Osteoblast Deletion of Exon 3 of the Androgen Receptor Gene Results in Trabecular Bone Loss in Adult Male Mice

Authors


  • The authors state that they have no conflicts of interest.

Abstract

The mechanism of androgen action on bone was studied in male mice with the AR deleted in mature osteoblasts. These mice had decreased trabecular bone volume associated with a decrease in trabecular number, suggesting that androgens may act directly on osteoblasts to maintain trabecular bone.

Introduction: Androgens modulate bone cell activity and are important for the maintenance of bone mass. However, the mechanisms by which they exert these actions on bone remain poorly defined. The aim of this study was to investigate the role of androgens acting through the classical androgen receptor (AR) signaling pathways (i.e., DNA-binding dependent pathways) in osteoblasts using male mice in which exon 3 of the AR gene was deleted specifically in mature osteoblasts.

Materials and Methods: Mice with a floxed exon 3 of the AR gene were bred with Col 2.3-cre transgenic mice, in which Cre recombinase is expressed in mineralizing osteoblasts. The skeletal phenotype of mutant mice was assessed by histomorphometry and quantitative μCT at 6, 12, and 32 weeks of age (n = 8 per group). Wildtype, hemizygous exon 3 floxed and hemizygous Col 2.3-cre male littermates were used as controls. Data were analyzed by one-way ANOVA and Tukey's posthoc test.

Results: μCT analysis of the fifth lumbar vertebral body showed that these mice had reduced trabecular bone volume (p < 0.05) at 32 weeks of age compared with controls. This was associated with a decrease in trabecular number (p < 0.01) at 12 and 32 weeks of age, suggesting increased bone resorption. These effects were accompanied by a reduction in connectivity density (p < 0.01) and an increase in trabecular separation (p < 0.01). A similar pattern of trabecular bone loss was observed in the distal femoral metaphysis at 32 weeks of age.

Conclusions: These findings show that inactivation of the DNA binding–dependent functions of the AR, specifically in mature osteoblasts in male mice, results in increased bone resorption and decreased structural integrity of the bone, leading to a reduction in trabecular bone volume at 32 weeks of age. These data provide evidence of a role for androgens in the maintenance of trabecular bone volume directly through DNA binding–dependent actions of the AR in mature osteoblasts.

INTRODUCTION

Androgens play an important role in the acquisition of peak bone mass during puberty and in the maintenance of trabecular bone mass during aging.(1,2) The actions of androgens are mediated through the androgen receptor (AR), a ligand-dependent transcription factor. Despite the large number of clinical and animal studies that have investigated the effects of androgens on bone,(3) the molecular mechanisms of action and the target cells involved remain poorly defined.

AR expression has been shown in a number of bone cells including osteoblasts, osteoclasts, osteocytes, and chondrocytes.(4–7) The identification of functional ARs in bone strongly suggests that bone is a direct androgen target tissue. Orchidectomy of young and aged male rats results in cortical and trabecular bone loss caused by an imbalance of bone formation and resorption leading to increased bone turnover.(8,9) In growing male rats, the reduction in cortical bone mass caused by orchidectomy seems to be caused by a decrease in periosteal bone formation(8) and is partially prevented by treatment with the nonaromatizable androgen, 5α-dihydrotestosterone (DHT), verifying that androgen action on bone is at least in part, mediated directly through the AR.(10) In vitro studies have shown that DHT stimulates the proliferation and differentiation of isolated osteoblasts and that these effects are blocked by AR antagonists,(11,12) further supporting the hypothesis that androgen effects on bone are at least in part mediated through the AR in osteoblasts. Other studies have shown that androgens inhibit osteoclast formation,(13,14) osteoclast activity,(15,16) and bone resorption.(17)

The well-defined classical mechanism of androgen action involves androgen binding to the AR, which in turn regulates target gene transcription through interaction with specific response elements known as androgen response elements (AREs), which are located within the promoter regions of target genes.(18,19) The AR can also regulate gene transcription without binding DNA. This mechanism occurs through protein–protein interactions with other DNA-binding transcription factors. Androgens can also induce transcription-independent (nongenomic) signaling. Nongenomic mechanisms of androgen action are often associated with the rapid induction of second messenger signal transduction cascades and also do not involve DNA binding of the AR.(20,21) It has been suggested that the classical signaling pathways, essential for normal male sexual differentiation,(22,23) are not required for their bone protective effects, and that nongenomic pathways are largely responsible for skeletal effects of androgens.(24) Androgens may also exert significant effects on bone indirectly through aromatization to estradiol and subsequent activation of the estrogen receptor (ER).(25)

