Induction of Osteoblast Differentiation by Selective Activation of Kinase-Mediated Actions of the Estrogen Receptor

Abstract

Estrogens control gene transcription by cis or trans interactions of the estrogen receptor (ER) with target DNA or via the activation of cytoplasmic kinases. We report that selective activation of kinase-mediated actions of the ER with 4-estren-3α,17β-diol (estren) or an estradiol-dendrimer conjugate, each a synthetic compound that stimulates kinase-mediated ER actions 1,000 to 10,000 times more potently than direct DNA interactions, induced osteoblastic differentiation in established cell lines of uncommitted osteoblast precursors and primary cultures of osteoblast progenitors by stimulating Wnt and BMP-2 signaling in a kinase-dependent manner. In sharp contrast, 17β-estradiol (E₂) suppressed BMP-2-induced osteoblast progenitor commitment and differentiation. Consistent with the in vitro findings, estren, but not E₂, stimulated Wnt/β-catenin-mediated transcription in T-cell factor-lacZ transgenic mice. Moreover, E₂ stimulated BMP signaling in mice in which ERα lacks DNA binding activity and classical estrogen response element-mediated transcription (ERα^NERKI/−) but not in wild-type controls. This evidence reveals for the first time the existence of a large signalosome in which inputs from the ER, kinases, bone morphogenetic proteins, and Wnt signaling converge to induce differentiation of osteoblast precursors. ER can either induce it or repress it, depending on whether the activating ligand (and presumably the resulting conformation of the receptor protein) precludes or accommodates ERE-mediated transcription.

Steroid hormones, including estrogens and androgens, can induce cellular responses not only through direct activation of gene transcription resulting from cis or trans interactions of their receptors with DNA binding sequences on target gene promoters but also indirectly. The latter responses result from extranuclear actions of the same hormone receptors, causing activation of various cytoplasmic protein kinase cascades, which in turn alter the activities of other transcription factors or coactivators of the receptors themselves (5, 6, 27, 40, 43, 46, 53, 67). During the last few years, we and others have identified several synthetic compounds that, unlike estradiol, can selectively activate kinase-initiated routes by which the estrogen receptor (ER) controls gene transcription (28, 38, 60). By use of these compounds, substantial evidence that kinase-mediated effects on gene transcription are dissociable from direct cis or trans effects on transcription has been obtained (28, 38, 39, 54). Moreover, the use of selective activators of kinases, as well as compounds that dissociate the classical genomic actions of the estrogen receptor from cross talk with other transcription factors (8, 56), has provided extensive proof of the general concept that it is possible to eliminate the uterotropic activity of estrogens while retaining other nonreproductive actions. Together with cell and murine models expressing mutant receptors that cannot enter the nucleus and interact with DNA directly, these tools have become increasingly important for our understanding of nuclear receptor pharmacology (25, 37, 47, 48, 64). Heretofore, however, it has remained unknown whether kinase-mediated actions of the ER or androgen receptor (AR), in the absence of classical genotropic actions, could result in unique biologic outcomes that cannot be elicited by the natural sex steroids.

4-Estren-3α,17β-diol (estren) is a synthetic ligand of the ER or AR which in standard equilibrium assays binds the ER with an affinity that is about 0.15% of that of 17β-estradiol (E₂) (38) and AR with an affinity which is about 2% of that of the potent androgen R1881 (J. A. Katzenellenbogen, University of Illinois, personal communication). Estren potently activates kinase-mediated actions of the ER or AR and downstream transcriptional events at concentrations 3 to 4 orders of magnitude lower than those required to stimulate classical genotropic transcription (38, 40). Evidently, such kinase-mediated actions are responsible for the anti- and proapoptotic effects of estren, as well as estrogens and androgens, in osteoblasts/osteocytes and osteoclasts, respectively. Moreover, estren reversed the loss of bone mass and strength in ovariectomized (ovx) female or orchidectomized male mice, while it had either no effect (39) or a blunted effect (31, 49, 65) on reproductive organs.

Using HeLa cells transduced with wild-type (wt) ERα or the ligand binding domain of ERα localized to the cell membrane, the OB-6 osteoblastic cell line, MCF-7 breast carcinoma cells, and uteri from mice treated with E₂ or estren, we have shown that nongenotropic ER actions regulated a population of genes distinct from those regulated by genotropic ER actions (2). Specifically, estren and E₂ acting via membrane-localized ERα upregulated the expression of Wnt members and their Frizzled receptors as well as extracellular signal-regulated kinase (ERK)-regulated transcriptional targets but had no effect on several estrogen response element (ERE)- or AP-1-containing genes. In agreement with these observations, a cell-impermeable estrogen dendrimer conjugate (EDC) comprising abiotic nondegradable poly(amido)amine macromolecules and multiple estrogen molecules stimulated ERK, Shc, and Src phosphorylation in MCF-7 breast cancer cells at low concentrations (28). Yet, EDC was ineffective in stimulating transcription of endogenous estrogen target genes, being approximately 10,000-fold less potent than estradiol in genomic actions.

Genetic studies of humans and mice have established that the lipoprotein receptor-related protein 5/Wnt and bone morphogenetic protein 2 (BMP-2) signaling cascades are potent anabolic stimuli for bone. Wnts promote osteoblast differentiation, proliferation, or survival via inactivation of glycogen synthase kinase 3β (GSK-3β), stabilization of β-catenin, and β-catenin-mediated activation of transcription factors, such as the T-cell factor (TCF) (1, 24, 61). BMPs, acting via their receptors, induce phosphorylation of Smad1, Smad5, and Smad8 and subsequent recruitment of Smad4 to the phosphorylated Smad complex. Eventually, the complex translocates to the nucleus and induces expression of Runx2, a transcription factor that is absolutely required for osteoblast differentiation (for a review, see reference 9).

Heretofore, our mechanistic understanding of the actions of estren on bone has been limited to the demonstration of similar effects of estren and classical sex steroids in prolonging osteoblast survival, stimulating osteoclast apoptosis, and preventing the loss of bone mass and strength in gonadectomized mice. In view of our earlier work showing that estrogens attenuate osteoblastogenesis (13) and that estren has a transcriptional signature distinct from that of classical estrogens or androgens on osteoblastic cells and especially the ability to upregulate the expression of Wnt proteins and Frizzled receptors (2, 2b), we explored the hypothesis that selective activation of kinase-initiated routes of gene transcription may lead to a unique biologic outcome, i.e., promotion of the differentiation of uncommitted osteoblast progenitors.

We report that estren does induce differentiation of uncommitted osteoblast progenitors by stimulating Wnt as well as BMP signaling in a kinase-dependent manner in vitro and in vivo. This effect can be reproduced in vitro by EDC, and it involves a large signalosome in which inputs from the ER, kinases, BMPs, and Wnt signaling converge. Natural estrogens or androgens cannot replicate these effects. In fact, in sharp contrast to estren, E₂ suppresses exogenous or endogenous BMP-induced osteoblast commitment and differentiation. Elimination of ERE-mediated transcription in mice in which ERα lacks DNA binding activity (ERα^NERKI/−) allowed E₂ to induce phosphorylation of Smad1/5/8 in bone in vivo and in calvarium cells in vitro. Thus, the ER can either induce or repress osteoblastogenesis, depending on whether the activating ligand (and presumably the resulting conformation of the receptor protein) precludes or accommodates ERE-mediated transcription.

