Developmental Programming: Postnatal Steroids Complete Prenatal Steroid Actions to Differentially Organize the GnRH Surge Mechanism and Reproductive Behavior in Female Sheep

Abstract

In female sheep, estradiol (E₂) stimulates the preovulatory GnRH/LH surge and receptive behavior, whereas progesterone blocks these effects. Prenatal exposure to testosterone disrupts both the positive feedback action of E₂ and sexual behavior although the mechanisms remain unknown. The current study tested the hypothesis that both prenatal and postnatal steroids are required to organize the surge and sex differences in reproductive behavior. Our approach was to characterize the LH surge and mating behavior in prenatally untreated (Control) and testosterone-treated (T) female sheep subsequently exposed to one of three postnatal steroid manipulations: endogenous E₂, excess E₂ from a chronic implant, or no E₂ due to neonatal ovariectomy (OVX). All females were then perfused at the time of the expected surge and brains processed for estrogen receptor and Fos immunoreactivity. None of the T females exposed postnatally to E₂ exhibited an E₂-induced LH surge, but a surge was produced in five of six T/OVX and all Control females. No surges were produced when progesterone was administered concomitantly with E₂. All Control females were mounted by males, but significantly fewer T females were mounted by a male, including the T/OVX females that exhibited LH surges. The percentage of estrogen receptor neurons containing Fos was significantly influenced in a brain region-, developmental stage-, and steroid-specific fashion by testosterone and E₂ treatments. These findings support the hypothesis that the feedback controls of the GnRH surge are sensitive to programming by prenatal and postnatal steroids in a precocial species.

In female sheep, the ovarian steroids estradiol (E₂) and progesterone (P₄) provide feedback control of reproductive neuroendocrine function, with each hormone affecting both the tonic and surge modes of GnRH and LH secretion (1). Negative feedback actions of E₂ and P₄ control both modes of gonadotropin secretion (2, 3). The positive feedback action of E₂ during the late follicular phase initiates the preovulatory GnRH/LH surge (4); this action can be blocked by luteal phase concentrations of P₄ (5). In males, sensitivity to these steroid feedback controls is reduced due to the organizational actions of testosterone and its metabolites during development (1). Thus, differentiation of reproductive neuroendocrine function in normal adult males is classically characterized by the absence of an LH surge in response to follicular phase concentrations of E₂ (6), although the mechanisms underlying the absence of a surge response in males have not been elucidated.

Testosterone and E₂ have organizational actions during late gestation and early postnatal life in altricial species such as the mouse and rat (6–9). In precocial species such as sheep, the hypothalamus is sensitive to the organizing effects of steroids earlier, during midgestation (10, 11). However, the severity of the effects of prenatal testosterone on E₂-positive feedback in sheep may depend on the pattern of steroid exposure during postnatal life. The preovulatory LH surge is ablated in prenatally testosterone-treated female sheep that are neonatally ovariectomized and given exogenous E₂ via a constant-release implant (ie, a neuroendocrine model) (11, 12). In contrast, testosterone-treated females that remain gonadally intact exhibit an LH surge (albeit atypical) and ovarian cyclicity at the expected time of puberty, but cyclicity gradually deteriorates during the first 2 years of life (13–15). Because there are quantitative and qualitative differences in the pattern of postnatal E₂ exposure in these two groups of studies, the contrasting findings raise the possibility that postnatal exposure to E₂ plays a role in fully defeminizing the positive feedback action of E₂ after prenatal testosterone exposure. We recently reported that exposure of ovary-intact prenatally testosterone-treated sheep to excess E₂ during the first year of postnatal life accelerated the ablation of the LH surge (16). Thus, we have evidence to suggest that there is a period other than midgestation during which E₂ can organize reproductive neuroendocrine function.

In the male, neuroendocrine and behavioral differentiation is dependent on prenatal testosterone or its metabolites (6, 7). The precise reason for the lack of an LH surge in male sheep is not clear and may differ from rodents and primates in which E₂ is able to produce an LH surge in neonatally castrated males (17). One possibility is that the developing sheep testis is more steroidogenic than the immature ovary (18–20), and sexual differentiation of the developing hypothalamus is completed during gestation. A second possibility is that, as in rodents, postnatal steroids are necessary to fully differentiate the surge mechanism. Newborn male sheep are exposed to significant amounts of testosterone (21) and, presumably E₂ via peripheral aromatization, and this normal postnatal exposure could continue to organize the neural circuits contributing to the loss of a surge mechanism in males. However, this possibility has not been explored.

