Journal of Cellular Biochemistry 89:633–646 (2003)

Phytoestrogen Genistein Acts as an Estrogen Agoniston Human Osteoblastic Cells Through EstrogenReceptors a and b

D.J. Rickard,1 D.G. Monroe,1 T.J. Ruesink,1 S. Khosla,2 B.L. Riggs,2 and T.C. Spelsberg1*1Department of Biochemistry and Molecular Biology, Mayo Clinic, Rochester, Minnesota 559052Endocrine Research Unit, Mayo Clinic, Rochester, Minnesota 55905

Abstract Genistein, a soybean isoflavone, has estrogen-like activity inmammals, including the prevention of boneloss. However, whether its mechanism of action on bone turnover is distinct from that of estrogen or raloxifene isunknown. Although genistein has been reported to bind both estrogen receptor (ER) isoforms (a and b), little is knownconcerning differential activation of gene expression via these ER isoforms. To examine this question, comparison of theresponses of normal fetal osteoblast (hFOB) cells stably expressing either ERa (hFOB/ERa9) or ERb (hFOB/ERb6), totreatment with genistein, 17b-estradiol (E2) or raloxifene were conducted. In hFOB/ERa9 cells, both genistein and E2increased the endogenous gene expression of the progesterone receptor (PR), the proteoglycan versican, and alkalinephosphatase (AP), but inhibited osteopontin (OP) gene expression and interleukin-6 (IL-6) protein levels. Raloxifene hadno effect on these bonemarkers. Genistein, but not raloxifene, alsomimicked E2 action in the hFOB/ERb6 cells increasingPR gene expression and inhibiting IL-6 production. To determine whether the gene regulatory actions of genistein inhuman osteoblast cells occur at the level of transcription, its action on the transcriptional activity of a PR-A promoter-reporter constructwas assessed. Both genistein and E2were found to stimulate the PR promoter in the hFOB cell linewhentransiently co-transfected with either ERa or ERb. Whereas hFOB cell proliferation was unaffected by E2, raloxifene orgenistein at low concentrations, higher concentrations of genistein, displayed significant inhibition. Together, thesefindings demonstrate that genistein behaves as a weak E2 agonist in osteoblasts and can utilize both ERa and ERb. J. Cell.Biochem. 89: 633–646, 2003. � 2003 Wiley-Liss, Inc.

Key words: phytoestrogens; genistein; estrogen receptor; osteoblasts; progesterone receptor; IL-6; versican; alkalinephosphatase; osteoporosis

Phytoestrogens are a structurally diversegroup of plant-derived compounds that includethe isoflavones, lignans, and coumestans. Dueto their estrogen-like activity in mammals,there is considerable interest in phytoestrogensas potential natural alternatives or supple-ments to traditional hormone replacementtherapies for postmenopausal osteoporosis, car-diovascular disease, as well as the treatment of

reproductive cancers [Glazier and Bowman,2001]. The lower incidence of osteoporosis inJapan and of hip fractures in Asian populationsin general, compared to Caucasians, may becorrelated with the higher consumption of foodsrich in phytoestrogens [Cooper and Campion,1992; Russell-Aulet et al., 1993]. However, acausal link between phytoestrogens and bonemass maintenance in humans has yet to beestablished. Nevertheless, short-term treat-ment of peri- and post-menopausal womenwithsoy protein diets enriched in isoflavones hasbeen demonstrated to reduce bone loss from thelumbar spine as well as provide a potentialcardioprotective effect through improvement ofthe serum cholesterol profile [Potter et al., 1998;Alekel et al., 2000]. Long-term (1–2 year)treatment with the synthetic isoflavone, ipri-flavone, not only potentiated the effect of lowdose estrogen in reducing bone loss in healthypostmenopausal women [Agnusdei et al., 1995],

� 2003 Wiley-Liss, Inc.

Grant sponsor: NIH; Grant number: AG04875 (to TCS, SK,and BLR); Grant sponsor: NIH; Grant number: DK07352(to DJR); Grant sponsor: The Mayo Foundation.

*Correspondence to: Dr. T.C. Spelsberg, Department ofBiochemistry and Molecular Biology, Mayo Clinic andFoundation, 1601 Guggenheim, 200 1st Street SW, Roche-ster, MN 55905. E-mail: spelsberg.thomas@mayo.edu

Received 5 February 2003; Accepted 7 March 2003

DOI 10.1002/jcb.10539

but also prevented the loss of cortical andcancellous bone when administered alone toosteopenic postmenopausal women [Passeriet al., 1992; Gennari et al., 1997]. Soybeanprotein and natural and synthetic isoflavoneshave also been demonstrated to prevent sex-steroid deficiency-induced bone loss and ele-vated blood cholesterol in ovariectomizedrodents [Arjmandi et al., 1998; Uesugi et al.,2001]. However, as with estrogen, these com-pounds cannot reverse established osteopenia[Arjmandi et al., 1998; Picherit et al., 2001].

The mechanism of action of isoflavones in theprevention of bone loss appears to differ fromthat of estrogen and selective estrogen receptormodulators (SERMs, i.e., raloxifene). Whereasboth estrogen and raloxifene prevent bone lossby reducing bone resorption [Wronski et al.,1988; Evans et al., 1996], the isoflavonesgenistein and ipriflavone, are antiresorptiveonly at chronic high doses and therefore havebeen reported to either inhibit [Gennari et al.,1997; Picherit et al., 2001] or not affect [Passeriet al., 1992; Agnusdei et al., 1995; Uesugiet al., 2001] the bone turnover rate. Indeed,high dose (>1 mM) ipriflavone and ipriflavonemetabolites have been reported to inhibitmulti-nucleated osteoclast formation and activityin vitro [Notoya et al., 1993; Giossi et al.,1996]. Interestingly, neither of these isofla-vones blocks the elevated bone resorption ratesinduced by ovariectomy ormenopause [Gennariet al., 1997; Arjmandi et al., 2000]. Thus, whileisoflavone treatment may prevent bone loss, inpart, because the enhanced bone formationexceeds resorption, the mechanism(s) under-lying the enhancedbone formationby isoflavonetreatment in vivo is largely unknown. In im-mature and mature osteoblast cell culturescertain isoflavones, including genistein, havebeen shown to modestly stimulate differen-tiated osteoblast characteristics, such as thesynthesis of total proteins [Yamaguchi andSugimoto, 2000], type I collagen, alkalinephosphatase (AP) activity [Benvenuti et al.,1991], and mineralized matrix deposition[Cheng et al., 1994].

Mechanistically, genistein is considered to beone of the most estrogenic of the phytoestro-gens, and is able to bind both a and b isoforms ofthe estrogen receptor (ER).Both osteoblasts andosteoclasts expressERs and respond to estrogentreatment in vivo and in vitro [for a review seeOursler, 1998]. Although the affinity of genis-

tein for ERa and ERb is lower than that of 17b-estradiol (E2), the relative binding affinity ofgenistein for ERb is greater than for ERa[Kuiper et al., 1998]. The two ER isoformsexhibit differential ligand-dependent regula-tion not only of transcription from syntheticinducible promoter elements [Paech et al., 1997;Saville et al., 2000] but also of endogenous genesand cellular functions in osteoblasts in vitro[Waters et al., 2001]. Thus, the repertoire oftarget genes affected, as well as the response ofany specific gene, will be dependent on theparticular ligand bound to the receptor iso-forms and the ratio of ERa to ERb. Thus, thedivergent effects of isoflavones on bone meta-bolism in vivo, including their unique regula-tion of osteoblast and osteoclast functions, couldbe caused by the relative concentrations of theERa and ERb isoforms in the osteoblasts andother skeletal cells. To examine this possibility,the effects of genistein on human osteoblast celllines, stably expressing either the ERa or ERbisoform, were compared to the effects of E2 andthe SERM raloxifene.

