Androgen Receptor in the Ovary Theca Cells Plays a Critical Role in Androgen-Induced Reproductive Dysfunction

Abstract

Androgen and its receptor (AR) play a critical role in reproductive function under both physiological and pathophysiological conditions. Female AR global knockout mice are subfertile due to both neuroendocrine and ovarian defects. Female offspring from prenatally androgenized heterozygous AR pregnant mice showed rescued estrous cyclicity and fertility. Ar is expressed in granulosa cells, theca interstitial cells, and oocytes in the ovary. We created mice with theca-specific deletion of Ar (ThARKO) by crossing Cyp17-iCre mice that express Cre recombinase under cytochrome P450 17A1 (Cyp17) promoter with Ar^fl/fl mice. ThARKO mice exhibited no significant differences in pubertal onset or fertility compared with control littermates, and neither estrogen or testosterone levels were different between these groups. Therefore, Ar expression in theca cells likely does not influence fertility nor androgen levels in female mice. We then tested the role of AR in theca cells under hyperandrogenemic condition. After treatment with a pathophysiological level of dihydrotestosterone (DHT), control mice (control-DHT) showed acyclicity and infertility. However, estrous cycles and fertility were altered to a significantly less degree in ThARKO-DHT mice than in control-DHT mice. Messenger RNA (mRNA) levels of Lhcgr (luteinizing hormone receptor) and Timp1 (tissue inhibitor of metalloproteinase 1, and inhibitor of matrix metalloproteinase) were significantly lower in control-DHT ovary compared with control-no DHT ovaries, whereas mRNA levels of Fshr (follicle-stimulating hormone receptor) were significantly higher. Timp1 gene expression was comparable in the ThARKO-DHT and the control-no DHT ovary. We speculate that the preserved level of Timp1 in ThARKO-DHT mice contributes to retained reproductive function.

Androgen receptor (AR) signaling plays a critical role in reproductive function in both males and females. For females, reduced androgen signaling in animal models of gene deletion or hypomorphia demonstrate that androgen signaling contributes to normal follicle development, ovulation, and fertility (1–5).

The AR is a member of the nuclear receptor superfamily and is coded by the Ar gene on the X chromosome (6). The AR is expressed in each of the 3 types of cells in the ovary: theca interstitial cells where androgens are produced, granulosa cells where testosterone is converted to estrogens, and the germ cells, the oocytes (7–9). Although female mice without Ar expression in oocytes do not exhibit impaired reproductive function, female mice with Ar knocked out in granulosa cells (GCARKO) have impaired fertility, with reduced numbers of litters and smaller litter sizes after 2 or 6 months of age and irregular cyclicity after 6 months of age (10, 11). However, the role of androgen signaling in the theca cells of normal females is not known.

While androgens in humans are produced in the ovary and adrenal glands, in rodents, they are produced solely by the ovary (12). In both mouse and human, normal androgen production of theca cells maintains follicular growth via promotion of early-stage folliculogenesis and prevention of follicular atresia (10). However, androgen excess (pathological doses) leads to abnormal follicular growth and infertility (13).

In women, hyperandrogenemia-associated infertility is found in diseases such as congenital adrenal hyperplasia, Cushing syndrome, and polycystic ovary syndrome (PCOS) (14, 15). Meanwhile, increased androgen signaling due to androgen excess is associated with reproductive dysfunction in primates and lower mammals. Many studies that investigated the reproductive disorder associated with androgen excess used models of prenatal and prepubertal androgen exposure in rhesus monkeys (16, 17), sheep (18–21), rats (22), and mice (23). However, in these models, the loci of hyperandrogenic effects could not be elucidated. For modeling androgen exposure, a more potent agonist to the AR than testosterone is dihydrotestosterone (DHT), which is enzymatically produced from testosterone by 5α-reductase. DHT is an attractive agent in studies of the singular role of androgen signaling in health and disease, because, unlike testosterone, it cannot be aromatized to estrogen.

Loss of 50% of AR signaling in pregnant mice is sufficient to protect female offspring from infertility due to prenatal androgenization (24), and therefore, it is highly likely that excess androgens are acting through AR pathways. We previously reported that knockout of the insulin receptor in theca cells reduced testosterone secretion and partially rescued obesity-induced infertility (25). However, whether the ovary is also a target for androgen excess is less well understood.