The neural circuits that control male and female reproductive behaviors are also highly sensitive to the organizational actions of testicular steroids in a time- and duration-dependent manner (10, 22–24), and it is possible that postnatal steroid exposure may program behavioral differentiation as well. For example, female sheep exposed to testosterone throughout the recognized critical period for sexual differentiation (gestational day [GD] 30-90 of the 147-d gestation) have masculinized genitalia (25) and exhibit less female-typical behavior than do females subject to shorter duration exposures (eg, GD 60-90) (24). Importantly, testosterone exposure from GD 60-90 also alters behavior, but without virilizing the genitalia (11, 24), ablating the LH surge response to E₂ (11), or causing major disruptions in ovarian cyclicity (26). Therefore, sexual behavior in testosterone-treated females is arguably a highly sensitive indicator of programming by prenatal steroids, and despite the overt defeminization and masculinization of reproductive behavior by prenatal testosterone in both the ovary-intact and neuroendocrine models, the possibility of a postnatal contribution to sexual differentiation of behavior has not been examined directly.

Despite the extensive evidence supporting the hypothesis that E₂ feedback is disrupted by prenatal testosterone exposure in females, the precise neural mechanisms involved and their modulation by postnatal steroids remain unclear. Conceivably, the disruption of the surge by testosterone in females involves estrogen receptor (ER) neurons of the preoptic area (POA), arcuate nucleus (ARC), and ventromedial nucleus of the hypothalamus (VMN), because these regions are recognized as important centers for regulating reproductive neuroendocrine function and behavior (27–30). This hypothesis is supported by the recent observation that activation of somatostatin neurons in the VMN and tachykinin neurons in the ARC is suppressed after prenatal testosterone exposure (30). However, whether it is ER neurons specifically in these brain regions that are affected is not known.

Given the lack of information regarding extended or additional critical periods for sexual differentiation of the positive feedback action of E₂, reproductive behavior, and the neural circuits controlling reproductive function, the present study was performed to test the following hypotheses: 1) that postnatal E₂ is required to fully defeminize the surge mechanism in females exposed prenatally to testosterone; 2) that postnatal testicular steroids are necessary for differentiation of the surge mechanism in the male; 3) that the E₂ and P₄ feedback controls of the surge mechanism are programmed similarly (based on our overall working hypothesis (1), in which we predicted that the ability of P₄ to block the surge would be functional if postnatal E₂ is required to defeminize this feedback mechanism); 4) that postnatal E₂ exposure is necessary to facilitate the effects of prenatal testosterone on mating behavior; and 5) that prenatal testosterone and postnatal E₂ together are necessary to fully defeminize E₂-induced neural activation at the onset of the LH surge.

Materials and Methods

Animals and treatments

The University Committee for the Use and Care of Animals at the University of Michigan approved the protocol for all animal care and experimental procedures. Housing, feeding, husbandry, and prenatal testosterone treatment of female Suffolk sheep have been previously described (16, 31). Mated females, stratified by body weight (BW) and body condition score (32), were randomly assigned to an untreated (Control) or prenatal testosterone treatment (T) group such that the distributions of BW and body condition score did not differ between groups. Prenatal testosterone treatment (100 mg testosterone propionate [Sigma-Aldrich, St Louis, Missouri]) was administered im twice weekly from GD 30-90. This treatment regimen produces circulating levels of testosterone in the adult and fetal male ranges in the mother and female fetus, respectively (33).

Female lambs (22 Control and 18 T) were born between March 25 and April 11. At 2 weeks of age, lambs were stratified by BW and randomly assigned to one of three postnatal treatment groups, with twin females (Control: two sets; T: two sets) being assigned to different postnatal treatments. The groups differed with respect to their source of E₂ exposure: 1) ovary-intact control females exposed to endogenous ovarian E₂ (O), 2) a chronic E₂ treatment that exposed females to ovarian E₂ plus E₂ from a sc implant inserted at 2 weeks of age (E), and 3) neonatal ovariectomy at 2-3 weeks of age in which females matured in the absence of gonadal steroids (ovariectomy [OVX]). The E₂ implant was a small capsule made of SILASTIC tubing (outer diameter = 0.46 cm, inner diameter = 0.34 cm; Dow Corning Corp, Midland, Michigan) packed with a 3-cm column of crystalline 17β-E₂ (Sigma), and sealed with SILASTIC adhesive Type A (Dow Corning Corp). Implants of this size produce low physiologic circulating concentrations of E₂ (∼3–5 pg/mL) (34). Ovaries were removed via a midline abdominal incision under ketamine (20 mg/kg, im) and xylazine (0.1–0.2 mg/kg, im) anesthesia. The six experimental groups are designated by paired terms (prenatal/postnatal) representing the prenatal treatment and the principle source of E₂ during postnatal life (Control/O, n = 7; Control/E, n = 8; Control/OVX, n = 7; T/O, n = 5; T/E, n = 7; T/OVX, n = 6). The postpubertal LH surge response in the Control/O, Control/E, T/O, and T/E groups was tested at 14 months of age during the first seasonal anestrus, and those results have been reported previously (16). For the present study, all six groups were tested between 19 and 22 months of age for an LH surge response, blockade of the surge by P₄, and sexual behavior (see Figure 1).

