Maternal Testosterone Excess Contributes to Reproductive System Dysfunction of Female Offspring Mice

Abstract

Hyperandrogenism is considered 1 of the most important characteristics of polycystic ovary syndrome, which affects more than 10% of females of reproductive age and is a common cause of infertility. In addition to the effects on patients themselves, maternal androgen excess has also been reported to impair the growth and development of offspring. In our current study, we found that maternal testosterone (T) treatment during different gestational stages increased the percentage of atretic follicle and decreased corpus luteum formation in female offspring. In addition, decreased serum estradiol and increased T levels were also observed in female offspring of T-treated mice during late gestational stage. Further studies revealed that Forkhead box protein L2 (FOXL2) and Cytochrome P450 family 19 subfamily a member 1 (CYP19A1) expression in granulosa cells of these female offspring mice were decreased. By using mouse primary granulosa cells and the KGN cell line, we demonstrated that decreasing FOXL2 and CYP19A1 levels in ovarian granulosa cells partially may contribute to disturbed sex hormone synthesis in female offspring of T-treated mice during the late gestational stage. Findings from our current study highlight a critical role of excess maternal T exposure, especially during the late gestational stage, which could further lead to aberrant ovary development and sex hormone synthesis in female offspring.

maternal testosterone excess, offspring, follicle development, granulosa cell function

Androgens, mainly produced by testes, ovaries, and adrenal glands, play crucial roles during the organization and maintenance of normal physiological activities, including the development of sex organs as well as other peripheral tissues, such as skeletal muscle, skin, bone, brain, placenta, cardiovascular system, and adipose tissue (1–3). However, mechanisms of androgen-mediated activities and functions on these tissues or organs are still not completely understood. Androgen also acts as precursor of estrogen, which is critical for the development and regulation of female reproductive system (4). Ovarian theca cell–secreted testosterone (T), 1 of the main type of androgens in women, can be converted into estradiol (E2) under the catalyzation of aromatase in granulosa cell. Newly generated E2 further contributes to the regulation of various transcription factors that finally promote oocyte maturation and ovulation (5,6).

Despite such important functions of androgens in reproductive system, ovarian tumors or polycystic ovary syndrome (PCOS)-induced androgen excess, also termed hyperandrogenism, was reported to be responsible for a series of pathological symptoms in females of reproductive age, including acne, increased body or facial hair, changed ovarian morphology, and infrequent or absent menstruation (7–10). In addition to these effects themselves, clinical and basic studies revealed that androgen excess during gestational stage also affects the growth and development of their offspring (11–16). Phenotypic outcomes on offspring of maternal hyperandrogenism include intrauterine growth restriction, cardiovascular defects, behavior deficits, metabolic disorder, and reproductive dysfunctions (14, 17–19). Increased LH level, multifollicular ovary and anovulation can also be observed in offspring of women with maternal hyperandrogenism (16, 17, 19). In addition to these defects in humans, animal studies in rodents indicated that maternal androgen excess blocked steroid feedback, impaired follicular recruitment, and caused reproductive system malformation in female offspring (11, 16, 20).

Although multiple investigations demonstrated that maternal androgen excess may affect the reproductive system development, most animal studies were focused on hyperandrogenism during late gestational stage (14, 16, 17, 21). Effects of androgen excess on offspring during early gestational stage or the entire pregnancy have rarely been investigated. Thus, our current study treated maternal mice with T starting from different gestational days. Physiological indexes, sex hormone levels, and follicle development of female offspring were analyzed to evaluate the effects of maternal androgen excess on reproductive system development. Given the limited information on the mechanism of maternal androgen excess-induced reproductive system impairment of female offspring, we further investigated estradiol generation pathway inside the granulosa cell to elucidate disturbed sex hormone synthesis in female offspring.

Materials and Methods

Animals

All animal protocols were approved by the Animal Care and Use Committee of the Model Animal Research Center of Nanjing Medical University. Female C57BL/6 mice were purchased from Nanjing Medical University and housed in temperature-controlled (21–22°C) pathogen-free conditions with 12-hour light 12-hour dark cycle and free access to water and food. All animals were acclimatized to the laboratory environment for at least 1 week before experiments.

