Female Mice Haploinsufficient for an Inactivated Androgen Receptor (AR) Exhibit Age-Dependent Defects That Resemble the AR Null Phenotype of Dysfunctional Late Follicle Development, Ovulation, and Fertility

The role of classical genomic androgen receptor (AR) mediated actions in female reproductive physiology remains unclear. Female mice homozygous for an in-frame deletion of exon 3 of the Ar (AR^−/−) were subfertile, exhibiting delayed production of their first litter (AR^+/+ = 22 d vs. AR^−/− = 61 d, P < 0.05) and producing 60% fewer pups/litter (AR^+/+: 8.1 ± 0.4 vs. AR^−/−: 3.2 ± 0.9, P < 0.01). Heterozygous females (AR^+/−) exhibited an age-dependent 55% reduction (P < 0.01) in pups per litter, evident from 6 months of age (P < 0.05), compared with AR^+/+, indicating a significant gene dosage effect on female fertility. Ovulation was defective with a significant reduction in corpora lutea numbers (48–79%, P < 0.01) in 10- to 12- and 26-wk-old AR^+/− and AR^−/− females and a 57% reduction in oocytes recovered from naturally mated AR^−/− females (AR^+/+: 9.8 ± 1.0 vs. AR^−/−: 4.2 ± 1.2, P < 0.01); however, early embryo development to the two-cell stage was unaltered. The delay in first litter, reduction in natural ovulation rate, and aromatase expression in AR^+/− and AR^−/− ovaries, coupled with the restored ovulation rate by gonadotropin hyperstimulation in AR^−/− females, suggest aberrant gonadotropin regulation. A 2.7-fold increase (AR^+/+: 35.4 ± 13.4 vs. AR^−/−: 93.9 ± 6.1, P < 0.01) in morphologically unhealthy antral follicles demonstrated deficiencies in late follicular development, although growing follicle populations and growth rates were unaltered. This novel model reveals that classical genomic AR action is critical for normal ovarian function, although not for follicle depletion and that haploinsufficiency for an inactivated AR may contribute to a premature reduction in female fecundity.

THE ANDROGEN RECEPTOR (AR) gene, located on the X chromosome and a member of the nuclear receptor superfamily, has a pivotal role in male reproduction (1, 2). However, at present there remains limited understanding of its role(s) in female reproductive physiology, although both androgen excess and deficiency have ascribed impacts. Polycystic ovary syndrome (PCOS) is associated with excessive androgens and anovulation causing infertility (3, 4), which suggests that deleterious effects of androgens on follicle development and ovulation may lead to subfertility. Conversely, androgen deficiency due to adrenal insufficiency impacts on human female sexual function and well-being (5, 6), although effects on fertility remain unclear.

Although androgens are obligatory precursors for conversion to estrogens by aromatase (7), the importance and exact function(s) of additional direct effects of androgens mediated via genomic AR or nongenomic mechanisms on follicle or oocyte development and function remain unclear. However, AR-mediated actions within the ovary are thought to play important roles because androgens have been indicated as directly augmenting FSH signaling via a post-cAMP mechanism (8). The direct actions of androgens via ARs is supported by the universality of AR expression in mammalian ovaries such as in rodents (9, 10), domestic species (11, 12), and primates (13, 14). In rat and primate ovaries (14–16), AR is predominantly expressed in granulosa cells, with expression being highest in preantral/early antral follicles and gradually decreasing as the follicles mature (16, 17). Nevertheless, in vitro and in vivo investigations into the direct effect of androgens on follicular development have given conflicting findings showing inhibition of follicular development and increased granulosa apoptosis (18), modulation of gonadotrophin-stimulated ovarian function (19), antagonism of antiapoptotic effects of estradiol (E2) (20), but also stimulation of follicle growth and survival in primate (21, 22) or mouse (23) and IGF paracrine mechanisms during follicle growth (24). Furthermore, AR is known to be expressed in the brain and regulated by testosterone and estradiol (25). Hence, AR signaling has the potential to influence feedback mechanisms regulating the hypothalamic GnRH and pituitary LH and FSH release, including the ovulatory LH surge.

Development of genetic models of androgen insensitivity created by functional inactivation of the AR in females, allows in vivo investigation of the roles of AR mediated actions on reproductive function. However, because the AR gene is encoded on the X chromosome and hemizygous males with an inactive AR (the classical tfm mouse phenotype) are sterile, homozygous AR null females cannot be produced by natural mating. The first experimental model for female androgen insensitivity was created by breeding males chimeric for the tfm gene with female mice heterozygote for the tfm mutation to produce homozygous Tfm/Tfm female mice (26). These mice exhibited increased follicle atresia and reduced follicle numbers, but AR-mediated androgen action was not absolutely essential for ovulation, mating, pregnancy, or lactation (26, 27). More recently, using the Cre/LoxP system, female mice homozygous for deletion of the AR protein due to deletion of exon 1 (28) or 2 (29, 30) and insertion of a premature stop codon have been reported. In these models in which there is either no or a severely truncated AR protein produced, AR^−/− female mice remain fertile with fewer pups per litter and increased follicular atresia (28–30). Moreover, follicle development was reportedly unaffected in young (up to 16 wk of age) AR^−/− mice (28, 29), but older females exhibited accelerated follicle depletion, which led to a total loss of follicles at 40 wk of age (28). However, although gene alterations likely to impact follicle development were highlighted (28), the definitive mechanism of the subfertility at younger ages was not defined despite severe and accelerated age-related follicle depletion.

