This Article Abstract Full Text (PDF) All Versions of this Article: 145/7/3507    most recent Author Manuscript (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Wiren, K. M. Articles by Jepsen, K. J. Search for Related Content PubMed PubMed Citation Articles by Wiren, K. M. Articles by Jepsen, K. J.

Targeted Overexpression of Androgen Receptor in Osteoblasts: Unexpected Complex Bone Phenotype in Growing Animals

Bone and Mineral Research Unit (K.M.W., X.-W.Z., A.R.T.), Portland Veterans Affairs Medical Center; and Departments of Medicine (K.M.W.) and Behavioral Neuroscience (K.M.W., X.-W.Z.), Oregon Health & Science University, Portland, Oregon 97239; Department of Molecular Endocrinology/Bone Biology (V.K., M.A.G.), Merck Research Laboratories, West Point, Pennsylvania 19486; and Department of Orthopaedics (K.J.J.), Mt. Sinai School of Medicine, New York, New York 10029

Address all correspondence and requests for reprints to: Kristine Wiren, Ph.D., Portland Veterans Affairs Medical Center P3-R&D39, 3710 Southwest Veterans Hospital Road, Portland, Oregon 97239. E-mail: wirenk{at}ohsu.edu.

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
The androgen receptor (AR), as a classic steroid receptor, generally mediates biologic responses to androgens. In bone tissue, both AR and the estrogen receptor (ER) are expressed in a variety of cell types. Because androgens can be converted into estrogen via aromatase activity, the specific role of the AR in maintenance of skeletal homoeostasis remains controversial. The goal of this study was to use skeletally targeted overexpression of AR as a means of elucidating the specific role(s) for AR transactivation in bone homeostasis. Rat AR cDNA was cloned downstream of a 3.6-kb 1(I)-collagen promoter fragment and used to create AR-transgenic mice. AR-transgenic males gain less weight and body and femur length is shorter than wild-type controls, whereas females are not different. AR-transgenic males also demonstrate thickened calvaria and increased periosteal but reduced endosteal labeling by fluorescent labeling and reduced osteocalcin levels. High-resolution micro-computed tomography shows normal mineral content in both male and female AR-transgenic mice, but male AR-transgenics reveal a reduction in cortical area and moment of inertia. Male AR-transgenics also demonstrate an altered trabecular morphology with bulging at the metaphysis. Histomorphometric analysis of trabecular bone parameters confirmed the increased bone volume comprised of more trabeculae that are closer together but not thicker. Biomechanical analysis of the skeletal phenotype demonstrate reduced stiffness, maximum load, post-yield deflection, and work-to-failure in male AR-transgenic mice. Steady-state levels of selected osteoblastic and osteoclastic genes are reduced in tibia from both male and female transgenics, with the exception of increased osteoprotegerin expression in male AR-transgenic mice. These results indicate that AR action is important in the development of a sexually dimorphic skeleton and argue for a direct role for androgen transactivation of AR in osteoblasts in modulating skeletal development and homeostasis.

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
THE MOLECULAR PATHWAYS controlling bone formation in either normal individuals or in pathophysiologic disease states are not well understood. This question is of significant importance because osteoporosis, a low bone mass disease associated with an increased risk of fracture, is the most prevalent degenerative disease in developed countries (1). Osteoporosis is generally characterized by a relative decrease in bone formation (mediated by osteoblasts) vs. bone resorption (mediated by osteoclasts) and is often coupled with a hypogonadal state in both men (2) and women. Although both estrogen and androgen circulate in both genders, the influence of estrogen and androgen on the skeleton is distinct as shown by divergent responses to gonadectomy in either gender (3), particularly with respect to bone size and periosteal apposition. Furthermore, combination therapy with estrogen and androgen in postmenopausal women is more beneficial than either steroid alone (4, 5, 6), indicating nonparallel pathways of action. Estrogens are thought to act to maintain adult bone mass predominantly through an inhibition of bone resorption by the osteoclast, i.e. as antiresorptive agents, which protect the skeleton from further loss of bone. Nonaromatizable androgens such as 5-dihydrotestosterone (DHT), on the other hand, are characterized as anabolic agents that increase bone mass by stimulation of bone formation (7, 8), and thus represent an important therapeutic class that may have the potential to rebuild lost bone. However, an understanding of the pathways influenced by androgen in the osteoblast is currently very limited.

In general terms, the skeletal response to systemic androgen therapy has been characterized as increased trabecular and cortical bone mass, increased bone width with surface periosteal expansion and a lack of inner endosteal deposition. This is observed in the setting of inhibition of resorption due to reduced osteoclast activity, that may not be as significant as that seen with estrogen replacement (see Ref. 9). There has been speculation that the positive effects of androgens on the skeleton may be mediated indirectly through increased muscle mass in biomechanical linkage, thought to have beneficial effects on bone density. However, analysis of the myostatin null mouse, the so called "Mighty mouse," with dramatic hypertrophy of muscle that shows no difference in bone at the cortical midshaft, suggests this may not entirely be the case (10). In addition, fat mass, not lean mass, is better associated with improved bone mineral density (11). It has also been proposed that sex steroids can act nonspecifically through nongenomic actions at either ER or AR (12, 13), although recent data suggest that genomic signaling may be the more significant regulator in vivo (14, 15).

Analysis of AR signaling in vivo has been approached genetically both with global receptor knockouts (16, 17) and with the testicular feminization model (18). The global loss of AR results in high-turnover osteopenia and reduced trabecular bone volume, with a significant stimulatory effect on osteoclast function. Experimental strategies such as surgical or pharmacological intervention have also been employed to characterize androgen signaling. Distinct effects of androgen are seen with gonadectomy when comparing the effects of orchidectomy in male (ORX) vs. ovariectomy in the female (OVX) rats. OVX in the female results in decreased trabecular area in the metaphysis with increased osteoclast number and an increase in serum calcium, but in cortical bone at the diaphysis, an increase at the periosteal surface with circumferential enlargement but a decrease in endosteal labeling (3). These results suggest that estrogen protects trabecular bone predominantly through inhibition of osteoclast activity/recruitment, and exhibits an inhibitory action at the periosteal surface. In the male, ORX also resulted in decreased trabecular area in the metaphysis with increased osteoclast number, resulting in trabecular osteopenia in the secondary spongiosa (19). In contrast with the female, cortical bone periosteal formation was reduced. Replacement with nonaromatizable DHT pellets for 3 wk in the male prevented the loss of trabecular bone (19). In the intact animal, the stimulation of endosteal formation by estrogen compensates for the lack of periosteal formation, thus leading to no difference in biomechanical strength (20).

