Prostaglandin E₂ acts via bone marrow macrophages to block PTH-stimulated osteoblast differentiation in vitro

Abstract

Intermittent PTH is the major anabolic therapy for osteoporosis while continuous PTH causes bone loss. PTH acts on the osteoblast (OB) lineage to regulate bone resorption and formation. PTH also induces cyclooxygenase-2 (COX-2), producing prostaglandin E₂ (PGE₂) that can act on both OBs and osteoclasts (OCs). Because intermittent PTH is more anabolic in Cox-2 knockout (KO) than wild type (WT) mice, we hypothesized COX-2 might contribute to the effects of continuous PTH by suppressing PTH-stimulated differentiation of mesenchymal stem cells into OBs. We compared effects of continuous PTH on bone marrow stromal cells (BMSCs) and primary OBs (POBs) from Cox-2 KO mice, mice with deletion of PGE₂ receptors (Ptger4 and Ptger2 KO mice), and WT controls. PTH increased OB differentiation in BMSCs only in the absence of COX-2 expression or activity. In the absence of COX-2, PTH stimulated differentiation if added during the first week of culture. In Cox-2 KO BMSCs, PTH-stimulated differentiation was prevented by adding PGE₂ to cultures. Co-culture of POBs with M-CSF-expanded bone marrow macrophages (BMMs) showed that the inhibition of PTH-stimulated OB differentiation required not only COX-2 or PGE₂ but also BMMs. Sufficient PGE₂ to mediate the inhibitory effect was made by either WT POBs or WT BMMs. The inhibitory effect mediated by COX-2/PGE₂ was transferred by conditioned media from RANKL-treated BMMs and could be blocked by osteoprotegerin, which interferes with RANKL binding to its receptor on OC lineage cells. Deletion of Ptger4, but not Ptger2, in BMMs prevented the inhibition of PTH-stimulated OB differentiation. As expected, PGE₂ also stimulated OB differentiation, but when given in combination with PTH, the stimulatory effects of both were abrogated. These data suggest that PGE₂, acting via EP4R on BMMs committed to the OC lineage, stimulated secretion of a factor or factors that acted to suppress PTH-stimulated OB differentiation. This suppression of OB differentiation could contribute to the bone loss seen with continuous PTH in vivo.

Introduction

Parathyroid hormone (PTH) is the major regulator of calcium homeostasis through its actions on bone and kidney. PTH is critical for bone remodeling, exerting both anabolic and catabolic effects on bone in vivo by activating the PTH1 receptor, a G-protein coupled receptor, on osteoblast (OB) lineage cells [1, 2]. Intermittent PTH was the first anabolic agent approved for osteoporosis therapy in the USA [1, 3]. For reasons still not completely understood, daily injections of PTH increase bone formation more than resorption, thereby increasing bone mass, while continuous infusion increases bone resorption more than formation, resulting in bone loss [4–6]. Despite the anabolic effects of PTH in vivo and the demonstration that PTH can stimulate OB precursors or mesenchymal stem cells (MSCs) to differentiate into OBs [2, 7], it has been difficult to demonstrate osteogenic effects of PTH in vitro. A number of in vitro studies have reported that PTH present continuously in culture inhibits OB differentiation [8–11]. These observations suggest that the bone loss associated with continuous PTH is not simply the result of increased resorption but may also involve suppressed differentiation of bone-forming cells.

PTH is also a potent inducer of cyclooxygenase-2 (COX-2) and prostaglandin (PG) production, especially PGE₂, in OB lineage cells [12, 13]. PGs are locally produced lipids that have receptors on both OB and osteoclast (OC) lineage cells [14, 15]. PGE₂ is abundantly expressed in bone and can have important roles in skeletal metabolism. Although originally identified as a resorption agonist, PGE₂ also increases bone formation in vivo [16] and OB differentiation in vitro [14, 15]. Multiple regulators of bone metabolism induce COX-2, the major enzyme responsible for PG production. For some of these, their induction of COX-2 enhances or mediates their stimulation of OB differentiation in vitro [14, 17–19]. In addition, endogenous PGs are also necessary for normal bone repair [20] and a critical role for COX-2 and PGE₂ in triggering Wnt/β-catenin signaling in the anabolic response to mechanical loading has been proposed [21]. Four G-protein coupled receptors, EP1, EP2, EP3 and EP4, are associated with effects of PGE₂. EP2 and EP4, which activate Gα_s and stimulate cAMP formation, have predominant roles in both PGE₂-stimulated bone resorption and formation [15]. EP3 is coupled to Gα_i and inhibits cAMP, while EP1 acts largely by increasing calcium flux and perhaps protein kinase C (PKC) [22].

Because PTH induces PGE₂ production and because PTH and PGE₂ both have major actions via similar Gα_s/cAMP-activated pathways [23, 24], our initial hypothesis was that PGE₂ was the local mediator of some of the anabolic actions of PTH. However, we found intermittent PTH in vivo to be more anabolic in Cox-2 KO mice than in WT mice, suggesting an inhibitory interaction of PTH and PGs [25]. In the current study, we extend our initial findings on the inhibitory interaction of PTH and PGs in vitro [26] to show that the stimulatory effect of PTH on OB differentiation in BMSCs occurred only when COX-2 activity was absent in both mesenchymal and hematopoietic cells. Using co-cultures and conditioned media (CM) from bone marrow macrophages (BMMs), we show that the inhibition of PTH-stimulated OB differentiation was mediated by a factor or factors secreted by hematopoietic cells committed to the OC lineage in response to COX-2 produced PGs or to added PGE₂. This study reveals a new role for COX-2 and PGE₂ in regulating PTH-stimulated responses in bone and a new example of regulation of OB differentiation by OCs.

Materials and methods

Materials

PGE₂, NS398, MRE-269 (prostaglandin IP receptor agonist), dinoprost (PGF_2α receptor agonist) and all other prostanoids used were from Cayman Chemical Company (Ann Arbor, MI). Recombinant mouse macrophage-colony stimulating factor (M-CSF), osteoprotegerin (OPG)/Fc-chimera and RANKL were from R&D systems (Minneapolis, MN). Bovine PTH (bPTH; 1–34) and all other chemicals were from Sigma (St. Louis, MO), unless otherwise noted.

