Developmental Programming: Prenatal and Postnatal Contribution of Androgens and Insulin in the Reprogramming of Estradiol Positive Feedback Disruptions in Prenatal Testosterone-Treated Sheep

Abstract

Prenatal testosterone (T) excess compromises the estradiol (E₂) positive feedback. This study tested the hypothesis that antagonizing androgen action or improving insulin sensitivity prenatally would prevent positive feedback disruptions from developing, whereas postnatal intervention with androgen antagonist or insulin sensitizer would ameliorate the severity of disruptions in prenatal T-treated females. The E₂ positive feedback response was tested at 16 wk of age in the following groups of animals: 1) control, 2) prenatal T, 3) prenatal T plus the androgen antagonist, flutamide, 4) prenatal T plus insulin sensitizer, rosiglitazone, 5) prenatal T and postnatal androgen antagonist, and 6) prenatal T and postnatal insulin sensitizer (n = 7–21 animals/group). Prenatal T treatment involved the administration of T propionate (100 mg, im) twice weekly from d 30 to 90 of gestation. Prenatal interventions involved daily sc administration of androgen antagonist (15 mg/kg) or oral administration of insulin sensitizer (8 mg) for the same duration. Postnatal treatments began at 8 wk of age and involved daily oral administration of androgen antagonist (15 mg/kg) or insulin sensitizer (0.11 mg/kg). None of the prenatal/postnatal interventions increased number of animals responding or prevented the time delay in LH surge response to the E₂ positive feedback challenge. In contrast, the postnatal treatment with androgen antagonist or insulin sensitizer increased total LH released in response to E₂ positive feedback challenge, compared with the T animals. Overall, these interventional studies indicate that timing and magnitude of the LH surge are programmed by different neuroendocrine mechanisms with postnatal androgens and insulin determining the size and prenatal estrogen likely the timing of the LH surge.

Prenatal treatment with testosterone (T) leads to reproductive neuroendocrine, ovarian, and metabolic dysfunction in mice, rats, sheep, and monkeys (1, 2). At the neuroendocrine level, these disruptions are manifested at the level of estradiol (E₂) negative (3, 4), E₂ positive (4–6), and progesterone negative (7, 8) feedbacks. Comparative studies involving prenatal T and dihydrotestosterone (DHT) treatment found that the impact on E₂ negative feedback was facilitated by androgenic actions of T (4, 9). This was subsequently confirmed by cotreatment with an androgen antagonist (10), prenatally. E₂ positive feedback disruptions on the other hand appear to be mediated by estrogenic actions of T, because DHT, the nonaromatizable androgen, failed to produce positive feedback disruptions (4, 11). Because DHT can be converted to 3β diol and act through estrogen receptor (ER)β (12), negation of androgenic programming of E₂ positive feedback requires documentation that blockade of androgen action will not overcome positive feedback disruptions.

Because 1) insulin is an essential contributor of brain (13, 14) and pituitary (15) development, 2) the prenatal T treatment period encompasses the time of organization of the GnRH neuronal network (16) and pituitary gonadotrope differentiation (17), and 3) prenatal T treatment decreases fetal pancreatic weight and percentage of α cells (18), E₂ positive feedback defects may also be mediated by metabolic disruptions. As such, improving insulin sensitivity prenatally may help prevent/reduce E₂ positive feedback disruptions.

Once prenatally reprogrammed, manifestation of E₂ positive feedback defects may be maintained postnatally via altered steroid or insulin signaling. An increase in androgen receptor (AR) expression at the level of the hypothalamus (19) and pituitary (Nada, S., and V. Padmanabhan, unpublished data) is evident in prenatal T-treated females. Prenatal T-treated animals are insulin resistant (20, 21). Insulin plays a facilitatory role in the enhancement of neuroendocrine function: it augments GnRH and LH secretion (22–24). Interestingly, diabetic female rats have reduced LH secretion suggestive of insulin resistance at the brain/pituitary level (25). Therefore, negation of androgen action or improving insulin sensitivity postnatally may help ameliorate the degree of E₂ positive feedback disruptions. This study tested the following two hypotheses: 1) cotreatment with an androgen antagonist or an insulin sensitizer prenatally would prevent the mal-programming of E₂ positive feedback disruptions induced by prenatal T treatment and 2) postnatal treatment with androgen antagonist or insulin sensitizer would ameliorate severity of E₂ positive feedback disruptions programmed by prenatal T treatment.

