The role of the medial and central amygdala in stress-induced suppression of pulsatile LH secretion in female rats.

Stress exerts profound inhibitory effects on reproductive function by suppressing the pulsatile release of GnRH and therefore LH. Although the mechanisms by which stressors disrupt the hypothalamic GnRH pulse generator remain to be fully elucidated, numerous studies have implicated the amygdala, especially its medial (MeA) and central nuclei (CeA), as key modulators of the neuroendocrine response to stress. In the present study, we investigated the roles of the MeA and CeA in stress-induced suppression of LH pulses. Ovariectomized rats received bilateral ibotenic acid or sham lesions targeting the MeA or CeA; blood samples (25 μl) were taken via chronically implanted cardiac catheters every 5 min for 6 h for the measurement of LH pulses. After 2 h of baseline sampling, the rats were exposed to either: restraint (1 h), insulin-induced hypoglycemia (IIH) (0.3 U/kg, iv), or lipopolysaccharide (LPS) (25 μg/kg, iv) stress. The restraint but not IIH or LPS stress-induced suppression of LH pulses was markedly attenuated by the MeA lesions. In contrast, CeA lesioning attenuated LPS, but not restraint or IIH stress-induced suppression of LH pulses. Moreover, after restraint stress, the number of Fos-positive neurons and the percentage of glutamic acid decarboxylase(67) neurons expressing Fos was significantly greater in the GnRH-rich medial preoptic area (mPOA) of rats with intact, rather than lesioned, MeA. These data indicate that the MeA and CeA play key roles in psychogenic and immunological stress-induced suppression of the GnRH pulse generator, respectively, and the MeA-mediated effect may involve γ-aminobutyric acid ergic signaling within the mPOA.

Stress exerts profound inhibitory effects on reproductive function by suppressing the pulsatile release of gonadotrophin-releasing hormone and therefore LH, however the mechanisms by which stressors disrupt the hypothalamic GnRH pulse generator remain to be fully elucidated. The amygdaloid complex, a prominent limbic brain structure, has been implicated in many aspects of the neuroendocrine responses during stress, including activation of the hypothalamo-pituitary-adrenal (HPA) axis; with numerous studies demonstrating that while its lesioning leads to decreases in the stress-induced release of corticosterone (1), its stimulation has the opposite effect (2). Furthermore, in accordance with the notion that psychogenic and systemic stressors activate distinct neuronal pathways, it is now the understanding that like other limbic structures, the effect of the amygdala on the HPA axis is also stressor- and region-specific (3). Of a number of subdivisions of the amygdala, the medial nucleus (MeA) which projects to basal forebrain, hypothalamic, and brain stem structures demonstrates the most neuronal activation in response to psychogenic stressors such as restraint, forced swim, noise, or predator exposure (4–6) rather than systemic stressors like interleukin-1β injection, hypoxia, or hemorrhage (6–8), as measured by c-fos expression. Moreover, consistent with its stressor-specific role, selective MeA lesioning has been found to block the activation of hypothalamic paraventricular neuroendocrine cell responses to acute restraint (9) and the increase in corticosterone in response to predator odor in rats (10). In contrast, the central nucleus (CeA) of the amygdala has been shown to be highly responsive to systemic stressors like inflammation or hemorrhage (7, 8), with specific lesions of this subnucleus resulting in an attenuation of the HPA response after interleukin-1β injection (11) but not restraint stress (9).

There is accumulating evidence describing a role for limbic brain structures as modulators of the hypothalamo-pituitary-gonadal (HPG) axis. The amygdala with its medial subdivision has been strongly implicated in regulating reproductive processes, such as ovulation, estrous cyclicity, pituitary LH and FSH production, as well as sexual behavior (12–15). A recent report indicating that lowering stress levels with cognitive behavioral therapy restored ovulation in more than 80% of women with functional hypothalamic amenorrhoea (16) may also underlie the clinical importance of the limbic system in the suppression of the HPG axis. More specifically, it has been shown that plasma LH levels increase after MeA lesioning (17) while electrical stimulation of the MeA results in delayed puberty in female rats (18), suggesting this nucleus exerts an inhibitory influence on reproductive function. Such an influence of the MeA on the HPG axis is further supported by neuroanatomical studies demonstrating the existence of extensive direct projections from the MeA to the hypothalamic GnRH-rich medial preoptic area (mPOA) (19, 20) although the neurochemical phenotype of these efferents is unknown. The CeA contains a major corticotropin-releasing factor (CRF) neuronal population (21), which is a key inhibitory neuropeptide of the GnRH pulse generator (22). Moreover, recent studies indicate that CRF within the CeA negatively impacts reproductive physiology in female rats by disrupting the estrus cycle and decreasing GnRH expression in the mPOA (23).