Although these approaches have advanced our understanding of steroid action in bone, they have limitations. For example, surgical removal of gonadal tissue (i.e. castration) to eliminate hormone production or pharmacological treatment with a receptor antagonist such as hydroxyflutamide may affect multiple organ systems and result in secondary effects. Global knockout of AR necessarily results in loss of AR function in all tissues, which can have developmental consequences and indirect effects on bone balance. Additional confusion regarding the specific action of androgens on the skeleton results from the fact that testosterone can be metabolized to estrogen via aromatase activity. Thus, some androgen action may result from ER-dependent activation after conversion to 17ß-estradiol. For these reasons, creation of transgenic lines with bone-targeted overexpression of AR should enhance our understanding of the specific role for androgen through AR transactivation in skeletal tissue. Our data show that AR overexpression in the osteoblast lineage has dramatic effect on bone quality and trabecular morphology in male AR-transgenic mice, and argue for a direct role for androgen transactivation of AR in osteoblasts in modulating skeletal development and homeostasis.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Cloning of expression plasmids
The pBR327-based plasmid col3.6-ßgal-ClaPa containing the rat type I ₁ collagen (ColI1) promoter sequence –3518 to +115, served as the starting vector (generously provided by Dr. David Rowe, University of Connecticut Health Center, Farmington, CT). The BamHI site at –3145 in col3.6-ßgal-ClaPa was removed by introducing a point mutation using in vitro mutagenesis to create the modified col3.6E-ßgal-ClaPa. Briefly, a 50-µl thermal cycling elongation reaction consisted of 50 ng col3.6-ßgal-ClaPa plasmid, 0.4 µM primers each, 200 µM deoxynucleotide triphosphate, and 2.5 U PfuTurbo DNA high-fidelity polymerase. The primers used contain a point mutation (bold/italicized letters) that destroys the BamHI site (forward: 5'-CACCACACACCTAGGACCCACCCACAGATTTTGC-3' and reverse: 5'-GCAAAATCTGTGGGTGGGTCCTAGGTGTGTGGGTG-3'). The reaction was 94 C for 45 sec, 55 C for 45 sec, and 68 C for 20 min for 20 cycles. The combination of low cycle number and high-fidelity polymerase minimizes unwanted mutations. After the reaction, 10 U DpnI was added for 1 h at 37 C. DpnI is specific for methylated and hemimethylated DNA, and selectively digests the parental and hybrid plasmid DNAs. Positive clones were identified by the BamHI restriction pattern. To add BamHI sites to the rat AR cDNA (provided by Dr. Shutsung Liao, University of Chicago, Chicago, IL), PCR primers were designed with BamHI ends as follows: forward 5'-GGATCCATGGAGGTGCAGTTAGGGCT-3', reverse 5'-GGATCCTCACTGTGTGGAAATAGA-3'. PCR conditions were 94 C for 30 sec, 55 C for 30 sec, and 68 C for 3 min for 30 cycles with 2.6 U Expand High Fidelity PCR Systemenzyme mix (Roche Applied Science, Indianapolis, IN) in 50 µl reaction. The PCR product was T/A cloned in pCR 2.1-TOPO vector (Invitrogen Life Technologies, Carlsbad, CA). The BamHI-rAR was released by digestion with BamHI. Finally the BamHI-rAR fragment was cloned into the BamHI site in the modified col3.6E-ßgal-ClaPa (after removal of the ßgal cDNA sequences), to give the expression construct termed col3.6-AR (hereafter referred to as colAR) shown in Fig. 1A. The correct sequence and orientation of the AR insert was verified by direct DNA sequencing.

View larger version (41K): [in this window] [in a new window]   FIG. 1. Generation of transgenic mice with bone-targeted AR overexpression. A, A schematic representation of the col3.6 AR transgene. B, Southern analysis for characterization of transgenic animals. Genomic DNA was isolated from both AR-transgenic founder lines (104 and 106), digested with EcoRV, PstI, or SalI and subjected to Southern blot analysis after hybridization with the colAR transgene product. Analysis indicates a single insertion site. C, Copy number of the AR transgene (tg-AR) relative to the endogenous AR (endog-AR) gene was estimated by real-time PCR analysis of genomic DNA (5–7 for both lines). bGH, Bovine GH; PA, polyadenylation signal.

DNA extraction, Southern analysis, and PCR genotyping
Genomic DNA was purified from a small piece of tail tissue obtained at the time of weaning using a standard proteinase K and phenol-chloroform extraction procedure. Genotyping was performed by PCR analysis using colAR-GT primers forward 5'-TAGCACCTCTGGCCCATGTA-3' and reverse 5'-TCCTGCCGCTGCTGTAAACA-3'. These primers were designed to specifically amplify the transgene by including part of the rat collagen sequence in the ClaPa constructs and part of the AR sequence as shown in Fig. 1A. Primers were derived using OLIGO Software from Molecular Biology Insights, Inc. (Cascade, CO) and purchased from Fisher Scientific (Pittsburgh, PA). A single insertion site for the transgene was confirmed by Southern blot using the PCR fragment as a probe. Copy number was determined by real-time PCR analysis using genomic DNA employing colAR-RT primers described in real-time RT-PCR analysis and primers for the endogenous mouse AR gene forward 5'-GGAATTCGGTGGAAGCTACA-3' and reverse 5'-CCGGGAGGTGCTATGT-3'. Eight mice were found to be positive for the presence of the transgene, and founders of both genders were mated to B6D2F1 mice to establish transgenic lines. Results are presented using two independent AR-transgenic lines (104 and 106) with moderate copy number and single integration site.

Animals
AR-transgenic mice were bred to B6D2F1 mice (The Jackson Laboratory); both genders were employed. The litters were housed with the dam until weaning at 21 ± 2 d of age, at which time they were housed three to four per cage in isosexual groups with mice of the same genotype. The mice had free access to tap water and were fed a diet containing 1.14% calcium, 0.8% phosphorous, 2200 IU/kg vitamin D3, 6.2% fat, and 25% protein (Purina PMI Nutrition International, St. Louis, MO). All animals were weighed weekly, and body length (nose to rump) was determined at weekly or monthly intervals, respectively, over 6 months (n = 4–5).