Materials and Methods

Animal breeding and diet

All study procedures were approved by the Institutional Animal Care and Use Committee of the University of Michigan and are consistent with National Research Council's Guide for the Care and Use of Laboratory Animals. The study was carried out at the Sheep Research Facility of the University of Michigan (42°18′ north latitude). Mature Suffolk ewes of proven fertility were purchased from local sources. Starting 2–3 wk before and throughout the breeding period, each ewe received 0.5 kg of shelled corn and 1.0–1.5 kg of alfalfa hay to increase energy balance. During the breeding period, fertility-proven raddled Suffolk rams mated the ewes. Mating was monitored on daily basis and confirmed via paint markings left on the rump of ewes by raddled rams. Pregnant ewes were kept on pasture and received in addition 1.25 kg of alfalfa/brome mix hay/ewe. For the last 6 wk of gestation, pregnant ewes were provided with 0.5 kg of shelled corn and 2 kg of alfala of hay/day per ewe. In addition, pregnant ewes were preventively treated with 250 mg of aureomycin (chlortetracycline) per ewe/day to reduce the prevalence of abortion caused by Campylobacter fetus infection. Parturition of all lambs took place in early 2010 from March through April. Lactating ewes were fed 1 kg of shelled corn and 2–2.5 kg of alfalfa hay/ewe per day. Lambs were provided ad libitum with commercial feed pellets (Shur-Gain, Elma, NY) consisting of 18% of crude protein. Breeder ewes and lambs had ad libitum access to water and minerals, and they were regularly treated with antihelminthics to minimize parasitic infection.

Prenatal and postnatal interventions

Prenatal treatment groups included 1) control (C) (n = 7), 2) insulin sensitizer only (CR) (n = 7), 3) T (n = 7), 4) T plus androgen antagonist (TF) (n = 11), and 5) T plus insulin sensitizer (TR) (n = 7). A prenatal flutamide only group was not included, because in a pilot study, we found no differences in cycle attributes between C and flutamide-treated groups. Postnatal treatment groups included 1) C (n = 7, same as above), 2) C plus androgen antagonist (C+F) (n = 7), 3) C plus insulin sensitizer (C+R) (n = 7), 4) T (n = 7, same as above), 5) T plus androgen antagonist (T+F) (n = 8), and 6) T plus insulin sensitizer (T+R) (n = 8).

Prenatal T treatment involved twice weekly im injections of 100 mg of T propionate (Sigma-Aldrich Corp., St. Louis, MO) suspended in corn oil from d 30 to 90 of gestation (term, ∼147 d). The C group received similar volume of vehicle injections. Prenatal insulin sensitizer treatment involved oral treatment of insulin sensitizer (Rosiglitazone, Avandia; GlaxoSmithKline, Durham, NC) at a dose of 8 mg/ewe·d. This dose of insulin sensitizer administered postpubertally improved reproductive function in prenatal T-treated females (26). Prenatal androgen antagonist treatment involved sc injections of flutamide (15 mg/kg·d) (Sigma-Aldrich Corp.) as described earlier (10). In addition, to determine whether oral treatment would be as efficacious as sc injections, a small subset of animals was administered androgen antagonist orally (TF group, sc n = 9; oral n = 2). The mean birth weight of female offspring used in this study (C, 4.9 ± 0.7; CR, 5.1 ± 0.6; C+F, 4.0 ± 0.6; C+R, 4.7 ± 0.6; T, 4.6 ± 0.6; TF, 5.4 ± 0.5; TR, 4.5 ± 0.6; T+F, 4.4 ± 0.5; and T+R, 4.7 ± 0.5 kg) did not differ among groups.