MATERIALS AND METHODS

Reagents

The phenol red-free Dulbecco’s MEM-Ham’sF12 (1:1) medium (DMEM-F12), DMEM med-ium, and all other reagents were obtained fromSigma Chemical Company (St. Louis, MO) un-less otherwise stated. Charcoal-stripped fetalbovine serum (CS-FBS) was purchased fromHyclone Laboratories (Logan, UT), hygromycinB from Boehringer Mannheim (Indianapolis,IN), zeocin from Invitrogen (Carlsbad, MA),geneticin disulphate (G418) and LipofectaminePLUS transfection reagent from Gibco BRL(Gaithersburg,MD). Human recombinant basicfibroblast growth factor (bFGF) was purchasedfrom Promega Corporation (Madison, WI), [3H]methyl-thymidine from DuPont-New EnglandNuclear (Boston, MA), and a-[32P]-dCTP fromAmersham Pharmacia (Piscataway, NJ).

ICI 182,780 was generously provided byZeneca Pharmaceuticals (Macclesfield, Che-shire, UK). Plasmid reporter constructs con-taining the human PRA promoter region wasgenerously provided by Prof. Pierre Chambon(University Louis Pasteur, INSERM, Stras-bourg, France) [Kastner et al., 1990]. Full-length human versican cDNA and crude rab-bit anti-human versican polyclonal antiserum

634 Rickard et al.

were a generous gift from Dr. Erkki Ruoslahti(TheBurnhamInstitute,LaJolla,CA) [LeBaronet al., 1992]. The human AP cDNA probe was agift from Dr. Gideon Rodan (Merck, Sharp andDohme, West Point, PA), and the osteonectincDNA probe was provided by Dr. George Long(University of Vermont, Burlington, VT).

Purification of Raloxifene

Raloxifene (LY 139,481) was purified fromEvista1 tablets (Eli Lilly, Indianapolis, IN).Each tablet containing 60 mg raloxifene HCl(MW510.05) was pulverized and then extractedin absolute ethanol for 1 h at 48C. Insolublematerial was removed by filtration and thesupernatant dried under vacuum. The driedresidue was dissolved in 10% (v/v) acetonitrilein water, 0.1% (v/v) trifluoroacetic acid (TFA),and fractionated on a C-18 HPLC column usinga 4% to 64% (v/v) acetonitrile/water gradientcontaining 0.1% (v/v) TFA with detection at220 nm. A single peak eluting around 40%acetonitrile was collected and pooled fractionsdesiccated, resuspended inmethanol/water andthen re-desiccated. The resultant pale yellowpowder was then analyzed by mass spectro-metry and produced a single spike with MW474.1, the expected molecular mass of raloxi-fene. Purified raloxifene was then dissolved inabsolute ethanol to yield a 1 mM stock solution.The biological activity of the purified raloxifenewas verified by testing its antagonism of E2-stimulated MCF-7 breast cancer cell prolifera-tion (Fig. 1) essentially as described [Wakelinget al., 1984].

Cell Culture

The hFOB/ERa9 and hFOB/ERb6 humanfetal osteoblast cell lines were developed fromthe parent temperature-sensitive SV40 (tsA58)large T antigen-immortalized hFOB1.19 cellline by stable transfection with ERa and ERbexpression plasmids, respectively, and havebeen described previously [Harris et al., 1995;Waters et al., 2001]. These cells express onlythe single ER isoform, as appropriate, and ex-hibit E2 responsiveness. Cells were cultured inphenol red-free DMEM-F12 medium supple-mented with 10% (v/v) CS-FBS with continualselection in alternating antibiotics [Waterset al., 2001]. The hFOB1.19 cells were culturedin DMEM-F12 supplemented with 10% (v/v)FBS and G418. All experiments were carriedout at the permissive temperature of 348C.

MCF-7 human breast carcinoma cells were cul-tured in phenol red-free DMEM-F12 mediumcontaining 10% (v/v) FBS and MG63 humanosteosarcoma cells in DMEM medium supple-mented with 10% (v/v) FBS, both at 378C.

Cell Proliferation

Cells were seeded into 12-well dishes (4 cm2/well) at a density of 2� 104 (hFOB) or 2� 105

(MCF-7) cells per well, and allowed to recoverfor 24h.Cellswere then pretreated for 48hwithICI 182,780 (10�8 M) in DMEM-F12þ 10% (v/v)CS-FBS (5% CS-FBS for MCF-7 cells) withoutantibiotics. The ICI 182,780 treatment serves todecrease any estrogen-independent signaling(basal signaling). After rinsing three times withserum-freemedium, cellswere treatedwith ICI,E2, raloxifene, or genistein in the presence andabsence of bFGF, in DMEM-F12þ 1% (v/v) CS-FBSand cultured either for 3days (MCF-7 cells)or 5 days (hFOB cells). Cells were labeled with[3H]-thymidine (1 mCi/well) for the final 20 h oftreatment. For the hFOB cells, all incubationswere at 348C. As an indicator of DNA synthesis,the radioactivity incorporated into trichloroa-cetic acid-precipitablematerial was determinedas described [Robinson et al., 1997].

Fig. 1. The HPLC purified raloxifene antagonism of the E2induction of MCF-7 cell proliferation. MCF-7 cells were pre-treated with ICI and treated with E2, raloxifene (Ral), or ICI inDMEM-F12þ1% (v/v) CS-FBS. DNA synthesis was measuredafter 3 days treatment by the incorporation of [3H]-thymidine.The data are presented as the mean� SEM of six determinationsfor each treatment. A representative experiment repeated fourtimes with similar results is shown. Significant differencesbetween the control (i.e., treatment with ICI alone) and the othertreatments are indicated by asterisks (*) while the significantdifferences between treatment with E2 alone and estrogenco-treatments with either raloxifene or ICI are denoted by000P< 0.001.

Genistein Effects on ERs in Human Osteoblasts 635

Northern Blotting

Total RNA was extracted from cells using aphenol-guanidine isothiocyanate solution (Tri-Reagent, Molecular Research Center, Cincin-nati, OH). RNA, 10 or 15 mg, was separated on1% (w/v) glyoxal agarose gels and transferredovernight to nylon membranes (Osmonics, Inc.,Intermountain Scientific, Bountiful, UT) bydownward capillary action in 10� SSC (1.5 MNaCl, 0.15 M sodium citrate, pH 7.0). Mem-branes were analyzed for steady state mRNAlevels by hybridization with cDNA probeslabeled by random priming with [a32P]-dCTPusing a commercially available kit (DuPont-New England Nuclear). Hybridizations wereperformed in50% (v/v) deionized formamide, 3�SSPE (0.45 M NaCl, 30 mM NaH2PO4, 3 mMEDTA, pH 7.4), 5� Denhardt’s solution (0.25%(w/v) SDS, 10 mg/ml polyadenylic acid), and0.2 mg/ml denatured salmon sperm DNA. Thefollowing cDNAprobes were used: progesteronereceptor (PR) [Rickard et al., 2002], full lengthhuman versican, bone/liver/kidney AP, osteo-pontin (OP), and osteonectin (ON). Hybridiza-tions and post-hybridization washes wereperformed at 428C, except for PR, which wasperformed at 378C. Loading and integrity ofthe RNA was assessed by a final hybridizationwith a GAPDH cDNA probe. Densitometry ofbands on autoradiograms was performed usinga DC120 zoom digital camera and 1D imageanalysis software (Eastman Kodak, Rochester,NY).