AR action plays a crucial role in normal ovarian function, but the cellular sites of action have yet to be fully elucidated. Additionally, the cellular targets that account for altered ovarian function during hyperandrogenic states have not been teased out. Therefore, we developed a theca-specific AR knockout (ThARKO) mouse to study the role of androgens in the function of this crucial ovarian cell. This new model allows us to examine the cell-specific pathways, and molecular mechanisms, involved in the pathogenesis of female infertility. We studied the ThARKO mouse in both physiological and pathophysiological androgen states to determine the ovarian contribution of androgen/AR function related to the development of infertility due to hyperandrogenic disorders.

Materials and Methods

Generation and genotyping of ThARKO mice

Floxed AR and cyp17iCre mice (Cre driven by the promoter to CPY17) were maintained in our laboratory as previously described (5, 25) in a mixed background (C57/B6, CD1, and 129Sv). We generated ovarian theca AR knockout (ThARKO) mice (AR^fl/fl; cyp17icre^+/−) by mating heterozygous female (AR^fl/wt; cyp17icre^+/−) with homozygous male (AR^fl/y; cyp17icre^−/−) mice. As controls, we used female littermates with genotype AR^fl/wt; cyp17icre^−/− and/or AR^fl/fl; cyp17icre^−/−. Genomic DNA isolation and primers to detect AR gene and cyp17iCre were described previously (5, 25). All procedures were performed with approval of the Johns Hopkins Animal Care and Use Committee.

Generation of hyperandrogenemic females

To prepare pellets of DHT for insertion into the mice, Dow Corning Silastic tubing (0.04 in inner diameter × 0.085 in outer diameter, Fisher Scientific, Grand Island, NY) was filled with DHT or without DHT to a length of 4 mm and sealed with 2 mm of medical adhesive silicone (Factor II, Inc., Lakeside, AZ) on each end. The resulting pellets were incubated in saline for 24 hour at 37°C for equilibration before insertion (26, 27).

At 2 months of age, female mice were treated by insertion of a DHT pellet (DHT mice) or an empty pellet (control mice); pellets were replaced every month. The DHT level in blood serum was measured by both enzyme-linked immunosorbent assay (Alpha Diagnostics International, San Antonio, TX) (28) every week and by liquid chromatography tandem mass spectrometry in the laboratory of Dr. Brian Keevil at the University Hospital of South Manchester (29). The DHT release depends on the length of DHT, the diameter, and the material of the tubing. Three months after insertion of the pellets, body weight was recorded and body composition was measured by EchoMRI (EchoMedical Systems) to determine total body fat and lean mass.

Assessment of puberty, estrous cyclicity, and reproductive phenotypes in ThARKO females

Puberty was assessed beginning at 21 days of age by visual inspection of vaginal opening and assessing the age of first estrus (5). Estrous cyclicity was determined, beginning at 8 weeks of age and continuously for 16 days, by assessing vaginal cytology (5). Fertility was assessed by mating 2- to 3-month-old female mice with proven fertile wild-type male mice for 100 days and recording the number of pups and number of litters per female (5, 25). The examiners were blinded to genotypes during all data collection.

Ex vivo ovary culture

Serum levels are tightly controlled by feedback on hypothalamic-pituitary hormone output and thus may not reflect differences in theca cell steroidogenesis. We therefore examined ovarian steroid secretion ex vivo to determine if AR in theca cells is involved in ovary responses to luteinizing hormone (LH) stimulation. Ovaries were extracted at diestrus and incubated in a 24-well tissue culture plate with McCoy’s 5A medium at 37°C (25). After 3 hours of incubation, the medium was replaced with fresh medium with or without 1.6 IU/mL human chorionic gonadotropin (hCG). After a further 24 hours of incubation, medium was collected, and the concentration of androstenedione was measured at the Center for Research in Reproduction, University of Virginia.

