The production of prostaglandins by osteoblasts is an important mechanism for the regulation of bone turnover. Bone cells contain both inducible and constitutive prostaglandin G/H synthase (PGHS-2 and PGHS-1) and these are differentially regulated. Nonsteroidal anti-inflammatory drugs (NSAIDs), which selectively inhibit one of these enzymes, would be useful in assessing their relative roles in bone metabolism. By Northern analysis, only PGHS-2 is expressed by the immortalized rat osteoblastic cell line, Py1a, while only PGHS-1 is expressed by the rat osteosarcoma cell line, ROS 17/2.8. We tested the relative inhibitory potency (IC50) of seven different NSAIDs on these two cell lines. A recently described selective inhibitor of PGHS-2, NS-398, was approximately 30 times more potent in inhibiting PGHS-2 than PGHS-1, and diclofenac was approximately 10 times more potent. Both had IC50's of approximately 3 nM for PGHS-2 in Py1a cells. Indomethacin, flurbiprofen, naproxen, and piroxicam were relatively nonselective with IC50's ranging from 30 nM to 1 μM, while 6-methoxy-2 naphthyl acetic acid, the active metabolite of nabumetone, was inhibitory only at concentrations greater than 1 μM. These results indicate that the presently available NSAIDs are unlikely to distinguish completely between effects mediated by PGHS-2 or PGHS-1. However, the cell systems employed could provide a model for the analysis of new compounds with greater selective activity.
Prostaglandins are potent regulators of skeletal metabolism which can be produced by cells of the osteoblastic lineage.1 Regulation of prostaglandin production by mechanical forces, hormones, cytokines, and growth factors appears to be due in large part to effects on the transcriptional regulation of the inducible prostaglandin G/H synthase (PGHS-2) or cyclo-oxygenase-2.2,3 However, bone also contains the constitutive enzyme, PGHS-1 or cyclo-oxygenase-1. Selective inhibitors of PGHS-1 or PGHS-2 would facilitate the analysis of the relative roles of these two enzymes in physiologic and pathologic responses of the skeleton. Marked differences in the effects of nonsteroidal anti-inflammatory drugs (NSAIDs) on isolated preparations of the two enzymes have been reported,4–6 but in vivo studies indicate that the majority of NSAIDs inhibit both PGHS-1 and PGHS-2.7 Recently, compounds that may be selective inhibitors of PGHS-2 have been reported.8–11
To examine the regulation of the inducible and constitutive enzymes in osteoblasts, we have compared an immortalized rat osteoblastic cell line, Py1a,12 which expresses only PGHS-2 on Northern analysis, with a rat osteosarcoma cell line, ROS 17/2.8, which expresses only PGHS-1. ROS 17/2.8 has osteoblastic features and is constitutively active in prostaglandin synthesis.13 The availability of these cells has given us the opportunity to estimate the relative inhibitory potency (IC50) of different NSAIDs on prostaglandins produced only by the inducible or constitutive enzyme. The majority of NSAIDs tested in the present study showed similar activities against both enzymes. However, the compound NS-398, which was previously shown to be PGHS-2 selective,8–11 showed greater potency in Py1a than in ROS 17/2.8 cells. A smaller difference was also found with diclofenac, which is ordinarily classified as a nonselective NSAID. Changes in endogenous prostaglandin production had little functional effect on ROS 17/2.8 cells but did affect collagen and noncollagen protein synthesis in Py1a cells in which exogenous prostaglandins have been shown to inhibit collagen synthesis.12,14
MATERIALS AND METHODS
Murine PGHS-1 cDNA, which cross-reacts with rat PGHS-1 mRNA, was a gift of Drs. David DeWitt and William Smith (Michigan State University, East Lansing, MI, U.S.A.). Murine PGHS-2 cDNA, which also cross-reacts with rat mRNA, was obtained from Oxford Biomedical Research Inc. (Oxford, MI, U.S.A.). Murine glyceraldehyde-3-phosphate dihydrogenase (GAPDH) cDNA was amplified by PCR using an amplimer set from Clontech (Palo Alto, CA, U.S.A.). The prostaglandin E2 (PGE2) antibody was purchased from Dr. Lawrence Levine, Waltham, MA, U.S.A.). Enzyme-linked immunoassay (ELISA) kits for PGE2, PGF2α, and 6-keto-PGF1α were obtained from Elisa Technologies (Lexington, KY, U.S.A.). NS-398 was obtained from Cayman Chemical (Ann Arbor, MI, U.S.A.). Transforming growth factor β (TGF-β) was the gift of Genentech, Inc. (South San Francisco, CA, U.S.A.). Other prostanoids and chemicals were obtained from Sigma Chemical Co. (St. Louis, MO, U.S.A.). The culture media were obtained from GIBCO BRL (Gaithersburg, MD, U.S.A.).
