 50% formamide solution in rotating cylinders at 42°C, filters were hybridized overnight at 42°C in a similar solution with random primer [32P]deoxycytidine triphosphate (dCTP; Perkin Elmer Boston, MA, USA)-labeled cDNA probes. Filters were washed once in a 1⫻ SSC and 1% SDS solution at room temperature, once in a 0.1⫻ SSC and 0.1% SDS solution at 65°C, and then three times in the latter solution at room temperature. After washing, filters were exposed to Eastman Kodak Co. XAR-5 film (Eastman Kodak Co., Rochester, NY, USA) at ⫺70°C, X-rays were scanned (ScanJet; Hewlett-Packard Co., Corvallis, OR, USA), and band density was determined using NIH Image 1.62 (National Institutes of Health, Bethesda, MD, USA).

PGE2 assay Medium was removed from cultured cells and PGE2 accumulation was measured by radioimmunoassay as described previously.(28) The lower limit of detection of this assay is 0.1 nM. One hundred microliters of unmodified medium was assayed for each sample.

In vivo mineralization A solution of 0.4 ml of rat type I collagen (BD Biosciences, Bedford, MA, USA) with BMP-2 (10 ␮g) or vehicle PBS with 0.1% BSA was freeze-dried and formed into pellets. Pellets were implanted subcutaneously in the flanks of 7-week-old COX-2⫺/⫺ and COX-2⫹/⫹ mice under general anesthesia. Mice were killed and pellets were removed 3 weeks later. Each mouse carried one BMP-2 pellet and one vehicle pellet in opposing flanks. Mineralized bone volume in the pellets was analyzed by microcomputed tomography (␮CT; microCT20; Scanco Medical AG, Zurich, Switzerland). All animal protocols were approved by the Animal Care and Use Committee at UCHC.

Statistical analysis Means of groups were compared by ANOVA (SigmaStat, San Rafael, CA, USA). The significance of differences was determined by post hoc testing using the Bonferroni method.

BMP-2 INDUCTION OF COX-2

1433

FIG. 1. BMP-2 induction of COX-2 mRNA and PGE2 in calvarial osteoblastic cultures. Primary osteoblastic cultures were prepared from calvariae of COX-2⫹/⫹ and COX-2⫺/⫺ mice. Cells were treated with vehicle (control) or BMP-2 (100 ng/ml) for the indicated periods. (A) Cells were cultured for 6 days and serum-deprived for 24 h before being treated. Steady-state mRNA levels of COX-2 were measured by Northern analysis and compared with mRNA levels of COX-1 and Cbfa1. mRNA levels for GAPDH are shown to assess loading. (B) Cells were cultured with 10% FCS for 2 weeks. Media were removed for measurement of PGE2 levels and fresh media were added on days 3, 7, and 10. Medium PGE2 values were measured by radioimmunoassay (RIA) and cumulative levels were calculated. Data are means ⫾ SEM for n ⫽ 6 samples. a Significant difference from control group; p ⬍ 0.01.

RESULTS BMP-2 induction of COX-2 and PGE2 production BMP-2 (100 ng/ml) rapidly induced both COX-2 mRNA and PGE2 production in primary calvarial osteoblastic cell cultures derived from COX-2⫹/⫹ mice (Fig. 1A). Both COX-1 mRNA and Cbfa1mRNA were expressed constitutively and not increased by BMP-2. COX-2 expression and PGE 2 production are induced in calvarial osteoblasts by addition of fresh serum at the time of medium changes (Pilbeam, data not shown). In calvarial osteoblastic cells cultured for 14 days, treatment with BMP-2 (100 ng/ml) increased medium PGE 2 accumulation by 45% in cells derived from COX-2⫹/⫹ mice relative to cells from COX2⫺/⫺ mice (Fig. 1B).

BMP-2 induction of luciferase activity In calvarial osteoblasts from neonatal mice transgenic for ⫺371/⫹70 bp of the murine COX-2 promoter fused to a luciferase reporter (Pluc), BMP-2 induced luciferase activity in a dose-related manner with peak effects at 7 h (Figs. 2A and 2B). In MC3T3-E1 cells stably transfected with Pluc, BMP-2 dose-dependently increased luciferase activity with peak effects at 3 h (Figs. 2C and 2D). The delayed peak response in primary cells compared with MC3T3-E1 cells has been observed for all COX-2 agonists we have studied (Pilbeam, data not shown). For both primary cells and MC3T3-E1 cells, the maximal stimulation of luciferase activity by BMP-2 at 100 ng/ml was fourfold.