Physiological evidence that supports a direct role for the AR in male skeletal homeostasis has been shown using genetically engineered global AR knockout (ARKO) mice, in which there is a complete absence of AR protein.(16) Global ARKO males display high turnover osteopenia with increased bone resorption, leading to reduced trabecular bone mass and reduced cortical thickness, suggesting that AR function is essential for normal bone formation and remodeling, at least in male mice.(16)

We tested the hypothesis that the effects of androgens on bone are mediated predominantly through direct DNA binding–dependent actions of the AR in osteoblasts. To achieve this, we generated mice lacking the second zinc finger of the DNA-binding domain of the AR, specifically in mature osteoblasts using the Cre/loxP system. These mice provide an opportunity to dissect the relative contributions of the DNA binding–dependent and DNA binding–independent mechanisms of androgen action within bone.

MATERIALS AND METHODS

Animal care

All transgenic mice were backcrossed for four generations onto a C57BL/6J background, resulting in mice whose genetic composition was >93% C57BL/6J. All mice were supplied with water and standard chow (Barastoc GR2 rat and mouse breeder ration; Ridley Agriproducts) ad libitum and were housed at 22°C on a 12-h light/dark cycle in standard cages. All procedures involving animals were approved by the Austin Health Animal Ethics Committee.

Generation of transgenic mice

A mouse line, in which exon 3 of the mouse AR gene was flanked by loxP sites, was generated using standard techniques of cloning followed by homologous recombination in embryonic stem (ES) cells.(23) Col 2.3-cre transgenic mice were generated by conventional techniques.(26) Cre recombinase expression in these mice is directed to mature osteoblasts under the control of a 2.3-kb fragment of the rat type 1a1 collagen (Col1a1) promoter. Female mice, heterozygous for the exon 3 floxed AR allele (ARflox(ex3)/+) were bred with male mice, hemizygous for the Col 2.3-cre transgene resulting in male offspring hemizygous for osteoblast-specific deletion of exon 3 of the AR gene (ARflox(ex3)/Y;Col 2.3-cre). Offspring were weaned at 3 weeks of age, and males were genotyped by PCR as described previously.(23) Wildtype (AR+/Y), hemizygous exon 3 floxed (ARflox(ex3)/Y), and hemizygous Col 2.3-cre (AR+/Y;Col 2.3-cre) male littermates were used as controls.

RNA isolation, cDNA synthesis, and RT-PCR

Total RNA was isolated from the long bones of ARflox(ex3)/Y;Col 2.3-cre mice by phenol chloroform extraction as described previously.(27) mRNA was reverse transcribed using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Promega) according to the manufacturer's instructions. The resulting cDNA was subjected to PCR using AR-specific primer pairs that flank exon 3:exon 2, 5′-GACAGTACCAGGGACCATGTT-3′ and exon 4, 5′-CTCAATGGCTTCCAGGACGTT-3′. PCR was performed for 35 cycles at 94°C for 30 s, 54°C for 30 s, and 72°C for 45 s.

Serum and bone collection

For dynamic histomorphometry experiments, mice received two intraperitoneal injections of 20 mg/ml calcein (Sigma), with a 7-day interval between injections. Three days after the second injection, mice were anesthetized by intraperitoneal injection of 80 mg/kg ketamine and 80 mg/kg xylazine, and blood was collected by terminal cardiac puncture. Blood samples were allowed to clot and were centrifuged at 4500 rpm for 10 minutes at room temperature. Serum was aspirated, transferred to a clean tube, and stored at −20°C until required. Femora and the spine were dissected, and the surrounding soft tissue was removed.