MATERIALS AND METHODS

Chemicals.

E₂, dihydrotestosterone (DHT), flutamide, wortmannin, LiCl, and Alizarin red (AR-S) were purchased from Sigma Chemical Co. (St. Louis, MO). BMP-2, Dkk-1, and Wnt3a were purchased from R&D Systems (Minneapolis, MN). ICI 182,780 was purchased from Tocris Cookson Inc. (Ellisville, MO). PP1 was obtained from Biomol International P.A. (Plymouth Meeting, PA), PD98059 from Promega (Madison, WI), and SP600125 from Calbiochem (San Diego, CA). Estren, 19-nortestosterone, and dihydronandrolone were purchased from Steraloids Inc. (Newport, RI). Propyl-pyrazole-triol (PPT), EDC, and the EDC control were kindly provided by John A. Katzenellenbogen (University of Illinois at Urbana, IL) (28, 57, 58). Slow-release pellets containing E₂, estren, or DHT were purchased from Innovative Research of America (Sarasota, FL).

Generation, breeding, and care of animals.

Transgenic mice expressing a TCF-β-galactosidase (TCF-β-Gal) reporter construct were obtained from Jackson Laboratories (Bar Harbor, ME). Mice heterozygous for an ERα knock-in mutant that cannot bind to the estrogen response element (ERα^NERKI/+) were provided by J. Larry Jameson (Division of Reproductive Endocrinology and Infertility, Feinberg School of Medicine, Northwestern University, Chicago, IL) (35). Mice carrying an inactivating mutation in the ERα locus (ERα^+/−) were provided by Andree Krust and Pierre Chambon (Institute for Genetics and Cellular and Molecular Biology, Strasbourg, France) (15). ERα^NERKI/+ mice were crossed with heterozygote ERα^+/− female mice to produce animals in which ERα is incapable of DNA binding (ERα^NERKI/−). Female ERα^NERKI/− mice and their wild-type ERα^+/+ littermates were used for experiments. For the experiments with ovariectomized animals, the mice were anesthetized with an intraperitoneal injection of sodium pentobarbital (80 mg/kg body weight). The back was shaved and cleansed with betadine, and an incision was made to expose the ovaries, which were then removed or left intact in the case of sham-operated animals. The incision was then closed with a wound clip. All animals were housed in a temperature-controlled room (22 ± 2°C), with a daily 12-h-light/12-h-dark schedule. During the study, animals had free access to water and were fed a phytoestrogen-free diet (2014 Teklad Global 14% protein rodent maintenance diet; Harland Teklad, Madison, WI). Pups were genotyped at 4 to 5 weeks of age by PCR as described previously (15, 35). Animals were euthanized by inhalation of 5% isoflurane, followed by cervical dislocation. All procedures were approved by the Institutional Animal Care and Use Committee.

Cell cultures.

C2C12 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). MC3T3-E1 cells were maintained in minimum essential medium (MEM) and 10% FBS. ST2 and UAMS32 cells were maintained in α-MEM supplemented with 10% FBS (18). Calvarium-derived osteoblastic cells were obtained and cultured as described previously (38, 40).

Plasmids.

Murine Dkk-1 was provided by C. Niehrs, Division of Molecular Embryology, Deutsches Krebsforschungszentrum, Germany (22). A reporter plasmid carrying three TCF binding sites upstream of a minimal c-fos promoter driving the firefly luciferase gene (TOPFLASH) was provided by B. Vogelstein, Johns Hopkins University Medical Institutions, Baltimore, MD (29). The Smad6-Luc reporter plasmid was obtained from Kohei Miyazono, Cancer Institute of the Japanese Foundation for Cancer Research, Tokyo, Japan (33).

Transient transfections and reporter assays.

MC3T3-E1 cells were transfected with 0.1 μg of the TCF-Luc or Smad6-Luc reporter construct and 0.2 μg of empty vector or Dkk-1 plasmid along with 0.01 μg of the Renilla-Luc reporter, using Lipofectamine Plus (Invitrogen, Carlsbad, CA). Cells were allowed to recover for 24 hours in the presence of 10% serum and then treated with the appropriate compounds for an additional 24 h under serum-free conditions. Luciferase activity was determined using a dual luciferase kit (Promega) and was normalized for Renilla luciferase activity.

Measurement of AP activity and osteocalcin production.

C2C12 cells were seeded at a density of 2 × 10⁴/cm² in 10% FBS. The following day, cells were placed in 5% serum-containing medium and were pretreated with either vehicle, ICI 182,780, flutamide, PP1, wortmannin, SP600125, noggin, or human recombinant Dkk-1 for 1 h. Vehicle, BMP-2, and steroids were added to the cells, and cultures were continued for 3 to 5 days. Cells were lysed in 100 mM glycine, 1 mM MgCl₂, and 1% Triton X-100 at pH 10. Alkaline phosphatase (AP) activity in the cell lysate was determined using a buffer containing 2-amino-2-methylpropanol and p-nitrophenylphosphate (Sigma-Aldrich Inc). The amount of osteocalcin secreted in the medium was determined by a radioimmunoassay (Biomedical Technologies, Inc., Stoughton, MA). Both activities were normalized for total protein concentration, determined using a Bio-Rad DC protein assay kit (Hercules, CA).

Mineralization assay.

MC3T3-E1 or ST2 cells (5 × 10⁴/cm²) were treated with vehicle, BMP-2, E₂, or estren in the presence of mineralization medium containing 10% FBS, 50 μg/ml ascorbic acid, and 10 mM β-glycerolphosphate for 11 or 23 days, respectively. The mineralized matrix was stained with 40 mM AR-S and destained with 10% (wt/vol) cetylpyridinium chloride in 10 mM sodium phosphate (pH 7.0) for quantification. The AR-S concentration was determined by absorbance measured at 570 nm using an AR-S standard curve.

Preparation of bone extracts for Western blot analysis.

The fifth lumbar vertebrae (L5) were obtained from each experimental mouse. Lysates were prepared and used for Western blot analysis as described previously (40).

Western blot analysis.

The phosphorylation status of GSK-3β, Smad1/5/8, ERK1/2, and Elk-1 was analyzed by immunoblotting, using antibodies recognizing Ser9-phosphorylated GSK-3β, phosphorylated Smad1/5 (Ser463/465), and phosphorylated Smad8 (Ser426/428) (Cell Signaling); GSK-3β (BD Biosciences); and Smad4, phosphorylated ERK1/2, ERK1/2, Ser383-phosphorylated Elk-1, and Elk-1 (Santa Cruz Biotechnology). Primary antibodies were detected with a horseradish peroxidase-conjugated secondary antibody (Santa Cruz) and SuperSignal West Pico chemiluminescent substrate (Pierce).

Real-time PCR.