Figure 1.

Schematic Representation of the Series of Experiments Performed to Study the Role of Postnatal Steroids in Organizing the Feedback Controls of the LH Surge, Sexual Behavior, and Neural Responses at the Onset of the LH Surge in Untreated and Prenatally Testosterone-Treated Female Sheep.

An additional group of spring-born, prenatally untreated lambs (8 females and 18 males) was used to examine the possibility that the LH surge is not fully defeminized in newborn male sheep. Females were neonatally ovariectomized, and nine males were castrated at approximately 2 weeks of age by the routine agricultural practice of banding of the scrotal stalk. A 3-cm E₂ implant was placed in these males and all females at the time of gonadectomy. The nine remaining males were surgically castrated under ketamine (10 mg/kg) and xylazine (0.1 mg/kg) anesthesia within 2 days of birth and received no additional steroid treatment.

Role of postnatal E₂ in defeminizing the surge mechanism in females exposed prenatally to testosterone

To test the hypothesis that postnatal E₂ is necessary to defeminize the GnRH surge mechanism, the LH surge response was evaluated in females at 19 months of age. The endocrine status of all females was standardized by removing the ovaries and E₂ implants (when necessary) from Control/O, Control/E, T/O, and T/E females 1 month before the surge induction experiment. At the start of the experiment, an artificial luteal phase was created by sc insertion of a SILASTIC capsule containing a 1-cm column of E₂ and two controlled-release P₄ implants (CIDR; DEC International, Hamilton, New Zealand). The T-shaped CIDRs, designed for intravaginal use, were modified for sc insertion by cutting off the plastic retrieval tail and tying the side extensions together to form a cylindrical implant approximately 9 cm in length (35). Luteal phase implants were removed after 10 days, and 24 hours later, four 3-cm E₂ implants were inserted sc to produce late follicular phase concentrations of E₂ (36). Blood samples were collected by jugular venipuncture every 2 hours from 4 hours before until 72 hours after the E₂ implants were inserted; additional blood samples for measurement of steroid concentrations were collected immediately before inserting implants, and after collection of the last sample at 72 hours. All implants were removed at the conclusion of sample collection. Plasma was separated and stored at −20°C until assayed for LH or steroids.

Role of postnatal testicular steroids in defeminizing the surge mechanism in the male

The LH surge response in neonatally gonadectomized males and females was tested at 15 weeks of age following the procedure described above, except that the artificial luteal phase was maintained for 8 days.

Programming of the P₄ feedback control of the surge mechanism

A second surge induction was initiated 2 weeks after the first to examine the ability of P₄ to block the E₂-induced surge. The implant and sampling procedures were identical to the first surge induction except that the P₄ implants were not removed before inserting the four E₂ implants. To determine that absence of an LH surge was not due to insufficient stores of LH, pituitary responsiveness to GnRH was tested at t = 74 hours, before removing the steroid implants. A bolus iv injection of GnRH (Sigma, 10 ng/kg) was administered, and blood samples were collected at 10-minute intervals from immediately before until 40 minutes after the injection. This dose of GnRH is approximately twice the amount previously used to produce GnRH concentrations in the hypophyseal portal circulation that are similar to peak endogenous concentrations (3).

Role of postnatal E₂ in mediating the effects of prenatal testosterone on female mating behavior

The effects of postnatal E₂ exposure on mating behavior were studied during the first surge induction. Mating behavior was observed in a subset of individual ewes (Control/O = 6, Control/E = 6, Control/OVX = 7, T/O = 3, T/E = 5, T/OVX = 6) approximately 24 hours after insertion of the four E₂ implants, when females were considered to be in estrus. The female was placed in a 30-m² room with a vasectomized adult male, and all interactions between the female and male were videotaped for 5 minutes. Videotapes were analyzed for the frequency of behaviors previously described by Roberts et al (24), including all proceptive (eg, approach, passby, look back, ear wiggle, and tail wag) and receptive (ie, standing still upon male approach) mating behaviors exhibited by the female, and mounting by the male.