In the present study, homozygous AR^−/− female mice were generated using Cre/LoxP recombination for an in-frame excision of exon 3, encoding the second zinc finger essential for DNA-binding. Exon 3 of the AR gene was targeted because this encoding region is highly conserved among species (31), and several independent spontaneous mutational deletions in humans all produced complete androgen insensitivity syndrome (32, 34), a finding replicated in the mouse with this exon 3 deletion (31). (Also see the McGill University Androgen Receptor Gene Mutations Database at http://androgendb.mcgill.ca/.) Therefore, by the in-frame deletion of exon 3, the normal reading frame is retained allowing the protein produced to have seven of the eight exons intact, unlike previous AR^−/− studies (28–30) that deleted at least six exons. In vitro studies have revealed that the protein produced by point mutations or deletion in exon 3 has normal androgen binding affinity and is localized in the nucleus but has markedly reduced DNA-binding affinity and fails to activate transcription of an androgen-responsive reporter gene (32, 34, 35). Therefore, the exon 3 deleted mutant AR protein produced is nonfunctional as a nuclear transcription factor and can feasibly be held responsible for the results exhibited in this model. Hence, the present model provides a novel system for a detailed analysis of the direct effects of a deficiency in classical genomic AR signaling on female fertility and ovarian follicular growth and health. This study provides evidence that classical genomic AR mediated signaling plays important roles in regulating late stages of follicular development and in particular ovulation but is not necessary for preservation of ovarian follicle populations during aging.

Materials and Methods

Mice

Mice were housed under standard conditions (19–22 C, 12 h light, 12 h dark cycle) in cages with ad libitum access to water and food. Female mice were killed by cardiac exsanguination under ketamine/xylazine anesthesia, then organs dissected, and weighed before fixation (4% paraformaldehyde at 4 C overnight) for histological processing and collected serum stored frozen −20 C. All experiments and procedures were approved by the Sydney South West Area Health Service Animal Welfare Committee within National Health and Medical Research Council guidelines for animal experimentation.

Generation of AR knockout mice

Androgen receptor knockout mice (AR^−/−) and heterozygous mice (AR^+/−) were generated by crossing ARflox mice [exon 3 flanked by loxP sites, as previously described (31)] with transgenic cytomegalovirus (CMV)-Cre mice as a universal deletor. Because the CMV-cre transgene (36) as well as the AR are both located on the X chromosome, a two-step breeding strategy was required. First, ARflox and CMV-cre mice were crossed to obtain heterozygous (AR^+/−) female offspring expressing CMV-cre. Second, these AR^+/− females carrying CMV-cre were mated with males containing ARflox, with meiotic cross-over allowing production of females with complete homozygous knockout of the AR exon 3.

DNA and RNA extraction, genotyping, and RT-PCR

Genomic DNA isolated from toe clip or tail biopsy, as previously described (37), was used as a template for PCR genotyping to detect rearrangements in the mouse AR gene. Forward PCR primers upstream of the first LoxP site (ARExcised-forward, CAGAAATCCACCTGCCTCTACC), within mouse AR exon 3 (AREx3-forward, CTTCTCTCAGGGAAACAGAAGT) and within NEO cassette (ARNeo-forward, TAGATCTCTCGTGGGATCATTG) were used with a common reverse primer located within intron 3 (AR-reverse, GGGAGACACAGGATAGGAAATT) (Fig. 1A). The conditions for the PCR were as follows: 94 C for 5 min followed by 39 cycles of 94 C for 1 min, 58 C for 1 min, 72 C for 1 min, and final extension at 72 C for 5 min. Mice containing the CMV-cre gene were detected using primers and PCR conditions as described (36). AR^−/− males and females were distinguished, using primers and PCR conditions for the mouse Y chromosome sry gene as previously described (31).

Fig. 1

Characterization of AR-deficient mice. A, Schematic diagram of intact, floxed, and excised AR exon 3, showing location of primer pairs used for PCR genotype identification from genomic DNA. Representation of PCR genotyping by which all mice (AR^+/+, AR^+/−, and AR^−/−) were genotyped using genomic DNA. The intact AR exon 3 product (AREx3) was 613 bp, the floxed AR exon 3 product was 289 bp, and the excised AR exon 3 showed no product. Representation of PCR genotyping for AR^+/+ and AR^−/− mice using genomic DNA shows positive results for the genotyping of AR^−/− mice. The intact AR exon 3 product was 1650 bp, and the excised AR exon 3 had a product of 510 bp. B, Schematic diagram of intact and excised AR exon 3, showing location of primer pairs used for PCR genotype identification from RNA. Representation of RT-PCR analyses using cDNA from ovaries of 6-wk-old mice confirm that ovaries from AR^−/− females carried the global exon 3-deleted transcripts. The exon 3 PCR product is 262 bp, intact AR exon 3 product is 288 bp, and excised AR exon 3 product is 171 bp. Mouse β-actin, used as an internal control, had a product of 431 bp. C, AR immunohistochemistry in AR^+/+ (A), AR^+/− (B), and AR^−/− (C) ovaries. D is the negative control. Sections were counterstained with Toludine Blue.

Genotyping of AR^−/−, AR^+/−, and AR^+/+ female mice was confirmed by RT-PCR on the RNA extracted from ovaries of 6- and 17-wk old mice (n = 3/genotype). Reverse transcription was performed with 2 μg of total RNA using SuperScript III first-strand synthesis system (Invitrogen, Carlsbad, CA). Subsequent PCR was carried out on 0.5 μl of the cDNA, and the products were analyzed by electrophoresis on a 1.5% agarose gel. The 5′-GGA CAG TAC CAG GGA CCA T-3′ and 5′GTC TCA GTA GGG ACG AAG TA-3′ primers identified the presence (product size 262 bp) or absence (no band observed) of exon 3, and 5′-GGA CAG TAC CAG GGA CCA T-3′ and 5′-CCA AGT TTC TTC AGC TTA CGA-3′ primers identified intact and excised exon 3 AR (Fig. 1B). The conditions for the PCR were as follows: 94 C for 2 min followed by 35 cycles of 94 C for 5 sec, 60 C for 15 sec, 72 C for 30 sec, and final extension at 72 C for 2 min. Two product sizes were obtained, 288 bp for intact AR and 171 bp for Cre-mediated exon 3-excised AR. Primers specific for mouse β-actin were used as an internal control.

