Fetal Programming: Excess Prenatal Testosterone Reduces Postnatal Luteinizing Hormone, But Not Follicle-Stimulating Hormone Responsiveness,
http://www.100md.com
《内分泌学杂志》
Departments of Pediatrics (H.N.S., C.H., M.M., J.D., V.P.), Obstetrics and Gynecology (D.L.F., V.P.), Molecular and Integrative Physiology (V.P.), and Ecology and Evolutionary Biology (D.L.F.), Reproductive Sciences Program (H.S., M.M., C.H., J.D., D.L.F., V.P.), and Center for Statistical Consultation and Research (K.B.W.), University of Michigan, Ann Arbor, Michigan 48109
Abstract
Exposure of female sheep fetuses to excess testosterone (T) during early to midgestation produces postnatal hypergonadotropism manifest as a selective increase in LH. This hypergonadotropism may result from reduced sensitivity to estradiol (E2) negative feedback and/or increased pituitary sensitivity to GnRH. We tested the hypothesis that excess T before birth reduces responsiveness of LH and FSH to E2 negative feedback after birth. Pregnant ewes were treated with T propionate (100 mg/kg in cotton seed oil) or vehicle twice weekly from d 30–90 gestation. Responsiveness to E2 negative feedback was assessed at 12 and 24 wk of age in the ovary-intact female offspring. Our experimental strategy was first to arrest follicular growth and reduce endogenous E2 by administering the GnRH antagonist (GnRH-A), Nal-Glu (50 μg/kg sc every 12 h for 72 h), and then provide a fixed amount of exogenous E2 via an implant. Blood samples were obtained every 20 min at 12 wk and every 10 min at 24 wk before treatment, during and after GnRH-A treatment both before and after E2 implant. GnRH-A ablated LH pulsatility, reduced FSH by approximately 25%, and E2 production diminished to near detection limit of assay at both ages in both groups. Prenatal T treatment produced a precocious and selective reduction in responsiveness of LH but not FSH to E2 negative feedback, which was manifest mainly at the level of LH/GnRH pulse frequency. Collectively, these findings support the hypothesis that prenatal exposure to excess T decreases postnatal responsiveness to E2 inhibitory feedback of LH/GnRH secretion to contribute to the development of hypergonadotropism.
Introduction
STRONG EPIDEMIOLOGICAL evidence reveals that the environment to which the fetus is exposed during gestation influences prenatal development and subsequent expression of physiology and behavior during adulthood (1, 2, 3, 4, 5). In mammals, such programming of the reproductive system by hormones is striking. This can occur naturally as exemplified by the finding that female rats are masculinized in utero by hormones emanating from male litter mates sharing the same uterine horn (6, 7). Such females are less attractive to males, more aggressive, and reproductively compromised. Similarly, exposure to excess testosterone (T) in utero experimentally can lead to sterility in rodents, monkeys, and sheep (7, 8, 9, 10). Our studies, and those of others, using the sheep as a model reveal that prenatal exposure of the female to excess T during the middle third of gestation produces a suite of adult disorders (9, 10, 11, 12, 13, 14, 15, 16, 17, 18), culminating in early reproductive failure (9, 10). As such, manipulation of the prenatal steroid environment provides a powerful experimental tool for understanding the neuroendocrine and ovarian mechanisms that underlie prenatal programming of the reproductive axis.
At the neuroendocrine level, it is well established that prenatal T treatment reduces hypothalamic sensitivity to estradiol (E2) positive feedback (11, 12, 14, 16) and progesterone negative feedback (17), the two major feedback systems involved in the control of cyclic changes in GnRH/gonadotropin secretion. From a developmental perspective, E2 negative feedback is the predominant feedback system operational before puberty (19, 20, 21), and the reduction in sensitivity to this steroid feedback inhibition is responsible for the pubertal increase in GnRH (19, 20, 21). Studies in the female sheep have demonstrated this mechanism using the ovariectomized E2 replaced model (20), in which peripheral steroid concentrations are maintained at physiological levels by means of a constant release device (E2 capsule sc) from shortly after birth. In this model, circulating LH levels remain suppressed in response to chronically low exogenous E2 until about 30 wk of age when they rise markedly. The timing of this robust LH increase, reflecting the escape from E2 negative feedback, occurs at the same time as the initiation of ovulations and estrous behavior (puberty) in normal intact lambs (19, 21). Prenatal exposure to T markedly advances the time of the pubertal LH rise to 10 wk, an age that corresponds to the initiation of puberty in the male (11, 12, 22). This prenatal programming by T would explain the much earlier decrease in sensitivity to negative feedback in the male. Interestingly, in the prenatal T female, when the ovaries are left in situ instead of replacing them with an E2 capsule as a source of steroids, the time of puberty is not advanced (10, 14), as would be predicted from the finding of a precocial LH rise in the prenatal T, postnatally ovariectomized E2 model. Rather, in the prenatal T female left ovary intact (no exogenous E2), initiation of progestogenic cycles occurred at the same time as in the controls. This raises two possibilities: 1) the constant release E2 in the ovariectomized E2-clamped model may exert further organizational effects postnatally to combine with those of the prenatal T to result in the precocious reduction in sensitivity to E2 negative feedback; 2) alternatively, ovarian factors may play a protective role and not allow reduction in sensitivity to E2 negative feedback (early postnatal LH rise) to be expressed. There is some evidence that the ovary-intact prenatal T female is hypergonadotropic during the prepubertal period (14), raising the possibility that such females do express a reduced sensitivity to E2 negative feedback. However, an increased LH secretion alone does not provide evidence for reduced E2 feedback (hypothalamic effect) because this may reflect a reduced amount of ovarian steroid secretion (ovarian effect) or alternatively increased pituitary gonadotropin responsiveness to GnRH (pituitary effect) in such females. In the present study, the primary hypothesis we tested is that prenatal exposure to T produces a precocial reduction in sensitivity to E2 negative feedback after birth. Our results indicate that this indeed is the case and further that development of hypergonadotropism occurs progressively over time. In addition, because E2, along with inhibin (23, 24, 25), is a major negative feedback regulator of FSH, the second hypothesis we tested is that the reduced sensitivity to E2 feedback will also be reflected as increased FSH secretion in the prenatal T females. Our results indicate that this is not the case and leads to the conclusion that prenatal T treatment has differential effects on feedback responsiveness of LH and FSH to E2.
Materials and Methods
Generation of prenatal T females
All procedures were approved by the Institutional Animal Care and Use Committee of University of Michigan and were consistent with the National Institutes of Health Guide for Use and Care of Animals. Suffolk sheep used in this study were prenatally treated with T or cottonseed oil (control) during 30–90 d gestation (n = 7 per treatment group). The details of the prenatal treatments, husbandry and nutrition of maternal ewes, newborn measures, and growth characteristics until 4 months of age of the larger cohort of females of which these are part of have been published (26). Briefly, pregnant ewes were administered im 100 mg T propionate (Sigma Chemical Co., St. Louis, MO) suspended in 2.0 ml cottonseed oil twice weekly from d 30–90 gestation (term, 147 d). Control ewes received the same volume of cottonseed oil in a similar regimen as T. Lambs were born between March 15 and April 20. They were weaned at 8 wk of age and maintained outdoors at the Sheep Research Facility (Ann Arbor, MI; 42°, 18'N) under natural photoperiod and were provided ad libitum access to commercial feed pellets (Shur-Gain, Elma, NY) consisting of 3.6 MCal/kg digestible energy and 18% crude protein. When they reached approximately 40 kg body weight, they were switched to a pellet feed with 15% crude protein to avoid excessive fat deposition as the rate of growth naturally diminished and adult size was approached. Trace mineralized salt with selenium and vitamins A, D, and E (Armada Grain Co., Armada, MI) were freely accessible throughout the study. The postnatal growth of all female lambs was monitored by weighing every 2 wk before feeding. Samples were collected twice weekly from 20 wk of age for measurement of progesterone to assess the onset of puberty.
Experimental design
To determine developmental changes in gonadotropin responsiveness to E2 negative feedback, feedback tests were conducted at approximately 12 wk of age (early prepubertal) and at approximately 24 wk (shortly before the expected time of puberty). Typically, E2 negative feedback tests are conducted in the absence of the ovaries to avoid confounding from endogenous E2, which could differ between individuals depending on the stage of follicular development. Thus, in our study of ovary-intact females, we needed to produce a reduced but similar starting follicular E2 milieu in all females. To do this, we reduced and normalized endogenous E2 production by blocking LH- and GnRH-induced FSH release with the GnRH antagonist (GnRH-A), Nal-Glu (Nal-Glu [Ac-D2Nal1, D4ClPhe2, D3Pal3, Arg5, 4-(methoxybenzoyl)-D-2-aminobutyric acid6, D-Ala10]), as described previously (27).
In experimental phase 3, the response to E2 feedback inhibition was tested, and a 3-mm SILASTIC brand (Dow Corning Corp., Midland, MI) E2 capsule filled with crystalline E2-17 was inserted sc in the axillary region of both control and prenatal T lambs. This size implant produces circulating concentrations of E2 of less than 1 pg/ml. Blood samples were collected frequently at 12 wk (20-min intervals) and 24 wk (10-min intervals) for 6 h, from 66–72 h, after implant insertion, which was 138–144 h after cessation of the GnRH-A.
All blood samples were collected in heparinized tubes, and plasma was separated at 5 C. Plasma was stored at –20 C until assay. Plasma LH concentrations were determined in all samples. Plasma E2 and FSH levels were determined in pooled samples obtained by combining equal aliquots of plasma from all samples of a given female during each phase (n = 19 for 12 wk where the sampling frequency was every 20 min and 37 for 24 wk where the sampling frequency was 10 min).
RIA
Plasma LH concentrations were determined using a validated competitive double-antibody RIA (28) in duplicate 10- to 200-μl samples. The assay sensitivity (2 SD from the buffer control) of LH assay was 0.34 ± 0.07 ng/ml (n = 19 assays). Mean intraassay coefficient of variations at 80 and 20% displacement points were 6.35 ± 0.27 and 3.15 ± 0.13%, respectively, and the mean median variance ratio was 0.04 ± 0.003. The interassay coefficient of variations in three quality control pools averaging 1.0, 13, and 23 ng/ml were 19.5, 3.9, and 3.5%, respectively. FSH was measured in duplicate 200-μl samples (n = 2 assays) with a validated RIA (29) using reagents from the National Hormone and Pituitary Program. Assay sensitivity averaged 0.05 and 0.03 ng/ml, respectively, and average intraassay coefficients of variation at 80 and 20% displacement points were 7.97 and 3.98%. Interassay coefficient of variation for two plasma quality control pools averaging 5.4 and 30 ng/ml was 0.37 and 1.1%, respectively. Plasma concentrations of E2 were measured using a commercial RIA kit (Coat-A-Count E2, MAIA, Polymedco Inc., Cortland Manor, NY) validated for use in sheep (30). Assay sensitivity averaged 0.26 ± 0.06 pg/ml (n = 3 assays), and mean median variance ratio averaged 0.05 ± 0.03. Mean intraassay coefficient of variation at 80 and 20% displacement points averaged 14.9 ± 3.9 and 7.5 ± 2.0%, respectively. Interassay coefficient of variations based on two quality control pools averaging 5 and 40 pg/ml averaged 19.4 and 9.3%, respectively. Each assay included both control and prenatal T females. Plasma concentrations of progesterone were measured by a commercial RIA kit (Coat-A-Count P4; Diagnostic Products Corp., Los Angeles, CA) in daily samples. This assay has been validated for use in sheep (31). Sensitivity of progesterone assay averaged 0.03 ± 0.01 ng/ml (n = 8 assays). The intraassay coefficient of variation, based on two quality control pools averaging 1.99 and 13.7 ng/ml averaged 8.7, and 10.4%, respectively. The interassay coefficients of variation for the same quality control pools were 10.5 and 10.0%, respectively.