Role of prenatal testosterone and postnatal E₂ in defeminizing E₂-induced neural activation at surge onset

One month after the second surge induction (to test the P₄ block of an E₂-induced surge), a third LH surge induction was initiated during which all female sheep were euthanized between 22 and 23 hours after insertion of the four E₂ implants, and the brains were perfused with 4% paraformaldehyde in 0.1 M PBS (pH 7.4) (37). Euthanasia was timed to occur at the expected onset of the LH surge, based on LH profiles from the first surge induction experiment. Free-floating sections (55 μm) were processed for dual immunofluorescence as described previously (37) using antibodies to ER-α (mouse monoclonal, 1D5, diluted 1:100; DAKO Corp., Carpinteria, California) and c-Fos (rabbit polyclonal, SC-253, diluted 1:500; Santa Cruz Biotechnology, Santa Cruz, California). Secondary antibodies (diluted 1:200) used were: donkey antirabbit IgG conjugate to Alexa 488 and donkey antimouse IgG conjugated to Alexa 594 (Jackson Immunologicals, West Chester, Pennsylvania). For analysis, three tissue sections (∼200-300 μm apart) from each of three brain regions (POA, ARC, and VMN) were selected according to standardized drawings (38). For each section, the number of ER, Fos, and ER+Fos (dual-labeled) cells was determined by an investigator who was blind to the treatment associated with the tissue.

Hormone assays

Concentrations of LH were measured in duplicate 20-200 μL aliquots of plasma using a modified (39, 40) RIA developed by Niswender et al. (41). Assay sensitivity, defined as 2 SDs from the buffer control, averaged 0.51 ± 0.03 ng/mL NIH ovine (o)LH-S12 for 200 μL plasma (n = 23 assays). Intraassay coefficients of variation (CV) calculated from six replicates of three standard sera binding at 29%, 51%, and 84% of buffer controls averaged 2.9%, 3.4%, and 16.1%, respectively. Interassay CVs for the same standards were 2.7%, 2.6%, and 7.6%. Concentrations of E₂ were measured in duplicate diethyl ether extracts of 200 μL of plasma using a commercially available E₂ magnetic immunoassay kit (Adaltis Italia, Bologna, Italy) modified for use in the sheep as previously described (42, 43). Sensitivity of the assay averaged 0.51 ± 0.09 pg/mL for 200 μL of plasma (n = 2 assays). Mean intra- and interassay CVs were 4.0% and 2.2%. Concentrations of P₄ were measured in duplicate 100-μL aliquots of unextracted plasma using a commercially available RIA kit (Coat-A-Count P4; Diagnostic Products Corp, Los Angeles, California) previously validated for use in sheep (44). Sensitivity of this assay averaged 0.18 ± 0.06 ng/mL for 100 μL of plasma (n = 2 assays). Mean intra- and interassay CVs were 1.7% and 5.1%. The limit of assay sensitivity was assigned to those samples in which the concentration of LH, E₂, or P₄ was below the sensitivity of the assay.

Data and statistical analyses

Using criteria previously described (13, 45), a surge was defined as an increase in concentrations of LH that exceeded baseline by 2 SDs for at least 8 hours, with peak concentrations of LH exceeding twice the average concentrations measured in samples collected before insertion of the E₂ implants. The onset of the surge was the time of the first sample that exceeded baseline by 2 SDs, duration of the surge was the time between onset and the last sample that exceeded baseline by 2 SDs, and peak amplitude was the maximum concentration of LH measured during the surge.

The proportion of individuals exhibiting an LH surge response was analyzed using a χ² statistic. The time to surge onset, duration of the surge, peak amplitude, baseline (nonsurge) concentrations of LH during the E₂-induced surge, and mean concentrations of LH during the P₄ block of the surge were compared among treatment groups using ANOVA, including a Bonferroni correction for multiple comparisons. Concentrations of LH after administration of GnRH were analyzed by repeated-measures ANOVA, and compared with pre-GnRH values within a group using Scheffe's multiple comparison. The proportion of females exhibiting female-typical behavior or being mounted by a male was analyzed using Fisher's exact test. Frequency of female-typical behaviors was compared between groups with ANOVA and Tukey's post hoc test for group differences. Neuroanatomic data (cell counts and percentage of dual-labeled cells) were analyzed by two-way ANOVA, with post hoc comparisons made using Tukey's multiple comparison test. All data are expressed as mean ± SEM; statistical significance for all analyses was defined as P < .05.