Postnatal treatments began at 8 wk of age. Postnatal androgen antagonist treatment involved daily oral administration of flutamide pills (15 mg/kg·d). Flutamide dose was adjusted based on weekly body weight measures. Postnatal insulin sensitizer treatment involved oral daily administration of the insulin sensitizer (0.11 mg/kg). The administered dose is comparable with that given to women with polycystic ovary syndrome (PCOS) (27, 28). The same dose administered postpubertally was effective in improving reproductive cycle attributes of prenatal T-treated females (26).

E₂ positive feedback testing

E₂ positive feedback responses were tested at 16 wk of age. All female lambs were implanted sc with four 30-mm SILASTIC E₂ implants (inner diameter, 3.35 mm and outer diameter, 4.65 mm; Dow Corning Corp., Midland, MI) to ensure achievement of follicular phase levels of E₂ (29). Starting 2 h before the insertion of E₂ implants, blood samples were collected from the jugular vein every 2 h for 72 h. Plasma concentrations of LH were measured in duplicate using validated assays (30). The sensitivity of LH assay was 0.48 ± 0.07 ng/ml (n = 11 assays; mean ± sem). The mean intraassay coefficients of variation calculated based on four quality control pools measuring 3.6 ± 0.1, 9.7 ± 0.1, 13.0 ± 0.2, and 23.4 ± 0.3 ng/ml were 7.9, 4.0, 6.4, and 7.1%, respectively. The interassay coefficients of variation for the same four quality control pools were 4.0, 7.4, 6.0, and 11.3%, respectively.

Statistical analysis

For all data series, values below assay sensitivity were replaced with assay sensitivity. Onset and end of the LH surge was defined as the time when LH concentration increased two times the assay sensitivity above the baseline. The end of the surge was defined as the first time point below the onset threshold. The LH concentrations had to remain above the threshold for at least 8 h to be defined as a surge. For all animals that elicited an LH surge response, the onset of surge, surge duration, surge amplitude (difference between peak level and baseline), timing of the surge peak, surge slope (time from onset to peak), and total LH concentration (surge magnitude) during surge were determined. For analyses of prenatal treatment effects, because there were no significant differences in LH surge parameters between the C and the CR groups, both groups were combined (C, n = 14). For analyses of postnatal treatment effects, because there were no differences between postnatal C, C+F, C+R, they were merged into one group referred from now on as C′ (n = 21). Similarly, because administration of flutamide sc or orally reversed the phenotype (all females exhibited female genitalia, a bioassay for blockade of androgen action) and did not differ in their attributes, these groups were combined in the analyses. Fisher's exact test was used to determine differences in percentages of female lambs exhibiting LH surges. A general linear model and post hoc Tukey's significance test after appropriate transformations to account for heterogeneity of variances and corrections for multiple comparisons were applied for all surge variables.

When variances were skewed, Kruskal-Wallis ANOVA test was performed to assess statistical significance among different prenatal and postnatal treatment cohorts. Mann-Whitney U tests of variance were conducted to assess statistical significance among treatment groups, when variances were not skewed. SPSS for Windows release version 18.00 was used to conduct all statistical analysis. All results are presented as mean ± sem. A P value less than 0.05 was considered significant.

Results

Impact of prenatal interventions

Representative profiles of LH responses to E₂ positive feedback challenges from prenatal intervention groups are shown in Fig. 1 and summary results in Figs. 2, 3, and 4. All 14 lambs in C responded to the E₂ positive feedback challenge by eliciting an LH surge (Fig. 2). Only three out of seven T, four out of seven TR, and five out of 11 TF group exhibited LH surge in response to E₂ positive feedback challenge (Fig. 2). Percentages of animals responding to E₂ positive feedback were significantly lower in T (P < 0.05), TF (P < 0.05), and TR (P < 0.05), relative to C, and did not differ among treatment groups.

Fig. 1.

Representative profiles of circulating LH concentrations during the E₂ positive feedback testing of prenatal intervention groups.

Fig. 2.

Percentage of prepubertal female lambs exhibiting LH surge response to E₂ challenge in the prenatal intervention groups. Because there were no differences in C and CR groups, they are collapsed into one group (C). *, P < 0.05 differs significantly from C group.