Pregnant C57BL/6 mice were assigned to 1 of 6 groups for 3 different experiments. In experiment 1, mice were injected daily subcutaneously with either T (Sigma) in a dose of 2.5 mg/kg body weight (22) or equivalent solvent as control from days post coitum (dpc) 1 to 19. In experiments 2 and 3, mice were injected with the same dose of T or solvent from dpc 8 to 19 and dpc 14 to 19, respectively. The body weight and blood glucose level of maternal mice were measured every 2 days from 8:00 to 9:00 am.

For offspring mice, all litters were weaned after parturition and females were separated from males at 3 weeks of age. The weight gain of female offspring was measured every 2 weeks. Blood glucose level was measured at 12 weeks of age. All female offspring were sacrificed at 3 months of age. Serum samples were collected and ovaries were fixed in 10% formalin for further experiments. Granulosa cells were isolated for real-time PCR or western blotting analysis.

Chemical preparation

For animal injection, T was purchased from Sigma and dissolved in sesame oil (Sigma)/benzyl benzoate (Sigma) to a final concentration of 1 mg/mL. For cell treatment, T was dissolved in 100% dimethyl sulfoxide (Sigma) and diluted in the cell culture medium to a final concentration of 10^–7 mol/L.

Cell culture and treatment

Mouse primary granulosa cells were isolated using the follicular puncture method (23, 24). Each mouse was intraperitoneally injected with 5 IU pregnant mare serum gonadotropin 48 hours before granulosa cell isolation. Ovaries were immediately isolated from the mice upon death by cervical dislocation and immersed in DMEM/Nutrient Mixture F-12 (DMEM/F-12, Invitrogen) in a 60-mm culture dish. Then, granulosa cells were released into the medium by manual puncture of the ovaries with a 25-gauge needle under a dissecting microscope. The resulting cell suspensions were filtered using a 40-μm filter (Thermo Fisher Scientific) to remove large tissue clumps and oocytes.

Mouse primary granulosa cell and KGN granulosa cell line (BeNa Culture Collection) was cultured in DMEM/F-12 (with phenol red) or DMEM (with phenol red) supplemented with 10% (v/v) fetal bovine serum (HyClone) and 100 IU/mL penicillin and 100 µg/mL streptomycin (Invitrogen), at 37°C in a humidified atmosphere with 5% CO_2.

Forkhead box protein L2 (FoxL2) small interfering RNA (siRNA) and negative control scramble siRNA were synthesized by GenePharma. Details of oligo sequences are listed in Table 1. siRNAs were transfected into different cell types using lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions. Cells were either collected for western blot analysis 48 hours after transfection or further treated with 10^–7 mol/L testosterone for another 24 hours to stimulate E2 generation.

Table 1.