Results

Defeminization of the surge mechanism

An LH surge was evident in all Control/O, Control/E, Control/OVX females, and in five of six T/OVX females (Figures 2 and 3A). In contrast, none of the T/O or T/E females exhibited an LH surge. Circulating concentrations of E₂ during the surge induction were similar in all groups, and P₄ concentrations were below 1 ng/mL when the E₂ implants were inserted (Figure 2). There was no effect of postnatal treatment on surge onset, duration, or peak amplitude in Control females (Figure 3, B–D). Peak amplitude of the surge was lower in T/OVX females compared with Control/O females (Figure 3D) but did not differ between T/OVX and Control/E or Control/OVX females. The surge lasted longer in T/OVX females than in any other group. Mean baseline concentrations of LH were higher in Control/OVX females than in Control/O and Control/E females (Figure 3E), but there were no differences among any other groups.

Figure 2.

Profiles of LH Secretion In Response To Exogenous E₂ during an Artificial Ovarian Cycle In Representative Control (left) and Prenatally T-Treated (right) Female Sheep Exposed Postnatally to Endogenous Ovarian E₂ (Ovary, top panel), Excess E₂ from a Chronic sc Implant (+ E, middle panel), or no E due to Neonatal Ovariectomy (Ovx, bottom panel). Bar graphs to the right of each set of LH profiles show mean (+SEM) concentrations of E₂ (white bars) measured immediately before and 72 hours after insertion of E₂ implants, and mean concentrations of P₄ (black bars) measured immediately before insertion of E₂ implants (24 h after removal of P₄ implants).

Figure 3.

Summary of LH Surge Responses in Control Female Sheep Exposed Postnatally to Endogenous Ovarian E₂ (C/O, white bars), Excess E₂ from a Chronic sc Implant (C/E, gray bars), or no E₂ due to Neonatal Ovariectomy (C/OVX, black bars), and Prenatally T-Treated Females with the Same Postnatal Exposures to E₂ (T/O, white striped bars; T/E, gray striped bars; T/OVX, black striped bars). A, Percent of females in each group that exhibited an LH surge response. The numbers on each bar represent the number of responsive females expressed as a proportion of the total number of females in the group. B, Mean (+SEM) number of hours between insertion of E₂ implants at t = 0 hours and the onset of the LH surge. C, Mean (+SEM) duration of the LH surge. D, Mean (+SEM) amplitude of the peak of the LH surge. E, Mean (+ SEM) concentration of LH measured in samples collected from 2 to 72 hours, excluding samples collected during a surge. Asterisks indicate significant differences in the surge response relative to control females. Different letters denote significant differences (P < .05) among groups.

No surges were detected in any of the males, regardless of postnatal steroid exposure. By contrast, all females produced an LH surge (data not shown). Time to the onset of the surge was 18.9 ± 0.9 hours, maximum concentration of LH during the surge was 110 ± 9.2 ng/mL, and duration of the surge was 8.7 ± 0.8 hours.

Programming of the P₄ feedback control

No LH surges were produced in response to the E₂ stimulus in the presence of luteal phase concentrations of P_4. (Figure 4A). Concentrations of E₂ produced by the implants (Figure 4A, bar graphs) were similar to those measured during the previous surge induction (Figure 3), and concentrations of P₄ at t = 72 hours exceeded 3.5 ng/mL. Mean concentration of LH from 2 to 74 hours after E₂ was greater (P < .05) in Control/OVX females than in Control/O or Control/E females (Figure 4B, left). Mean LH in T-treated females was similar to that in the Control/OVX group and did not differ by postnatal treatment. The GnRH challenge produced a significant increase in concentrations of LH within 10 minutes in all females, with no difference in the maximum amount of LH released in response to GnRH (Figure 4B, right).

Figure 4.

A. Profiles of LH Secretion during a Surge Block Experiment And Concentrations of LH Measured following an iv Injection of GnRH at t = 74 h (insets) in Representative Control (left) and Prenatally T-treated (right) Female Sheep Exposed Postnatally to Endogenous Ovarian E₂ (Ovary, top panel), Excess E₂ from a Chronic sc Implant (+ E, middle panel), or no E₂ due to Neonatal Ovariectomy (Ovx, bottom panel). Implants providing luteal phase concentrations of P₄ were left in place during an attempt to induce an LH surge with E₂ implants inserted at t = 0 hours. Bar graphs to the right of each LH profile show mean (+SEM) concentrations of E₂ (white bars) and P₄ (black bars) measured at t = 0 hours and t = 72 hours. B, Mean (+ SEM) concentration of LH measured from 2–72 hours (left) and maximum concentration of LH measured from 10-40 minutes after administration of GnRH (right). Different letters denote significant differences among groups.