Reverse Transcriptase PCR

Aliquots of total RNA (�30 mg) were digestedwith RNase-free DNase1, and 4 mg of the digest-ed RNA reverse transcribed and amplified induplicate by PCR using previously describedmethods [Rickard et al., 1998]. The sensitivityand accuracy of PCR was improved by theinclusion of Taq Start antibody in the reactionmixture to provide a ‘hot start’ (Clontech, PaloAlto, CA). Amplification reactions were per-formed for the following sequences: GAPDH;PR hormone binding domain (50 primer: 50-AGAGTTGTGAGAGCACTGGAT, and 30 pri-mer: 50-GATTCTTTCATCCGCTGTTCAT); andthe V0, V1, and V2 splice variants of humanversican using published primer sequences[Dours-Zimmermann and Zimmerman, 1994].The reactions were terminated during thelinear amplification phase; after 24 cycles for

GAPDH, 38 cycles for PR, and 30 cycles forthe versican isoforms. Similar reaction profileswere used for each primer set: an initial de-naturation at 948C for 2 min was followed bycycles of denaturation at 948C for 30 s, anneal-ing at 558C for 2 min (508C for versican V0, and528C for versicans V1 and V2), and polymeriza-tion at 728C for 2 min. Amplification productswere visualized on 1.5% (w/v) agarose gelsstained with ethidium bromide.

Transient Transfection

The hFOB cells seeded into six well plateswere transiently transfected at �70% conflu-ence using the Lipofectamine PLUS reagent(Invitrogen). Each well received 0.5 mg of eitherERa or ERb expression vectors, 2 mg of thereporter construct PR-(464,1105)-CAT contain-ing the human PRA promoter [Kastner et al.,1990], and 0.5 mg CMV-Luciferase vector tocontrol for transfection efficiency. The ER ex-pression vectors encoded N-terminally FLAGtagged receptor and were constructed in thevector pcDNA4/TO (Invitrogen). After a 3 h in-cubation, the transfection medium was re-moved and cells treated overnight with ICI182,780 (10�8M) inDMEM-F12þ 10% (v/v) CS-FBS. The cells were then rinsed three timeswith serum-free medium and treated witheither E2 or genistein for 24 h in medium con-taining 0.1% (w/v) BSA. Each treatment wasadded to triplicate wells. The CAT and lucifer-ase activities were then assayed in 25 mg and10 mg aliquots of cell lysate, respectively, asdetailed previously [Harris et al., 1995].

Indirect Immunofluorescencefor Versican Protein

Versican protein was detected by indirectimmunofluoresence of cells stimulated for 48 h.The hFOB/ERa9 cells were seeded at 1.5� 104

cells/well ontoPermanoxplastic 8-well chamberslides (Nunc, Inc., Naperville, IL) previouslycoated with 5 mg/cm2 human fibronectin. Cellswere incubated for 48 h in mediumþ 10% (v/v)CS-FBS to recover and then stimulated for 48 hwith or without E2, genistein or ICI 182,780 inmediumþ 0.1%(w/v) BSA. MG63 cells seededinto chamber slides coated with poly-L-lysine(10 mg/cm2) were used as a positive control[LeBaron et al., 1992]. After fixation and per-meabilization with methanol for 15 min at�208C, cells were rehydrated with PBSþ 2%(v/v) FBS, blocked in PBSþ 10% (v/v) FBS, and

636 Rickard et al.

incubated with rabbit polyclonal anti-humanversican antiserum diluted 1:100 in PBSþ 3%(w/v) BSA for 1 h at room temperature. Purifiedrabbit IgG (5 mg/ml) was used as a negativecontrol. The cells were washed in PBSþ 0.5%(v/v) Triton X-100, incubated with goat anti-rabbit FITC conjugated secondary antibodydiluted 1:200 in PBSþ 3%(w/v) BSA, andwashed extensively with PBS before beingmounted with 90% (v/v) glycerol, 10% (v/v)PBS and 1% (w/v) DABCO. The staining wasvisualized with an Axioplan 2 fluorescencemicroscope (Carl Zeiss, Germany).

IL-6 Protein Assay

The hFOB/ER cells were plated into 12 wellplates at 4� 104 cells/cm2, cultured to con-fluency and treated for a total of 4 days witheither vehicle, ICI 182,780, E2, raloxifene orgenistein at the concentrations indicated in thefigure legends, in mediumþ 10% (v/v) CS-FBS.Themediumwas changed and fresh treatmentsadded after 2 days. The concentration of IL-6 inconditioned medium harvested on day 4 wasmeasured using anELISA specific for IL-6, witha sensitivity of 0.7 pg/ml (R&D Systems, Inc.,Minneapolis, MN). Each sample was assayed induplicate with inclusion of a standard curve ofrecombinant human IL-6 in each assay run.

Statistical Analyses

The results are presented as the mean�SEM. The non-paired Student’s t-test was usedto calculate statistical significance betweenthe control group and each treatment group. Avalue of P equal to or below 0.05 was consider-ed to be significant. Statistical significance isdenoted in the figures by asterisks: *P< 0.05,**P< 0.01, and ***P< 0.001.

RESULTS

Verification of the Bioactivityof Purified Raloxifene

Raloxifene, purified from Evista1 tablets,exhibited biological activity and behaved as anERantagonist inMCF-7breast cancer cells. Theproliferation of these cells was induced in adose-dependent manner by E2 and the responsewas inhibited by co-treatment with raloxifene(Fig. 1). A 100-fold molar excess of raloxifeneor ICI 182,780 reduced the E2-stimulated pro-liferation to basal levels. Raloxifene alone hadno effect on the proliferation of MCF-7 cells. In

the absence of ICI pretreatment, the prolifera-tive response to E2 was severely diminished orundetectable, despite the relatively abundantexpression of ER in this cell type.

Regulation of PR Gene Expressionand Promoter Activity

Expression of the PR gene is stimulated byE2 in osteoblasts via ERa [Harris et al., 1995;MacNamara and Loughrey, 1998] and to aminor extent by ERb but only after long-termtreatment [Rickard et al., 2002]. To determinewhether genistein can also induce PR geneexpression in a similar fashion by either ERaand ERb, the steady state mRNA levels for bothPR isoformswere first examined in hFOB/ERa9(Fig. 2A) and hFOB/ERb6 (Fig. 2B) cells byNorthern blot and RT-PCR analysis, respec-tively. Genistein (10�7 and 10�6 M) and E2

(10�10 to 10�8 M), but not raloxifene, markedlyinduce the concentrations of multiple PR tran-scripts in the hFOB/ERa9 cells after 24-htreatment. The major mRNA species of �11 kbwas increased approximately 10-fold comparedto levels in untreated cells. Genistein at 10�8 Mwas non-stimulatory (data not shown), but100� higher concentrations of genistein thanE2 concentration was shown to stimulate PRmRNA expression by a magnitude similar tothat produced by E2.

As previously reported, the expression of PRmRNA is stimulated in the hFOB/ERb6 cell byE2 to considerably less degree and only afterlengthy treatment periods compared to thehFOB/ERa9 cells [Rickard et al., 2002]. ThePR mRNA in the hFOB/ERb6 cells was detect-able only by RT-PCR following long-term treat-ment with the steroid for 4 to 14 days. Theinduction of PR mRNA expression occurredwith both genistein (10�6 M) and E2 (10�9 to10�8 M) in the hFOB/ERb6 cell line followingtreatment for 10 days (Fig. 2B). Genistein at10�6 M was also stimulatory, as was 10�7 Mgenistein (datanot shown).Raloxifeneat10�7Mfailed to induce PR gene expression in theERb cells even after 10 days of treatment. Thus,in the hFOB/ERb6 cells, the effects of genisteinon PR mRNA concentrations was weakly sti-mulatory, but only at approximately 100-foldhigher concentrations than E2. The above re-sults support the notion that genistein is aweaker stimulator of PR gene transcriptioncompared to E2 when acting through either ERisoform.Further, as reportedpreviouslywithE2

Genistein Effects on ERs in Human Osteoblasts 637

[Rickard et al., 2002], the ERa isoform bound bygenistein appears to be a more efficient stimu-lator of PR gene expression than ERb boundgenistein. However, due to the lack of specific,

high affinity antibodies to ERb, the relativeabundance of ER protein between the twohFOB/ER cell lines is unknown.

