Introduction
An adverse environment during prenatal life programs alterations in the developing organs. In particular, since gametes and the growing fetal ovary are vulnerable to prenatal events,Reference Chan, Tsoulis and Sloboda1 an adverse environment during fetal life could affect the developing gonads and impact on postnatal fertility.
Polycystic ovary syndrome (PCOS) is one of the main causes of female ovarian infertility. Although the etiopathogenesis of PCOS is not yet fully understood, current theories emphasize on genetic and environmental factors during intrauterine life.Reference Franks2 It has been reported that prenatal androgen exposure induces polycystic ovaries and metabolic alterations in several species.Reference Demissie, Lazic, Foecking, Aird, Dunaif and Levine3–Reference Ortega, Rey, Velazquez and Padmanabhan9 In previous works, we found that androgen excess in utero programs two different phenotypes regarding the estrous cycle (anovulatory and irregular ovulatory) with biochemical hyperandrogenism and alterations in ovarian morphology.Reference Abruzzese, Heber and Ferreira7 These phenotypes show differences in metabolic features, being the anovulatory phenotype (PHanov) more affected.Reference Amalfi, Velez and Heber5,Reference Abruzzese, Heber and Ferreira7
In female mammals, alterations in the metabolic status could impact on the energy balance of the ovary, affecting its functions and leading to fertility issues.Reference Torre, Benedusi, Fontana and Maggi10
Ovarian steroidogenesis is strictly regulated by both neuroendocrine negative feedback systems and by the response to the metabolic status.Reference Walker and Gore11 Several molecules such as peroxisome proliferator-activated receptors (PPARs) and adipokines act signaling and responding to the metabolic status in ovaries.Reference Fernandez-Fernandez, Martini and Navarro12,Reference Faut, Elia, Parborell, Cugnata, Tesone and Motta13 Thus, they have been referred to as “fuel sensors” as they monitor energy status and regulate metabolic pathways and also those processes that depend on these.Reference Sandoval, Cota and Seeley14,Reference Shiue, Chen, Tsai, Yeh and Huang15
PPARs are a family of transcriptional nuclear factors with three isoforms—alpha, beta, and gamma—which regulate the gene expression of several molecules.Reference Issemann and Green16,Reference Komar17 The three isoforms are present in the ovaries of different species, but PPAR gamma (PPARG) is the one that has been associated with female fertility and PCOS.Reference Froment, Gizard, Defever, Staels, Dupont and Monget18,Reference Vitti, Di Emidio and Di Carlo19 The activation of PPARG regulates the synthesis of steroid hormones in granulosa cells,Reference Froment, Gizard, Defever, Staels, Dupont and Monget18,Reference Velez, Heber, Ferreira, Abruzzese, Reynoso and Motta20 whereas the disruption of PPARG in the ovary leads to female subfertility.Reference Cui, Miyoshi and Claudio21 In addition, PPARG has two isoforms (PPARG1 and PPARG2), both of which are found in the ovaries but whose regulation and action are still unknown.
Adipokines regulate reproductive functions in gonads and through the hypothalamic-pituitary axis.Reference Mitchell, Armstrong, Robker and Norman22,Reference Reverchon, Ramé, Bertoldo and Dupont23 The adipokines secreted by ovaries play autocrine and paracrine functions, directly affecting ovarian signaling.Reference Bharati, Bharti, Kar and Sahoo24 In PCOS patients, the pattern of serum adipokines is altered independently of the body mass index and body weight. Thus, it has been proposed that they contribute to the pathology of PCOS as well as to the related metabolic abnormalities.Reference Reverchon, Ramé, Bertoldo and Dupont23,Reference Chen, Jia, Qiao, Guan and Kang25
The transcription of adipokines such as adiponectin and chemerin seems to be regulated by PPARG because they show PPAR response elements on their promoter.Reference Bharati, Bharti, Kar and Sahoo24,Reference Muruganandan, Parlee, Rourke, Ernst, Goralski and Sinal26 Moreover, PPARG agonists decrease the levels of the adipokine leptin, another energy metabolism regulator also involved in the functionality of the hypothalamic-pituitary-gonadal axis.Reference Considine27–Reference Elias and Purohit30
Based on this, we hypothesized that prenatal androgen excess programs alterations in the metabolic signaling within the ovary, which impact on ovarian functions. Thus, we aimed to study the effect of prenatal hyperandrogenization on the ovarian fuel sensors: PPARG and the adipokines leptin, adiponectin and chemerin, and on the steroidogenesis pathway.
Material and methods
Animals and treatments
Virgin female rats (90–110 days of age) of the Sprague Dawley strain were mated with fertile males of the same strain. Three females and one male were housed in each cage under controlled conditions of light (12 h light, 12 h dark) and temperature (23–25 °C). Animals received food and water ad libitum. Day 1 of pregnancy was defined as the day on which spermatozoa were observed in the vaginal fluid. To study the fetal programming effect of androgens, we used a rodent model already established in our laboratory.Reference Abruzzese, Heber and Ferreira7 Briefly, pregnant rats (N = 15) received subcutaneous injections of 1 mg of free testosterone (T-1500; Sigma, St. Louis, MO, USA) dissolved in 100 µl sesame oil from day 16 to day 19 of pregnancy. This hormonal paradigm mimics the fetal testosterone surge observed in male rats when the reproductive axis in the fetus is established. Another group (N = 15) received only 100 µl of sesame oil. Under the conditions of our animal facilities, spontaneous term labor occurs on day 22 of gestation. Female offspring were separated from males at 21 days of age. Those from hyperandrogenized mothers were assigned as the prenatally hyperandrogenized (PH) group, whereas those from mothers injected with sesame oil were the control group. Animals were allowed free access to Purina rat chow (Cooperación SRL, Argentina) and water.