Defeminization of mating behavior

During the individual behavior tests, look-back behavior was the only proceptive female behavior observed in every group. Look-back behavior was observed in every C/O, C/OVX, and T/OVX female, 80% of C/E females, 40% of T/E females, and 33% of T/O females. However, the mean number of displays was lower (P < .05) in prenatally T-treated females, with no effect of postnatal treatment in either Control or T females (Figure 5, top). All Control females were mounted by the male during the test period, but a significantly lower percent of prenatally T-treated females were mounted (Figure 5, bottom; P < .05), with no effect due to postnatal treatment.

Figure 5.

Summary of Sexual Behavior Observed in Control/O, Control/E, Control/OVX, T/O, T/E, and T/OVX Female Sheep during Estrus Induced with Hormone Implants. Top, Mean (+SEM) number of look-back behaviors observed when a female was paired with a male for 5 minutes. Bottom, Percent of female sheep that were mounted by a male during a 5-minute behavior test. Different letters denote significant differences (P < .05) among groups.

Defeminization of E₂-induced neural activation

Brain tissue from 14 Control (4 C/O, 4 C/E, 6 C/OVX) and 16 T (4 T/O, 6 T/E, 6 T/OVX) was analyzed; tissue from the remaining animals was rendered unusable by freezing without adequate cryoprotection. Representative sections from a C/OVX brain illustrating ER and Fos nuclear immunolabeling in the ARC (left), POA (middle), and VMN (right) are shown in Figure 6. Results of analysis of single and dual-labeled neurons are presented below and in Figure 7. Because no statistical differences due to postnatal treatment were observed for behavior, endocrine profiles (see Figures 3–5), or anatomic data (not shown) between ovary-intact and ovary-intact plus postnatal E₂ groups, the data for these C and T groups were combined (hereinafter referred to as C/O-E and T/O-E) for all subsequent anatomic analyses. All values represent the mean (±SEM) number of single- or dual-labeled cells per section.

Figure 6.

Representative Photomicrographs Illustrating ER (red) and Fos (green) Immunofluorescent Labeling of the POA, ARC, and VMN Regions in a C/OVX Sheep at Low-Power Magnification (A–C) and Higher Magnification (D–I). Arrows indicate examples of dual-labeled cells. 3V, third ventricle. Scale bar = 100 μm (A–C), 50 μm (D–I).

Figure 7.

Summary of Mean (+SEM) Percent of ER Neurons Expressing Fos in the POA (A), ARC (B), and VMN (C) at the Predicted Time of the LH Surge in Control and T-Treated Female Sheep with (C/O-E, T/O-E) and without (C/OVX, T/OVX) Exposure to Postnatal E₂ (see Materials and Methods for additional details). Asterisks denote significant differences relative to postnatal treatment. Numbers of animals used for analysis are indicated by the boxes in panel A.

ER neurons.

No significant main effect of prenatal or postnatal treatment on the number of ER neurons was observed in the POA (C/OVX = 1614 ± 256; C/O-E = 1637 ± 223; T/OVX = 1593 ± 277; T/O-E = 1329 ± 202) or the ARC (C/OVX = 478 ± 17; C/O-E = 535 ± 46; T/OVX = 453 ± 73; T/O-E = 401 ± 68; all comparisons P > .05). For the VMN, only a main effect of prenatal treatment was observed for number of ER neurons (C/OVX = 1726 ± 162; C/O-E = 1650 ± 181; T/OVX = 1307 ± 179; T/O-E = 1347 ± 160; control vs T − F_1,26 = 4.12; P = .05).

Fos neurons.

No significant main effect of prenatal or postnatal treatment on the number of Fos neurons was observed in the POA (C/OVX = 334 ± 106; C/O-E = 253 ± 66; T/OVX = 434 ± 166; T/O-E = 294 ± 90) or ARC (C/OVX = 80 ± 10; C/O-E = 83 ± 12; T/OVX = 75 ± 20; T/O-E = 35 ± 14; all comparisons, P > .05). For the VMN, only a main effect of postnatal treatment was observed for numbers of Fos neurons (C/OVX = 1288 ± 144; C/O-E = 576 ± 184; T/OVX = 885 ± 160; T/O-E = 610 ± 176; OVX vs O-E − F_1,26 = 7.5, P < .01).

ER/Fos colocalization.

