Expression of an estrogen agonist in differentiating cultures

Thomas L. McCarthy*†, Mary E. Clough‡, Caren M. Gundberg‡, and Michael Centrella*†

Departments of *Surgery (Plastic Surgery) and ‡Orthopedics and Rehabilitation, Yale University School of Medicine, New Haven, CT 06520

Edited by Jan-Åke Gustafsson, Karolinska Institutet, Stockholm, Sweden, and approved March 14, 2008 (received for review January 4, 2008) respond in direct and indirect ways to estrogens, and interactions with other factors. In this regard, age-dependent changes in hormone levels and bone health can be osteoblasts from fetal rat bone express little functional sex limited by focused hormone replacement therapy. In this study, we steroid receptor but acquire the ability to drive complex direct report the release and isolation of an agonist and indirect responses after transgenic receptor expression (3, from osteoblast cultures. This entity reprises many aspects of 12, 20, 22, 23). Still, some studies suggest that endogenous estradiol activity in isolated osteoblasts, but differs from authentic estrogen receptor ␣ (ER␣) levels increase during late-stage estradiol by several biochemical and physical criteria. At levels that osteoblast differentiation (24, 25). To address this further, we occur in conditioned medium from differentiating osteoblast cul- investigated whether osteoblasts acquire a functional sex steroid tures, the agonist directly drives expression through estro- response during differentiation in vitro. In this study we report gen-sensitive response elements, activates the obligate osteoblast an increase in ER␣ dependent activity, through expression of an Runx2, and potently enhances Smad-depen- endogenous agonist, which occurs during this process. dent in response to TGF-␤, but exhibits relatively lesser suppressive effects on gene expression through C/EBP and Results AP-1-binding transcription factors. Estrogen receptor ag- Endogenous ER␣-Dependent Gene Expression in Differentiating Os- onist activity is resistant to heating at 100°C and separable from the teoblast Cultures. In an effort to determine whether osteoblasts bulk of the remaining alcohol- and hexane-soluble molecules by express functional ER␣ during early stages of in vitro differen- C18 chromatography. MS and molecular fragmentation analyses tiation, we found no significant endogenous gene expression predict a Mr of 415.2 to 437.2. Therefore, in addition to earlier through canonical estrogen response elements (ERE) without studies showing that osteoblasts readily respond to and metabo- transgenic ER␣ expression in young (7-day-old) or differentiat- lize various sex steroid-like substrates, we find that they also ing (21-day-old) cultures. Surprisingly, reporter gene activity generate a potent estrogen receptor agonist during differentiation through ER␣ increased significantly without added ligand in the in vitro. Changes in the availability of a molecule like this within 21-day-old cultures. No appreciable endogenous ER␣- bone may relate to differences in skeletal integrity with aging or dependent activity occurred in 7-day-old cultures or in 21-day- metabolic disease. old cultures that express (AR) and reporter driven by androgen response elements (ARE) (Fig. 1A Upper). intracrine ͉ selective estrogen receptor modulator ͉ steroid At all times, however, the cells were fully competent to respond to added estradiol (17␤E) or dihydrotestosterone (DHT) (Fig. ex steroids cause direct genomic transcriptional effects 1A Lower, and other studies below). These findings suggested Sthrough specific receptors that contain ligand binding, that later stage osteoblast cultures acquired the ability to elab- dimerization, DNA binding, and gene transactivation domains. orate an endogenous ER␣ agonist-like activity. This activity was They also cause indirect stimulatory or inhibitory effects through fully dependent on ER␣ expression (Fig. 1B) and was maximal complexes composed of activated and other between 21 and 35 days of culturing (Fig. 1C Left) when no transcription factors or through variations in the activation significant AR-dependent activity was observed (Fig. 1C Right). potential of other transcriptional components (1–6) The skele- Like the effect of 17␤E, ER␣ agonist-like activity was potently ton is a well recognized target for sex steroids. Bone fragility is suppressed by the ER antagonist ICI 182780 (ICI) (Fig. 1D). notable when sex steroid levels fall in women after menopause, ER␣ activation may also occur indirectly by cascades in elderly males, or after sex organ ablation (7–11). In such cases, induced by intrinsic osteoblast growth factors like insulin-like bone loss follows a release from native constraints on bone growth factor (IGF)-I (26, 27). However, endogenous ER␣ resorption that largely result from changes in growth regulators activation was not suppressed by staurosporine, PD-98059, or expressed by osteoblasts and from opposing effects on osteoblast SB-203580, which inhibit multiple including protein and osteoclast activation and apoptosis, leading to an overall kinase C, p38 MAP kinase, or MEK [supporting information increase in bone remodeling (12–16). Importantly, imbalances in (SI) Fig. S1]. bone remodeling are restored by sex hormone replacement therapy. Regardless of a clear benefit for bone, use of native sex Release of a Soluble ER␣ Agonist. Based on these observations, we steroids risks inappropriate gene activation in other tissues. The surmised that osteoblasts in later stage culture could express and possibility of breast, cardiovascular, and prostate disease that possibly release an endogenous ER agonist. Correspondingly, can occur in this context (17–19) has driven the search for modulators with tissue- or function-restricted effects. However, the value of agents like these may be compli- Author contributions: T.L.M., C.M.G., and M.C. designed research; T.L.M., M.E.C., and M.C. performed research; T.L.M., C.M.G., and M.C. analyzed data; and T.L.M. and M.C. wrote the cated by steroid and precursor metabolizing expressed paper. by osteoblasts (3, 20, 21). The authors declare no conflict of interest. Many cells used to evaluate sex steroid receptor modulators †To whom correspondence should be addressed at MS 208041, 333 Cedar Street, Yale express little or no endogenous sex steroid receptors. This offers University School of Medicine, New Haven, CT 06520. E-mail: [email protected] a select advantage to introduce native or mutated receptors or [email protected]. transgenically and thereby discriminate effects through individ- This article contains supporting information online at www.pnas.org/cgi/content/full/ ual estrogen or androgen receptors, through direct events that 0800085105/DCSupplemental. require sex steroid receptor binding to DNA, or through indirect © 2008 by The National Academy of Sciences of the USA