Statistical analysis
The first sustained rise in circulating progesterone was used to determine puberty, and this was based on two criteria: the time when circulating progesterone concentrations first exceeded baseline by 2 times assay sensitivity, and this must be followed by a progestogenic cycle of greater than 0.5 ng. Differences in timing of the onset of puberty were determined by ANOVA. Growths of lambs were compared using a linear mixed model in which a separate regression line (random intercept and random slope) was calculated for lamb weights of each mother. Overall effect of age, treatment, and the age by treatment interactions were examined.
Serial LH data from control and prenatal T females from each time period were subjected to pulse analysis using the Cluster algorithm (32). The Cluster algorithm identifies pulses using criteria that define a pulse such that the peak of the pulse differs significantly from both the preceding and following nadirs according to two-sample Student’s t tests. For analysis with Cluster, the minimum number of data points in a peak and nadir were set at 1 and 1, respectively, when blood samples were obtained at 20-min intervals and 2 and 2, respectively, when blood samples were obtained at 10-min intervals. The Student’s t statistic values used to identify a significant increase from preceding nadir and a decrease to following nadir were both 1.0 and 2.0 for the 10- and 20-min sampling frequencies, respectively.
Hormone data for the control and prenatal T-treated females were compared at 12 wk of age for the treatment periods pre-; 72 h after GnRH-A; and E2; and at 24 wk of age for the periods pre; 24, 48, 72 h after GnRH-A; and E2 (Fig. 1). The variables compared were E2 (concentration), LH (concentration, pulse frequency, and pulse amplitude), FSH (concentration), and LH to FSH ratio. A repeated-measures ANOVA was conducted for comparing concentrations of LH and FSH, E2, and the LH pulse amplitude. The repeated-measures ANOVA had one between-subjects factor (control vs. prenatal T) and one within-subjects factor (period), with period having three levels for the 12-wk analysis and five levels for the 24-wk analysis. The main effects of treatment and period and the interactions between treatment and period were examined. Post hoc tests were used to compare treatment means within each period, and the change between the post-GnRH-A at 72 h and E2 periods for C- vs. T-treated females. Residuals from these analyses were checked for normality. A nonparametric Wilcoxon rank sum test with exact P values was used to compare LH to FSH ratios for C- vs. T-treated females within each treatment period because normality could not be assumed for these measures. For LH pulse frequency, a count variable, which could not be assumed to be normally distributed, a Poisson regression with repeated measures was carried out, using the generalized estimating equations method (33). This analysis allowed us to take into account the correlation among observations on the same female for the different bleeds and also allowed us to take the nonnormality of the response into account. The percent suppression from the GnRH-A 72 h to E2 periods for each of the hormones was compared for the control vs. prenatal T-treated females at 12 and 24 wk. A nonparametric Wilcoxon rank sum test was used for the analysis of the percent suppression data because we could not assume normality for these measures.
To allow comparisons of the age-dependent changes in E2 feedback of gonadotropins at 12 and 24 wk, alternate data points from the 10-min LH data series at 24 wk of age were deleted so that they would be comparable to the 20-min data series at 12 wk, and the normalized data series was then subjected to a second set of Cluster analyses. To examine the effect of age on percent suppression, the difference between the normalized 12- and 24-wk suppression was calculated for each animal and a Wilcoxon signed rank test was used to assess the significance of the change in percent suppression at the two ages. We also examined the overall effect of treatment on the normalized 12- vs. 24-wk LH suppression data by first calculating the average of the normalized percent suppression from 12–24 wk for each animal and then comparing this mean for C- vs. T-treated animals using a Wilcoxon rank sum test. All analyses were carried out using SAS for Windows release 9.1.3 (2002–2003; SAS Institute, Cary, NC).
Results
Growth rate and onset of puberty
Birth weights and growth rate during the first 16 wk of postnatal development for the larger group of animals of which these were part of have been published (26). These showed that prenatal T females have lower birth weight and exhibit catch up growth between 8 and 12 wk of age. Analyses of biweekly weights beginning at 16 wk of age revealed no significant treatment effect indicating that growth rates for young females within both treatments was similar after 16 wk (not shown). The estimated slope for the regression of weight on age was 0.15 ± 0.01 for controls and 0.14 ± 0.01 for the prenatal T females. When E2 negative feedback tests were conducted, control and prenatal T lambs weighed 30.3 ± 5.6 and 31.6 ± 2.6 kg, respectively, at 12 wk and 57.3 ± 7.3 and 63.3 ± 3.4 kg, respectively at 24 wk.
The age of puberty for females used in this study was delayed in prenatal T females compared with controls (Fig. 2). The first E2 negative feedback test at 12 wk of age was 15 and 19 wk before puberty in the control and prenatal T females, respectively, during the age of relatively high sensitivity to estrogen negative feedback. The second E2 feedback test at 24 wk of age was about 3 and 8 wk before puberty, respectively, when sensitivity to E2 negative feedback normally reduces in control females.
Circulating E2 concentrations in 12- and 24-wk-old females
Effects of prenatal T treatment on LH responsiveness to E2 negative feedback at 12 wk of age
Representative patterns of LH pulses and circulating E2 levels (shaded area) from three control and three prenatal T females are shown in Fig. 4; mean changes in circulating LH, LH pulse frequency, amplitude, and degree of E2 feedback suppression of LH are presented in Fig. 5. Before treatment, control females exhibited zero to one pulses per 6 h, and the prenatal T females had one to five pulses. Administration of the GnRH-A abolished LH pulsatility completely regardless of prenatal treatment. Cessation of the GnRH-A treatment resulted in restoration of pulses with both control and prenatal T females having high-frequency LH pulses between 66 and 72 h after the GnRH-A. E2 treatment reduced LH pulse frequency to zero to one pulse in control females and to one to three pulses in prenatal T females. Group statistics (Fig. 5) revealed that mean LH levels were not different between control and prenatal T females before treatment (Fig. 5; Pre), but circulating LH was higher (P < 0.01) in prenatal T females before (72 h after GnRH-A, Fig. 5) and during E2 treatment (Fig. 5). LH pulse frequency was higher (P < 0.05) in prenatal T animals compared with controls during the pretreatment period and before E2 treatment (72 h after GnRH-A). This was also the case during E2 treatment (P < 0.01) (Fig. 5) despite the higher levels of E2 in prenatal T females (Fig. 2). Comparison of pre-E2 (72 h after GnRH-A) and E2 treatment periods revealed reduced (P < 0.05) E2 suppression of LH pulse frequency but not mean LH concentrations. The greater suppression of pulse frequency in control females after E2 treatment was not reflected by a statistically significant increase (P = 0.07) in LH pulse amplitude, possibly because of absence of pulses from which to calculate amplitude from in four of the seven control females.
Twenty-four weeks of age
Circulating LH pulse patterns and circulating E2 levels (shaded area) from the same three control and three prenatal T females are shown in Fig. 6 as were presented in Fig. 5; mean LH concentrations, pulse frequencies, and amplitudes are in Fig. 7. Before treatment with the GnRH-A, despite the higher E2 levels at 24 wk of age in prenatal T females, circulating LH levels and LH pulse frequency were higher (P < 0.01) in prenatal T females compared with controls. In both groups of 24-wk-old females, as at their younger age, the GnRH-A reduced circulating LH levels and completely abolished LH pulsatility (P < 0.01). Cessation of the GnRH-A treatment resulted in a gradual restoration of circulating LH and return of LH pulses. During the post-GnRH-A period, LH pulse frequency was higher in the prenatal T females during all three-time points (24, 48 and 72 h) (P < 0.05). These higher pulse frequencies occurred in the presence of similar circulating levels of E2 (Fig. 2, bottom panel). The increase in LH pulse frequency was reflected as an increase in circulating LH levels at 24 (P < 0.01), 48 (P < 0.05), and 72 h (P < 0.05) after GnRH treatment (top panel). E2 treatment reduced the number of LH pulses in the control and prenatal T females. LH concentration (P < 0.01) and LH pulse frequency (P < 0.05) were both higher in the prenatal T females compared with controls during E2 treatment. LH pulse amplitude of prenatal T females also tended to be higher compared with controls during the E2 treatment period (P = 0.06). Comparison of pre-E2 (72 h after GnRH-A) and E2 treatment periods revealed a tendency for reduced E2 suppression of LH pulse frequency (P = 0.09) but not mean LH concentrations or amplitude in prenatal T females compared with controls (right panels).
Age-associated changes in E2 negative feedback
Comparison of results between 12 and 24 wk, after adjusting the data series for differences in sampling frequency, revealed a significant treatment and age effect with LH pulse frequency but not mean LH concentrations (Fig. 8). This was reflected as reduced E2 suppression of LH frequency at 24 than 12 wk across treatment (P < 0.05) and reduced E2 suppression of LH frequency in prenatal T females compared with controls across ages (P < 0.05). There was no age by treatment interaction.
Effect of prenatal T treatment on circulating FSH at 12 and 24 wk
Plasma FSH concentrations of control and prenatal T females at 12 and 24 wk of age are depicted in Fig. 9. In contrast to near complete ablation of LH, the GnRH-A suppressed (P < 0.01) circulating FSH concentrations minimally (<25%) at both ages, and there were no treatment or age effects. Circulating levels of FSH were restored to pretreatment levels by 48 h after GnRH-A in the 24-wk-old females and by 72 h after GnRH-A in the 12-wk-old females (48 h was not studied at 12 wk). E2 treatment decreased FSH secretion in both control and prenatal T females at both ages (12 wk, P < 0.01; 24 wk, P = 0.06). Changes in the LH to FSH ratio, in general, reflected changes in LH (Table 1). During the pretreatment period, the LH to FSH ratio was higher in prenatal T females compared with controls reaching significance at 24 wk (P < 0.05). The GnRH-A reduced the LH to FSH ratio at both age groups in both treatment groups (range of suppression, 68–83%). Cessation of the GnRH-A treatment restored the LH to FSH ratio to pretreatment levels by 48 h in 24-wk-old females and by 72 h in 12-wk-old females (other times were not studied at 12 wk). LH to FSH ratio during the E2 treatment period was significantly higher (P < 0.05) in prenatal T females than controls at both ages.
Discussion
The findings from this study support the hypothesis that prenatal treatment of the developing female sheep with T during the middle third of gestation reduces postnatal LH responsiveness to E2 negative feedback, thus contributing at least in part to the development of hypergonadotropism. Based on LH pulse frequency, the reduced responsiveness to E2 negative feedback was evident as early as 12 wk of age (younger ages not studied) when control females are yet highly sensitive to E2 negative feedback (19, 20, 21). At 24 wk, the E2 feedback suppression of LH pulse frequency in the prenatal T females (20%) was lower compared with the 12-wk-old females (60%). Age-dependent reduction in responsiveness to E2 feedback was also evident in controls (90 vs. 40% at 12 vs. 24 wk, respectively), although the sensitivity to E2 feedback regulation of LH frequency across both ages was greater in controls than prenatal T females (Fig. 8). This, in concert with the fact that prenatal T females lagged a month behind control females in their initiation of reproductive cycles, is supportive of a general reduction in sensitivity of prenatal females to E2 negative feedback. Interestingly, the reduced responsiveness to E2 negative feedback of prenatal T females at 12 wk was manifest at the level of LH pulse frequency but not FSH. The validity of the GnRH-A approach that made it possible to study responsiveness to E2 feedback in the ovary-intact female and the potential implications of the findings from this study as they relate to the timing of puberty, GnRH feedback control mechanisms, development of adult hypergonadotropism and infertility, and differential regulatory control of LH and FSH are discussed below.