To better assess the actions of genisteinunderconditions of comparable ERa and ERb receptorlevels, and to determine if genistein regulatesPR expression at the level of transcription, asdoes E2, the effects of genistein and E2 on PRgene promoter activity were compared by tran-sient transfection assays. Parental hFOB cells,which lack endogenous ER expression, were co-transfected with a PRA promoter-CAT reportergene construct and an ER expression vector,encoding either ERa or ERb. The ERs wereFLAG-tagged at their N-termini, and Westernblotting using an anti-FLAG antibody revealedsimilar protein levels of each ER isoform incells 48 h after transfection (data not shown).Genistein and E2 both stimulated the activityof PRA through binding to either ER isoform(Fig. 2C).TheERa isoform,however,wasamoreefficient stimulator of PRA transcription com-pared with ERb in the presence of both ligands.The E2 and genistein increased the PRA pro-moter activity by a maximum of 8- and 4-fold,respectively, when actingwith the ERa isoform,but only by a maximum of 3- and 2.5-fold,respectively, with the ERb isoform. In sum-mary, using conditions of equivalent ER con-centrations, the ERa isoform displays a more

Fig. 2. Genistein and E2, but not raloxifene, induce PR geneexpression and promoter activity via ERa and ERb. Panel A:Northern analysis of hFOB/ERa9 cells for both A and B isoformsof PR. Cells were pretreated with ICI 182,780 (ICI) prior totreatment for 24 h with genistein (Gen), E2, raloxifene (Ral) or ICIat the indicated concentrations in DMEM-F12þ0.1% (w/v) BSAmedium. Densitometry of the major PR transcript of �11 kb(normalized to GAPDH) is shown below, expressed relativeto the level in cells treated with ICI alone. Values are themean� SEM of three separate experiments. Panel B: RT-PCR forPR (both A and B isoforms) and GAPDH in hFOB/ERb6 cellscontinuously treated for 10 days inmediumcontaining 10% (v/v)CS-FBS. T47D human breast cancer cells were used as a positivecontrol for PR expression. Panel C: Stimulation of PRA promoteractivity by genistein and E2 in the presence of ERa (top) and ERb(bottom). The hFOB cells were transiently co-transfected withthe appropriate ER expression vector encoding FLAG-taggedreceptor together with the PRA-CAT reporter gene construct PR-(464,1105)-CAT and CMV-luciferase. After pretreating with ICI,transfected cultures were stimulated with genistein or E2 for 24 hin DMEM-F12 mediumþ0.1% (w/v) BSA. Results are presentedas fold induction of CAT activity compared to untreated controlcultures, after normalizing with luciferase for transfection effi-ciency, and are the mean� SEM of at least four separate ex-periments. A statistically significant difference betweengenisteinand E2 treatment at the same concentration is indicated by anasterisk (*).

638 Rickard et al.

potent action on the PR promoter activity thandoes the ERb isoform. Further, E2 was over 10-fold more potent than genistein at stimulatingPRA promoter activity irrespective of the ERisoform involved. The EC50 with ERa was,E2¼ 4� 10�10 M, genistein¼ 5.7� 10�9 M;while the EC50 with ERbwas E2¼ 2� 10�10 M,genistein¼ 2.2� 10�9 M. These data, using thetransient transfection approach, are in generalagreement with the regulation of endogenousPR gene expression observed in the above-described stably transfected hFOB/ER celllines. These data demonstrate that genisteinbehaves as a weak E2 agonist when signalingthrough either ERa or ERb at the level oftranscription (i.e., PRA promoter activity).

Regulation of Versican Expression

Screening for possible estrogenic regulationof extracellular matrix constituents in hFOB/ERa9 cells revealed amarked stimulation of themRNA levels for versican by E2. Versican is alarge chondroitin sulphate proteoglycan relatedto aggrecan and neurocan [Kimata et al., 1986;Zimmermann and Ruoslahti, 1989]. After a24 h treatment period, both E2 and genisteinincreased the versican mRNA expression inhFOB/ERa9 cells (Fig. 3A). Transcripts encod-ing two versican splice variants, designated V0and V1 that differ in their glycosaminoglycan(GAG) attachment domains, were detectable byNorthern blotting. A third splice variant, V2,was not induced by either E2 or genistein in thehFOB/ERa9 or hFOB/ERb6 cell lines (data notshown). Densitometry analysis showed thatboth genistein (10�7 and 10�6 M) and E2 (10

�9

to 10�8 M) produced a 2-fold increase in V0 andV1 mRNA levels relative to the ICI-treatedcontrol in the hFOB/ERa9 cells (Fig. 4B). In-terestingly, treatment of hFOB/ERb6 cells

with genistein or E2 from 24 h to 10 days,failed to modulate versican mRNA abundance(Fig. 3A,B).Raloxifene (10�8 and10�7M)hadnoeffect on the expression of versican in either ofthe cell lines. Fig. 3C shows the results of RT-PCR using isoform-specific primers for V0 andV1. The data support theNorthern blot analysis(Fig. 3A,B).

Fig. 3. Genistein and E2 increase versicanmRNA expression inhFOB/ERa9 cells but not in hFOB/ERb6 cells. The hFOB/ERa9cells were ICI pretreated before stimulation with genistein (Gen),E2, raloxifene (Ral) or ICI for 24 h in DMEM-F12mediumþ0.1%(w/v) BSA. The hFOB/ERb6 cells were treated with the sameagents continuously for 10 days in mediumþ 10% (v/v) CS-FBS.Panel A: Total RNA was extracted and analyzed for versicanexpression by Northern blot hybridization using a full-lengthcDNA probe. Panel B: Densitometry of the V0 and V1 versicanisoforms in hFOB/ERa9 cells from Panel A is shown relative tocells treated with ICI alone, after normalizing to the GAPDHlevel. A representative experiment, repeated two times withsimilar results, is presented. Panel C: RT-PCR analysis for the V0and V1 splice variants of versican.

Genistein Effects on ERs in Human Osteoblasts 639

The E2 and genistein regulation of versicanprotein concentration in hFOB/ERa9 cells wasthen assessed by indirect immunofluorescence.Staining of versican protein human osteosar-coma cells (Fig. 4A) served as a positive controlwith widespread and intense staining. Withno steroid treatment, the hFOB/ERa9 cellsshowed a minority of the cells with perinuclearand punctate cytoplasmic staining (Fig. 4C).Treatment of hFOB/ERa9 cells with E2

(Fig. 4D) or with genistein (Fig. 4E) for 48 hincreased both the staining intensity and thefraction of positively stained cells. Co-treatmentof these cells with E2 plus a 100-fold molarexcess of ICI 182,780 prevented the E2-mediated increase in versican protein staining,results which support the ICI antagonist effectat the mRNA level.

Responses of Other Bone MatrixProtein Genes

The effects of genistein, E2, and raloxifene onthe expression of the osteoblast marker genes,AP, OP, and ON, were also investigated in thetwo hFOB/ER cell lines. As previously reportedin hFOB/ERa9 cells [Robinson et al., 1997;Waters et al., 2001], E2 at the highest concen-tration tested (10�8 M) in hFOB/ERa9 (Fig. 5A)

stimulated the level of the 2.5 kb mRNA of AP.Genistein at 10�6 M, but not raloxifene, alsoincreased the APmRNA expression to a similardegree as E2. Conversely, both genistein and E2

inhibited the OP steady state mRNA levels(1.6 kb) in hFOB/ERa9 cells. The steady statemRNA levels for the two ON transcripts at3.0 kb and 2.2 kb were unaffected by any of theER ligands tested. As shown in Fig. 5B, theexpression of none of these genes inhFOB/ERb6cells was modulated by genistein, E2, or ralox-ifene, following prolonged treatment for 4–10 days.