Primer and Oligo Sequences

Gene Name	Forward (5′-3′) (bp)	Reverse (5′-3′) (bp)
Quantitative RT-PCR primer sequences
FoxL2	ACAACACCGGAGAAACCAGAC (21)	CGTAGAACGGGAACTTGGCTA (21)
Cyp19a1	ATGTTCTTGGAAATGCTGAACCC (23)	AGGACCTGGTATTGAAGACGAG (22)
Ar	CTGGGAAGGGTCTACCCAC (19)	GGTGCTATGTTAGCGGCCTC (20)
Foxo3	CTGGGGGAACCTGTCCTATG (20)	TCATTCTGAACGCGCATGAAG (21)
Rspo1	GGGATCAAGGGCAAGAGACAG (21)	CTGGCGGATGTCGTTCCTC (19)
Wnt4	AGACGTGCGAGAAACTCAAAG (21)	GGAACTGGTATTGGCACTCCT (21)
Hsd17b1	ACTTGGCTGTTCGCCTAGC (19)	GAGGGCATCCTTGAGTCCTG (20)
Fshr	CCTTGCTCCTGGTCTCCTTG (20)	CTCGGTCACCTTGCTATCTTG (21)
Glut4	GTGACTGGAACACTGGTCCTA (21)	CCAGCCACGTTGCATTGTAG (20)
Gapdh	TGGATTTGGACGCATTGGTC (20)	TTTGCACTGGTACGTGTTGAT (21)
FoxL2 small interfering RNA oligo sequences
FoxL2-Homo-804	GCGUAGUGAACUCGUACAATT (21)	UUGUACGAGUUCACUACGCTT (21)
FoxL2-Homo-1011	GCUUCCUCAACAACUCGUGTT (21)	CACGAGUUGUUGAGGAAGCTT (21)
FoxL2-Mus-299	CCAAGUUCCCGUUCUACGATT (21)	UCGUAGAACGGGAACUUGGTT (21)
FoxL2-Mus-1648	GGAAAGAACGUGUCUGGUCTT (21)	GACCAGACACGUUCUUUCCTT (21)

Quantitative RT-PCR

Quantitative RT-PCR analysis was performed as previously described (25). Generally, the total RNA of mouse primary granulosa cells was extracted by TRIZOL (Invitrogen). RT was performed to generate the first-strand complementary DNA (Takara), then quantitative RT-PCR was conducted using the SYBR-Green Kit (TakaRa) on Applied Biosystems StepOnePlus. Data were normalized to the expression levels of Gapdh in each sample and the details of primers were listed in Table 1.

Western blotting

Total protein of collected cells was extracted using a RIPA lysis buffer (Cell Signaling Technology), and western blotting procedures were performed as previously described (25). The following primary antibodies were used: FOXL2 (1:1000, Abcam) (26), CYP19A1 (1:500, Abcam) (27), and B-ACTIN (1:5000, Santa Cruz) (28). Horseradish peroxidase (HRP)-linked anti-goat IgG (1:2000, Santa Cruz) (29), anti-rabbit IgG (1:2000, CST) (30), or anti-mouse IgG (1:5000, CST) (31) was used as a secondary antibody. Finally, protein bands were developed using Immobilon Western Chemiluminescent HRP Substrate (Thermo Fisher Scientific).

ELISA

An estradiol (mouse) ELISA kit was purchased from BioVision (catalog number: K3830) (32, 33), testosterone and triglyceride ELISA kits were purchased from Abcam (catalog number: ab108666 and ab65336) (34, 35), mouse insulin ELISA kit was purchased from Mercodia (catalog number: 10-1247-01) (36), and mouse progesterone, LH, and FSH ELISA kits were purchased from NOVUS Biologicals (catalog number: NBP2-60125-1, NBP2-60086-1 and KA2330) (37, 38). All ELISA procedures were performed according to the manufacturer’s instructions.

Histological analysis and immunohistochemical staining

Hematoxylin and eosin staining and immunohistochemical (IHC) staining were performed as previously described (25, 39). For IHC staining of FOXL2 and CYP19A1, mouse ovaries were fixed in 10% formalin and embedded in paraffin after dehydration. Next, 4-μm sections were cut and rehydrated followed by 3% H₂O₂ treatment. Permeabilization was conducted using a 0.01 mol/L citrated buffer. FOXL2 and CYP19A1 antibody (Abcam) were applied to the sections in a dilution of 1:200 at 4°C overnight after a blocking procedure. HRP-linked mouse anti-goat or goat anti-rabbit (1:400, Dako) were used as a secondary antibody.

Statistical analysis

Data are presented as mean ± SEM unless otherwise indicated. Differences between the groups were analyzed by a 2-tailed unpaired Student t-test with Welch correction or 1-way ANOVA, followed by Tukey post hoc test where appropriate. Statistical comparisons were made using GraphPad Prism 6 (GraphPad Software, San Diego, CA). A P value <.05 was considered as statistically significant.

