Prenatal Exposure to Low Levels of Androgen Accelerates Female Puberty Onset and Reproductive Senescence in Mice

Abstract

Sex steroid hormone production and feedback mechanisms are critical components of the hypothalamic-pituitary-gonadal (HPG) axis and regulate fetal development, puberty, fertility, and menopause. In female mammals, developmental exposure to excess androgens alters the development of the HPG axis and has pathophysiological effects on adult reproductive function. This study presents an in-depth reproductive analysis of a murine model of prenatal androgenization (PNA) in which females are exposed to a low dose of dihydrotestosterone during late prenatal development on embryonic d 16.5–18.5. We determined that PNA females had advanced pubertal onset and a delay in the time to first litter, compared with vehicle-treated controls. The PNA mice also had elevated testosterone, irregular estrous cyclicity, and advanced reproductive senescence. To assess the importance of the window of androgen exposure, dihydrotestosterone was administered to a separate cohort of female mice on postnatal d 21–23 [prepubertal androgenization (PPA)]. PPA significantly advanced the timing of pubertal onset, as observed by age of the vaginal opening, yet had no effects on testosterone or estrous cycling in adulthood. The absence of kisspeptin receptor in Kiss1r-null mice did not change the acceleration of puberty by the PNA and PPA paradigms, indicating that kisspeptin signaling is not required for androgens to advance puberty. Thus, prenatal, but not prepubertal, exposure to low levels of androgens disrupts normal reproductive function throughout life from puberty to reproductive senescence.

The hypothalamic-pituitary-gonadal (HPG) axis controls all stages of the reproductive lifespan and is active from puberty through menopause. Hypothalamic GnRH neurons activate pituitary gonadotropes to produce LH and FSH, which then stimulate gonadal steroid hormone production. In females, the ovaries produce testosterone (T), which is aromatized by follicle-derived aromatase into estradiol (E₂) (1). E₂ and T exert feedback effects (E₂, positive and negative; T, negative) at the level of the hypothalamus and pituitary to control reproductive hormone production (2). The tight regulation of a complex interplay of positive and negative feedback loops is required for many reproductive functions including follicle maturation, ovulation, and estrous cyclicity.

GnRH neurons are the central regulators of the HPG axis. During the juvenile stage, the HPG axis is quiescent, with the subsequent onset of puberty reflecting an increase in GnRH secretion, leading to secondary sex characteristics, estrous or menstrual cycling, and ovulation in females (3). A key hallmark of puberty is the activation of the GnRH neurons, which occurs, in part, through the upstream kisspeptin neurons. Kisspeptin, a neuropeptide encoded by the Kiss1 gene, potently stimulates GnRH neurons by signaling through the kisspeptin receptor [encoded by Kiss1r, formerly termed G protein-coupled receptor 54 (GPR54)] (3, 4). Puberty is impaired in the absence of kisspeptin signaling, such as in Kiss1 or Kiss1r-null mice (5), or in humans lacking functional kisspeptin or its receptor (6, 7).

The release of GnRH not only drives puberty but has also been suggested to play a role in reproductive senescence (8). In aging rodents, LH pulsatility is altered, the preovulatory LH surge (caused by increased GnRH) is attenuated, and functional activity of the GnRH neurons is decreased (8). This reduction in GnRH neuronal activity during senescence is associated with decreased ovarian sex steroid levels and a concomitant loss of sex steroid negative feedback in the brain (8).

The negative feedback by sex steroids, such as androgens, is necessary for normal functioning of the reproductive neuroendocrine axis. The importance of androgens and their receptor (AR) in reproductive development and function has long been known in males but is less well understood in females. Evidence from female AR-null mice has demonstrated that AR is necessary for normal estrous cyclicity, ovarian morphology, progesterone production, and fertility (9). Exposure of females to exogenous, supraphysiological levels of androgens, however, may disrupt the normal role of androgens and have negative impacts on female reproductive outcomes (10).