Identification of the cis-acting site(s) mediating response to BMP-2 To determine regions within the 371-bp 5⬘-flanking region that mediated the response to BMP-2, MC3T3-E1 cells were stably transfected with a series of 5⬘-deletion constructs and treated with BMP-2 (100 ng/ml) for 3 h (Fig. 3A). There was no reduction in BMP-2–stimulated luciferase activity with deletion to ⫺300/⫹70 bp but BMP-2 stimulation was abrogated by deletion to ⫺213/⫹70 bp. Hence, site(s) mediating the BMP-2 stimulation of COX-2 promoter activity lay between ⫺300 and ⫺213 bp. Although distant from the ⫺150/⫺40 bp region previously shown to contain several important cis-acting elements regulating COX-2 activity, including the AP-1 and CRE sites (Fig. 3B), it is similar to the region found to mediate the COX-2 promoter response to TGF-␤.(25,26,27) Examination of the DNA sequence identified a Cbfa1 consensus sequence (5⬘-AACCACA-3⬘) at ⫺267/⫺261 bp (Fig. 4B). MC3T3-E1 cells were transiently transfected with either the wild-type Pluc construct or the Pluc construct carrying a 2-bp muCbfa1 and treated with BMP-2 (100 ng/ml) for 3 h (Fig. 4A). In cells carrying muCbfa1, the BMP-2 stimulation of luciferase activity was reduced 82% and was not significantly different from the untreated control. To determine if the mutation was specific for the response to BMP-2, we treated cells with PMA (1 ␮M), which stimulates COX-2 promoter activity through the AP-1 and CRE sites.(26) PMA-induced luciferase activity was not decreased in cells carrying the muCbfa1 promoter construct (Fig. 4A). Although other transactivating sites in the pro-

1434

CHIKAZU ET AL.

FIG. 2. BMP-2 stimulation of luciferase activity in osteoblasts carrying ⫺371/⫹70 bp of the COX-2 5⬘-flanking DNA fused to a luciferase reporter gene. Cells were treated with vehicle (control) or BMP-2 (100 ng/ml) and luciferase activity was calculated as counts per second (cps) per microgram of total protein. In primary calvarial osteoblasts from transgenic mice, (A) cells were treated with BMP-2 for the indicated times and (B) cells were treated with increasing concentrations of BMP-2 for 7 h, and the fold increase in luciferase activity, normalized to total protein, was calculated as treated/control ratios. Data are means ⫾ SEM for n ⫽ 3 samples. aSignificant difference from control group; p ⬍ 0.01. In stably transfected MC3T3-E1 cells, (C) cells were treated with BMP-2 for the indicated times and (D) cells were treated with increasing concentrations of BMP-2 for 3 h. Luciferase activity was normalized to total protein and the fold increase in activity was calculated as treated/control ratios. Data are means ⫾ SEM for n ⫽ 6 samples. aSignificant difference from control group; p ⬍ 0.01.

moter might not be sufficient to permit BMP-2 stimulation of promoter activity in the absence of the Cbfa1 site, they might, nevertheless, be necessary for BMP-2 stimulation of activity. However, 2-bp mutations of the AP-1 and CRE sites, which were previously shown to inhibit PMA stimulation of luciferase activity, did not reduce BMP-2– stimulated luciferase activity (Fig. 4B).

Binding to the Cbfa1 site on EMSA Nuclear extracts were prepared from MC3T3-E1 cells treated for 1 h with or without BMP-2 (100 ng/ml) and run on gel with a 32P-labeled oligonucleotide spanning the Cbfa1 site (⫺278/⫺253 bp of the COX-2 5⬘-flanking sequence). Binding to this probe was constitutive (Fig. 5A, lanes 2 and 3, arrows a– c). All bands were competed by the unlabeled probe (Fig. 5A; lanes 4 and 5). Only the upper band (Fig. 5B, arrow a) was competed by the unlabeled probe containing the muCbfa1 sequence (Fig. 5B; lanes 4 and 5), indicating that protein binding was unaffected by the mutation. Proteins in the lower two bands (Fig. 5B, arrows b and c) could only bind the intact Cbfa1 site. The lower bands were inhibited by an antibody

specific for Cbfa1 as shown in Fig. 6C and a light supershifted band was evident (Fig. 5C, lanes 4 and 5, arrow d). Nonspecific immunoglobulin G (IgG) antibody did not inhibit the binding (Fig. 5B, lanes 5 and 6).