Results

Physiological indexes of maternal mice are not affected by T treatment

To evaluate the effects of T treatment on physiological indexes of maternal mice, we measured the body weight and blood glucose of these gestational mice every 2 days. The body weight of maternal mice was increased with gestation, but it was not affected by T treatment in all 3 experiments. Similarly, the blood glucose level was also stable even in the presence of T treatment (Fig. 1A–F).

Figure 1.

Physiological indexes of maternal mice are not affected by T treatment. Maternal C57BL/6 mice were injected subcutaneously daily with T in a dose of 2.5 mg/kg body weight or equivalent solvent as control from dpc 1 to 19 (experiment 1), 8 to 19 (experiment 2), or 14 to 19 (experiment 3). The (A-C) body weight and (D-F) blood glucose level were measured every 2 days. n = 12–15. Data are presented as mean ± SEM. Abbreviations: Dpc, days post coitum; T, testosterone.

On gestational day 20, 5 maternal mice of each group were randomly selected and sacrificed for serum hormone level measurement. As expected, the serum T level of each androgen-treated mice were significantly increased compared with solvent-treated ones (Fig. 2A). Consistently, serum insulin level was also elevated in experiments 1 and 2 after T treatment (Fig. 2B). In experiment 3, serum insulin was also slightly but not significantly increased (Fig. 2B). However, other hormones, including E2 and progesterone, were not affected (Fig. 2C and D).

Figure 2.

T treatment during different gestational stages increases serum T and insulin levels of maternal mice. At gestational day 19, 5 maternal mice of each group were sacrificed, and serum was collected for (A) T, (B) insulin, (C) E2, and (D) progesterone measurement. n = 5. Data are presented as mean ± SEM. *P < .05, **P < .01 when evaluated using a 2-tailed unpaired Student t-test. Abbreviations: E2, estradiol; T, testosterone.

T treatment during late gestational stage disturbs the sex hormone level in female offspring

Maternal androgen excess was proposed to be associated with poor pregnancy outcome, so we first evaluated the successful parturition rate of these mice after T treatment during different gestational stages. In experiments 2 and 3, all pregnant mice had successful parturition regardless of T or solvent treatment (Table 2). However, the parturition rate was dramatically decreased after T treatment in experiment 1 (25% vs. 100%; Table 2), indicating that maintenance of normal T level during early gestational stage was critical for embryo development and elevated maternal T may cause embryo loss.

Table 2.

Effects of Maternal Androgen Treatment on Pregnancy Outcome

Experiment	Treatment	Successful Parturition No. (Total No.)	No. of Offspring Per Litter (Mean ± SEM)	Weight of Neonatal Mice (g, Mean ± SEM)
Experiment 1 (dpc 1–19)	Solvent	6 (6)	6.67 ± 0.72	1.36 ± 0.13
	T	2 (8)	6.33 ± 0.41	1.27 ± 0.14
Experiment 2 (dpc 8–19)	Solvent	6 (6)	6.45 ± 0.39	1.29 ± 0.07
	T	7 (7)	7.21 ± 0.95	1.37 ± 0.10*
Experiment 3 (dpc 14–19)	Solvent	6 (6)	6.87 ± 0.48	1.27 ± 0.12
	T	6 (6)	7.33 ± 0.40	1.34 ± 0.09†

*P <0.01 †P < .05, when compared with each solvent treatment group using a 2-tailed unpaired Student t-test.

dpc, days post coitum.

Although parturition rate was not affected in experiments 2 and 3, the body weight of newborn mice was significantly higher in offspring of T-treated mice than control ones (1.37 ± 0.10 vs. 1.29 ± 0.07 in experiment 2 and 1.34 ± 0.09 vs. 1.27 ± 0.12 in experiment 3; Table 2). Nevertheless, this difference no longer exists with increased age. Four weeks after birth, the body weight of offspring mice in all groups was similar (Fig. 3A–C). The blood glucose level of female offspring mice was measured at 12 weeks of age and it was also not affected by maternal T treatment in all experiments (Fig. 3D–F).