Bone histomorphometry

Femur length was measured using a digital caliper (Etalon). Distal femora from 6-, 12-, and 32-week-old mice were prepared for quantitative histomorphometry using established resin-embedding techniques as described previously.(28) Five-micrometer-thick longitudinal sections were cut using a Jung K motorized microtome (Reichert). Sections were stained using a modified von Kossa silver technique(29) and subsequently counterstained with H&E for osteoclast and osteoid surface calculations. Trabecular bone volume, thickness, and number were calculated in the metaphyseal region below the growth plate excluding the primary spongiosa, using a Quantimet 500 image analysis system (Cambridge Instruments). To exclude primary spongiosa from analysis in the metaphysis, measurements started 0.55 mm from the growth plate for mice at 6 weeks of age and 0.27 mm from the growth plate for mice at 12 and 32 weeks of age. The region of the metaphysis analyzed was a function of femur length and extended 1.6 mm for mice 6 weeks of age and 1.94 mm for mice 12 and 32 weeks of age from the starting point above the growth plate. This region included all of the secondary spongiosa of the metaphysis ending at the border of the metaphyseal/diaphyseal region where the cortical bone shaft begins to narrow. Dynamic markers of bone turnover were estimated in the same area of the secondary spongiosa of the distal femoral metaphysis. Osteoclast surface was determined manually as a percentage of trabecular bone surface occupied by osteoclasts using an ocular mounted Weibel II graticule at ×200 magnification. Osteoclasts were defined as multinucleated cells, with foamy cytoplasm, located adjacent to trabecular bone surfaces in a resorption pit except those covered with osteoid. Unstained sections were used for evaluation of calcein fluorescence. Mineralizing surface (MS) was estimated using the equation, MS (%) = [(0.5 × sLS) + dLS] × 100/BS, where BS = bone surface, sLS = single-labeled surface (% of BS), and dLS = double-labeled surface (% of BS).(30) Mineral apposition rate (MAR) was estimated using the equation, MAR (μm/day) = mean distance between labels/time interval between injection of labels. Bone formation rate was estimated using the following equation: BFR = MS × MAR.

Quantitative μCT

Femora and vertebrae from 12- and 32-week-old mice were evaluated using a desktop microtomographic imaging system (μCT40; Scanco Medical) equipped with a 10-mm focal spot microfocus X-ray tube. Transverse CT slices of the distal femoral metaphysis and vertebral body (L5) were acquired using 12-μm isotropic voxel size. For the vertebral body, ∼220 CT slices were obtained from the cranial to caudal growth plates. Trabecular bone parameters were evaluated in the volume of interest extending from just below (∼60 μm) the cranial growth plate to just above (∼60 μm) the caudal growth plate. The trabecular bone in the vertebral body was delineated manually. For the distal femur, 200 CT slices were acquired, and trabecular bone properties were evaluated in a region starting 0.36 mm proximal to the growth plate and extending 1.8 mm proximally. Images were reconstructed, filtered, and thresholded using a specimen-specific threshold determined by an adaptive, iterative algorithm.(31) Morphometric parameters were computed using a direct 3D approach that does not rely on any assumptions about the underlying structure.(32,33) For trabecular morphology, the following variables were assessed: bone volume/tissue volume (BV/TV, %), trabecular thickness (Tb.Th, μm), trabecular separation (Tb.Sp, μm), trabecular number (Tb.N, /mm), and connectivity density (Conn.D/mm3). Transverse CT slices were also acquired at the femoral midshaft using 12-μm slice increments (50 μCT slices per specimen). For this cortical region, the total cross-sectional area, cortical bone area, and medullary area (TA, BA, and MA, respectively, mm2), bone area fraction (BA/TA, %), and cortical thickness (μm) were assessed.

Biochemical analyses

Serum hormone levels were measured using serum from 9- and 32-week-old mice. Testosterone levels were assayed in duplicate by radioimmunoassay as previously described.(34) Estradiol, mouse luteinizing hormone (LH), and mouse follicle-stimulating hormone (FSH) were measured by immunofluorometric assay as previously described.(34–36) Urine was collected for 12 h before death in individual metabolic cages. Urine creatinine was measured on an automated chemical analyser using manufacturer recommended methods. Total deoxypyridinoline (DPD) cross-links were measured by competitive enzyme immunoassay using a commercially available kit (Metra Total DPD EIA Kit; Quidel) according to the manufacturer's instructions in urine collected from 6- and 12-week-old mice and in serum collected from 32-week-old mice. Urine DPD was expressed as a ratio to urine creatinine. Serum osteocalcin was measured by immunoradiometric assay using a commercially available kit (Immuntopics, San Clemente, CA, USA) according to the manufacturer's instructions.