Animals

Mice with disruption of Ptgs2, which produce no functional COX-2 protein, called Cox-2 knockout (KO) mice, in a C57BL/6, 129SV background were the gift of Scott Morham [27]. Ptger2 and Ptger4 KO mice in C57BL/6, 129 backgrounds were gifts from Richard and Matthew Breyer [28, 29]. All KO mice were backcrossed more than 16 generations into the CD-1 (outbred) background. Breeding colonies were refreshed twice a year by regenerating maintenance colonies from mice heterozygous for the deleted or disrupted gene mated with WT mice from Jackson Laboratory (Bar Harbor, ME). For experiments, Cox-2 KO mice were bred by KO × KO mating, and Ptger2 and Ptger4 KO mice were bred by heterozygous × heterozygous mating. Genotyping protocols were as described previously [30–32]. All animal studies were conducted in accordance to the approved protocols by the Animal Care and Use Committee of the University of Connecticut Health Center.

Cell cultures

All cells were cultured in a humidified atmosphere of 5% CO₂ at 37 °C. Basic medium was α-MEM (Invitrogen, Carlsbad, CA), 10% heat inactivated fetal calf serum (HIFCS), 100 U/ml penicillin, and 50 µg/ml streptomycin. Vehicles for the various treatments were as follows: 0.1% ethanol for PGE₂, all other prostanoid receptor agonists, and NS398; 0.1% bovine serum albumin (BSA) in 1× phosphate buffered saline (PBS) for RANKL, M-CSF and OPG; dimethyl sulfoxide for isobutyl methyl xanthine (IBMX); and 0.001 N hydrochloric acid-acidified 0.1% BSA in 1× PBS for PTH.

To make bone marrow stromal cell (BMSC) cultures, whole marrow flushed from tibiae and femora of 6–8 week old mice, plated at 10⁶ nucleated cells/well in 6-well tissue culture dishes and cultured in OB differentiation medium from the time of plating onward. Differentiation medium consisted of basic medium plus 50 µg/ml phosphoascorbate (Wako Pure Chemical Industry, Osaka, Japan). To study mineralization, 8 mM of β-glycerophosphate was added on day 7. Media were changed every 3–4 days. Unless specified, all agents were added from the beginning of culture and with each medium change. To make primary osteoblast (POB) cultures, calvariae from 5 to 6 neonatal mice were dissected free of sutures, minced, washed with 1× PBS and digested with 0.5 mg/ml of collagenase P (Roche Diagnostics, Indianapolis, IN) in a solution of 1 ml 0.25% trypsin/EDTA and 4 ml PBS at 37 °C. Four digests were performed for 10 min each and a final digest for 90 min. Digests 2–5 were pooled and plated at 4 × 10⁴ cells/well in 6-well dishes and cultured in differentiation media. To make bone marrow macrophage (BMM) cultures, we followed the protocols of R. Faccio http://www.orthoresearch.wustl.edu/content/Laboratories/2978/Roberta-Faccio/Faccio-Lab/Protocols.aspx. Briefly, 10⁷ nucleated bone marrow cells/well were plated in 150 mm Petri dishes (Fisher Scientific, Pittsburgh, PA) in basic medium plus 100 ng/ml M-CSF and expanded twice, each for three days, before being used for co-culture or conditioned media experiments.

Co-culture and conditioned media (CM) experiments

For co-culture of BMMs and POBs, POBs were plated at 4 × 10⁴ with 4 × 10⁵ BMMs (1:10 ratio) per well in 6-well tissue culture dishes and cultured in OB differentiation medium. For co-culture of BMMs and BMSCs, BMMs were plated at 1:3 with BMSCs and cultured in OB differentiation medium. To obtain CM, BMMs were re-plated at 6 × 10⁴ cells/well in 12 well tissue culture dishes in basic medium plus 30 ng/ml M-CSF with/without RANKL (30 ng/ml). CM were collected, pooled and centrifuged at 800 rpm for 5 min at 4 °C to get rid of debris and kept frozen until use. For differentiation studies with CM, freshly isolated POBs were plated at 4 × 10⁴/well in 6-well dishes and cultured in 3 parts of CM and 1 part OB differentiation media with 50 ng/ml of OPG to block RANKL–RANK interactions that might generate osteoclasts.

Real-time (quantitative) PCR analysis

Total RNA was extracted with Trizol (Invitrogen) following manufacturer's instructions. 2–5 µg of total RNA was DNase treated (Ambion, Inc., Austin, TX) and converted to cDNA by the High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA). PCR was performed in 96-well plates. Both Assays-on-Demand Gene Expression Taqman primers (Applied Biosystems) and validated Syber Green primers (http://pga.mgh.harvard.edu/primerbank) were used for PCR. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or β-actin served as endogenous control. All primers were checked for equal efficiency over a range of target gene concentrations. Each sample was amplified in duplicate. PCR reaction mixture was run in Applied Biosystems Prism 7300 Sequence Detection System instrument utilizing universal thermal cycling parameters. Data analysis was done using relative quantification (RQ, ΔΔCt) or the relative standard curve method.

OB and osteoclast-like cell (OCL) staining

For alkaline phosphatase (ALP) staining, cells were fixed and stained with an alkaline phosphatase kit (Sigma) using the manufacturer's instructions. Dishes were air dried and scanned into the computer. To assess mineralization, cells were washed with PBS, fixed in 100% V/Vmethanol on ice for 30 min and stained with 40 mM alizarin red-S pH 4.2 for 10 min at room temperature. Dishes were washed with water, air dried and scanned into the computer. For tartrate resistant acid phosphatase (TRAP) staining, cells were fixed with 2.5% glutaraldehyde in PBS for 30 min at room temperature and stained by the Leukocyte Acid Phosphatase Kit (Sigma) following company's instructions.

Oil red O staining

BMSCs were cultured for 14 days under similar conditions used for OB differentiation but without phosphoascorbate and β-glycerophosphate in the culture medium. Instead, 1 µM of insulin was added to the medium on day 7 to induce formation of fat bodies. For staining, cells were washed twice with 1× PBS, fixed with 4% paraformaldehyde for 15 min at room temperature, rinsed with water and then incubated with oil red O working solution (3 parts to 2 parts water) for 1 h at room temperature. Dishes were washed with water, air dried and scanned into the computer.

Prostaglandin (PG)E₂ assay

Media were removed from cultured cells and frozen until assay. PGE₂ accumulation was measured using an enzyme immunoassay (correlate-EIA™) kit following the manufacturer's instructions (Assay Designs, Ann Arbor, MI).