Quantitative real-time RT-PCR

Using real-time RT-PCR, mRNA expression for steroidogenic acute regulatory protein, P450side-chain cleavage, P450c17, and P450aromatase was carried out on ovaries collect after natural mating and the presence of a copulatory plug to evaluate possible impairments in steroidogenic enzymes. Ovaries collected at diestrus were assessed histologically for follicle health and gene expression was analyzed for B cell leukemia/lymphoma-2 (Bcl-2) (prosurvival) and Bcl-2-associated X protein (Bax) (proapoptotic). To assess whether the observed effects present in this model may be due to estrogenic effects mediated by 5α-androstandiols, mRNA expression for the following enzymes that carry out the conversion of androgens to estrogens were analyzed: 5α-reductase type I, 5α-reductase type II, 3β-hydroxysteroid dehydrogenase (HSD), and 3α-HSD. Total RNA was extracted from whole ovaries from all genotypes [(diestrous samples (n = 6/genotype) and estrous samples (n = 5/genotype)] using TriReagent (Sigma, Castle Hill, Australia) according to the manufacturer’s protocol. Reverse transcription was performed with 2 μg of total RNA using SuperScript III first-strand synthesis system (Invitrogen). Quantitative RT-PCR analysis on resulting cDNA was performed in duplicate using Corbett RotorGene 2000 real-time amplification system (Corbett Research, Sydney, Australia). Specific commercially purchased SYBR Green primers were used with SYBR Green master mix (Superarray Bioscience Corp., Bethesda, MD). Mouse ribosomal protein was used as a housekeeping gene. There was no significant differences in mouse ribosomal protein mRNA levels between treatment groups (P > 0.05). Reaction steps were 15 min at 95 C, followed by 40 cycles of denaturation at 95 C for 30 sec, annealing at 55 C for 30 sec, and extension at 72 C for 30 sec. A standard curve was generated for each gene from five serial dilutions of purified PCR product. Standards were assigned an arbitrary value and mRNA expression determined relative to housekeeping gene expression in each sample. No template controls, substituting water for cDNA, and a negative reverse transcription were included in each run.

AR immunohistochemistry

Ovaries from 10- to 12-wk-old mice were collected, fixed in 4% paraformaldehyde, dehydrated in alcohols, cleared in xylene, embedded in paraffin, and serially sectioned (5 μm). Ovarian sections were subjected to microwave (600 W) antigen retrieval in 10 mm citrate buffer (pH 6.0) for 15 min. Endogenous peroxidase activity was blocked by incubating sections in 3% hydrogen peroxide for 5 min. To block nonspecific binding, sections were incubated in 1% BSA for 1 h at room temperature followed by goat serum for 30 min. Sections were then incubated with a polyclonal anti-AR antibody (N-20) (Santa Cruz Biotechnology, Santa Cruz, CA) (1:100 dilution) in 1% BSA/PBS for 1 h at 37 C. Negative controls were incubated with a control primary antibody (Santa Cruz rabbit IgG). Antibody binding was visualized using the Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA) followed by color development using liquid DAB+ (Dako, Botany, Australia), according to the manufacturer’s instructions. Sections were briefly counterstained with 0.5% Toludine Blue, dehydrated through ethanol series and xylene, and mounted.

Fertility

To estimate natural fertility 8- to 10-wk-old females [AR^+/+ (n = 5), AR^+/− (n = 8), and AR^−/− (n = 9)] were continuously mated with an individual ARflox male (at least 8 wk old) for a 10-month period. Cages were monitored daily and the number of pups and litters were recorded.

Classification and enumeration of follicles

The follicle classification system used was based on the system used by Myers et al. (38). Briefly, primordial consisted of an oocyte surrounded by flattened granulosa cells, primary consisted of an oocyte surrounded by one layer of cuboidal granulosa cells, small preantral contained an oocyte with 1.5–2 layers of cuboidal granulosa cells, large preantral were classified by having an oocyte surrounded by more than two and up to five layers of cuboidal granulosa cells, small antral contained an oocyte surrounded with more than five layers of cuboidal granulosa cells, and/or one or two small areas of follicular fluid whereas large antral contained a single large antral cavity, and preovulatory possessed a single large antrum and an oocyte surrounded by cumulus cells at the end of a stalk of mural granulosa cells. Corpus lutea (CL) were identified by morphological properties consistent with luteinized follicles and by being visible throughout several serial sections. Zona pellucida remnants (ZPRs) (representing end-stage atretic follicles) were defined as previously described (38).

One ovary per animal was collected from 10- to 12-, 26-, and 52-wk-old mice at diestrus, fixed, and then processed through graded alcohols into glycolmethacrylate resin (Technovit 7100; Heraeus Kulzer, Chatswood, Australia). Ovaries were serially sectioned at 20 μm, stained with periodic acid-Schiff, and counterstained with hematoxylin. Follicles were counted using a microscope (Olympus, Tokyo, Japan) with Stereo Investigator software (MicroBrightField, Williston, VT). Total number of growing follicles per ovary at different developmental stages for each of the genotypes (minimum three ovaries counted per genotype per time point) was determined as previously described (38). For all histological analysis, repetitive counting of follicles was avoided by counting/measuring only follicles containing an oocyte with a visible nucleolus. To avoid bias, all ovaries were analyzed without knowledge of genotype.