Statistical analysis

Bone histomorphometry, μCT analyses, and serum hormone data were analyzed by one-way ANOVA. The effects of genotype and age on urinary DPD and serum osteocalcin were analyzed by two-way ANOVA. When the ANOVA indicated a significant effect, the specific differences were identified by Tukey's posthoc test. A value of p < 0.05 was considered significant. All tests were performed using SPSS 11 for Mac OS X software.

RESULTS

Confirmation of exon 3 deletion in bone

Deletion of exon 3 of the AR was assessed in the long bones of ARflox(ex3)/Y;Col 2.3-cre male mice. Cre-mediated deletion of exon 3 was shown by RT-PCR using primers homologous to exons 2 and 4 of the mouse AR. A wildtype band of 426 bp and an exon 3–deleted band of 309 bp was seen in the long bones, whereas only the wildtype band was observed in the brain (control tissue) of an ARflox(ex3)/Y;Col 2.3-cre male mouse (Fig. 1).

Figure Figure 1.

(A) RT-PCR using cDNA isolated from the long bones, brain, and colon (negative control tissue) of an ARflox(ex3)/Y;Col 2.3-cre male with AR-specific primers located in exons 2 and 4. (B) Schematic diagram of wildtype and exon 3–deleted AR mRNA showing size of expected PCR products: 426 and 309 bp, respectively.

Vertebra

Trabecular bone volume of the fifth lumbar vertebral body was reduced (p < 0.05) in ARflox(ex3)/Y;Col 2.3-cre mice compared with ARflox(ex3)/Y controls at 32 weeks of age (Figs. 2A and 2B). μCT analysis showed a decrease in trabecular number (p < 0.01) and an increase in trabecular thickness (p < 0.05) compared with control mice at 12 and 32 weeks of age (Figs. 2C and 2D). This was associated with an increase in trabecular separation (p < 0.01) and a decrease in connectivity density (p < 0.01) compared with control mice (Figs. 2E and 2F).

Figure Figure 2.

μCT analysis of the fifth lumbar vertebral body at 12 and 32 weeks of age. (A) Representative μCT images of the fifth lumbar vertebral body from 12-week-old (left) and 32-week-old (right) AR+/Y, ARflox(ex3)/Y, AR+/Y;Col 2.3-cre, and ARflox(ex3)/Y;Col 2.3-cre male mice. (B) Trabecular bone volume (BV/TV). (C) Trabecular thickness (Tb.Th). (D) Trabecular number (Tb.N). (E) Trabecular separation (Tb.Sp). (F) Connectivity density (Conn.D). Values are mean ± SE, n ≥ 6.ap < 0.05 vs. 32-week-old ARflox(ex3)/Y, bp < 0.01 vs. 12-week-old AR+/Y;Col 2.3-cre, cp < 0.05 vs. 32-week-old AR+/Y and AR+/Y;Col 2.3-cre, dp < 0.01 vs. all 12-week-old controls, ep < 0.001 vs. all 32-week-old controls, fp < 0.001 vs. all 12-week-old controls, gp < 0.01 vs. 32-week-old ARflox(ex3)/Y.

No changes were detected in cortical thickness of the fifth lumbar vertebral body (data not shown).

Femur

ARflox(ex3)/Y;Col 2.3-cre mice had normal femur length compared with control mice at 6, 12, and 32 weeks of age (Table 1). Trabecular bone of the distal femoral metaphysis showed a similar pattern of bone loss to that observed in the fifth lumbar vertebral body. Bone volume was reduced (p < 0.05) in ARflox(ex3)/Y;Col 2.3-cre mice compared with ARflox(ex3)/Y controls at 32 weeks of age (Table 1). μCT analysis showed a reduction in trabecular number (p < 0.05) in ARflox(ex3)/Y;Col 2.3-cre mice compared with ARflox(ex3)/Y controls at 32 weeks of age (Fig. 3D). This was associated with an increase in trabecular separation (p < 0.05) and a decrease in connectivity density (p < 0.01) compared with ARflox(ex3)/Y controls at 32 weeks of age (Figs. 3E and 3F). No changes in any parameters were detected at 6 or 12 weeks of age by histomorphometry (Table 1) or μCT analysis (Figs. 3A–3F). Dynamic histomorphometry in the metaphyseal region of the distal femur showed no changes in osteoid surface, osteoclast surface, MAR, or BFR in ARflox(ex3)/Y;Col 2.3-cre male mice compared with controls (Table 2).