The reproductive neuroendocrine system is particularly vulnerable to environmental and hormonal disturbances. Exposure to endocrine-disrupting chemicals (EDC; such as steroidogenic compounds) during critical periods of development can have profound effects on reproduction and fertility later in life (11). The mechanisms by which EDC alter neuroendocrine function is not fully understood, but it has been suggested that prenatal sex steroid exposure can increase the risk of reproductive disorders, such as precocious puberty and polycystic ovarian syndrome (PCOS), the leading cause of infertility in women (10, 12). The study of prenatal androgenization in animal models may therefore lead to a better understanding of the endocrine abnormalities and underlying mechanisms resulting from EDC exposure and PCOS in females. Indeed, studies in rats, sheep, and non-human primates suggest that low-level exposure to androgens during prenatal development can preprogram females for reproductive abnormalities later in life, disrupting hormonal dynamics that influence cyclicity and sensitivity to sex steroid feedback (13–16). A prenatal androgen model has been described in female rats in which pregnant females are treated with T on embryonic d (e) 16–19 (17). This paradigm results in a disruption in neuroendocrine feedback in the prenatally androgenized female rats, but because T can be aromatized to E₂, the hormonal mechanisms and pathways underlying these effects may be mediated by either estrogen receptor (ER) or AR (17).

Using dihydrotestosterone (DHT) instead of T avoids potential effects resulting from aromatization of T to E₂. In a similar prenatal androgenization model (termed PNA) in mice (18), pregnant dams are treated with low-dose DHT on e16.5–18.5. PNA female offspring have normal genitalia but do not cycle regularly in adulthood and have increased serum levels of T (18), mirroring oligomenorrhea and increased androgen levels observed in women with PCOS (10). Although the extent to which the PNA model fully recapitulates the clinical characteristics of PCOS is not fully established, it still offers important insights into the mechanisms and pathophysiology of female reproductive disorders caused by early androgen exposure.

The overall goal of this study was to determine the effects of developmental DHT exposure on female reproductive outcomes. To accomplish this, our study had two main objectives. First, we compared the reproductive phenotype resulting from prenatal and peripubertal exposure to DHT in females. Second, we used neuron-specific AR knockout and Kiss1r-null mice to investigate the role of neuroendocrine mechanisms in response to developmental DHT exposure.

Materials and Methods

Animals

All animal procedures were performed in accordance with the University of California, San Diego (UCSD) Institutional Animal Care and Use Committee regulations. All mice were on a C57BL/6J background and were group housed on a 12-h light, 12-h dark cycle with ad libitum chow (11% of calories fat, 17% of calories protein) and water. Kiss1r-null animals were from Omeros, Inc. (Seattle, WA) (19, 20). Heterozygous Kiss1r animals bred to generate homozygous Kiss1r knockout (KO) and wild-type (WT) littermates for study. To generate mice lacking AR exclusively in neurons (AR/Syn-Cre), AR flox males (21) in which the second exon of the AR gene is flanked by loxP sites) were bred to Synapsin-Cre females [in which Cre is expressed exclusively in neurons (21, 22)]. Offspring, termed AR/Syn-Cre, have glial but not neuronal AR because this is a neuron-specific knockout (data not shown). Littermate homozygous AR flox with and without Syn-Cre were compared.

PNA and PPA (prepubertal androgenization) paradigms

To generate PNA females, pregnant dams were injected (sc) with 100 μl of vehicle (sesame oil; Sigma, St. Louis, MO) or 250 μg of DHT (5α-androstan-17β-ol-3-one; Sigma) dissolved in 100 μl of sesame oil. Injections were given once daily on gestational days e16.5–18.5. This low dose of prenatal DHT can influence the functioning of the HPG axis of the female offspring without secondary effects resulting from virilized female genitalia (data not shown) (18). To generate PPA females, mice were injected once daily on postnatal days (PND) 21–23 with 100 μl of vehicle (sesame oil) or 250 μg of DHT dissolved in 100 μl of sesame oil. The dose of 250 μg of DHT was based on a half-log dose response study and calculation of equivalent dose per gram weight as the PNA treatments (weight of the pregnant dam).

Pubertal assessment and estrous cycles

All experimental animals were monitored for vaginal opening status, an external marker of puberty onset beginning after weaning on PND 21 (Fig. 1A). Of note, Kiss1r-null females normally have either severely delayed or absent vaginal opening; therefore, PND 35 was used as the upper limit of vaginal opening assessment for statistical purposes.

Fig. 1.

VO was advanced in PNA and PPA but not in PNA progeny. A, Diagram illustrates the time line for DHT administration in the PNA and PPA paradigms and the time frame for monitoring pubertal onset by VO. B, Wild-type females were subjected to the PNA paradigm and were monitored daily for VO. ***, P < 0.001 by Student's t test. C, Wild-type females were subjected to the PPA paradigm and were monitored daily for VO. ***, P < 0.001 by Student's t test. D, Adult PNA females were mated to wild-type males and the female progeny were monitored daily for VO (P > 0.05 by Student's t test).