Measurement of follicular growth and health

One ovary per mouse was collected at diestrus from mice at 10–12 wk of age, fixed, dehydrated in alcohols, cleared in xylene, embedded in paraffin, serially sectioned (5 μm), stained with hematoxylin and eosin or proliferating cell nuclear antigen (PCNA; minimum three ovaries/genotype), and every 10th section of each ovary was analyzed.

The growth rate of follicles was estimated by measuring diameters of all follicles, and their contained oocyte, in two perpendicular axes using a light microscope at ×20, calibrated using Stereo Investigator computer software (MicroBrightfFeld). The mean diameter at each follicular (or oocyte) developmental stage was calculated.

Cellular proliferation in the ovary was estimated by PCNA staining using a commercially available kit (Zymed PCNA detection kit; Invitrogen) according to the manufacturer’s instructions. In all experiments mouse primary antibody isotype control (Invitrogen) as a negative control and mouse intestinal tissue as a positive control were run in parallel. Follicles were classified, as above and then granulosa cell nuclei within each follicle positively stained for PCNA (positive staining was defined as intense brown staining of the nucleus) were counted. The average percentage of proliferating granulosa cells for each follicular development stage/ovary, and similarly, the percentage of theca proliferating cells/ovary were calculated.

Unhealthy follicles at different developmental stages were classified according to strict morphometric criteria if they contained a degenerate oocyte, disorganized granulosa cell layers, and/or more than 10% of the granulosa cells being pyknotic in appearance as previously described (39). The proportion of unhealthy follicles per ovary was estimated as the percentage of all follicles.

Measurement of ovulation and oocyte viability

To quantify numbers of ovulated oocytes a minimum of six mice per genotype (6–8 wk old) were treated with 10 IU pregnant mare serum gonadotropin and 46 h later 10 IU pregnant mare serum gonadotropin (CenVet, Artarmon, New South Wales, Australia), mimicking an endogenous LH surge to induce ovulation. Ovulation was examined 16 h after human chorionic gonadotropin injection. Oocyte-cumulus complexes (OCCs) were collected from the oviducts and counted.

To evaluate in vivo fertilization after natural mating, 8-wk-old female mice (minimum of five per genotype) were housed with a known fertile wild-type (AR^+/Y) male mouse. Cages were monitored daily, and on the identification of a copulatory plug, female mice were killed immediately, OCCs collected from the oviducts, counted, and placed in M2 medium (Sigma). Fertilization was identified by the extrusion of the second polar body before embryos were cultured overnight at 37 C with 5% CO₂ in M16 medium (Sigma) to assess the number of embryos reaching cleavage to the two-cell stage.

Hormone assays

Serum samples (at least six per genotype and time point) for hormone assay were obtained at diestrus as determined by light microscopic analysis of vaginal epithelial cell smears collected daily in 20 μl of sterile PBS and then transferred to glass slides, air dried, and stained with 0.05% Trypan Blue.

Mouse serum FSH and LH were determined using species-specific immunofluorometric assays (DELFIA; PerkinElmer, Turku, Finland) as described previously (40, 41). Serum estradiol levels were measured by time-resolved immunofluorometric (DELFIA) and testosterone assays by in-house RIA in extracts of serum as described previously (42, 43) with detection limits of 20 pm and 0.1 nm and intraassay variability of 12.7 and 7.5% for estradiol and testosterone assays, respectively. Cross-reactivities for testosterone were: 5α-androstane-3β-diol, 0.96%; 5α-androstane-3α-17β-diol, 0.51%; androstanedione, 0.29%; androstenediol, 0.20%; dehydroisoandrostenne, 0.011%; cortisol, E2, androsterone, dehydroepiandrosterone sulfate, less than 0.009%. Estradiol cross-reactivities were: estrone, estrone-sulfate, estriol, 2-hydroxyestriol, 16-OxoE2, 16OH-estrone, E2-SO4, E-glucuronide, less than 1%; testosterone, progesterone, dehydroepiandrosterone, cortisol, less than 0.01%. Serum or ovarian homogenates were extracted with (1:100 volumes) diethyl ether for estradiol or hexane-ethyl acetate (3:2) for testosterone.

Ovaries were homogenized in 100 μl volume of PBS + 0.5% BSA and 0.5 mm EDTA buffer on ice using a pestle. Homogenates were extracted with 10 volumes of organic solvent (1000 μl of organic solvent per 100 μl of tissue homogenate), vortexed thoroughly for approximately 2 min and allowed to stand at 4 C for 1 h. Samples were then centrifuged for 10 min (4 C, 3000 rpm) before freezing the aqueous layer in a dry ice-ethanol mixture. The organic layer was removed into a clean glass test tube and dried overnight at 37 C. The extract was reconstituted in 100 μl of 1% gelatin PBS (pH 7.2) by thorough vortexing before reextraction by adding a further 1000 μl of organic solvent and repeating the extraction and drying procedure. The dried residue was then gently dissolved in 50 μl of estradiol or testosterone assay buffer, and the reconstituted extract was assayed as for the serum extracts (42, 43).

Statistical analysis

Statistical analysis was performed using NCSS 2004 software (NCSS Statistical Software, Kaysville, UT). Unless otherwise stated, all results were expressed as means ± sem. Statistical differences were tested by ANOVA with post hoc test using Fisher’s least significant differences multiple-comparison test. All parametric tests were confirmed by nonparametric equivalent tests. The proportion of females to have produced their first litter and10th pup was analyzed by Fishers exact test, whereas the time in days each breeder took to produce their first litter and 10th pup was analyzed with the use of Kaplan-Meier analysis using log rank test to compare groups. P values smaller than 0.05 were considered statistically significant.