Intracellular cAMP measurement

Confluent POBs were treated with 3 parts CM and 1 part of OB differentiation medium containing 0.5 mM isobutyl methyl xanthine (IBMX) 1 h prior to adding PTH or PGE₂ for 20 min. Cells were scraped off in 400 µl/well of ice-cold ethanol. The ethanolic cell suspension was collected in tubes and centrifuged at 1500 ×g for 10 min at 4 °C. Supernatants were collected and evaporated to dryness using a lyophilizer. cAMP was measured using an enzyme immunoassay kit (Cayman Chemical, Ann Arbor, MI).

Statistics

All data are presented as means ± SEM. Analysis was performed using Sigma Plot 11.0 (Systat Software, Inc.). Experiments involving several genotypes (or combinations of genotypes in co-cultures) and treatments were examined by two-way ANOVA, followed by post hoc Bonferroni pairwise multiple comparison. If data were not normally distributed, they were transformed (log 10) before ANOVA. Comparison of multiple treatments to a single control was examined by one-way ANOVA, followed by Bonferroni pairwise multiple comparison. If these data were not normally distributed, they were examined by one-way ANOVA on ranks, followed by Dunn's Test for all pairwise multiple comparisons.

Results

Endogenous PGs suppressed PTH-stimulated OB differentiation in BMSC cultures

To study effects of endogenous PGs on PTH-stimulated OB differentiation, we used BMSCs from WT and Cox-2 KO mice. Despite the constitutive expression of Cox-1, very little PGE₂ is measurable in the media of Cox-2 KO BMSC cultures [14, 33]. It is expected that there will be “basal” production of PGE₂ in WT BMSC cultures because fresh serum stimulates Cox-2 expression [34]. Because PGE₂ can stimulate OB differentiation, this basal production often leads to increased OB differentiation in vehicle-treated WT compared to KO or NSAID-treated WT cultures, as seen here (e.g., Figs. 1A–E). PTH is expected to further induce Cox-2 expression and PGE₂ production in these cultures [12, 13].

Fig. 1.

Effects of COX-2 expression or activity on PTH-stimulated OB differentiation in bone marrow stromal cell (BMSC) cultures. BMSCs from WT and Cox-2 knockout (KO) mice were plated in osteogenic media and treated with vehicle (VEH) or PTH (10 nM) begun at the time of plating and given with each media change. Gene expression was measured by real time PCR (qPCR). (A) Time course for PTH effects on Alp mRNA expression in WT and KO cells. (B) Osteocalcin mRNA expression at day 21 of the culture shown in (A). Effects of NS398 (0.1 µM), a selective inhibitor of COX-2 activity, on PTH-stimulated Alp mRNA (C), Osteocalcin mRNA (D) and mineralization (E) in WT BMSCs, measured at 14 days of culture. Bars are means ± SEM for n = 3 wells of cells. ^aSignificant effect of treatment relative to vehicle, p < 0.01, ^bp < 0.05. ^cSignificant effect of genotype, p < 0.01, ^dp < 0.05.

BMSCs were cultured with PTH (10 nM) added at plating of cells and with each media change. This protocol should provide continuous exposure to PTH because PTH has been shown to be stable in culture up to 72 h between medium changes [35]. As we showed previously [26], PTH stimulated OB differentiation in Cox-2 KO, but not WT, BMSC cultures. PTH stimulated marked increases in Alp and Osteocalcin mRNA (Figs. 1A,B) and alizarin red staining (data not shown) in KO cultures, but not in WT cultures. In WT cultures, PTH decreased, or tended to decrease, markers of OB differentiation relative to vehicle treatment. The stimulatory effect of PTH in Cox-2 KO cultures was seen by day 7 of culture and was maintained throughout 3 weeks of culture (Fig. 1A).

To determine if the inhibitory effect of COX-2 was due to COX-2 activity, we examined treatment with a selective inhibitor of COX-2 activity, NS398. NS398 restored the ability of PTH to stimulate Alp and Osteocalcin mRNA expression and alizarin red staining in WT cultures, confirming that the inhibitory effects were due to PG production (Figs. 1C–E).

Because there may be reciprocal effects between OB and adipocyte differentiation [36, 37] and because PTH can regulate adipocyte differentiation [35], we examined expression of Adiponectin, a marker of adipocytes, and Pparγ, a transcription factor that may be important not only for stimulating adipogenesis but also for suppressing osteogenesis [38]. PTH inhibited both Adiponectin and Pparγ expression on day 14 of culture in WT, but not Cox-2 KO, cultures (Figs. 2A,B). Similar patterns were seen on day 21 (data not shown). Oil red O staining for mature adipocytes was consistent with the gene expression (Fig. 2C). Hence, differentiation of both OBs and adipocytes in these cultures was inhibited by endogenous PGs.

Fig. 2.

PTH regulation of adipogenesis and osteoclast-like cell (OCL) formation in BMSCs from WT and Cox-2 KO mice. BMSCs were cultured in osteogenic media, except as noted, and treated with vehicle (VEH) or PTH (10 nM). mRNA expression was measured by qPCR. (A) Adiponectin and (B) Pparγ mRNA measured at 14 days of culture. Data are from same experiment as shown in Figs. 1 (A,B). Bars are means ± SEM for n = 3. ^aSignificant effect of treatment relative to vehicle, p < 0.01. ^bSignificant effect of genotype, p < 0.01. (C) Oil red Ostaining in BMSCs cultured for 14 days. Cells were cultured without phosphoascorbate and β-glycerophosphate. Insulin (1 µM) was added to the medium on day 7. (D) Tartrate resistant acid phosphatase (TRAP) staining in BMSCs at day 7 of culture. (E) Alkaline phosphatase (ALP) staining of colonies in PTH-treated Cox-2 KO BMSCs at day 9 of culture. (F) TRAP staining (40× magnification) at day 7.

BMSC cultures differ from the marrow cultures used for studying OC differentiation in that they are plated at lower density and have phosphoascorbate in the media. PTH stimulated formation of osteoclast-like cells (OCLs), defined as tartrate resistant acid phosphatase (TRAP) multinucleated cells, during the first week of culture in both WT and Cox-2 KO BMSCs. OCLs were seen at days 4–5 of culture and were abundant by day 7, resulting in the appearance of “empty” areas in the center of ALP stained colonies (Figs. 2D–F). No OCLs were formed in control cultures (Fig. 2D). OCLs had largely disappeared by days 12–14 (data not shown). It was not possible to quantify OCL number in these cultures since most were covered by a canopy of cells. Although there appeared grossly to be little difference in TRAP staining between WT and KO cells, these observations raised the possibility that differences in PTH-simulated OB differentiation between WT and KO cultures might be due to space-occupying OCLs.