ROS 17/2.8 and Py1a cells were grown to confluence in F-12 containing 5% fetal calf serum (FCS) and MC3T3-E1 cells in Dulbecco's modified Eagle's medium (DMEM) with 10% FCS. Cells were then serum deprived for 24 h. The cells were treated for varying periods with TGF-β (10 ng/ml), arachidonic acid (AA, 10−5 M), or FCS. Combined treatment with AA and FCS was used to produce maximal induction of PGHS-2 and maximal prostaglandin production in Py1a cells. Although ROS 17/2.8 cells can produce large amounts of PGE2 simply with the addition of AA, these cells were treated in the same way as Py1a cells to minimize any differences in culture conditions.
The inhibitory effects of NSAIDs were generally tested after 24 h of treatment with a stimulator and varying concentrations of the drug. The concentration that produced 50% inhibition of prostaglandin production (IC50) was estimated graphically from plots of PGE2 levels in treated cultures expressed as percent of control untreated cultures. Each NSAID was tested in at least three experiments.
For estimation of IC50, the PGE2 concentration of the medium was measured by radioimmunoassay, as described previously.2 For analysis of the relative proportions of PGE2, PGF2α, and 6-keto-PGF1α, the metabolite of prostacyclin, produced by Py1a cells, ELISA kits were used. Values are the means of duplicate or triplicate analyses.
RNA extraction and Northern blot analysis
Total RNA was extracted by the method of Chomczynski and Sacchi15 in 4 M guanidine isothiocyanate followed by phenol/chloroform-isoamyl alcohol (24:1). RNA was precipitated with isopropanol and washed with 80% ethanol.
After quantitation at 260 nm, 20–25 μg of total RNA was fractionated on a 1% agarose/2.2 M formaldehyde gel and transferred to nylon membrane (Genescreen; New England Nuclear, Boston, MA, U.S.A.) by positive pressure blotting (Posiblotter; Stratagene Inc., La Jolla, CA, U.S.A.). After prehybridization in a 50% formamide solution at 42°C, filters were hybridized overnight in a similar solution in rotating cylinders (Hybridizer; Techne, Inc., Princeton, NJ, U.S.A.) at 42°C.
Collagen and noncollagen protein synthesis
Cells treated with or without FCS and AA were cultured for 24 h and pulsed with 5 mCi/ml of [3H]proline (15–40 Ci/mm, Amersham, Arlington Heights, IL, U.S.A.) for the final 4 h of culture. The medium was removed and the cells scraped into extraction buffer (1 M NaCl, 2.25 mM EDTA, 1 mM N-ethyl-maleimide, and 0.2 mM phenylmethylsulfonylfluoride). The medium and cells were pooled and sonicated and the protein precipitated with 15% trichloroacetic acid (TCA). After repeated washing, the pellets were dissolved in 0.5 M NaOH and an aliquot was digested with purified bacterial collagenase to determine collagen-digestible protein (CDP) and noncollagen protein (NCP) labeling. The percentage of total protein synthesis represented by collagen (percentage collagen synthesis, PCS) was calculated after correcting for the relative abundance of proline in collagen compared with noncollagen protein. DNA content was determined by fluorimetry using diaminophenylindole.