The animals were killed under CO₂ narcosis by decapitation. Before killing, 8-wk-old mice received two fluorochrome labels by ip injection for evaluation of bone dynamics [oxytetracycline hydrochloride (Sigma, St. Louis, MO) at 30 mg/kg 10 d before killing and calcein green (Sigma) at 10 mg/kg 3 d before killing]. The calvaria and femora were dissected and cleaned from surrounding tissue. The left femur was immersed in 4% paraformaldehyde fixative for 24 h at 4 C and subsequently kept in 100% ethanol until histomorphometric analysis. The right femur was used for measurement of cortical and trabecular volumetric density and geometry by micro-computed tomography (micro-CT) ex vivo, followed by destructive analysis of whole bone biomechanical properties described below. The length of the femur was measured from the femoral head to the distal condyles. In addition, calvaria, liver, kidney, tendon, ear, muscle, skin, heart, fat, and spleen were collected for RNA isolation or immunocytochemical analyses. For RNA isolation, tibia was cleaned of muscle tissue and aseptically dissected. After removal of the epiphyseal area, marrow is briefly flushed with sterile saline and the bone frozen in liquid nitrogen and stored at –80 C until RNA isolation as described below.

Real-time RT-PCR analysis
Analysis of colAR transgene expression in different tissues was performed with the iCycler IQ Real Time PCR detection system (Bio-Rad Laboratories, Inc., Hercules, CA) using a one-step procedure on deoxyribonuclease (DNase)-treated total RNA (see Table 1). RNA was isolated using the RNA Stat-60 kit (Tel-Test, Inc., Friendswood, TX). Contaminating DNA was removed by RQ1-DNase (Promega, Madison, WI) digestion and phenol-chloroform extraction or by Zymo-spin column purification following manufacturer’s recommendations (Zymo Research, Orange, CA). Twenty nanograms of RNA were reverse transcribed and amplified in a 25-µl reaction mix containing 1x QuantiTect SYBR Green RT-PCR Master Mix (QIAGEN, Valencia, CA) and 0.5 µM each primer. ColAR-RT primers were 5'-GCATGAGCCGAAG-CTAAC-3' and 5'-GAACGCTCCTCGATAGGTCTTG-3' and specifically amplified colAR using sites in the collagen untranslated region and AR near to those used for genotyping (Fig. 1A). After PCR, reaction products were melted over the temperature range 55–95 C in 0.5 C increments, 10 sec per increment to ensure only the expected PCR product was amplified per reaction. The efficiency of amplification was determined for each primer set from serial dilutions and did not vary significantly from 2. Expression of osteoblast and osteoclast-specific genes was determined by real-time RT-PCR analysis using Applied Biosystems 7700 Sequence Detector System (Applied Biosystems, Foster City, CA). Total RNA was extracted from tibia diaphysis using Purescript RNA Isolation Kit (Gentra Systems, Minneapolis, MN) followed by DNase treatment on RNeasy Micro Columns (QIAGEN) according to manufacturer’s instructions. cDNA was prepared from pooled RNA (n = 5) using TaqMan Reverse Transcription Reagents (Applied Biosystems). The PCR was performed from cDNA in duplicates using TaqMan PCR Core Reagent Kit (Applied Biosystems) with 200 nM primers and a probe, and 10 µl of cDNA template. Gene-specific primers and fluorescence-labeled probes [5'-reporter dye: FAM (6-carboxyfluorescein), 3'-quencher dye: TAMRA (6-carboxymethyl rhodamine)] (Table 2) were designed using Primer Express (version 1.5) software (Applied Biosystems) and were synthesized by Applied Biosystems. Relative expression of the RT-PCR product was then determined using the comparative C_t method (21). 18S rRNA was used to normalize expression in each sample. Fold regulation was then determined by normalizing all values to the mean of the relative expression for the control group.

Histochemical analysis of calvaria
Histochemical analysis was performed on representative calvaria from offspring of two independent founder lines (104 and 106). Calvaria were fixed in freshly prepared 4% paraformaldehyde in borate buffer for 48 h. After decalcification in Immunocal (Decal Corp., Tallman, NY) for 2–3 wk, sections were processed by dehydration, paraffin infiltration, and embedding (melting point: 58–62 C). Tissue sections (5–6 µm) were cut with a microtome and floated onto positively charged slides (Superfrost, Fisher Scientific). Sections were deparaffinized and hydrated through a xylene and graded ethanol series. For hematoxylin and eosin (H&E) staining, the sections were placed in hematoxylin for 5 min, in 1% acid alcohol for a few seconds, and in eosin for 5 min.

For AR immunocytochemistry, slides were placed in freshly prepared 3% H₂O₂ in methanol for 10 min to inhibit endogenous peroxidase activity. A high-temperature antigen unmasking technique was performed by immersing slides in boiling 0.01 M citrate buffer (pH 6.0) for 15 min, then slides were subjected to immunohistochemical staining. Polyclonal rabbit AR antibody (PA1-111A) was purchased from Affinity Bioreagents Inc. (Golden, CO) and used at 4 µg/ml. The PA1-111A AR antibody maps to the N terminus of the receptor and does not recognize other members of the steroid receptor family. Controls for nonspecific binding were incubated with rabbit nonimmune IgG. The sections were incubated with antibody overnight at 4 C. Secondary biotinylated antirabbit antibody was applied at a dilution of 1:200. Sections were incubated with ABC reagent (Vector Laboratories, Inc., Burlingame, CA) for 30 min and then processed for horseradish peroxidase/3,3'-diaminobenzidine tetramethyl chloride (DAB) using the ABC elite system (Vector Laboratories) according to the manufacturer’s instructions. Slides were counterstained with hematoxylin followed by ethanol dehydration, and then cleared in xylene and mounted in Permount (Vector Laboratories, Inc.).

Micro-CT and bone histomorphometry
The biomechanical and morphological consequences of osteoblastic and osteocytic AR elevation in AR-transgenic animals were evaluated in 8-wk-old male and female mice (n = 7–14). Right femurs from each genotype were examined for diaphyseal cross-sectional morphology and mineralization using micro-CT. Bones were placed in small polypropylene tubes and immersed in PBS during the scan. Three-dimensional micro-CT images of the diaphyseal region were obtained using an EVS MS-8 micro-CT system (EVS-GE Medical Systems, London, Ontario, Canada). Scans were performed at 6.62-µm voxel size resolution. Before each bone scan, a calibration scan was performed using a three-point calibration phantom corresponding to the density range from air to cortical bone. Bone and non-bone were differentiated using an adaptive thresholding technique (22). Computer-based stereological analyses of the full three-dimensional data set were performed to characterize femoral cross-sectional morphology using methods described previously (23). From the reconstructed and rendered micro-CT images, a 3-mm region of the mid-diaphysis, corresponding to the typical failure region for four-point bending (see below), was examined for cortical area, polar moment of inertia, and mineral content. Morphological traits were quantified for every transverse plane and averaged. The average mineralization of the diaphyseal region for each bone was determined by converting the grayscale value of each voxel to a mineral density value and then averaging mineral density values over all of the voxels for each respective region.