Table Table 1.. Femoral Length and Static Bone Histomorphometry of the Distal Femur in 6-, 12-, and 32-Week-Old Mice
original image
Table Table 2.. Dynamic Bone Histomorphometry of the Distal Femur in 6-, 12-, and 32-Week-Old Mice
original image
Figure Figure 3.

μCT analysis of the distal femoral metaphysis at 12 and 32 weeks of age. (A) Representative μCT images of the distal femur from 12-week-old (left) and 32-week-old (right) AR+/Y, ARflox(ex3)/Y, AR+/Y;Col 2.3-cre, and ARflox(ex3)/Y;Col 2.3-cre male mice. (B) Trabecular bone volume (BV/TV). (C) Trabecular thickness (Tb.Th). (D) Trabecular number (Tb.N). (E) Trabecular separation (Tb.Sp). (F) Connectivity density (Conn.D). Values are mean ± SE, n ≥ 6. ap < 0.05 and bp < 0.01 vs. 32-week-old ARflox(ex3)/Y control mice.

No changes were detected in cortical thickness or periosteal circumference of the femoral midshaft of 12- or 32-week-old ARflox(ex3)/Y;Col 2.3-cre mice (Fig. 4).

Figure Figure 4.

Cortical bone of the femoral midshaft at 12 and 32 weeks of age. Representative μCT images, cortical thickness, and periosteal circumference of the femoral midshaft from AR+/Y, ARflox(ex3)/Y, AR+/Y;Col 2.3-cre, and ARflox(ex3)/Y;Col 2.3-cre male mice at (A) 12 and (B) 32 weeks of age. Values are mean ± SE.

Biochemical analyses

There was no difference in testosterone, estradiol, LH, or FSH levels between ARflox(ex3)/Y;Col 2.3-cre and control mice at 9 or 32 weeks of age (Table 3). There was no difference in urinary DPD levels between ARflox(ex3)/Y;Col 2.3-cre and control mice at 6 or 12 weeks of age (Table 4) or in serum DPD levels at 32 weeks of age (data not shown). Serum osteocalcin was decreased in ARflox(ex3)/Y, AR+/Y;Col 2.3-cre, and ARflox(ex3)/Y;Col 2.3-cre mice at 6 weeks of age compared with AR+/Y mice (p < 0.05; Table 4). No differences in serum osteocalcin were observed between ARflox(ex3)/Y;Col 2.3-cre and control mice at 12 or 32 weeks of age (Table 4). As expected, urine DPD and serum osteocalcin levels decreased with increasing age, independent of genotype (p < 0.001; Table 4).

Table Table 3.. Serum Hormone Levels in 9- and 32-Week-Old Mice
original image
Table Table 4.. Urinary DPD Levels Corrected for Creatinine (Cr) Levels in 6- and 12-Week-Old Mice and Serum Osteocalcin Levels in 6-, 12-, and 32-Week-Old Mice
original image

DISCUSSION

A large number of clinical and animal studies have shown the importance of androgens in male bone development and homeostasis. Despite evidence suggesting that estrogen (derived from testosterone) is an important sex steroid regulator of male bone metabolism, there is strong data to suggest that significant skeletal effects of androgens in males are mediated through the AR. For example, the male testicular feminized (Tfm) rat that is resistant to androgens because of a single base mutation in the ligand-binding domain of the AR gene(37) displays a female-like bone structure with reduced femoral length, diameter, and cortical thickness compared with wildtype male littermates.(38) Furthermore, the ER antagonist ICI 182780, which shows high affinity for both ERα and ERβ,(39) has no effect on bone growth in young male rats, suggesting that, in growing males, androgens act mainly through the AR in the accumulation of normal peak bone mass.(40)