To assess estrous cyclicity, adult PNA and PPA females were housed one to two per cage approximately 1 wk before analysis (single housing did not affect cyclicity). Vaginal smears were taken once a day for 15–25 consecutive days, and vaginal cytology determined the cycle stage (diestrus, proestrus, estrus, metestrus). To evaluate the onset of reproductive senescence, mice were classified as having persistent vaginal cornification (PVC) if their vaginal smears contained cornified cells on 75% or more of days checked.

Testosterone implantation

The estrous cycles of 16 2-month-old wild-type females were monitored for 15 d, followed by a sc implantation of a SILASTIC brand capsule (Dow Corning Corp., Midland, MI; inner diameter 1.47 mm; outer diameter 1.95 mm) filled with testosterone (17β-hydroxy-3-oxo-4-androstene; Sigma) diluted 1:40 in cholesterol (Sigma). Pilot studies showed that this dose resulted in serum T levels comparable with PNA females when measured at 1 wk after implantation. Eight females each were implanted with T capsules or cholesterol vehicle capsules. Animals were allowed to recover for 1 wk, and then estrous cycles were measured again for 15 d.

Single-label in situ hybridization (ISH)

Brains from 4-month-old control diestrus PNA females and male mice were harvested between 1030 and 1230 h, fresh frozen on dry ice, cryosectioned, and probed for Kiss1 via ISH as described (23). Briefly, slide-mounted sections were fixed in 4% paraformaldehyde, treated with acetic anhydride, washed in 2× sodium citrate, sodium chloride, delipidated in chloroform, dehydrated in ethanol, and air dried. The Kiss1 radioactive probe was denatured and added to hybridization buffer at 0.05 pmol/ml. One hundred microliters of the probe-buffer mix were added to each slide and allowed to hybridize at 55 C for 16 h. Slides were washed and then treated with ribonuclease. After several more washes, the slides were dipped in Kodak emulsion (Kodak, Rochester, NY) and stored at 4 C for 9 d before developing. For quantification, slides were analyzed with custom grain-counting software (Don Clifton, University of Washington, Seattle, WA). The number of Kiss1-expressing neurons and silver grains per cell (a semiquantitative measure of mRNA levels per cell) were counted.

Serum collection and hormone assays

Blood was collected at the time the animals were killed from 4-month-old PNA and PPA females in diestrus. Briefly, mice were anesthetized with isoflurane (Abbott Laboratories, Abbott Park, IL), and the blood was collected through the abdominal aorta by syringe before 1300 h. After coagulation, the blood samples were centrifuged at 2655 × g for 15 min, and sera were transferred to a fresh tube and stored at −20 C. Sera were analyzed by RIA at the University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core (Charlottesville, VA).

Ovarian morphology

Ovaries were dissected from 2-, 3-, 7-, and 10-month-old PNA females, fixed in 4% paraformaldehyde, washed in PBS, and stored in 70% ethanol. One ovary per animal was embedded in paraffin (UCSD histology core facility). Ten-micrometer sections were stained with hematoxylin and eosin (UCSD histology core facility). Three separated sections per ovary were quantified for corpora lutea, preantral follicles, and antral follicles.

Glucose and insulin tolerance tests

Oral glucose tolerance tests (GTT) were performed on 1- and 6-month-old PNA females and oil-treated controls. Mice were administered 1 g/kg of an oral gavage of 50% glucose after 7 h of fasting. Insulin tolerance tests (ITT) were performed on 1-month-old PNA females, using 6 U/kg of insulin-BSA (0.5 U/ml). For both the GTT and ITT, blood glucose levels were measured from tail blood samples using a glucose meter. Animals were then transferred to a high-fat diet (60% calories from fat, 20% calories from protein) for 6 months and subjected to another GTT.

Statistical analyses

All results are presented as mean ±sem. Student's t test, one-way ANOVA, two-way ANOVA, χ², and survival analysis were used where indicated. P < 0.05 was considered statistically significant. Because Kiss1r-null females do not normally display vaginal opening (VO) within the normal time frame of wild-type mice, VO was not recorded past PND 35. We therefore determined statistical significance using survival analysis, followed by a log-rank χ²test.