Results

Generation and verification of AR^+/+, AR^+/−, and AR^−/− female mice

There was no sex ratio bias with all breeders producing pups with a male to female ratio of approximately 50:50 (percent of females/litter: AR^+/+: 49.9; AR^+/−: 49.1; and AR^−/−: 45.1) (data not shown). The AR deletion in exon 3 of the AR gene was verified using genomic DNA by PCR using a primer within exon 3, which is deleted during recombination. AR with intact exon 3 was undetectable in the genomic DNA from AR^−/− toe or tail clippings by PCR (Fig. 1A). To confirm the global excision of exon 3 was also true of the ovary, ovarian RNA was subjected to RT-PCR, showing no AR with intact exon 3 but only AR with excised exon 3 (Fig. 1B). To confirm that the in-frame deletion of exon 3 did not disrupt protein production, AR immunohistochemistry showed the presence of AR protein expression expressed in follicles in all three genotypes (Fig. 1C).

Fertility of AR-deficient females

Litter size was markedly reduced by 60% in AR^−/− (3.2 ± 0.9 pups/litter) (P < 0.01), compared with AR^+/+ (average 8.1 ± 0.4 pups/litter) and AR^+/− (average 6.4 ± 0.6 pups/litter) breeders over a 10-month breeding trial (Fig. 2A). AR^−/− female breeders also took longer to produce their first litter (median AR^+/+ = 22 d, AR^+/− = 21 d, and AR^−/− = 61 d, P < 0.05) (Fig. 2B) with 30% of AR^−/− breeders never producing a litter. Moreover, both AR^−/− and AR^+/− female breeders took longer than AR^+/+ to produce 10 pups (median AR^+/+ = 32 d, AR^+/− = 73 d, and AR^−/− = 87 d, P < 0.01) (Fig. 2C). There was no difference in the frequency of litters per month among the three genotypes.

Fig. 2

Effect of AR deficiency on breeding efficiency. In all graphs, AR^+/+ = black, AR^+/− = gray, and AR^−/− = white. A, Average number of pups/litter. Data are the mean ± sem (n ≥ 5/genotype). Different superscripts denote statistically significant differences (P < 0.01). B, Percentage of females to have produced their first litter and median time in days to first litter (n ≥ 5/genotype). Different superscripts denote statistically significant differences (P < 0.05). C, Percentage of females to have produced their 10th pup and median time in days to 10th pup (n ≥ 5/genotype). Different superscripts denote statistically significant differences (P < 0.01). D, Average cumulative number of pups/month over a 10-month period of continual mating. Data are the mean ± sem (n ≥ 5/genotype). Different superscripts denote statistically significant differences (P < 0.01).

Cumulative number of pups produced over the 10-month breeding trial period was reduced by 55% in AR^−/− (P < 0.01) and 44% in AR^+/− (P < 0.01) female breeders, both significantly lower than AR^+/+ female breeders (AR^+/+: 71 ± 8.1; AR^+/−: 40 ± 5.7 and AR^−/−: 32 ± 8.5, P < 0.01) (Fig. 2D). The AR^+/− females exhibited an age-dependent reduction in fertility evident from 6 months of age (35% reduction), compared with AR^+/+ (P < 0.05), indicating a gene dosage effect.

Phenotype of AR-deficient females

Ovarian weights were significantly reduced by 23% in AR^−/− and 26% AR^+/− female mice when compared with AR^+/+ controls at 5–7 wk of age (P < 0.01) and 17% in both AR^−/− and AR^+/− at 10–12 wk of age (P < 0.01) but not at 26 or 52 wk of age (Fig. 3A). Uterine and body weights did not differ according to genotype or age (data not shown). There was no overt effect of genotype on the macroscopic appearance of the ovaries and uterus. Furthermore, there was no difference in weights of somatic organs (kidney, heart, and liver) according to genotype or age nor among the AR^+/+, AR^+/−, or AR^−/− genotypes (data not shown).

Fig. 3

Effect of AR deficiency on ovary weights and follicle and CL numbers. In all graphs, AR^+/+ = black, AR^+/− = gray, and AR^−/− = white. A, Comparisons between AR^+/+, AR^+/−, and AR^−/− ovary weights at 5–7, 10–12, 26, and 52 wk of age. Data are the mean ± sem (n ≥ 6/genotype). Different superscripts denote statistically significant differences (P < 0.01). B, Average number of CL/ovary at the diestrous stage in AR^+/+, AR^+/−, and AR^−/− mice at 10–12, 26, and 52 wk of age. Numbers represent total counts from serially sectioned ovaries. Data are the mean ± sem (n ≥ 3/genotype). Different superscripts denote statistically significant differences (P < 0.01). Histological ovarian sections of AR^+/+, AR^+/−, and AR^−/− ovaries, CL are indicated by an asterisk. C, Average number of growing follicles/ovary at the diestrous stage in AR^+/+, AR^+/−, and AR^−/− mice at 10–12, 26, and 52 wk of age. Numbers represent total counts from serially sectioned ovaries. Data are the mean ± sem (n ≥ 3/genotype).

Effect of AR deficiency on follicle and CL numbers

Analysis of histological sections of ovaries showed that the full range of expected stages of follicle development was present, with overall normal morphology regardless of genotype. At 10–12 wk of age, CL numbers, at the diestrous stage of the estrous cycle, were significantly reduced by 79% in AR^−/− (P < 0.01) and 48% in AR^+/− (P < 0.01) female mice and 68% in AR^−/− (P < 0.01) and 55% in AR^+/− (P < 0.01) female mice at 26 wk of age, compared with AR^+/+ (Fig. 3B). Quantitative analysis showed there was no significant difference in follicle population numbers between genotypes in ovaries from 10- to 12-, 26-, and 52-wk-old mice (Fig. 3C).