Figure 3.

The body weight and blood glucose level of female offspring are not affected by maternal T treatment. (A-C) The body weight of female offspring from different experiments was measured every 2 weeks. (D-F) The blood glucose level of female offspring from different experiments was measured at 12 weeks age from 8:00 to 9:00 am. n = 10 for treatment group of experiment 1, n = 16–18 for remaining groups of experiments 1–3. Data are presented as mean ± SEM.

At 3 months of age, female offspring were sacrificed and serum was collected for hormone level measurement. Serum E2 and T levels were not affected by maternal T treatment in both experiments 1 and 2 (Fig. 4A and B). Interestingly, there was a marked decline of E2 level in female offspring of experiment 3 (Fig. 4A), which was accompanied by increased serum T level (Fig. 4B). In addition to these 2 hormones, other indexes, including LH, FSH, triglyceride, and insulin, were not affected by maternal T treatment in all experiments (Fig. 4C–F).

Figure 4.

T treatment during late gestational stage disturbs the sex hormone balance in female offspring. At 3 months age, female offspring of maternal T-treated mice were sacrificed and serum samples were collected for (A) E2, (B) T, (C) LH, (D) FSH, (E) triglyceride, and (F) insulin measurement. n = 10 for treatment group of experiment 1, n = 16–18 for remaining groups of experiments 1–3. Data are presented as mean ± SEM. *P < .05, **P < .01 when evaluated using a 2-tailed unpaired Student t-test. E2, estradiol; T, testosterone.

Maternal T treatment impairs ovarian follicle development of female offspring

It is widely recognized that ovarian hyperandrogenism impairs follicular development by inducing follicular atresia, disturbing oocyte meiosis and developmental competence, and causing follicular arrest and degeneration (40, 41). However, effects of maternal androgen excess on follicular development of female offspring were rarely investigated. In our current study, we performed hematoxylin and eosin staining on the ovary of female offspring from all 3 different experiments and calculated the percentage of follicles undergoing different developmental stages (Fig. 5). Primary follicle, secondary follicle, and graafian follicle generation were not affected in all 3 experiments, but the percentage of primordial follicle in offspring of experiment 1 was significantly decreased compared with the control group (Fig. 5A–E). Atretic follicles were increased in all 3 experiments (Fig. 5G), and this impairment was accompanied by reduced corpus luteum percentage (Fig. 5F). These results suggested that maternal T excess may cause follicular atresia and failure of ovulation, which disturbs the normal follicular development of female offspring. During early pregnancy, a high internal T environment will lead to the reduction of the number of progenitor follicles.

Figure 5.

Maternal T treatment impairs ovarian follicle development of female offspring. At 12 weeks age, female offspring of maternal testosterone treated mice were sacrificed and ovary tissues were fixed in formalin. (A) Hematoxylin and eosin staining of the ovary was performed to evaluate the ovarian morphology and follicular development. Percentage of (B) primordial follicle (PrF), (C) primary follicle (PF), (D) secondary follicle (SF), (E) graafian follicle (GF), (F) corpus luteum (CL), and (AF) atretic follicle (G) in each ovary was calculated. Scale bar = 500 μm. n = 5. Data are presented as mean ± SEM. *P < .05, **P < .01, ***P < .001 when evaluated using a 2-tailed unpaired Student t-test. T, testosterone.

Maternal T treatment impairs granulosa cell function of female offspring

Maintenance of granulosa cell function is important for normal follicular development. Since the progression of follicular maturation was disturbed in female offspring of T-treated mice, we isolated the ovarian granulosa cell and evaluated the mRNA expression level of cell survival- and function-related markers. Surprisingly, none of the selected markers in granulosa cells of female offspring was affected by T treatment during the entire gestational stage (experiment 1, Fig. 6). Nevertheless, Androgen receptor (AR) mRNA expression was impaired in both experiments 2 and 3 compared with their control groups (Fig. 6C). In addition, FoxL2 and cytochrome P450 family 19, subfamily member 1 (Cyp19a1), mRNA levels were both decreased about 50% in granulosa cells of experiment 3 (Fig.6A and B). Other markers, including forkhead box O3 (Foxo3), R-spondin-1 (Rspo1), Wnt family member 4 (Wnt4), hydroxysteroid 17-beta dehydrogenase 1 (Hsd17b1), FSH (Fshr), and glucose transporter 4 (Glut4), were not affected by maternal testosterone treatment (Fig. 6D–I).