7022–7027 ͉ PNAS ͉ May 13, 2008 ͉ vol. 105 ͉ no. 19 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800085105 Downloaded by guest on September 26, 2021 A B ERα ERE C ERα ERE AR ARE untreated untreated untreated untreated 100 100 100

50 100 50 50 er activity

21 d 7 14 21 28 35 42 49 7 14 21 28 35 17βE 17βE DHT days in culture 100 50

7 d ERα ERE Relative report 50 D 21 d 0 ERα 0 ERα 0 AR 100 0 8040 ERE ERE ARE ERα, ng 50 72121 days in culture 0 ICI 0 ICI untreated 17βE

Fig. 1. Endogenous ER␣ dependent gene expression in differentiating osteoblast cultures. Osteoblasts were grown for the numbers of days (d) indicated and transfected with luciferase reporter plasmid driven by ERE or ARE, and with empty vectors (0) or expression plasmids encoding ER␣ or AR in A, C, and D and with reporter plasmid driven by ERE and various amounts of empty vector or expression plasmid encoding ER␣ in B. Transfected cultures were untreated or treated for 24 h with 1 nM 17␤E or DHT, and diluent or 100 nM of the ER␣ antagonist ICI 182780 (ICI) as shown. Cell extracts were assayed for relative reporter activity and corrected for protein content.

21-day-old cultures were washed exhaustively to deplete medium progesterone, or glucocorticoid sensitive gene promoter activity and serum components and refed serum-free, phenol red-free in cells that were clearly responsive to added steroid (Fig. 2D). medium for 24 h. Conditioned medium (CM) recovered from Other studies suggest that ER agonist-like activity might result refed cultures was then tested on 7-day-old cells, where it from cell culture components such as bisphenol plastic deriva- potently drove ER␣-dependent gene expression (Fig. 2A). Elab- tives (28) or perhaps phenol red added to many culture media oration of ER␣ agonist activity in 24 h CM increased throughout (29). However, cells grown on glass rather than plastic also at least 28 days of culture, and its effectiveness was also blocked released potent ER␣ agonist activity (Fig. 2E), and in agreement by ICI 182780 (Fig. 2B). Using 21-day-old cultures as a model of with other studies (30), addition of phenol red had no effect ER␣ agonist expression, we found virtually no activity in CM without or with 17␤E in 7-day-old cultures, or on endogenous collected during the first hour after refeeding, whereas it was ER␣ agonist activity or 17␤E in 21-day-old cultures (Fig. 2F). released continuously during 6 to 96 h of incubation in serum- CM also enhanced gene expression in 7-day-old cultures through free medium. In contrast, although alcohol soluble extract from ER␤ (31), which may function independently of ER␣ in bone the cell layer of 21-day-old cultures contained significant agonist remodeling in females (32). Whereas basal activity was Ϸ10-fold activity immediately after exhaustive washing, this remained higher with ER␤ relative to ER␣, the relative stimulatory effect relatively constant throughout 96 h (Fig. 2C). Therefore, later of CM or 17␤E was significantly less, and no activity through stage cells accrued a pool of ER␣ agonist that appeared to be ER␤ was evident in 21-day-old cultures (Fig. S2). Also, CM released and replenished during this period. Although CM activated ER␣ in human SaOs-2 osteoblast-like cells, and CMs caused potent ER␣ agonist activity, it failed to induce androgen, from confluent 7-day-old SaOs-2 and MG63 human osteoblast