Fluorochrome-based dynamic histomorphometric measurements of bone formation were determined from cross sections at the femoral diaphysis or from frontal sections through the metaphysis. Distal femurs were fixed in ethanol and embedded in methyl methacrylate without prior demineralization. Cross sections (100 µm) through the central portion the of diaphysis were prepared using an SP1600 saw microtome (Leica Microsystems Inc., Bannockburn, IL) and then polished. Frontal sections (5 µm) through the central portion of the distal methaphysis were also prepared using a Poycut sledge microtome (Leica Microsystems Inc.). Dynamic histomorphometric analyses were carried out using a light/epifluorescent microscope with a charge-coupled device camera interfaced to a semiautomatic image analysis system (Bioquant NOVA 4.00.8b, Bioquant, Nashville, TN). Some sections were also stained with Masson’s Trichrome (24) for histological evaluation. Static histomorphometric analysis of trabecular bone was performed from the micro-CT image at the metaphysis with computer-aided analysis using the automated trabecular analysis system (25). All measurements were two dimensional. The terminology and units used were those recommended by the Histomorphometry Nomenclature Committee of the American Society for Bone and Mineral Research (26).

Mechanical testing
After micro-CT analysis, the right femurs were subjected to destructive testing to establish whole bone mechanical properties. To test for differences in whole bone mechanical properties among genotypes, femurs were loaded to failure in four-point bending at 0.05 mm/sec using a servohydraulic materials test system (Instron Corp., Canton, MA). Based on methods described previously [as described by Jepsen et al. (27)], all whole bone bending tests were conducted by loading the femurs in the posterior to anterior direction, such that the anterior quadrant was subjected to tensile loads. The load-deflection curves were analyzed for stiffness (the slope of the initial portion of the curve), maximum load, post-yield deflection (PYD), and work-to-failure. PYD, a measure of bone brittleness, was defined as the deflection at failure minus the deflection at yield. Yield was defined as a 10% reduction of the secant stiffness (load range normalized for deflection range) relative to the initial (tangent) stiffness. Work-to-failure was defined as the area under the load-deflection curve. Femurs were tested at room temperature and kept moist with PBS during all loading procedures.

Statistical analysis
All data were analyzed using Prism software (GraphPad Software, Inc., San Diego, CA). Significance of difference between wild-type and AR-transgenic mice was assessed by an unpaired two-tailed t test using Welch’s correction. One-way ANOVA was carried out to detect overall differences followed by Neumann Keuls multiple-comparison test to calculate intergroup differences. Body lengths and weights were analyzed by repeated measures two-way ANOVA for the effects of gender and genotype. All data are expressed as mean ± SEM.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
Generation of transgenic mice
AR-transgenic mice were created with full-length rat AR under the control of the 3.6 kb type I collagen promoter (Fig. 1). The colAR transgene was cloned as described in Materials and Methods, using full-length rat AR cDNA and rat type I ₁ promoter sequence from –3518 to +115. To create col3.6 AR-transgenic mice, the linearized colAR transgene was microinjected into pronuclei of fertilized oocytes from B6D2F1 mice, and then transferred into pseudopregnant mice (performed by the OHSU Transgenic Core Facility). Positive founders were identified by genotyping by PCR with primers at locations indicated in Fig. 1A. Founder mice were bred to wild-type B6D2F1 mice; two AR-transgenic lines (lines 104 and 106) derived from independent founders have been retained. Southern analysis confirmed a single insertion site for the AR transgene (Fig. 1B), with five to seven copies of the transgene in each line as determined by real-time PCR quantitation (Fig. 1C). Table 1 lists quantitative real-time RT-PCR analysis of expression of the colAR transgene in various tissues, showing bone targeting with highest levels in calvaria but approximately 100- to 500-fold lower in muscle, skin, heart, intestine, kidney, liver, lung, and spleen. These mice, therefore, display the expected bone-targeted AR expression, consistent with the expression patterns observed by other investigators employing this promoter construct for the generation of transgenic mice (for example, see Refs. 28 and 29).

Phenotype in AR-transgenic mice with bone-targeted AR overexpression
We first determined the effect of bone-targeted AR overexpression on body weight gain and nose-rump length over a 6-month period. At birth, animals were indistinguishable. However, as the mice aged, AR-transgenic males were significantly shorter and weighed less than wild-type littermates (Fig. 2, A and B). In contrast, AR-transgenic females were no different from wild-type controls (Fig. 2, C and D).

View larger version (17K): [in this window] [in a new window]   FIG. 2. Age-related changes in body weight and nose-rump length in AR-transgenic mice. Body weight and nose-rump-length determinations were carried out weekly or monthly, respectively, over 6 months in both genders in both wild-type (wt) and col3.6 AR-transgenic (tg) mice (n = 4–5). A, Weight gain in growing male mice. Analysis for the effects of time and genotype by repeated measures two-way ANOVA revealed an extremely significant effect of genotype (F = 54.57; P < 0.0001) and time (F = 36.51; P < 0.0001) with no interaction. B, Nose-rump length in male mice. Analysis revealed a significant effect of genotype (F = 6.21; P < 0.05) and an extremely significant effect of time (F = 15.95; P < 0.0001) with no interaction. C, Weight gain in female mice. In contrast to the male mice, analysis revealed no effect of genotype but an extremely significant effect of time (F = 31.32; P < 0.0001) with no interaction. D, Nose-rump length in female mice. Again in contrast to the male mice, analysis revealed no effect of genotype but an extremely significant effect of time (F = 25.56; P < 0.0001) with no interaction. All data are expressed as mean ± SEM.

View larger version (31K): [in this window] [in a new window]   FIG. 3. Biochemical analyses of parameters of calcium metabolism and hormone levels in AR-transgenic animals. Comparisons were performed between littermate control (wt) and AR-transgenic (tg) animals. Serum from 8-wk-old mice (n = 2–7) was analyzed to determine levels of markers of calcium metabolism. Assays were performed in duplicate by RIA for 17ß-estradiol, and by EIA for testosterone and intact mouse osteocalcin, and for calcium by the colorimetric cresolphthalein-binding method. A, 17ß-estradiol; B, testosterone; C, calcium. There were no statistical differences between 17ß-estradiol, testosterone or calcium levels. D, Osteocalcin levels were significantly reduced in both male and female AR-transgenic mice. E, OPG circulating levels were elevated in males but significantly reduced in female AR-transgenic mice. Values are expressed as mean ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (vs. gender-appropriate wild-type control).