Role of COX-2 in BMP-2–induced osteoblastic cell differentiation Primary calvarial osteoblastic cells derived from COX-2⫹/⫹ and COX-2⫺/⫺ mice were cultured for 14 days with and without BMP-2 (100 ng/ml). ALP activity was similar in COX-2⫺/⫺ and COX-2⫹/⫹ control cultures, but treatment with BMP-2 stimulated ALP activity only in COX-2⫹/⫹ cultures (Fig. 6A). In this experiment, treatment with PGE 2 (1 ␮M) alone did not significantly increase ALP activity in either COX-2⫹/⫹ or COX-2⫺/⫺ cultures. However, treatment with both PGE 2 and BMP-2 increased ALP activity in COX-2⫺/⫺ cultures to the same level as seen in COX-2⫹/⫹ cultures. Treatment of COX-2⫹/⫹ cultures with NS-398 (0.1 ␮M), a selective inhibitor of COX-2 activity in osteoblasts at this dose,(29) did not significantly inhibit ALP activity in control

BMP-2 INDUCTION OF COX-2

1435

FIG. 3. Sequential deletion analysis. (A) MC3T3-E1 cells were stably transfected with the wild-type construct (⫺371/⫹70 bp of COX-2 fused to a luciferase reporter) or 5⬘ deletions of this construct. Cells were treated with vehicle (control) or BMP-2 (100 ng/ml) for 3 h. Luciferase activity was normalized to total protein and the fold increase in activity was calculated as treated/control ratios. Data are means ⫾ SEM for n ⫽ 3 samples. a Significant difference from control group; p ⬍ 0.01. Similar results were obtained in a second experiment. (B) Potential transactivating factor binding sites in the 5⬘-flanking DNA of the murine COX-2 gene. A 2-bp mutation (muCbfa1) was made in the putative Cbfa1-binding site as described in the Materials and Methods section.

FIG. 4. Effect of mutating the Cbfa1 site on BMP-2–stimulated luciferase activity in MC3T3-E1 cells. MC3T3-E1 cells were transiently transfected with either the unmodified ⫺371/⫹70-bp construct (wild type) or with the ⫺371/⫹70-bp construct carrying a 2-bp muCbfa1, a 2-bp mutation of the CRE site (muCRE), or a 2-bp mutation of the AP-1 site (muAP-1). Cells were treated for 3 h with vehicle (control), BMP-2 (100 ng/ml), or PMA (1 ␮M). Data are means ⫾ SEM for n ⫽ 3 samples. aSignificant difference from control; p ⬍ 0.01. (A) Luciferase activity is expressed as counts per second (cps) per microgram of total protein. Similar results were obtained in three additional experiments. (B) Luciferase activity was normalized to total protein and the fold increase in activity was calculated as treated/control ratios. Similar results were obtained in a second experiment.

cultures but inhibited the fold stimulation of ALP activity by BMP-2 by 38% (Fig. 6B). In addition, BMP-2 stimulated osteocalcin mRNA expression at day 14 of culture in primary calvarial cells derived from COX-2⫹/⫹ mice but not in COX2⫺/⫺ cells (Fig. 6C).

Role of COX-2 in BMP-2–induced osteoclast-like cell formation To investigate osteoclast formation, we cultured bone marrow cells from COX-2⫹/⫹ and COX-2⫺/⫺ mice and

counted the number of TRAP⫹ MNCs formed after 7 days of culture with or without BMP-2 (100 ng/ml). BMP-2 stimulated osteoclast formation in marrow from COX-2⫹/⫹ mice but not in marrow from COX-2⫺/⫺ mice (Figs. 7A and 7B). The TRAP⫹ MNCs formed in response to BMP-2 were able to resorb pits on dentine slices (data not shown). In a similar experiment, PGE2 (1 ␮M) added to BMP-2–treated cultures synergistically enhanced the formation of TRAP⫹ MNCs in both COX-2⫹/⫹ and COX-2⫺/⫺ cultures and reversed the difference between COX-2⫹/⫹ and COX-2⫺/⫺ cultures (Table 1).

1436

CHIKAZU ET AL.

FIG. 5. EMSA analysis of nuclear binding to a 32P-labeled COX-2 oligonucleotide (Cbfa1) spanning the putative Cbfa1-inding site. MC3T3-E1 cells were treated with vehicle or BMP-2 (100 ng/ml) for 1 h and nuclear proteins were extracted as described in the Materials and Methods section. Lane 1 is the free probe. (A) Constitutive bands (arrows a– c) were seen with the 32P-labeled COX-2 oligonucleotide (lanes 2 and3). Bands were competed by addition of 100-fold molar excess of the unlabeled COX-2 oligonucleotide (lanes 4 and 5). (B) Competition by 100-fold molar excess of unlabeled COX-2 oligonucleotide carrying the 2-bp mutation of the Cbfa1 sequence (muCbfa1) inhibited only the top band (arrow a; lanes 4 and 5). Nonspecific IgG antibody did not inhibit the remaining bands (lanes 6 and 7). (C) Specific antibody for Cbfa1 inhibited the lower band(s) and shifted binding (arrow d; lanes 4 and 5).