ERα ERE A ERα ERE B ERα ERE C total ERα ERE, 7 d cultures 80 F 80 160 CM 40 80 40 extract 40 80 0 10 15 21 28 0 10 15 21 28 0 20 60 100 days in culture before 24 hour CM collection 24 hour CM, % ERα ERE, 21 d cultures from 21 day cultures

control + ICI 182780 MEDICAL SCIENCES 0 24 48 72 96 80 hours of CM production D ERα ERE AR ARE PRB PRE MMTV-Luc by 21 d cultures 40 Relative reporter activity 80 0 0 E ERα ERE E β 40 E β 17 17 80 Phenol red Phenol red [-] [+] x 0 0 0 0 E

β 40 CM CM CM CM De DHT Prog 17 24 hour CM from 21 d cultures or 1 nM hormone 70 14 21 24 hour CM, days in glass culture

Fig. 2. Release of a soluble ER␣ agonist by differentiating osteoblasts. Osteoblasts were grown for 7 days (d) in A–E and F (Upper), or 21 d in F (Lower), and transfected with luciferase reporter plasmid driven by ERE, ARE, progesterone response elements (PRE), or glucocorticoid response elements (MMTV-Luc) and expression plasmid encoding ER␣, AR, or B (PRB; cultures express endogenous ). Transfected cultures were untreated or treated for 24 h with osteoblast-derived conditioned medium (CM) at various dilutions in A or undiluted in B–F) or with cell layer extract in C, with 1nMof17␤E, DHT, promogestin (Prog), dexamethasone (Dex), and/or diluent in D, and without or with 100 nM ICI 182780 in B, or 15 mg/liter phenol red in F, as indicated. Cell extracts were assayed for relative reporter activity and corrected for protein content.

McCarthy et al. PNAS ͉ May 13, 2008 ͉ vol. 105 ͉ no. 19 ͉ 7023 Downloaded by guest on September 26, 2021 B Runx2 Activity C Smad Activity A ERE 250 200

tivitca A 125 100 80 ey rr 40 IGF-I Promoter AP-1 Activity r evi r tpot ale 80 100

Re 0 50 100 0 0.5 1.0 CM, % 17βE, nM 40 50 [+] ERα α 0 0 0 0 E E E [-] ER E M MC M MC β β β β C C 71 71 71 71

control PGE2 control TGF-β

Fig. 3. Differential effects by osteoblast-derived ER␣ agonist on osteoblast activity. Osteoblasts were grown for 7 days and transfected with luciferase reporter plasmid driven by ERE without or with expression plasmid encoding ER␣ in A; reporter plasmid driven by 5 GAL4 DNA binding domain sensitive response elements and expression plasmids encoding transcription factor Runx2 fused to the GAL4 DNA-binding domain and ER␣ in B (Upper); reporter plasmid driven by 1.7 kb of the rat IGF1 gene promoter and expression plasmid encoding ER␣ in B (Lower); reporter plasmid driven by four Smad response elements and expression plasmid encoding ER␣ in C (Upper); reporter plasmid driven by 3 AP-1 binding protein response elements and expression plasmid encoding ER␣ in C (Lower). Cells were untreated (0) or treated for 24 h with CM at various dilutions (A) or undiluted (B and C)or17␤E at various dilutions (A)orat1nM(B and C), and diluent, 1 ␮M PGE2, or 120 pM TGF-␤1, as shown. Cell extracts were assayed for relative reporter activity and corrected for protein content.