The present study was designed to investigate whether neurotoxic lesions specific to the MeA block psychological (restraint) stress-induced suppression of pulsatile LH release in the rat and compare the response with commonly used systemic stressors such as insulin-induced hypoglycemia (IIH) or lipopolysaccharide (LPS), which readily suppress LH pulses in a variety of species (24–27). A differential response to these three stress paradigms in CeA lesioned animals was also investigated. In addition, immunostaining for Fos protein, a marker of neuronal activation, was used to establish whether MeA lesions affected activation of the mPOA in response to restraint stress. Because of the ubiquity of γ-aminobutyric acid (GABA) neurons in the mPOA (28), their dense connectivity with GnRH neurons (28), the inhibitory effect of GABA on the excitability of GnRH neurons in vitro (29) and the suppression of pulsatile LH secretion in response to intra-mPOA administration of GABA in vivo (30, 31), we further examined whether these activated neurons are GABAergic by double labeling for glutamic acid decarboxylase (GAD67), a marker for GABAergic neurons, and Fos.

Materials and Methods

Animals and surgical procedures

Adult female Sprague Dawley rats, weighing 220–250 g, obtained from Charles River (Margate, UK), were housed under controlled conditions (12-h light, 12-h dark, with lights on at 0700 h; temperature at 22 ± 2 C) and provided with food and water ad libitum. All animal procedures were performed in accordance with the United Kingdom Home Office Regulations. All surgical procedures were carried out under ketamine (100 mg/kg ip; Pharmacia and Upjohn, Crawley, UK) and Rompun (10 mg/kg ip; Bayer, Leverkusen, Germany) anesthesia. Rats were ovariectomized and received bilateral MeA or CeA excitotoxic lesions via bilateral injections of 0.4 μl of ibotenic acid [10 μg/μl in sterile 0.1 m PBS (pH 7.4); Sigma-Aldrich Ltd, Poole, UK] using a 25 gauge 1 μl Hamilton injection syringe, at the following coordinates: 3.3 mm lateral, 2.6 mm posterior to Bregma, and 8.7 mm below the surface of the dura for the MeA; 4.1 mm lateral, 2.5 mm posterior to Bregma, and 7.7 mm below the surface of the dura for the CeA (32). Sham lesions were carried out using the same procedure but with sterile artificial cerebrospinal fluid. After a 10-d recovery period, the rats were fitted with two indwelling cardiac catheters via the jugular veins, as previously described (33). The catheters were exteriorized at the back of the head and secured to a cranial attachment; the rats were fitted with a 30-cm-long metal spring tether (Instec Laboratories Inc, Boulder, CO). The distal end of the tether was attached to a fluid swivel (Instec Laboratories), which allowed the rat freedom to move around the enclosure. Experimentation commenced 3 d later.

The role of the MeA and CeA in the restraint, LPS, hypoglycemic stress-induced suppression of pulsatile LH secretion

On the morning of experimentation, the rats were attached via one of the two cardiac catheters to a computer-controlled automated blood sampling system, which allows for the intermittent withdrawal of small blood samples (25 μl) for LH measurement without disturbing the animals (25). Once connected, the animals were left undisturbed for 1 h. Blood sampling commenced at approximately 1000 h, and samples were collected every 5 min for 6 h. Blood samples were frozen at −20 C for later assay to determine LH concentrations. After removal of each 25-μl blood sample, an equal volume of heparinized saline (5 U/ml normal saline, CP Pharmaceuticals Ltd, Wrexham, UK) was automatically infused into the animal to maintain patency of the catheter and blood volume. After 2 h of controlled blood sampling for baseline LH pulses, the animals were exposed to one of the three stressors. Control treatment matching a specific stressor was also given to separate groups of animals with bilateral MeA or CeA lesions. For restraint stress, the MeA lesioned (n = 12) or intact (n = 8) rats and CeA lesioned (n = 12) or corresponding intact (n = 8) animals were placed in restraint devices for 60 min (33), and automated blood sampling continued uninterrupted during restraint and the 3-h postrestraint period. Control MeA lesioned (n = 10) or CeA lesioned (n = 10) animals underwent the same procedures but were not restrained. For the immunological stressor, 2 h into the 6 h blood sampling procedure, LPS (25 μg/kg in 0.3 ml saline, Sigma-Aldrich) was injected via the second iv catheter (25) in MeA lesioned (n = 11), MeA intact (n = 8), CeA lesioned (n = 12), and CeA intact (n = 8) rats. Control animals with MeA lesions (n = 10) or CeA lesions (n = 10) received an iv injection of 0.3 ml sterile saline. For insulin-induced hypoglycemic stress, the rats were fasted overnight but were provided with water ad libitum throughout. After 2 h of controlled blood sampling insulin (0.3 IU/kg in 0.3 ml of saline) was administered through the second iv catheter (25) in MeA lesioned (n = 12), MeA intact (n = 8), CeA lesioned (n = 12), and CeA intact (n = 8) rats. Control MeA lesioned (n = 10) or CeA lesioned (n = 10) animals were also fasted overnight but given 0.3 ml sterile saline. Blood glucose was monitored during the experiment, and for this purpose manual blood samples (approximately 25 μl) were collected via the second catheter at −30 min, immediately before the insulin or saline administration, and +10, +20, +30, and +60 min after administration. Blood glucose concentrations were measured using a Reflolux S blood glucose monitor (Boehringer, Mannheim, Germany).