Regulation of Interleukin-6 Synthesis

Interleukin-6 synthesis was also examinedfor possible regulation by E2 and genistein. Theconstitutive synthesis of IL-6 by both hFOB/ERa9 and hFOB/ERb6 cells has previously beendemonstrated to be suppressed by E2 [Waterset al., 2001], as has both IL-6 production byendometrial cells expressing only ERb [Debet al., 1999] and cytokine-stimulated IL-6 pro-ductionbyosteoblast lineagecells [Passeri et al.,1993]. In agreement with these reports, E2

treatment of hFOB/ER cells, expressing eitherERa or ERb, caused an inhibition of IL-6 syn-thesis as measured by ELISA of conditioned

Fig. 4. Immunofluorescence detection of versican protein inMG63 and hFOB/ERa9 cells. Versican was detected by indirectimmunofluorescence using an anti-human versican polyclonalantiserum and FITC-conjugated secondary antibody. MG63human osteosarcoma cells were used as a positive control(Panel A). As a negative control, the cells were incubated with

purified rabbit IgG instead of versican antiserum (Panel B). ThehFOB/ERa9 cells were treated for 48 h in DMEM-F12 med-iumþ0.1% (w/v) BSA with either vehicle (Panel C), 10�9 M E2(Panel D), 10�6 M genistein (Panel E) or co-treated with 10�9 ME2 and 10�7 M ICI (Panel F). Magnification �50.

640 Rickard et al.

medium (Fig. 6). The synthesis of IL-6 wassuppressed by a similar magnitude in both celllines with E2 at 10

�8M decreasing IL-6 levels to50–60% of the ICI-treated control. Again, geni-stein acted as an E2 agonist and significantlydecreased IL-6 synthesis via bothERa andERb.Interestingly, the suppression of IL-6 by genis-tein was both of greater magnitude and occur-red at lower concentrations in the hFOB/ERb6cell line than in hFOB/ERa9 cells, correlatingwith the higher affinity of genistein for ERbover ERa [Kuiper et al., 1998]. The genistein-mediated inhibition of IL-6 synthesis in bothhFOB/ER lines was reversed by co-administra-tion of ICI 182,780, verifying that the inhibitoryeffects were indeed ER dependent. Althoughraloxifene significantly reduced IL-6 productionin thehFOB/ERa9 cells, the responsewasminorandwas not dose-dependent. Raloxifene did notaffect IL-6 levels in hFOB/ERb6 cells. Thesedata support the notion that the regulatoryactions of E2 and genistein are virtually iden-tical once the higher level of genistein is achiev-ed, that genistein is a less potent estrogenic

compound, and that the particular ER isoformdictates the overall response to any steroidanalogue.

Cell Proliferation

Since the bFGF stimulates osteoblast prolif-eration throughan intracellular tyrosinekinasedomain on the FGF receptor and since genisteinis an inhibitor of tyrosine kinase activity (IC50�2.6 mM) [Akiyama et al., 1987], the effect ofgenistein on both the basal and bFGF-stimu-lated proliferation of the ER-negative hFOBcells was examined and compared to the effectsof E2 and raloxifene. As expected, for 5 daysof treatment, raloxifene (10�8 to 10�7 M) or E2

(10�10 and 10�8 M) was shown to have no effecton proliferation in the absence or presence ofbFGF (Fig. 7B) since the hFOB cells do notexpress endogenous ER. Genistein at 10�7 Mwas also without effect, but higher concentra-tions did show an inhibitory effect (Fig. 7A).Genistein at 10�6 M reduced the basal prolif-eration of hFOB cells by �20% and 10�5 Mproduced an 80% reduction in both the basal

Fig. 5. Analysis of the alkaline phosphatase (AP), osteopontin(OP), and osteonectin (ON) mRNA expression by genistein, E2and raloxifene in hFOB/ER cell lines. The hFOB/ER cells weretreated as described in Figure 4 and steady state mRNA levels intotal RNA analyzed by Northern blotting. Panel A: hFOB/ERa9cells. Panel B: hFOB/ERb6 cells. Membranes were sequentially

hybridized to cDNA probes for AP, OP, ON, and then forGAPDH to correct for RNA loading. The band intensities weredetermined by densitometry, normalized for GAPDH, and arepresented as mean of two separate experiments expressedrelative to levels in cells treated with ICI alone.

Genistein Effects on ERs in Human Osteoblasts 641

(Fig. 7A) and bFGF-stimulated (Fig. 7B) pro-liferation.

The inhibitory effect of genistein on prolifera-tion of these ER deficient hFOB cells was shownto be largely due to reduced cell proliferationrather than increased cytotoxicity as determin-ed by the Trypan blue exclusion method. At theend of the 5-day treatment period, culturesadministered with 10�5 M genistein had ap-proximately twice the number of non-viablecells as untreated cultures or cultures treatedwith 10�8 M E2 or 10�7 M raloxifene. In con-trast, over the same time period, the number ofviable cells increased only 1.5-fold in the pre-sence of 10�5 M genistein, whereas viable cellnumber had increased by�15-fold for the otherconditions (data not shown). Thus, the dramaticinhibition of DNA synthesis by genistein, asquantitated in the [3H]-thymidine incorpora-tion assay, was almost completely due to reduc-ed cell division rather than to increased celldeath.

DISCUSSION

The effect of the natural isoflavone, genistein,on the activity of osteoblast cell lines thatexpress either ERa or ERb, was compared tothat of E2 and a SERM (raloxifene). In almostevery instance, genistein behaved as a weakER agonist. In addition, genistein displayed areceptor independent function by its effects oncell proliferation when added at high concen-trations. Both E2 and genistein induced PRgene expression and promoter activity, stimu-lated the steady state mRNA levels for AP andversican, and inhibitedOP gene expression andIL-6 protein production. The levels of IL-6 andPR mRNA were regulated in both hFOB/ERa9and hFOB/ERb6 cells, whereas these levels forAP, OP, and versican were only regulated inthe hFOB/ERa9 cells. Interestingly, raloxifenefailed to regulate any of the parameters in theOB cells expressing either ER isoform. Differ-ences in gene regulation between the two celllines to E2 and genistein may reflect unequalER expression, which we suspect is somewhatlower in theERb cell line.Nonetheless, this doesnot prevent comparison of the effects of E2 withgenistein and raloxifene in each of the cell linesseparately, as presented here.

Fig. 6. Dose-dependent inhibition of IL-6 protein levels bygenistein and E2, but not by raloxifene, in hFOB/ER cell lines.Cells were treated for 4 days with genistein (Gen), E2, raloxifene(Ral), and/or ICI 182,780 (ICI) at the concentrations indicated inmedium containing 10% (v/v) CS-FBS. Conditioned mediumharvested on day 4 was assayed for IL-6 using an ELISA. Panel A:hFOB/ERa9. Panel B: hFOB/ERb6 cells. The data are themean� SEM of three separate experiments. Asterisks denote astatistically significant difference compared to treatment with ICIalone (100%).

Fig. 7. Receptor independent effects of genistein, E2, andraloxifene on the hFOB cell proliferation. Cells pretreated withICI were then treated for 5 days with genistein, E2 and raloxifenein the absence (Panel A) or presence (Panel B) of bFGF (10 ng/ml)in medium supplemented with 1% (v/v) CS-FBS. Cells werelabeled with [3H]-thymidine for the final 20 h of treatment andthe incorporation of radioactivity into trichloroacetic acid-precipitable material was determined as a measure of DNAsynthesis. Each treatment was added to six wells. Results aregivenas themean� SEMof a representative experiment repeatedtwo times. Significant differences between treatments and thecontrol-treatment are indicated by asterisks and significantdifferences between bFGF alone and bFGF co-treatments areindicated by 000P< 0.001.