like cultures were also active (Fig. S3), albeit to a lesser extent. expressed by osteoblasts are known to be driven directly through The human osteoblasts did not survive long-term culture, how- ERE, the oxytocin gene promoter contains a well recognized ever, presumably due to their derivation from and ERE (33). When the HPLC fractions were assayed for activity unrestricted growth potential, preventing analysis of potentially through the native oxytocin gene promoter, the earlier eluting higher ER␣ activator levels at later times in serum-free medium. fraction induced a 40- to 60-fold increase in reporter gene expression (Fig. 4E Bottom). However, this was only Ϸ2–10% of Differential Effects on Osteoblast Activity. With ER␣, CM drove the increase that occurred in response to 0.3 nM 17␤E (data not ERE-dependent gene expression with a potency equivalent to shown). By contrast, the secondary peak was inert on oxytocin Ϸ1nM17␤E (Fig. 3A). Like 17␤E, it increased the activation gene promoter activity and is not addressed further. By both potential of the obligate osteoblast transcription factor Runx2 in Fourier transform and micromass Q-TOF MS, samples from the control and prostaglandin E2 (PGE2)-treated osteoblasts (Fig. primary activity peak revealed three minor species at Mr 353.27 3B Upper). It also decreased the stimulatory effect of PGE2 on (14% of total), 381.30 (14% of total), and 415.21 Da (12% of IGF1 gene promoter activity (Fig. 3B Lower), although with total), and a major species at Mr 437.19 (60% of total), with an somewhat reduced potency. Furthermore, CM enhanced the error range of 5 ppm. Of these, the only species common to the stimulatory effect of TGF-␤ on Smad-dependent gene expres- many molecules in the two fractions from open end C18 chro- sion (Fig. 3C Upper) to a relatively greater extent than 1 nM matography is the Mr 437.19 species. Typically, the difference 17␤E. By contrast to 17␤E however, the ER␣ agonist, at levels between the Mr 415.21 and 437.19 species is considered to be due which occur in CM, failed to suppress AP-1-dependent gene to a sodium atom, and together these species make up Ͼ70% of expression in TGF-␤ activated cells (Fig. 3C Lower). the active material from C18 HPLC. MS2 fragmentation yielded a new, predominant species at Mr 119.08. The 119.08 species may Characterization Studies. By various criteria described above, the represent a side chain or an internal fragmentation product, but ER␣ agonist in CM shares several, but not all, qualitative or at present no further structural information of these compounds quantitative characteristics with 17␤E, predicting expression of is definitive. By these several criteria, therefore, the most a selective estrogen receptor modulator. Initial studies showed purified sample of ER␣ agonist in CM from differentiating that the activity in CM was resistant to 30-min incubation at 100 osteoblasts contained a low Mr lipophilic molecule that was °C, soluble in 80% ethanol (Fig. 4A), removed by charcoal functionally and physically dissimilar to 17␤E(Mr 272.39). stripping (Fig. 4B), and extracted with methanol and hexane (Fig. 4C), consistent with a lipophilic compound. Material that Discussion survived 100°C and alcohol extraction was applied to an open- In this study, we report that differentiating osteoblasts express an end Sep-Pak C18 resin cartridge and step-gradient eluted with endogenous ER␣ agonist that reproduces some, but not all, of methanol, by which ER␣ agonist activity was released at 75–80% the biochemical effects induced by 17␤E in these cells. Earlier methanol (Fig. 4D). By electrospray MS, the two fractions with evidence showed that osteoblasts may control some aspects of the majority of ER␣ agonist activity each contained multiple, ER␣ activation locally within bone. Some studies revealed partly overlapping, species of various relative Mr. The two active greater ER␣ expression in parallel with osteoblast differentia- samples were dried and refractionated by C18 HPLC with a tion (24, 25), although no evidence correlated this with an linear acetonitrile gradient in 0.1% trifluoroacetic, by which increase in ER␣-dependent function. Other studies showed that ER␣ agonist activity resolved into a primary peak at Ϸ50% osteoblasts can metabolize various substrates into potent ER␣ acetonitrile, well removed from the bulk of UV light (254 agonists through endogenous aromatase (21) or 3-ketoreductase nm)-absorbing materials. The amount of ER␣ agonist recovered (20) activities or inactivate some ER␣ agonists through endog- in these highly purified fractions was equivalent to 0.3 nM 17␤E enous 3-hydroxysteroid dehydrogenase expression (3). Local or about 50–70% of maximal activity when assayed through production of estradiol-like molecules therefore completes the ERE. A second minor activity peak eluted at Ϸ53% acetonitrile loop of an endogenous ligand, receptor, and response system by this assay (Fig. 4E, Top and Middle). Whereas few if any within a single cell type.