Total RNA was extracted and real-time PCR performed as previously described (1, 2, 40). The primers and probes for alkaline phosphatase, Axin2, Frizzled 1, Smad6, and 18S rRNA (used as the housekeeping gene) were manufactured by the Assays-by-Demand service (Applied Biosystems, Foster City, CA). The probe and primers for BMP-2, osteocalcin, Runx2, Wnt1, Dkk-1, c-Jun, C3, and β-Gal were also obtained from Applied Biosystems and are described in Table 1.

TABLE 1.

Gene-specific primers and probes used for real-time PCR^a

Gene	Primer	Probe
BMP-2	F: CCCGCGTGCTTCTTAGAC	CATGGTCGACCTTTAG
	R: GAAGCAGCAACGCTAGAAGACA
C3	F: GGAGCCAGTGGACATCTGAGA	TCTCCATCAAGATTCC
	R: CCCTCCTTATCTGAGTTGAATTCCT
Dkk-1	F: GGGCTGTGTTGTGCAAGACA	TTCTGGTCCAAGATCT
	R: GGTGCACACCTGACCTTCTTTAA
c-Jun	F: CCGGTGGGCAGTCTGAAG	CCGAGTTCTTGGCGCG
	R: GTCGGGCGACGTGAGAAG
Osteocalcin	F: GCTGCGCTCTGTCTCTCTGA	AAGCCCAGCGGCC
	R: TGCTTGGACATGAAGGCTTTG
Runx2	F: AAGGTTCAACGATCTGAGATTTGTG	ACGAGGCAAGAGTTT
	R: GTGAAGACGGTTATGGTCAAGGT
Wnt1	F: AACCGCCGCTGGAACTG	TTCGGCAAGATCGTC
	R: GCGAAGATGAACGCTGTTTCTC
β-Gal	F: GGCTTACGGCGGTGATTTTG	ATCGTTCGGCGTATCGC
	R: CAAAGACCAGACCGTTCATACAGAA

^aF and R denote forward and reverse primers, respectively.

β-Galactosidase activity.

L5 vertebrae, tibiae, and femurs were obtained from each experimental mouse. The spinal cord was removed from the vertebrae, and bones were thoroughly cleaned to remove the attached soft tissues and homogenized in 500 μl lysis buffer (100 mM potassium phosphate [pH 7.8] and 0.2% Triton X-100). Fifty to 100 μg of protein was used for measuring β-galactosidase activity with a Galacto-Light kit (Applied Biosystems) according to the manufacturer's instructions.

Statistical analysis.

The data were analyzed by analysis of variance (ANOVA), and the Student-Newman-Keuls method was used to estimate the levels of significance of differences between means. Results shown were reproduced in at least three experiments. Dunn's test on ANOVA by ranks was used for comparisons in which variances were not equivalent.

RESULTS

Upregulation of Wnt signaling by estren, but not estradiol or DHT, in osteoblastic cells and uncommitted osteoblast progenitors and in bone in vivo.

Based on our observations that estren upregulates the expression of Wnt members and Frizzled receptors in ERα-expressing HeLa cells (2), we examined whether it could reproduce these effects in osteoblastic cells. For this purpose, MC3T3-E1 cells, a model of preosteoblastic cells committed to the lineage, were treated with estren or E₂ for the indicated time points and mRNA levels were quantified by real-time PCR (Fig. 1A). The expression of Wnt-1, a representative Wnt protein that activates β-catenin-mediated transcription, as well as Dkk-1, an inhibitor of Wnt signaling, and c-Jun, a reported regulator of Dkk-1 expression (26), was assessed. Estren downregulated the expression of both c-Jun and its target Dkk-1 by 50%. The suppressive effect of estren on both c-Jun and Dkk-1 expression in MC3T3-E1 cells was detectable within 1 h, preceding an upregulation of Wnt-1 at 24 h. E₂ also had an early suppressive effect on c-Jun and Dkk-1 at 1 h, but unlike estren, it did not upregulate Wnt-1 at the later time point.

FIG. 1.

Estren, but not E₂ or DHT, upregulates the expression of Wnt signaling components and activates Wnt/β-catenin-mediated transcription in osteoblast precursors. (A) MC3T3-E1 cells were treated with vehicle or 10⁻⁸ M E₂ or estren for the indicated periods of time. Expression of c-Jun, Dkk-1, and Wnt1 was determined by real-time PCR. (B) MC3T3-E1 cells were transiently transfected with TCF-Luc and either empty vector or Dkk-1. Twenty-four hours following transfection, cells were treated with vehicle, 40 mM LiCl, or 10⁻⁸ M of E₂, estren, or PPT for an additional 24 h. Luciferase activity was measured in cell lysates. TCF-Luc values were normalized with Renilla-Luc values for transfection efficiency. Normalized data are expressed as relative luciferase units (RLU). (C) C2C12 cells were treated with vehicle or 10⁻⁸ M estren, E₂, or DHT for 24 h. Axin2 and Fzd1 expression was quantified by real-time PCR. (D) MC3T3-E1 cells were treated with vehicle or PD98059 (PD; 25 μM) for 30 min prior to treatment with 10⁻⁸ M estren for an additional 24 h. Axin2 expression was quantified by real-time PCR. In panels A to D, bars indicate means ± standard deviations for triplicate determinations. *, P was <0.05 for comparison with vehicle by ANOVA. (E) Calvarium-derived osteoblastic cells were serum starved for 4 h and were then treated with vehicle or 10⁻⁸ M estren for 2 or 3 h. GSK-3β phosphorylation was determined by Western blot analysis. Blots were developed by chemiluminescence, and the intensities of the bands were quantified with an imaging system.

In addition to Wnt-1, several other Wnt proteins and Frizzled protein(s) are expressed by osteoblastic cells (23, 36, 55). However, it is unclear which Wnt and Frizzled protein(s) are responsible for the prodifferentiating effect of Wnt signaling on osteoblasts. However, it is well established that β-catenin/TCF-mediated transcription is an essential requirement for promotion of osteoblast differentiation by Wnts. Therefore, we focused on the ability of estren to activate ΤCF, using a reporter construct. We found that estren stimulated the activity of the TCF-luciferase (TCF-Luc) reporter construct (Fig. 1B). The effect of estren was greater than the effect of E₂ but similar to that of LiCl, which inactivates GSK-3β and was used as a positive control in these experiments. PPT, a synthetic compound that selectively activates genotropic actions of ERα (38, 40), was ineffective in inducing luciferase activity. The induction of TCF-Luc activity by estren was attenuated by transfection of a plasmid expressing Dkk-1, indicating that LRP5 and LRP6 are required for the ability of estren to stimulate TCF. Moreover, estren, but not E₂ or DHT, induced the expression of the Wnt receptor Fzd1 and the Wnt/β-catenin target Axin2 in cultures of both MC3T3-E1 (Fig. 1C) and C2C12 (Fig. 1D) cells. The latter is a murine cell line that can differentiate toward either myoblasts or osteoblasts under the appropriate stimuli. Upregulation of Axin2 expression by estren in MC3T3-E1 cells was abrogated by the inhibitor of ERK phosphorylation PD98059 (Fig. 1D), as well as the ER antagonist ICI 182,780, the AR antagonist flutamide, Dkk-1, or noggin, but not inhibitors of phosphatidylinositol 3-kinase (PI3K) or Src phosphorylation (data not shown). Inhibition of JNK phosphorylation with SP600125 upregulated by itself the basal levels of Axin2 expression. In agreement with the notion that estren activates Wnt signaling by actions upstream of β-catenin (Fig. 1B), treatment of calvarium-derived osteoblastic cells with estren resulted in the phosphorylation of GSK-3β at 2 h (Fig. 1E). However, it remains unclear whether inactivation of GSK-3β is the result of elevated Wnt levels or direct phosphorylation by ERKs, as it has been shown in other cell models (14).