Five μCi/ml of [3H]thymidine (New England Nuclear Corp.) was added to cultures for the last 2 h of the culture period. Cells were washed with cold phosphate-buffered saline and extracted with 1.0 ml of 10% trichloroacetic acid. Acid-extractable [3H]TdR was determined by measuring the radioactivity in the trichloroacetic acid washes. After the addition of 1 ml of 0.5 M NaOH to each well, cells were scraped from the dishes, put into glass tubes, and incubated overnight at 4°C. An aliquot of each cell digest was counted to quantitate thymidine incorporation (TdR) into DNA.
The data were subjected to analysis of variance, and the significance of differences was determined by post hoc testing using Bonferroni's method.
Expression of PGHS-1 and PGHS-2
The expression of mRNA for the inducible and constitutive enzymes in ROS 17/2.8 and Py1a cells is compared with the murine osteoblastic cell line, MC3T3-E1, in Fig. 1. Both enzymes were easily detectable in MC3T3-E1 cells, but only PGHS-2 showed a substantial stimulatory response to AA, TGF-β, or the combination. No PGHS-1 mRNA was detectable in Py1a cells, but these cells expressed PGHS-2 mRNA when stimulated by AA. In this experiment, TGF-β alone was not effective, but in other studies it was able to induce the enzyme.
In ROS 17/2.8 cells, PGHS-1 mRNA was present at high levels and no PGHS-2 mRNA was detected by Northern analysis. Freshly added FCS also induced PGHS-2 in MC3T3-E1 and Py1a cells, as previously demonstrated,3,16 but not in ROS 17/2.8 (data not shown).
To induce PGHS-2 expression and to supply exogenous substrate for both PGHS-1 and −2, cells were treated with FCS and AA. All three cell lines produced substantial amounts of PGE2, as measured by immunoassay. In previous studies, we showed that this was the major prostaglandin produced by ROS 17/2.8 and MC3T3-E1.13,17 However, the predominant product of Py1a cells after stimulation with FCS and AA was PGF2α. PGI2, as measured by its metabolite, 6-keto-PGF1α, and PGE2 were produced in somewhat smaller amounts (Table 1).
Table Table 1. Incorporation of [3H]Proline into Collagenase Digestible (CDP) and Noncollagen Protein (NCP), Percent Collagen Synthesis (PCS), and Prostaglandin Production in Py1a Cells Cultured in 5% Fetal Calf Serum (FCS) with Arachidonic Acid (AA, 10−5 M) with and Without Indomethacin (Indo)
Indomethacin produced a parallel inhibition of all three products and this was associated with changes in cellular function (see below). Similar results were obtained with Py1a cells treated with TGF-β and AA (data not shown).
Effects of NSAIDs
The estimated IC50's for the NSAIDs tested are listed in Table 2 and the effects shown graphically for NS-398 and indomethacin in Fig. 2. In Py1a cells, NS-398 was a potent inhibitor of PGHS-2 with an IC50 of approximately 3 nM. In contrast, the IC50 in ROS 17/2.8 cells expressing only PGHS-1 was about 80 nM. Of the other NSAIDs tested, only diclofenac showed any appreciable difference in potency in the two cell types, with an IC50 for PGHS-2 in Py1a of 3 nM and 30 nM for PGHS-1 in ROS 17/2.8. 6-methoxy-2-naphthyl acetic acid (6-MNA), the active metabolite of nabumetone, which was previously reported to have some selectivity for PGHS-2, was a weak inhibitor with an estimated IC50 of more than 1000 nM. Indomethacin showed similar values of IC50 of 30 nM for both PGHS-2 and PGHS-1.