7024 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800085105 McCarthy et al. Downloaded by guest on September 26, 2021 ABERα ERE ERα ERE 24 hCMfrom 21 daycultures

porter 80 80 activity tive re 40 40 Rela

0 4o 37o 100o 0 4o 37o 100o 0 4o 37o 100o 0 CM 17βE CM 17βE CM CM CM control charcoal supernatant 80% ethanol 80% ethanol stripped after incubation solute precipitate and spin

ERα ERE % C E C18 HPLC ac 3 70 etonitr 2 80 50 tive UV lighttive UV 1 30 ile A at 254 nM A at activity

40 Rela

Relative reporter ERα ERE 0 CM 0 CM control methanol and 80 hexane extract 40

ERα ERE er

D %

C18 (open) me

orter α

p ER Oxytocin promoter

80 80 thano Activity tive re l activity 40 40 Relative report 80 Rela 40 50 60 70 80 90 100 volume (ml) control 0 wash 2

H 15 25 35 45 55 65 75 85

flow throughflow Fraction number

Fig. 4. Characterization of osteoblast-derived ER␣ agonist. Osteoblasts were grown for 7 days and transfected with luciferase reporter plasmid driven by ERE in A–D and E (Middle), or a 1.1-kb fragment of the oxytocin gene promoter in E (Bottom), and expression plasmid encoding ER␣. Cells were untreated (0) or treated with undiluted CM or reconstituted to Ϸ100% equivalent, assuming no loss by processing, heating, and/or ethanol extraction in A; before and after charcoal stripping in B; by sequential methanol and hexane extractions in C; by open-end C18 chromatography in D; or by HPLC C18 chromatography in E. Cell extracts were assayed for relative reporter activity and corrected for protein content.

An important role for 17␤E in various aspects of skeletal occur in response to ICI 182780; and involve MAP kinase (39), growth and integrity is well known, and replacement therapy none of which are consistent with our results. ER␣ activators can with native or synthetic ER modulators has historically been a also increase Runx2 mRNA expression in subconfluent mouse potent adjunct to maintain bone mass and prevent fractures in osteoblasts (41), but we found no increase in nuclear Runx2 in postmenopausal women. Increased incidence of metabolic dis- 17␤E-treated rat osteoblasts (22). Still, future studies may reveal ease with certain ER modulators in nonskeletal tissues instills other effects in response to CM. valid concerns, however, and may limit the use of some of these These and characteristics derived from MS indicate that the agents (34, 35). ER modulators have complex direct and indirect CM-derived ER␣ agonist is functionally and physically dissimilar effects on gene expression that can differ in part from 17␤E and to 17␤E. Furthermore, assuming that virtually all of the 17␤Ein often appear tissue restricted (6, 36). In addition, even within culture medium before serum starvation accumulates as a osteoblasts, 17␤E itself can have complex stimulatory and in- cell-associated component and is released during CM prepara- hibitory effects. Although it directly enhances gene expression tion, this can account for no more than 2–5% of the activity in through ER␣ binding to ERE, it can also indirectly enhance the CM. Unlike effects that occurred with ER␣, no endogenous AR

transcriptional potential of nuclear factor Runx2, enhance or activator was evident in differentiating osteoblasts during 5 MEDICAL SCIENCES suppress independent aspects of TGF-␤ activity, and suppress weeks of culture, and treatment with aromatase inhibitors had no gene expression through members of the C/EBP transcription immediate or cumulative effect on endogenous ER␣ agonist factor family (12–14, 22, 23). In the case of Runx2 and TGF-␤, activity or its release into CM in 3-week-old cultures (Fig. S4). this may help to enhance bone mass through effects on select These findings together suggest that the CM-derived agonist phases of differentiated osteoblast function. In the case of might not originate from endogenous androgen-like precursors C/EBP␦, this may account in large part for reciprocal, inverse like testosterone, androstenedione, or their metabolites that can changes in local and systemic expression of IGF-I that occur activate ARE-driven gene expression through ARE (T.L.M. and before and after menopause, and after hormone replacement M.C., unpublished results). However, even more prolonged therapy (37, 38). The ER␣ agonist in osteoblast CM tends to exposure to aromatase inhibitors may be necessary to block reprise more fully the stimulatory effects of 17␤EonERE, conversion of these or other facile substrates that escape ready Runx2, and TGF-␤ activity, whereas it appears to be a less potent detection due to rapid metabolism. Still, transgenic overexpres- suppressor of C/EBP and AP-1 binding protein-dependent gene sion of 17␤-hydroxysteroid dehydrogenase reduced the stimu- expression. Some effects of ER agonists might involve other latory effect of either the CM-derived agonist or 17␤Eby membrane components such as G protein-coupled receptor 80%–90% (Fig. S5), predicting a degree of structural similarity GPR30 (39). However, GPR30 activation can increase cAMP, between them. The possibility that local release of an ER␣ which directly activates the IGF1 gene promoter (12, 27, 40); agonist has a protective role in bone during some phases of bone