To determine the window of time during which PTH needed to be present to stimulate OB differentiation, we cultured BMSCs for different periods of time with PTH and measured Alp mRNA at day 14 of culture. When PTH was given to Cox-2 KO BMSCs from days 0–3, 3–7 or 0–7 of culture, it increased Alp mRNA (Fig. 3A). However, when PTH was not started until day 7 of culture, it did not increase OB differentiation. PTH did not stimulate Alp mRNA expression in WT BMSCs when given for any period of time. As further confirmation that PTH acted during the first week of culture to stimulate OB differentiation, we treated WT BMSCs with NS398 from days 3 to 7 or from days 0 to 14 and measured mineralization on day 14. PTH stimulated mineralization to a similar extent in both cases (Fig. 3B).

Fig. 3.

Window of time for PTH stimulation of OB differentiation in BMSCs. BMSCs from WT and Cox-2 KO mice were plated in osteogenic media and treated with vehicle (VEH) or PTH (10 nM) begun at the time of plating and given with each media change. Gene expression as measured by qPCR. (A) Comparison of the Alp mRNA response to different periods of treatment with PTH. BMSCs were given PTH for the days indicated on x-axis. All cultures were extracted for RNA after 14 days. Bars are means ± SEM for n = 3 wells of cells. ^a Significant effect of PTH, p < 0.01. ^b Significant effect of genotype, p < 0.01. (B) Comparison of the mineralization response to PTH in WT BMSCs given NS398 for 3–7 days versus 0–14 days. All cultures were stained for alizarin red after 14 days of culture.

Because the window for PTH stimulation of OB differentiation in Cox-2 KO cultures was early in culture and because PGs cause PTH to decrease both OB and adipocyte differentiation, it is possible that PGs are modulating the actions of PTH on MSCs, which are likely to be available only early in culture.

Suppression of PTH-stimulated OB differentiation required both PGs and hematopoietic cells

Because OCLs formed early in BMSC cultures, beginning during the window of time for the stimulatory effects of PTH, we postulated that OC lineage cells might play a role in the inhibitory effects of PGs. If so, the inhibitory effect should not be seen in primary osteoblasts (POBs). However, in our previous study, we also observed an inhibitory effect of PGs on PTH-stimulated OB differentiation in POB cultures [26]. When we examined our POB cultures for the ability to form OCLs, we found that both PTH, which increases RANKL mRNA expression in POBs, and exogenous RANKL induced formation of cells that stained for TRAP in these cultures (Fig. 4A). If the minced calvarial pieces were carefully washed multiple times to eliminate marrow, many fewer TRAP staining cells were seen (Fig. 4A). With the increased washing of calvarial pieces, we found that PTH stimulated OB differentiation in WT POBs (Fig. 4B) and that NS398 had no effect on PTH-stimulated OB differentiation (Fig. 4C).

Fig. 4.

Effects of COX-2 expression or activity on PTH-stimulated differentiation in primary osteoblast (POB) cultures. POBs from WT and Cox-2 KO mice were plated in osteogenic media and treated with vehicle (VEH), PTH (10 nM), or other agents begun at the time of plating and given with each media change. Gene expression was measured by qPCR. (A) Stimulation of TRAP-staining cell formation in POBs cultured from WT mice. Cultures were treated with PTH or RANKL (30 ng/ml) for 8 days. (B) Alp mRNA expression in WT and Cox-2 KO POBs treated for 14 days. (C) Effect of NS398 (1 µM), an inhibitor of COX-2 activity, on Alp mRNA expression in WT POBs at 14 days. (D) Comparison of PTH or PGE₂ (1 µM) or their combination, PTH+PGE₂, all in the presence of NS398 (1 µM), on Alp mRNA expression at 14 days. (E) Measurement of intracellular cAMP in WT POB cultures in the presence of NS398 (1 µM). Cells were treated with the same agonists as in (D) for 15 min in the presence of IBMX (0.5 mM). (F) Treatment of Cox-2 KO POBs with PTH, PGE₂ (10 nM), or PTH + PGE₂ in the presence of osteoprotegerin (OPG, 50 ng/ml), an inhibitor of RANKL-mediated osteoclast formation. Osteocalcin mRNA was measured at 21 days. Bars are means ± SEM for n = 3 wells of cells. ^aSignificant effect of treatment, p < 0.01; ^bp < 0.05. ^cSignificant effect of genotype, p < 0.05. ^dSignificant effect of PTH + PGE₂ relative to either treatment alone, p < 0.01. Nd = below limits of assay.

On the assumption that PGE₂ might be the PG mediating the inhibitory effects of COX-2, we examined the effects of adding PGE₂ to PTH (Fig. 4D). (We continued to use either Cox-2 KO POBs or treat with NS398 because chronic exposure to PGE₂ in the media might down regulate responses to added PGE₂.) PTH or PGE₂ alone stimulated Alp mRNA in POBs at 14 days of culture, but the combination of PTH and PGE₂ had no greater effect than either agent alone, suggesting that some inhibition remained (Fig. 4D). However, treatment of POBs with PTH, PGE₂ and the combination for 15 min had an additive effect on cAMP production (Fig. 4E), the pathway through which both agents are supposed to produce anabolic effects. Because we had previously observed that the combination of PGE₂ and PTH had additive or greater effects on OCL formation in bone marrow cultures [31], we treated cultures with OPG, which interrupts the RANK–RANKL interaction. In the presence of OPG, the combination of PTH and PGE₂ had additive effects on PTH-stimulated Osteocalcin mRNA at 14 days (Fig. 4F).

These data suggest that RANKL-stimulated hematopoietic cells were necessary for suppression of PTH-stimulated OB differentiation. In addition, the data indicate that PGE₂ itself was not the factor that acted on POBs to inhibit PTH-stimulated OB differentiation.

Bone marrow macrophages (BMMs) expressing COX-2 were sufficient to prevent the PTH-stimulated OB differentiation

The addition of WT BMMs to Cox-2 KO BMSCs blocked the PTH-stimulation of OB differentiation (Fig. 5A). When Cox-2 KO POBs were co-cultured with BMMs from WT or Cox-2 KO mice, the presence of WT BMMs, but not KOBMMs, prevented the PTH-stimulated increase in OB mineralization (Fig. 5B). To confirm a role for cells committed to the OC lineage in mediating the inhibitory effect of PGs, we treated BMSCs with OPG. When OPG was present, PTH stimulated OB differentiation in WT as well as Cox-2 KO BMSCs (Figs. 5C–E). Although OPG is reported to have direct effects on OB differentiation [39], we did not see effects of OPG alone on OB differentiation. We considered the possibility that OPG might block inhibitory effects by suppressing PG production in these cultures. There was a reduction, not statistically significant, in PTH-stimulated medium PGE₂ accumulation in the presence of OPG from 7.3 ± 0.4 to 4.4 ± 1.6 nM, which, as will be discussed below, should not have prevented the inhibitory effects. These results are consistent with the previous data suggesting that the cells mediating the inhibition of PTH-stimulated OB differentiation are committed to the OC lineage.