FIG. 6. BMP-2–induced ALP activity and osteocalcin mRNA expression in calvarial osteoblasts from COX-2⫹/⫹ or COX-2⫺/⫺ mice. Cells were cultured for 14 days with vehicle (control) or BMP-2 (100 ng/ml) throughout the culture period. (A) COX-2⫹/⫹ and COX-2⫺/⫺ cells were treated with or without PGE 2 (1 ␮M). ALP activity, normalized to total protein, was calculated. Data are means ⫾ SEM for n ⫽ 3 samples. aSignificant effect of BMP-2, p ⬍ 0.01 and bp ⬍ 0.05. cSignificant increase relative to treatment with single agent, p ⬍ 0.01. dSignificant difference from COX-2⫹/⫹ group, p ⬍ 0.01. (B) COX-2⫹/⫹ cells were treated with or without NS-398 (0.1 ␮M). Data are means ⫾ SEM for n ⫽ 6 samples. a Significant effect of BMP-2, p ⬍ 0.01 and bp ⬍ 0.01. cSignificant effect of NS-398, p ⬍ 0.01. (C) Steady-state mRNA levels were examined by Northern blot analysis.

Role of COX-2 in BMP-2–induced mineralization in vivo Collagen pellets containing BMP-2 (10 ␮g/pellet) implanted in the flanks of 7-week-old mice for 3 weeks showed markedly increased mineralized volume, measured by ␮CT analysis, compared with pellets containing vehicle (Fig. 8). BMP-2–induced mineralization was reduced by 78% in pellets implanted in COX-2⫺/⫺ mice relative to pellets implanted in COX-2⫹/⫹ mice.

DISCUSSION Cbfa1 is an essential transcription factor for osteoblast differentiation and can regulate the expression of many genes associated with the development of bone matrix such as osteocalcin, type I collagen, and osteopontin.(12,30 –32) Recent studies have shown that parathyroid hormone (PTH) stimulates expression of MMP-13 via a Cbfa1-binding site.(33–35) Our study shows that BMP-2 also can induce COX-2 expression and PGE2 production in osteoblastic

BMP-2 INDUCTION OF COX-2

1437

FIG. 7. Effect of COX-2 expression on BMP-2–induced osteoclast formation in cultured bone marrow cells. Bone marrow from COX-2⫹/⫹ and COX-2⫺/⫺ mice was cultured for 7 days with vehicle (control) or BMP-2 (100 ng/ml). (A) Number of TRAP⫹ MNCs formed in the bone marrow cultures. Data are means ⫾ SEM for n ⫽ 12 wells. aSignificant difference from control group, p ⬍ 0.01. (B) Microphotograph of TRAP-stained bone marrow cells (bar ⫽ 25 ␮m).

TABLE 1. EFFECT OF EXOGENOUS PGE2 (1 ␮M) ON THE BMP-2 (500 NG/ML) INDUCTION OF OSTEOCLAST FORMATION IN COX-2⫹/⫹ AND COX-2⫺/⫺ MARROW CULTURES Treatment Control PGE2 BMP-2 ⫹PGE2

n

COX-2⫹/⫹

COX-2⫺/⫺

12 10–12 6 6

0 64.2 ⫾ 6.0 11.2 ⫾ 2.3 137.8 ⫾ 21.2*

0 46.3 ⫾ 4.5† 0 158.3 ⫾ 15.2*

Data are means and SEM for n wells. * Significant difference relative to treatment with single agent, p ⬍ 0.01; † significant effect of genotype, p ⬍ 0.05.

cells via a Cbfa1-binding site. Hence, Cbfa1 is likely to play a role not only in skeletal development but also in bone turnover in the adult skeleton in response to calciumregulating hormones and local factors. In some studies in which BMP-2 actions were mediated via Cbfa1, BMP-2 increased Cbfa1 mRNA expression.(17,18,36) However, other studies report that agonists considered to act via Cbfa1 did not increase expression of Cbfa1 mRNA or binding of Cbfa1on EMSA.(30 –34) Similarly, we found that Cbfa1 mRNA was expressed constitutively and that binding on EMSA to the oligonucleotide spanning the COX-2 Cbfa1 site was constitutive. The constitutive expression and binding of Cbfa1 does not preclude a role in regulating COX-2 transcription. As is seen with CREB and CREM, transcription factors that also are expressed and bound constitutively, Cbfa1 may require phosphorylation to become activated. Little is known about the signaling pathways that might lead to this activation, although one study has shown activation and phosphorylation of Cbfa1 by the MAPK (ERK) pathway(37) and another study has shown that the ERK pathway may be involved in BMP-2–induced osteoblastic proliferation.(38)