McCarthy et al. PNAS ͉ May 13, 2008 ͉ vol. 105 ͉ no. 19 ͉ 7025 Downloaded by guest on September 26, 2021 growth and remodeling, or that it may become a useful tool to was assessed in parallel in cells transfected with positive and negative reporter maintain bone in some metabolic states, is an intriguing exten- plasmids (3, 20, 22). sion of our current observations. CM Preparation. Cells were cultured for varying intervals in 10-cm-diameter Materials and Methods Petri dish cultures in 10 ml of growth medium supplemented with 10% FBS. Cells. Primary osteoblast-enriched cultures were isolated from parietal bones After various intervals, medium was aspirated, and cell layers were washed six of 22-day-old Sprague–Dawley rat fetuses (Charles River Breeding Laborato- times with phosphate-buffered isotonic saline and refed for 0–96 h with 5 ml ries) as approved by the Yale Institutional Animal Care and Use Committee. of phenol red-free, serum-free medium. CM was collected at various time Sutures were dissected and cells were released by five sequential collagenase points before assay or for further processing. Cell layers were rinsed six times digestions. Cells pooled from the last three digestions, which express features and extracted with 1 ml of 95% ethanol. of differentiating osteoblasts, were plated at 4,000/cm2 in Dulbecco’s modi- fied Eagle’s medium supplemented with 10% FBS and 100 ␮g/ml ascorbic acid Chromatography. CM was lyophilized to dryness, redissolved in water, heated (42, 43) and grown for various time intervals before transfection or treatment. to 100°C for 30 min in a boiling water bath, centrifuged at 1,000 ϫ g to remove insoluble material, and relyophilized. Residue was redissolved in 80% ethanol Plasmids. ER␣-, AR-, progesterone receptor-, and glucocorticoid receptor- and stored at Ϫ80°C overnight, and alcohol soluble material was collected by dependent gene expression were assessed with luciferase reporter plasmids centrifugation and redried in a vacuum centrifuge. The material was redis- containing specific hormone response elements, without or with coexpression solved in 80% ethanol, and an aliquot was diluted in water to Ͻ5% ethanol of homologous steroid receptors, as previously described (3, 20, 22). Runx and applied to a prewashed open-end Sep-Pak Vac 20cc cartridge packed with activity was assessed with luciferase reporter plasmid 5XGAL4 driven by five 5 g of C18 resin. Flow through was collected, and bound material was water GAL4 response elements in cells cotransfected with an expression plasmid washed and eluted with a step gradient of 50–100% methanol in 10% encoding a Runx2-GAL4 DNA binding domain fusion protein (3, 22). Smad- increments. Dried samples were redissolved and tested for ER␣ agonist activ- dependent gene expression was assessed with luciferase reporter plasmid ity. Active samples were dried and refractionated by C18 HPLC with a 60-ml, SBE4 driven by four Smad response elements (22), and IGF1 gene promoter 30–70% linear gradient of acetonitrile in 0.1% trifluoroacetic acid at 1.5 was assessed with luciferase reporter plasmid 1711b-Luc, which contains a ml/min. Samples were dried, redissolved, and tested for activity. highly sensitive C/EBP response element in exon 1 that is rapidly activated in a protein kinase A-dependent way by PGE2 (40), and is significantly repressed by activated ER␣ (12, 23). MS Analysis. Samples were supplied to the W. M. Keck Foundation Biotech- nology Resource Laboratory at Yale University, which analyzed them for Mr by Transfections. Promoter–reporter fusion plasmids, gene expression plasmids, Fourier transform and micromass Q-TOF spectrometry (3). or empty parental vectors were pretitrated for expression efficiency and transfected with reagent LT1 (Mirus). Cultures were exposed to an optimal Statistical Analysis. Differences were assessed by one-way ANOVA with Tukey amount of reporter plasmid (37.5 to 75 ng/cm2) or expression plasmid (10 to post hoc analysis in SigmaStat (Jandel Scientific) from six or more replicates 75 ng/cm2) in phenol red-free medium supplemented with 4% charcoal and two or more independent cell preparations. Significant difference was stripped serum. Transfected cells were cultured for another 24–48 h and then assumed by a P value Ͻ0.05. treated in phenol red-free, serum-free medium as indicated in the figures. The cells were rinsed and lysed, and supernatants were analyzed for reporter gene ACKNOWLEDGMENTS. We thank Dr. Richard B. Hochberg for informative activity and corrected for protein content. To account for competition among discussions (Yale University). These studies were supported by National Insti- plasmids for limiting transcriptional components, control cultures were trans- tute of Arthritis and Musculoskeletal and Skin Diseases Grants AR39201 and fected with a compensating amount of empty vector. Transfection efficiency AR38460.