This study describes the generation and bone phenotype of ARflox(ex3)/Y;Col 2.3-cre male mice, which express the second zinc finger–deficient AR specifically in mature osteoblasts. This model was designed to study the hypothesis that the major mechanism of androgen action on bone is through DNA binding–dependent actions of the AR in mature osteoblasts by deleting exon 3 of the AR.(23) We used Col 2.3-cre mice to drive expression of Cre recombinase specifically to mature osteoblasts and osteocytes. Col 2.3-cre mice express high levels of Cre in mature osteoblasts and osteocytes in trabecular and cortical compartments, but do not express Cre in preosteoblasts or the outer fibrous layer of the periosteum.(26) Examination of RNA isolated from the long bones of ARflox(ex3)/Y;Col 2.3-cre mice confirmed deletion of exon 3 within bone. The presence of the wildtype band in the bone sample was not unexpected and most likely represents AR expression in pre-osteoblasts, proliferating osteoblasts, bone marrow cells, and possibly osteoclasts.

We showed that 12- and 32-week-old ARflox(ex3)/Y;Col 2.3-cre male mice have decreased trabecular number and connectivity density, suggesting an increase in bone resorption. Despite a compensatory increase in trabecular thickness, this was insufficient to prevent a net loss of trabecular bone volume in the vertebra and the femur at 32 weeks of age. Our data are in accordance with previous studies using global ARKO male mice, which exhibited high bone turnover with increased bone resorption leading to reduced trabecular bone mass.(16) Furthermore, the nonaromatizable androgen DHT has been shown to inhibit biochemical bone turnover markers in mature female rats.(17) Taken together, these results suggest that androgens act on mature osteoblasts and/or osteocytes to modulate osteoclast activity. Although we did not detect any changes in the bone resorption marker, urine and serum DPD, the bone turnover marker, serum osteocalcin, or in the dynamic histomorphometric parameters measured in the distal femoral metaphysis between ARflox(ex3)/Y;Col 2.3-cre mice and the control groups, this may be because of the fact that the magnitude of the observed changes in bone structure in the mutant mice compared with control mice is small (e.g., trabecular BV/TV was reduced by 17% in ARflox(ex3)/Y;Col 2.3-cre compared with ARflox(ex3)/Y control mice in the vertebrae at 32 weeks of age) and was mostly only detected by μCT, the more sensitive technique. Furthermore, trabecular bone of the vertebrae seemed to be more sensitive to the effects of the mutation than the femora in our model. As expected, however, the biochemical markers, urine DPD and serum osteocalcin, were decreased at 12 and 32 weeks of age compared with levels measured at 6 weeks of age, consistent with the age-dependent reduction in bone turnover that occurs in mice on aging.(41,42) The unexpectedly high levels of serum osteocalcin in the wildtype mice at 6 weeks of age was most likely caused by the small numbers analyzed in each group, as a result of the limited availability of serum for the biochemical analyses.

During skeletal growth, androgens stimulate periosteal apposition, and therefore, males have larger bones with greater cortical thickness than females.(3) To what extent this is caused by direct activation of the AR remains unclear. We did not detect any changes in cortical thickness or periosteal circumference in our model, consistent with the lack of Cre expression at the periosteum in Col 2.3-Cre mice.(26)

In contrast to studies that have used global ARKO male mice (which are phenotypically female),(16,23,43) we generated a cell type- and stage-specific model that has male external genitalia, normal male body weight, and normal serum testosterone and estradiol levels. Although we observed differences in estradiol levels between groups in 32-week-old mice, we did not detect significant differences using the appropriate statistical analyses. This is most likely because of the large variation in values obtained within groups. It is possible that the lowered estradiol levels are contributing in part to the effect seen on BV/TV.(44–46)