At 60 days of age, female offspring were weighed and then anesthetized with carbon dioxide and killed by decapitation. Trunk blood was collected and serum and ovaries were separated, weighed, and kept at –80 °C for further studies. All animals were randomly assigned for each assay considering their litter precedence. Care was taken when assigning and equilibrating the number of animals from each littermate to all the assays, to prevent the maternal effect on the results. Most of the animals used for this work have been studied in previous projects from our laboratory.Reference Abruzzese, Heber and Ferreira7
All the procedures involving animals were conducted in accordance with the Animal Care and Use Committee of Consejo Nacional de Investigaciones Científicas y Tecnicas (CONICET) 1996, Argentina, and the study was approved by the Ethics Committee of the School of Medicine of University of Buenos Aires, Argentina.
Determination of the phenotype according to the regularity of the estrous cycle
The estrous cycle of all the animals was determined by vaginal smears taken daily from 45 to 60 days of age and classified as previously reportedReference Abruzzese, Heber and Ferreira7,Reference Karim, Landolfi and Christian31 (Fig. 1).
Fig. 1. Estrous cycle scheme for animals of each of the phenotypes defined from the postnatal day 45 to the time of sacrifice (around day 60). The animals defined as showing regular ovulatory phenotypes presented cycles that last between 4 and 6 days and respected the passage from proestrous, estrous, metaestrous, and diestrous. Rats, whose cycles lasted 7 days or more, showed some smears displaying an estrous stage, but other smears not following the progress of the cycle, as described above, were considered irregular ovulatory animals (PHiov). Rats that showed the metaestrous and diestrous stages or a combination of both for 4 consecutive days were considered to be noncycling and thus considered the anovulatory animals (PHanov).Reference Abruzzese, Heber and Ferreira7,Reference Karim, Landolfi and Christian31
The dose of testosterone used in the present study (1 mg) leads 123 to both the irregular ovulatory phenotype (PHiov) and the anovulatory phenotype (PHanov). As the PHanov group remained mainly in diestrous to allow the comparison between the phenotypes, all animals were sacrificed at this stage.
Follicle count and follicular growth
The ovaries were removed and immediately fixed in 4% (v/v) formaldehyde, dehydrated in ethanol, and embedded in paraffin. Serial 6-μm thick sections were mounted at 50 μm intervals onto microscope slides, to prevent counting the same structure twice,Reference Woodruff, Lyon, Hansen, Rice and Mather32 and stained with hematoxylin-eosin. Histological serial sections were analyzed independently by two of the authors, and ovarian follicles were classified and quantified. Follicles were classified as primordial, primary, preantral, antral, corpora lutea, and cysts.Reference Paixão, Velez and Santos33,Reference Abramovich, Irusta, Bas, Cataldi, Parborell and Tesone34 The number of different structures was determined in five ovaries per group.
Quantification of mRNA levels by real-time PCR
We assessed the mRNA expression of the Pparg2 isoform and Pparg1 + 2 isoforms, the adipokines (Leptin, Chemerin, and Adiponectin), the ovarian gonadotropin receptors Lh-r and Fsh-r, and the steroidogenic factors and enzymes: Steroidogenic acute regulatory protein (Star), cytochrome P450 family 11 subfamily A member 1 (Cyp11a1), 3B-Hydroxysteroid dehydrogenase (3bhsd), Cytochrome P450 family 17 subfamily A1 (Cyp17a1), and cytochrome P450 family 19 subfamily A1 and aromatase (Cyp19a1). All mRNA levels were evaluated by real-time PCR analysis. Briefly, total mRNA from ovarian tissue was extracted using RNAzol RT (MRC gene, Molecular Research Center, Cincinnati, OH, USA), following the manufacturer’s instructions. cDNA was synthesized from 1 μg mRNA using random primer hexamers (Invitrogen-Life Technologies, Buenos Aires, Argentina). Real-time PCR analysis was performed from this cDNA (2.5 μL) in 10 μL reaction buffer containing a 20 mM dNTPs mix, GoTaqPolymerase (Promega), Eva Green 20x (Biotium Hayward, CA, USA), and gene-specific primers in a total volume of 12.5 μL. The qPCR conditions started with a denaturation step at 95 °C for 5 min and followed by up to 40 cycles of denaturation (95 °C), annealing (see temperature for each primer in Table 1) and primer extension (72 °C). The amplified products were quantified by fluorescence, using the Rotor Gene 6000 Corbett, and mRNA abundance was normalized to the amount of 60s Ribosomal protein L32 (L32). L32 was validated as a reference gene as the variance between treatments did not differ. Gene expression was quantified using the 2 –ΔΔCt method.Reference Livak and Schmittgen35 Results are expressed as a fold value of the controls.