Analyzing estrous cyclicity and reproductive phenotypes in hyperandrogenemic females

Females implanted with DHT or an empty pellet for 15 days were divided into 2 groups. Group 1 underwent examination of estrous cyclicity by vaginal cytology for 16 days starting at day 15 after DHT treatment. Females in group 2 were mated with fertile males (1 female with 1 male per cage) for 100 days starting at day 15 of DHT treatment. Fertility was examined as described above.

Hormone assays and GnRH stimulation

Morning levels (9 to 10 am) of LH and follicle stimulating hormone from serum of mice at diestrus were measured by Luminex assay (MPTMAG-49K, Millipore, Billerica, MA) (5). Serum levels of estradiol and testosterone were measured at the University of Virginia Center for Research in Reproduction, Ligand Assay, and Analysis Core. Testosterone was also measured by liquid chromatography tandem mass spectrometry in Dr. Brian Keevil’s group at the University Hospital of South Manchester. Blood was collected 15 minutes after 100-ng/kg body weight. Gonadotropin releasing hormone (GnRH) administration per mouse and LH was measured as above and as previously described (5).

qRT-PCR

The RNA was extracted from hypothalamus, pituitary, granulosa, and theca-interstital cells by Trizol and reversed transcribed to complementary DNA (5). Granulosa and theca-interstital cells were separated by manually punctured ovary with a 26-gauge needle and fine-tip tweezers as previously described (25). The messenger RNA (mRNA) levels of 10 genes related to androgen production and ovary function in the ovary and Ar levels in hypothalamus, pituitary, and ovary were measured by quantitative real-time polymerase chain reaction (qRT-PCR) using iQSYBR green reagent according to the manufacturer’s protocol (Bio-Rad). GAPDH was used as the internal control. The genes were: Cyp17A1 (cytochrome P450, family 17, subfamily A, polypeptide 1), Cyp19 (cytochrome P450, family 19, subfamily A, polypeptide 1), StAR (steroidogenic acute regulatory protein), Amh (antimullerian hormone), Esr1 (estrogen receptor 1), Esr2 (estrogen receptor 2), Lhcgr (luteinizing hormone receptor), Fshr (follicle-stimulating hormone receptor), Cdkn1b (cyclin-dependent kinase inhibitor 1B), and Timp1 (tissue inhibitor of metalloproteinase 1, inhibitor of matrix metalloproteinase [MMP]). Primers are listed in Table 1.

Table 1.

Primers Used in this Study

F-Forward; R-Reverse		Primers
lhcgr	F	GACCAAAAGCTGAGGCTGAGA
	R	CAATGTGGCCATCAGGGTAGA
fshr	F	CATCACTGTGTCCAAGGCCA
	R	TGCGGAAGTTCTTGGTGAAAA
AR	F	GGCGGTCCTTCACTAATGTCAACT
	R	GAGACTTGTGCATGCGGTACTCAT
Timp1	F	TTCCCCAGAAATCAACGAGAC
	R	ATCCACAGAGGCTTTCCATG
Cdkn1b	F	CCCTTCCACCGCCATATTG
	R	GCCCCTCTCCAAACCTTG
Esr1	F	CGCCTAGCTCAGCTCCTTCT
	R	GATGCTCCATGCCTTTGTTAC
Esr2	F	GCCGACTTCGCAAGTGTTAC
	R	CGCTGGCACTTCTCTGTCT
Amh	F	CAGGCCCTGTTAGTGCTATACCCT
	R	GAAGTCCACGGTTAGCACCAAA
StAR	F	CCCAAAGAAGGCATAGCAAG
	R	GCTGAATCCCCCAAACTTCT
Cyp17	F	GATCTAAGAAGCGCTCAGGCA
	R	GGGCACTGCATCACGATAAA
Cyp19	F	TTGGAAATGCTGAACCCCAT
	R	CAAGAATCTGCCATGGGAAA
Gapdh	F	GGGCATCTTGGGCTACACT
	R	GGCATCGAAGGTGGAAGAGT

Histology and Western blot

Ovaries were dissected from mice at diestrus. One ovary was snap frozen in liquid nitrogen for RNA or protein assay. The other ovary was fixed in 10% formalin phosphate buffer, the paraffin embedded ovary was sectioned at 5-µm thickness, and every 10th section (total of 10 sections) was collected and stained with hematoxylin and eosin at the Johns Hopkins Histology Core Facility. Ovaries were examined, follicle development was determined, and follicles were counted with a Zeiss microscope (25).