Effect of AR deficiency on hyperstimulated and natural ovulation, fertilization, and early embryo development

The number of oocytes collected from the oviducts of superovulated mice was significantly increased in AR^+/− but did not differ between AR^−/− or AR^+/+ genotypes (AR^+/+: 29.5 ± 2.0; AR^+/−: 39.3 ± 3.3; AR^−/−: 25.2 ± 4.1, P < 0.05) (Fig. 4A).

Fig. 4

Effect of AR deficiency on hyperstimulated and natural ovulation, fertilization, and early embryo development. In all graphs, AR^+/+ = black, AR^+/− = gray, and AR^−/− = white. A, Average number of oocytes collected from superovulated AR^+/+, AR^+/−, and AR^−/− mice. Data are the mean ± sem (n ≥ 6/genotype). Different superscripts denote statistically significant differences (P < 0.05). B, Average number of naturally ovulated oocytes collected from AR^+/+, AR^+/−, and AR^−/− mice. Data are the mean ± sem (n ≥ 5/genotype). Different superscripts denote statistically significant differences (P < 0.01). C, Percentage of oocytes to be fertilized after natural mating and progress to the two-cell stage of early embryo development after culture overnight. Data are the mean ± sem (n ≥ 5/genotype). D, Real-time RT-PCR analysis of the expression levels of the major steroidogenic enzymes. Data are the mean ± sem (n ≥ 5/genotype). Different superscripts denote statistically significant differences (P < 0.05). StAR, Steroidogenic acute regulatory protein; Cyp11a1, cytochrome P450 side-chain cleavage; Cyp17a1, cytochrome P450 17α-hydroxylase; Cyp19a1, P450 aromatase.

The number of oocytes recovered from oviducts of naturally mated AR^−/− females was significantly decreased by 57%, compared with AR^+/+ controls (AR^+/+: 9.8 ± 21.0; AR^+/−: 7.3 ± 0.8; AR^−/−: 4.2 ± 1.2, P < 0.01) (Fig. 4B), whereas viability of eggs (proportion of ovulated eggs to be successfully fertilized) and embryos (proportion undergoing cleavage to the two-cell stage) did not differ between genotypes (Fig. 4C).

Real-time RT-PCR on ovaries collected from mice that have naturally mated detected that aromatase expression was significantly decreased by 82% in AR^+/− and 71% in AR^−/− ovaries, compared with AR^+/+ ovaries (P < 0.05). There was no difference in expression levels of steroidogenic acute regulatory protein, P450scc, or P450c17 between genotypes (Fig. 4D).

Effects of AR deficiency on serum FSH, LH E2, and testosterone (T) concentrations and intraovarian T and E2 levels

Levels of baseline serum FSH, LH, E2, and T were not significantly different among the three genotypes at 10–12 and 26 wk of age at the diestrous stage of the estrous cycle (Fig. 5A).

Fig. 5

Hormone levels of AR-deficient mice. In all graphs, AR^+/+ = black, AR^+/− = gray, and AR^−/− = white. A, Serum hormone levels at the diestrous stage in AR^+/+, AR^+/−, and AR^−/− mice at 10–12 and 26 wk of age. Data are the mean ± sem (n ≥ 6/genotype). B, Intraovarian T and E2 levels at the diestrous stage in AR^+/+, AR^+/−, and AR^−/− mice at 10–12, 26, and 52 wk of age. Data are the mean ± sem (n ≥ 6/genotype). Different superscripts denote statistically significant differences (P < 0.05). C, Real-time RT-PCR analysis of the expression levels of the steroidogenic enzymes involved in the conversion of androgens to compounds that can act via the estrogen receptors. Data are the mean ± sem (n ≥ 6/genotype). Srd5a1, 5α-Reductase type I; Srd5a2, 5α-reductase type II; Hsd3b1, 3β-HSD; Akr1c14, 3α-HSD.

There was a 3.2-fold increase in intraovarian levels of T in AR^−/− female mice, compared with AR^+/+ controls at 10–12 wk of age at diestrus (AR^+/+: 3.5 ± 0.9 nm/mg; AR^+/−: 4.9 ± 2.0 nm/mg; and AR^−/−: 11.2 ± 3.0 nm/mg, P < 0.05) (Fig. 5B) but not at later ages. Intraovarian E2 did not differ between genotypes or ages (Fig. 5B).

Using mRNA collected from the ovaries of mice at diestrous, real-time RT-PCR detected that there was no difference in expression levels of 5α-reductase type I, 5α-reductase type II, 3β-HSD, or 3α-HSD between genotypes (Fig. 5C).

Effect of AR genotype on follicle growth and health

To investigate possible abnormalities in follicle growth and somatic cell proliferation, oocyte and follicle diameters and PCNA immunohistochemistry were assessed. There were no significant differences between genotypes in oocyte (data not shown) and follicle diameters (Fig. 6A) or granulosa or theca cell (data not shown) PCNA staining (Fig. 6B) at any stage of follicle development.

Fig. 6

Effect of AR deficiency on ovarian follicle growth and health. In all graphs, AR^+/+ = black, AR^+/− = gray, and AR^−/− = white. A, Average follicle diameters at different follicular developmental stages at diestrus. Data are the mean ± sem (n ≥ 4/genotype). B, Average percentage of granulosa stained for PCNA at different follicular developmental stages at diestrus. Data are the mean ± sem (n ≥ 3/genotype). C, Average percentage of histologically unhealthy follicles at different follicular developmental stages at diestrus. Data are the mean ± sem (n ≥ 3/genotype). Different superscripts denote statistically significant differences (P < 0.01). D, Average ZPRs/ovary from AR^+/+, AR^+/−, and AR^−/− mice at 10–12, 26, and 52 wk of age at the diestrous stage. Data are the mean ± sem (n ≥ 5/genotype). E, Real-time RT-PCR analysis of the expression levels of Bax (proapoptotic) and Bcl2 (prosurvival). Data are the mean ± sem (n ≥ 6/genotype).