Results

Pubertal onset is advanced in PNA and PPA females

Prior studies of androgenization during critical developmental periods have focused on adult reproductive parameters, such as sex steroid feedback (13–16), but have not investigated the effects of developmental androgen exposure on puberty. To determine whether there are distinct sensitive periods of development when androgens act to alter sexual maturation and induce reproductive abnormalities, we assessed the onset of puberty in females exposed to DHT. Female mice were subjected to one of two DHT treatment paradigms: prenatal (PNA: DHT exposure during e16.5–18.5) or peripubertal (PPA: DHT exposure from PND 21–23) and were compared with oil-treated (control) females (12–15 animals/group) (Fig. 1A). VO age was significantly advanced in PNA females compared with vehicle-treated controls (P < 0.001, Fig. 1B). Because body weight can play a significant role in the timing of pubertal onset (24), we measured body weight in PNA at PND 20, 24, and 4 months of age and found no differences (Supplemental Fig. 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). Therefore, prenatal DHT does not advance VO due to increased body weight. Surprisingly, peripubertal DHT exposure significantly advanced pubertal onset as well (P < 0.001, Fig. 1C). The PPA mice also showed no differences in overall weight at 4 months of age (data not shown).

Some studies have shown transgenerational effects of endocrine disruption (11). We therefore next investigated whether early VO also occurred in the progeny of PNA females. The female offspring of PNA females displayed VO comparable with controls (Fig. 1D), suggesting that this pubertal effect of DHT exposure does not have a heritable component.

The role of neuronal AR and Kiss1r in DHT-mediated advanced pubertal onset

The initiation of puberty involves the activation of kisspeptin and GnRH signaling as well as an increase in sex steroids, which feed back on kisspeptin neurons via ERα and AR (3). Whether prenatal DHT exposure affects reproductive parameters by altering kisspeptin circuits is currently unknown. To answer this question, we tested whether developmental DHT exposure of Kiss1r-null mice, as well as mice conditionally lacking AR in neurons, would also impact pubertal and reproductive parameters. This allowed discrimination between effects of DHT on the hypothalamic component of the HPG axis vs. the pituitary or ovaries.

To test whether peripubertal DHT exposure regulates pubertal onset via neuronal AR, we subjected female mice with a neuron-specific AR knockout (AR/Syn-Cre) to the PPA paradigm. Vaginal opening was monitored in PPA AR/Syn-Cre and AR flox (control) females given peripubertal DHT or oil (six to eight animals per group). Because DHT is nonaromatizable, we hypothesized that any effects of DHT in the brain to induce early pubertal onset would be prevented by the absence of neuronal AR. However, PPA caused a significant advancement in VO in the AR/Syn-Cre mice as well as in the AR flox controls (P < 0.001 compared with vehicle-treated mice, Fig. 2A), indicating that peripubertal DHT is not acting through neuronal AR to induce early puberty.

Fig. 2.

VO was advanced in PPA AR/Syn-Cre and Kiss1r-null, and in PNA Kiss1r-null mice. A, AR/Syn-Cre females were mated to AR flox males. Female AR flox (Cre-) and AR/Syn-Cre (Cre+) offspring were subjected to the PPA paradigm and were monitored daily for VO. ***, P < 0.001 vs. control by two-way ANOVA. B, Heterozygous Kiss1r animals were mated. Female homozygous WT and KO Kiss1r-null offspring were subjected to the PPA paradigm and were monitored daily for VO. ***, P < 0.001 vs. control by log-rank χ². VO was not recorded after PND 35. C, Heterozygous Kiss1r dams were administered PNA. Female homozygous WT and Kiss1r-null (KO) offspring were subjected to the PNA paradigm and were monitored daily for VO. *, P < 0.05 vs. control by log-rank χ². ***, P < 0.001 vs. oil-treated knockout by log-rank χ². VO was not recorded after PND 35.

To further dissect the mechanism and site of action of DHT in PNA, we used a genetic model of HPG axis inactivation: Kiss1r-null mice (25–27). These mice display hypogonadotropic hypogonadism and fail to progress through normal sexual maturation (25). Study of PNA in the Kiss1r-null mice should yield more precise mechanistic results than the much broader AR/Syn-Cre elimination of AR in neurons. We first confirmed that vehicle-treated Kiss1r-null mice had absent VO compared with vehicle-treated wild-type females (P < 0.001, Fig. 2C). Interestingly, we found that PPA treatment caused VO to occur early in Kiss1r-null mice (KO; P < 0.001 vs. vehicle-treated mice, Fig. 2B). We next evaluated the effect of prenatal DHT on VO in Kiss1r-null mice. As in PPA females, PNA treatment resulted in early VO in Kiss1r-null mice (P < 0.05, Fig. 2C).