View larger version (60K): [in this window] [in a new window]   FIG. 4. Histochemical features and immunohistochemical analysis of AR levels in calvaria from AR-transgenic mice. Calvaria were isolated from 8-wk-old male and female mice from both AR-transgenic (AR-tg) lines (104 and 106) and wild-type (wt) littermate controls, and 5-µm sections were subjected to either H&E staining or immunocytochemical analysis after demineralization and paraffin embedding. Representative sections are shown. A, Calvaria from both 104 and 106 families were evaluated. Only male AR-tg animals from both AR-tg families exhibit an increase in calvarial thickness. B, AR was detected with rabbit polyclonal antisera after DAB incubation. AR is brown and the nucleus is purple after counterstaining with hematoxylin. The majority of osteoblasts and osteocytes demonstrated AR immunoreactivity (x63). As shown in the inset image, there were no observable differences in overall AR expression between male and female animals (x40). Bar, 50 µm.

View larger version (68K): [in this window] [in a new window]   FIG. 5. Radiographic and micro-CT analysis of bone with characterization of trabecular bone formation. A, Femurs were isolated from 8-wk-old male wild-type (wt) or AR-transgenic mice (AR-tg), and subjected to faxitron x-ray imaging. Differences observed in the AR-transgenic male animals (A, left lower panel) include bulging at the metaphysis indicated with white arrow, and a reduction in femur length compared with the wild-type littermate control (A, left upper panel). No difference was apparent in female AR-transgenic or littermate control animals (not shown). Right panels show mid-diaphysis of femurs from 8-wk-old male and female AR-transgenic and littermate control mice subjected to high-resolution micro-CT imaging. Mid-saggital section taken from three-dimensional reconstructions of representative mouse femurs from AR-tg and wt littermate controls. Male AR-transgenic mice show altered trabecular bone morphology as indicated by the asterisk. B, Histological analysis of metaphyseal bone in AR-tg mice. Femurs were isolated from 8-wk-old male wt or AR-tg, sectioned (5 µm) through the central portion of the distal metaphysis and then stained with Masson’s Trichrome. C, Dynamic histomorphometric analysis was performed in trabecular bone after fluorescent imaging microscopy. Surface labeling and mineral apposition were then determined to characterize bone formation (n = 6). Mineralizing surface as a percent of bone surface (MS/BS) and BFR were significantly inhibited in trabecular bone. MAR was not affected in AR-tg males. *, P < 0.05; **, P < 0.01 (vs. wt controls).

Altered cortical bone formation in AR-transgenic mice
Because of the changes in cortical bone observed in the micro-CT images, fluorescent imaging was also carried out at the femoral diaphysis as described for trabecular bone in Fig. 5C. Figure 6A shows patterns of bone formation in images of fluorochrome labeling from femoral cross sections. The AR-transgenic males (right panel) demonstrate both a dramatic lack of labeling at the endosteal surface and an increase on the anterior periosteal surface compared with wild-type controls (left panel). Consistent with these fluorescent images, dynamic histomorphometric analysis (Fig. 6B), demonstrated divergent responses at the endosteal and periosteal surfaces. MS/BS at the endosteum was dramatically inhibited in AR-transgenic mice (P < 0.05), analogous to the reduced bone turnover noted in trabecular bone (Fig. 5C). In contrast to these findings, a nonsignificant increase in formation was noted at the periosteal surface. Similar responses were seen in BFR, with trends for inhibition at the endosteal surface and stimulation at the periosteal surface. Finally, mineral apposition rate (MAR) was dramatically inhibited at the endosteal surface in male AR-transgenic males (P < 0.05).

View larger version (41K): [in this window] [in a new window]   FIG. 6. Characterization of cortical bone formation in AR-transgenic (AR-tg) mice. A, Fluorescent images of femur after double-label administration. Eight-week-old male AR-tg mice were pulsed with oxytetracycline followed 7 d later with calcein to fluorescently label mineralizing surfaces. Femurs were isolated and sectioned (100 µm) at the mid-diaphysis. Sections were subjected to fluorescent imaging microscopy to determine patterns of bone formation. Representative photomicrographs are shown, with sites of formation indicated by double arrows. Bands were photographed at comparable anatomic positions for each bone. In contrast to male wild-type (wt) mouse femur, the AR-tg demonstrates a dramatic lack of labeling at the endosteal surface and an increase on the anterior periosteal surface. Insets are higher power images demonstrating labeling on the endosteal (e) and periosteal (p) surfaces. B, Dynamic histomorphometric analysis was performed in cortical bone after fluorescent imaging microscopy in AR-tg males (n = 6–8). MS/BS, MAR, and BFR at both the endosteal and periosteal surface were determined in wt and AR-tg mice. C, Mineral content, cortical area (Ct. Ar.) and polar moment of inertia (Jo) (all measured in micro-CT image analysis at the diaphysis) in wt and AR-tg mice. All measures were weight adjusted. Data are mean ± SEM n = 6–9 females, 3–7 males. *, P < 0.05; **, P < 0.01 (vs. wt controls).

The dramatic alteration in fluorescent labeling and lack of cortical drift noted in the male AR-transgenic animals could result in changed bone shape. To evaluate bone shape, I_AP was divided by I_ML [rectangular moments of inertia about the anterior/posterior (AP) and medial/lateral (ML) axes as gross measures of shape relative to the AP and ML axes]. Each of these terms measures the distribution of bone about these axes; thus, the ratio of these terms provides a measure of shape (the closer the ratio is to 1, the more round the shape, and the further from 1, the more elliptical the structure). Wild-type and AR-transgenic animals were the same for both males and females (data not shown). Thus, the variation in growth patterns may have affected size but not shape during development. The fluorochrome labels are representative of mineralization patterns only for the period of time that the labels are present; thus, the patterns we observed likely do not reflect drift/modeling during early development.