RIA for LH measurement

A double-antibody RIA supplied by the National Institute of Diabetes, Digestive, and Kidney Diseases (NIDDK, Bethesda, MD, USA) was used to determine LH concentrations in the 25-μl whole-blood samples. Referenced preparation was rLH-RP-3. The sensitivity of the assay was 0.093 ng/ml. The intraassay variation was 6.8%, and the interassay variation was 8.0%.

The effect of MeA lesions on Fos and GAD67 expression in the mPOA after restraint stress

A separate group of MeA lesioned (n = 12) and sham lesioned controls (n = 8) were restrained for 1 h as described above. Then, 30 min postrestraint they were deeply anesthetized using 0.5 ml sodium pentobarbitone (60 mg/ml iv; Rhone Merieux Ltd., Harlow, UK) and transcardially perfused with heparinized saline (5 U/ml) followed by 4% paraformaldehyde in 0.1 m PBS (pH = 7.4). Another group of MeA lesioned (n = 12) and MeA intact (n = 8) rats underwent the same procedures but were not exposed to restraint stress. The brains were removed and directly put into a postfix solution. They were then transferred into a solution of 30% sucrose. Brains were stored at −80 C and coronally sectioned (30 μm) at a later point. Every sixth free-floating section was used for Fos and GAD67 dual immunostaining. Sections were processed for 45 min in 0.3% triton, 10 min in 0.02% H2O2, and then 10 min in 2% normal goat serum (NGS). Sections were then incubated in 1:250,000 polyclonal rabbit anti-fos primary antibody (AB-5 OBI, Oncogene Science, San Diego, CA) containing 1% NGS at 4 C for 48 h, after which they were processed with a biotinylated goat antirabbit secondary antibody (BA-1000; Vector Laboratories, Peterborough, UK) at a dilution of 1:1000 containing 1% NGS for 2 h. This was followed by incubation in Vectastain Elite ABC peroxidise system (PK-6100; Vector Laboratories, UK). The diaminobenzidine reaction was used to visualize Fos immunoreactivity, this was intensified with nickel ammonium chloride (Vector Laboratories, Burlingame, CA). Sections were then incubated for 10 min in 0.02% H2O2 and 30 min in 5% NGS, followed by a 36-h incubation at 4 C in mouse anti-GAD67 monoclonal antibody (MAB5406; Chemicon International, Temecula, CA) at a dilution of 1:4000 in 1% NGS. A further incubation in 1:1000 biotinylated goat antimouse antibody (BA-9200; Vector Laboratories, UK) containing 1% NGS for 90 min, was followed by incubation in streptaridin peroxidise tertiary reagent (PK-6100; Vector Laboratories) at a dilution of 1:200 in 0.1 m PBS. Visualization of GAD immunoreactivity was achieved using the diaminobenzidine reaction. The sections were mounted on slides, dehydrated, and coverslipped. Omission of the Fos or GAD67 primary antibody resulted in the absence of specific staining. Several brains from each experimental group were processed on the same day to control for interbatch variability.