To further investigate the role of the kisspeptin system in mediating the effects of PNA, we examined Kiss1 expression in the anteroventral paraventricular nucleus (AVPV) of adult PNA and control mice by ISH. Kiss1 is expressed in the AVPV in a sexually dimorphic manner, at much higher levels in females than males (23). We found no differences in Kiss1 expression in the AVPV between PNA and control females and confirmed that the control males had low levels (P < 0.01, Supplemental Fig. 2). Thus, the development and sexual dimorphism of the Kiss1 neurons is not affected by PNA treatment. Furthermore, Kiss1 expression in the arcuate nucleus was also unchanged in these females, suggesting that PNA treatment does not alter the sex steroid negative feedback on kisspeptin neurons (data not shown).

PNA, but not PPA, causes irregular estrous cycling in adults

Having established that both PNA and PPA advanced pubertal onset, we next determined whether cyclicity was similarly affected in the two models. Estrous cycling is initiated in female rodents following VO, and previous evidence has suggested that PNA can cause irregular estrous cycling between 4 and 6 months of age (18). We investigated whether PNA can cause irregular cycling at an earlier age when fertility is robust (3 months). To evaluate whether DHT acts to influence reproductive parameters in both developmental androgen exposure paradigms, we compared PNA with age-matched PPA females. The cycles of DHT- and oil-treated PNA and PPA females were followed up. To quantify the regularity of cycling, the proportion of time spent in each cycle stage was compared between DHT- and oil-treated groups (seven to eight animals per group). Figure 3A shows that 3-month-old PNA females had irregular cycling compared with oil-treated controls. In contrast, 3-month-old PPA females showed normal estrous cycling and were therefore not further studied. Quantification of the proportion of time spent in each cycle stage (diestrus, proestrus, estrus, and metestrus) showed significant differences in that PNA mice spent less time in diestrus and more time in metestrus than control females (P < 0.001, Fig. 3B). This suggests a mechanistic divergence of DHT action between these two time frames: PNA has a lasting effect by adversely influencing estrous cycling in adulthood, whereas PPA does not. To further investigate this idea, testosterone levels were measured in 4-month-old PNA and PPA females (this time point was chosen because the animals have sufficiently progressed into adulthood but have not yet reached reproductive senescence). We found a modest increase of T in the serum of PNA mice compared with control females but no change in T levels in PPA females (Fig. 3C).

Fig. 3.

PNA exposure resulted in irregular estrus cycles and increased T in females. A, Estrous stages for 3-month-old PNA and PPA females were tested daily for 15–20 d. Representative cycle traces are shown. M, Metestrus; E, estrus; P, proestrus; D, diestrus. B, The proportion of time spent in each cycle stage was quantified for the PNA females (n = 8 oil and n = 7 DHT). ***, P < 0.001 by Student's t test. C, Four-month-old PNA and PPA females were killed in diestrus and sera were analyzed for T levels. *, P = 0.05 by Student's t test.

The role of testosterone in the irregularity of estrous cycling in PNA mice

Given that cycling irregularities, as well as increased T, were found in PNA females but not PPA females, we hypothesized that the higher T maintained by the PNA females adversely affects estrous cycling. To test this, T implants, at a dose resulting in levels of T similar to those measured in PNA females, were administered to adult, wild-type females, and estrous cycling was monitored both before and after the implantation. The appropriate dose of T was established by extensive pilot studies measuring T levels 1 wk after the implantation (data not shown). Vaginal cytology was quantified and the T-implanted females spent less time in proestrus and more time in metestrus compared with controls (P < 0.05, Fig. 4) mirroring the effects of the elevated endogenous T in the PNA mice.

Fig. 4.

Testosterone implantation resulted in disrupted estrous cycling in wild-type females. Estrous stages were tested in wild-type females (2–3 months old) daily for 15 d both before (A) and after (B) T or cholesterol implantation, and the proportion of time spent in each cycle stage was quantified (n = 8 cholesterol and n = 8 T). *, P < 0.05 by Student's t test.