On the other hand, Cbfa1 constitutively bound to the COX-2 promoter could become active only on proteinprotein interaction, at the level of DNA binding, with other BMP-2–stimulated coactivators. Although BMP-2 has been shown to promote osteoblast apoptosis through a Smadindependent protein kinase C– dependent pathway,(39) BMP-2 signaling generally is thought to be mediated via receptor activation of Smads 1, 5, and 8, which oligomerize with Smad 4 and translocate into the nucleus to bind DNA.(10,11) However, Smad-regulated Smad-DNA binding alone generally is insufficient for gene activation and requires cooperation from DNA binding partners.(11) Although some of these Smad partners can bind DNA, they cannot transactivate unless complexed with activated Smads.(11) Other Smad DNA-binding partners may be active on their own, requiring input from different signaling pathways and only having their activity modulated by interactions with Smads.(11) A number of studies have shown that BMP-2–activated Smads can form complexes with Cbfa1 to enhance Cbfa1 transactivation.(19,40,41) It also is possible that interaction with Smads could relieve an autoinhibitory conformation of Cbfa1. Exogenous PGs can be anabolic for bone. In vitro, PGE 2 stimulates the formation and differentiation of osteoblastic colonies.(42,43) In vivo, PGs given to rats produce substantial increases in bone mass(44) and enhance osteoblastic differentiation in explanted marrow stromal cells.(45) Less is known about the role of endogenous PGs, but marrow stromal cells cultured from COX-2⫺/⫺ mice show decreased osteoblastic differentiation compared with cells from COX-2⫹/⫹ mice and this decrease can be reversed by adding back PGE2.(46) Small decreases in trabecular bone volume and parameters of bone formation in COX-2⫺/⫺ mice compared with COX-2⫹/⫹ mice suggest that a deficiency of endogenous PGs may decrease bone formation more than resorption.(46) Endogenous PGs also appear to play a role in acutely stimulated bone formation in vivo. Studies have shown that perioperative treatment with nonsteroidal anti-

1438

CHIKAZU ET AL.

FIG. 8. Measurement by ␮CT of mineralized volume of pellets implanted in COX-2⫹/⫹ and COX-2⫺/⫺ mice. Collagen pellets were implanted subcutaneously in the flanks of 7-week-old COX-2⫹/⫹ (n ⫽ 5) and COX-2⫺/⫺ (n ⫽ 5) mice and removed 3 weeks after implantation. Each mouse carried a BMP-2 (10 ␮g) pellet and a vehicle pellet. (A) ␮CT analysis of BMP-2– containing pellets. Three representative pellets, including the smallest and largest mineralized volumes, are shown for each group (bar ⫽ 1 mm). (B) Quantitation of pellet mineralization. Data are means ⫾ SEM for n ⫽ 5 samples. aSignificant difference from vehicle group, p ⬍ 0.01. bSignificant difference from wild-type group, p ⬍ 0.01.

inflammatory drugs (NSAIDs) can prevent heterotrophic ossification after hip surgery,(47) suggesting that this problem is caused by PG-mediated bone formation. In addition, NSAIDs can inhibit fracture repair(48) and spinal fusion.(49) This study suggests that the effects of BMP-2 on osteoblastic differentiation are enhanced by BMP-2–induced COX-2 expression and PG production and that BMP-2–stimulated PG production also may be important for the effects of BMP-2 on ectopic bone formation and fracture healing. The substrate for COX-2, arachidonic acid, also can be metabolized via lipoxygenase to leukotrienes and hydroxyeicosatetraenoic acids (HETEs). In COX-2⫺/⫺ mice, more arachidonic acid may be available for metabolism via the lipoxygenase pathway than in COX-2⫹/⫹ mice. Several studies have suggested that the products of the lipoxygenase pathway can inhibit bone formation.(1) In one study, leukotriene B4 and 5-HETE blocked the BMP-2–induced formation of mineralized nodules in fetal rat calvarial culture.(50) Hence, it is possible that an increase in the products of the lipoxygenase pathway caused by COX-2 depletion might be involved in the decreased osteoblastic differentiation response to BMP-2 in COX-2⫺/⫺ cultures and the decreased ectopic calcification response to BMP-2 in COX-2⫺/⫺ mice. Marrow cultures from some strains of mice, such as C57Bl/6 or C57Bl/6 ⫻ 129 (the background of our COX2⫹/⫹ and COX-2⫺/⫺ mice), produce osteoclasts in response to multiple agonists without the need for additional treatment with RANKL or macrophage colony–stimulating factor (M-CSF) or the need for coculture with osteoblasts (e.g., see the study by Okada et al.(24)). BMP-2 alone was able to stimulate osteoclast formation in COX-2⫹/⫹ marrow cultures but not in COX-2⫺/⫺ marrow cultures, indicating a dependence of BMP-2–stimulated osteoclastogenesis on COX-2 expression. PGs, which are themselves potent stimulators of resorption, can enhance osteoclastogenesis stimulated by many agonists.(1,24) Similarly, PGE2 added back