Semiquantitative analysis of immunostaining data for Fos and GAD67 was carried out on an AxioVision microscope image system (Zeiss, Oberkochen, Germany). All analysis was performed on coded slides by two individual investigators without knowledge of experimental treatment conditions. The boundaries of the mPOA counted were determined by comparing the Paxinos and Watson's rat brain atlas (32) with neuroanatomical and cytoarchitectural landmarks. Sections used for mPOA analysis were taken from the region corresponding to bregma +0.20 to −0.40 mm. Neurons expressing both Fos and GAD67 immunoreactivity were counted with bright-field microscopy at ×40 magnification. The number of Fos-positive cells was determined bilaterally in four sections from each rat. The number of cells displaying the GAD67 protein was counted, and the percentage of these cells colocalizing Fos was calculated. Fine focusing was performed to ensure counting of all immunostained cells throughout the thickness of the sections.

Statistical analysis

The algorithm ULTRA was used to establish the detection of LH pulses (34). Two intraassay coefficients of variation of the assay were used as the reference threshold for the pulse detection. The inhibitory effect of restraint, hypoglycemic, or LPS stress on LH pulses, in the presence or absence of either MeA or CeA lesions, was calculated by comparing the mean LH pulse interval before stress with the first prolonged interval after stress onset. For the restraint and hypoglycemic stress the first prolonged interval corresponded to the first LH pulse interval after stress onset. However, in the case of LPS stress, the first prolonged LH pulse interval was delayed by approximately 30 min post LPS injection. The poststress recovery period was defined as the remaining observation period after the first prolonged LH pulse interval. In the absence of stressors, the LH interpulse intervals within the corresponding time periods were compared. The decrease in blood glucose concentration in response to insulin was determined by comparing the mean glucose levels during the 30 min before insulin injection with the mean glucose concentration during the 60-min period after injection. Statistical significance was tested using ANOVA and Dunnett's test. For Fos and GAD67 immunostaining, individual rat means were combined to provide group means and analyzed using ANOVA followed by the Mann–Whitney Rank Sum test. P < 0.05 was considered statistically significant.

Results

MeA and CeA lesion verification

The extent and location of the lesions were confirmed by microscopic histological inspection, using Nissl staining. The presence of extensive neuronal loss, gliosis infiltration, and injector trajectory within the MeA or CeA were used as parameters to determine the existence of significant lesions. Animals were excluded from the lesion groups as a result of unsuccessful lesioning, unilateral damage, or damage extending outside of the MeA or CeA. All artificial cerebrospinal fluid injected rats sustained no damage to the MeA or CeA. Figure 1, A and D shows the variation in MeA and CeA lesion extents, respectively. Figure 1, B and C shows representative examples of intact MeA and lesioned MeA, respectively. Figure 1, E and F shows representative examples of intact CeA and lesioned CeA, respectively.

Fig. 1.

Schematic diagrams and representative examples illustrating bilateral lesions of the medial nucleus (MeA) or central nucleus (CeA) of the amygdala. A and D show the variation in extent of the bilateral MeA and CeA lesions targeted by ibotenic acid injections, respectively; the gray areas highlight the smallest extent of the lesions, while the black bold line denotes the largest extent of the lesions. B and E, Intact MeA and CeA, respectively, shown by Nissl staining. C and F, Lesioned MeA and CeA, respectively, shown by Nissl staining and outlined by the broken line. Bars, 500 μm. opt, optic tract.

The effect of MeA or CeA lesions on the restraint stress-induced suppression of pulsatile LH secretion

The LH pulse intervals in the prerestraint stress control period with and without MeA lesions were 22.39 ± 0.99 and 24.63 ± 1.02 min (mean ± sem, n = 8 per group), respectively. There was no significant difference between the LH pulse intervals within the 2-h baseline period of control (n = 7) and restrained rats with bilateral MeA lesions (P = 0.73). Similarly, the LH pulse intervals in the prerestraint stress control period with and without CeA lesions were 25.10 ± 1.49 and 23.33 ± 0.89 min (mean ± sem, n = 8 per group), respectively. There was no significant difference between the LH pulse intervals within the 2-h baseline period of control (n = 7) and restrained rats with bilateral CeA lesions (P = 0.30). Restraint stress induced an immediate suppression of pulsatile LH in intact MeA rats (Fig. 2, B and D; P < 0.05) that was markedly attenuated in MeA lesioned animals (Fig. 2, C and D; P < 0.05). In contrast, bilateral CeA damages had no effect on restraint stress-induced suppression of LH pulses (Fig. 3, C and D). The MeA or CeA lesions per se had no effect on LH pulse frequency with regular pulsatile LH secretion observed throughout the entire 6-h blood sampling period (Fig. 2, A and D; Fig. 3, A and D).