Frozen tissues of hypothalamus, pituitary, and ovary were homogenized, and western blot was performed to determine AR expression as described previously (5).

Immunostaining

Following euthanasia, mice were perfused with 4% paraformaldehyde (30). Ovaries were post fixed in 4% paraformaldehyde overnight at 4^oC and dehydrated in 30% sucrose for 24 hours or until they had sunk to the bottom. Ovaries were then embedded in Optical Cutting Temperature Compound and sectioned at 6-µm thickness using a cryostat (HM550, MICROM). Sections were incubated with primary AR antibody (Table 2) for 36 hours and then incubated with the VECTASTAIN Elite ABC Kit (Rabbit immunoglobulin G: PK6101) according to the manufacture protocol. After a triple wash with phosphate-buffered saline, sections were stained with DAB Peroxidase Substrate (Vector Laboratories, SK-4100) for 20 minutes. Sections were washed again and incubated with neutral red (0.5%) for 10 minutes. Sections were washed with distilled water and covered by coverslip with mounting medium (VectaMount AQ, H-5501). The immunostained sections were photographed by an AxioCCamMR camera and exported to AxioVision software.

Table 2.

Antibody Used in This Study

Peptide/Protein Target	Antigen Sequence (If Known)	Name of Antibody	Manufacturer, Catalog No., and/or Name of Individual Providing the Antibody	Species Raised in; Monoclonal or Polyclonal	Dilution Used
AR		AR Antibody (N-20)	Santa Cruz, sc-816	rabbit; polyclonal	1/500

Statistical analysis

Data were analyzed by 2-tailed, unpaired Student t tests using Prism software (GraphPad, Inc.) or by 1-way analysis of variance (ANOVA) followed by Tukey’s post hoc test. All results are expressed as means ± standard error of mean (SEM). Statistical significance was defined as P < 0.05.

Results

Generation of ThARKO mice

Excess androgens induced infertility, and AR conditional knockout in granulosa cells revealed impaired estrous cyclicity and fertility. We therefore examined whether AR in theca cells also may play a role in both physiological and pathophysiological conditions (Fig. 1). We successfully generated mice with Ar deletion in their ovarian theca cells. We confirmed AR mRNA and protein reduction by qRT-PCR, western blot, and immunostaining [Fig. 2(A–C)].

Figure 1.

Animal models to study the reproductive role of AR in theca cells under physiological (normal androgen level) and pathophysiological (androgen excess) condition. WT, wild type.

Figure 2.

AR expression in ovaries. (A) AR mRNA level, measured by qRT-PCR, was significantly reduced in the theca-interstitial cells (TI) of ThARKO compared with control littermates (Con). There were no significant differences in AR expression between control and ThARKO mice for granulosa cells, hypothalamus (Hypo), or pituitary. (B) Western blot was performed, and AR protein levels were quantified by densitometry in 3 independent experiments. The AR protein level was significantly reduced in the ovary of ThARKO compared with controls, but there was no change in the hypothalamus and pituitary. (C) Immunohistology staining of AR (5x and 20x objectives) in the ovaries showed AR expressed intensively in the granulosa cells in both ThARKO and control mice. AR was barely detectable in the theca cells of ThARKO ovary but was expressed widely in the theca-interstitial cells of control ovary (dark brown). Insets are the higher magnifications of the areas shown in the dashed boxes. Values are mean ± SEM; n = 4 to 7/group.

The Ar mRNA levels were significantly reduced in the theca-interstitial cells of ThARKO mice compared with controls but not in the granulosa cells, hypothalamus, and pituitary, as quantified by qRT-PCR [Fig. 2(A)]. The AR protein levels were also examined by western blotting, and the protein levels were significantly reduced in the ovary of ThARKO mice [Fig. 2(B)] as compared with control animals. Finally, to examine the specificity of AR deletion, AR protein expression was detected by immunostaining, and ovaries from controls were used to test for antibody specificity. AR was expressed in the theca cells of the controls but was rarely seen in theca cells of the ThARKO ovary [Fig. 2(C)]. This was a cell-specific disruption because AR expression was detected in granulosa cells of ovaries in both ThARKO and control mice [Fig. 2(C)].