Figure 6.

Maternal T treatment impairs granulosa cell function of female offspring. Ovarian granulosa cells were isolated from female offspring of maternal testosterone treated mice for RNA extraction. Quantitative RT-PCR analysis was performed to evaluate the mRNA expression level of granulosa cell function related markers. n = 5 for treatment group of experiment 1, n = 11–13 for remaining groups of experiments 1–3. Data are presented as mean ± SEM. *P < .05, **P < .01 when evaluated using a 2-tailed unpaired Student t-test.

Because FoxL2 and Cyp19a1 mRNA level were impaired in experiment 3, we further performed IHC staining to confirm the expression levels of these 2 proteins in granulosa cells. Consistent with quantitative RT-PCR results, FOXL2 protein was robustly expressed in the control group of experiment 3, but dramatically declined in granulosa cells of treatment group (Fig. 7A and B). Similarly, expression of CYP19A1 protein was also significantly impaired in granulosa cells of maternal T-treated female offspring (Fig. 7A and C).

Figure 7.

Testosterone treatment during late gestational stage impairs FOXL2 and CYP19A1 protein expression in granulosa cells of female offspring. (A) Immunohistochemistry of FOXL2 (upper) and CYP19A1 (lower) was performed on ovaries isolated from female offspring of experiment 3. In negative control experiments, primary antibodies were replaced with phosphate-buffered saline. (B, C) Quantification of the ratio of FOXL2- and CYP19A1-positive areas to granulosa cell area. Scale bar = 500 μm. n = 5. Data are presented as mean ± SEM. ***P < .001 when evaluated using a 2-tailed unpaired Student t-test.

Decreased FOXL2 downregulates CYP19A1 expression and contributes to granulosa cell dysfunction

Previous results indicated that serum E2 level was decreased, but serum T level was increased in female offspring in experiment 3. Meanwhile, FOXL2 and CYP19A1 expression in granulosa cells of experiment 3 were also impaired. Because CYP19A1 acts as a key catalyst that promotes the conversion of T to E2, and FOXL2 was reported to be a central transcription factor which regulates the expression of CYP19A1 in granulosa cells (42, 43), we speculated that disturbed sex hormone level in experiment 3 was probably caused by abnormal expression of FOXL2 and CYP19A1.

To confirm this hypothesis, FoxL2 siRNA was transfected into mouse primary granulosa cell and KGN human granulosa cell line. Knock-down of FOXL2 was confirmed by western blot (Fig. 8A–C). As expected, decreased CYP19A1 expression was also observed in cells transfected with FoxL2 siRNA (Fig. 8A, D and E). These results suggested that CYP19A1 was located downstream of FOXL2 in granulosa cells. To assess the effects of decreased FOXL2 expression on granulosa cell function, ELISA was performed to evaluate T-induced E2 generation. Consistent with our hypothesis, T-induced E2 secretion was impaired after FoxL2 siRNA transfection in both cell types (Fig. 8F and G). These results indicated that abnormal sex hormone level in experiment 3 was, at least partially, contributed to by decreased FOXL2 and CYP19A1 expression in granulosa cells.

Figure 8.

Decreased FOXL2 and CYP19A1 expression contributes to granulosa cell dysfunction. Two FoxL2 siRNAs were individually transfected into mouse primary granulosa cell (left) and KGN cell line (right). (A-E) FOXL2 and CYP19A1 protein expression in different cell types were analyzed by western blot 48 hours after transfection. Experiments were repeated 3 times. Testosterone (10^–7 mol/L) was added to the cell culture system for another 24 hours. Cell culture medium from mouse (F) primary granulosa cell and (G) KGN cell line was then collected for ELISA of estradiol. Data are presented as mean ± SD. *P < .05, ***P < .001 when evaluated using a 2-tailed unpaired Student t-test.