PNA delays fertility in early adulthood

Because PNA females have irregular cycling at 3 months of age, developmental DHT exposure could have significant adverse effects on reproductive output. A 6-month-long fertility assessment was therefore conducted in PNA females compared with oil-treated controls, beginning at 2 months of age. DHT and oil females (2 months old, six animals per group) were each housed with a single C57BL6 male continuously for 6 months. The time to first litter and the number of litters produced were determined. After the first month of the fertility assessment, more control mice had given birth to their first litter than PNA females (P < 0.05, Fig. 5A). As the fertility assessment progressed, the PNA females produced numbers of litters, as well as litter sizes (data not shown), similar to the controls (Fig. 5B), suggesting that PNA only temporarily affects fertility.

Fig. 5.

PNA females had an initial delay in fertility. PNA females were pair housed continuously with single wild-type males for 6 months. A, The number of days until birth of the first litter was measured. *, P < 0.05 by Student's t test. B, The total number of litters born during each month of the fertility assessment was quantified. *, P < 0.05 by Student's t test.

Ovarian morphology and metabolic factors in PNA mice

Polycystic ovaries and impaired insulin sensitivity are two common hallmarks of PCOS (10). To complement our findings from the estrous cycling and fertility assessments and to further evaluate the extent to which PNA represents PCOS in a murine model, ovarian morphology was examined in 2-, 3-, 7-, and 10-month-old PNA mice by observing the presence or absence of ovarian cysts, corpora lutea, and follicles.

Ovaries from 2-, 3-, and 7-, and 10-month-old PNA and control mice were analyzed for morphological features. The numbers of corpora lutea, preantral follicles, and antral follicles were comparable between PNA and control mice at each age (Fig. 6). Furthermore, there were no detectable ovarian cysts, suggesting that PNA did not result in a persistent ovarian phenotype. The presence of follicles and corpora lutea is consistent with the overall ability of the PNA females to produce litters after prolonged exposure to a male, as was seen in the fertility assessment.

Fig. 6.

PNA females exhibited ovarian morphology consistent with the ability to produce litters. PNA females were killed in diestrus at 2 (A), 3 (B), 7 (C), and 10 (D) months of age. Ovaries were fixed, sectioned, and examined for normal follicles, Graffian follicles (GF), and corpora lutea (CL) (n = 3 for 2, 3, and 10 months; n = 2 for 7 months). CL, preantral follicles, and antral follicles were quantified and no differences were found.

Despite the absence of ovarian cysts, the PNA mice displayed precocious puberty, irregular cycling accompanied by a decreased fertility, and elevated levels of T, all of which are associated with PCOS (28). Because PCOS patients also commonly present with metabolic changes such as increased body weight and insulin resistance (10), we further tested whether PNA mice exhibited changes in these metabolic end points. We found no group differences in body weight in adulthood at 4 or 6 months of age (Supplemental Fig. 1B) or at PND 20 or 24 (Supplemental Fig. 1A) or in adult PPA females (data not shown). As indicators of metabolic function, an oral GTT and an ITT were performed in PNA females vs. controls to test whether PNA adversely affected insulin sensitivity and blood glucose regulation in PNA females at 1 month of age. However, no differences were detected between PNA and controls in glucose clearance or insulin sensitivity (Supplemental Fig. 3). The PNA animals were then transferred to a high-fat diet for 6 months. A GTT after the high-fat diet revealed a greater change in glucose metabolism in the controls compared with PNA animals. A GTT was also performed on 6-month-old PNA females that were not subjected to a high-fat diet, and no differences were detected (Supplemental Fig. 3).

PNA advances reproductive aging

To our knowledge, studies of prenatal androgenization in experimental animals have not yet investigated the effect of early androgen exposure on reproductive senescence in females. We hypothesized that prenatal DHT would have negative effects on estrous cycling not only in early adulthood but continuing as the animals aged. Therefore, because puberty onset was advanced and cycling was irregular in PNA females compared with controls, we investigated whether the onset of reproductive aging occurred earlier in PNA mice. In female mice, reproductive aging with regard to estrous cycling is characterized by persistent cornification of the vaginal epithelial cells as determined by cytology (29). Cytology was monitored in PNA mice at 8 and 10 months of age, comparing oil- and DHT-treated females (20–23 animals/group). Estrous cyclicity was measured in the female mice for 25 d, determining that the incidence of PVC occurred more frequently in PNA mice compared with controls at both ages (P < 0.01, Fig. 7). This higher incidence of aberrant cycling in older PNA mice indicates that PNA may advance the onset of reproductive aging.