Fig. 3.

Timing relationships (mean ± sem) of E₂ positive feedback responses in the prenatal intervention groups. Time of LH surge onset (left), peak time of LH surge (middle), and time interval between LH surge start and peak (right). Because there were no differences in C and CR groups, they are collapsed into one group (C). *, P < 0.05 differs significantly from C group.

Fig. 4.

Characteristics of LH surge response to E₂ positive feedback challenge (mean ± sem) in prenatal intervention groups. LH surge duration (left), LH surge amplitude (middle), and total LH release (right). Because there were no differences in C and CR groups, they are collapsed into one group (C). *, P < 0.05 differs significantly from C group.

The timing relationships of the E₂ positive feedback response in the prenatal intervention groups are shown in Fig. 3. The onset of the LH surge response occurred later (P < 0.05) in the T group (35.0 ± 2.7 h) compared with C (14.3 ± 0.5 h) (Fig. 3, left). Prenatal androgen antagonist or insulin sensitizer treatment did not reverse the delayed E₂ positive feedback response in the T group (28.0 ± 3.8 h for TF and 39.0 ± 5.4 h for TR groups, respectively); the responses of TR and TF groups differed from C (P < 0.05). LH surge peak also occurred significantly later (P < 0.05) in T (39.3 ± 2.9 h), TF (37.6 ± 2.8 h), and TR (43.0 ± 4.9 h) groups compared with the C (17.6 ± 0.5 h) (Fig. 3, middle). The time interval between onset and peak of the LH surge was similar for the T (4.0 ± 1.1 h), TR (4.0 ± 1.3 h), and C (3.6 ± 0.2 h) groups, although the TF group took longer (P < 0.05) to peak (11.0 ± 1.1 h) (Fig. 3, right).

The dynamics of LH surge response to the E₂ positive feedback challenge are summarized in Fig. 4. The duration of the LH surge did not differ among C (16.0 ± 1.2 h), T (10.7 ± 1.8 h), and TR (10.0 ± 2.0 h) groups (Fig. 4, left). However, the duration of LH surge response was longer in the TF group (21.0 ± 1.5 h) compared with the C group (P < 0.05) (Fig. 4, left). As expected from earlier studies (4, 6), the amplitude of LH surge was reduced significantly (P < 0.05) in the prenatal T treatment group (T, 11.1 ± 3.7 ng/ml) compared with the C (61.4 ± 4.9 ng/ml). Cotreatment with androgen antagonist or insulin sensitizer did not overcome this compromise (TF, 15.6 ± 4.6 ng/ml; and TR, 4.3 ± 0.4 ng/ml) (Fig. 4, middle). Total LH secretion during the LH surge response was also lower (P < 0.05) in prenatal T-treated group (T, 35.7 ± 6.3 ng/ml) compared with the C (166.3 ± 12.0 ng/ml) (Fig. 4, right). Neither cotreatment completely overcame this defect (TR, 19.1 ± 3.9 ng/ml; TF, 79.6 ± 16.6 ng/ml), although total LH released in the TF group was intermediate between C and T groups.

Impact of postnatal interventions

The representative E₂ positive feedback response profiles and percentage of animals responding to postnatal interventions are summarized in Figs. 5 and 6, respectively. All C′ animals responded with an LH surge to the E₂ positive feedback challenge. Only three out of seven females of the T group, four out of eight of the T+F group, and three out of eight of the T+R group exhibited LH surge response to the E₂ positive feedback challenge. Percentages of animals responding to the E₂ positive feedback challenge were lower in T (P < 0.05), T+F (P < 0.05), and T+R (P < 0.05), relative to C′ (Fig. 6), and did not differ among treatment groups.

Fig. 5.

Representative profiles of circulating LH concentrations during the E₂ positive feedback testing of postnatal intervention groups.

Fig. 6.

Percentage of prepubertal female lambs exhibiting LH surge response to the E₂ challenge in the postnatal intervention groups. Because there were no differences in responses of C, C+F, and C+R groups, they were merged into one group (C'). *, P < 0.05 differs significantly from C′ group.