Next, the ability of estren to upregulate Wnt signaling in vivo was examined using a transgenic “reporter” mouse expressing ubiquitously a TCF-β-Gal reporter construct in which three TCF sites upstream of the minimal c-fos promoter drive the expression of β-galactosidase (30). The results presented in Fig. 1D indicate that activation of Wnt signaling by estren requires ERKs. We have previously shown that activation of ERKs by estren or E₂ can be detected in bone in vivo 2 days following implantation of estren- or E₂-containing pellets into adult ovariectomized mice (40). The administered dose of either of the two compounds can prevent ovariectomy-induced bone loss (39). Based on these observations and the fact that transcriptional activation of Wnt/β-catenin targets can be efficiently detected 24 h following application of stimulus, estren was administered to ovariectomized TCF-β-Gal mice for 3 days. Estren, with this treatment regimen, stimulated β-galactosidase activity in long bones as well as in vertebrae (Fig. 2A). The effect of estren was similar to that of a dose of LiCl that increases bone mass in vivo (11). In a separate experiment, estren, similar to LiCl, increased expression of β-Gal reporter transcripts in the vertebrae of TCF-β-Gal mice (Fig. 2B). On the other hand, E₂ or DHT, at doses that prevent gonadectomy-induced bone loss in female or male mice, had no effect on TCF-β-Gal expression (39).

FIG. 2.

Estren, but not E₂ or DHT, stimulates Wnt/β-catenin-mediated transcription in bone in the TCF-β-Gal mouse. Six-month-old TCF-β-Gal transgenic mice were randomly assigned to groups of 4 to 6 animals. With the exception of one group, which was sham operated, all other animals were ovx, left untreated for 5 days, and then implanted with 21-day-slow-release pellets containing estren (2.6 mg), E₂ (0.01 mg), or DHT (3.5 mg). As a control, mice received daily intraperitoneal injections of LiCl at a dose of 200 mg/kg of body weight. Three days later, β-galactosidase activities (A) were determined in triplicate aliquots of lysates obtained from femurs and tibiae combined or vertebrae. (B) mRNA levels were determined in a single sample from each animal. Each bar indicates the mean ± standard deviation for values obtained from all the animals in each group. *, P was <0.05 for comparison with untreated ovx mice by ANOVA.

Induction of osteoblast differentiation by estren, but not E₂, DHT, or potential androgenic metabolites of estren, in uncommitted osteoblast progenitors and osteoblastic cells.

We then investigated the possibility that estren may promote the differentiation of uncommitted progenitors to the osteoblastic lineage. Estren, at 10⁻⁸ M, similar to BMP-2, induced differentiation of C2C12 cells to the osteoblastic lineage as determined by an increase in AP activity and osteocalcin protein levels (Fig. 3A and C) and also stimulated AP activity in primary cultures of calvarium-derived osteoblastic cells (Fig. 3D). In contrast, E₂ had no effect by itself on the osteoblastic differentiation of C2C12 cells (Fig. 3A and C), but it attenuated BMP-2-induced AP activity in C2C12 cells (Fig. 3E). Moreover, E₂ suppressed AP activity in murine bone marrow cell cultures maintained in the presence of ascorbic acid, an agent that induces osteoblast differentiation by facilitating the effects of endogenous BMPs (66). PPT, the ERα ligand that activates genotropic but not kinase-mediated actions of the receptor, was unable to stimulate AP activity (Fig. 3F).

FIG. 3.

Induction of AP and osteocalcin by estren, but not E₂ or DHT, in bi- or multipotential osteoblast progenitors and osteoblastic cells. C2C12 cells were pretreated with ICI 182,780 (10⁻⁷ M), flutamide (10⁻⁷ M), PP1 (10⁻⁶ M), wortmannin (WM; 30 nM), or SP600125 (SP6; 10 μM) (A and C) or noggin (10, 100, and 600 ng/ml) or human recombinant Dkk-1 (25, 50, and 100 ng/ml) (B) for 1 h. Vehicle, BMP-2 (300 ng/ml), E₂ (10⁻⁸ M), or estren (10⁻⁸ M) was added to the cells, and cultures were continued for 3 to 5 days. (D) Calvarium cells were treated with vehicle (veh), BMP-2 (100 ng/ml), E₂ (10⁻⁸ M), or estren (10⁻⁸ M) for 3 days. (E) C2C12 cells were treated with vehicle, BMP-2 (100 ng/ml), E₂ (10⁻⁸ M), or E₂ in the presence of BMP-2 for 3 days. Bone marrow cells were treated with vehicle, E₂ (10⁻⁸ M), or 50 μg/ml ascorbic acid (AA) in the presence or absence of E₂ for 14 days. C2C12 cells were treated with vehicle, BMP-2 (300 ng/ml), and estren (10⁻⁸ M) or PPT (F) or 10⁻¹² to 10⁻⁵ M of estren, E₂, DHT, 19-nortestosterone, or dihydronandrolone (G) for 3 days. AP activity in cell lysates and osteocalcin secreted in the medium were determined as described in Materials and Methods. ST2 cells (I) or MC3T3-E1 cells (H) were treated with vehicle, BMP-2 (300 ng/ml), E₂, or estren in the presence of ascorbic acid and β-glycerophosphate for 11 or 23 days, respectively. Mineralized matrix visualization and quantification were accomplished following staining with Alizarin red. Bars indicate means ± standard deviations for triplicate determinations. *, P was <0.05 for comparison with vehicle by ANOVA.

The estren-induced AP activity as well as osteocalcin production in C2C12 cells was blocked by antagonists of the ER or the AR (Fig. 3A and C) and inhibitors of Src, PI3K, and JNK kinases (Fig. 1A and C). Induction of AP activity by estren in C2C12 cells was completely abrogated by the BMP-2 antagonist noggin as well as Dkk-1 (Fig. 3B), indicating that both BMP-2 and Wnt signaling are required for its effects on mesenchymal cell commitment. These results, along with our previous observations that estrogens attenuate the self-renewal of transit-amplifying osteoblast progenitors in the bone marrow (13), are in full agreement with extensive experimental and clinical evidence that loss of estrogens increases both osteoclastogenesis and osteoblastogenesis and thereby the rate of bone turnover. Moreover, these findings strongly suggest that whereas kinase-mediated actions of the ER stimulate differentiation of uncommitted progenitors to the osteoblastic lineage, genotropic actions exert a diametrically opposite effect by interfering with the prodifferentiating actions of potent osteogenic agents.