Table 1. List of primers used in real-time PCR
F, forward sequence; R, reverse sequence
Protein expression analysis
Protein expression in ovarian tissue was determined by Western blot analysis. Briefly, ovarian tissue was lysed for 20 min at 4 °C in lysis buffer (20 mM Tris-HCl, pH = 8.0, 137 mM NaCl, 1% Nonidet P-40, and 10% glycerol) supplemented with protease inhibitors (Sigma–Aldrich, St. Louis, MO, USA). The lysate was centrifuged at 4 °C for 10 min at 10,000 g and the pellet discarded. Protein concentrations in the supernatant were measured by the Bradford assay (Bio-Rad, Hercules, CA, USA). After boiling for 5 min, 50 μg of each protein was applied to an SDS-polyacrylamide gel (12% for PPARG and 3BHSD and 15% for adipokines) and electrophoresis was performed at 80 volts for 1.5 h. The separated proteins were transferred onto nitrocellulose membranes in transfer buffer (20% methanol, vol/vol; 0.19 M glycine; 0.025 M Tris-Base, pH = 8.3) for 1 h at 400 mA and 4 °C. Blocking was carried out for 1 h at room temperature in 5% (w/v) nonfat dry milk in phosphate-buffered saline (PBS) and membranes were incubated with the primary antibody (diluted in 1% (w/v) bovine serum albumin in PBS) overnight at 4 °C. Anti-PPARG 1 + 2 1:1000 (ab41928, Abcam, Cambridge, UK), anti-LEPTIN (sc-393043 1:400, Santa Cruz Biotechnology, CA, USA), anti-ADIPONECTIN (ab22554 1:1000, Abcam), anti-CHEMERIN 1:1000 (ab112520, Abcam), and anti-3BHSD 1:200 (sc-28206, Santa Cruz Biotechnology) were used as primary antibodies. The protein bands were visualized by incubating the blots with horseradish peroxidase-conjugated secondary antibody (Rabbit anti-mouse IgG H&L, HRP, 1:5000, ab6728, Abcam or Goat anti-rabbit IgG H&L, HRP 1:2000, 1706515, BioRad) for 1 h, followed by ECL Western Blotting Substrate (Thermo Scientific, IL, USA). Rainbow-colored protein mass markers (14.3-200 kDa, Bio-Rad) were applied to samples as molecular weight standards.
The consistency of protein loading was evaluated by staining the membranes with Ponceau-S and normalized applying the protein B-TUBULIN (ab131205, Abcam) or the protein B-ACTIN (Sigma. St. Louis, MO, USA). The intensities (area × density) of the individual bands were quantified by ImageQuant LAS 4000 (GE Healthcare Life Sciences, NJ, USA). Results are expressed as arbitrary units.
Serum determinations
LH and FSH were determined with radioimmunoassay (RIA) kits (n = 7 in duplicates for each of the phenotypes established), following the protocols previously described.Reference Lacau de Mengido, Becú-Villalobos and Libertun36,Reference Lacau-Mengido, Libertun and Becú-Villalobos37 For these hormones, the intra- and interassay coefficients were less than 10% and 13%, respectively.
The estradiol to testosterone ratio (E2/T) was determined as a marker of ovarian function.Reference Horng, Wang and Wang38 Serum testosterone and estradiol were already measured in the animal model and reported elsewhere before.Reference Abruzzese, Heber and Ferreira7 Animals from both the phenotypes (PHiov and PHanov) showed higher levels of testosterone than control group (Control = 35,9 ± 13,4 pg/mL; PHiov = 291,0 ± 82,7 pg/mL; PHanov = 344,5 ± 15,0 pg/mL), but only PHanov phenotype showed a significant difference in estradiol levels as compared with controls values (Control = 12,25 ± 2,31 pg/mL; PHiov = 9,69 ± 1,93 pg/mL; PHanov = 8,05 ± 1,24 pg/mL).Reference Abruzzese, Heber and Ferreira7
Serum testosterone was quantified by RIA as previously described,Reference Abruzzese, Heber and Ferreira7 whereas serum estradiol levels were quantified by Cobase immunoassay analyzers using an Electro Chemiluminescence ImmunoAssay, following the manufacturer’s instructions.Reference Abruzzese, Heber and Ferreira7 Progesterone serum levels were measured by RIA, as previously described.Reference Abraham, Swerdloff, Tulchinsky and Odell39 Progesterone was extracted twice with diethyl ether. Progesterone antiserum was highly specific and showed low cross-reactivity. The intra- and interassay coefficients of variation were 10.9 and 12.8%, respectively. Values were expressed as ng/ml of serum progesterone.
Statistical analysis
Statistical analyses were carried out using the Instat program (GraphPad software, San Diego, CA, USA). T-student test or ANOVA with post hoc Tukey test were used to compare between two or three groups, respectively.
A generalized linear model (GLM) with binomial distribution was used to study the association between the proportion of anovulatory animals (classified as anovulatory or not) per female offspring per litter and the number of female dams, the number of male dams, and the total number of offspring per litter. These analyses were performed using the R environment (R Core team 2014).
Statistical significance was considered as p < 0.05.
Results
Litters analysis
The litters’ details are shown in Table 2. Prenatal hyperandrogenization did not affect gestation time, the number of offspring neither the number of female nor male offspring. The prenatal hyperandrogenized animals showed alterations in their estrous cycles. When classifying the animals as anovulatory or not, the analysis showed that the proportion of the PHanov per females per litter was not associated with the total number of offspring (Fig. 2A, p = 0.875) nor with the number of males or females per offspring (Fig. 2B, p = 0.4202 and Fig. 2C, p = 0.0581, respectively).