The timing relationships of the E₂ positive feedback response in the postnatal intervention groups are shown in Fig. 7. The onset of the LH surge was delayed (P < 0.05) in the postnatal treatment groups (T+F, 29.0 ± 2.6 h; T+R, 38 ± 13.0 h) relative to C′ (14.9 ± 0.7 h), as was the case with the T group (Fig 7, left). Peak of LH surge response also occurred later in T+F (38.0 ± 2.6 h) and T+R (44.7 ± 12.3 h) groups, relative to C′ (18.3 ± 0.8 h) (P < 0.05), but similar to the T group (Fig. 7, middle). The time interval from start to peak of the LH surge was longer (P < 0.05) in the T+F (9.0 ± 2.1 h) but not the T+R group (6.7 ± 3.7 h) compared with C′ (3.4 ± 0.3 h) and T groups (Fig. 7, right).

Fig. 7.

Timing relationships (mean ± sem) of E₂ positive feedback responses in the postnatal intervention groups. time of LH surge onset (left), peak time of LH surge (middle), and time interval between LH surge start and peak (right). Because there were no differences in responses of C, C+F, and C+R groups, they were merged into one group (C'). *, P < 0.05 differs significantly from C′ group.

The characteristics of the LH surge response to E₂ positive feedback challenge are summarized in Fig. 8. The duration of the LH surge did not differ among groups: C′ (15.2 ± 0.5 h), T, T+F (17.0 ± 3.4), and T+R (16.0 ± 4.2 h) groups (Fig. 8, left). The amplitude of LH surge was significantly lower (P < 0.05) in the T+F (18.7 ± 2.2 ng/ml) compared with the C′ group (51.8 ± 5.2ng/ml) but similar to the T group (Fig. 8, middle). The LH surge amplitude of T+R group (23.1 ± 5.9 ng/ml) did not differ from C′. Total LH release during the LH surge response in the T+R (88.2 ± 31.7 ng/ml) and T+F (84.02 ± 12.26 ng/ml) groups did not differ from the C′ group (140.7 ± 9.9 ng/ml) (Fig. 8, right) but differed significantly (P < 0.05) from the T group.

Fig. 8.

Characteristics of LH surge response to E₂ positive feedback challenge (mean ± sem) in postnatal intervention groups. LH surge duration (left), LH surge amplitude (middle), and total LH release (right). Because there were no differences in responses of C, C+F, and C+R groups, they were merged into one group (C'). *, P < 0.05 differs significantly from C′ group.

Discussion

The findings from this study indicate that the delayed and reduced LH surge magnitude of prenatal T-treated females is programmed by different mechanisms, with postnatal androgens and insulin playing a role in determining the size of the LH surge and prenatal estrogen likely the timing of the surge. These findings are discussed in the context of steroidal programming and postnatal amplifications mechanisms.

Prenatal interventions

Failure of cotreatment with androgen antagonist to overcome the time delay in LH surge onset suggests that this aspect of the surge is not organized by androgenic actions but more likely by estrogenic actions of T. Increase in estrogen in fetal circulation of T fetuses (31) after gestational T treatment does provide a means for such estrogenic mediation. Absence of such disruptions in prenatal DHT-treated animals (4, 10, 11) also negates androgenic mediation. It needs to be recognized that DHT has the potential to be converted to 3β diol and act through ERβ receptors (12, 32). ERβ is expressed in the ovine hypothalamus during prenatal life (33). Studies with ERβ-knock out (βERKO) females indicate that lower magnitude LH surge seen in ovary-intact ERβ-knock out mice is mediated via ovarian ERβ, because the magnitude of E₂-induced LH surge did not differ when an ovariectomized model was used (34, 35). Studies of androgen antagonism in other species have focused on E₂ negative [monkeys (36) and spotted hyena (37)], but not E₂ positive, feedback mechanisms.