Fig. 5.

Effects of bone marrow macrophages (BMMs) on PTH-stimulated OB differentiation. BMSCs or POBs from WT and Cox-2 KO mice were cultured alone or with bone marrow macrophages (BMMs). All cultures were in osteogenic media and treated with vehicle (VEH), PTH (10 nM), or other agents began at the time of plating and given with each media change. Gene expression was measured by qPCR. (A) Alizarin red staining at 14 days in Cox-2 KO BMSCs co-cultured with BMMs from KO and WT mice. BMMs were plated at a ratio of 1:3 to BMSCs. (B) Alizarin red staining at 14 days in Cox-2 KO POBs co-cultured with BMMs from WT and KO mice. POBs were plated at a ratio of 1:10 to BMMs. (C) Alp and (D) Osteocalcin mRNA expression in WT and Cox-2 KO BMSCs at 14 days following treatment with PTH, plus/minus osteoprotegerin (OPG, 30 ng/ml), which interferes with RANKL-RANK binding. (E) Alizarin red staining at 14 days of culture in WT BMSCs treated with PTH plus/minus OPG (50 ng/ml). Bars are means and SEM for n = 3 wells of cells. ^aSignificant effect of PTH, p < 0.01. ^bSignificant effect of genotype, p < 0.01. ^cSignificant effect of OPG, p < 0.01.

Although OBs are generally assumed to be the major source of PGs in bone, these co-culture results suggested that WT BMMs produced sufficient PGs to mediate the inhibitory effects. To examine the relative roles of OB and OC lineage cells in producing PGE₂ in these cultures, we measured medium PGE₂ accumulation in co-cultures of POBs and BMMs from WT and Cox-2 KO mice and compared with OB differentiation measured by Osteocalcin mRNA (Table 1). As expected, when KO POBs were co-cultured with KO BMMs, medium PGE₂ was undetectable in vehicle or PTH-stimulated cultures [31, 33]. WT BMMs (plated at 10:1 ratio with POBs) made more PGE₂ under basal conditions than WT POBs. The basal level of PGE₂ production by POBs was likely due to the serum induction of COX-2 [34]. PTH stimulated PGE₂ production 2- to 3-fold in co-cultures with WT POBs but had little effect in cultures with KO POBs, consistent with the expected absence of PTH receptors on BMMs. The small increase in PGE₂ in the WT BMM, KO POB co-culture might be due to PTH-stimulated RANKL expression in the POBs, which subsequently induced COX-2 in BMMs [40].

Table 1.

Medium PGE₂ and Osteocalcin mRNA expression in co-cultured POBs and BMMs from WT and COX-2 KO mice treated with vehicle or PTH (10 nM).

	WT BMMs		KO BMMs
	WT POBs	KO POBs	WT POBs	KO POBs
(1) PGE2 (nM) accumulated in media during first week of culture
Vehicle	6.8 ± 0.1	6.4 ± 0.2	3.8 ± 0.1a,b	nd
PTH	14.2 ± 0.4c	7.2 ± 0.2a,c	11.3 ± 0.2a,b,c	nd
(2) Osteocalcin mRNA at day 14 of culture
Vehicle	1.71 ± 0.13d	1.33 ± 0.07d	1.07 ± 0.12d	0.55 ± 0.05
PTH	0.12 ± 0.01c,d	0.09 ± 0.01c,d	0.19 ± 0.02c,d	10.9 ± 1.24c

(1) PGE₂ data are means ± SEM for n = 3. Nd is not detectable. Media were collected at time of media change (day 3 and day 7) and equal aliquots pooled for measurement.

(2) Osteocalcin data are means ± SEM for n = 4. For both sets, data were log 10 transformed before two-way ANOVA analysis to achieve normal distribution.

^aSignificantly different from WT BMMs + WT POBs, p < 0.01.

^bSignificantly different from WT BMMs + KO POBs, p < 0.01.

^cSignificant effect of PTH, p < 0.01, p < 0.05.

^dSignificantly different from KO BMMs + KO POBs, p < 0.01.

In vehicle-treated cultures, the Osteocalcin levels decreased as PGE₂ levels decreased (Table 1). PTH-stimulated Osteocalcin mRNA expression was increased 20-fold relative to vehicle treatment in KO BMM-KO POB cultures, which had no detectable PGE₂ production. In all other combinations, which contained WT POBs or WT BMMs and did produce measurable PGE₂, PTH-stimulated Osteocalcin expression was inhibited relative to the KO-KO combination. Hence, either POBs or BMMs expressing COX-2 were sufficient to prevent the PTH-stimulated OB differentiation in this culture system.

Exogenous PGE₂ suppressed PTH-stimulated OB differentiation in BMSCs

In many of our experiments in BMSC cultures (Figs. 1, 3) or in cultures with both POBs and BMMs (Table 1), but not in POBs cultured alone (Fig. 5), PTH given in the presence of COX-2 expression resulted in decreased Alp or Osteocalcin expression relative to vehicle-treated cultures. Since some of the OB differentiation in vehicle-treated cultures is explainable by the serum induction of COX-2 expression and endogenous PGE₂ production (Table 1) [34], this observation suggests that, in the presence of BMMs, the stimulatory effect of endogenous PGE₂ on OB differentiation was suppressed in the presence of PTH.

To look at this possibility more directly, we treated BMSC cultures with PTH (10 nM), PGE₂ (10 nM) and the combination (Fig. 6A). PGE₂ stimulated Bone sialoprotein (Bsp) mRNA at 14 days in both WT and Cox-2 KO BMSCs. (The small but significant increase in the effects of PGE₂ in KO cells has been seen before and may be due to down regulation of PGE₂ receptors due to chronic exposure to endogenous PGE₂ in WT cultures). Although both PTH and PGE₂ individually stimulated Bsp mRNA expression in KO cultures, the combination of PTH and PGE₂ had no stimulatory effect.

Fig. 6.