Fig. 2.

Representative examples illustrating the effect of restraint, LPS, and insulin-induced hypoglycemic stress on pulsatile LH secretion in ovariectomized rats with intact medial nuclei of the amygdala (MeA) or bilateral lesions of the MeA. Rats with intact (sham lesions) MeA (B) and MeA lesions (C) restrained for 1 h. A, MeA-lesioned rat without restraint. F and J, Rats with intact MeA treated with LPS and insulin, respectively. G and K, Rats with MeA lesions treated with LPS and insulin, respectively. MeA-lesioned rats treated with saline control (E, I). Summary showing the MeA lesions attenuate the inhibitory effect of restraint (D), but not LPS (H) and insulin-induced hypoglycemic (L) stress, on pulsatile LH secretion. #, P < 0.05 vs. prestress control period within the same group. *, P < 0.05 vs. nonlesion group post restraint at the same time point.

Fig. 3.

Representative examples illustrating the effect of restraint, LPS, and insulin-induced hypoglycemic stress on pulsatile LH secretion in ovariectomized rats with intact central nuclei of the amygdala (CeA) or bilateral lesions of the CeA. Rats with intact (sham lesions) CeA (B) and CeA lesions (C) restrained for 1 h. A, CeA-lesioned rat without restraint. F and J, Rats with intact CeA treated with LPS and insulin, respectively. G and K, Rats with CeA lesions treated with LPS and insulin respectively. CeA-lesioned rats treated with saline control (E and I). Summary showing the CeA lesions attenuate the inhibitory effect of LPS (H), but not restraint (D) and insulin-induced hypoglycemic (L) stress, on pulsatile LH secretion. #, P < 0.05 vs. prestress control period within the same group. *, P < 0.05 vs. nonlesion group post restraint at the same time point.

The data of four rats with unilateral damage or lesions dorsal to the MeA were analyzed separately. The restraint stress-induced suppression of LH pulses was not attenuated in these animals, with a group mean ± sem of 47.32 ± 4.78 min for the first prolonged LH pulse interval. Similarly, restraint induced suppression of LH pulses in four CeA misplaced lesion rats, with a group mean ± sem of 43.75 ± 6.35 min for the first prolonged LH pulse interval.

The effect of MeA or CeA lesions on the LPS stress-induced suppression of pulsatile LH secretion

In the 2-h basal control period, no differences in LH pulse interval were detected among the three groups in the MeA and CeA experiments (Figs. 2, E–H and 3, E–H). Administration of LPS (25 μg/kg, iv) evoked a profound inhibitory effect on the pulsatile release of LH which was observed after a delay of approximately 30 min with the first prolonged LH pulse interval lasting more than 1 h, in both MeA intact (Fig. 2, F and H, n = 7, P < 0.05) and CeA intact rats (Fig. 3, F and H, n = 8, P < 0.05), followed by a recovery to normal pulse frequency. Bilateral CeA damage significantly attenuated LPS-induced suppression of LH pulses (Fig. 3, G and H, n = 8, P < 0.05), but bilateral MeA lesions had no effect on the inhibitory response (Fig. 2, G and H, n = 7). Control saline infusions had no effect on LH pulse frequency in both MeA-lesioned rats (Fig. 2, E and H, n = 8, P = 0.72) or CeA-lesioned animals (Fig. 3, E and H, n = 8, P = 0.65). LPS induced similar suppression of LH pulses in three MeA and four CeA misplaced lesion rats (data not shown).

The effect of MeA or CeA lesions on the hypoglycemic stress-induced suppression of pulsatile LH secretion

Insulin (0.3 U/kg, iv) administration resulted in a significant decrease in blood glucose; mean blood glucose levels during the pretreatment control period vs. the first hour after insulin injection were 55.26 ± 4.13 vs. 23.27 ± 3.91 mg/dl (mean ± sem, P < 0.05) in sham MeA-lesioned (n = 8) rats and 57.41 ± 3.18 vs. 24.77 ± 4.16 mg/dl (mean ± sem, P < 0.05) in MeA-lesioned (n = 9) rats, and 60.31 ± 4.25 vs. 22.13 ± 4.05 mg/dl (mean ± sem, P < 0.05) in sham CeA-lesioned (n = 8) rats and 55.13 ± 3.75 vs. 21.11 ± 3.46 mg/dl (mean ± sem, P < 0.05) in CeA-lesioned (n = 9) rats. Hypoglycemia induced an immediate and significant interruption of the LH pulses in MeA intact (Fig. 2, J and L, P < 0.05) and CeA intact (Fig. 3, J and L, P < 0.05) animals. Neither MeA nor CeA lesioning affected the hypoglycemic stress-induced suppression of pulsatile LH secretion (Figs. 2, K and L and 3, K and L). MeA lesioned rats (n = 8) treated with saline showed no significant changes in either blood glucose (data not shown) or LH pulse intervals (Fig. 2, I and L, P = 0.44). Similarly, CeA lesioned rats (n = 7) treated with saline showed no significant changes in either blood glucose (data not shown) or LH pulse intervals (Fig. 3, I and L, P = 0.39). IIH induced similar suppression of LH pulses in three rats with misplaced MeA lesions and two rats with misplaced CeA lesions (data not shown).