Bone strength in AR-transgenic mice is reduced
To analyze bone quality, whole bone failure properties were determined by loading femurs to failure in four-point bending at 0.05 mm/sec. Analysis shown in Fig. 7 revealed significant differences between wild-type and male AR-transgenic animals (family 104) in whole bone biomechanical properties at 8 wk. Male AR-transgenic mice showed significant and relatively large decreases in maximum load (P < 0.01, Fig. 7A). A much more modest but also significant decline in female AR-transgenic mice (P < 0.05, Fig. 7A) was also seen. The effect on the female AR-transgenic mice is subtle with only a 13–14% reduction in maximum load when values are body weight corrected. In all other measures, female AR-transgenic mice were not significantly different. In comparison to wild-type and AR-transgenic females, male AR-transgenic mice showed a significant decrease in stiffness (P < 0.05, Fig. 7B), PYD (P < 0.05, Fig. 7C) and work-to-failure (P < 0.01, Fig. 7D). Furthermore, male AR-transgenic mice weighed less and had significantly shorter femur length (P < 0.001, Fig. 7, E and F) consistent with the faxitron and micro-CT image in Fig. 6A, indicating that AR action may also have a role in determining closure at the epiphysis or other aspects of longitudinal growth. Similar results were obtained with AR-transgenic family 106.

View larger version (39K): [in this window] [in a new window]   FIG. 7. Biomechanical analyses of bone quality in AR-transgenic mice. Femurs from wild-type (wt) and col3.6 AR-transgenic (tg) mice were isolated from 8-wk-old mice to determine whole bone failure properties. Femurs were loaded to failure in four-point bending and the stiffness, maximum load, and PYD were calculated from the load-deflection curves and adjusted for body weight differences. A, Adjusted maximum load; B, adjusted stiffness; C, PYD; D, work-to-failure; E, femoral length; F, weight. The whole bone biomechanical properties are shown as mean ± SEM, n = 7–14. Differences between genotypes were determined by Student’s t test. *, P < 0.05; **, P < 0.01, ***, P < 0.001 (vs. gender-appropriate wt controls).

View larger version (40K): [in this window] [in a new window]   FIG. 8. Gene expression in AR-transgenic mice. Analysis of steady-state mRNA expression of osteoblast or osteoclast genes was determined by real-time RT-PCR analysis using tibial RNA isolated from wild-type (wt) or AR-transgenic mice (tg), n = 5. Osteoblast genes involved in bone formation and matrix production examined included ColI, osterix, osteocalcin (OC), cyclin D1. Osteoclast genes involved in bone resorption were TRAP, OPG, RANKL, and cathepsin K (CatK). A and B, Analysis of expression of osteoblastic (A) and osteoclastic (B) genes in males. C and D, Analysis of expression of osteoblastic (C) and osteoclastic (D) genes in females.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The col3.6 promoter was chosen for several reasons: the skeletal expression patterns for this promoter are both well characterized and bone selective (for example, see Refs. 31, 32, 33); the col3.6 promoter is active in the periosteum (29), a site of particular interest for androgen responsiveness; and androgens do not inhibit expression from the 3.6-kb promoter fragment (data not shown). The col3.6 promoter fragment directs expression of the fused transgene very early in periosteal, endosteal, and trabecular surfaces and in cortical bone in the collar region of the growth plate (29). In addition, the col3.6 promoter is more active than the osteocalcin promoter in the control of expression of transgene (32) and is the only known promoter able to drive Cre expression in osteoblasts at a high enough level to induce effective recombination activity (34).

The identification of mechanisms involved in androgen-mediated changes in osteoblast function has important ramifications for both basic and applied knowledge of bone physiology. The discovery of pathways that androgens influence in bone is particularly significant because, as with the recently approved PTH therapy, androgen remains as a promising therapeutic agent with a demonstrated anabolic effect on the skeleton (35). A better understanding of the mechanisms of androgen action in bone will also be relevant for the development of a class of drugs termed selective AR modulators, for the treatment of osteoporosis and other disorders (36). These drugs, analogous to the selective ER modulators (37), are being developed to stimulate anabolic actions in bone, but to not display the detrimental changes in lipid profiles, increases in prostate growth and facial hair growth in women that have been associated negatively with androgen therapy. Thus, the elucidation of the mechanism(s) mediating the effect of androgens on osteoblasts will provide for a better understanding of the cascade of molecular events associated with androgen exposure in bone, may provide a framework for improvements in the diagnosis and treatment of metabolic bone diseases, and could help to identify approaches to overcome the deleterious consequences of hypogonadism on bone mass. Because of an enhanced responsiveness in the AR-transgenic animals, the AR-transgenic mice represent a useful model to identify important actions of androgen that influence skeletal modeling and remodeling.

Analysis by micro-CT and faxitron imaging demonstrated altered trabecular bone morphology only in the male AR-transgenic mice. A particularly striking "bulge" at the metaphysis is observed, and femur length is reduced. The trabecular tissue phenotype is dramatic and is seen in the metaphyses at both ends of the femur. In the male AR-transgenic animals, static histomorphometric analysis in the metaphysis confirmed the increased trabecular bone volume, comprised of more trabeculae that are smaller in size and closer together. Because bone volume fraction was increased, AR overexpression appears to be beneficial, but further biomechanical analysis will be necessary to confirm that these morphological changes result in superior mechanical properties. These results are consistent with the histological, micro-CT, and faxitron images and are analogous to the phenotype observed in growing animals treated with antiresorptive drugs because a similar morphology is observed in both human and rodent growing skeleton after administration of bisphosphonates. These effects of treatment noted in these studies include increased metaphyseal bone probably due to higher trabecular number not thickness, a "club-shaped" metaphysis, reduced surface-based indicators of trabecular bone modeling or remodeling, and decreased femur lengths in some studies (38, 39, 40). In the adult skeleton during remodeling, bone resorption and formation are physically closely coupled, and as a consequence both processes would be inhibited with antiresorptive drug treatment. However during modeling in the growing animal, osteoclasts and osteoblasts are active on different and distinct surfaces and are thus uncoupled. It is thus interesting to speculate that AR overexpression in the osteoblast lineage can alter the way trabecular bone is remodeled or constructed in the metaphyseal region during growth.