ThARKO mice exhibited no alteration in pubertal onset, estrous cycling, fertility, and hormonal secretion

We did not observe any difference in measured markers of puberty (VO and first estrus) between control and ThARKO mice [Fig. 3(A)].

Figure 3.

Female puberty and fertility. (A) ThARKO mice exhibited similar age of puberty onset, assessed by examination of vaginal opening and first estrus, as control mice. (B) Percentage of time spent in each of the stages (5 month old) was not significantly different between control and ThARKO mice. (C and D) There was no significant differences between ThARKO and controls in either (C) total numbers of litters per female or (D) numbers of pups per female. Values are mean ± SEM; n = 6 to 10/group. Con, control; D, diestrus; E, estrus; M, metestrus; NS, nonsignificant; P, proestrus.

There was no difference in estrous cycling between control and ThARKO mice at 2 months of age or between control and ThARKO mice at 5 to 6 months of age, as assessed by the percentage of time in each of the stages of estrous [Fig. 3(B)]. The mean time for a cycle was approximately 6.5 days in both control and ThARKO mice, which was not significantly different between the 2 groups, and was similar to other mice in the colony where they are housed (2).

Fertility was examined by recording the number of pups and litters as previously described (30). During the mating period of 100 days, there was no significant difference between control and ThARKO mice in either the number of litters [control: 3.6 ± 0.3 versus ThARKO: 3.3 ± 0.3; Fig. 3(C)] or number of pups per female [control: 36 ± 3.1 versus ThARKO: 29 ± 3.5; Fig. 3(D)].

Morning levels (9 to 10 am) of LH, follicle stimulating hormone, estradiol, and testosterone at diestrus were not significantly different between control and ThARKO mice (Supplemental Fig. 1 ^{(88.3KB, docx)} ). Furthermore, ThARKO and control mice exhibited equal gonadotropin responses to GnRH stimulation [Fig. 4(A)].

Figure 4.

Hormone levels. (A) Pituitary responses to GnRH stimulation were tested in the mornings of diestrus. There was no significant difference in LH secretion at either 10 or 20 minutes post GnRH stimulation between control and ThARKO mice. n = 8 to 10. (B) Androstenedione secretion with and without hCG. Ovary was collected at diestrus and cultured in McCoy5A medium for 3 hours. Medium was replaced with/without hCG. At 24 hours, medium was collected and androstenedione was measured. There was no significant difference in androstenedione secretion between control and ThARKO mice either without hCG or with hCG. For both groups of mice, androstenedione secretion was significantly increased by hCG addition compared with no hCG addition. n = 4/group. Values are mean ± SEM. Con, control; NS, nonsignificant.

Ovarian steroid secretion was similar in control and ThARKO mice in ex vivo culture

Although baseline serum androstenedione was not altered in ThARKO mice compared with controls, androstenedione was measured in the culture medium in order to eliminate the effects from hypothalamus and pituitary. The amounts of androstenedione in ovaries from control and ThARKO mice were not significantly different in medium without hCG. Furthermore, there were no significant differences in the amounts of androstenedione in ovaries from ThARKO and control mice after hCG (24-hour) stimulation [Fig. 4(B)].

Expression of genes important for ovarian function was not altered in ThARKO mice

Gene expression (Lhcgr, Fshr, Cyp17A1, Cyp19, StAR, and Esr2) was assessed from ovaries collected at 2 pm of diestrus or proestrus. There were no significant differences in expression of 8 genes between control and ThARKO mice at diestrus [Fig. 5(A)] or at proestrus [Fig. 5(B)].

Figure 5.

Expressions patterns of 8 genes necessary for steroidogenesis and ovary function at (A) diestrus and (B) proestrus. There were no significant differences in mRNA level between control (Con) and ThARKO mice for any gene at either estrous stage. Values are mean ± SEM; n = 5 to 11/group.