The effect MeA lesions on Fos and GAD67 expression in the mPOA after restraint stress

Fos-positive neurons were identified by black reaction product restricted to the nucleus of the cells. In the absence of restraint stress, Fos-positive cells within the mPOA of the MeA intact and MeA lesioned rats were very rare, with groups mean ± sem of 6.88 ± 1.34 and 5.88 ± 1.13, respectively (Fig. 4, A, B, and F, n = 8 per group, P = 0.58). After restraint stress, a relatively high number of Fos-positive cells were detected in the mPOA (Fig. 4, C and F, 99.09 ± 16.98, mean ± sem) in intact MeA rats (n = 8). In contrast, MeA lesioned rats demonstrated a dramatic reduction in the number of mPOA neurons expressing Fos after restraint stress (Fig. 4, D and F, P < 0.001). Similarly, the percentage of mPOA GAD67-immunoreactive neurons expressing Fos was significantly reduced in MeA lesioned rats (Fig. 4, D and G), when compared with non-lesioned animals after restraint stress (Fig. 4, C and G, P < 0.001). In the absence of restraint stress, the percentage of mPOA GAD67-immunoreactive neurons expressing Fos was very low and showed no significant difference between sham MeA lesioned and MeA lesioned rats (Fig. 4G, P = 0.77). Between MeA lesioned and nonlesioned rats, there was no significant difference in the number GAD67-positive neurons within the mPOA (data not shown).

Fig. 4.

Representative examples of dual-labeling immunoreactivity for Fos and GAD67 neurons within the mPOA in response to restraint stress. A and B, mPOA of rats with intact (sham lesions) medial nuclei of the amygdala (MeA) or bilateral lesions of the MeA in the absence of restraint stress, respectively. C and D, mPOA of intact MeA or MeA lesioned rats after restraint stress, respectively. Bars, 200 μm. E, Enlarged view showing a Fos and GAD67 dual-labeled neuron (black arrowhead), GAD67 (white arrowhead), or Fos (black arrow) single-labeled neurons. Bars, 20 μm. F and G, Summary showing the effect of MeA lesions on Fos expression and percentage of GAD67 neurons expressing Fos in the mPOA, respectively, in the presence or absence (control) of exposure to restraint stress (1 h). **, P < 0.001 vs. sham-lesion group post restraint.

Discussion

The present study provides the first evidence revealing the pivotal roles of the MeA and CeA in the psychogenic and immunological stress-induced suppression of the GnRH pulse generator, respectively. Neurotoxic lesions of the MeA markedly attenuated the inhibitory effect of restraint, but not LPS or IIH stress, on pulsatile LH secretion. In contrast, LPS but not restraint or IIH stress induced suppression of pulsatile LH secretion was attenuated by the CeA lesions. The differential role of the MeA and CeA in stress-induced suppression of pulsatile LH secretion supports the hypothesis that different neural pathways mediate reproductive dysfunction to different types of stressors. Moreover, because MeA or CeA lesioning did not affect basal pulsatile LH secretion, it appears these amygdala nuclei are only involved in regulating GnRH pulse generator activity under stressful conditions. These data suggest that restraint stress-induced MeA activity and immunological stress-induced CeA activity can impact the HPG axis by exerting an inhibitory effect on the GnRH pulse generator. The HPG-inhibitory role for the MeA and CeA is consistent with their well-known function in regulating the stress response following stress exposure (35). Indeed damage to the MeA has been shown to decrease anxiety-related behavior and corticosterone responses to stressors (9, 10, 36, 37), while MeA stimulation has been shown to increase HPA axis output (2) and facilitate its stress-induced activation (38). Electrical stimulation of the CeA results in the activation of the autonomic nervous system (39), while lesions of the CeA attenuate corticosterone release evoked by interleukin-1β administration and autonomic activation (11, 40).