Main effects of prenatal (F_1,26 = 24; P < .001) and postnatal (F_1,26 = 21.85, P < .0001) treatments on the percentage of ER cells colocalizing Fos in the POA at the predicted onset of the LH surge were observed. A significant interaction was also observed. (F_1,26 = 6.03; P < .05). For the ARC, only a main effect of prenatal treatment on the number of dual-labeled cells was found (F_1,26 = 17.39; P < .001); no significant interaction was found. Similar to the POA, ER neurons colocalizing Fos in the VMN were influenced by both prenatal (F_1,26 = 42.92; P < .0001) and postnatal (F_1,26 = 39.38; P < .0001) treatments; however, no significant interaction was found.

Discussion

The primary goal of the present study was to determine whether postnatal steroid exposure contributes to the prenatal programming of sexual differentiation. Based on neuroendocrine and neuroanatomic endpoints, our finding that the GnRH surge mechanism is not completely defeminized in prenatally testosterone-treated females without exposure to ovarian steroids during postnatal life supports the hypothesis that the developmental period for programming the positive feedback action of E₂ in sheep extends beyond GD 30–90. Furthermore, combined with our previous results demonstrating that excess E₂ during postnatal life causes further deterioration of the surge mechanism and ovarian cyclicity in prenatally testosterone-treated females (14, 16, 46), these findings strongly support the existence of an extended (or additional) critical period during which neuroendocrine circuits are sensitive to the organizing actions of steroids in a precocial species. The current findings also support the explanation that the different neuroendocrine consequences of prenatal testosterone treatment in the neuroendocrine and ovary-intact sheep models reflect differences in the nature of postnatal steroid exposure.

Steroid feedback controls of neuroendocrine function

Positive feedback actions of E₂ were sufficient to generate a surge in all Control females and in T females that were ovariectomized postnatally, but these results should not be interpreted as evidence that the gonadotropin surge in females is relatively refractory to prenatal steroid programming. On the contrary, our observations that T females exposed to E₂ postnatally failed to produce a surge (present study) or exhibited decrements in surge quality (16) support the hypothesis that additional postnatal events are required to fully defeminize the positive feedback actions of E₂ in T females or, alternatively, to unmask a prenatally programmed event.

This argument is clearly insufficient to explain the absence (ie, defeminization) of a surge response in males because no surge was evident in males that were castrated within 2 days of birth. It has been established that castrated adult male sheep do not exhibit an LH surge response to E₂ (47), but the absence of this response in neonatally castrated male sheep is somewhat unexpected given the presence of the response in male rats castrated at birth (48). Nonetheless, our findings are consistent with a precocial species exhibiting accelerated prenatal development (49). The sheep testis develops its steroidogenic capacity after approximately one-quarter of gestation (20), similar to the onset of testicular steroidogenesis in the developing human (50) and nonhuman primates (51). However, neural development at birth is more advanced in the precocial sheep than in altricial species, thereby moving the window of responsiveness to experimental manipulation of organizing factors into the prenatal period. The neural circuits underlying a surge response in both males and females could remain sensitive to organization by steroids beyond the established critical period of GD 30–90, but steroids produced by the fetal testes between GD 90 and birth (or postnatal day 2, when males in this study were castrated) complete the process of defeminizing the surge mechanism in the male. Because there is no sex difference in circulating concentrations of either testosterone or E₂ at GD 90, or testosterone at GD 140 (33), we suggest that E₂, which is higher in males than in females at GD 140, is responsible for full defeminization between GD 90 and birth at GD 147. However, additional experiments using aromatase inhibitors during late gestation are needed to confirm or refute this possibility. In females treated with testosterone from GD 30–90, it is possible that the endogenous steroids present between GD 90 and birth are insufficient to fully defeminize the surge mechanism during prenatal development. Thus, postnatal steroids continue to organize reproductive neuroendocrine function in T-treated females, but not in males.

Consistent with the earlier prediction that both E₂ and P₄ feedback controls are programmed similarly (1), we found that P₄ blocked the surge in all Control and T/OVX sheep that had previously expressed an LH surge. However, because T/O and T/E females did not exhibit LH surges when tested at 19 months, additional studies of younger females are needed to determine whether the ability of P₄ to block an E₂-induced surge exists early in reproductive life. If the P₄ block of the surge is further programmed by postnatal E₂, we predict that this inhibitory action of P₄ would be absent or reduced in young T females exhibiting a surge response.