A possible reduced turnover phenotype in AR-transgenic mice is also suggested from the decline in serum osteocalcin levels, analogous to the observation that DHT treatment reduces osteocalcin levels in gonadectomized rat models (41, 42). In addition, reduced serum osteocalcin levels have been observed with antiresorptive therapy (43). Serum osteocalcin is a complex measure reflecting bone turnover and could represent either decreased formation or decreased resorption or both (44). Because of the resemblance of the phenotype to that seen after antiresorptive drug administration in growing animals described above and because of the increased formation noted in the calvaria and periosteal surface where there are few osteoclasts, these findings are consistent with an anabolic and/or antiresorptive response in AR-transgenic mice depending on the site. Changes in osteoclastic gene expression with increased OPG and decreased cathepsin K, TRAP, and RANKL levels in AR-transgenic bones are also consistent with suppression of bone resorption in these animals, likely mediated through communication with AR-targeted osteoblastic cells. Evaluation of serum levels of OPG also revealed an increase in males but decrease in female AR-transgenic mice, consistent with the changes observed in gene expression. The effects of androgens on OPG levels are controversial because they have been associated both positively (45) and negatively with androgen levels (46, 47). Furthermore, these results suggest a possible role for the RANKL/RANK/OPG signaling pathway in the AR-transgenic mice and are consistent with reports that androgen inhibits osteoclast activity (48) and osteoclastogeneisis (16, 49, 50). The effect of androgens on osteoclast survival is also controversial, with either enhancement (13) or no effect (16) reported. Ongoing histomorphometric and in vitro analyses, designed to characterize osteoblast and osteoclast number and activity, will be helpful in addressing these issues.

We have obtained no evidence that estrogen action plays a role in the phenotype that is observed in the AR-transgenic animals. Because the bone phenotype (size, cortical thickness etc.) of the transgenic male is smaller than the wild-type female, the male phenotype is not readily explained by increased transactivation of ER (because activation of ER in the male would not make the male skeleton smaller than the female). There is no evidence that circulating estrogen levels are affected in these animals because the serum estrogen concentrations are not different between wild-type and AR-transgenic males. It is possible that local aromatization may increase estrogen concentrations in bone without influencing circulating levels. Although both aromatase mRNA and activity have been detected in cultured osteoblasts, treatment with androgen (i.e. R1881 or DHEA) had no effect on osteoblast aromatase activity (51). Consistent with this result, we observed no detectable differences in the very low levels of aromatase mRNA present in tibial samples between male AR-transgenic and wild-type animals.

The bone phenotype observed in AR-transgenic mice is also consistent with many of the known effects of androgen treatment on the skeleton. The skeletal response to androgen treatment has been characterized as increased cortical and trabecular bone mass (with increased trabecular number but not thickness), increased bone width with surface periosteal expansion and a lack of inner endosteal deposition, in the setting of inhibition of resorption due to reduced osteoclast activity (9, 20, 52, 53, 54). Although AR-transgenic males demonstrate an altered trabecular morphology with increased trabecular bone volume and increased periosteal apposition, they also exhibit dramatic inhibition at the endosteal envelope that may be responsible for the decreased cortical bone area and changes in biomechanical properties we have observed. Not surprisingly, DHT replacement does not preserve mechanical strength in orchidectomized mice (54). Inhibition of osteoclastic resorption may be responsible for bulging at the metaphysis and altered trabecular morphology in AR-transgenic males and would be consistent with reduced osteoclast activity and increased trabecular bone observed with androgen therapy.

Some characteristics of the phenotype observed in 8-wk-old male AR-transgenic mice cannot be explained by the known physiological actions of androgens in adults. For example, biomechanical analysis of the skeletal phenotype demonstrated significantly reduced stiffness, maximum load, PYD, and work-to-failure in male AR-transgenic mice. The reduced stiffness and maximum load are consistent with a reduction in cross-sectional area (27, 55). Significant reductions in PYD and work are indications that bone quality and/or matrix properties may be altered in these mice. The reduced PYD observed in the male AR-transgenic mice cannot be explained by an increase in mineral content because the male AR-transgenic show a similar mineral content value compared with male wild-type by high-resolution micro-CT analysis. Therefore, other changes in bone matrix quality are likely responsible (e.g. analysis of MOV13 mouse, see Ref. 56). A potential change in bone matrix quality in AR-transgenic mice is also consistent with the reduced osteoblastic gene expression noted in the tibial RNA samples. Significantly shorter femurs observed in male AR-transgenic mice suggest premature closure at the epiphysis and could represent direct effects in the cartilage because it has been reported that there is a low-level col3.6 transgene activation in hypertrophic cartilage (29). However, this phenotype is unlikely to be a result of direct effects in cartilage because testosterone injections result in stimulatory effects on epiphyseal growth in castrated rats (57). It is also possible that the effects of androgens on epiphyseal closure are due to accelerated differentiation of osteoblasts toward osteocytes, leading to early ossification and mineralization in the secondary spongiosa. We have previously demonstrated increased AR expression as osteoblasts differentiate into osteocytes, suggesting a role for androgen in osteoblast differentiation (58). Androgen treatment has also been shown to enhance mineralization (59).

Not surprisingly, the overall phenotype observed in the male AR-transgenic mice represents in some ways the converse of that characterized in null AR mice with global elimination of functional AR. With the null AR mutation, mice develop a high turnover osteopenia, with increased formation in trabecular bone and at the endosteal surface, but even higher bone resorption resulting in reduced trabecular volume (16, 17). Conversely, we have demonstrated that with bone-targeted AR overexpression, males develop a phenotype consistent with reduced turnover and inhibition of osteoclast activity, suggested by the changes in osteoblastic and osteoclastic gene expression, and reduced serum osteocalcin levels, elevated OPG levels, and the altered morphology noted at the metaphysis with increased trabecular bone volume but a reduction in MAR. Loss of AR function in the null AR model also increased RANKL expression (16), opposite to the inhibition in RANKL expression we observed in males with AR overexpression. Thus, AR-transgenic mice and AR null mice both have altered bone quality, but likely through different cellular mechanisms.

Evidence of the anabolic actions of androgens can be observed at the calvaria and periosteum. Male AR-transgenic mice develop widening of the calvaria that could reflect periosteal expansion. In addition, enhanced periosteal apposition after fluorochrome labeling in the femur was noted, but with a dramatic reduction in endosteal labeling characterized by decreased MAR and BFR and thus a change in the normal cortical drift pattern at the time of labeling. These data, particularly in cortical bone and at the calvaria, are consistent with the documented anabolic effects of androgens on periosteum. Although the reduction of endosteal apposition could reflect an effect on bone resorption because the marrow space undergoes net bone resorption as the bone is growing, it is also conceivable that transgenic expression of AR suppresses endosteal bone formation. AR transactivation directly in osteoblastic cells may thus play a primary role in determining sexual dimorphism in the skeleton, i.e. that male bones tend to be wider but not thicker (20).