Validity of approach used to assess negative feedback
To assess E2 negative feedback in the presence of the ovaries, we attempted to standardize circulating E2 by reducing endogenous production and replacing this with exogenous E2 from a constant release device, the well-characterized SILASTIC brand capsule implanted sc. A GnRH-A was used to block GnRH action, thereby reducing gonadotropic drive, which, in turn, arrested follicular development and reduced endogenous E2 production. This provided a short window of time in which to conduct the E2 feedback test. Although the approach proved to be technically useful to study the responsiveness of the prenatal T females to E2 negative feedback, there are some limitations in applying this approach to other ovary-intact situations. The GnRH-A does not completely block FSH production or endogenous estrogen production. It ablates GnRH action (27), blocks LH release (27, 32), blocks production of GnRH-induced release of FSH (less acidic and more bioactive) (27), induces follicular atresia, and arrests follicular development to the gonadotropin-independent stage (34), thus limiting E2 production to near detection levels (27, 34) of the assay. Although release from GnRH-A inhibition allows a new wave of follicular growth, it does not follow that the number of follicles recruited will be the same and result in similar endogenous levels of E2. Because prenatal T females are multifollicular, we expected levels of E2 production and, hence, the feedback drive, to be equal or greater in prenatal T females compared with control females. This proved to be the case. The converse, namely increased E2 drive in the control females, would have not allowed us to test the hypothesis.
Altered sensitivity to E2 negative feedback and pubertal timing
Central to the neuroendocrine mechanisms associated with puberty in the sheep, as in other species, is the reduction in sensitivity to E2 negative feedback inhibition that allows the GnRH pulse generator to express high activity and produce the gonadotropin secretion at levels that stimulate the gonads to function like an adult (19, 20, 21). Sex differences often occur in the timing of puberty. In the female sheep, the pubertal increase in GnRH secretion occurs around 25–30 wk of age, whereas in males, this begins much earlier, at 8–10 wk of age (19). This earlier pubertal increase in GnRH secretion in the male occurs in response to the earlier decrease in sensitivity to estrogen feedback inhibition brought about by the prenatal programming of this system by exposure to steroids from the developing testes. Experimentally, the timing of the pubertal GnRH rise in ovariectomized females exposed to constant E2 can be advanced by treatment with T in utero (11, 12, 22). Such females, like males, reduce their responsiveness to E2 negative feedback, leading to an early increase in circulating LH (11, 12, 22). Thus, one would predict from the precocious initiation of neuroendocrine puberty in this model that precocious initiation of ovulatory cycles would occur had the ovaries not been removed in such prenatal T females. This is not the case as determined by our earlier studies in which progestogenic cycles occurred at the normal time (10, 14, 16, 18) or this study, where progestogenic cycles began later in the prenatal T animals. Although onset of progestogenic cycles did not occur early, circulating levels of LH, based on infrequent sampling (twice weekly), were found to be higher postnatally in prenatal T females compared with control females (14). LH pulse frequency was also found to be higher during the follicular phase of sheep treated prenatally with T from d 60–90 gestation (18). In the present study, the increased LH pulse frequency and mean LH concentrations during the pretreatment period both at 12 and 24 wk of age in the prenatal T females provides additional evidence for hypergonadotropism. Because this occurred in these prenatal T females in the face of similar or higher endogenous E2 concentrations, it provides evidence that an early reduction in LH sensitivity to E2 feedback is expressed before puberty in prenatal T females, even in the presence of the ovary. Further evidence for this decreased sensitivity was provided by their response to estrogen, when the steroid was provided by a constant release device (implant); prenatal T females were hypergonadotropic relative to control females. Finally, further corroboration for this contention is provided in the present study by the early increase in LH pulse frequency in prenatal T females during the post-GnRH-A period compared with control females in the presence of similar or lower concentrations of E2. Alternatively, the elevated LH secretion found 24 h after GnRH-A in 24-wk-old prenatal T females, when increase in endogenous E2 is negligible, may likely represent steroid-independent stimulatory effect on LH by prenatal T excess, analogous to that seen between gonadectomized male and female lambs (35) or the seasonal effects seen in ovariectomized adult ewes (36).
Considering that a reduced sensitivity to E2 feedback was evident in ovary-intact prenatal T females by 12 wk, around the time of early initiation of neuroendocrine puberty in the ovariectomized females exposed to constant E2 (11, 12, 22), the absence of initiation of progestogenic cycles at this early age is paradoxical. The delay in onset of progestogenic cycles, on the one hand, may imply that the reduction in E2 sensitivity is not of sufficient magnitude. A second possibility is that although the neuroendocrine mechanisms controlling negative feedback inhibition of E2 have been altered by prenatal T treatment, the E2 positive feedback mechanism may not be operative until a later time. In the ovariectomized females exposed to constant E2, this is the case (11, 12), and the ability of the prenatal T female to produce a surge of GnRH, as measured in the portal vasculature, is impaired or absent (37). A recent study (16) using a different breed of sheep (Dorset) found no definable LH surges in prenatal T females at all ages studied (11, 15, 19, 23, and 27 wk). In the Suffolk breed that we use, the surge mechanism was found to remain operative in most prenatal T-treated ovary-intact females (14, 38). Recently, we found follicular phase levels of exogenous estrogen produced a LH surge in three of seven females at 12 wk, two of seven at 23 wk, and five of six at 54 wk, although the latency was increased and surge magnitude reduced (38). The most salient explanations for these differing results using the same dose of T and period of treatment are genetic differences in sensitivity to T programming. The early escape from E2 inhibitory feedback (high LH pulse frequency) in this study, combined with a potentially operative positive feedback mechanism in at least in some of T females of this breed, should have led to early puberty. Failure of prenatal T females to initiate the predicted early puberty raises the consideration that the ovary may not be capable of producing a preovulatory E2 rise. Our studies in very young females found that as early as 5 wk, the ovaries of prenatal T females were multifollicular and express abundant follistatin mRNA and relatively low activin B mRNA, all features suggestive of a compromised ovarian follicular function (15). Whatever the ovarian deficiency, it becomes at least partially restored by the time normal females achieve puberty at 25–30 wk of age because most prenatal T females also begin to exhibit reproductive cycles at that time (10, 14). Thus, prenatal T appears to dissociate neuroendocrine puberty, which is achieved at a relatively young age (11, 12, 22) compared with ovarian puberty (10, 14), which occurs much later.
Prenatal programming of steroid feedback disruptions
Disruption of E2 negative feedback evidenced by prenatal T treatment in the present study in conjunction with earlier findings of disruption of progesterone negative feedback (17) and E2 positive feedback (11, 12, 14, 16) is suggestive of more generalized disruption of the steroid feedback systems controlling pulsatile and surge release of GnRH by prenatal exposure to T and its metabolites. If other neuroendocrine systems responsive to E2 feedback are similarly involved remains to be determined. For instance, prenatal T treatment programs male-typical behavior in these animals (Lee, T., unpublished data), suggesting that neuroendocrine feedback mechanisms regulating behavioral centers may be similarly affected. Generalized disruptions of neuroendocrine systems may be programmed by altering neurogenesis, neuron survival, and even angiogenesis in the central nervous system, where androgens have been implicated (39, 40, 41).
Progressive deterioration of the neuroendocrine axis
Earlier studies of others and ours found progressive deterioration of reproductive cyclicity with majority of animals cycling the first year but becoming anovulatory the next year (9, 10). Findings from the present study indicate that development of hypergonadotropism in prenatal T females may also be a progressive event as reflected by abnormally increased pretreatment levels of circulating LH at 24 wk of age compared with 12 wk. Further corroboration for this premise also comes from the greater increase in LH to FSH ratio of prenatal T females compared with controls during the pretreatment, post-GnRH-A (72 h) and E2 treatment periods at 24 wk of age than 12 wk.
Differential programming of E2 feedback control of LH and FSH
Considering that LH and FSH are produced by the same gonadotrope (42, 43), and E2 is a major negative feedback regulator of both LH and FSH at the pituitary level (23, 24, 25), one would expect changes in feedback sensitivity to E2 be reflected by changes in circulating FSH. Although the reduced responsiveness to E2 negative feedback was reflected as an increase in circulating LH concentrations and LH pulse frequency in the prenatal T females after GnRH-A and E2 treatment, no such differences in circulating levels of FSH were evident in prenatal T females at either 12 or 24 wk. Such findings are also not consistent with slow frequency GnRH pulse favoring FSH (44) and raise the possibility that other pituitary paracrine modulators of FSH production/release (23, 24, 25, 45, 46, 47) and/or changes in FSH heterogeneity (48) may be involved. Because LH and FSH are produced and released from the same gonadotrope, such differences in regulation are likely to involve different secretory pathways. It is well documented FSH is predominantly secreted via a constitutive pathway and LH via a regulated pathway involving coupling to a stimulus, GnRH (23, 24, 25, 49, 50). Alternatively, because of the integrated nature of FSH measures involved, we cannot assess if there are changes in temporal pattern of FSH release between control and prenatal T females. It should be noted, however, that FSH secretory dynamics could not be assessed reliably from peripheral measurements because of its long circulatory half-life (48). This would require measurements closer to the site of release (51, 52).
Implications of reduced E2 negative feedback
The reduced responsiveness to E2 negative feedback is likely to contribute, at least in part, to the developing hypergonadotropism in the prenatal T females (14, 18), although other mechanisms such as increased pituitary gonadotropic responsiveness to GnRH may be involved. That the prenatal T female becomes hypergonadotropic is evidenced by 1) the higher circulating levels of LH during the prepubertal period (14), 2) the increased follicular phase LH pulse frequency (18), and 3) the absence of progesterone negative feedback (17). To what extent reduced sensitivity to E2 negative feedback contributes to the hypergonadotropism and what are the consequences of this feedback disruption in terms of fertility remain to be determined. The prenatal T females exhibit several characteristic features that are typical of the majority of women with polycystic ovary syndrome (PCOS), namely hypergonadotropism, and an increase in LH to FSH ratio and multifollicular ovaries (53, 54, 55). The hypergonadotropism in women with PCOS is a contributing factor in development of hyperandrogenism, multifollicular ovaries, and infertility because treatment with GnRH-A helps correct hyperandrogenism of ovarian origin in women with PCOS (56, 57, 58, 59). Whether early perturbations in gonadotropin dynamics leading to altered LH to FSH ratio underlie the functional hyperandrogenism and multifollicular ovarian development (15) and loss of cyclicity (9, 10) in the prenatal T female and other hyperandrogenic disorders remain to be determined. If hypergonadotropism is the basis, normalization of gonadotropic milieu should restore fertility in prenatal T females.
Acknowledgments
We are grateful to Mr. Douglas D. Doop and Mr. Gary McCalla for help with animal breeding and/or maintenance, Drs. Gordon D. Niswender and Leo E. Reichert, Jr. for the generous supply of LH RIA reagents, Dr. Teresa Steckler for assistance with formatting of figures, and Dr. Albert F. Parlow and the National Hormone and Pituitary Program for the FSH reagents. The GnRH-A and Nal Glu used in this study were obtained from the Contraceptive Development Branch, Center for Population Research, National Institute of Child Health and Human Development and synthesized by the Salk Institute under Contract 1-HD-02906 with the National Institutes of Health.
Footnotes
This work was supported by United States Public Health Service Grants HD 41098 and P01 HD44232 (to V.P.) and by a Department of Biotechnology Overseas Associateship, India (to H.N.S.).
Abbreviations: E2, Estradiol; GnRH-A, GnRH antagonist; PCOS, polycystic ovary syndrome; T, testosterone.