Great consideration was taken when selecting control groups for this study. Because the Cre/loxP system is comprised of two components, a Cre line and a floxed line, we included hemizygous floxed and hemizygous Cre control groups in our analyses. In theory, floxed genes should function normally until excision after expression of Cre. Some studies, however, have shown that the presence of a selection marker (such as a floxed pgk-neo cassette) can interfere with normal gene function and result in the generation of a hypomorphic allele.(47–50) We have previously shown that our exon 3 floxed line (which retains the pgk-neo cassette) have normal external genitalia and are fertile, characteristics that are documented to be significantly affected by small decreases in AR function.(23,51) In fact, we recently showed that the ARflox(ex3)/Y mice show a disproportionate increase in some androgen-dependent organ weights, despite normal serum testosterone levels (HE MacLean, WSM Chiv, C Ma, AJ Notini, RA Davey, JF McManus, JO Zajac, unpublished data, 2006), suggesting that the floxed AR may actually increase AR sensitivity. As such, we believe that ARflox(ex3)/Y mice are the most appropriate control group for our study because the ARflox(ex3)/Y;Col 2.3-cre mice are on a floxed background. This may explain why some of the changes we observed are significantly different only compared with the floxed control group.

The skeletal phenotype of our ARflox(ex3)/Y;Col 2.3-cre mice was less marked than predicted by previous studies, suggesting that the classical AR nuclear signaling pathways in mature osteoblasts and osteocytes are not as important for skeletal status as AR activity during growth in immature osteoblasts or other bone cells such as osteoclasts. As noted earlier, we predict that the nonclassical mechanisms of androgen action are still functional in our model and may have some impact on mature osteoblasts. Furthermore, we used a Cre line in which Cre is expressed only in mature osteoblasts. Most in vitro studies have shown that androgens (including DHT) enhance proliferation of osteoblast progenitors.(11,52) It is possible, therefore, that the effect of inactivating AR signaling pathways in osteoblast precursor cells may be more dramatic than inactivating the AR in mature osteoblasts. Supportive evidence for this hypothesis is provided from recent findings in transgenic mouse lines in which the AR is overexpressed in all cells of the osteoblast lineage, including those located at the periosteum or only in mature osteoblasts and osteocytes using the Col3.6 and Col2.3 promoters, respectively. Overexpression of the AR in stromal cells and throughout the osteoblast lineage (Col3.6) resulted in a more dramatic bone phenotype compared with overexpression in mature osteoblasts (Col2.3), including marked calvarial thickening, reduced femur length, and reduced whole bone strength.(53) As discussed earlier, previous studies using ROSA26 indicator mice have shown that the Col 2.3 promoter is a strong promoter that is activated in mature osteoblasts and osteocytes. We therefore predict that osteoblast-specific deletion of exon 3 is virtually complete in our model. However, it is difficult to directly quantitate the degree of AR deletion within mature osteoblasts in our model in vivo because bone is a heterogeneous tissue comprised of a number of AR-expressing cells and deletion of exon 3 does not result in absence of AR mRNA or protein. Therefore, it is possible that the deletion of exon 3 in our model is mosaic, which may contribute to the mild phenotype. Finally, androgens may also act directly on other cells within bone, such as osteoclasts, although evidence for AR expression in osteoclasts has been conflicting.(6,15,54,55)

The skeletal phenotype of models such as Tfm rodents and global ARKO male mice could, in part, be explained by low levels of testosterone.(25) In contrast, testosterone levels are normal in our model. Further study is needed to assess the contribution of estrogen and the ER in ARflox(ex3)/Y;Col 2.3-cre male mice.

In conclusion, this study showed that the effects of androgens on trabecular bone are, at least in part, mediated directly through the AR in mature osteoblasts. Deletion of exon 3 of the AR in male mice resulted in reduced trabecular bone volume, which was associated with decreased trabecular number and connectivity density, suggesting increased bone resorption and reduced bone quality in adult mice. Furthermore, the mechanism of action requires the second zinc finger and is therefore dependent on DNA-binding activities of the AR.

Acknowledgements

The authors thank Sonia Dunn and Adnan Mulaibrahimovic of the Hanson Institute for excellent technical assistance and Helen MacLean of the Department of Medicine (AH/NH) for expert advice and helpful discussions. This study was supported by The National Health and Medical Research Council, The Cass Foundation Ltd., Sir Edward Dunlop Medical Research Foundation, The Austin Hospital Medical Research Foundation, an Endocrine Research Grant provided by Eli Lilly Australia and an Eva and Les Erdi Major Research Grant. AN was supported by an Australian Postgraduate Award.

Ancillary