Although the small number of animals studied precludes definitive conclusions from being drawn, the finding that cotreatment with an androgen antagonist leads to an intermediate LH response between C and prenatal T females is provocative and raises the possibility of androgenic programming of LH surge magnitude. On the other hand, potential for estrogenic programming of surge magnitude is suggested by the absence of an effect of prenatal DHT in reducing LH surge magnitude (4, 11). Interestingly, a decrease in LH surge magnitude was evidenced in sheep treated prenatally with bisphenol-A (38), an endocrine disruptor with estrogenic as well as antiandrogenic properties (39). As such, organizational effects on LH surge magnitude may involve both androgenic and estrogenic mediation.

Postnatal interventions

The failure of postnatal androgen antagonist as well as insulin sensitizer treatment to affect timing of surge is not surprising, because this likely involves organizational changes occurring during prenatal/early postnatal life at the neural level. The increase in total LH released in response to the E₂ positive feedback challenge after postnatal androgen antagonist treatment is supportive of increased androgen signaling underlying reduced LH surge in prenatal T-treated females. Prenatal T treatment indeed increases AR protein expression in the hypothalamus (19) and mRNA expression in pituitary (Nada, S., and V. Padmanabhan, unpublished data) of adult females. One possibility is that postnatal blockade of AR action may have normalized LH pulsatile secretion to levels seen in C′ group thereby increasing availability of the releasable LH pool, although this premise remains to be experimentally proven. In women with PCOS, the reproductive phenotype of whom prenatal T-treated females recapitulates (3, 4, 40), androgen antagonist treatment does normalize LH pulse frequency (41). The possibility that prenatal cotreatment with androgen antagonist may increase total LH released in response to the E₂ positive feedback suggests that the effects of prenatal and postnatal androgen antagonist treatment may be transduced through a common mediary. If so, prenatal intervention may do so by reducing AR expression to C levels and postnatal treatment by blocking the effect of increased AR expression (19). It is tempting to speculate that setting the level of AR expression is part of the reorganization cascade programmed by prenatal T excess. Postnatal treatment effects are likely to be activational and disappear, once the treatment is stopped.

The fact that postnatal insulin sensitizer treatment also increases the magnitude (total LH released in response to positive feedback challenge) as well as amplitude (difference between peak and nadir) of the surge suggests an activational role for insulin in modulating LH surge magnitude. Earlier findings that insulin infusion dampens the LH surge magnitude (42) in sheep are consistent with the reduced magnitude of the LH surge evidenced in prenatal T-treated sheep, which are hyperinsulinemic (20, 21). It is unclear whether high insulin levels render the brain insulin resistant by down-regulating insulin receptor expression in hypothalamic centers controlling E₂ positive feedback responses. Insulin receptors are expressed in the ovine hypothalamus (43), and prenatal T treatment causes insulin resistance and abolishes the GnRH surge (44). If brain insulin resistance contributes to LH surge defects, insulin sensitizer treatment should help restore GnRH/LH surge magnitude. Evidence in rodents supports a facilitatory role for insulin in stimulating GnRH synthesis in the hypothalamus (22–24).

Partial restoration of the LH surge magnitude by insulin sensitizer treatment may also be achieved by insulin having a direct effect at the pituitary and increasing LH synthesis or reducing pulsatile LH release. This would lead to an increase in the storage pool of LH available for release. Evidence exists in support of insulin synergizing with GnRH in increasing LHβ transcription (45). However, an increase in LHβ mRNA expression was not evident in prenatal T-treated females (46), although an effect at the level of protein translation cannot be ruled out.