Estren has 5% relative binding affinity for the AR compared to DHT (J. A. Katzenellenbogen, University of Illinois, personal communication), raising the possibility that it may be a weak androgen exerting its effects on bone via weak genotropic actions mediated by the androgen receptor. However, at concentrations as high as 4 to 5 orders of magnitude greater than the effective concentration of estren, E₂ and DHT as well as 19-nortestosterone and dihydronandrolone, which are potential androgenic metabolites of estren, had no effect on AP activity in C2C12 cells (Fig. 3G). These results strongly argue that the effects of estren on bone cannot result from a weak androgenic action. Further, estren, similar to BMP-2 but unlike E₂, induced mineralization in cultures of MC3T3-E1 preosteoblastic cells (Fig. 3H) and the pluripotent ST2 cells (Fig. 3I).

In agreement with its ability to increase AP activity and osteocalcin production, estren, but not E₂ or DHT, upregulated the expression of AP and osteocalcin genes in C2C12 cells (Fig. 4A). Similarly, estren, but not E₂, increased the expression of the osteoblast-specific transcription factor Runx2 and BMP-2 in MC3T3-E1 cells (Fig. 4B). As shown earlier (20, 21), the expression of endogenous BMP-2 in response to exogenous BMP-2 was initially stimulated (day 1) and then returned to basal levels (day 6).

FIG. 4.

Upregulation of the expression of osteoblast differentiation markers by estren but not E₂ or DHT. (A) C2C12 cells were treated with vehicle or 10⁻⁸ M E₂, DHT, or estren for 48 h. AP and osteocalcin expression was measured by real-time PCR. (B) MC3T3-E1 cells were treated with vehicle, BMP-2 (300 ng/ml), E₂ (10⁻⁸ M), or estren (10⁻⁸ M) for 1 or 6 days. Expression of Runx-2 and BMP-2 was assessed by real-time PCR. Bars indicate means ± standard deviations for triplicate determinations. *, P was <0.05 for comparison with vehicle by ANOVA.

Induction of Smad1/5/8 phosphorylation in uncommitted and committed osteoblastic cells is a property unique to estren and requires ERK, BMP-2, and Wnt signaling.

We next examined whether estren interacts with components of the BMP-2 signaling cascade. We found that estren stimulated the phosphorylation of Smad1/5/8 within 30 min in cultures of C2C12 (Fig. 5A) and calvarium (data not shown) cells. E₂, DHT, and 19-nortestosterone did not exhibit this property. PPT also had no effect on Smad 1/5/8 phosphorylation (Fig. 5B). Consistent with its inhibitory effect on AP activity, E₂, but not estren, attenuated the stimulatory action of BMP-2 on Smad phosphorylation (Fig. 5C). The induction of Smad1/5/8 phosphorylation in both cell types by estren was blocked by ICI 182,780, flutamide, noggin, and Dkk-1 as well as inhibitors of PI3K, ERK, or JNK activation, indicating that estren utilizes kinases as well as BMP and Wnt receptors to stimulate Smad phosphorylation. Estren also upregulated the expression of Smad6, a Smad1/5/8-specific transcriptional target, following 24 and 48 h of treatment of MC3T3-E1 cells (Fig. 5D). Neither E₂ nor DHT affected Smad6 expression. In fact, E₂, consistent with its antiosteoblastogenic properties and its suppressive effect on BMP-2-induced AP activity (Fig. 3E) and Smad phosphorylation (Fig. 5B), inhibited the stimulatory effect of BMP-2 on Smad6-Luc activity in UAMS32 stromal cells (18) transfected with the Smad6-Luc reported construct (Fig. 5E). Estren did not interfere with BMP-2-induced luciferase activity.

FIG. 5.

Estren, but not E₂, DHT, or potential androgenic metabolites of estren, induces Smad1/5/8 phosphorylation in C2C12 or calvarium cells via ERKs as well as BMP-2 and Wnt receptors. Following 4 h of serum starvation, C2C12 cells were treated with vehicle (V) or the indicated inhibitors at doses described in Fig. 2A and C or PD98059 (PD; 50 μM) (A) or vehicle or the indicated steriod (B) or vehicle, E₂, or estren (C). Thirty minutes later, BMP-2 (100 ng/ml) or 10⁻⁸ M of each of the different steroid compounds was added and the cultures were continued for 1 h or 30 min, respectively. (D) MC3T3-E1 cells were treated with vehicle (veh) or 10⁻⁸ M of E₂, DHT, or estren. Smad6 expression was quantified by real-time PCR at 24 and 48 h of treatment. (E) UAMS32 cells transiently transfected with a Smad6-Luc reporter were treated as indicated for 24 h. Bars indicate means ± standard deviations for triplicate determinations. *, P was <0.05 for comparison with vehicle by ANOVA. C2C12 cells were serum starved and pretreated with vehicle or the indicated inhibitors as described for panels A and B for 30 min prior to the addition of BMP-2 (100 ng/ml) for an additional 1 h (F) or Wnt3a, recombinant protein (50 ng/ml), for the indicated periods of time (G). In panels A to C and F to G, Smad1/5/8 phosphorylation was determined in cell lysates by Western blot analysis.

The requirement for ERK and Wnts in Smad activation was unique to estren, as BMP-2-induced Smad phosphorylation, as expected, was inhibited by noggin but was not affected by Dkk-1 or any of the kinase inhibitors (Fig. 5F). Treatment of C2C12 cells with Wnt3a protein for as long as 4 h did not induce Smad1/5/8 phosphorylation (Fig. 5G).

The increase in ERK phosphorylation by estren in C2C12 cells was attenuated not only by ICI 182,780 and flutamide but also by Dkk-1 and noggin (Fig. 6A). BMP-2 also induced ERK phosphorylation (Fig. 6B). In contrast to what was found for estren, this effect was not inhibited by ICI 182,780, flutamide, or Dkk-1 but was further enhanced by noggin. Noggin by itself had no effect on ERK phosphorylation (data not shown). These results indicate that in addition to Smad phosphorylation, ERK activation is another point where kinases, BMP, and Wnt signaling converge in response to estren.

FIG. 6.

BMP-2 and Wnt signaling are required for the induction of ERK phosphorylation by estren, but not by BMP-2, in C2C12 cells. Cells were cultured in the absence of serum for 4 h and were then treated with vehicle (V) or the following inhibitors: ICI 182,780 (ICI; 10⁻⁷ M), flutamide (Flu; 10⁻⁷ M), human recombinant Dkk-1 (25 ng/ml), noggin (100 ng/ml), or PD98059 (PD; 50 μM) for 30 min prior to the addition of estren (10⁻⁸ M) (A) or BMP-2 (100 ng/ml) (B) for an additional 5 min. Cells were harvested, and the status of ERK phosphorylation was determined by Western blot analysis.

Simultaneous activation of the ER and AR is not sufficient to induce osteoblast commitment and differentiation.