Fig. 7.

Advanced reproductive senescence in aging PNA females. PNA females were cycled daily for 25 d at 8 and 10 months of age, and the incidence of PVC was quantified. **, P < 0.01 by χ².

Discussion

Exposure to EDC, including androgens, during critical periods of development in females can dramatically alter HPG axis function and reproductive physiology later in life (11). The present work has characterized the reproductive system of a prenatal androgenization mouse model (18) by studying the effects of PNA on puberty onset, fertility, and reproductive senescence. Exposure to low-level DHT prenatally accelerates onset of puberty, delays fertility, causes irregular estrous cycling, increases T, and advances reproductive senescence, whereas exposure prepubertally advances only puberty. Furthermore, the advanced pubertal onset induced by DHT administered prenatally and prepubertally occurs independently of the kisspeptin system and AR in neurons.

Advanced pubertal onset in PNA and PPA mice

Increased levels of androgens are associated with premature puberty in girls (28). Exposure to androgens during critical developmental periods can predispose females to reproductive disorders, but the mechanisms by which this occurs are not fully understood. The PNA mouse represents a useful model (18) for studying the effects and mechanisms of prenatal DHT exposure and reflects many aspects of reproductive disorders in women (18).

The use of DHT in the PNA and PPA models is advantageous, as opposed to T, in that DHT is not aromatized and therefore acts through AR only and not ERα. A caveat of this is that the DHT metabolite, 5α-androstane-3β,17β-diol, acts through ERβ (30), and therefore, a possible contribution of ERβ signaling should be considered. Indeed, administration of an ERβ agonist to female rats resulted in advanced pubertal onset (31). Future studies using the PNA and PPA models could avoid the possibility of DHT having secondary effects through ERβ signaling by additionally treating with an ERβ antagonist such as raloxifene or an AR antagonist such as flutamide. However, we chose not to use these antagonists, which have extensive activities of their own to ensure that our models best reflect androgen exposure in women.

The studies presented herein show that pubertal onset is advanced in PNA female mice, as previously noted (32). However, whether other time frames exist during which the female reproductive axis is sensitive to sex steroid modulation has not been studied (to our knowledge). We therefore compared pubertal onset in PNA to a novel model of DHT exposure, PPA. That VO is advanced in both PNA and PPA indicates that it can act during both the late prenatal and prepubertal stages to influence the timing of pubertal onset. Because VO in PNA mice occurs 3–4 wk after DHT treatment, whereas VO in PPA mice occurs immediately after treatment, it is possible that these effects are occurring through different mechanisms. PNA may have permanent, organizational effects on the neuronal circuitry driving the HPG axis and/or sex steroid feedback, although sexual dimorphism of the kisspeptin system is not affected, whereas PPA may act only transiently and peripherally to induce puberty onset. Given the critical importance of Kiss1r in puberty, we sought to determine whether kisspeptin signaling (which responds to sex steroid feedback and drives GnRH and the rest of the reproductive axis) was involved in either or both of the PNA and PPA paradigms.

Advanced puberty onset in PNA is Kiss1r independent

Kiss1r is critical for the onset of puberty both in humans and mice (5, 7, 25), but its role in mediating the advanced VO in PNA was unknown. Our finding that VO occurred early after PNA regardless of the absence of Kiss1r indicates an alternative pathway influencing pubertal onset, possibly through a peripheral mechanism. Furthermore, the effects of PNA are not mediated by masculinization of the sexual dimorphism of the Kiss1 system, suggesting the kisspeptin system is not playing a major role in the actions of PNA. This lack of alterations in Kiss1 sexual dimorphism was not surprising, given that our animals were treated with nonaromatizable DHT, thus avoiding the effects of testosterone-derived estradiol. Although the effects of PNA appear to be independent of kisspeptin signaling, it is possible that GnRH may still be involved in the early puberty onset because it can be regulated by factors other than kisspeptin, such as neurokinin B and steroid-responsive glial inputs (33, 34).