Estrous cyclicity was partially recovered in ThARKO-DHT mice compared with control-DHT mice

Although AR knocked out in theca cells did not affect estrous cyclicity or fertility, we investigated if AR in theca cells plays any role in hyperandrogenemia-induced infertility. We treated mice with 4-mm DHT pellets to produce serum androgen levels that mimic those in human hyperandrogenic phenotypes. The DHT serum levels from DHT treated mice were increased approximately twofold compared with that of non-DHT treated mice 2 weeks post insertion of the pellet, and this level was consistent when tested 4 weeks after insertion. When measured by enzyme-linked immunosorbent assay, the serum DHT level was 0.10 ± 0.01 ng/mL in non-DHT treated mice versus 0.19 ± 0.01 ng/mL in DHT-treated mice after 2 weeks post insertion. To further confirm the relative levels of DHT, samples were also measured by mass spectrometry, and DHT levels were 0.52 ± 0.09 nmol/L (or 0.15 ± 0.03 ng/mL) in non-DHT treated mice versus 1.16 ± 0.17 nmol/L (or 0.34 ± 0.05 ng/mL) in DHT-treated mice after 2 weeks post insertion (Supplemental Fig. 2 ^{(88.3KB, docx)} ). During the 3 months of DHT treatment, body mass (Supplemental Fig. 3 ^{(88.3KB, docx)} ) and body composition were not altered compared with controls (data not shown).

We assessed estrous cyclicity in adult mice beginning 16 days after DHT treatment, and examples of the estrous cyclicity patterns in the 4 groups of mice (control-no DHT, control-DHT, and ThARKO-DHT) are shown in Figure 6(A). Control-DHT mice [Fig. 6(B), right-striped column] had very few days in proestrus and estrus and were predominantly in diestrus and metestrus compared with control-no DHT mice [Fig. 6(B), open column]. However, ThARKO-DHT mice [Fig. 6(B), black column] showed improved estrous cyclicity with significantly more days in proestrus and fewer days spent in diestrus and metestrus compared with control-DHT mice [Fig. 6(B)]. This result indicates that estrous cycling was partially restored in ThARKO-DHT mice. There is no difference of the estrous cyclicity between control-no DHT and ThARKO-no DHT [Fig. 6(B)], and the cycling pattern is similar to Figure 3(B).

Figure 6.

Estrous cycles. (A) Representative data for vaginal cytology from individual control (Con)-no DHT, ThARKO-no DHT, control-DHT, and ThARKO-DHT mice. (B) Percentage time spent in each stage of the estrous cycle was significantly differently among the 3 groups, analyzed by 1-way ANOVA followed by Tukey posttest. Percentage of time in proestrus (P) and estrus (E) was reduced, and time in metestrus (M)/diestrus (D) was increased in control-DHT compared with control-no DHT. The percentage time spent in each stage in ThARKO-DHT was in between that of control-no DHT and control-DHT. Bars with different letters represented values that are significantly different to those analyzed within the same cycle stage (P < 0.05). Values are mean ± SEM; n = 5 to 11.

Fertility was partially rescued in ThARKO-DHT mice compared with control-DHT mice

Number of litters and average litter size were compared among experimental groups and graphed in Figure 7(A). Control-no DHT data are displayed as a reference. Examples of fertility profiles for control-DHT and ThARKO-DHT mice are plotted in Supplemental Figure 4 ^{(88.3KB, docx)} . The number of litters [Fig. 7(A)] and pups per female [Fig. 7(B)] of DHT-ThARKO mice during the mating period was significantly increased compared with control-DHT mice but significantly reduced compared with control-no DHT mice, indicating fertility was partially recovered by Ar knockout in theca cells.

Figure 7.

Fertility with and without DHT. (A) Total number of litters and (B) pups per female were significantly reduced in control (Con)-DHT mice compared with control-no DHT mice during the 100 days of mating. ThARKO-DHT mice had a significantly improved fertility compared with control-DHT, although the fertility was still impaired compared with the control-no DHT mice. Bars with different letters represent significantly different values from each other with P < 0.05, analyzed by 1-way ANOVA followed by Tukey posttest. Values are mean ± SEM; n = 6 to 8/group.

Ovary weight and morphology and hormone secretion

Ovary weight was not significantly different between control and ThARKO mice and was not affected by DHT treatment (data not shown).