Fig. 2. Evaluation of the association between the proportion of anovulatory animals per litter and (A) the total number of offspring (p > 0.05), (B) the number of male dams (p > 0.05), and (C) the number of female dams (p > 0.05). Statistical analyses were made by a generalized linear model (GLM) with binomial distribution.
Table 2. Litter characterization between control and hyperandrogenized groups, n = 15 litters analyzed per group. Statistical analyses were made by t-students test. No differences were found between Control and hyperandrogenized litters for gestation time or litter analysis p > 0.05
Body and ovarian weight
Prenatal hyperandrogenization did not affect body weight or ovarian weight at 60 days of age. The PHanov phenotype showed a lower ovarian/body weight ratio (Table 3).
Table 3. Body weight, ovarian weight, and ovarian weight/body weight ratio for the prenatally hyperandrogenized (PH) and Control groups for eight rats per group
Statistical analyses were made by ANOVA, a vs. b, p < 0.05.
Ovarian follicle count and follicular development
To evaluate the effect of prenatal hyperandrogenism on folliculogenesis, we analyzed the ovarian morphology and the percentage of different follicular stages. No changes were found in the percentages of primordial follicles between groups (Fig. 3A, p > 0.05). The percentage of primary follicles was higher in the PHanov phenotype as compared with the Control group (Fig. 3B, p < 0.05). The percentage of preantral follicles was lower in the PHiov phenotype as compared with the Control group (Fig. 3C, p < 0.05). There were no differences between the percentages of antral follicles in the PH groups and the Control group (Fig. 3D, p > 0.05). The percentage of corpora lutea was lower in both the PH groups as compared with the Control group (Fig. 3E, p < 0.05). Regarding follicular cysts, we found that as expected control ovaries did not present follicular cysts, whereas both PH groups showed cysts (Fig. 3F, p < 0.05).
Fig. 3. Effects of prenatal hyperandrogenization on the quantification of the percentage of each follicular stage in the ovaries of the prenatally hyperandrogenized (PH) and Control groups. (A) Percentages of primordial follicles, (B) percentages of primary follicles, (C) percentages of preantral follicles, (D) percentages of antral follicles, (E) percentages of corpora lutea, and (F) percentage of cysts. Each column represents the mean ± SD from five animals per group. Statistical analyses were made by ANOVA; different letters mean statistically significant differences (a vs. b, p < 0.05).
PPARG as an ovarian fuel sensor marker
Prenatal hyperandrogenization affected the PPARG system. Both PH phenotypes showed higher levels of mRNA of Pparg1 + 2 (Fig. 4A, p < 0.01), but no alterations in the levels of Pparg2 compared with the Control group (Fig. 4B, p > 0.05). Thus, as Pparg has two isoforms, the levels of Pparg1 were the ones altered in the PH groups. To confirm this point, we evaluated the protein levels of both PPARG isoforms and found that the protein levels of PPARG1 were higher in both PH groups than in the Control group (Fig. 4C, p < 0.05) and that those of PPARG2 remained unaltered in the PH groups as compared with the Control group (Fig. 4D, p > 0.05).
Fig. 4. Effects of prenatal hyperandrogenization on the PPARG ovarian system in the prenatally hyperandrogenized (PH) and Control groups. The graphs correspond to (A) mRNA abundance of Pparg1 + 2 (p < 0.01) and (B) mRNA abundance of Pparg2 (p > 0.05), both relative to L32. (C) Protein levels of PPARG1 (p < 0.05) and (D) protein levels of PPARG2 (p > 0.05), both relative to B-TUBULIN levels. Each column represents the mean ± SD from six animals per group. Statistical analyses were made by ANOVA; different letters mean statistically significant differences (a vs. b, p < 0.05).
Characterization of the gene and protein expression of ovarian adipokines
The adipokine ovarian secretion pattern was found to be differentially deregulated in both PH phenotypes. Leptin mRNA levels were lower in the PHanov phenotype but not altered in the PHiov phenotype as compared with the Control group (Fig. 5A, p < 0.01). We found no alterations in the leptin protein levels (Fig. 5B, p > 0.05). Regarding Adiponectin mRNA expression, the PHiov phenotype showed an increase in its levels as compared with both the PHanov and Control groups (Fig. 5C, p < 0.01). Adiponectin protein levels were lower in both of the PH phenotypes (Fig. 5D, p < 0.01). Chemerin mRNA and protein levels were only altered in the PHiov phenotype, showing higher levels than the Control and PHanov groups (Fig. 5E, 5F, p < 0.05).
Fig. 5. Effects of prenatal hyperandrogenization on the adipokine ovarian pattern. The graphs correspond to (A) mRNA abundance of leptin (p < 0.01), (B) protein levels of leptin (p > 0.05), (C) mRNA levels of adiponectin (p < 0.01), (D) protein levels of adiponectin (p < 0.01), (E) mRNA levels of chemerin (p < 0.05), and (F) protein levels of chemerin (p < 0.05) of the prenatally hyperandrogenized (PH) and Control groups. Each column represents the mean ± SD from 6 animals per group for mRNA analysis and 10 animals per group for protein levels analysis. Statistical analyses were made by ANOVA; different letters mean statistically significant differences (a vs. b, p < 0.05).