We also investigated whether the prodifferentiating effects of estren result from its ability to bind and activate both the ER and the AR. For this purpose, we examined whether simultaneous activation of the two receptors with hermaphrodiol, a ligand of the ER and AR, or cotreatment of C2C12 cells with E₂ and DHT would reproduce the actions of estren on Smad1/5/8 phosphorylation, AP activity, or osteocalcin production. Unlike estren, neither hermaphrodiol nor a combination of E₂ and DHT had an effect on Smad1/5/8 phosphorylation (Fig. 7A), AP activity (Fig. 7B), or osteocalcin production (Fig. 7C).

FIG. 7.

Simultaneous activation of the ER and AR is not sufficient to induce Smad1/5/8 phosphorylation, AP activity, or osteocalcin production. (A) C2C12 cells were cultured in the absence of serum for 4 h. Cells were then treated with BMP-2 (100 ng/ml) for 1 h, 10⁻⁸ M hermaphrodiol for the indicated periods of time, or a combination of E₂ (10⁻⁸ M) and DHT (10⁻⁸ M) for 30 min. Cells were then harvested, and Smad1/5/8 activation was determined in cell lysates by Western blot analysis. V, vehicle. (B and C) C2C12 cells were treated with vehicle (veh), BMP-2 (100 ng/ml), estren (10⁻⁸ M), the indicated doses of hermaphrodiol, and a combination of E₂ (10⁻⁸ M) and DHT (10⁻⁸ M). Three or 5 days later, AP activity (B) or osteocalcin production (C) was determined as described in the legend to Fig. 3. Bars indicate means ± standard deviations for triplicate determinations. *, P was <0.05 for comparison with vehicle by ANOVA.

Stimulation of Wnt and BMP-2-like signaling by EDC.

In addition, we examined whether the prodifferentiating actions of estren were unique to this compound or could be mimicked by other ligands that preferentially activate kinase- as opposed to ERE-mediated actions of the ER. For this purpose, we used EDC, the non-cell-permeable estrogen conjugate which stimulates Shc/Src/ERK signaling in MCF-7 breast cancer cells at low concentrations but is approximately 10,000-fold less potent than E₂ in stimulating transcription of endogenous direct-estrogen-target genes (28). Similar to estren, but in contrast to E₂, EDC dose dependently upregulated Wnt/β-catenin-mediated transcription in C2C12 cells transiently transfected with the TCF-Luc reporter construct (Fig. 8A). Moreover, EDC (10⁻⁸ M) induced Smad1/5/8 phosphorylation within 30 min (Fig. 8B) and stimulated AP activity in C2C12 cells in a dose-dependent manner and with a potency similar to that of estren (Fig. 8C). An EDC control compound that does not contain the estrogen molecules had no effect in any of these end points.

FIG. 8.

Prodifferentiating actions of the EDC compound in C2C12 cells. (A) C2C12 cells were transiently transfected with TCF-Luc. Twenty-four hours following transfection, cells were treated with vehicle (veh), 50 ng/ml recombinant Wnt3a protein, or 10⁻¹² to 10⁻⁶ M of EDC, EDC control, or estren for an additional 24 h. Luciferase activity was measured in cell lysates. TCF-Luc values were normalized with Renilla-Luc values for transfection efficiency. Normalized data are expressed as relative luciferase units (RLU). (B) C2C12 cells were cultured in the absence of serum for 4 h. Cells were then treated with BMP-2 (100 ng/ml) for 1 h or 10⁻⁸ M of the indicated compounds for 30 min. Cells were then harvested, and Smad1/5/8 activation was determined in cell lysates by Western blot analysis. (C) C2C12 cells were treated with vehicle, BMP-2 (100 ng/ml), or 10⁻¹² to 10⁻⁶ M of EDC, EDC control, or estren. Three days later, AP activity was determined as described in the legend to Fig. 3. Bars indicate means ± standard deviations for triplicate determinations. *, P was <0.05 for comparison with vehicle by ANOVA.

Elimination of ERE-mediated transcription unleashes the ability of E₂ to stimulate Smad1/5/8 phosphorylation in vivo.

Finally, to test directly the hypothesis that ERE-mediated effects may be counteracting the ability of E₂ to elicit prodifferentiating actions similar to those of estren, we used a knock-in mouse model in which one endogenous ERα allele had been replaced by a mutant form of ERα which lacks DNA binding activity and classical ERE-mediated transcription (ERα^NERKI/+) (35). ERα^NERKI/+ animals were crossed with a heterozygote knockout mouse model of the ERα gene (15) to produce ERα^NERKI/− mice in which ERα should signal only through non-ERE pathways. Compared to 3-month-old wt control littermates, ERα^NERKI/− mice had atrophic uteri (weighing 4-fold less) and much lower levels of mRNA expression (20-fold) of the ERE-dependent complement 3 (C3) gene (Fig. 9A). Ovx reduced uterine weight and C3 expression in the wt control but not in the ERα^NERKI/−. E₂ replacement administered by daily subcutaneous injections (30 ng/g) for 5 days was effective in restoring uterine weight and C3 expression in the wt control but had a minimal effect in the ovx ERα^NERKI/− mice. However, 1 h following injection of E₂ into ovx wt controls or ERα^NERKI/− mice, we detected an increase in the phosphorylation of ERK and the ERK-regulated transcription factor Elk-1 in vertebral lysates from both types of mice (Fig. 9B). On the other hand, 1 h of treatment of ovx ERα^NERKI/− mice with E₂ induced phosphorylation of Smad1/5/8, but this effect was absent in the wild-type controls (Fig. 9C). Consistent with this, E₂ was as effective as estren and the EDC compound in stimulating Smad1/5/8 phosphorylation within 30 min of treatment in calvarium cells derived from ERα^NERKI/− but not wild-type littermates (Fig. 9D). Estren and the EDC compound induced Smad1/5/8 phosphorylation not only in ERα^NERKI/− but also in wild-type-control-derived calvarium cells.

FIG. 9.

Uterine weight is decreased but ERK/Elk-1 and Smad1/5/8 activities are stimulated in response to E₂ in ERα^NERKI/− mice. (A) Six-month-old wild-type control mice were left intact or were ovx, while ERα^NERKI/− mice were either sham operated or ovx. Ovx animals were left untreated or received an E₂ replacement administered by daily subcutaneous injections (30 ng/g) for 5 days. C3 mRNA expression in the uterus was assessed by real-time PCR (n = 4). Each bar indicates the mean ± standard deviation for all the animals in each group. *, P was <0.05 for comparison with untreated ovx mice by ANOVA. (B and C) Six-month-old ERα^NERKI/− or wild-type control mice were sham operated or ovx. Ovx animals were left untreated for 5 days and were then injected subcutaneously with vehicle or E₂ replacement (30 ng/g). One hour later, L5 was obtained and lysates were made for determination of ERK, Elk-1 (B), and Smad1/5/8 (C) phosphorylation by Western blot analysis. Each lane represents one mouse. In panels B and C, blots were developed by chemiluminescence and the intensities of the bands were quantified with an imaging system. Bars indicate means ± standard deviations for all animals in each group. , P was <0.05 for comparison with untreated ovx mice by ANOVA. (D) Calvarium cells isolated from ERα^NERKI/− or wild-type control mice were cultured in the absence of serum for 4 h. Cells were then treated with BMP-2 (100 ng/ml) for 1 h or 10⁻⁸ M of the indicated compounds for 30 min. Smad1/5/8 activation was determined in cell lysates by Western blot analysis. veh, vehicle.