In 10- to 12-wk-old mice, there was a 2.7-fold increase in morphologically unhealthy antral follicles within AR^−/− ovaries, compared with AR^+/+ controls (AR^+/+: 35.4 ± 13.4%; AR^+/−: 58.2 ± 4.8%; AR^−/−: 93.9 ± 6.1%, P < 0.01) (Fig. 6C).

ZPR numbers did not differ according to genotypes or age; however, ZPR numbers were nonsignificantly higher in the AR^−/− mouse ovaries at 26 week of age (AR^+/+: 28.2 ± 12.8; AR^+/−: 66.8 ± 22.9; AR^−/−: 206.6 ± 92, P = 0.09) (Fig. 6D).

Real-time RT-PCR analysis using mRNA collected from the ovaries of mice at diestrus showed that between genotypes there was no difference in expression levels of Bax or Bcl2 (Fig. 6E).

Discussion

Androgens have an indispensable role in the ovary as essential precursors of estrogen biosynthesis, therefore acting indirectly via the estrogen-α and -β receptors. In recent years, it has been identified that androgens also have additional direct effects mediated by the AR on follicle development and female fertility. The current study set out to selectively target classical genomic AR signaling pathways in vivo and investigate the importance of these AR-mediated actions on female fertility through effects on oocyte and follicular development and ovulation. AR^−/− mice were generated by an in-frame deletion of exon 3, which maintained the production of a largely intact AR protein, which is transcriptionally inactive and hence classically nonfunctional. Males hemizygous for this mutant AR confirmed the complete abolition of classic genomic AR function by exhibiting a classical tfm phenotype (31), similar to that observed in spontaneous (44) and targeted (30, 31) AR knockout models. The complete deletion of exon 3 of the AR in the ovaries of homozygous female (AR^−/−) mice used in this study was confirmed by PCR, which detected only AR with excised exon 3 and no intact exon 3 in genomic DNA or RNA.

In the present study, the observation that homozygous classical inactivation of the AR reduces but does not abolish female fertility is in agreement with previous studies (26, 28, 29). Our findings of a predominant defect in late follicle development and/or ovulation as the key mechanism(s) leading to the subfertility coupled with the novel finding that the reduction in fertility is without substantial loss in follicle populations, may also explain the previously observed reduced litter size in other studies of AR^−/− female mice produced either as embryonic chimeras (26) or major deletions in AR, resulting in production of no or extremely truncated AR protein (28–30). However, the relatively unaffected follicle development observed differs from the findings of an AR knockout model whereby the deletion within the AR gene results in no protein being produced, which reported that accelerated follicle depletion was the principal mechanism causing subfertility with age (28). By contrast, AR^−/− females maintaining the production of a transcriptional inactive AR protein in the present study produced an equally subfertile ovarian phenotype, but follicle populations remained undiminished with all follicular developmental stages present at 10 months of age.

The reduction in ovulation identified in the present and similar studies, by the reduction in CL numbers in AR^−/− (28, 29) and AR^+/− ovaries and decrease in naturally ovulated oocytes collected from oviducts of AR^−/− females in this study, may be due to intrinsic defects within the ovary, such as reduced responsiveness of the follicle or an extrinsic defect in the hypothalamic-pituitary regulation of gonadotropin secretion. This extraovarian defect is supported by the observed reduction in aromatase expression, an FSH regulated gene (8), suggesting a possible inhibition of FSH signaling; however, the aromatase down-regulation may also be due to disruption of other androgen-associated ovarian signaling pathways such as the IGF system. In the present study, there is an increase in unhealthy antral AR-deficient follicles, which may have either an impaired response to a normal or diminished ovulatory LH surge. Additionally, the novel observations in the present study that reduced ovulation rates can be overcome by gonadotrophin hyperstimulation together with the probable delayed onset of the first ovulatory estrus suggested by the later onset of first litter may indicate an extraovarian defect such as a defective LH surge, but further neuroendocrine investigation is required to clarify the details of this mechanism.

The identification of X-linked genes affecting ovulation rates in sheep highlights that ovulation rates and litter size can be genetically regulated (45). Striking examples of gene dosage affecting ovulation is the Inverdale (FecX) and Boroola (FecB) mutations in sheep, whereby homozygous female carriers of FecX are sterile with underdeveloped ovaries (46), whereas heterozygote FecX carriers and both homozygote and heterozygote carriers of the FecB gene have increased ovulation rates (47–49). AR^+/− female breeders were uniquely characterized in this study and found to display a reduction in CL numbers and an age-related subfertility, which was distinct from the pattern observed in the AR^−/−. AR^+/− breeders were observed to initially display similar fertility to homozygous normal (AR^+/+) females; however, over time their reproductive performance grew progressively poorer so that their breeding productivity curve converged with the AR^−/− females toward the end of the 10-month breeding experiment. This finding suggests that haploinsufficiency of the inactivated AR creates functional partial AR deficiency with significant effects on breeding efficiency, a previously unsuspected gene dosage effect of AR on female fertility. This may be consistent with observations that genetic polymorphisms determining different AR sensitivity correspond to androgen-sensitive disorders (50, 51). For example, the polymorphic CAG trinucleotide repeat in exon 1 of the AR determines AR sensitivity (or activity) and may influence the development of PCOS (52, 53). Hence, the AR^+/− female may prove to be a very useful tool in further characterization of the effects of AR deficiency in women. For example, our finding predicts that mothers of males with complete androgen insensitivity who are obligate heterozygotes for the same mutation may have curtailed reproductive function particularly at an older age.