Hormonal regulation of steroidogenesis
Neither of the PH groups showed altered levels of the gonadotropins FSH and LH as compared with the Control group (Fig. 6A, 6B, respectively, p > 0.05).
Fig. 6. Effect of prenatal hyperandrogenization on gonadotropin secrerum levels, ovarian receptors, and hormone ratios in the prenatally hyperandrogenized (PH) and control groups. (A) Serum FSH levels (p > 0.05) and (B) serum LH levels (p > 0.05); (C) mRNA abundance of ovarian Fsh-r (p < 0.05) and (D) mRNA abundance of ovarian Lh-r (p < 0.05), both relative to L32; (E) the LH to FSH ratio (p > 0.05), (F) estradiol to testosterone ratio (p < 0.05) of the prenatally hyperandrogenized (PH) and Control groups. Each column represents the mean ± SD from six animals per group for mRNA expression levels analysis and from seven animals per group for hormonal measurements. For the estradiol/testosterone ratio, seven animals per group were considered. Statistical analyses were made by ANOVA; different letters mean statistically significant differences (a vs. b, p < 0.05).
The PHiov phenotype showed lower levels of mRNA expression of Fsh-r than the Control group (Fig. 6C, p < 0.05), whereas the PHanov phenotype showed an increased expression of Lh-r mRNA levels as compared with the Control group (Fig. 6D, p < 0.05). The LH to FSH ratio, proposed as a marker of PCOS, presented no differences between the PH groups and the Control group (Fig. 6E, p > 0.05).
We have previously reported that testosterone levels are higher in both PH phenotypes and that estradiol levels are lower in the PHanov phenotype.Reference Abruzzese, Heber and Ferreira7 Here, we showed that both PH groups showed lower levels of the E2 to T ratio than the Control group (Fig. 6F, p < 0.05).
We analyzed progesterone serum levels and the expression of 3bhsd, the limiting enzyme of its synthesis. Both, the mRNA and protein levels of 3bhsd, were not affected in the PH groups as compared with the controls (Fig. 7A, 7B, p > 0.05). Progesterone serum levels were decreased in both the PH phenotypes as compared with the Control group levels (Fig. 7C, p < 0.05).
Fig. 7. Effect of prenatal hyperandrogenization on progesterone synthesis and serum levels in the prenatally hyperandrogenized (PH) and control groups. (A) mRNA abundance of ovarian 3bhsd relative to L32 (p > 0.05), (B) protein levels of ovarian 3BHSD relative to B-TUBULIN levels (p > 0.05), and (C) progesterone levels (p < 0.05) of the prenatally hyperandrogenized (PH) and Control groups. Each column represents the mean ± SD from 6 animals per group for mRNA expression, 10 animals per group for protein levels analysis, and from 7 animals per group for progesterone measurements. Statistical analyses were made by ANOVA; different letters mean statistically significant differences (a vs. b, p < 0.05).
Ovarian steroidogenic pathway
Prenatal hyperandrogenization affected the steroidogenic pathway. The mRNA levels of Star were higher in the PHanov phenotype than in the Control and PHiov groups (Fig. 8A, p < 0.05). Besides, the mRNA levels of Cyp11a1 were higher in the PHanov phenotype than in the Control group (Fig. 8B, p < 0.05). The mRNA levels of Cyp17a1, a limiting enzyme in androgen synthesis, were lower in the PHiov phenotype than in the Control group but showed no alterations in the PHanov phenotype (Fig. 8C, p < 0.05). The mRNA levels of Cyp19a1 (Aromatase) were lower in the PHanov phenotype than in the Control group (Fig. 8D, p < 0.05).
Fig. 8. Effect of prenatal hyperandrogenization on ovarian steroidogenic factors and enzymes. The graphs correspond to mRNA abundance of (A) Star (p < 0.05), (B) Cyp11a1 (p < 0.05), (C) Cyp17a1 (p < 0.05), and (D) Cyp19a1 (aromatase) (p < 0.05), all of them relative to mRNA abundance of L32 of the prenatally hyperandrogenized (PH) and Control groups. Each column represents the mean ± SD from six animals per group. Statistical analyses were made by ANOVA; different letters mean statistically significant differences (a vs. b, p < 0.05).
Discussion
Reproductive functions depend on the energy balance, and several metabolic misbalances may negatively impact on female reproduction.Reference Torre, Benedusi, Fontana and Maggi10 Such as in the case of PCOS, these patients display not only reproductive alterations, but also metabolic disturbances.
We have previously reported that prenatal hyperandrogenization could lead to two different phenotypes regarding the estrous cycle and that both showed altered systemic metabolism and ovarian morphological alterations.Reference Abruzzese, Heber and Ferreira7 In the present study, we show that, after prenatal hyperandrogenization, the ovary signaling of fuel sensors is impaired and that each phenotype presents a different deregulation that affects cyclicity and the steroidogenic pathway within the ovary.