In summary, skeletally targeted AR overexpression results in a complex phenotype characterized by increased periosteal bone formation and calvarial thickening, and decreased bone turnover/bone resorption on trabecular and endosteal surfaces, leading to increased trabecular bone volume with increased trabecular number and reduced trabecular separation. In addition, despite the increased periosteal growth, the size and length of long bone, cortical area and mechanical properties are all reduced in male AR-transgenic mice, which may result from inhibition of bone turnover during growth as is observed with antiresorptive therapies. Thus, whereas aromatization of androgen to estrogen does contribute to anabolic responses seen with testosterone (Ref. 41 , but also see Refs. 54 and 60), we postulate that the direct effects of androgen on osteoblasts through AR transactivation are nevertheless important. The skeletal phenotype is seen predominantly in male AR-transgenic mice with elevated circulating androgen levels relative to the females, and in many ways resembles that observed with androgen treatment. Therefore, the differences observed between littermate controls and the bone-targeted AR-transgenic lines suggest that many of the effects of androgen therapy are the consequence of direct androgen transactivation of the AR in bone. Analysis of the AR-transgenic mice may thus provide a proof-of-principle that AR transactivation directly in the osteoblastic lineage mediates at least some effects of androgens on skeletal homeostasis.

	Acknowledgments

 
The authors would like to thank Joel Hashimoto for excellent technical assistance, Steven Tommasini and Christopher Price for help with the micro-CT and histology, Dr. David Rowe (University of Connecticut Health Center, Farmington, CT) for providing the plasmid containing the rat ColI1 promoter sequence, and Dr. Shutsung Liao (University of Chicago, Chicago, IL) for the rat AR cDNA. We also want to express our gratitude to Dr. Gideon Rodan for continuing encouragement and support.

	Footnotes

 
This material is based upon work supported by the Department of Defense DAMD17-01-1-0806 (to K.J.J.) and the Office of Research and Development, Medical Research Service, Department of Veterans Affairs (to K.M.W.).

Abbreviations: AP, Anterior/posterior; AR, androgen receptor; BFR, bone formation rate; BS, bone surface; ColI₁, type I ₁ collagen; DAB, diaminobenzidine tetramethyl chloride; DHT, 5-dihydrotestosterone; DNase, deoxyribonuclease; ER, estrogen receptor; H&E, hematoxylin and eosin; MAR, mineral apposition rate; micro-CT, micro-computed tomography; ML, medial/lateral; OPG, osteoprotegerin; ORX, orchidectomy; OVX, ovariectomy; PYD, post-yield deflection; RANK, receptor activator of nuclear factor-B; RANKL, RANK ligand; TRAP, tartrate-resistant acid phosphatase.

Received August 7, 2003.

Accepted for publication March 11, 2004.

	References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Cooper C, Melton LR 1996 Magnitude and impact of osteoporosis and fractures. In: Marcus R, Feldman D, Kelsey J, eds. Osteoporosis. San Diego: Academic Press; 419–434
2. Greenspan S, Oppenheim D, Klibanski A 1989 Importance of gonadal steroids to bone mass in men with hyperprolactinemic hypogonadism. Ann Intern Med 110:526–531
3. Turner R, Hannon K, Demers L, Buchanan J, Bell N 1989 Differential effects of gonadal function on bone histomorphometry in male and female rats. J Bone Miner Res 4:557–563[Medline]
4. Miller B, De Souza M, Slade K, Luciano A 2000 Sublingual administration of micronized estradiol and progesterone, with and without micronized testosterone: effect on biochemical markers of bone metabolism and bone mineral density. Menopause 7:318–326[Medline]
5. Castelo-Branco C, Vicente J, Figueras F, Sanjuan A, Martinez de Osaba M, Casals E, Pons F, Balasch J, Vanrell J 2000 Comparative effects of estrogens plus androgens and tibolone on bone, lipid pattern and sexuality in postmenopausal women. Maturitas 34:161–168[CrossRef][Medline]
6. Raisz L, Wiita B, Artis A, Bowen A, Schwartz S, Trahiotis M, Shoukri K, Smith J 1996 Comparison of the effects of estrogen alone and estrogen plus androgen on biochemical markers of bone formation and resorption in postmenopausal women. J Clin Endocrinol Metab 81:37–43[Abstract]
7. Kenny A, Raisz L 2002 Mechanisms of bone remodeling: implications for clinical practice. J Reprod Med 47:63–70[Medline]
8. Liegibel U, Sommer U, Tomakidi P, Hilscher U, Van Den Heuvel L, Pirzer R, Hillmeier J, Nawroth P, Kasperk C 2002 Concerted action of androgens and mechanical strain shifts bone metabolism from high turnover into an osteoanabolic mode. J Exp Med 196:1387–1392[Abstract/Free Full Text]
9. Leder B, LeBlanc K, Schoenfeld D, Eastell R, Finkelstein J 2003 Differential effects of androgens and estrogens on bone turnover in normal men. J Clin Endocrinol Metab 88:204–210[Abstract/Free Full Text]
10. Hamrick M, McPherron A, Lovejoy C, Hudson J 2000 Femoral morphology and cross-sectional geometry of adult myostatin-deficient mice. Bone 27:343–349[Medline]
11. Reid I 2002 Relationships among body mass, its components, and bone. Bone 31:547–555[Medline]
12. Kousteni S, Bellido T, Plotkin L, O’Brien C, Bodenner D, Han L, Han K, DiGregorio G, Katzenellenbogen J, Katzenellenbogen B, Roberson P, Weinstein R, Jilka R, Manolagas S 2001 Nongenotropic, sex-nonspecific signaling through the estrogen or androgen receptors: dissociation from transcriptional activity. Cell 104:719–730[Medline]
13. Kousteni S, Chen J, Bellido T, Han L, Ali A, O’Brien C, Plotkin L, Fu Q, Mancino A, Wen Y, Vertino A, Powers C, Stewart S, Ebert R, Parfitt A, Weinstein R, Jilka R, Manolagas S 2002 Reversal of bone loss in mice by nongenotropic signaling of sex steroids. Science 298:843–846[Abstract/Free Full Text]
14. van der Eerden B, Emons J, Ahmed S, van Essen H, Lowik C, Wit J, Karperien M 2002 Evidence for genomic and nongenomic actions of estrogen in growth plate regulation in female and male rats at the onset of sexual maturation. J Endocrinol 175:277–288[Abstract]
15. Sims N, Clement-Lacroix P, Minet D, Fraslon-Vanhulle C, Gaillard-Kelly M, Resche-Rigon M, Baron R 2003 A functional androgen receptor is not sufficient to allow estradiol to protect bone after gonadectomy in estradiol receptor-deficient mice. J Clin Invest 111:1319–1327[CrossRef][Medline]
16. Kawano H, Sato T, Yamada T, Matsumoto T, Sekine K, Watanabe T, Nakamura T, Fukuda