Discussion

Hyperandrogenism is considered 1 of the most important characteristics of PCOS, which affects more than 10% of females of reproductive age and acts as the most common cause of infertility (44). In addition to the effects of androgen excess on patients themselves, high level of androgen during gestational stage was also reported to affect the embryonic and postnatal development of offspring (11–16). In our current study, we treated maternal mice with T during different gestational stages. Exposure to high levels of T during the whole gestational stage dramatically impaired the successful parturition rate, but sex hormone secretion was not affected in offspring of these mice. On the contrary, T treatment during late gestational stage significantly increased the serum T level, but decreased E2 secretion in female offspring. Besides, the follicle development of maternal T-treated offspring was also impaired. Because the sex hormone balance was disturbed, we further evaluated the ovarian granulosa cell function of these mice. FOXL2, as well as CYP19A1 expression were decreased in granulosa cells. Based on previous reports, we used the mouse primary granulosa cell and KGN granulosa cell line as cell models to evaluate the mechanism of impaired sex hormone synthesis. Our data demonstrated that decreased FOXL2 and CYP19A1 in ovarian granulosa cell might be responsible for impaired granulosa cell function and reduced E2 generation in female offspring of maternal T-treated mice.

Hyperandrogenism was reported to play an important role during recurrent miscarriages (45, 46). In our current study, T treatment during the whole gestational stage significantly impaired the successful parturition rate, which was not affected when the treatment started from median or late gestational stage. This result suggests that appropriate intrauterine androgen concentration was extremely important and critical for early fetal development and increased T level may be lethal for these embryos. Interestingly, although the parturition rate was dramatically decreased after T treatment during the whole gestational stage, the physiological indexes of those successfully delivered mice were completely normal. Quantitative RT-PCR results indicated that Cyp19a1 and Ar mRNA were slightly increased in granulosa cells of these mice. Upregulation of CYP19A1 and AR may promote the transformation of T to E2 (47, 48), but serum level of these 2 hormones were not affected. These mice may be spontaneously resistant to high level of T or some adaptive pathways were activated during early gestational stage to offset the negative effects of high T environment.

In our current study, T treatment during different gestational stages significantly increased serum insulin level of maternal mice. This result was consistent with previous reports that androgen treatment was able to stimulate insulin secretion from pancreatic β cells (49). However, it is still unclear whether an increased serum insulin level was caused by activated pancreatic β-cell function or a consequence of aggravated insulin resistance of those peripheral tissues, which may finally lead to gestational diabetes. The precise mechanism still needs further investigation.

Maternal androgen excess was found to be negatively associated with gestational age and birth size of offspring because of intrauterine growth restriction (50, 51). However, another study reported that prolonged gestational age was observed in women with PCOS (52). In our current study, gestational age of all maternal mice was not affected regardless of T treatment, but the body weight of neonatal mice was significantly increased in experiments 2 and 3 after maternal T treatment. The mouse embryo is delivered around the end of organogenesis, whereas the human embryo stays in the uterus for further growth and maturation of most organs (53). This distinction between human and mouse should be taken into consideration during the interpretation of these results. In addition, a different genetic background and environment may also lead to different pregnancy outcome even in the presence of hyperandrogenism (11).