In the present study, we evaluated a period of transition to adulthood (60 days of age).Reference Sengupta40–Reference Spear42 Our results showed that the reproductive phenotype did not depend on the number of males or females per litter. In the same litter, we could detect animals that showed PHanov while their siblings not. Thus suggesting a plastic response to prenatal androgens. A plastic phenotype has also been described in PCOS patients and their relatives. Particularly, Jahanfar et al. (1995)Reference Jahanfar, Eden, Warren, Seppälä and Nguyen43 showed in a study of monozygotic and dizygotic twins that while one of the twins may show polycystic ovaries with hyperandrogenism and/or menstrual dysfunction, the other twin shows only some or none of the features. However, the complex network of mechanisms that leads to the described two phenotypes found in this work may involve not only developmental plasticity phenomena,Reference Bateson, Gluckman and Hanson44 but also embryo position in utero and differential placental dysfunctions. The position of a fetus in relation to the sexes of its neighboring intrauterine littermates has been reported to influence its exposure to gonadal hormones and therefore its development.Reference Zielinski, Vandenbergh and Montano45 Moreover, regarding placental dysfunction, it has been reported a deficit in placental aromatase in PCOS women, thus compromising placental protection mechanisms to androgen actions.Reference Maliqueo, Lara, Sánchez, Echiburú, Crisosto and Sir-Petermann46 Sun et al. (2012) showed that prenatal hyperandrogenized rats presented altered placental steroidogenesis without affections on the maternal metabolic status or circulating estradiol, suggesting that the effect of testosterone on placental and fetal growth in rats is direct and not a consequence of alterations in metabolism and steroid hormones.Reference Sun, Maliqueo and Benrick47 However, in other species, for example, in sheep and monkeys, the effects of testosterone on programming reproductive alterations might be caused by direct androgenic effects or by an indirect action via estrogenic pathways as a result of testosterone aromatization.Reference Puttabyatappa, Cardoso, Herkimer, Veiga-Lopez and Padmanabhan48–Reference Luense, Veiga-Lopez, Padmanabhan and Christenson50 Thereby, the response to androgens in different species needs to be further studied.
The ovary weight/body weight ratio, usually used as a marker of prenatal growth when the gonads are developed, has been described as a better marker than organ weight.Reference Fujikura and Froehlich51,Reference Mitropoulos, Scurry and Cussen52 Here, we found that this ratio was impaired in the PHanov phenotype but not in the PHiov phenotype. This lower ratio in the PHanov phenotype may reflect a harmful effect of the prenatal exposure of the developing organ to testosterone, which conditions its growth and functions during postnatal life. This may explain, at least in part, the differences in estrous cyclicity between the two PH phenotypes.
Ovarian energetic signaling involves two important master energetic fuel sensors: PPARG and adipokines. PPARG may regulate the gene expression of several adipokines and is in turn regulated by complex networks of adipokine pattern.Reference Bharati, Bharti, Kar and Sahoo24,Reference Muruganandan, Parlee, Rourke, Ernst, Goralski and Sinal26,Reference Cabrero, Cubero and Llaverías53,Reference Mirzaei, Hossein-Nezhad and Keshavarz54
In the present study, we found an altered expression of Pparg in both PH groups. Specifically, we found that both PPARG1 and PPARG2 were expressed in the ovaries of the PH groups and that PPARG1, the most abundant isoform in the ovary,Reference Komar17,Reference Froment, Gizard, Defever, Staels, Dupont and Monget18 was impaired in both PH groups. It has been reported that PPARG may influence the development of granulosa cells and that alterations in its expression may affect oocyte developmental competence, thus compromising ovary functions.Reference Komar17,Reference Froment, Gizard, Defever, Staels, Dupont and Monget18 Thus, the impairment of ovarian PPARG may be contributing to the dysregulation of the estrous cycle found in the PH groups. On the other hand, PPARs may also modulate estradiol levels by regulating Aromatase expression.Reference Keller, Givel, Perroud and Wahli55–Reference Lovekamp-Swan, Jetten and Davis58 In agreement with these findings, we have previously reported that estradiol levels were lower in the PHanov phenotype than in controls,Reference Abruzzese, Heber and Ferreira7 whereas here we found that Aromatase mRNA expression was also decreased in this phenotype. These findings can be due to the action of the increased PPARG levels that contribute to the hormonal imbalance of the estradiol to testosterone ratio. These results are in accordance with previous reports which have shown that, in a hyperandrogenic context, Pparg is upregulated, contributing to alterations in androgen production.Reference Amalfi, Velez and Heber5,Reference Ortega, Rey, Velazquez and Padmanabhan9 Our results add evidence to the idea that PPARG may be one of the mediators in the developmental programming of ovarian dysfunctions in a context of androgen excess.Reference Ortega, Rey, Velazquez and Padmanabhan9