Effects of exogenous PGs on PTH-stimulated differentiation in BMSCs. BMSCs were treated with vehicle (VEH), PTH, PG or the combination of PTH and PG at plating and at every media change. (A) Bone sialoprotein (Bsp)mRNA expression in WT and Cox-2 KOBMSCs treated for 14 days with PTH (10 nM) and PGE₂ (10 nM). Bars are means ± SEM for n = 3 wells. ^aSignificant effect of treatment relative to vehicle, p < 0.01, ^bp < 0.05. ^cSignificantly different from PTH alone, p < 0.01. ^dSignificantly different from PGE₂ alone, p < 0.01. ^eSignificant effect of genotype, p < 0.01. (B) Alp mRNA in Cox-2 KOBMSCs at 14 days of culture: effect of adding varying doses of PGE2 to PTH (10 nM). Symbols are means ± SEM for n = 3 wells. PTH and all doses of PGE2 alone increased Alp mRNA relative to vehicle (p < 0.01). The combination of PTH and PGE₂ at all doses reduced Alp mRNA relative to either agent alone (p < 0.01). (C) Alp mRNA in Cox-2 KOBMSCs at 14 days of culture: effect of adding varying doses of PTH to PGE2 (10 nM). Symbols are means ± SEM for n = 3 wells. PGE2 and all doses of PTH increased Alp mRNA relative to vehicle (p < 0.01). The combination of PTH and PGE2 at all doses reduced Alp mRNA relative to either agent alone (p < 0.01). (D) Osteocalcin mRNA in Cox-2 KOBMSCs at 14 days of culture: effect of OPG (50 ng/ml). Symbols are means ± SEM for n = 3 wells. ^aSignificant effect of treatment relative to vehicle, p < 0.01. ^bSignificantly different from PTH or PGE2 alone, p < 0.01. ^cSignificant effect of OPG, p < 0.01. (E) Osteocalcin mRNA in Cox-2 KO BMSCs at 14 days of culture: effect of different PGs (all at 0.1 µM). Symbols are means ± SEM for n = 3 wells. ^aSignificant effect of treatment relative to vehicle, p < 0.01. ^bSignificantly different from all other PGs alone, p < 0.01. ^cSignificantly different from PTH or PGE2 alone and all other combinations of PTH and PG, p < 0.01.

To better understand the dose range over which these effects occurred, we treated Cox-2 KO BMSCs with PTH (10 nM) ± PGE₂ (0.1 nM to 0.1 µM) for 14 days (Fig. 6B). PTH and all doses of PGE₂ alone increased Alp mRNA relative to vehicle, but PTH combined with PGE₂, at any dose, resulted in decreased Alp mRNA expression relative to either PTH or PGE₂ alone. Similarly, the combination of a single dose of PGE₂ (10 nM) with several doses of PTH (0.1 nM to 10 nM) decreased Alp mRNA expression relative to PGE₂ or PTH alone (Fig. 6C). To examine a role for BMMs in the inhibition of OB differentiation by the combination of PTH and PGE₂, we examined the effects of OPG (Fig. 6D). In the presence of OPG, the combination of PTH and PGE₂ had additive stimulatory effects on Osteocalcin mRNA.

Other PGs could be involved in the inhibitory effect of COX-2. To screen for some other likely candidates, we treated Cox-2 KO BMSCs with PGE₂ and compared with other PG receptor agonists, all at 0.1 µM (Fig. 6E). Because PGI₂ is unstable, we used MRE-269, a stable IP receptor agonist. For PGF₂α, we used dinoprost, an FP receptor agonist. All cultures were with Cox-2 KO cells because PGs can induce COX-2 expression and make PGE₂, which could confound the comparison [41]. PGE₂ was the only prostanoid that stimulated Osteocalcin mRNA, and the only prostanoid that resulted in loss of the stimulatory effect when added to PTH.

These data on exogenous PGE₂, along with the previous data on endogenous PGs, can be summarized as follows (Table 2). The inhibition of PTH-stimulated OB differentiation was only seen in the presence of both BMMs and endogenous or exogenous PGs. In the absence of BMMs, there was no inhibitory effect of COX-2 or PGE₂, and PTH and PGE₂ were additive. In the presence of BMMs, treatment with the combination of PTH and PGE₂, each of which was stimulatory alone, produced no stimulatory effect.

Table 2.

Summary of effects of PTH, COX-2/PGE2 and BMMs on OB differentiation.

Treatments	Cells
	OBs with BMMs		OBs without BMMs
	COX-2 in OB or BMM	No COX-2	COX-2	No COX-2
PTH	−	+	+	+
PGE2	+	+	+	+
PTH + PGE2	−	−	++	++

Inhibition of PTH-stimulated OB differentiation required expression of EP4 receptors (EP4) on BMMs

The need for BMMs to be present in order to see inhibition of PTH effects suggests that PGs are acting on BMMs to cause the inhibition. As indicated by the studies above, PGE₂ is a likely candidate for the PG involved. The effects of PGE₂ in bone have been most often associated with cAMP production and protein kinase A (PKA) activation, suggesting an important role for the PGE₂ receptors EP2 and EP4, which are both coupled to Gαs. Both EP2 and EP4 are reported to be expressed by bone marrow macrophage OC precursors [42]. To examine the roles of these receptors, BMSCs from WT and Ptger2 or Ptger4 KO mice were cultured with PTH (Figs. 7A,B). PTH stimulated OB differentiation in Ptger4 KO cultures but inhibited in WT and Ptger2 KO cultures. For comparison, we treated these cultures with PGE₂. PGE₂ stimulated Osteocalcin expression in both WT and Ptger2 KO BMSC (Figs. 7A,B). As expected from previous experiments, which showed a major role for EP4 in the osteogenic effects of PGE₂ [43, 44], deletion of Ptger4 greatly reduced PGE₂-mediated OB differentiation.

Fig. 7.

Effects of deleting Ptger2 and Ptger4 on PTH-stimulated OB differentiation. All cells were cultured in osteogenic media. Treatments were begun at the time of plating and given with each media change. Gene expression was measured by qPCR. (A) Osteocalcin mRNA and (B) alizarin red staining in WT, Ptger2 KO and Ptger4 KO BMSCs at day 14 of culture. BMSCs were treated with vehicle (VEH), PTH (10 nM) and PGE₂ (10 nM). (C) Osteocalcin mRNA in Cox-2 KO POBs co-cultured with BMMs from WT, Cox-2 KO and Ptger4 KO mice for 14 days. POBs were plated at a ratio of 1:10 to BMMs. (D) TRAP staining in POB and BMM co-cultures at day 8. Bars are means ± SEM for n = 3. ^aSignificant effect of treatment, p < 0.01. ^bSignificant effect of genotype compared to WT, p < 0.01, ^cp < 0.05.