References
Barker DJP 1994 Programming the baby. In: Mothers, babies, and disease in later life. London: BMJ; 14–36
Rhind SM, Rae MT, Brooks AN 2001 Effects of nutrition and environmental factors on the fetal programming of the reproductive axis. Reproduction 122:205–214
Godfrey KM 1998 Maternal regulation of fetal development and health in adult life. Eur J Obstet Gynecol Reprod Biol. 78:141–150
Lucas A 1991 Programming by early nutrition in man. In: Block GR, Whelan J, eds. The childhood environment and adult disease. Chichester, UK: John Wiley and Sons; 38–55
Nathanielsz PW 1999 Life in the womb: the origin of health and disease. New York: Promethean Press
Meisel RL, Ward IL 1981 Fetal female rats are masculinized by male littermates located caudally in the uterus. Science 213:239–242
Vom Saal FS, Even MD, Quadagno DM 1991 Effects of maternal stress on puberty, fertility and aggressive behavior of female mice from different intrauterine positions. Physiol Behav. 49:1073–1078
Abbott DH, Dumesic DA, Eisner JR, Kemnitz JW, Goy RW 1997 The prenatally androgenized female Rhesus monkey as a model for PCOS. In: Azziz R, Nestler JE, Dewailly D, eds. Androgen excess disorders in women. Philadelphia: Lippincott-Raven; 369–384
Clarke IJ, Scaramuzzi RJ, Short RV 1977 Ovulation in prenatally androgenized ewes. J Endocrinol. 73:385–389
Birch RA, Padmanabhan V, Foster DL, Robinson JE 2003 Prenatal programming of reproductive neuroendocrine function: fetal androgen exposure produces progressive disruption of reproductive cycles in sheep. Endocrinology. 144:1426–1434
Wood RI, Foster DL 1998 Sexual differentiation of reproductive neuroendocrine function in sheep. Rev Reprod. 3:130–140
Robinson JE, Birch RA, Foster DL, Padmanabhan V 2002 Prenatal exposure of the ovine fetus to androgens sexually differentiates the steroid feedback mechanisms that control gonadotropin releasing hormone secretion and disrupts ovarian cycles. Arch Sex Behav. 31:35–41
DeHaan KC, Berger LL, Kesler DJ, McKeith FK, Thomas DL, Nash TG 1987 Effect of prenatal androgenization on lamb performance, carcass composition and reproductive function. J Anim Sci. 65:1465–1470
Sharma TP, Herkimer C, West C, Ye W, Birch R, Robinson JE, Foster DL, Padmanabhan V 2002 Fetal programming: prenatal androgen disrupts positive feedback actions of estradiol but does not affect timing of puberty in female sheep. Biol Reprod. 66:924–933
West C, Foster DL, Evans NP, Robinson J, Padmanabhan V 2001 Intra follicular activin availability is altered in prenatally-androgenized lambs. Mol Cell Endocrinol. 185:51–59
Unsworth WP, Taylor JA, Robinson JE 2005 Prenatal programming of reproductive neuroendocrine function: the effect of prenatal androgens on the development of estrogen positive feedback and ovarian cycles in the ewe. Biol Reprod. 72:619–627
Robinson JE, Forsdike RA, Taylor JA 1999 In utero exposure of female lambs to testosterone reduces the sensitivity of the GnRH neuronal network to inhibition by progesterone. Endocrinology 140:5797–5805
Savabieasfahani M, Lee JS, Herkimer C, Sharma TP, Foster DL, Padmanabhan V 2005 Fetal programming: testosterone exposure of the female sheep during mid-gestation disrupts the dynamics of its adult gonadotropin secretion during the periovulatory period. Biol Reprod. 72:221–229
Foster DL 1994 Puberty in the sheep. In: Knobil E, Neill JD, eds. Physiology of reproduction. Vol 2. New York: Raven; 411–451
Foster DL, Ryan KD 1979 Endocrine mechanisms governing transition into adulthood: a marked decrease in inhibitory feedback action of estradiol on tonic secretion of luteinizing hormone in the lamb during puberty. Endocrinology 105:896–904
Huffman LJ, Inskeep EK, Goodman RL 1987 Changes in episodic luteinizing hormone secretion leading to puberty in the lamb. Biol Reprod. 37:755–761
Kosut SS, Wood RI, Herbosa-Encarnacion C, Foster DL 1997 Prenatal androgens time neuroendocrine puberty in the sheep: effect of testosterone dose. Endocrinology 138:1072–1077
McNeilly AS 1991 The ovarian follicle and fertility. J Steroid Biochem Mol Biol. 40:29–33
Padmanabhan V, Sharma TP 2001 Neuroendocrine vs. paracrine control of follicle-stimulating hormone. Arch Med Res. 32:533–543
Padmanabhan V, West C 2001 Endocrine, autocrine, and paracrine actions of inhibin/activin/follistatin on follicle-stimulating hormone. In: Muttukrishna S, Ledger W, eds. Inhibin, activin and follistatin in human reproductive physiology. Chap III. London: Imperial College; 61–90
Manikkam M, Crespi EJ, Doop DD, Herkimer C, Lee JS, Yu S, Brown MB, Foster DL, Padmanabhan V 2004 Fetal programming: prenatal testosterone excess leads to fetal growth retardation and postnatal catch-up growth in sheep. Endocrinology 145:790–798
West CR, Carlson NE, Lee JS, McNeilly A, Sharma TP, Ye W, Padmanabhan V 2002 Acidic mix of FSH isoforms are better facilitators of ovarian follicular maturation and estradiol production than the less acidic. Endocrinology 143:107–116
Niswender GD, Reichert Jr LE, Midgley Jr AR, Nalbandov AV 1969 Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84:1166–1173
Padmanabhan V, McFadden K, Mauger DT, Karsch FJ, Midgley Jr AR 1997 Neuroendocrine control of follicle-stimulating hormone (FSH) secretion: I. Direct evidence for separate episodic and basal components of FSH secretion. Endocrinology 138:424–432
Evans NP, Dahl GE, Mauger DT, Padmanabhan V, Thrun LA, Karsch FJ 1995 Does estradiol induce the preovulatory gonadotropin-releasing hormone (GnRH) surge in the ewe by inducing a progressive change in the mode of operation of the GnRH neurosecretory system Endocrinology 136:5511–5519
Padmanabhan V, Evans NP, Dahl GE, McFadden KL, Mauger DT, Karsch FJ 1995 Evidence for short or ultrashort loop negative feedback of gonadotropin-releasing hormone secretion. Neuroendocrinology 62:248–258
Veldhuis JD, Johnson ML 1986 Cluster analysis: a simple versatile, and robust algorithm for endocrine pulse detection. Am J Physiol. 250:E486–E493
Diggle PJ, Heagerty P, Liang K-Y, Zeger S 2002 Analysis of longitudinal data. 2nd ed. Oxford, UK: Oxford University Press
Campbell BK, Dobson H, Scaramuzzi RJ 1998 Ovarian function in ewes made hypogonadal with GnRH antagonist and stimulated with FSH in the presence or absence of low amplitude LH pulses. J Endocrinol. 156:213–222
Claypool LE, Foster DL 1990 Sexual differentiation of the mechanism controlling pulsatile secretion of luteinizing hormone contributes to sexual differences in timing of puberty in sheep. Endocrinology 126:1206–1215
Robinson JE, Radford HM, Karsch FJ 1985 Seasonal changes in pulsatile luteinizing hormone secretion in the ewe: relationship of frequency of LH pulses to day length and response to estradiol negative feedback. Biol Reprod. 33:324–334
Herbosa CG, Dahl GE, Evans NP, Pelt J, Wood RI, Foster DL 1996 Sexual differentiation of the surge mode of gonadotropin secretion: prenatal androgens abolish the gonadotropin-releasing hormone surge in the sheep. J Neuroendocrinol. 8:627–633
Lee JS, Herkimer C, Sarma HN, Manikkam M, Foster DL, Padmanabhan V 2004 Fetal programming: developmental progression of estradiol positive feedback defects in prenatally T-treated sheep. Biol Reprod. 120:121
Desai M, Hales CN 1997 Role of fetal and infant growth in programming metabolism in later life. Biol Rev. 72:329–348
Yu WH 1989 Administration of testosterone attenuates neuronal loss following axotomy in the brain-stem motor nuclei of female rats. J Neurosci. 9:3908–3914
Rasika S, Alvarez-Buylla A, Nottebohm F 1999 BDNF mediates the effects of testosterone on the survival of new neurons in an adult brain. Neuron. 22:53–62
Molter-Gerard C, Fontaine J, Guerin S, Taragnat C 1999 Differential regulation of the gonadotropin storage pattern by gonadotropin releasing hormone pulse frequency in the ewe. Biol Reprod. 60:1224–1230
Thomas SG. Clarke IJ 1997 The positive feedback action of estrogen mobilizes LH-containing, but not FSH-containing secretory granules in ovine gonadotropes. Endocrinology 138:1347–1350
Marshall JC, Dalkin AC, Haisenleder DJ, Griffin ML, Kelch RP 1992 GnRH pulses-the regulators of human reproduction. Trans Am Clin Climatol Assoc. 104:31–46
Bilezikjian LM, Blount AL, Corrigan AZ, Leal A, Chen Y, Vale WW 2001 Actions of activins, inhibins and follistatins: implications in anterior pituitary function. Clin Exp Pharmacol Physiol. 28:244–248
DePaolo LV, Bicsak TA, Erickson GF, Shimasaki S, Ling N 1991 Follistatin and activin: a potential intrinsic regulatory system within diverse tissues. Proc Soc Exp Biol Med. [ Erratum (1992) 200:447] 198:500–512
Phillips DJ, Woodruff TK 2004 Inhibin: actions and signalling. Growth Factors 22:13–18
Padmanabhan V, Lee JS, Beitins IZ 1998 Follicle-stimulating isohormones: regulation and biological significance. In: Thatcher WW, Inskeep EK, Niswender GD, Doberska C, eds. Reproduction in domestic ruminants IV. J Reprod Fertil Suppl 54 London: Cambridge; 87–99
Fanworth PG 1995 Gonadotropin secretion revisited. How many ways can FSH leave a gonadotroph. J Endocrinol. 145:387–395
Muyan M, Ryzmkiewicz DM, Boime I 1994 Secretion of lutropin and follitropin from transfected GH3 cells: evidence for separate secretory pathways. Mol Endocrinol. 8:1789–1797
Padmanabhan V, Battaglia D, Brown MB, Karsch FJ, Lee JS, Pan W, Phillips DJ, VanCleeff J 2002 Neuroendocrine control of follicle-stimulating hormone secretion: II. Is follistatin-induced suppression of follicle-stimulating hormone secretion mediated via changes in activin availability and does it involve changes in GnRH secretion Biol Reprod. 66:1395–1402
Padmanabhan V, Brown M, Dahl G, Evans NP, Karsch FJ, Mauger DT, Neill JD, Van Cleeff J 2003 Neuroendocrine control of FSH secretion: III. Is there a GnRH-independent component of episodic FSH secretion in ovariectomized and luteal phase ewes Endocrinology 144:1380–1392
Marshall JC, Eagleson CA 1999 Neuroendocrine aspects of polycystic ovary syndrome. Endocrinol Metab Clin North Am. 28:295–324
Hall JE, Taylor AE, Hayes FJ Crowley Jr WF 1998 Insights into hypothalamic-pituitary dysfunction in polycystic ovary syndrome. J Endocrinol Invest. 21:602–611
Rosenfield RL 1997 Is polycystic ovary syndrome a neuroendocrine or an ovarian disorder Clin Endocrinol (Oxf). 47:423–424
McCartney CR, Eagleson CA, Marshall JC 2002 Regulation of gonadotropin secretion: implications for polycystic ovary syndrome. Semin Reprod Med. 20:317–326
Cardone VS 2003 GnRH antagonists for treatment of polycystic ovarian syndrome. Fertil Steril. 80(Suppl 1):S25–S31; discussion S32–S34
Lubin V, Charbonnel B, Bouchard P 1998 The use of gonadotrophin-releasing hormone antagonists in polycystic ovarian disease. Baill Clin Obstet Gynaecol. 12:607–618
Hayes FJ, Taylor AE, Martin KA, Hall JE 1998 Use of a gonadotropin-releasing hormone antagonist as a physiologic probe in polycystic ovary syndrome: assessment of neuroendocrine and androgen dynamics. J Clin Endocrinol Metab. 83:2343–2349(Hirendra N. Sarma, Mohan )
Abstract
Exposure of female sheep fetuses to excess testosterone (T) during early to midgestation produces postnatal hypergonadotropism manifest as a selective increase in LH. This hypergonadotropism may result from reduced sensitivity to estradiol (E2) negative feedback and/or increased pituitary sensitivity to GnRH. We tested the hypothesis that excess T before birth reduces responsiveness of LH and FSH to E2 negative feedback after birth. Pregnant ewes were treated with T propionate (100 mg/kg in cotton seed oil) or vehicle twice weekly from d 30–90 gestation. Responsiveness to E2 negative feedback was assessed at 12 and 24 wk of age in the ovary-intact female offspring. Our experimental strategy was first to arrest follicular growth and reduce endogenous E2 by administering the GnRH antagonist (GnRH-A), Nal-Glu (50 μg/kg sc every 12 h for 72 h), and then provide a fixed amount of exogenous E2 via an implant. Blood samples were obtained every 20 min at 12 wk and every 10 min at 24 wk before treatment, during and after GnRH-A treatment both before and after E2 implant. GnRH-A ablated LH pulsatility, reduced FSH by approximately 25%, and E2 production diminished to near detection limit of assay at both ages in both groups. Prenatal T treatment produced a precocious and selective reduction in responsiveness of LH but not FSH to E2 negative feedback, which was manifest mainly at the level of LH/GnRH pulse frequency. Collectively, these findings support the hypothesis that prenatal exposure to excess T decreases postnatal responsiveness to E2 inhibitory feedback of LH/GnRH secretion to contribute to the development of hypergonadotropism.