Adipokines can regulate ovarian steroidogenesis, oocyte maturation, and embryo development.Reference Reverchon, Ramé, Bertoldo and Dupont23 Our present research was focused on three adipokines: leptin, adiponectin, and chemerin, which are known to be involved in the regulation of ovarian steroidogenesis and related to the master metabolic regulator PPARG.Reference Reverchon, Ramé, Bertoldo and Dupont23,Reference Muruganandan, Parlee, Rourke, Ernst, Goralski and Sinal26,Reference Barkan, Jia, Dantes, Vardimon, Amsterdam and Rubinstein59–Reference Wang, Leader and Tsang63 Regarding Leptin levels, the PHanov phenotype showed a decrease in its mRNA expression. In fact, decreased leptin levels are related to an infertility status.Reference Cornejo, Hentges, Maliqueo, Coirini, Becu-Villalobos and Elias64 In addition, leptin levels have been shown to be inversely correlated to PPARG levels in several tissues.Reference Cabrero, Cubero and Llaverías53,Reference Wang, Shao and Ballock65,Reference Abbasi, Moghadam, Kahrarian, Abbsavaran, Yari and Alizadeh66 We did not find alterations in leptin protein expression, possibly due to a deregulation of its secretion. It has been shown that androgens might impact negatively on leptin exocytosis.Reference Jenks, Fairfield, Johnson, Morrison and Muday67 Thus, as in the PHanov phenotype, we found an increase in androgen production, which may impair leptin secretion. Regarding the other adipokines, in the PHanov phenotype, we found no alterations in chemerin levels, but a decrease in the protein levels of adiponectin without affecting the mRNA levels of this adipokine. It has been reported that adiponectin modulates follicle maturation and particularly preovulatory changes in the ovary, but also steroidogenesis.Reference Dobrzyn, Smolinska and Kiezun68,Reference Ledoux, Campos, Lopes, Dobias-Goff, Palin and Murphy69 These findings are in agreement with our results as low levels of both of these adipokines may be related to the increase of primary follicles found in this phenotype, thus contributing to the reproductive issues and ovulatory dysfunctions. It is also important to highlight that adiponectin acts as an insulin-sensitizing agent in several tissues,Reference Dupont, Chabrolle, Ramé, Tosca and Coyral-Castel28,Reference Michalakis and Segars70 and it has been found that insulin-resistant PCOS patients have lower adiponectin levels.Reference Michalakis and Segars70 In accordance with these results, we have already reported that the PHanov phenotype showed insulin resistance.Reference Abruzzese, Heber and Ferreira7 Thereby, these results suggest that impaired ovarian adiponectin expression may be associated with metabolic alterations.
In the PHiov phenotype, the leptin levels remained unaltered compared with the Control group. Regarding adiponectin expression, we found that, although the protein levels of this adipokine were low, there was a high expression of the mRNA levels. These results suggest an increase in Adiponectin mRNA transcription, possibly in response to the low ovarian protein levels. It is important to highlight that many post-transcriptional mechanisms may modulate mRNA translation. In accordance to this, miR-378 may modulate Adiponectin expression in adipose tissueReference Ishida, Shimabukuro and Yagi71; this miRNA is expressed in ovary and also downregulates Aromatase expression,Reference Xu, Linher-Melville, Yang, Wu and Li72 but to our knowledge, there have been no studies on miR-378 in PCOS and neither on ovarian adiponectin regulation. Thus, the regulation of this adipokine in the ovary in an androgenic context needs further study.
The PHiov phenotype also showed increased levels of chemerin expression. It is known that this adipokine may inhibit steroidogenesis.Reference Wang, Kim, Xue, Liu, Leader and Tsang62 Therefore, adiponectin and chemerin may jointly contribute to the ovarian alterations observed in the PHiov phenotype.
Ovarian steroidogenesis is a process that depends on energy expenditure and neuroendocrine control and is regulated by autocrine signaling. Our results showed that the levels of both pituitary hormones LH and FSH were not altered in the PH groups as compared with the Control group. However, we found altered expression of the ovarian receptors Fsh-r and Lh-r, suggesting a possible impairment of the LH and FSH signaling in the ovaries. These data are in agreement with those of other authors who showed that in a hyperandrogenic context, gonadotropin levels may remain unaltered, although alterations in their pulsatility may be contributing to neuroendocrine dysfunction.Reference Foecking, Szabo, Schwartz and Levine73,Reference Yan, Yuan, Zhao, Cui and Liu74 Here, we found that although LH and FSH secretion may not be impaired by prenatal hyperandrogenization, the ovarian response is affected, a fact also reflected in steroidogenic alterations and cyclicity deregulations found. In addition, it has to be considered that the secretion of these gonadotropins is pulsatile and that we only evaluated the levels at the diestrous stage. It is important to point out that within the ovary, FSH is necessary to induce the maturation of ovarian follicles and that together with estrogen it leads to ovulation.Reference Hunzicker-Dunn and Maizels75 Here, we found that the PHiov phenotype showed a decreased expression of Fsh-r, which affected the ovarian response to this hormone and low levels of the estradiol to testosterone ratio, which contributed to the irregular estrous cycles present in this phenotype. On the other hand, we found that the PHanov phenotype showed high levels of Lh-r, which allowed increased response to this hormone and contributed to an exacerbated androgen production,Reference Hunzicker-Dunn and Maizels75,Reference Raju, Chavan and Deenadayal76 which is reflected on the high serum testosterone levels and an impaired E2/T ratio found.