642 Rickard et al.

In general, genistein concentrations of 10- to100-fold higher than for E2 were required toobserve responses of comparable magnitude.Similarly, using transient transfection assays,maximal activity of the PRA promoter wasachieved with an approximately 10-fold higherconcentration of genistein compared to E2, how-ever, the maximal promoter activity was stilllower with genistein. These findings thereforecorrespond, in part, to the reported�20-fold and3-fold lower binding affinity of genistein forERaand b, respectively, compared to E2 [Kuiperet al., 1998]. An additional explanation for thelower potency of genistein may include differ-ences in activity of the two genistein-complexedreceptor isoforms towards the endogenous PRpromoters compared to the transfected PRA

promoter fragment.The suppression of OP mRNA by genistein

and E2 in hFOB/ERa9 cells coincides with aprevious report demonstrating down-regula-tion of bone matrix protein expression, includ-ingOP, inbones ofE2-treated rats [Turner et al.,1990]. In addition, E2 has been shown to inhibitboth osteocalcin synthesis [Robinson et al.,1997] and mineralized extracellular matrixdeposition [Waters et al., 2001] in hFOB/ERa9cells. OP is an acidic glycoprotein important forcell attachment and considered a late marker ofosteoblastic differentiation and an earlymarkerof mineralization [Robey and Boskey, 1996].Overall, these findings suggest that in thisosteoblast cell line, ER agonists includinggenistein, arrest osteoblast differentiation priorto the onset of both mineralization and the ex-pression of late differentiation markers. Al-though the precise role of OP in bone is unclear,OPhas been shown to be required for osteoclast-mediated bone resorption: binding of extracel-lular OP to the integrin avb3 on the surface ofosteoclasts activates cytoskeletal reorganiza-tion in podosomes, stimulating cell motility andresorption [Ross et al., 1993]. Since OP caninduce osteoclastic bone resorption, at leastin vitro, the ER-mediated inhibition of OP ex-pression observed here may therefore contri-bute to the overall anti-resorptive effects ofestrogen. It remains unknown whether theinhibition ofOPexpression in osteoblasts occursby a direct suppression of gene transcription orindirectly as the result of arrested osteoblastdifferentiation.Both genistein and E2 stimulated the mRNA

levels for AP in hFOB/ERa9 cells, in accordance

with previous reports of the effects of estrogenon osteoblastic cells by our laboratory andothers [Robinson et al., 1997]. AP is generallyconsidered to be an earlier marker of osteoblastdifferentiation than OP, the levels of whichincrease in primary osteoblast cultures follow-ing cessation of cell division and continue to risethereafter [Owen et al., 1990]. Consequently, ifER agonists do arrest differentiation in hFOB/ERa9 cells, they do so at a stage subsequent toAP expression and before the induction of OP.

The synthesis of IL-6 protein byhFOB/ER celllines was inhibited by genistein and estrogenacting through both ER isoforms. The effectivegenistein concentration required to inhibit IL-6was lower in the ERb expressing osteoblastscompared to the ERa containing cells whichprobably reflects the greater affinity of genis-tein for ERb [Kuiper et al., 1998]. The anti-resorptive effects of high-dose isoflavones suchas genistein in vivo, may be due to the inhibi-tion of multinucleated osteoclast formation andactivity, as reported for high concentrations(>1 mM) of the synthetic isoflavone, ipriflavone,and select metabolites when added to boneorgan and unfractionated bone cell culturesin vitro [Notoya et al., 1993; Giossi et al., 1996].Since enhanced IL-6 production by osteoblastlineage cells has been strongly implicated incausing osteoclast differentiation and thus in-creased bone resorption and rate of bone loss inestrogen deficiency [Jilka et al., 1992; Passeriet al., 1993], the genistein-mediated suppres-sion of IL-6 may be an important additionalmechanism explaining the anti-resorptiveeffects of isoflavones. More recently, genisteinhas been reported to increase the ratio ofOPG:OPGL (RANKL) expression by osteogenicstromal cells, which would also be expected tosuppress osteoclastogenesis [Yamagishi et al.,2001]. The IL-6 promoter is suppressed by E2-ER complexes through the binding of thetranscription factors,NF-kBandC/EBPb [Steinand Yang, 1995]. It would therefore be of in-terest to determinewhether genistein-activatedERa and ERb suppress transcription of the IL-6gene via interactions with the same set offactors and response elements.

The identification of versican (PG-M inchickens) as a target for estrogen regulationrepresents another novel finding of this inves-tigation.Versican is a large,multidomain, chon-droitin sulphate-linked proteoglycan, related toaggrecan, brevican, and neurocan [Kimata

Genistein Effects on ERs in Human Osteoblasts 643

et al., 1986; Zimmermann and Ruoslahti, 1989].Wehave previously demonstrated that estrogensignificantly displays mineralized matrixdeposition in the hFOB/ERa9 cell line, but notin hFOB/ERb6 cells [Waters et al., 2001]. Wespeculate that one mechanism for this delay isthe ER regulation of different bone matrixprotein genes. Versican has been shown to bepredominantly expressed during embryonicskeletal development, and later in cartilageand unmineralized embryonic bone, as well asin numerous adult connective tissue cell types[Shinomura et al., 1990; Bode-Lesniewka et al.,1996]. It is possible that the estrogen inhibitionof mineralization of bone matrix is due to thealtered matrix proteins in the matrix due to theaction of estrogen, fibroblasts, keratinocytes,and arterial smooth muscle cells [Shinomuraet al., 1990; Bode-Lesniewka et al., 1996].During embryonic skeletal development, versi-can is localized to prechondrogenic condensa-tions and subsequently to the cartilage coreand immature bone, althoughversican is absentfrom mature cartilage [Kimata et al., 1986;Shinomura et al., 1990]. It is unknown whetherversican is present in adult human bone al-though the protein is synthesized and secretedby MG63 human osteosarcoma cells [LeBaronet al., 1992].

Versican has been shown to inhibit cellattachment to matrix as well as modulate cellmotility and differentiation through binding tohyaluronan, type I collagen and fibronectin,[Perissinotto et al., 2000]. In addition, the GAGside chains of versican can bind chemokines,modifying their activity. Since the function ofversican in cartilage and other connective tis-sues remains obscure, and its presence in adultmineralized bone is unknown, it is difficult tospeculate on the possible physiological signifi-cance of the estrogenic regulation of versicansynthesis in bone. Alternatively, versican ex-pression by the hFOB/ERa9 and hFOB/ERb6cells may be a consequence of their fetal origin,implying a role for estrogen in early stages ofbone formation.

Both E2 and genistein stimulated the expres-sion of the same splice variants of versican,designated V0 and V1, which are the iso-forms also expressed by fibroblasts [Dours-Zimmermann and Zimmerman, 1994]. A thirdsplice variant that has been detected in brain,V2, was neither constitutively expressed inhFOB/ER cells nor induced by the ER ligands.

Versican expression in arterial smooth musclecells has been shown to be stimulated byplatelet-derived growth factor (PDGF) andTGFb, and genistein has been shown to antag-onize the effect of PDGFby inhibiting thePDGFreceptor tyrosine kinase activity [Schonherret al., 1997]. This highlights the dual mechan-ism of genistein action on cellular responses,wherein the effects of low genistein concentra-tions are likely to be ER-mediated whereaseffects at high concentrations (>1 mM) may bedue to inhibition of tyrosine kinases (the IC50for the EGFR tyrosine kinase is 2.6 mM). In thisregard, in our study, the proliferation of the ER-negative hFOB cell line was potently inhibitedby 10�5 M genistein whereas lower concentra-tions of genistein as well as ER-saturatingconcentrations of both E2 and raloxifene werewithout effect.