Table Table 2. Approximate IC50s for NSAIDs in 24 h Cultures of Py1a or ROS 17/2.8 Cells Stimulated with 5% FCS and AA (10−5 M)
Effects of endogenous prostaglandins on cell function
Stimulation of prostaglandin synthesis in Py1a cells with AA and FCS resulted in a reduction in collagen synthesis as measured by the incorporation of [3H]proline. This is similar to the effects of exogenous prostaglandins on this cell line.12,14 Addition of indomethacin resulted in a relative increase in proline incorporation into collagen and an increase in percent collagen synthesis (Tables 1 and 3). In cultures treated with AA plus TGF-β, an increase in percent collagen synthesis was also observed with indomethacin (Table 3). The values for percent collagen synthesis were similar to those obtained in Py1a cells, cultured serum-free with bovine serum albumin (BSA), which do not produce measurable PGs.12,14 ROS 17/2.8 cells showed a lower level of percent collagen synthesis (0.9–1.3) with no significant effect of indomethacin. With respect to [3H]thymidine incorporation, indomethacin appeared to reduce the response in cultures treated with AA plus TGF-β, suggesting that some of the mitogenic effect of TGF-β might be mediated by endogenous prostaglandins. In contrast, there was no inhibition of TdR incorporation by indomethacin in cultures treated with FCS.
Table Table 3. Incorporation of [3H]Thymidine into DNA or [3H]Proline into CDP or NCP in Py1a Cells Cultured with AA (10−5 M) and TGF-β (10 ng/ml) or 5% FCS with or Without Indomethacin (Indo, 10−6M)
In this study, we have taken advantage of the observation that two rat osteoblastic cell lines, Py1a and ROS 17/2.8, differ in their expression of the critical enzymes for prostaglandin production. In Py1a, an osteoblastic cell line derived from rat calvaria and immortalized by transfection with polyoma virus T antigen,12 only the inducible PGHS-2 mRNA is detectable. In ROS 17/2.8, a rat osteosarcoma-derived cell line with osteoblastic phenotype,13 we are able to detect only the mRNA for the constitutive enzyme PGHS-1. A murine osteoblastic cell line, which we have studied previously, MC3T3-E1, expresses mRNA and proteins for both enzymes.3 Both enzymes are present in organ cultures of bone,2 but they have not yet been localized by immunocytochemistry or in situ hybridization.
In the present studies, we tested seven NSAIDs for their inhibitory effects on prostaglandin production stimulated by FCS and AA in the two cell types bearing different PGHS isoenzymes. The only differential effects we observed were with NS-398, a compound that has been shown selectively to inhibit PGHS-2 in other cell systems, and with diclofenac, which was not considered previously to be a selective inhibitor of either enzyme. It is of interest that diclofenac has been shown to decrease bone resorption in postmenopausal women.18 These results differ from those obtained by looking at instantaneous inhibition of cyclo-oxygenase activity with isolated enzymes.4 In those studies, indomethacin and piroxicam preferentially inhibited PGHS-1, while flurbiprofen showed no differential effect and 6-MNA preferentially inhibited PGHS-2. In our studies, 6-MNA, which is the active metabolite of nabumetone, was relatively ineffective in inhibiting prostaglandin synthesis in either cell line. There are a number of possible explanations for the differences that we observed in these prolonged cell culture experiments and those observed with isolated enzymes. Selective inhibition of PGHS-2 appears to be related to a time-dependent increase in the inactivation of these enzymes.19 Lack of uptake by the cell could explain the lack of efficacy of 6-MNA and differences in uptake might also affect the differences observed with other NSAIDs. There are also possible interactions involving autoamplification of the enzymes.16 While these are not relevant in the present models in which only one form is expressed, they could be explored in bone cells that express both enzymes.
Currently, selective inhibitors of PGHS-2 are being explored for their possible clinical use. If compounds that can safely produce long-term inhibition in vivo can be obtained, this will make it possible to explore the relative importance of the constitutive and inducible enzyme in the regulation of bone turnover. Based on our studies of the response of the two enzymes to cytokines, hormones, growth factors, and mechanical forces, the major regulator does appear to be PGHS-2. However, the constitutive enzyme, PGHS-1, may be important in initial responses to perturbations that result in the release of AA. In any event, the availability of two osteoblastic cell lines from the same species that have differential expression of the two enzymes will be useful in assessing new inhibitors for their potential effects on bone.
These studies were supported by grants AR-18063, AR-38933, and DK-48361 from the National Institutes of Health. We thank Ms. Lisa Godin for her careful preparation of the manuscript.