to the marrow cultures synergistically enhanced the BMP-2 stimulation of osteoclastogenesis to equal levels in COX2⫹/⫹ and COX-2⫺/⫺ cultures. Other studies have also implicated BMP-2 in osteoclast formation. BMP-2 stimulated osteoclast-like cell formation and bone resorbing activity in mouse bone cell cultures.(7) Antagonism of BMP-2 and BMP-4 action by treatment with noggin inhibited both osteoblast and osteoclast formation in mouse bone marrow cultures.(9) A role for COX-2 in the BMP-2 induction of osteoclast formation was suggested by a study showing that BMP-2 induced COX-2 expression and enhanced IL-1– induced osteoclastogenesis in ddY marrow stromal cells.(8) In summary, we have shown that BMP-2 induced COX-2 in osteoblasts transcriptionally via a Cbfa1 DNA-binding site and that the induction of COX-2 may play a role in the effects of BMP-2 on bone metabolism both in vitro and in vivo.

ACKNOWLEDGMENTS We appreciate the excellent technical assistance of Cynthia Alander. This work was supported by NIH grant DK 48361 to CCP.

REFERENCES 1. Pilbeam CC, Harrison JR, Raisz LG 2002 Prostaglandins and bone metabolism. In: Bilezikian JP, Raisz LG, Rodan GA (eds.) Principles of Bone Biology. Academic Press, New York, NY, USA, pp. 979 –994. 2. Smith WL, Garavito RM, DeWitt DL 1966 Prostaglandin endoperoxide H synthases (cyclooxygenases)-1 and -2. J Biol Chem 271:33157–33160. 3. Herschman HR 1996 Prostaglandin synthase 2. Biochim Biophys Acta 1299:125–140.

BMP-2 INDUCTION OF COX-2 4. Rosen V, Cox K, Hattersley G 1996 Bone morphogenetic proteins. In: Bilezikian JP, Raisz LG, Rodan GA (eds.) Principles of Bone Biology. Academic Press, New York, NY, USA, pp. 661– 671. 5. Yamaguchi A, Komori T, Suda T 2000 Regulation of osteoblast differentiation mediated by bone morphogenetic proteins, hedgehogs, and Cbfa1. Endocr Rev 21:393– 411. 6. Reddi AH 1994 Bone and cartilage differentiation. Curr Opin Genet Dev 4:737–744. 7. Kanatani M, Sugimoto T, Kaji H, Kobayashi T, Nishiyama K, Fukase M, Kumegawa M, Chihara K 1995 Stimulatory effect of bone morphogenetic protein-2 on osteoclast-like cell formation and bone-resorbing activity. J Bone Miner Res 10: 1681–1690. 8. Koide M, Murase Y, Yamato K, Noguchi T, Okahashi N, Nishihara T 1999 Bone morphogenetic protein-2 enhances osteoclast formation mediated by interleukin-1alpha through upregulation of osteoclast differentiation factor and cyclooxygenase-2. Biochem Biophys Res Commun 259:97– 102. 9. Abe E, Yamamoto M, Taguchi Y, Lecka-Czernik B, O’Brien CA, Economides AN, Stahl N, Jilka RL, Manolagas SC 2000 Essential requirement of BMPs-2/4 for both osteoblast and osteoclast formation in murine bone marrow cultures from adult mice: Antagonism by noggin. J Bone Miner Res 15:663– 673. 10. Nishimura R, Kato Y, Chen D, Harris SE, Mundy GR, Yoneda T 1998 Smad5 and DPC4 are key molecules in mediating BMP-2-induced osteoblastic differentiation of the pluripotent mesenchymal precursor cell line C2C12. J Biol Chem 273: 1872–1879. 11. Attisano L, Wrana JL 2000 Smads as transcriptional comodulators. Curr Opin Cell Biol 12:235–243. 12. Ducy P, Zhang R, Geoffroy V, Ridall AL, Karsenty G 1997 Osf2/Cbfa1: A transcriptional activator of osteoblast differentiation. Cell 89:747–754. 13. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, Shimizu Y, Bronson RT, Gao YH, Inada M, Sato M, Okamoto R, Kitamura Y, Yoshiki S, Kishimoto T 1997 Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell 89:755–764. 14. Hoshi K, Komori T, Ozawa H 1999 Morphological characterization of skeletal cells in Cbfa1-deficient mice. Bone 25:639 – 651. 15. Kobayashi H, Gao YH, Ueta C, Yamaguchi A, Komori T 2000 Multilineage differentiation of Cbfa1-deficient calvarial cells in vitro. Biochem Biophys Res Commun 273:630 – 636. 16. Gao YH, Shinki T, Yuasa T, Kataoka-Enomoto H, Komori T, Suda T, Yamaguchi A 1998 Potential role of cbfa1, an essential transcriptional factor for osteoblast differentiation, in osteoclastogenesis: Regulation of mRNA expression of osteoclast differentiation factor (ODF). Biochem Biophys Res Commun 252:697–702. 17. Gori F, Thomas T, Hicok KC, Spelsberg TC, Riggs BL 1999 Differentiation of human marrow stromal precursor cells: Bone morphogenetic protein-2 increases OSF2/CBFA1, enhances osteoblast commitment, and inhibits late adipocyte maturation. J Bone Miner Res 14:1522–1535. 18. Chen D, Ji X, Harris MA, Feng JQ, Karsenty G, Celeste AJ, Rosen V, Mundy GR, Harris SE 1998 Differential roles for bone morphogenetic protein (BMP) receptor type IB and IA in differentiation and specification of mesenchymal precursor cells to osteoblast and adipocyte lineages. J Cell Biol 142: 295–305. 19. Lee KS, Kim HJ, Li QL, Chi XZ, Ueta C, Komori T, Wozney JM, Kim EG, Choi JY, Ryoo HM, Bae SC 2000 Runx2 is a common target of transforming growth factor beta 1 and bone morphogenetic protein 2, and cooperation between Runx2 and