The MeA has also been implicated in regulating reproduction and therefore the HPG axis, however its precise role in modulating this neuroendocrine axis is controversial. Electrical stimulation of the MeA delays puberty in female rats (18), suggesting an inhibitory influence on gonadotropin secretion. Conversely, MeA lesioning has also been shown to reduce ovulation (12) and the mating-induced enhancement of lordosis in female rats (41). Furthermore, the precise role of the MeA in the control of LH secretion is also controversial with reports of increases in LH secretion after both MeA lesioning (17) and MeA stimulation in rats (13, 42, 43) and reports indicating that MeA lesioning has no effect on LH secretion in the rhesus monkey (44), immature rat (45), and deer-mice (46). Nonetheless the studies presented above, regardless of whether they demonstrate an inhibitory or stimulatory role for the MeA on LH secretion, focus on either changes in mean circulating levels of LH or the LH surge. Although electrochemical stimulation of the CeA has no effect on mean circulating levels of LH (13), the CeA has been shown to be involved in maternal (47) and sexual behavior (23). In the present study, we found that MeA or CeA lesions do not affect LH pulse frequency under stress-free conditions. The conflicting data presented above, which highlight a controversial role for the MeA on LH secretion, may result from the use of different experimental techniques, different LH measurement time points after MeA damage or activation, or unintentional damage to the stria terminalis or to fibers that run through the MeA. Thus, based on the results from this study, it appears that the MeA and CeA per se are not involved in the normal regulation of LH pulses under nonstressed conditions, but nonetheless their activation under stressful conditions consequently inhibits GnRH pulse generator activity.

In accordance with studies demonstrating the specific role of the MeA in regulating the stress response to psychogenic but not systemic stressors evidenced by intense c-fos expression within the MeA in response to restraint, swim, noise, or predator exposure (4–6), but not interleukin-1β injection, hypoxia, or hemorrhage (6–8), the present data suggest the MeA also exerts an HPG-inhibitory effect in a stressor-type specific manner. Bilateral lesions of the MeA markedly attenuated the psychogenic (restraint) stress-induced suppression of pulsatile LH release without affecting the inhibitory response induced by immunological (LPS) or metabolic (IIH) stressors, suggesting that the MeA suppresses reproductive function only when specifically activated by psychogenic stressors. With regards to systemic stressors, the CeA has been shown to be highly responsive to stressors like inflammation or hemorrhage (7, 8), with specific lesions of this sub-nucleus resulting in an attenuation of the HPA response after interleukin-1β injection (11) but not restraint stress (9). In addition, it was recently shown that CRF within the CeA may play a significant role in stress-related reproductive disorders by demonstrating that increased expression of CRF in the CeA emulate the neuroendocrine effects, normally caused by stress, that negatively impact reproductive physiology in female rats (23); thereby indicating that the CeA may also play a role in stress-induced suppression of LH pulses. Indeed, the present study demonstrated that lesions of the CeA attenuate LPS, but not restraint stress-induced suppression of LH pulses in the rat, supporting the stressor selective nature of the CeA.

Although the results of the present study demonstrate that under restraint stress the MeA has an important role in inhibiting the HPG axis, the neuroanatomical circuitry involved remains to be elucidated. MeA-mediated activation of GABAergic neurons after restraint stress, as shown by a greater percentage of GAD67 neurons expressing Fos immunoreactivity in the mPOA of intact compared with MeA lesioned animals, is consistent with the suggestion of a functional relationship between these two areas of the brain and the possibility that preoptic GABAergic neurons may underlie the reduction in GnRH pulse generator activity. Numerous studies indicate that GABA plays an important inhibitory role in the regulation of GnRH/LH secretion, because endogenous GABA release has been shown to exert a powerful inhibitory effect on the activation of GnRH neurons in vitro (29) and local infusion of GABA into the mPOA to suppress GnRH expression (48) and pulsatile LH release (30). More importantly, intra-mPOA administration of GABA receptor antagonist blocks stress-induced suppression of LH pulses (Lin, Li, & O'Byrne, unpublished observation), indicating a crucial role for endogenous mPOA GABAergic signaling in the stress-induced suppression of the GnRH pulse generator. Furthermore, the levels of Fos and percentage of GAD67 neurons expressing Fos within the mPOA were very low and indifferent between the MeA lesioned and intact animals in the absence of restraint stress, which further emphasizes a role for the MeA in regulation of the reproductive neuroendocrine system during stress but not under nonstress conditions.