Introduction
STRONG EPIDEMIOLOGICAL evidence reveals that the environment to which the fetus is exposed during gestation influences prenatal development and subsequent expression of physiology and behavior during adulthood (1, 2, 3, 4, 5). In mammals, such programming of the reproductive system by hormones is striking. This can occur naturally as exemplified by the finding that female rats are masculinized in utero by hormones emanating from male litter mates sharing the same uterine horn (6, 7). Such females are less attractive to males, more aggressive, and reproductively compromised. Similarly, exposure to excess testosterone (T) in utero experimentally can lead to sterility in rodents, monkeys, and sheep (7, 8, 9, 10). Our studies, and those of others, using the sheep as a model reveal that prenatal exposure of the female to excess T during the middle third of gestation produces a suite of adult disorders (9, 10, 11, 12, 13, 14, 15, 16, 17, 18), culminating in early reproductive failure (9, 10). As such, manipulation of the prenatal steroid environment provides a powerful experimental tool for understanding the neuroendocrine and ovarian mechanisms that underlie prenatal programming of the reproductive axis.
At the neuroendocrine level, it is well established that prenatal T treatment reduces hypothalamic sensitivity to estradiol (E2) positive feedback (11, 12, 14, 16) and progesterone negative feedback (17), the two major feedback systems involved in the control of cyclic changes in GnRH/gonadotropin secretion. From a developmental perspective, E2 negative feedback is the predominant feedback system operational before puberty (19, 20, 21), and the reduction in sensitivity to this steroid feedback inhibition is responsible for the pubertal increase in GnRH (19, 20, 21). Studies in the female sheep have demonstrated this mechanism using the ovariectomized E2 replaced model (20), in which peripheral steroid concentrations are maintained at physiological levels by means of a constant release device (E2 capsule sc) from shortly after birth. In this model, circulating LH levels remain suppressed in response to chronically low exogenous E2 until about 30 wk of age when they rise markedly. The timing of this robust LH increase, reflecting the escape from E2 negative feedback, occurs at the same time as the initiation of ovulations and estrous behavior (puberty) in normal intact lambs (19, 21). Prenatal exposure to T markedly advances the time of the pubertal LH rise to 10 wk, an age that corresponds to the initiation of puberty in the male (11, 12, 22). This prenatal programming by T would explain the much earlier decrease in sensitivity to negative feedback in the male. Interestingly, in the prenatal T female, when the ovaries are left in situ instead of replacing them with an E2 capsule as a source of steroids, the time of puberty is not advanced (10, 14), as would be predicted from the finding of a precocial LH rise in the prenatal T, postnatally ovariectomized E2 model. Rather, in the prenatal T female left ovary intact (no exogenous E2), initiation of progestogenic cycles occurred at the same time as in the controls. This raises two possibilities: 1) the constant release E2 in the ovariectomized E2-clamped model may exert further organizational effects postnatally to combine with those of the prenatal T to result in the precocious reduction in sensitivity to E2 negative feedback; 2) alternatively, ovarian factors may play a protective role and not allow reduction in sensitivity to E2 negative feedback (early postnatal LH rise) to be expressed. There is some evidence that the ovary-intact prenatal T female is hypergonadotropic during the prepubertal period (14), raising the possibility that such females do express a reduced sensitivity to E2 negative feedback. However, an increased LH secretion alone does not provide evidence for reduced E2 feedback (hypothalamic effect) because this may reflect a reduced amount of ovarian steroid secretion (ovarian effect) or alternatively increased pituitary gonadotropin responsiveness to GnRH (pituitary effect) in such females. In the present study, the primary hypothesis we tested is that prenatal exposure to T produces a precocial reduction in sensitivity to E2 negative feedback after birth. Our results indicate that this indeed is the case and further that development of hypergonadotropism occurs progressively over time. In addition, because E2, along with inhibin (23, 24, 25), is a major negative feedback regulator of FSH, the second hypothesis we tested is that the reduced sensitivity to E2 feedback will also be reflected as increased FSH secretion in the prenatal T females. Our results indicate that this is not the case and leads to the conclusion that prenatal T treatment has differential effects on feedback responsiveness of LH and FSH to E2.
Materials and Methods
Generation of prenatal T females
All procedures were approved by the Institutional Animal Care and Use Committee of University of Michigan and were consistent with the National Institutes of Health Guide for Use and Care of Animals. Suffolk sheep used in this study were prenatally treated with T or cottonseed oil (control) during 30–90 d gestation (n = 7 per treatment group). The details of the prenatal treatments, husbandry and nutrition of maternal ewes, newborn measures, and growth characteristics until 4 months of age of the larger cohort of females of which these are part of have been published (26). Briefly, pregnant ewes were administered im 100 mg T propionate (Sigma Chemical Co., St. Louis, MO) suspended in 2.0 ml cottonseed oil twice weekly from d 30–90 gestation (term, 147 d). Control ewes received the same volume of cottonseed oil in a similar regimen as T. Lambs were born between March 15 and April 20. They were weaned at 8 wk of age and maintained outdoors at the Sheep Research Facility (Ann Arbor, MI; 42°, 18'N) under natural photoperiod and were provided ad libitum access to commercial feed pellets (Shur-Gain, Elma, NY) consisting of 3.6 MCal/kg digestible energy and 18% crude protein. When they reached approximately 40 kg body weight, they were switched to a pellet feed with 15% crude protein to avoid excessive fat deposition as the rate of growth naturally diminished and adult size was approached. Trace mineralized salt with selenium and vitamins A, D, and E (Armada Grain Co., Armada, MI) were freely accessible throughout the study. The postnatal growth of all female lambs was monitored by weighing every 2 wk before feeding. Samples were collected twice weekly from 20 wk of age for measurement of progesterone to assess the onset of puberty.
Experimental design
To determine developmental changes in gonadotropin responsiveness to E2 negative feedback, feedback tests were conducted at approximately 12 wk of age (early prepubertal) and at approximately 24 wk (shortly before the expected time of puberty). Typically, E2 negative feedback tests are conducted in the absence of the ovaries to avoid confounding from endogenous E2, which could differ between individuals depending on the stage of follicular development. Thus, in our study of ovary-intact females, we needed to produce a reduced but similar starting follicular E2 milieu in all females. To do this, we reduced and normalized endogenous E2 production by blocking LH- and GnRH-induced FSH release with the GnRH antagonist (GnRH-A), Nal-Glu (Nal-Glu [Ac-D2Nal1, D4ClPhe2, D3Pal3, Arg5, 4-(methoxybenzoyl)-D-2-aminobutyric acid6, D-Ala10]), as described previously (27).
In experimental phase 3, the response to E2 feedback inhibition was tested, and a 3-mm SILASTIC brand (Dow Corning Corp., Midland, MI) E2 capsule filled with crystalline E2-17 was inserted sc in the axillary region of both control and prenatal T lambs. This size implant produces circulating concentrations of E2 of less than 1 pg/ml. Blood samples were collected frequently at 12 wk (20-min intervals) and 24 wk (10-min intervals) for 6 h, from 66–72 h, after implant insertion, which was 138–144 h after cessation of the GnRH-A.
All blood samples were collected in heparinized tubes, and plasma was separated at 5 C. Plasma was stored at –20 C until assay. Plasma LH concentrations were determined in all samples. Plasma E2 and FSH levels were determined in pooled samples obtained by combining equal aliquots of plasma from all samples of a given female during each phase (n = 19 for 12 wk where the sampling frequency was every 20 min and 37 for 24 wk where the sampling frequency was 10 min).
RIA
Plasma LH concentrations were determined using a validated competitive double-antibody RIA (28) in duplicate 10- to 200-μl samples. The assay sensitivity (2 SD from the buffer control) of LH assay was 0.34 ± 0.07 ng/ml (n = 19 assays). Mean intraassay coefficient of variations at 80 and 20% displacement points were 6.35 ± 0.27 and 3.15 ± 0.13%, respectively, and the mean median variance ratio was 0.04 ± 0.003. The interassay coefficient of variations in three quality control pools averaging 1.0, 13, and 23 ng/ml were 19.5, 3.9, and 3.5%, respectively. FSH was measured in duplicate 200-μl samples (n = 2 assays) with a validated RIA (29) using reagents from the National Hormone and Pituitary Program. Assay sensitivity averaged 0.05 and 0.03 ng/ml, respectively, and average intraassay coefficients of variation at 80 and 20% displacement points were 7.97 and 3.98%. Interassay coefficient of variation for two plasma quality control pools averaging 5.4 and 30 ng/ml was 0.37 and 1.1%, respectively. Plasma concentrations of E2 were measured using a commercial RIA kit (Coat-A-Count E2, MAIA, Polymedco Inc., Cortland Manor, NY) validated for use in sheep (30). Assay sensitivity averaged 0.26 ± 0.06 pg/ml (n = 3 assays), and mean median variance ratio averaged 0.05 ± 0.03. Mean intraassay coefficient of variation at 80 and 20% displacement points averaged 14.9 ± 3.9 and 7.5 ± 2.0%, respectively. Interassay coefficient of variations based on two quality control pools averaging 5 and 40 pg/ml averaged 19.4 and 9.3%, respectively. Each assay included both control and prenatal T females. Plasma concentrations of progesterone were measured by a commercial RIA kit (Coat-A-Count P4; Diagnostic Products Corp., Los Angeles, CA) in daily samples. This assay has been validated for use in sheep (31). Sensitivity of progesterone assay averaged 0.03 ± 0.01 ng/ml (n = 8 assays). The intraassay coefficient of variation, based on two quality control pools averaging 1.99 and 13.7 ng/ml averaged 8.7, and 10.4%, respectively. The interassay coefficients of variation for the same quality control pools were 10.5 and 10.0%, respectively.