1439

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

Smad5 induces osteoblast-specific gene expression in the pluripotent mesenchymal precursor cell line C2C12 Mol Cell Biol 20: 8783– 8792. Kujubu DA, Fletcher BS, Varnum BC, Lim RW, Herschman HR 1991 TIS10, a phorbol ester tumor promoter-inducible mRNA from Swiss 3T3 cells, encodes a novel prostaglandin synthase/cyclooxygenase homologue. J Biol Chem 266: 12866 –12872. Fletcher BS, Kujubu DA, Perrin DM, Herschman HR 1992 Structure of the mitogen-inducible TIS10 gene and demonstration that the TIS10-encoded protein is a functional prostaglandin G/H synthase. J Biol Chem 267:4338 – 4344. Bazan NG, Fletcher BS, Herschman HR, Mukherjee PK 1994 Platelet-activating factor and retinoic acid synergistically activate the inducible prostaglandin synthase gene. Proc Natl Acad Sci USA 91:5252–5256. Morham SG, Langenbach R, Loftin CD, Tiano HF, Vouloumanos N, Jennette JC, Mahler JF, Kluckman KD, Ledford A, Lee CA 1995 Prostaglandin synthase 2 gene disruption causes severe renal pathology in the mouse. Cell 83:473– 482. Okada Y, Lorenzo JA, Freeman AM, Tomita M, Morham SG, Raisz LG, Pilbeam CC 2000 Prostaglandin G/H synthase-2 is required for maximal formation of osteoclast-like cells in culture J Clin Invest 105:823– 832. Pilbeam CC, Rao Y, Voznesensky O, Kawaguchi H, Alander CB, Raisz LG, Herschman HR 1997 Transforming growth factor-beta1 regulation of prostaglandin G/H synthase-2 expression in osteoblastic MC3T3–E1 cells. Endocrinology 138: 4672– 4682. Okada Y, Voznesensky O, Herschman HR, Harrison JR, Pilbeam CC 2000 Identification of multiple cis-acting elements mediating the induction of prostaglandin G/H synthase-2 by phorbol ester in murine osteoblastic cells. J Cell Biochem 78:197–209. Xie W, Herschman HR 1996 Transcriptional regulation of prostaglandin synthase 2 gene expression by platelet-derived growth factor and serum. J Biol Chem 271:31742–31748. Raisz LG, Simmons HA 1985 Effects of parathyroid hormone and cortisol on prostaglandin production by neonatal rat calvaria in vitro. Endocr Res 11:59 –74. Pilbeam CC, Fall PM, Alander CB, Raisz LG 1997 Differential effects of nonsteroidal anti-inflammatory drugs on constitutive and inducible prostaglandin G/H synthase in cultured bone cells. J Bone Miner Res 12:1198 –1203. Schinke T, Karsenty G 1999 Characterization of Osf1, an osteoblast-specific transcription factor binding to a critical cis-acting element in the mouse osteocalcin promoters. J Biol Chem 274:30182–30189. Kern B, Shen J, Starbuck M, Karsenty G 2001 Cbfa1 contributes to the osteoblast-specific expression of type I collagen genes J Biol Chem 276:7101–7107. Sato M, Morii E, Komori T, Kawahata H, Sugimoto M, Terai K, Shimizu H, Yasui T, Ogihara H, Yasui N, Ochi T, Kitamura Y, Ito Y, Nomura S 1998 Transcriptional regulation of osteopontin gene in vivo by PEBP2alphaA/CBFA1 and ETS1 in the skeletal tissues. Oncogene 17:1517–1525. Porte D, Tuckermann J, Becker M, Baumann B, Teurich S, Higgins T, Owen MJ, Schorpp-Kistner M, Angel P 1999 Both AP-1 and Cbfa1-like factors are required for the induction of interstitial collagenase by parathyroid hormone. Oncogene 18: 667– 678. Selvamurugan N, Brown RJ, Partridge NC 2000 Regulation of collagenase-3 gene expression in osteoblastic and nonosteoblastic cell lines. J Cell Biochem 79:182–190. Jimenez MJ, Balbin M, Lopez JM, Alvarez J, Komori T, Lopez-Otin C 1999 Collagenase 3 is a target of Cbfa1, a transcription factor of the runt gene family involved in bone formation. Mol Cell Biol 19:4431– 4442.