Behavior

The defeminization of mating behavior in T females in this study is consistent with previous reports (10, 24). Importantly, the finding that T/OVX females exhibited a female-like surge, but only one of six females in this group was mounted, suggests that prenatal testosterone alone is sufficient to program sex differences in reproductive behavior, but not the surge mechanism. It is possible that the masculinized external genitalia of T females reduced their attractiveness to the male, although virilization of T females did not completely eliminate male mounting behavior. Postnatal E₂ treatment had no significant effect on look-back behavior or the percentage of T females mounted by the male, and because mounting of C/OVX females does not differ from mounting of Control females with ovaries (with or without E implants), postnatal E exposure does not appear to be critical for mating behavior. Indeed, although not systematically measured, we noted that C/OVX females were most responsive to the presence of males and their odors.

Neuroanatomy

The main effects of prenatal treatment to suppress activation of ER neurons in all three brain regions examined supports the idea that T programs adult neuroendocrine function through direct actions on these important neuroendocrine cell populations. Neurons of the VMN appear to be particularly important for both the generation of a LH surge and behavior in response to E₂ (27). Together with a significant reduction in Fos activation in the POA of T/OVX females relative to Control females, this may relate to the reduction in proceptive behavior in T/OVX females relative to Controls, suggesting that the appropriate programming of both brain regions is necessary for expression of the full complement of sexual responses. Consistent with this hypothesis are studies showing that disruption of the VMN-POA connections greatly suppresses or eliminates the surge in sheep (52, 53). The expression of an LH surge in T/OVX ewes with lower numbers of activated neurons above a minimal threshold could represent a partial defeminization of the POA induced by prenatal testosterone excess. Importantly, the addition of postnatal E appears sufficient to further differentiate (masculinize) the surge mechanism by lowering the numbers of activated neurons below this threshold in the VMN, rendering the surge inoperable.

We were surprised to find no additional action of postnatal E on activation of ARC ER neurons because the ARC is important source of fibers in the external layer of the median eminence where GnRH fibers terminate. For example, both ER neurons (54) and Kisspeptin neurons of the ARC send projections to the external zone of the median eminence in sheep (55). Because most Kisspeptin neurons of the ovine ARC also contain ER (56), this population is likely a direct target of prenatal effects of T, but this remains to be determined. Why postnatal treatment caused no significant additional reduction in ER/Fos is unclear, although two possibilities are likely. First, as was the case for the POA, a threshold may be required for the full defeminization of ARC neurons to occur. Second, ARC neurons may be involved in other aspects of neuroendocrine control that are impacted by prenatal T treatment, such as steroid negative feedback (11, 16, 31, 57); however, this remains speculative because we did not directly address this in the current study.

The ability of prenatal testosterone to suppress the activation of ER neurons in our study is similar to that found recently for somatostatin neurons of the VMN (58). Although approximately 30% of VMN somatostatin neurons contain ER (58), whether the somatostatin-ER population is the sole target of organization by prenatal testosterone remains to be determined. Regardless of neuronal phenotype, it appears that prenatal testosterone treatment disrupts the LH surge mechanism in part by reducing the ability of VMN ER neurons to be activated by E₂. This defeminization appears to require continued postnatal E₂ exposure, and the significantly greater number of ER neurons colocalizing Fos in T/OVX females that expressed normal surges supports this hypothesis.

Summary

The present findings expand our understanding of the critical developmental periods for sexual differentiation to include a postnatal period of sensitivity to E₂ in the female sheep, a model precocial species. We hypothesize that the function of this extended or additional critical period is to facilitate the complete defeminization of the female reproductive neuroendocrine system exposed to testosterone in utero. This is consistent with the concept that the default condition is female and recruitment of additional mechanisms is required to produce a male phenotype (59). Furthermore, there are numerous studies that demonstrate seasonal neuronal plasticity in the adult sheep brain (60, 61), and the present study suggests that the postnatal brain retains a certain amount of plasticity to modify reproductive neuroendocrine function. Of particular interest are our findings that in prenatally untreated females, postnatal OVX and the subsequent absence of gonadal steroids during postnatal life resulted in patterns of tonic LH secretion (Figures 3E and 4B) and neural activation (Figure 7, A and C) that were different from the responses in ovary-intact females. Although our hypothesis is clearly based on results from an experimental treatment, its physiologic relevance extends to considering the potential consequences of inappropriate steroid exposure of exogenous or pathologic origin, or the actions of endocrine disruptors, in developing individuals. An extended period of differentiation underscores the importance of recognizing the effects of unwanted prenatal and postnatal steroid exposure on adult sexual function, particularly when detrimental effects are not evident until after puberty.