Statistical analysis
The first sustained rise in circulating progesterone was used to determine puberty, and this was based on two criteria: the time when circulating progesterone concentrations first exceeded baseline by 2 times assay sensitivity, and this must be followed by a progestogenic cycle of greater than 0.5 ng. Differences in timing of the onset of puberty were determined by ANOVA. Growths of lambs were compared using a linear mixed model in which a separate regression line (random intercept and random slope) was calculated for lamb weights of each mother. Overall effect of age, treatment, and the age by treatment interactions were examined.
Serial LH data from control and prenatal T females from each time period were subjected to pulse analysis using the Cluster algorithm (32). The Cluster algorithm identifies pulses using criteria that define a pulse such that the peak of the pulse differs significantly from both the preceding and following nadirs according to two-sample Student’s t tests. For analysis with Cluster, the minimum number of data points in a peak and nadir were set at 1 and 1, respectively, when blood samples were obtained at 20-min intervals and 2 and 2, respectively, when blood samples were obtained at 10-min intervals. The Student’s t statistic values used to identify a significant increase from preceding nadir and a decrease to following nadir were both 1.0 and 2.0 for the 10- and 20-min sampling frequencies, respectively.
Hormone data for the control and prenatal T-treated females were compared at 12 wk of age for the treatment periods pre-; 72 h after GnRH-A; and E2; and at 24 wk of age for the periods pre; 24, 48, 72 h after GnRH-A; and E2 (Fig. 1). The variables compared were E2 (concentration), LH (concentration, pulse frequency, and pulse amplitude), FSH (concentration), and LH to FSH ratio. A repeated-measures ANOVA was conducted for comparing concentrations of LH and FSH, E2, and the LH pulse amplitude. The repeated-measures ANOVA had one between-subjects factor (control vs. prenatal T) and one within-subjects factor (period), with period having three levels for the 12-wk analysis and five levels for the 24-wk analysis. The main effects of treatment and period and the interactions between treatment and period were examined. Post hoc tests were used to compare treatment means within each period, and the change between the post-GnRH-A at 72 h and E2 periods for C- vs. T-treated females. Residuals from these analyses were checked for normality. A nonparametric Wilcoxon rank sum test with exact P values was used to compare LH to FSH ratios for C- vs. T-treated females within each treatment period because normality could not be assumed for these measures. For LH pulse frequency, a count variable, which could not be assumed to be normally distributed, a Poisson regression with repeated measures was carried out, using the generalized estimating equations method (33). This analysis allowed us to take into account the correlation among observations on the same female for the different bleeds and also allowed us to take the nonnormality of the response into account. The percent suppression from the GnRH-A 72 h to E2 periods for each of the hormones was compared for the control vs. prenatal T-treated females at 12 and 24 wk. A nonparametric Wilcoxon rank sum test was used for the analysis of the percent suppression data because we could not assume normality for these measures.
To allow comparisons of the age-dependent changes in E2 feedback of gonadotropins at 12 and 24 wk, alternate data points from the 10-min LH data series at 24 wk of age were deleted so that they would be comparable to the 20-min data series at 12 wk, and the normalized data series was then subjected to a second set of Cluster analyses. To examine the effect of age on percent suppression, the difference between the normalized 12- and 24-wk suppression was calculated for each animal and a Wilcoxon signed rank test was used to assess the significance of the change in percent suppression at the two ages. We also examined the overall effect of treatment on the normalized 12- vs. 24-wk LH suppression data by first calculating the average of the normalized percent suppression from 12–24 wk for each animal and then comparing this mean for C- vs. T-treated animals using a Wilcoxon rank sum test. All analyses were carried out using SAS for Windows release 9.1.3 (2002–2003; SAS Institute, Cary, NC).
Results
Growth rate and onset of puberty
Birth weights and growth rate during the first 16 wk of postnatal development for the larger group of animals of which these were part of have been published (26). These showed that prenatal T females have lower birth weight and exhibit catch up growth between 8 and 12 wk of age. Analyses of biweekly weights beginning at 16 wk of age revealed no significant treatment effect indicating that growth rates for young females within both treatments was similar after 16 wk (not shown). The estimated slope for the regression of weight on age was 0.15 ± 0.01 for controls and 0.14 ± 0.01 for the prenatal T females. When E2 negative feedback tests were conducted, control and prenatal T lambs weighed 30.3 ± 5.6 and 31.6 ± 2.6 kg, respectively, at 12 wk and 57.3 ± 7.3 and 63.3 ± 3.4 kg, respectively at 24 wk.
The age of puberty for females used in this study was delayed in prenatal T females compared with controls (Fig. 2). The first E2 negative feedback test at 12 wk of age was 15 and 19 wk before puberty in the control and prenatal T females, respectively, during the age of relatively high sensitivity to estrogen negative feedback. The second E2 feedback test at 24 wk of age was about 3 and 8 wk before puberty, respectively, when sensitivity to E2 negative feedback normally reduces in control females.
Circulating E2 concentrations in 12- and 24-wk-old females
Effects of prenatal T treatment on LH responsiveness to E2 negative feedback at 12 wk of age
Representative patterns of LH pulses and circulating E2 levels (shaded area) from three control and three prenatal T females are shown in Fig. 4; mean changes in circulating LH, LH pulse frequency, amplitude, and degree of E2 feedback suppression of LH are presented in Fig. 5. Before treatment, control females exhibited zero to one pulses per 6 h, and the prenatal T females had one to five pulses. Administration of the GnRH-A abolished LH pulsatility completely regardless of prenatal treatment. Cessation of the GnRH-A treatment resulted in restoration of pulses with both control and prenatal T females having high-frequency LH pulses between 66 and 72 h after the GnRH-A. E2 treatment reduced LH pulse frequency to zero to one pulse in control females and to one to three pulses in prenatal T females. Group statistics (Fig. 5) revealed that mean LH levels were not different between control and prenatal T females before treatment (Fig. 5; Pre), but circulating LH was higher (P < 0.01) in prenatal T females before (72 h after GnRH-A, Fig. 5) and during E2 treatment (Fig. 5). LH pulse frequency was higher (P < 0.05) in prenatal T animals compared with controls during the pretreatment period and before E2 treatment (72 h after GnRH-A). This was also the case during E2 treatment (P < 0.01) (Fig. 5) despite the higher levels of E2 in prenatal T females (Fig. 2). Comparison of pre-E2 (72 h after GnRH-A) and E2 treatment periods revealed reduced (P < 0.05) E2 suppression of LH pulse frequency but not mean LH concentrations. The greater suppression of pulse frequency in control females after E2 treatment was not reflected by a statistically significant increase (P = 0.07) in LH pulse amplitude, possibly because of absence of pulses from which to calculate amplitude from in four of the seven control females.
Twenty-four weeks of age
Circulating LH pulse patterns and circulating E2 levels (shaded area) from the same three control and three prenatal T females are shown in Fig. 6 as were presented in Fig. 5; mean LH concentrations, pulse frequencies, and amplitudes are in Fig. 7. Before treatment with the GnRH-A, despite the higher E2 levels at 24 wk of age in prenatal T females, circulating LH levels and LH pulse frequency were higher (P < 0.01) in prenatal T females compared with controls. In both groups of 24-wk-old females, as at their younger age, the GnRH-A reduced circulating LH levels and completely abolished LH pulsatility (P < 0.01). Cessation of the GnRH-A treatment resulted in a gradual restoration of circulating LH and return of LH pulses. During the post-GnRH-A period, LH pulse frequency was higher in the prenatal T females during all three-time points (24, 48 and 72 h) (P < 0.05). These higher pulse frequencies occurred in the presence of similar circulating levels of E2 (Fig. 2, bottom panel). The increase in LH pulse frequency was reflected as an increase in circulating LH levels at 24 (P < 0.01), 48 (P < 0.05), and 72 h (P < 0.05) after GnRH treatment (top panel). E2 treatment reduced the number of LH pulses in the control and prenatal T females. LH concentration (P < 0.01) and LH pulse frequency (P < 0.05) were both higher in the prenatal T females compared with controls during E2 treatment. LH pulse amplitude of prenatal T females also tended to be higher compared with controls during the E2 treatment period (P = 0.06). Comparison of pre-E2 (72 h after GnRH-A) and E2 treatment periods revealed a tendency for reduced E2 suppression of LH pulse frequency (P = 0.09) but not mean LH concentrations or amplitude in prenatal T females compared with controls (right panels).
Age-associated changes in E2 negative feedback
Comparison of results between 12 and 24 wk, after adjusting the data series for differences in sampling frequency, revealed a significant treatment and age effect with LH pulse frequency but not mean LH concentrations (Fig. 8). This was reflected as reduced E2 suppression of LH frequency at 24 than 12 wk across treatment (P < 0.05) and reduced E2 suppression of LH frequency in prenatal T females compared with controls across ages (P < 0.05). There was no age by treatment interaction.
Effect of prenatal T treatment on circulating FSH at 12 and 24 wk
Plasma FSH concentrations of control and prenatal T females at 12 and 24 wk of age are depicted in Fig. 9. In contrast to near complete ablation of LH, the GnRH-A suppressed (P < 0.01) circulating FSH concentrations minimally (<25%) at both ages, and there were no treatment or age effects. Circulating levels of FSH were restored to pretreatment levels by 48 h after GnRH-A in the 24-wk-old females and by 72 h after GnRH-A in the 12-wk-old females (48 h was not studied at 12 wk). E2 treatment decreased FSH secretion in both control and prenatal T females at both ages (12 wk, P < 0.01; 24 wk, P = 0.06). Changes in the LH to FSH ratio, in general, reflected changes in LH (Table 1). During the pretreatment period, the LH to FSH ratio was higher in prenatal T females compared with controls reaching significance at 24 wk (P < 0.05). The GnRH-A reduced the LH to FSH ratio at both age groups in both treatment groups (range of suppression, 68–83%). Cessation of the GnRH-A treatment restored the LH to FSH ratio to pretreatment levels by 48 h in 24-wk-old females and by 72 h in 12-wk-old females (other times were not studied at 12 wk). LH to FSH ratio during the E2 treatment period was significantly higher (P < 0.05) in prenatal T females than controls at both ages.
Discussion
The findings from this study support the hypothesis that prenatal treatment of the developing female sheep with T during the middle third of gestation reduces postnatal LH responsiveness to E2 negative feedback, thus contributing at least in part to the development of hypergonadotropism. Based on LH pulse frequency, the reduced responsiveness to E2 negative feedback was evident as early as 12 wk of age (younger ages not studied) when control females are yet highly sensitive to E2 negative feedback (19, 20, 21). At 24 wk, the E2 feedback suppression of LH pulse frequency in the prenatal T females (20%) was lower compared with the 12-wk-old females (60%). Age-dependent reduction in responsiveness to E2 feedback was also evident in controls (90 vs. 40% at 12 vs. 24 wk, respectively), although the sensitivity to E2 feedback regulation of LH frequency across both ages was greater in controls than prenatal T females (Fig. 8). This, in concert with the fact that prenatal T females lagged a month behind control females in their initiation of reproductive cycles, is supportive of a general reduction in sensitivity of prenatal females to E2 negative feedback. Interestingly, the reduced responsiveness to E2 negative feedback of prenatal T females at 12 wk was manifest at the level of LH pulse frequency but not FSH. The validity of the GnRH-A approach that made it possible to study responsiveness to E2 feedback in the ovary-intact female and the potential implications of the findings from this study as they relate to the timing of puberty, GnRH feedback control mechanisms, development of adult hypergonadotropism and infertility, and differential regulatory control of LH and FSH are discussed below.