Morphology of ovaries from control-no DHT, control-DHT, ThARKO-no DHT, and ThARKO-DHT mice is shown in Figure 8(A). The most marked difference was abundance of the corporus luteum (CL), which was much less common in the ovaries of the control-DHT mice than in any of the other 3 groups.

Figure 8.

Ovary morphology and follicular numbers. (A) Histological sections of ovary from control and ThARKO mice with and without DHT 3 months after insertion of pellets (hematoxylin and eosin stain). (B) Numbers of follicles in each group. CL, antral follicles, and preantral follicles (primodial, primary, and secondary follicles) were examined in each group of ovaries. Control-DHT mice showed a significant reduction in numbers of CL compared with control-no DHT and ThARKO-DHT. In ThARKO-DHT mice, the number of CL was intermediate between control-no DHT and control-DHT mice. There were no significant differences in number of antral and preantral follicles among the 4 groups as analyzed by 1-way ANOVA followed by Tukey posttest. Bars with different letters represent significantly different values from each other with P < 0.05. Values are mean ± SEM; n = 5 to 21/group. Con, control; FC, follicle cyst; NS, nonsignificant.

There were significant differences in the number of CL in the ovaries of the 4 different groups of mice [Fig. 8(B)], with few CL in the control-DHT mice. In contrast, there were similar numbers of preantral follicles (primodial, primary, and secondary) in ovaries from control-no DHT, control-DHT, ThARKO-no DHT, and ThARKO-DHT mice [Fig. 8(B)]. Furthermore, the numbers of antral follicles were not significantly different among the 4 groups [Fig. 8(B)]. In the control-DHT mice, there are multiple follicular cysts, which are not present in the control-no DHT mice [Fig. 8(A)].

There was no significant difference in serum levels of estradiol and testosterone between groups (Supplemental Fig. 5 ^{(88.3KB, docx)} ).

Ovarian mRNA expression in ThARKO-DHT compared with control-DHT

Genes related to steroidogenesis and ovulation were examined at diestrus by qRT-PCR. The mRNA levels of 8 of the 10 genes studied were significantly different between control-DHT and control-no DHT mice: Lhcgr, Fshr, Cyp17A1, Cyp19, StAR, Esr2, Timp1, and Cdkn1b [Fig. 9(A)]; the mRNA levels of Fshr and Esr2 were significantly higher in control-DHT than in control-no DHT mice, whereas for the other 6 genes, mRNA levels were significantly lower.

Figure 9.

Gene expression patterns in ovary at dietrus. (A) Among the 10 genes related to steroid production and ovarian function examined by qRT-PCR, mRNA levels were significantly reduced in 6 (Lhcgr, Cyp17, Cyp19, StAR, Timp1, and Cdkn1b) and significantly increased in 2 (Fshr and Esr2) in control (Con)-DHT ovary compared with that of control-no DHT ovary 3 months after insertion of pellet. (B) Only Timp1 was recovered in ThARKO-DHT ovary compared with that of control-no DHT ovary. Bars with different letters represent significantly different values from each other with P < 0.05, analyzed by 1-way ANOVA followed by Tukey posttest. Values are mean ± SEM; n = 5 to 7/group.

The mRNA levels of Lhcgr, Fshr, Cyp17A1, Cyp19, StAR, Esr2, and Cdkn1b were similar between THARKO-DHT and control-DHT mice (data not shown); however, Timp1 expression in ThARKO-DHT mice was not significantly different from control-no DHT mice [Fig. 9(B)]. The mRNA level of Timp1 was significantly higher in ThARKO-DHT mice than in control-DHT mice.

Discussion

In this work, we have explored the role of AR in theca cells in ovarian function and the role of theca cell AR in hyperandrogenemia-induced ovarian dysfunction. We have shown that AR signaling in the theca cells is not required for steroid secretion, development, or reproductive function of the ovary under physiological androgen levels, as indicated by normal age of puberty, estrous cyclicity, and fertility in ThARKO mice (Fig. 3). Expression of genes associated with steroidogenesis and ovary function is not significantly altered at either diestrus or proestrus (Fig. 5) between control and ThARKO mice, indicating that AR signaling is not required for normal theca cell function, although it could still play a role. These observations indicate that the subfertility exhibited by the female global ARKO mice is likely due to the functions of AR in granulosa cells (10), pituitary gonadotropes (5), and/or brain, rather than the ovarian theca cell. However, it could not exclude that residual remaining AR in the theca cells may be sufficient for its reproductive functions.