Additionally, anatomical studies indicate that the MeA has extensive direct projections to the GnRH-rich mPOA (19, 20). Because extrinsic projections of the MeA are predominantly GABAergic (49), this raises the possibility that the attenuation of the restraint stress-induced suppression of pulsatile LH secretion, observed in MeA lesioned rats, may result from removal of inhibitory GABAergic inputs to the mPOA. However, the neurochemical phenotype of the direct projections from the MeA to the mPOA is unknown. In fact, as suggested by the immunohistochemical results of the present study, the MeA may suppress pulsatile LH secretion via efferents, which interact with intrinsic mPOA GABAergic neurons.

Furthermore, it is worth noting that the suppressive effect of restraint stress on pulsatile LH secretion was only attenuated by bilateral MeA lesions, not completely eliminated, putting forward the possibility that the MeA might just play a partial role in psychogenic stress-induced suppression of pulsatile LH secretion. In effect, there are other brain regions, which have been shown to play significant roles in the stress-induced suppression of reproduction. We have previously shown that CRF administration into the bed nucleus of stria terminalis (BNST) (50) or locus coeruleus (LC) (51) also results in the concomitant suppression of pulsatile LH secretion with activation of GABAergic neurons within the mPOA. Moreover, microinfusion of CRF antagonist into either the BNST (50) or the LC (51) blocked the psychogenic but not metabolic stress–induced suppression of LH secretion. Therefore, because the MeA sends extensive GABAergic projections to the BNST (52), the LC receives dense innervation from the BNST and sends projections directly to the mPOA (53, 54), it is reasonable to speculate that a neural MeA→BNST→LC→mPOA multi-synaptic circuit may underline an indirect pathway which contributes to the psychogenic stress evoked suppression of the GnRH pulse generator. This multi-synaptic pathway could also provide an explanation as to why the psychogenic stress-induced suppression of LH pulses was partially but not completely eliminated by MeA lesioning. Similar to the partial effect of MeA lesioning on restraint stress-induced suppression of LH pulses, the inhibitory effect of LPS stress on pulsatile LH secretion was not completely blocked by bilateral CeA lesions. It has been reported that there are very few (55) or no (20, 52) neural projections from the CeA to the GnRH-rich mPOA. Nevertheless, the CeA is known to densely innervate the BNST, primarily with CRF containing neurons (21, 56). Further, LPS stress induces strong Fos expression in both the BNST (57) and LC (58). Thus, the CeA may similarly provide input to the postulated multi-synaptic pathway to the mPOA described above. Interestingly, both MeA and CeA lesions did not affect the inhibitory response to hypoglycemic stress. Further, the BNST (50) and LC (51) appear not to be involved in mediating the inhibitory effect of this metabolic stressor on GnRH pulse generator activity, and the neural circuitry underlying this response remains to be established.

Although we have postulated that projections from the amygdala that mediate stress effects on the GnRH pulse generator terminate on targets in the mPOA, a caveat to consider is the overwhelming evidence that the neural construct comprising this hypothalamic oscillator resides in the arcuate nucleus of the mediobasal hypothalamus in rats (59) in common with other species including primates (60). The very sparse direct projections from the MeA/CeA to the arcuate nucleus (52, 61) might suggest a rather limited level of interaction between these brain regions. However, the robust connections between the amygdala and the BNST (52, 56), which in turn projects heavily to the arcuate nucleus (62), may provide an indirect functional route for amygdaloid influence on GnRH pulse generator activity; this is a postulate that remains to be explored. Nevertheless, the observation that microinfusion of a wide range of neuropeptides, including calcitonin gene-related peptide, CRF, or GABA, directly into the mPOA suppresses pulsatile LH secretion (30, 63, 64) is not inconsistent with the notion of a neurocircuit relay via the mPOA to the mediobasal hypothalamus to modulate GnRH pulse generator activity.

In the present study an ovariectomized rat model without gonadal steroid replacement was used. Because there is substantial evidence that gonadal steroids, in particular estradiol, sensitize the GnRH pulse generator to the inhibitory effects of stress (65–67) and augment the inhibitory effects of various neuropeptides such as CRF (68) and noradrenaline (69) on pulsatile LH secretion, we cannot rule out the possibility that the results of the present study might have been different with the use of a gonadal steroid replaced or ovary intact rat model.