1440 36. Lee MH, Javed A, Kim HJ, Shin HI, Gutierrez S, Choi JY, Rosen V, Stein JL, van Wijnen AJ, Stein GS, Lian JB, Ryoo HM 1999 Transient upregulation of CBFA1 in response to bone morphogenetic protein-2 and transforming growth factor beta1 in C2C12 myogenic cells coincides with suppression of the myogenic phenotype but is not sufficient for osteoblast differentiation. J Cell Biochem 73:114 –125. 37. Xiao G, Jiang D, Thomas P, Benson MD, Guan K, Karsenty G, Franceschi RT 2000 MAPK pathways activate and phosphorylate the osteoblast-specific transcription factor, Cbfa1 J Biol Chem 275:4453– 4459. 38. Lou J, Tu Y, Li S, Manske PR 2000 Involvement of ERK in BMP-2 induced osteoblastic differentiation of mesenchymal progenitor cell line C3H10T1/2 Biochem Biophys Res Commun 268:757–762. 39. Hay E, Lemonnier J, Fromigue O, Marie PJ 2001 Bone morphogenetic protein-2 promotes osteoblast apoptosis through a Smad-independent, protein kinase C-dependent signaling pathway. J Biol Chem 276:29028 –29036. 40. Hanai J, Chen LF, Kanno T, Ohtani-Fujita N, Kim WY, Guo W-H, Imamura T, Ishidou Y, Fukuchi M, Shi M-J, Stavnezer J, Kawabata M, Miyazono K, Ito Y 1999 Interaction and functional cooperation of PEBP2/CBF with Smads. J Biol Chem 274:31577–31582. 41. Zhang Y-W, Yasui N, Ito K, Huang G, Fujii M, Hanai J, Nogami H, Ochi T, Miyazono K, Ito Y 2000 A RUNX2/ PEBP2␣/CBFA1 mutation displaying impaired transactivation and Smad interaction in cleidocranial dysplasia. Proc Natl Acad Sci USA 97:10549 –10554. 42. Flanagan AM, Chambers TJ 1992 Stimulation of bone nodule formation in vitro by prostaglandins E1 and E2. Endocrinology 130:443– 438. 43. Scutt A, Bertram P 1995 Bone marrow cells are targets for the anabolic actions of prostaglandin E2 on bone: Induction of a transition from nonadherent to adherent osteoblast precursors. J Bone Miner Res 10:474 – 487.

CHIKAZU ET AL. 44. Jee WS, Ma YF 1997 The in vivo anabolic actions of prostaglandins in bone. Bone 21:297–304. 45. Weinreb M, Suponitzky I, Keila S 1997 Systemic administration of an anabolic dose of PGE 2 in young rats increases the osteogenic capacity of bone marrow. Bone 20:521–526. 46. Okada Y, Tomita M, Gronowicz G, Kawaguchi H, Sohn J, Tanaka Y, Morimoto I, Nakamura T, Raisz LG, Pilbeam CC 2000 Effect of cyclooxygenase-2 gene disruption on osteoblastic function. J Bone Miner Res 15:S1;S217. 47. Nilsson OS, Persson PE 1999 Heterotopic bone formation after joint replacement. Curr Opin Rheumatol 11:127–131. 48. Reikeraas O, Engebretsen L 1998 Effects of ketoralac tromethamine and indomethacin on primary and secondary bone healing. An experimental study in rats. Arch Orthop Trauma Surg 118:50 –52. 49. Dimar JR, Ante WA, Zhang YP, Glassman SD 1996 The effects of nonsteroidal anti-inflammatory drugs on posterior spinal fusions in the rat. Spine 21:1870 –1876. 50. Traianedes K, Shima N, Mochizuki S, Yamaguchi K, Kinosaki M, Yano K, Shibata O, Udagawa N, Yasuda H, Suda T, Bonewald LF 1998 5-Lipoxygenase metabolites inhibit bone formation in vitro. Endocrinology 139:3178 –3184.

Address reprint requests to: Carol C. Pilbeam, M.D. Room AM047 University of Connecticut Health Center 263 Farmington Avenue Farmington, CT 06030, USA

Received in original form September 24, 2001; in revised form February 28, 2002; accepted April 2, 2002.