Whether insulin sensitizer restores LH pulsatility back to C levels in prenatal T-treated females, thus reducing depletion of the releasable pool, has not been tested experimentally in this model. However, our recent findings (47) that overfeeding renders sheep to be insulin resistant/hyperinsulinemic while increasing LH pulsatility are consistent with this premise. The majority of studies in women with PCOS, the reproductive phenotype of whom prenatal T-treated sheep recapitulates (3, 4, 40, 48), found that insulin sensitizer treatment decreases circulating LH (49–52) and LH pulse amplitude (51, 52). This is likely to increase the storage pool of LH available for release in response to an E₂ positive feedback challenge. Whether insulin sensitizer alters pituitary sensitivity to GnRH is unclear. Although 20-wk treatment with pioglitazone, an insulin sensitizer, failed to alter GnRH-induced LH secretion in women with PCOS (53), treatment with metformin, another insulin sensitizer, decreased GnRH neuronal firing in prenatal DHT-treated mice (54), which do not manifest peripheral insulin resistance. Relative to insulin having a role in increasing GnRH sensitivity, an effect of prenatal T on GnRH receptor mRNA level was not evident in prenatal T-treated sheep (46), in spite of these animals being hyperinsulinemic (20, 21). However, this does not rule out the possibility of an effect at the level of protein translation.

The finding that neither postnatal treatment restored the LH surge magnitude/amplitude to levels seen in C′ group raises the possibility that the amount of androgen antagonist and insulin sensitizer delivered may be insufficient. This appears unlikely, because the dose of flutamide used was able to completely overcome phenotypic virilization as well as E₂ negative feedback disruptions seen in prenatal T-treated females (10). The same holds true for the insulin sensitizer treatment as well, because the dose used normalized insulin sensitivity in prenatal T-treated females (20) as well as gestational T-associated increase in insulin in pregnant sheep (Abi Salloum, B., and V. Padmanabhan, unpublished data). It is conceivable that the effect of these two treatments in increasing LH surge magnitude involves different mechanisms. Rosiglitazone, the insulin sensitizer used, acts through peroxisome proliferator-activated receptors-γ (55), and flutamide, the androgen antagonist used, blocks AR action (56). Combined treatment involving both may restore full magnitude/amplitude LH surge. A difference in the slope of LH increase, the timing of the surge, and the amount of LH secreted has been evidenced in prenatal E₂-treated sheep (E₂ treatment from before conception throughout pregnancy) when tested prepubertally but not postpubertally (57). Studies involving cotreatment with estrogen antagonist would be required to delineate estrogenic contribution to the LH surge magnitude.

Translational relevance

The preovulatory LH surge is an obligatory signal for normal ovulation (58) and regulates oocyte maturation (59). As such, the delayed LH surge, although capable of luteinizing and sustaining corpus luteum function in prenatal T-treated females (albeit compromised), would likely compromise quality of fertilizable oocyte. A delayed LH surge has been found to reduce fertilization and implantation rates and increases embryonic death and chromosome abnormalities (60–62). In this regard, it is interesting that although interventions with androgen antagonist (63), insulin sensitizer (64–68), or combined (69) are able to improve ovulation rate in subjects with polycystic ovarian disease, optimal success with pregnancy rate are yet to be realized. If time delay in LH surge dynamics is a feature of women with PCOS, this would explain the disparity in ovulatory and pregnancy success rates. Unfortunately, studies testing the E₂ positive feedback response have not been carried out in women with PCOS to address whether the LH surge is indeed inadequate. Future studies involving combined androgen antagonist and insulin sensitizer treatment are required to determine whether full amplitude LH surge and normal corpus luteum function can be achieved in prenatal T-treated females.

From the perspective of LH surge magnitude, our earlier findings of short/inadequate luteal progestogenic response (8, 70, 71), and presence of larger than ovulatory-sized follicles that fail to ovulate (69), indicate that the amplitude of LH surge in prenatal T-treated sheep may be insufficient for ovulation and launching an adequate progestogenic response to support pregnancy, if fertilization occurs. The findings that postnatal androgen antagonist and insulin sensitizer treatment increases the total LH surge release are therefore of translational value and have implications for gonadotropin-stimulation protocols, relative to the amount of human chorionic gonadotropin required to induce normal corpus luteum function.

In summary, the findings from this study provide evidence in support of different neuroendocrine mechanisms regulating timing and amplitude of the LH surge response to E₂ positive feedback, with postnatal androgens and insulin playing a role in determining the size of the LH surge and prenatal estrogen likely the timing of the LH surge. These studies are the first to provide a neuroendocrine basis for the beneficial effects of agents that are already used in the treatment of women with PCOS.