DISCUSSION

The results reported in this paper demonstrate that estren, but not classical estrogens or androgens, induces osteoblast differentiation in a variety of established cell lines of uncommitted or committed osteoblast progenitors as well as in primary cultures of bone marrow and calvarium cells. These effects result from an extranuclear function of the ER, causing activation of ERKs and downstream potentiation of both the BMP and the Wnt signaling cascades. The Src, PI3K, and JNK kinases may also be involved, but this has not been shown directly in this work; therefore, future studies will be required to establish their role in the phenomena described here. Evidently, the effects of estren resulted from an extranuclear function of the ER independent of direct ER/DNA interaction, as evidenced by the fact that EDC, an estradiol-dendrimer conjugate that cannot enter the cell, produced effects identical to those of estren on Wnt transcription, Smad phosphorylation, and differentiation of uncommitted osteoblast progenitors.

The effect of estren on Smad phosphorylation required ERK signaling. Likewise, estren stimulated Wnt/β-catenin-mediated transcription, at least in part, via ERK-dependent phosphorylation and inactivation of GSK-3β. However, ERK phosphorylation by estren required both Smad and Wnt signaling. In addition, Smad phosphorylation was mediated not only via ERK but also via Wnt activation. Collectively, these observations suggest that kinase-initiated actions of the ER, in the absence of counteracting effects from direct interactions of the receptors with DNA, induce the assembly of a large signalosome in which inputs from the ER (or AR), BMPs, Wnt signaling, and kinases converge (Fig. 10). The data of the present report also suggest that estren upregulates the expression of BMP-2, as well as Wnt family members and receptors, either via ERK-dependent transcription or via Smad- or β-catenin-mediated transcription. Hence, increased levels of BMP-2 and Wnts may in turn promote the activation of this signalosome.

FIG. 10.

Potentiation of an ER, ERK, BMP, and Wnt signalosome by kinase-initiated signaling of the ER. Estren rapidly activates ERKs and ERK-mediated transcription by actions of the estrogen receptor. Estren also rapidly induces ERK-dependent Smad phosphorylation and stimulates Wnt/β-catenin-mediated transcription, probably by ERK-dependent inactivation of GSK-3β. In parallel, ERK phosphorylation requires both Smad and Wnt signaling, and Smad phosphorylation is mediated not only by ERK but also by Wnt activation. These observations suggest the existence of a large signalosome in which inputs from steroid receptors, BMPs, Wnt signaling, and kinases converge. Finally, estren upregulates the expression of BMP-2, as well as Wnt family members, either via ERK-dependent transcription or via Smad- or β-catenin-mediated transcription. Increased levels of BMP-2 and Wnts may in turn promote the activation of the signalosome. SRE, serum response element; TFs, transcription factors; APC, adenomatosis polyposis coli.

In line with the evidence suggesting that estren's ability to induce differentiation of uncommitted osteoblast progenitors resulted from the convergence of kinase (Src, PI3K, and JNK), BMP, and Wnt signaling, induction of osteoblast commitment and differentiation of C2C12 and other osteoblast precursor cells by BMP-2 involves modulation of cytosolic kinases, such as PI3K, p38, and ERKs (19, 44, 51, 59, 66). ERKs, PI3K, and protein kinase C have been also implicated in Wnt signaling. Indeed, several reports have suggested that these three kinases may regulate or cooperate with components of the Wnt signaling, such as the Frizzled receptors, GSK-3β, or β-catenin, in mammalian cells as well as in Caenorhabditis elegans and in Drosophila spp. (10, 14, 17, 32). Furthermore, there is considerable evidence that BMPs and the LRP5/Wnt signaling cascades cooperate in the induction of osteoblastic cell differentiation and maturation by controlling different stages of the process (36, 36, 42, 55, 63). Specifically, it has been shown that LRP5/Wnt signaling acting via β-catenin mediates BMP-2-induced osteoblast differentiation (3, 55) and that β-catenin and TCF can form a complex with Smad4 (52).

In contrast to the regulation of bone cell survival, which is a property of both estren and classical ER ligands, induction of osteoblast differentiation is a property of estren or EDC, each a compound which selectively activates kinase-initiated actions of the ER, but not a property of classical estrogens. In fact, the prodifferentiating effects of estren and EDC on osteoblast precursors are opposite to the ability of E₂ to attenuate the self-renewal of transit-amplifying osteoblast progenitors in the bone marrow (13) and also to attenuate BMP-2 (Fig. 3E and 5C and E)- or parathyroid hormone (PTH)-induced osteoblast differentiation (27a). Moreover, the prodifferentiating actions of estren cannot be reproduced by the combination of classical ER and AR ligands or by PPT, the compound that can activate only the genotropic function of the ER. This striking dichotomy between the effects of ligands that selectively activate kinases and those of ligands that activate both kinases and genomic actions raises the possibility that at least some responses of target cells to sex steroids may be the result of a balance between nongenotropic and genotropic actions. Removal of genotropic counterregulatory actions on Wnts, Wnt receptors or antagonists, BMPs, and/or regulators of BMP signaling may “unleash” the differentiation process. In support of this hypothesis, we found that a single injection of estradiol into mice in which the ERα lacks DNA binding activity and classical ERE-mediated transcription (ERα^NERKI/−) induced Smad1/5/8 phosphorylation within 1 h, but this effect could not be seen in wild-type controls, in which the ERα is capable of both ERE-dependent and -independent actions. Moreover, there is evidence that ERE-mediated actions of the ER are enhanced by physical association of the receptor with β-catenin, and ER/β-catenin association attenuates Wnt/β-catenin-mediated transcription (50).

Others have suggested that the effects of estren may result from its ability to weakly activate classical genotropic actions of the AR (7, 31, 41, 49, 65), perhaps through conversion to the androgenic metabolite 19-nortestosterone (7, 34, 41, 49), or that estren is a weak genotropic ER ligand (31, 49). In contrast to these suggestions, we found that E₂, DHT, dihydronandrolone, and 19-nortestosterone could not reproduce the effects of estren on differentiation of uncommitted osteoblast precursors even at concentrations 5 orders of magnitude higher. In line with our findings, White and colleagues have found that estren is a weak regulator of AR-mediated transcription but a moderate promoter of Xenopus oocyte maturation, a process that involves nongenotropic actions of androgens (45, 62). In contrast, 19-nortestosterone is a highly potent regulator of AR-mediated transcription but a very weak activator of Xenopus oocyte maturation. Moreover, a recent study of the effects of estren on the murine uterus revealed that the transcriptional profile of estren is indeed distinct from that of DHT or 19-nortestosterone (31). In the same study, a dose of estren threefold higher than the dose that we have previously reported to exert a beneficial effect on bone (39) produced only 50% of the effect of E₂ on the uterine weights of adult ovx mice. Yet, under the same conditions, estren stimulated the expression of rapid, presumably kinase-regulated, response genes as robustly as E₂ but was a weak activator of the expression of late, ER/DNA-regulated response genes.