Validity of approach used to assess negative feedback
To assess E2 negative feedback in the presence of the ovaries, we attempted to standardize circulating E2 by reducing endogenous production and replacing this with exogenous E2 from a constant release device, the well-characterized SILASTIC brand capsule implanted sc. A GnRH-A was used to block GnRH action, thereby reducing gonadotropic drive, which, in turn, arrested follicular development and reduced endogenous E2 production. This provided a short window of time in which to conduct the E2 feedback test. Although the approach proved to be technically useful to study the responsiveness of the prenatal T females to E2 negative feedback, there are some limitations in applying this approach to other ovary-intact situations. The GnRH-A does not completely block FSH production or endogenous estrogen production. It ablates GnRH action (27), blocks LH release (27, 32), blocks production of GnRH-induced release of FSH (less acidic and more bioactive) (27), induces follicular atresia, and arrests follicular development to the gonadotropin-independent stage (34), thus limiting E2 production to near detection levels (27, 34) of the assay. Although release from GnRH-A inhibition allows a new wave of follicular growth, it does not follow that the number of follicles recruited will be the same and result in similar endogenous levels of E2. Because prenatal T females are multifollicular, we expected levels of E2 production and, hence, the feedback drive, to be equal or greater in prenatal T females compared with control females. This proved to be the case. The converse, namely increased E2 drive in the control females, would have not allowed us to test the hypothesis.
Altered sensitivity to E2 negative feedback and pubertal timing
Central to the neuroendocrine mechanisms associated with puberty in the sheep, as in other species, is the reduction in sensitivity to E2 negative feedback inhibition that allows the GnRH pulse generator to express high activity and produce the gonadotropin secretion at levels that stimulate the gonads to function like an adult (19, 20, 21). Sex differences often occur in the timing of puberty. In the female sheep, the pubertal increase in GnRH secretion occurs around 25–30 wk of age, whereas in males, this begins much earlier, at 8–10 wk of age (19). This earlier pubertal increase in GnRH secretion in the male occurs in response to the earlier decrease in sensitivity to estrogen feedback inhibition brought about by the prenatal programming of this system by exposure to steroids from the developing testes. Experimentally, the timing of the pubertal GnRH rise in ovariectomized females exposed to constant E2 can be advanced by treatment with T in utero (11, 12, 22). Such females, like males, reduce their responsiveness to E2 negative feedback, leading to an early increase in circulating LH (11, 12, 22). Thus, one would predict from the precocious initiation of neuroendocrine puberty in this model that precocious initiation of ovulatory cycles would occur had the ovaries not been removed in such prenatal T females. This is not the case as determined by our earlier studies in which progestogenic cycles occurred at the normal time (10, 14, 16, 18) or this study, where progestogenic cycles began later in the prenatal T animals. Although onset of progestogenic cycles did not occur early, circulating levels of LH, based on infrequent sampling (twice weekly), were found to be higher postnatally in prenatal T females compared with control females (14). LH pulse frequency was also found to be higher during the follicular phase of sheep treated prenatally with T from d 60–90 gestation (18). In the present study, the increased LH pulse frequency and mean LH concentrations during the pretreatment period both at 12 and 24 wk of age in the prenatal T females provides additional evidence for hypergonadotropism. Because this occurred in these prenatal T females in the face of similar or higher endogenous E2 concentrations, it provides evidence that an early reduction in LH sensitivity to E2 feedback is expressed before puberty in prenatal T females, even in the presence of the ovary. Further evidence for this decreased sensitivity was provided by their response to estrogen, when the steroid was provided by a constant release device (implant); prenatal T females were hypergonadotropic relative to control females. Finally, further corroboration for this contention is provided in the present study by the early increase in LH pulse frequency in prenatal T females during the post-GnRH-A period compared with control females in the presence of similar or lower concentrations of E2. Alternatively, the elevated LH secretion found 24 h after GnRH-A in 24-wk-old prenatal T females, when increase in endogenous E2 is negligible, may likely represent steroid-independent stimulatory effect on LH by prenatal T excess, analogous to that seen between gonadectomized male and female lambs (35) or the seasonal effects seen in ovariectomized adult ewes (36).
Considering that a reduced sensitivity to E2 feedback was evident in ovary-intact prenatal T females by 12 wk, around the time of early initiation of neuroendocrine puberty in the ovariectomized females exposed to constant E2 (11, 12, 22), the absence of initiation of progestogenic cycles at this early age is paradoxical. The delay in onset of progestogenic cycles, on the one hand, may imply that the reduction in E2 sensitivity is not of sufficient magnitude. A second possibility is that although the neuroendocrine mechanisms controlling negative feedback inhibition of E2 have been altered by prenatal T treatment, the E2 positive feedback mechanism may not be operative until a later time. In the ovariectomized females exposed to constant E2, this is the case (11, 12), and the ability of the prenatal T female to produce a surge of GnRH, as measured in the portal vasculature, is impaired or absent (37). A recent study (16) using a different breed of sheep (Dorset) found no definable LH surges in prenatal T females at all ages studied (11, 15, 19, 23, and 27 wk). In the Suffolk breed that we use, the surge mechanism was found to remain operative in most prenatal T-treated ovary-intact females (14, 38). Recently, we found follicular phase levels of exogenous estrogen produced a LH surge in three of seven females at 12 wk, two of seven at 23 wk, and five of six at 54 wk, although the latency was increased and surge magnitude reduced (38). The most salient explanations for these differing results using the same dose of T and period of treatment are genetic differences in sensitivity to T programming. The early escape from E2 inhibitory feedback (high LH pulse frequency) in this study, combined with a potentially operative positive feedback mechanism in at least in some of T females of this breed, should have led to early puberty. Failure of prenatal T females to initiate the predicted early puberty raises the consideration that the ovary may not be capable of producing a preovulatory E2 rise. Our studies in very young females found that as early as 5 wk, the ovaries of prenatal T females were multifollicular and express abundant follistatin mRNA and relatively low activin B mRNA, all features suggestive of a compromised ovarian follicular function (15). Whatever the ovarian deficiency, it becomes at least partially restored by the time normal females achieve puberty at 25–30 wk of age because most prenatal T females also begin to exhibit reproductive cycles at that time (10, 14). Thus, prenatal T appears to dissociate neuroendocrine puberty, which is achieved at a relatively young age (11, 12, 22) compared with ovarian puberty (10, 14), which occurs much later.
Prenatal programming of steroid feedback disruptions
Disruption of E2 negative feedback evidenced by prenatal T treatment in the present study in conjunction with earlier findings of disruption of progesterone negative feedback (17) and E2 positive feedback (11, 12, 14, 16) is suggestive of more generalized disruption of the steroid feedback systems controlling pulsatile and surge release of GnRH by prenatal exposure to T and its metabolites. If other neuroendocrine systems responsive to E2 feedback are similarly involved remains to be determined. For instance, prenatal T treatment programs male-typical behavior in these animals (Lee, T., unpublished data), suggesting that neuroendocrine feedback mechanisms regulating behavioral centers may be similarly affected. Generalized disruptions of neuroendocrine systems may be programmed by altering neurogenesis, neuron survival, and even angiogenesis in the central nervous system, where androgens have been implicated (39, 40, 41).
Progressive deterioration of the neuroendocrine axis
Earlier studies of others and ours found progressive deterioration of reproductive cyclicity with majority of animals cycling the first year but becoming anovulatory the next year (9, 10). Findings from the present study indicate that development of hypergonadotropism in prenatal T females may also be a progressive event as reflected by abnormally increased pretreatment levels of circulating LH at 24 wk of age compared with 12 wk. Further corroboration for this premise also comes from the greater increase in LH to FSH ratio of prenatal T females compared with controls during the pretreatment, post-GnRH-A (72 h) and E2 treatment periods at 24 wk of age than 12 wk.
Differential programming of E2 feedback control of LH and FSH
Considering that LH and FSH are produced by the same gonadotrope (42, 43), and E2 is a major negative feedback regulator of both LH and FSH at the pituitary level (23, 24, 25), one would expect changes in feedback sensitivity to E2 be reflected by changes in circulating FSH. Although the reduced responsiveness to E2 negative feedback was reflected as an increase in circulating LH concentrations and LH pulse frequency in the prenatal T females after GnRH-A and E2 treatment, no such differences in circulating levels of FSH were evident in prenatal T females at either 12 or 24 wk. Such findings are also not consistent with slow frequency GnRH pulse favoring FSH (44) and raise the possibility that other pituitary paracrine modulators of FSH production/release (23, 24, 25, 45, 46, 47) and/or changes in FSH heterogeneity (48) may be involved. Because LH and FSH are produced and released from the same gonadotrope, such differences in regulation are likely to involve different secretory pathways. It is well documented FSH is predominantly secreted via a constitutive pathway and LH via a regulated pathway involving coupling to a stimulus, GnRH (23, 24, 25, 49, 50). Alternatively, because of the integrated nature of FSH measures involved, we cannot assess if there are changes in temporal pattern of FSH release between control and prenatal T females. It should be noted, however, that FSH secretory dynamics could not be assessed reliably from peripheral measurements because of its long circulatory half-life (48). This would require measurements closer to the site of release (51, 52).
Implications of reduced E2 negative feedback
The reduced responsiveness to E2 negative feedback is likely to contribute, at least in part, to the developing hypergonadotropism in the prenatal T females (14, 18), although other mechanisms such as increased pituitary gonadotropic responsiveness to GnRH may be involved. That the prenatal T female becomes hypergonadotropic is evidenced by 1) the higher circulating levels of LH during the prepubertal period (14), 2) the increased follicular phase LH pulse frequency (18), and 3) the absence of progesterone negative feedback (17). To what extent reduced sensitivity to E2 negative feedback contributes to the hypergonadotropism and what are the consequences of this feedback disruption in terms of fertility remain to be determined. The prenatal T females exhibit several characteristic features that are typical of the majority of women with polycystic ovary syndrome (PCOS), namely hypergonadotropism, and an increase in LH to FSH ratio and multifollicular ovaries (53, 54, 55). The hypergonadotropism in women with PCOS is a contributing factor in development of hyperandrogenism, multifollicular ovaries, and infertility because treatment with GnRH-A helps correct hyperandrogenism of ovarian origin in women with PCOS (56, 57, 58, 59). Whether early perturbations in gonadotropin dynamics leading to altered LH to FSH ratio underlie the functional hyperandrogenism and multifollicular ovarian development (15) and loss of cyclicity (9, 10) in the prenatal T female and other hyperandrogenic disorders remain to be determined. If hypergonadotropism is the basis, normalization of gonadotropic milieu should restore fertility in prenatal T females.
Acknowledgments
We are grateful to Mr. Douglas D. Doop and Mr. Gary McCalla for help with animal breeding and/or maintenance, Drs. Gordon D. Niswender and Leo E. Reichert, Jr. for the generous supply of LH RIA reagents, Dr. Teresa Steckler for assistance with formatting of figures, and Dr. Albert F. Parlow and the National Hormone and Pituitary Program for the FSH reagents. The GnRH-A and Nal Glu used in this study were obtained from the Contraceptive Development Branch, Center for Population Research, National Institute of Child Health and Human Development and synthesized by the Salk Institute under Contract 1-HD-02906 with the National Institutes of Health.
Footnotes
This work was supported by United States Public Health Service Grants HD 41098 and P01 HD44232 (to V.P.) and by a Department of Biotechnology Overseas Associateship, India (to H.N.S.).
Abbreviations: E2, Estradiol; GnRH-A, GnRH antagonist; PCOS, polycystic ovary syndrome; T, testosterone.