Results from the current study showed that egg (fertilization) and early embryo viability (progression to two-cell stage) from naturally mated AR^−/− and AR^+/− females were unaltered, supporting our hypothesis that the major physiological role for androgens in follicular dynamics is regulating late stages of follicular development and ovulation. The observed effects of androgens during the later stages of follicle growth, oocyte maturation, and ovulation (54–58) along with the persistent high AR expression in the OCC and granulosa cells lying in close proximity to the OCC are consistent with a paracrine role of androgens in oocyte health and ovulation (59). Kit ligand and bone morphogenetic protein 15 gene expression, which are both involved in the oocyte-granulosa cell regulatory loop, have previously been observed to be reduced in AR^−/− ovaries (28). In the present study, oocyte health appeared to be maintained by the observed normal growth, fertilization, and cleavage rates. However, future investigation should analyze whether the crucial oocyte-somatic cell interactions, which ensure correct signaling for key ovulatory processes such as oocyte maturation and cumulus cell expansion, are maintained because in a previous AR knockout model, cumulus cells were found to be disassociated from the oocyte in preovulatory follicles (29).

Our results clearly showed that there was no drastic effect of AR deficiency on follicle populations present within the AR^−/− ovaries. At diestrus follicle numbers were equivalent to that observed in the AR^+/+ controls. However, this does not fully exclude difference in follicle dynamics at other stages of the estrous cycle. Despite androgens being implicated in enhancing follicle growth (60, 61), the present detailed analysis of follicle growth showed no differences. Hence, classical genomic AR-mediated actions do not appear to play essential roles in somatic cell proliferation. The overexpression of AR in estrogen receptor-β knockout mouse ovaries is associated with high levels of follicular atresia but can be reversed by the addition of an AR antagonist (9), suggesting that AR could play a vital role in regulating granulosa cell survival and hence protecting the follicle from undergoing follicular atresia. In the present study, a significant increase in the incidence of unhealthy antral follicles within AR^−/− ovaries at 10–12 and 26 wk of age suggests an increase in follicular atresia; however, this was not confirmed by ZPR counts or gene expression for the proapoptotic protein Bax, suggesting that follicle atresia is not the major cause of the observed reduced fertility in this model.

The normal serum levels of FSH, LH, E2, and T at diestrus remain consistent with our hypothesis of impaired late stages of follicle development and ovulation. However, taken together the high intraovarian T levels observed, especially in the youngest AR^−/− mice, with the present observations of poor follicle health and disrupted ovulation, suggests that ovarian androgens may initiate effects on the developing follicle by mechanisms not requiring the AR, such as T inducing maturation in mouse oocytes arrested in meiosis, independently of transcription (55). Our gene expression analysis observed no change in expression of androgen metabolizing enzymes (5α-reductase and 3α- and 3β-HSD), indicating that the intraovarian T is unlikely to exert its effects via conversion to dihydrotestosterone, or its further 3-hydroxyated metabolites (3α- or 3β-androstanediols), which might exert actions via an estrogen receptor. Hence, our finding supports the possibility that the high levels of T observed in the AR^−/− ovaries, which may be due to a reduction in aromatase, may exert long-term detrimental affects on follicle health and ovulatory processes, consistent with previous findings that treatment of rat granulosa cells with dihydrotestosterone at doses greater than 1 mg/rat inhibits LH receptor induction and hence ovulatory responses (18). Furthermore, in women with elevated levels of ovarian androgens that promote abnormal ovarian growth and development of polycystic ovaries containing follicles unable to ovulate, treatment with the AR antagonist flutamide improves fertility in some individuals, which is consistent with AR-mediating critical actions during late follicle development and ovulation (33, 62).

In summary, we have shown that a loss of classical genomic AR signaling in the ovary leads to subfertility, dysfunctional late follicle development, and a major disruption in ovulation rates. Hence, classical genomic AR-mediated actions have a key involvement in optimizing late follicle development and processes required for successful ovulation. Furthermore, this study provides a clearer understanding of the role of androgens in ovarian physiology and highlights that AR might be of significance clinically as a target gene in the treatment of androgen-associated fertility disorders such as PCOS. The discovery that AR^+/− females develop age-dependent subfertility could be of significance in further characterizing AR function in women with the possibility that impaired AR signaling may contribute to a premature reduction in female fecundity.

Acknowledgments

The authors thank Jenny Spaliviero for histological slide preparation, Danny Liske for mouse genotyping, and Karsten Kleo and Maggie Ma for their technical assistance in the laboratory.

This work was supported by the National Health and Medical Research Council.

Disclosure Statement: The authors declare that there is no conflict of interest that would prejudice the impartiality of this scientific work. K.A.W., C.M.A., M.J., P.R.L., R.A.D., J.D.Z., and D.J.H. have nothing to disclose. P.I. previously received lecture fees from Serono Ply.

Abbreviations

• AR

  Androgen receptor

• Bax

  Bcl-2-associated X protein

• Bcl-2

  B cell leukemia/lymphoma-2

• CL

  corpus lutea

• CMV

  cytomegalovirus

• E2

  estradiol

• HSD

  hydroxysteroid dehydrogenase

• OCC

  oocyte-cumulus complex

• PCNA

  proliferating cell nuclear antigen

• PCOS

  polycystic ovary syndrome

• T

  testosterone

• ZPR

  zona pellucida remnant.