Perhaps surprisingwas the lack of response ofthe hFOB/ER cell lines to raloxifene for any ofthe parameters examined here. The bioactivityof the HPLC-purified raloxifene preparationwas verified by antagonism of E2-stimulatedMCF-7 cell growth. Thus, at concentrationswhich should saturate the ER, raloxifene failedto affect either hFOB/ER cell line, despite thefact that the actions of E2 and raloxifene on bonemetabolism in vivo and on osteoclast lineagecells in vitro are very similar [Bryant et al.,1999]. Whereas some effects of raloxifene onosteoblastic cells are similar to those producedby E2, differences have also been reported. Forexample, the induction of TGFb3 expression byraloxifene in intact bone and cultured bone cellsis also produced by metabolites of E2 but onlyweakly by E2 itself [Yang et al., 1996]. More-over, raloxifene, but not E2 or idoxifene, antag-onized ERE-dependent transcriptional activityin osteoblastic cells [Nuttall et al., 2000]. Thus,it is possible that raloxifene and E2 may dif-ferentially regulate ER target genes in osteo-blasts and raloxifene alone may regulate aunique subset of genes not affected by E2.

In vivo treatment with isoflavones, in con-trast to estrogen and SERMs, has little effect onthe high bone turnover rate that is character-istic of sex steroid deficiency and reportedlyenhances the rate of bone formation [Gennariet al., 1997; Arjmandi et al., 2000]. Differencesin behavior between genistein and E2 couldtherefore be due to a differential regulation ofER-dependent mechanisms in bone cells. Usinghuman osteoblast cell lines that over-express

644 Rickard et al.

either ERa or ERb, we have shown that thenatural isoflavone, genistein, functions as anER agonist. However, at high concentration(10 mM) the effect of genistein diverges fromthose of E2 with regard to ER dependency andare speculated here to be due to the inhibition oftyrosine kinases. In general, our data demon-strates that genistein mimics the ER-mediatedeffects of E2 in osteoblasts. Although we cannotexclude the possibility that responses otherthan those examined may be differentially re-gulated by genistein and E2 acting via ER,divergent signaling of non-genomic and nuclearER-independent mechanisms may insteadaccount for the effects in vivo. Further investi-gation is necessary to fully define the actions ofgenistein on osteogenic cells.

ACKNOWLEDGMENTS

Dr. Malayannan Subramaniam and StevenA. Johnsen are thanked for their insightfulcomments and suggestions during the course ofthis work. This project was supported by NIHgrants AG04875 (to TCS, SK, and BLR), NIHtraining grant DK07352 (to DJR.), and theMayo Foundation.

REFERENCES

Agnusdei D, Gennari C, Bufalino L. 1995. Prevention ofearly postmenopausal bone loss using low doses of con-jugated estrogens and the non-hormonal, bone-activedrug ipriflavone. Osteoporos Int 5:462–466.

Akiyama T, Ishida J, Nakagawa S, Ogawara H, WatanabeS, Itoh N, Shibuya M, Fukami Y. 1987. Genistein, a spe-cific inhibitor of tyrosine-specific protein kinases. J BiolChem 262:5592–5595.

Alekel DL, St. Germain A, Peterson CT, Hanson KB,Stewart JW, Toda T. 2000. Isoflavone-rich soy proteinisolate attenuates bone loss in the lumbar spine of peri-menopausal women. Am J Clin Nutr 72:844–852.

Arjmandi BH, Getlinger MJ, Goyal NV, Alekel L, HaslerCM, Juma S, Drum ML, Hollis BW, Kukreja SC. 1998.Role of soy protein with normal or reduced isoflavonecontent in reversing bone loss induced by ovarian hor-mone deficiency in rats. Am J Clin Nutr 68:1358S–1363S.

Arjmandi BH, Birnbaum RS, Juma S, Barengolts E,Kukreja SC. 2000. The synthetic phytoestrogen, ipri-flavone, and estrogen prevent bone loss by differentmechanisms. Calcif Tissue Int 66:61–65.

Benvenuti S, Tanini A, Frediani U, Bianchi S, Masi L,Casana R, Bufalino L, Serio M, Brandi ML. 1991. Effectsof ipriflavone and its metabolites on a clonal osteoblasticcell line. J Bone Miner Res 6:987–996.

Bode-Lesniewka B, Dours-Zimmermann MT, OdermattBF, Briner J, Heitz PU, Zimmerman DR. 1996. Distribu-tion of the large aggregating proteoglycan versican inadult human tissues. J Histochem Cytochem 44:303–312.

Bryant HU, Glasebrook AL, Yang NN, Sato M. 1999. Anestrogen receptor basis for raloxifene action in bone.J Steroid Biochem Mol Biol 69:37–44.

Cheng S-L, Zhange S-F, Nelson TL,Warlow PM, Civitelli R.1994. Stimulation of human osteoblast differentiationand function by ipriflavone and its metabolites. CalcifTissue Int 55:356–362.

Cooper C, Campion G. 1992. Hip fractures in the elderly: Aworld-wide projection. Osteoporos Int 2:285–289.

Deb S, Tessier C, Prigent-Tessier A, Barkai U, Ferguson-Gottschall S, Srivastava RK, Faliszek J, Gibori G. 1999.The expression of interleukin-6 (IL-6), IL-6 receptor, andgp-130-Kilodalton glycoprotein in the rat decidua and adecidual cell line: Regulation by 17b-estradiol andprolactin. Endocrinology 140:4442–4450.

Dours-Zimmermann MT, Zimmerman DR. 1994. A novelglycosaminoglycan attachment domain identified in twoalternative splice variants of human versican. J BiolChem 269:32992–32998.

Evans GL, Bryant HU, Magee DE, Turner RT. 1996.Raloxifene inhibits bone turnover and prevents furthercancellous bone loss in adult ovariectomized rats withestablished osteopenia. Endocrinology 137:4139–4144.

Gennari C, Adami S, Agnusdei D, Bufalino L, Cervetti R,Crepaldi G, Di Marco C, Di Munno O, Fantasia L, IsaiaGC, Mazzuoli GF, Ortolani S, Passeri M, Serni U,Vecchiet L. 1997. Effect of chronic treatment with ipri-flavone in postmenopausal women with low bone mass.Calcif Tissue Int 61:S19–S22.

Giossi M, Caruso P, Civelli M, Bongrani S. 1996. Inhibitionof parathyroid hormone-stimulated resorption in cul-tured fetal rat long bones by the main metabolites ofipriflavone. Calcif Tissue Int 58:419–422.

Glazier MG, Bowman MA. 2001. A review of the evidencefor the use of phytoestrogens as a replacement for tradi-tional estrogen replacement therapy. Arch Intern Med161:1161–1171.

Harris SA, Tau KR, Enger RJ, Toft DO, Riggs BL,Spelsberg TC. 1995. Estrogen response in hFOB1.19human fetal osteoblastic cell line stably transfected withthe human estrogen receptor gene. J Cell Biochem 59:193–201.

Jilka RL, Hangoc G, Girasole G, Passeri M, Williams DC,Abrams JS, Boyce B, Broxmeyer H, Manolagas SC. 1992.Increased osteoclast development after estrogen loss:Mediation by interleukin-6. Science 257:88–91.

Kastner P, Krust A, Turcotte B, Stropp U, Tora L,Gronemeyer H, Chambon P. 1990. Two distinct estro-gen-regulated promoters generate transcripts encodingthe two functionally different progesterone receptorforms A and B. EMBO J 9:1603–1614.

Kimata K, Oike Y, Tani K, Shinomura T, Yamagata M,Uritani M, Suzuki S. 1986. A large chondroitin sulfphateproteoglycan (PG-M) synthesized before chondrogenesisin the limb bud of chick embryo. J Biol Chem 261:13517–13525.

Kuiper GGJM, Lemmen JG, Carlsson B, Corton JC, SafeSH, van der Saag PT, van der Burg B, Gustafsson J-A.1998. Interaction of estrogenic chemicals and phytoestro-gens with estrogen receptor b. Endocrinology 139:4252–4263.