In full agreement with the relative binding affinities of estren for the ER (300-fold lower than that of estradiol) (39) and the AR (approximately 1/40 that of R1881 or 1/25 that of DHT), we and others have found that at very high concentrations, for example, 3,000- to 7,000-fold higher than the replacement dose of estradiol, estren may also exert genotropic activity (data not shown). Indeed, using estren at doses 100-, 300-, 1,000-, and 3,000-fold higher than the replacement dose of estradiol, we have confirmed the lack of an effect on the uteri of 6- to 8-month-old mice at the dose that we had used in our published studies. Nonetheless, consistent with its affinity for the ER, at a dose that exceeded by 10-fold the K_d for the ER (3,000-fold higher than the replacement dose of E₂), estren restored uterine weight. Similarly, in a 1-week experiment with rats, at a dose 3,000-fold higher than the E₂ replacement dose, estren exhibited 30% of the effect of E₂ on uterine weight.

Moverare et al. have reported an effect of estren on the uteri of 11-month-old mice, which was one-fifth that of E₂, as well as a small effect on bone (49). In that study, the authors also found that estren had a transcriptional activity in kidney cells transfected with ER; however, the transcriptional effect occurred at concentrations which were 3 orders of magnitude higher than that of 17β-estradiol and even 1 order of magnitude higher than that of 17α-estradiol, the biologically inactive isomer of 17β-estradiol (4). These findings are consistent with the contention that the weak reproductive actions of estren may indeed reflect transcriptional activity corresponding to its weak affinity for the ER.

Most importantly, in another report, Gallea et al. (19) had found that subcutaneous injections of 4-estren-3β,17β-diol, the enantiomer of estren, also had a potent effect on bone but, in contrast to treatment with estren, induced only very moderate increases in seminal vesicle and uterine weight compared to what was found for gonadectomized control mice. However, when given by pellets, it did have an effect on uterine weight and also prevented bone loss. This observation confirms our essential finding that estren has favorable effects on bone without affecting reproductive organs, at least under certain experimental conditions. The ages and sizes of the animals, different routes of estren administration (daily subcutaneous injections versus pellet implantation), and different doses of E₂ and/or estren, compared to the ones used in our experiments, may explain the discrepancies between different studies (46a).

Collective evidence from us and others has led us to propose that estren (i) exerts its effects as a result of the selective activation of kinase-mediated actions of the ER or AR; (ii) has a biologic profile that is distinct from that of estradiol, SERMs, or androgens; and (iii) exerts minimal or blunted effects on reproductive organs under conditions in which it exhibits a desirable effect on bone. Based on these lines of evidence, we had suggested earlier that estren represents the prototype of a function-specific ER/AR ligand with unique biologic properties not exhibited by the natural ligands of these receptors. Three significant advances argue in favor of this contention. First, cell-impermeable estrogen conjugates produce effects identical to those of estren, while they are ineffective in stimulating transcription of endogenous ERE-containing genes (54). Second, screening of focused libraries has led to the identification of additional synthetic compounds that, like estren, can dissociate kinase- from ERE-mediated transcription, even though, unlike estren, they have affinities for ER comparable to that of estradiol and no AR binding capability (60). Thus, the function-dissociating properties of estren and other such compounds cannot be due to weak estrogenic activity or binding to AR. Instead, as we had proposed originally, these compounds most likely induce a distinct conformation of the receptor protein in such a manner that they can initiate kinase cascades but not classical cis- or trans-mediated transcription. Third, function-specific ligands of the ER other than estren have been identified, and they also lack uterotropic effects but retain other nonreproductive actions, such as immunomodulation (8) and ischemic neuroprotection (56). Hence, selective activators of ER-mediated kinase cascade activation and perhaps activators of nongenotropic signals of other nuclear receptors may represent a novel class of pharmacotherapeutics with the potential for biologic outcomes distinct from those produced with natural ligands that activate genotropic and nongenotropic signals alike or synthetic ligands that activate only the former.

In studies not described here, we have compared the effects of estren to those of intermittent PTH, a proven bone anabolic agent, on the skeletons of mice and rats and have been unable to show that estren has bone anabolic properties comparable to those of PTH, as determined by measuring bone mineral density or microcomputed tomography. Whether the apparent inability of estren to manifest in vivo a bone anabolic effect, consistent with its prodifferentiation effects on osteoblast progenitors in vitro, is due to its more potent antiremodeling/antiresorptive action, which overwhelms or conceals its putative bone-forming property, or is due to unknown pharmacokinetic shortcomings, a suboptimal route or even frequency of administration will, of course, require extensive additional work. Nonetheless, our observations and evidence in the literature (8, 28, 38, 39, 54, 56, 60) continue to support our proposition that selective activators of ER-mediated kinase cascade activation and perhaps activators of nongenotropic signals of other nuclear receptors may represent a novel class of pharmacotherapeutics with the potential for biologic outcomes distinct from those produced with natural ligands that activate genotropic and nongenotropic signals alike or synthetic ligands that activate only the former. Irrespective of whether the window of this therapeutic opportunity for a beneficial effect on bone without adverse effects on reproductive organs is small or large, the existing evidence from the work with these novel compounds supports the contention that there is indeed such a window.

During the preparation of our manuscript, Cvoro and colleagues reported a heretofore unknown mechanism of transcriptional repression by estrogens (12). Specifically, they showed that in human osteoblastic cells, among others, tumor necrosis factor alpha (TNF-α) induces the assembly of a transcriptional activation complex at the TNF-α promoter in which unliganded ERα acts as a potent coactivator of transcription. Remarkably, in the presence of E₂, the ligand-independent activation of transcription by ERα is reversed and ERα becomes a repressor of TNF-α-induced gene expression by recruiting GRIP1. In agreement with the findings of Cvoro et al., the results presented herein indicate that in the presence of E₂, the ER suppresses the prodifferentiating actions of BMP-2 by inhibiting BMP-2-induced activation of Smads and Smad-mediated transcription as well as BMP-2-induced upregulation of AP activity. On the other hand, in preliminary studies reported elsewhere (2a), we have found that ICI 182,780, which causes ER degradation (16), attenuates BMP-2-induced alkaline phosphatase activity in C2C12 as well as 2T3 osteoblastic cells. We are, therefore, tempted to speculate that unliganded ERα promotes the prodifferentiating actions of BMP-2 by interacting with components of the BMP-2 or Wnt signaling pathways, while the liganded ER has the opposite effect. Hence, at least part of the inhibitory actions of E₂ on osteoblastogenesis may result from a repressive effect of the hormone on the differentiation-promoting actions of the unliganded receptor. Future studies will, of course, be required to test this hypothesis.