1. Boonyaratanakornkit V, Edwards DP (2007) Receptor mechanisms mediating non- 17. Lemay A (2002) The relevance of the Women’s Health Initiative results on combined genomic actions of sex steroids. Semin Reprod Med 25:139–153. hormone replacement therapy in clinical practice. J Obstet Gynaecol Can 24:711–715. 2. Syed FA, et al. (2005) Skeletal effects of estrogen are mediated by opposing actions of 18. Brand TC, Canby-Hagino E, Thompson IM (2007) Testosterone replacement therapy classical and nonclassical estrogen receptor pathways. J Bone Miner Res 20:1992–2001. and : A word of caution. Curr Urol Rep 8:185–189. 3. Centrella M, McCarthy TL, Chang W, Labaree DC, Hochberg RB (2004) Estren (4-Estren- 19. Raynaud JP (2006) Prostate cancer risk in testosterone-treated men. J Steroid Biochem 3alpha,17beta-diol) is a prohormone that regulates both androgenic and estrogenic Mol Biol 102:261–266. transcriptional effects through the androgen receptor. Mol Endo 18:1120–1130. 20. McCarthy TL, Hochberg RB, Labaree DC, Centrella M (2007) 3-ketosteroid reductase 4. Manolagas SC, et al. (2004) Kinase-mediated transcription, activators of nongeno- activity and expression by fetal rat osteoblasts. J Biol Chem 282:34003–34012. tropic estrogen-like signaling (ANGELS), and osteoporosis: A different perspective on 21. Yanase T, et al. (2003) Aromatase in bone: Roles of Vitamin D3 and androgens. J Steroid the HRT dilemma. Kidney Int Suppl S41–49. Biochem Mol Biol 86:393–397. 5. Centrella M, Christakos S, McCarthy TL (2004) Skeletal hormones and the C/EBP and 22. McCarthy TL, Chang, W-Z, Liu Y, Centrella M (2003) Runx2 integrates estrogen activity Runx transcription factors: Interactions that integrate and redefine gene expression. in osteoblasts. J Biol Chem 278:43121–43129. Gene 342:13–24. 23. Chang W, Parra M, Centrella M, McCarthy TL (2005) Interactions between CCAAT 6. Barkhem T, Nilsson S, Gustafsson JA (2004) Molecular mechanisms, physiological enhancer binding protein delta and control insulin-like consequences and pharmacological implications of estrogen receptor action. Am J growth factor I (igf1) and estrogen receptor-dependent gene expression in osteo- Pharmacogenomics 4:19–28. blasts. Gene 345:225–235. 7. McClung MR (2005) The relationship between bone mineral density and fracture risk. 24. Wiren KM, Chapman Evans A, Zhang XW (2002) Osteoblast differentiation influences Curr Osteoporos Rep 3:57–63. androgen and estrogen receptor-alpha and -beta expression. J Endocrinol 175:683– 8. Sorensen MG, Henriksen K, Schaller S, Karsdal MA (2007) Biochemical markers in 694. preclinical models of osteoporosis. Biomarkers 12:266–286. 25. Wang Q, et al. (2006) Temporal expression of estrogen receptor alpha in rat bone 9. Rosen CJ (2003) The cellular and clinical parameters of anabolic therapy for osteopo- marrow mesenchymal stem cells. Biochem Biophys Res Commun 347:117–123. rosis. Crit Rev Eukaryot Gene Expr 13:25–38. 26. Levin ER (2005) Integration of the extranuclear and nuclear actions of estrogen. Mol 10. Windahl SH, Andersson G, Gustafsson JA (2002) Elucidation of estrogen receptor Endocrinol 19:1951–1959. function in bone with the use of mouse models. Trends Endocrinol Metab 13:195–200. 27. McCarthy TL, Centrella M (2001) Local IGF-I expression and bone formation. Growth 11. Windahl SH, et al. (2007) Identification of target cells for the genomic effects of estrogens in bone. Endocrinology 148:5688–5695. Horm IGF Res 11:213–219. 12. McCarthy TL, et al. (1997) 17beta-estradiol potently suppresses cAMP-induced insulin- 28. Welshons WV, Nagel SC, vom Saal FS (2006) Large effects from small exposures. III. like growth factor-I gene activation in primary rat osteoblast cultures. J Biol Chem Endocrine mechanisms mediating effects of bisphenol A at levels of human exposure. 272:18132–18139. Endocrinology 147:S56–S69. 13. Ray A, Prefontaine KE, Ray P (1994) Down-modulation of interleukin-6 gene expression 29. Soto AM, Justicia H, Wray JW, Sonnenschein C (1991) p-Nonyl-phenol: An estrogenic by 17 beta-estradiol in the absence of high affinity DNA binding by the estrogen xenobiotic released from ‘‘modified’’ polystyrene. Environ Health Perspect 92:167– receptor. J Biol Chem 269:12940–12946. 173. 14. Stein B, Yang MX (1995) Repression of the interleukin-6 promoter by estrogen receptor 30. Moreno-Cuevas JE, Sirbasku DA (2000) Estrogen mitogenic action. III. Is phenol red a is mediated by NF-kappa B and C/EBP beta. Mol Cell Biol 15:4971–4979. ‘‘red herring’’? In Vitro Cell Dev Biol Anim 36:447–464. 15. Suda T, Nakamura I, Jimi E, Takahashi N (1997) Regulation of osteoclast function. 31. Koehler KF, Helguero LA, Haldosen LA, Warner M, Gustafsson JA (2005) Reflections on J Bone Miner Res 12:869–879. the discovery and significance of . Endocr Rev 26:465–478. 16. Jilka RL, Weinstein RS, Parfitt AM, Manolagas SC (2007) Quantifying osteoblast and 32. Sims NA, et al. (2002) Deletion of estrogen receptors reveals a regulatory role for estrogen osteocyte apoptosis: Challenges and rewards. J Bone Miner Res 22:1492–1501. receptors-beta in bone remodeling in females but not in males. Bone 30:18–25.