References
Barker DJP 1994 Programming the baby. In: Mothers, babies, and disease in later life. London: BMJ; 14–36
Rhind SM, Rae MT, Brooks AN 2001 Effects of nutrition and environmental factors on the fetal programming of the reproductive axis. Reproduction 122:205–214
Godfrey KM 1998 Maternal regulation of fetal development and health in adult life. Eur J Obstet Gynecol Reprod Biol. 78:141–150
Lucas A 1991 Programming by early nutrition in man. In: Block GR, Whelan J, eds. The childhood environment and adult disease. Chichester, UK: John Wiley and Sons; 38–55
Nathanielsz PW 1999 Life in the womb: the origin of health and disease. New York: Promethean Press
Meisel RL, Ward IL 1981 Fetal female rats are masculinized by male littermates located caudally in the uterus. Science 213:239–242
Vom Saal FS, Even MD, Quadagno DM 1991 Effects of maternal stress on puberty, fertility and aggressive behavior of female mice from different intrauterine positions. Physiol Behav. 49:1073–1078
Abbott DH, Dumesic DA, Eisner JR, Kemnitz JW, Goy RW 1997 The prenatally androgenized female Rhesus monkey as a model for PCOS. In: Azziz R, Nestler JE, Dewailly D, eds. Androgen excess disorders in women. Philadelphia: Lippincott-Raven; 369–384
Clarke IJ, Scaramuzzi RJ, Short RV 1977 Ovulation in prenatally androgenized ewes. J Endocrinol. 73:385–389
Birch RA, Padmanabhan V, Foster DL, Robinson JE 2003 Prenatal programming of reproductive neuroendocrine function: fetal androgen exposure produces progressive disruption of reproductive cycles in sheep. Endocrinology. 144:1426–1434
Wood RI, Foster DL 1998 Sexual differentiation of reproductive neuroendocrine function in sheep. Rev Reprod. 3:130–140
Robinson JE, Birch RA, Foster DL, Padmanabhan V 2002 Prenatal exposure of the ovine fetus to androgens sexually differentiates the steroid feedback mechanisms that control gonadotropin releasing hormone secretion and disrupts ovarian cycles. Arch Sex Behav. 31:35–41
DeHaan KC, Berger LL, Kesler DJ, McKeith FK, Thomas DL, Nash TG 1987 Effect of prenatal androgenization on lamb performance, carcass composition and reproductive function. J Anim Sci. 65:1465–1470
Sharma TP, Herkimer C, West C, Ye W, Birch R, Robinson JE, Foster DL, Padmanabhan V 2002 Fetal programming: prenatal androgen disrupts positive feedback actions of estradiol but does not affect timing of puberty in female sheep. Biol Reprod. 66:924–933
West C, Foster DL, Evans NP, Robinson J, Padmanabhan V 2001 Intra follicular activin availability is altered in prenatally-androgenized lambs. Mol Cell Endocrinol. 185:51–59
Unsworth WP, Taylor JA, Robinson JE 2005 Prenatal programming of reproductive neuroendocrine function: the effect of prenatal androgens on the development of estrogen positive feedback and ovarian cycles in the ewe. Biol Reprod. 72:619–627
Robinson JE, Forsdike RA, Taylor JA 1999 In utero exposure of female lambs to testosterone reduces the sensitivity of the GnRH neuronal network to inhibition by progesterone. Endocrinology 140:5797–5805
Savabieasfahani M, Lee JS, Herkimer C, Sharma TP, Foster DL, Padmanabhan V 2005 Fetal programming: testosterone exposure of the female sheep during mid-gestation disrupts the dynamics of its adult gonadotropin secretion during the periovulatory period. Biol Reprod. 72:221–229
Foster DL 1994 Puberty in the sheep. In: Knobil E, Neill JD, eds. Physiology of reproduction. Vol 2. New York: Raven; 411–451
Foster DL, Ryan KD 1979 Endocrine mechanisms governing transition into adulthood: a marked decrease in inhibitory feedback action of estradiol on tonic secretion of luteinizing hormone in the lamb during puberty. Endocrinology 105:896–904
Huffman LJ, Inskeep EK, Goodman RL 1987 Changes in episodic luteinizing hormone secretion leading to puberty in the lamb. Biol Reprod. 37:755–761
Kosut SS, Wood RI, Herbosa-Encarnacion C, Foster DL 1997 Prenatal androgens time neuroendocrine puberty in the sheep: effect of testosterone dose. Endocrinology 138:1072–1077
McNeilly AS 1991 The ovarian follicle and fertility. J Steroid Biochem Mol Biol. 40:29–33
Padmanabhan V, Sharma TP 2001 Neuroendocrine vs. paracrine control of follicle-stimulating hormone. Arch Med Res. 32:533–543
Padmanabhan V, West C 2001 Endocrine, autocrine, and paracrine actions of inhibin/activin/follistatin on follicle-stimulating hormone. In: Muttukrishna S, Ledger W, eds. Inhibin, activin and follistatin in human reproductive physiology. Chap III. London: Imperial College; 61–90
Manikkam M, Crespi EJ, Doop DD, Herkimer C, Lee JS, Yu S, Brown MB, Foster DL, Padmanabhan V 2004 Fetal programming: prenatal testosterone excess leads to fetal growth retardation and postnatal catch-up growth in sheep. Endocrinology 145:790–798
West CR, Carlson NE, Lee JS, McNeilly A, Sharma TP, Ye W, Padmanabhan V 2002 Acidic mix of FSH isoforms are better facilitators of ovarian follicular maturation and estradiol production than the less acidic. Endocrinology 143:107–116
Niswender GD, Reichert Jr LE, Midgley Jr AR, Nalbandov AV 1969 Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84:1166–1173
Padmanabhan V, McFadden K, Mauger DT, Karsch FJ, Midgley Jr AR 1997 Neuroendocrine control of follicle-stimulating hormone (FSH) secretion: I. Direct evidence for separate episodic and basal components of FSH secretion. Endocrinology 138:424–432
Evans NP, Dahl GE, Mauger DT, Padmanabhan V, Thrun LA, Karsch FJ 1995 Does estradiol induce the preovulatory gonadotropin-releasing hormone (GnRH) surge in the ewe by inducing a progressive change in the mode of operation of the GnRH neurosecretory system Endocrinology 136:5511–5519
Padmanabhan V, Evans NP, Dahl GE, McFadden KL, Mauger DT, Karsch FJ 1995 Evidence for short or ultrashort loop negative feedback of gonadotropin-releasing hormone secretion. Neuroendocrinology 62:248–258
Veldhuis JD, Johnson ML 1986 Cluster analysis: a simple versatile, and robust algorithm for endocrine pulse detection. Am J Physiol. 250:E486–E493
Diggle PJ, Heagerty P, Liang K-Y, Zeger S 2002 Analysis of longitudinal data. 2nd ed. Oxford, UK: Oxford University Press
Campbell BK, Dobson H, Scaramuzzi RJ 1998 Ovarian function in ewes made hypogonadal with GnRH antagonist and stimulated with FSH in the presence or absence of low amplitude LH pulses. J Endocrinol. 156:213–222
Claypool LE, Foster DL 1990 Sexual differentiation of the mechanism controlling pulsatile secretion of luteinizing hormone contributes to sexual differences in timing of puberty in sheep. Endocrinology 126:1206–1215
Robinson JE, Radford HM, Karsch FJ 1985 Seasonal changes in pulsatile luteinizing hormone secretion in the ewe: relationship of frequency of LH pulses to day length and response to estradiol negative feedback. Biol Reprod. 33:324–334
Herbosa CG, Dahl GE, Evans NP, Pelt J, Wood RI, Foster DL 1996 Sexual differentiation of the surge mode of gonadotropin secretion: prenatal androgens abolish the gonadotropin-releasing hormone surge in the sheep. J Neuroendocrinol. 8:627–633
Lee JS, Herkimer C, Sarma HN, Manikkam M, Foster DL, Padmanabhan V 2004 Fetal programming: developmental progression of estradiol positive feedback defects in prenatally T-treated sheep. Biol Reprod. 120:121
Desai M, Hales CN 1997 Role of fetal and infant growth in programming metabolism in later life. Biol Rev. 72:329–348
Yu WH 1989 Administration of testosterone attenuates neuronal loss following axotomy in the brain-stem motor nuclei of female rats. J Neurosci. 9:3908–3914
Rasika S, Alvarez-Buylla A, Nottebohm F 1999 BDNF mediates the effects of testosterone on the survival of new neurons in an adult brain. Neuron. 22:53–62
Molter-Gerard C, Fontaine J, Guerin S, Taragnat C 1999 Differential regulation of the gonadotropin storage pattern by gonadotropin releasing hormone pulse frequency in the ewe. Biol Reprod. 60:1224–1230
Thomas SG. Clarke IJ 1997 The positive feedback action of estrogen mobilizes LH-containing, but not FSH-containing secretory granules in ovine gonadotropes. Endocrinology 138:1347–1350
Marshall JC, Dalkin AC, Haisenleder DJ, Griffin ML, Kelch RP 1992 GnRH pulses-the regulators of human reproduction. Trans Am Clin Climatol Assoc. 104:31–46
Bilezikjian LM, Blount AL, Corrigan AZ, Leal A, Chen Y, Vale WW 2001 Actions of activins, inhibins and follistatins: implications in anterior pituitary function. Clin Exp Pharmacol Physiol. 28:244–248
DePaolo LV, Bicsak TA, Erickson GF, Shimasaki S, Ling N 1991 Follistatin and activin: a potential intrinsic regulatory system within diverse tissues. Proc Soc Exp Biol Med. [ Erratum (1992) 200:447] 198:500–512
Phillips DJ, Woodruff TK 2004 Inhibin: actions and signalling. Growth Factors 22:13–18
Padmanabhan V, Lee JS, Beitins IZ 1998 Follicle-stimulating isohormones: regulation and biological significance. In: Thatcher WW, Inskeep EK, Niswender GD, Doberska C, eds. Reproduction in domestic ruminants IV. J Reprod Fertil Suppl 54 London: Cambridge; 87–99
Fanworth PG 1995 Gonadotropin secretion revisited. How many ways can FSH leave a gonadotroph. J Endocrinol. 145:387–395
Muyan M, Ryzmkiewicz DM, Boime I 1994 Secretion of lutropin and follitropin from transfected GH3 cells: evidence for separate secretory pathways. Mol Endocrinol. 8:1789–1797
Padmanabhan V, Battaglia D, Brown MB, Karsch FJ, Lee JS, Pan W, Phillips DJ, VanCleeff J 2002 Neuroendocrine control of follicle-stimulating hormone secretion: II. Is follistatin-induced suppression of follicle-stimulating hormone secretion mediated via changes in activin availability and does it involve changes in GnRH secretion Biol Reprod. 66:1395–1402
Padmanabhan V, Brown M, Dahl G, Evans NP, Karsch FJ, Mauger DT, Neill JD, Van Cleeff J 2003 Neuroendocrine control of FSH secretion: III. Is there a GnRH-independent component of episodic FSH secretion in ovariectomized and luteal phase ewes Endocrinology 144:1380–1392
Marshall JC, Eagleson CA 1999 Neuroendocrine aspects of polycystic ovary syndrome. Endocrinol Metab Clin North Am. 28:295–324
Hall JE, Taylor AE, Hayes FJ Crowley Jr WF 1998 Insights into hypothalamic-pituitary dysfunction in polycystic ovary syndrome. J Endocrinol Invest. 21:602–611
Rosenfield RL 1997 Is polycystic ovary syndrome a neuroendocrine or an ovarian disorder Clin Endocrinol (Oxf). 47:423–424
McCartney CR, Eagleson CA, Marshall JC 2002 Regulation of gonadotropin secretion: implications for polycystic ovary syndrome. Semin Reprod Med. 20:317–326
Cardone VS 2003 GnRH antagonists for treatment of polycystic ovarian syndrome. Fertil Steril. 80(Suppl 1):S25–S31; discussion S32–S34
Lubin V, Charbonnel B, Bouchard P 1998 The use of gonadotrophin-releasing hormone antagonists in polycystic ovarian disease. Baill Clin Obstet Gynaecol. 12:607–618
Hayes FJ, Taylor AE, Martin KA, Hall JE 1998 Use of a gonadotropin-releasing hormone antagonist as a physiologic probe in polycystic ovary syndrome: assessment of neuroendocrine and androgen dynamics. J Clin Endocrinol Metab. 83:2343–2349(Hirendra N. Sarma, Mohan )