Alterations in hormone and metabolic sensors levels may lead to defects in follicular development. To further characterize these changes, we evaluated the ovarian percentage of each follicular stage, corpora lutea, and cysts. In ovaries from PH rats, we found more primary follicles and less preantral follicles than in ovaries from control rats. Moreover, PH ovaries showed an increase in the percentage of cysts. These results suggest that follicular development would be impaired in PH rats. Contrary to our findings, in PCOS women it has been shown that preantral follicles are increased.Reference Chang and Cook-Andersen77 However, the decrease of preantral follicles has also been reported in other androgenic contexts and PCOS models in rodents (Reviewed in Noroozzadeh et al. (2017)Reference Noroozzadeh, Behboudi-Gandevani, Zadeh-Vakili and Ramezani Tehrani78). It is likely that preantral follicles in rats start to grow, although this development is abnormal and tend to form cysts. Surprisingly, we found corpora lutea in the PHanov animals. This has already been reported by other authors, who described the presence of corpora lutea in rats that were acyclic and did not present vaginal opening.Reference Amalfi, Velez and Heber5,Reference Wu, Li and Wu79 This phenomenon may be explained, as these animals showed their vaginal opening around day 33 of life, which matches the first ovulation.Reference Caligioni80 However, as we started following the vaginal smears at day 45 of life, the observed corpora lutea may belong to those previous cycles that were not assessed and can be detected because they persist for three consecutive cycles (12–14 days).Reference Greenwald and Rothchild81 In accordance with our findings, in young PCOS patients, the menarche usually occurs, but then develops secondary amenorrhea as they grow up.Reference Carmina, Oberfield and Lobo82 It is important to highlight that we found the corpora lutea of PH animals show alterations in their structure (see Abruzzese et al. 2016Reference Abruzzese, Heber and Ferreira7). Moreover, in a previous work from our group,Reference Amalfi, Velez and Heber5 we found that acyclic rats also showed corpora lutea with abnormal luteinization; here, we found a decrease in progesterone levels of PH animals, thus suggesting a dysfunction in corpora lutea.
It is known that steroidogenic enzymes are controlled by the ovarian fuel sensors PPARG and adipokines. Our results showed that Star and Cyp11a1, the two enzymes that participate in the first steps of the steroidogenic process, increased in the PHanov phenotype and slightly decreased in the PHiov phenotype. This indicates that in the PH groups, the steroidogenesis process is impaired from the very first steps. We also found a phenotype-specific pattern in the expression of Cyp17a1 and Cyp19a1, the limiting enzymes of androgen and estradiol synthesis, respectively. The PHiov phenotype showed low levels of Cyp17a1 and a tendency to low levels of Cyp19a1. It has been reported that adiponectin may regulate Cyp17a1 by inhibiting androgen production in cattle,Reference Lagaly, Aad, Grado-Ahuir, Hulsey and Spicer83 but also PPARG system may modulate Cyp17a1 expression.Reference Velez, Heber, Ferreira, Abruzzese, Reynoso and Motta20,Reference Veldhuis, Zhang and Garmey84 Thus, in this phenotype, steroid synthesis may be partly buffered by the action of adiponectin and of PPARG. On the other hand, chemerin may act downregulating Cyp19a1 expression and inhibiting estradiol secretion.Reference Reverchon, Cornuau, Ramé, Guerif, Royère and Dupont85 In accordance with these results, we found a tendency to a decrease in Cyp19a1 levels, a result that agrees with our previous findings of a tendency to low levels of serum estradiol in this phenotype.Reference Abruzzese, Heber and Ferreira7
The PHanov group showed no alterations in Cyp17a1 but a decrease in Cyp19a1 levels. It is known that leptin may regulate aromatase activity by stimulating estrogen production. As already mentioned, PPARG activation upregulates Star and downregulates Cyp19a1, and both PPARG and leptin are implicated in the dysregulation of the steroidogenic processes in the PHanov phenotype.Reference Kitawaki, Kusuki, Koshiba, Tsukamoto and Honjo86,Reference Fan, Yanase and Morinaga87
These results show that, despite the absence of obesity or overweight, prenatal hyperandrogenization leads to an impaired ovarian energy status. We found that prenatal hyperandrogenism affects the energy expenditure within the ovary, altering the master fuel sensor PPARG, the adipokines leptin, adiponectin, and chemerin, and also impacts on ovarian steroidogenesis and the regulation of ovarian gonadotropin signaling leading to hormonal imbalances (Fig. 9).
Fig. 9. Effects of prenatal hyperandrogenization on ovarian folliculogenesis and steroidogenesis. Ovarian folliculogenesis and steroidogenesis depend on gonadotropins (LH and FSH) and fuel sensors such as PPARG, leptin, adiponectin, and chemerin. An impaired profile of ovarian steroidogenic factors and enzymes as well as altered steroid production was found on PH groups. Moreover, prenatal hyperandrogenization also led to alterations in folliculogenesis. Gray arrows and equals signs correspond to PHiov group. Black arrows and equals signs correspond to PHanov group. Gray and black gradient arrows show the tendency to decrease (down arrow) or to increase (up arrow) for the different parameters in the PHiov and PHanov phenotypes, respectively. Check marks indicate the presence of cysts. PCO = polycystic ovaries. *Serum testosterone and estradiol were already measured in the animal model and reported elsewhere.
Author ORCID
Giselle Adriana Abruzzese 0000-0003-3498-7010
Acknowledgments
We thank Enzo Cuba and Marcela Marquez for their technical support in the animal care and Pablo Milla Carmona for his assistant with statistical analysis.
Funding
The present study was supported by grants from Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) (Grants PICT 689/2013 and PICT 632/2016), Argentina. GAA, MFH, FCVA, and SRF are supported by fellowships awarded by CONICET. ABM is a PhD principal investigator from CONICET.
Conflicts of Interest
The authors have no conflicts of interest to declare.
Ethical Standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national guides on the care and use of laboratory animals committee of Consejo Nacional de Investigaciones Científicas y Tecnicas (CONICET) 1996, Argentina, and the study was approved by the Ethics Committee of the School of Medicine of University of Buenos Aires, Argentina.