7026 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800085105 McCarthy et al. Downloaded by guest on September 26, 2021 33. Richard S, Zingg HH (1990) The human oxytocin gene promoter is regulated by 39. Filardo EJ, Quinn JA, Frackelton AR, Jr., Bland KI (2002) Estrogen action via the G estrogens. J Biol Chem 265:6098–6103. protein-coupled receptor, GPR30: Stimulation of adenylyl cyclase and cAMP-mediated 34. Gambacciani M, Vacca F (2004) Postmenopausal osteoporosis and hormone replace- attenuation of the epidermal growth factor receptor-to-MAPK signaling axis. Mol ment therapy. Minerva Med 95:507–520. Endocrinol 16:70–84. 35. Armitage M, Nooney J, Evans S (2003) Recent concerns surrounding HRT. Clin Endo- 40. McCarthy TL, Ji C, Chen Y, Kim K, Centrella M (2000) Time- and dose-related interac- crinol 59:145–155. tions between glucocorticoid and cyclic adenosine 3Ј,5Ј-monophosphate on CCAAT/ 36. Cheskis BJ, Greger JG, Nagpal S, Freedman LP (2007) Signaling by estrogens. J Cell enhancer-binding protein-dependent insulin-like growth factor I expression by osteo- Physiol 213:610–617. blasts. Endocrinology 141:127–137. 37. Wilson ME (1998) Regulation of the growth hormone-insulin-like growth factor I axis 41. Taranta A, et al. (2002) The selective estrogen receptor modulator raloxifene regulates in developing and adult monkeys is affected by estradiol replacement and supplemen- osteoclast and osteoblast activity in vitro. Bone 30:368–376. tation with insulin-like growth factor I. J Clin Endocrinol Metab 83:2018–2028. 42. McCarthy TL, Centrella M, Canalis E (1988) Further biochemical and molecular charac- 38. Balogh A, Kauf E, Vollanth R, Graser G, Klinger G, Oettel M (2000) Effects of two oral terization of primary rat parietal bone cell cultures. J Bone Miner Res 3:401–408. contraceptives on plasma levels of insulin-like growth factor I (IGF-I) and growth 43. Carpenter TO, et al. (1998) Osteocalcin production in primary osteoblast cultures hormone (hGH). Contraception 62:259–269. derived from normal and Hyp mice. Endocrinology 139:35–43. MEDICAL SCIENCES

McCarthy et al. PNAS ͉ May 13, 2008 ͉ vol. 105 ͉ no. 19 ͉ 7027 Downloaded by guest on September 26, 2021