Advanced puberty onset in PPA is independent of neuronal AR and Kiss1r

We used the PPA paradigm to determine whether this model also did not require Kiss1r as well as whether PPA was exerting a broad effect through neuronal AR to induce early puberty. The acceleration of VO by PPA in the absence of neuronal AR and Kiss1r indicated that the advancement of puberty in this model may be occurring through a nonneuronal mechanism. Because the Kiss1r-null animals have little or no hypothalamic activation of the reproductive axis, it appears that PPA is bypassing the kisspeptin neuronal circuitry to cause early VO. Because Kiss1r-null mice do not have VO or puberty onset, the induction of VO by PPA and PNA in these mice indicates that DHT is capable of activating VO in the absence of normal puberty. Whether there is a contribution from the pituitary and the ovaries remains to be determined.

PNA resulted in long-term effects on fertility, cycling, and aging

Studies of the PNA mouse had not previously investigated the effects on fertility, so we conducted a fertility assessment to determine whether PNA treatment could influence reproductive function. Despite an early delay, the PNA females were capable of producing litters as the fertility assessment progressed, suggesting that the PNA paradigm was not sufficient to render females infertile. This is consistent with our investigation of ovarian morphology, which provided evidence of ovulation. Interestingly, the time frame during which the early decrease in fertility was observed (when the females were 3 months old) coincided with the age at which they showed irregular cycling. The reduction in litters produced by PNA females at this age could be explained by abnormalities in estrous cycling.

Estrous cycling of PNA and PPA females showed irregularities at 3 months of age in PNA but not PPA. Thus, despite a similar advancement in pubertal onset between the two paradigms, the effects of PNA persist into adulthood to influence estrous cycling, whereas PPA effects do not. As a complex interplay of sex steroid feedback effects is necessary to promote progression through the estrous cycle, we therefore propose that PNA (but not PPA) results in organizational changes during HPG axis development (17), possibly affecting sex steroid production, which persist throughout the reproductive life span. The increased T in PNA mice is consistent with PNA, but not PPA, affecting reproductive outcomes in adulthood (18). We directly tested the role of increased T in estrous cyclicity and found that even low-level exposure to T, comparable with the levels in PNA, caused irregularities in estrous cycling. Sex steroids can increase cornification of the vaginal epithelium (35), which may explain the increased incidence of metestrus (which is partly characterized by cornified epithelial cells) in both PNA and T-implanted females. We therefore conclude that an increase in T resulting from PNA treatment is at least partly responsible for the alterations in estrous cycling.

These effects on reproductive parameters were not associated with increases in body weight or changes in glucose metabolism. A high-fat diet in 7-month-old PNA animals also did not adversely affect glucose metabolism as strongly as controls. This is in contrast with a previous study (32) that found impaired glucose tolerance in PNA females. Although the cause of these dissimilar results has not been fully determined, there may be a possible contribution of mouse strain and background differences. Other studies have found strain differences to be the cause of variation in metabolic parameters (36). Roland et al. (32) studied animals that were originally bred from CBB6/F1 transgenic mice, whereas the mice in the current study were 100% C57BL/6J. The difference in strains may partly account for the differences seen in glucose tolerance.

Because PNA influenced the timing of pubertal onset and fertility in early adulthood, we asked whether PNA could also influence reproduction later in life by altering the timing of reproductive aging. PNA females displayed evidence of reproductive senescence earlier than controls, demonstrating that PNA can influence reproductive parameters even 8–10 months after exposure. This finding, combined with the irregular cycling in PNA but not PPA, suggests a mechanistic difference in DHT action between the PNA and PPA models, in which PNA has organizational, lasting effects, whereas PPA acts only transiently to induce VO. The HPG axis is more sensitive to DHT during the late prenatal period than the peripubertal period, and changes that occur in utero affect reproductive function late into adulthood.

In conclusion, we have studied the PNA mouse model in depth to show a kisspeptin-independent advancement of pubertal onset, irregular cycling in early adulthood, with a concomitant delay in fertility, and early reproductive senescence. We have compared PNA to a novel model of precocious puberty, PPA. The advanced pubertal onset in PPA occurs independently of neuronal AR and Kiss1r. PPA has only a transient effect to accelerate the onset of puberty, whereas PNA has lasting effects throughout the reproductive life span, influencing, not only puberty but also fertility, cycling, and reproductive senescence. Thus, prenatal androgen exposure alters reproductive function throughout life, indicating that androgen exposure during this stage of development may be important for the etiology of female reproductive disorders.