To determine if EP4 on BMMs was necessary for the suppression of PTH effects, we co-cultured Cox-2 KO POBs with BMMs from WT, Cox-2 KO and Ptger4 KO mice (Fig. 7C). As expected, PTH stimulated Osteocalcin expression in POBs cultured without BMMs and in POBs co-cultured with Cox-2 KO BMMs but not with WT BMMs. There was no inhibition of PTH-stimulated Osteocalcin expression in POBs co-cultured with Ptger4 KO BMMs. To rule out the possibility that the effect of Ptger4 deletion was due to preventing formation of OC precursors, we compared the co-cultures for TRAP staining. There was no increase in TRAP staining with PTH in cultures without BMMs. PTH increased TRAP similarly in all the other co-cultures (Fig. 7D). Hence, Ptger4 in BMMs was required for the inhibitory effects of PGs on PTH-stimulated OB differentiation.

The inhibitory effect was transferred by conditioned media (CM) from RANKL-stimulated BMMs

To determine if the inhibition was mediated by cell–cell contact or by secretion of a soluble factor, POBs were co-cultured with CM collected from WT and Cox-2 KO BMMs. Cox-2 KO POBs were used in all experiments, and Alp or Osteocalcin mRNA was measured after 14 days of culture. Because RANKL was added to most BMM cultures before obtaining the CM, all POB cultures were done in the presence of OPG to prevent OCL formation.

In the first experiment, CM was collected from BMMs expanded for 5 days with M-CSF and compared with CM from BMMs treated with both M-CSF and RANKL for 0–3 days or 3–5 days (Fig. 8A). CM from WT, but not Cox-2 KO, BMMs treated with both M-CSF and RANKL inhibited the PTH stimulation of Osteocalcin in POBs. CM from WT BMMs treated only with M-CSF did not significantly inhibit. Inhibition by CM from WT BMMs cultured for 0–3 days was similar to that from BMMs cultured for 3–5 days. The 3 day BMM culture, treated with both M-CSF and RANKL, was used in all further experiments. Some TRAP + multinucleated cells were present in both WT and KO BMM cultures treated for 3 days with M-CSF and RANKL (data not shown).

Fig. 8.

Effect of conditioned media (CM) from BMM cultures on PTH-stimulated POB differentiation. POBs in all experiments were from Cox-2 KO mice. CM were taken from BMMs treated with M-CSF (30 ng/ml) alone or with both M-CSF and RANKL (30 ng/ml). POBs ± CM were treated with vehicle (VEH), PTH (10 nM) or the combination for 14 days. All POB cultures were done in the presence of OPG (50 ng/ml) to block osteoclast formation. Gene expression was measured by qPCR. (A) Comparison of effects of CM, pooled from different periods of BMM culture, on PTH-stimulated Osteocalcin expression in POBs. BMMs were cultured with M-CSF only for 5 days or with both M-CSF and RANKL for 0–3 or 3–5 days. After this experiment, all further experiments used CM from BMMs cultured for 3 days with both M-CSF and RANKL. (B) Comparison of effects of WT and Cox-2 KO CM on vehicle- and PTH-stimulated Osteocalcin expression in POBs. (C) Effects of WT and Cox-2 KO CM on the ability of exogenous PGE₂ (0.1 µM) and PTH + PGE₂ to stimulate Osteocalcin expression in POBs. (D) Effects of CM from WT, Ptger2 KO, and Ptger4 KO BMMs on Alp mRNA expression in POBs. Bars are means ± SEM for n = 3. ^aSignificant effect of treatment (compared to vehicle), p < 0.01. ^bSignificant effect of genotype (compared to WT) of BMMs from which CM was obtained, p < 0.01. ^cSignificantly different from PGE₂ and PTH alone, p < 0.01. ^dSignificantly different from PGE₂ alone, p < 0.01.

Although CM from WT BMMs inhibited PTH-stimulated Osteocalcin expression, WT CM did not inhibit Osteocalcin in vehicle-treated cultures compared to cultures without CM (Fig. 8B). In addition, CM from Cox-2 KO BMMs had no effect on vehicle-treated POBs.

To look at the effects of CM on responses to exogenous PGE₂, we examined effects of WT and Cox-2 KO CM on PGE₂-and PTH + PGE₂-stimulated Osteocalcin expression (Fig. 8C). WT CM did not inhibit PGE₂ stimulated Osteocalcin expression but did inhibit the stimulation of expression by PTH and PTH + PGE₂. In the presence of Cox-2 KO CM, the combination of PTH and PGE₂ had additive effects on Osteocalcin mRNA, confirming that a factor (or factors) made by BMMs expressing COX-2, not only inhibited PTH-stimulated Osteocalcin but also caused the inhibitory interaction of PTH and PGE₂.

To confirm the role of EP4 in the inhibitory effect, we treated Cox-2 KO POBs with CM from WT, Ptger2 and Ptger4 KO BMMs (Fig. 8D). PTH inhibited Alp expression relative to vehicle in the presence of CM from WT BMMs or Ptger2 KO BMMs. In contrast, in the presence of CM from Ptger4 KO BMMs, PTH stimulated Alp expression. Hence, it seems likely that PGs produced by BMMs acted on BMMs via EP4 to produce one or more soluble factors that inhibited the osteogenic effects of PTH on POBs. This observation supports the likelihood that the PG involved in the inhibitory effects is PGE₂.

Discussion

This study confirms and expands upon our previous observation that COX-2 produced PGs inhibit PTH-stimulated OB differentiation in BMSCs [26]. When COX-2 expression or PG production was absent, PTH markedly stimulated OB differentiation in BMSCs. The window for the stimulatory effect was the first week of culture, and this observation, in conjunction with similar effects of PTH on both OB and adipocyte differentiation, suggests that PTH was acting on OB precursors or MSCs, consistent with reported effects of PTH on OB precursors or MSCs in vivo [2, 7].

Because PTH is stable in culture up to 72 h between medium changes [35], our culture conditions provided continuous exposure of cells to PTH, which in most in vitro studies has resulted in inhibition of OB differentiation. Because intermittent PTH is anabolic in vivo but continuous PTH is catabolic, it is often assumed that PTH must be applied intermittently in vitro in order to be osteogenic. This assumption was strengthened by positive effects on OB differentiation when cells had short, transient exposure to PTH [8, 10, 45]. However, the brief duration of PTH exposure is usually accomplished by removing PTH-containing media and replacing with fresh media. Since this procedure also removes PGs that accumulate in the media, it is possible that the osteogenic effects in such experiments were really due to the removal of PGs that inhibited osteogenic effects of PTH.