LeBaron RG, Zimmerman DR, Ruoslahti E. 1992. Hyalur-onate binding properties of versican. J Biol Chem 267:10003–10010.

Genistein Effects on ERs in Human Osteoblasts 645

MacNamara P, Loughrey HC. 1998. Oestrogen regulatesthe expression of progesterone receptor A and B isoformsin human osteoblast cells. Calcif Tissue Int 63:39–45.

Notoya K, Yoshida K, Taketomi S, Yamazaki I, KumegawaM. 1993. Inhibitory effect of ipriflavone on osteclast-mediated bone resorption and new osteoclast formationin long-term cultures of mouse unfractionated bone cells.Calcif Tissue Int 53:206–209.

Nuttall ME, Stroup GB, Fisher PW, Nadeau DP, GowenM,Suva LJ. 2000. Distinct mechanisms of action of selectiveestrogen receptor modulators in breast and osteoblasticcells. Am J Physiol Cell Physiol 279:C1550–C1557.

Oursler MJ. 1998. Estrogen regulation of gene expressionin osteoblasts and osteoclasts. Crit Rev Euk Gene Expr8:125–140.

Owen TA, Aronow M, Shalhoub V, Barone LM, Wilming L,Tassinari MS, KennedyMB, Pockwinse S, Lian JB, SteinGS. 1990. Progressive development of the rat osteoblastsphenotype in vitro: Reciprocal relationships in expres-sion of genes associated with osteoblast proliferation anddifferentiation during formation of the bone extracellularmatrix. J Cell Physiol 143:420–430.

Paech K, Webb P, Kuiper GGJM, Nilsson S, Gustafsson J-A, Kushner PJ, Scanlan TS. 1997. Differential ligandactivation of estrogen receptors ERa and ERb at AP-1sites. Science 277:1508–1510.

Passeri M, Biondi M, Costi D, Bufalino L, Castiglione GN,Di Peppe C, Abate G. 1992. Effect of ipriflavone on bonemass in elderly osteoporotic women. BoneMiner 19:S57–S62.

Passeri G, Girasole G, Jilka RL, Manolagas SC. 1993.Increased interleukin-6 production by murine bonemarrow and bone cells after estrogen withdrawal. Endo-crinology 133:822–828.

Perissinotto D, Iacopetti P, Bellina I, Doliana R, ColombattiA, Pettway Z, Bronner-Fraser M, Shinomura T, KimataK, Morgelin M, Lofberg J, Perris R. 2000. Avian neuralcrest cell migration is diversely regulated by the twomajor hyaluronan-binding proteoglycans PG-M/versicanand aggrecan. Development 127:2823–2842.

Picherit C, Bennetau-Pelissero C, Chanteranne B, Lebec-que P, Davicco M-J, Barlet J-P, Coxam V. 2001. Soybeanisoflavones dose-dependently reduce bone turnover butdo not reverse established osteopenia in adult ovariecto-mized rats. J Nutr 131:723–728.

Potter SM, Baum JA, Teng H, Stillman RJ, Shay NF,Erdman JWJ. 1998. Soy protein and isoflavones: Theireffects on blood lipids and bone density in postmenopau-sal women. Am J Clin Nutr 68:1375S–1379S.

Rickard DJ, Hofbauer LC, Bonde SK, Gori F, Spelsberg TC,Riggs BL. 1998. Bone morphogenetic protein-6 produc-tion in human osteoblastic cell lines. Selective regulationby estrogen. J Clin Invest 101:413–422.

Rickard DJ, Waters KM, Ruesink TJ, Khosla S, Katzenel-lenbogen JA, Katzenellenbogen BS, Riggs BL, SpelsbergTC. 2002. Estrogen receptor isoform-specific induction ofprogesterone receptors in human osteoblasts. J BoneMiner Res 17:580–592.

Robey PG, Boskey AL. 1996. The biochemistry of bone. In:Marcus R, Feldman D, Kelsey J, editors. Osteoporosis.San Diego, CA; Academic Press, Inc. pp 95–183.

Robinson JA, Harris SA, Riggs BL, Spelsberg TC. 1997.Estrogen regulation of human osteoblastic cell prolifera-tion and differentiation. Endocrinology 138:2919–2927.

Ross FP, Chappel J, Alvarez JI, Sander D, Butler WT,Farach-Carson MC, Mintz KA, Robey PG, TeitelbaumSL, Cheresh DA. 1993. Interaction between the bonematrix proteins osteopontin and bone sialoprotein andthe osteoclast integrin alpha V beta 3 potentiate boneresorption. J Biol Chem 268:9901–9907.

Russell-Aulet M, Wang J, Thornton JC, Colt EW, PiersonRNJ. 1993. Bone mineral density and mass in a cross-sectional study of white and Asian women. J Bone MinerRes 8:575–582.

Saville B, Wormke M, Wang F, Nguyen T, Enmark E,Kuiper G, Gustafsson J-A, Safe S. 2000. Ligand-, cell-,and estrogen receptor subtype (a/b)-dependent activationat GC-rich (Sp1) promoter elements. J Biol Chem 275:5379–5387.

Schonherr E, KinsellaMG,Wight TN. 1997. Genistein selec-tively inhibits platelet-derived growth factor-stimulatedversican biosynthesis in monkey arterial smooth musclecells. Arch Biochem Biophys 339:353–361.

Shinomura T, Jensen KL, Yamagata M, Kimata K, SolurshM. 1990. The distribution of mesenchyme proteoglycan(PG-M) during wing bud outgrowth. Anat Embryol 181:227–233.

Stein B, Yang MX. 1995. Repression of the interleukin-6promoter by estrogen receptor is mediated by NF-kB andC/EBPb. Mol Cell Biol 15:4971–4979.

Turner RT, Colvard DS, Spelsberg TC. 1990. Estrogeninhibition of periosteal bone formation in rat long bone:Down regulation of gene expression for bone matrixproteins. Endocrinology 127:1346–1351.

UesugiT,TodaT,TsujiK, IshidaH. 2001.Comparative studyon reduction of bone loss and lipidmetabolismabnormalityin ovariectomized rats by soy isoflavones,Daidzin, genistinand glycitin. Biol Pharm Bull 24:368–372.

Wakeling AE, Valcaccia B, Newboult E, Green LR. 1984.Non-steroidal antioestrogens-receptor binding and biolo-gical response in rat uterus, rat mammary carcinomacells and human breast cancer cells. J Steroid BiochemMol Biol 20:111–120.

Waters KM, Rickard DJ, Riggs BL, Khosla S, Katzenellen-bogen JA, Katzenellenbogen BS, Moore J, Spelsberg TC.2001. Estrogen regulation of human osteoblast functionis determined by the stage of differentiation and theestrogen receptor isoform. J Cell Biochem 83:448–462.

Wronski TJ, Cintron M, Doherty AL, Dann LM. 1988.Estrogen treatment prevents osteopenia and depressesbone turnover in ovariectomized rats. Endocrinology123:681–686.

Yamagishi T, Otsuka E, Hagiwara H. 2001. Reciprocalcontrol of expression of mRNAs for osteoclast differentia-tion factor and OPG in osteogenic stromal cells bygenistein: Evidence for the involvement of topoisomeraseII in osteoclastogenesis. Endocrinology 142:3632–3637.

Yamaguchi M, Sugimoto E. 2000. Stimulatory effect ofgenistein and daidzein on protein synthesis in osteo-blastic MC3T3-E1 cells: Activation of aminoacyl-tRNAsynthetase. Mol Cell Biochem 214:97–102.

Yang NN, Venugopalan M, Hardikar S, Glasebrook A.1996. Identification of an estrogen response elementactivated by metabolites of 17beta-estradiol and ralox-ifene. Science 273:1222–1225.

Zimmermann DR, Ruoslahti E. 1989. Multiple domains ofthe large fibroblast proteoglycan, versican. EMBO J 8:2975–2981.

646 Rickard et al.