A previous study has already reported that prenatal T exposure could result in aberrant anogenital distance, nipple development, and reproductive tract morphology in female rats (22). Therefore, our current study focused mainly on the internal reproductive system development instead of external genitals. Impaired follicular development and oocyte quality were observed in the ovaries of PCOS patients; this is at least partially contributed by increased androgen production (54, 55). Although clinical and basic studies revealed that hyperandrogenism during gestational stage may lead to abnormal reproductive system development in offspring of these subjects (16, 56, 57), mechanisms were rarely investigated. In our current study, we analyzed ovarian follicular development in offspring of maternal T-treated mice. The percentage of atretic follicle was increased and corpus luteum was decreased in offspring of all 3 experiments. Because all these female offspring were sacrificed at 3 months of age without a standardization of their cycle, the statistical results may have some biases. This is a drawback of our study design. Nevertheless, we can draw a preliminary conclusion that both follicular development and ovulation process might be impaired based on these results. Regan et al. discovered that ovarian granulosa cells play critical roles during the maintenance of folliculogenesis and ovulation (58). Therefore, we evaluated some granulosa cell function–related markers expression level. Both mRNA and protein expression of FOXL2 were decreased in offspring of experiment 3. FOXL2 was reported to be a central transcription factor, which is involved in multiple biological processes in the ovary, including early ovarian development, folliculogenesis, granulosa cell differentiation, and proliferation (59). CYP19A1, which is an important enzyme that contributes to the catalyzation of T to E2 in granulosa cell, was also reported as a direct target of FOXL2 (60). According to our present data, both FOXL2 and CYP19A1 expression were decreased. Because we have already observed impaired E2 and increased T level in offspring of T-treated mice during a late gestational stage. We speculated that disturbed sex hormone balance was caused by altered expression of FOXL2 and CYP19A1 in granulosa cells. This hypothesis was confirmed by our current results that knock-down of FOXL2 in both primary granulosa cell and KGN cell line impaired CYP19A1 expression and T-induced E2 generation. In addition to CYP19A1, decreased FOXL2 was also reported to promote the transdifferentiation of granulosa cell to Sertoli-like cell, which may also be responsible for ovarian insufficiency in female offspring (61). This could be another potential mechanism of disturbed hormone balance and impaired ovarian follicle development in these mice. But this hypothesis still needs further investigation.

As already indicated, the percentage of atretic follicle was increased and corpus luteum was decreased in all 3 experiments, but the mRNA expression of detected markers in granulosa cell was not affected in experiments 1 and 2. These results suggested that, in addition to granulosa cell function, impaired folliculogenesis in experiments 1 and 2 might be contributed by other factors, such as compromised angiogenesis during follicular development (62) or disrupted hypothalamic-pituitary-gonadal axis (63). But these hypotheses still need further investigations.

In conclusion, our current study demonstrated that maternal testosterone excess, especially during late gestational stage, contributes to impaired follicular development and disturbed sex hormone synthesis in female offspring. These impairments were, at least partially, caused by decreased FOXL2 and CYP19A1 expression in ovarian granulosa cell. These findings may have profound implications during the treatment of maternal hyperandrogenism, but because this study was based on animal and cell models, further clinical investigations and clinical data are needed to verify these findings.

Abbreviations

Abbreviations
 
• AR

  androgen receptor

 
• Cyp19a1

  cytochrome P450 family 19 subfamily a member 1

 
• dpc

  days post coitum

 
• E2

  estradiol

 
• FOXL2

  Forkhead box protein L2

 
• Foxo3

  forkhead box O3

 
• Glut4

  glucose transporter 4

 
• HRP

  horseradish peroxidase

 
• Hsd17b1

  hydroxysteroid 17-beta dehydrogenase 1

 
• IHC

  immunohistochemical

 
• PCOS

  polycystic ovary syndrome

 
• Rspo1

  R-spondin-1

 
• siRNA

  small interfering RNA

 
• T

  testosterone

 
• Wnt4

  Wnt family member 4

Acknowledgments

Financial Support: This study was supported by Medical Research Grant of Jiangsu Commission of Health (H2017043 to R.J.).

Author Contributions: Y.Z., J.C., and R.J. designed and performed the experiments, analyzed the data, and wrote the article. A.Z., M.G., Y.L., C.Z., and X.S. performed the experiments and analyzed the data. X.Z. and L.W. performed the IHC staining. The manuscript is guaranteed by J.C. and R.J.

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.