The inhibitory effects of PGs on OB formation did not occur in vehicle-treated BMSC cultures but only in PTH-treated BMSCs. In these cultures, OCLs were formed in response to PTH during the same “window” of time that PTH had its stimulatory effect. The inhibitory effects of PGs did not occur in POBs washed free of hematopoietic cells or in OPG-treated BMSCs. Co-cultures of POBs with BMMs or with CM from BMMs demonstrated that RANKL-treated BMMs were required to see the inhibitory effects of PGs. The need for RANKL in order to see the inhibitory effects and the reversal by OPG suggest that the BMMs involved were committed to the OC lineage. Finally, using these same co-cultures, we showed that PGs acted on BMMs to cause them to produce a soluble factor or factors that then acted on OBs to suppress PTH-stimulated OB differentiation.

We could find no precedent for a soluble factor produced in OC lineage cells in response to PGs that inhibited PTH-stimulated OB differentiation. A number of studies have shown that soluble factors produced by monocytes and non-resorbing OCs can regulate OB differentiation in a stimulatory, but not inhibitory, manner [46–51]. Osteal macrophages (osteomacs), resident macrophages in bone-lining tissues that interact with OBs, would seem to be ruled out as candidates for producing the inhibitory factor observed in our study because studies indicate that they do not become OCs (or at least do not have their regulatory functions if they commit to become OCs) [52–54]. Several studies have proposed OC-produced factors that, unlike our findings, are not specific for PTH-treated cultures but can inhibit OB differentiation in general. These factors include cardiotropin-1 [55], semaphorin 4D [56], and sclerostin [47]. We have done several microarray studies on the BMMs under our culture conditions and did not find differential expression of any of these factors by COX-2 expression/activity or PGE₂ addition (data not shown), but this does not rule out their regulation at the protein level.

The inhibition of PTH-stimulated differentiation mediated by endogenous PGs could be generated by addition of PGE₂, but not other agonists for other PG receptors, to cultures. Moreover, production of the inhibitory CM required expression on BMMs of EP4, one of two receptors for PGE₂ that activates cAMP signaling. Hence, it seems likely that the endogenous PG mediating the inhibitory action under our conditions is PGE₂. PGE₂ is expected to have its major actions via cAMP/PKA signaling pathways similar to those stimulated by PTH. Exogenous PGE₂ concentrations as low as 0.1 nM were sufficient to inhibit osteogenic effects of PTH, and levels ≥4 nM were seen in vehicle treated co-cultures of POBs and BMMs as long as one cell type expressed COX-2. PGE₂ itself stimulates OB differentiation in vitro, as shown in the current studies. For a number of agents, such as TGFβ, BMP2, strontium ranelate and fresh serum[14, 17–19], the induction of COX-2 expression and PGE₂ production enhances their stimulation of OB differentiation in vitro. In contrast to PTH, these agents all have major actions via signaling pathways other than cAMP/PKA. Hence, other agonists that act via cAMP signaling pathways might also be inhibited by PGE₂ in this culture model.

CM from COX-2 expressing BMMs did not block the stimulatory effects of endogenous PGs or exogenous PGE₂ unless the cultures were also treated with PTH. In the absence of BMMs, the combination of PTH with PGE₂ had additive effects on OB differentiation, as expected of two osteogenic agents. In contrast, in the presence of the as yet unidentified factor or factors secreted by BMMs, the stimulatory effect of the combination of PTH and PGE₂ was abrogated. Assuming that the stimulatory effects of PTH and PGE₂ on OBs are mediated via stimulation of cAMP, it is possible that the CM contains a factor that acts via Gα_i to inhibit production of PTH- and PGE₂-stimulated cAMP. PGE₂ in WT CM can act via EP3, which is coupled to Gα_i. However, it is unclear why this effect would only occur in the presence of PTH. The factor that blocks PTH-stimulated differentiation produced by BMMs is unlikely to be PGE₂ itself because the addition of PGE₂ to PTH, in the absence of BMMs or WT CM, resulted in additive stimulatory effects. Another possibility is that PTH induced activity of a phosphodiesterase that rapidly degraded PGE₂-stimulated cAMP, but it is unclear why this should happen only in the presence of WT BMMs or CM. Although the explanation awaits further studies, this observation might explain why it has also been difficult to demonstrate an anabolic effect of systemically applied PGE₂ in mice [57]. Because the inhibitory factor made by BMMs blocks the stimulatory effects of PGE₂ in the presence of PTH and because endogenous PTH is present continuously in vivo, PGE₂ given in vivo might act on BMMs to suppress not only PTH-stimulated OB differentiation but also its own ability to stimulate OB differentiation.

In our in vitro study, PGE₂ is stable in the media (personal observation), unlike the conditions expected in vivo. PGs in vivo are not stored but are synthesized, released as needed and rapidly metabolized in their passage through the lung [58]. COX-2 protein is estimated to have a half-life on the order of 2 h [59, 60], and the local level of PGs in vivo is highly dependent on new production of Cox-2, which is a rapidly inducible and transiently expressed gene [14]. However, even when PTH was given intermittently, where the interaction of PTH and PGE₂ is expected to be brief, we found that PTH in vivo was more anabolic in Cox-2 KO mice than in WT mice [25]. A more marked effect of the inhibitory interaction of PTH and PGs on OB differentiation is expected in the continuous PTH infusion protocol, because both PTH and PGs should be continuously elevated. In addition, there should be an abundance of OCs generated by continuous PTH in vivo to produce the inhibitory factor(s). It is possible, therefore, that the PTH induction of COX-2 could account for some of the bone loss seen with continuous PTH in vivo.

Our findings suggest a novel role for COX-2 produced PGE₂ in vitro to inhibit PTH-stimulated osteogenic/anabolic activity via actions through EP4 on early osteoclastic lineage cells. PGE₂ is likely to be generated by COX-2 induction in many types of culture, and these findings suggest that it may have important modulatory roles that are overlooked. A better understanding of how PGs modulate the actions of PTH may help us be more effective in targeting bone remodeling for the treatment of osteoporosis and lead to the future development of new anabolic agents or protocols to improve therapy for osteoporosis and other skeletal defects.