To study how and where hyperandrogenemia induces infertility, we implanted adult female mice with a low dose of DHT that mimics the androgen levels in serum of women with hyperandrogenemia (31–33). The serum DHT levels (twofold that in no DHT–treated mice) used in this study were much lower than those reported by other investigators in previous studies (31, 34, 35), where the level of serum DHT in the treated animals was 6- to eightfold that in vehicle-treated animals. Further, levels of testosterone and/or androstenedione in ovarian follicle fluid (FF) and serum of women with PCOS are 1.5- to 4.0-fold higher than that of normal women (31–33, 36–41), whether the fluid is assayed directly from FF (36, 37, 41) or indirectly by ex vivo theca cell culture (38–30), except in one that reported androstenedione secreted from cultured theca cells of women with PCOS was 25-fold higher than that from normal women (42). By extension, it is expected that the magnitude of the increased testosterone concentrations in FF and serum of hyperandrogenic mice will be proportional to each other.

In our study, weight of the mice was not significantly different between DHT and no DHT groups, indicating that obesity or being overweight was not a confounding variable in the fertility. This is in contrast to previous studies (31, 34, 35) in which there was a significant difference in weight between androgen-treated and untreated animals.

Adult control-DHT mice showed diminished estrous cyclicity, the presence of cystic follicles in the ovary, few CLs, and infertility (Figs. 6–8). This result indicates that this lower-dose DHT model, which more closely resembles the androgen levels in women with PCOS, recapitulates the reproductive phenotype in the absence of obesity. We observed that ThARKO-DHT mice had improved estrous cycling and increased number of CLs and thus more frequent ovulations compared with control-DHT mice. The increased number of ovulations likely resulted in the increased number of litters and larger litter sizes in the ThARKO-DHT mice compared with the control-DHT mice. However, in our study, the fertility rate of the ThARKO-DHT mice remained significantly lower than in control-no DHT mice, perhaps due to the pathogenic effects of androgen excess on multiple tissues such as pituitary, hypothalamus, and placenta (43, 44).

To investigate which ovarian factors underlie the increased fertility of the ThARKO-DHT mice, we examined expression of key ovarian genes involved in follicular development. Ovulation, follicle maturation, rupture, and egg release are tightly controlled by genes such as lhcgr, fshr, Timp1, and cdkn1b (45–51). In normal ovarian function, Timp1 mRNA is localized to the stroma and theca of developing follicles and to the granulosa and thecal-interstitial cells of large preovulatory and ovulating follicles (46, 52). An increase in MMP and in the MMP/TIMP ratio appears to determine early selection of follicles destined for atresia in PCOS (46, 52, 53). Decreased level of TIMPs, especially TIMP1, may be one of the mechanisms responsible for ovarian dysfunction and early regression of the CL (46). Global knockout of the Timp1 gene results in altered reproductive function (54). Cdkn1b is the regulator of the cell cycle and is required for proper functioning of luteinizing/luteinized cells in vivo (51), and animals with global knockout of Cdkn1b exhibit impaired ovulation, cyclicity, and fertility (50).

Although Timp1 levels were significantly reduced in ovaries of control-DHT compared with control-no DHT mice, the mRNA levels of Timp1 in the ovary of ThARKO-DHT mice were not significantly difference compared with control-no DHT mice. We speculate that the higher expression of this gene in ThARKO-DHT mice contributes to their improved estrous cyclicity and ovulation. Considering the heterogeneous ovarian cell population, further examination of separated theca-interstitial, granulosa, and luteinizing CL cells are needed for future gene expression analysis.

Our findings suggest that although the theca AR is not essential in responding to the typical female hormonal milieu, it plays an important role in mediating the pathological effects of androgen excess via AR-induced infertility. These findings may contribute to the development of new therapies for the treatment of hyperandrogenemia-related reproductive disorders by targeting treatment to ovarian theca cells.