Evidence that γ-Aminobutyric Acid Is Part of the Neural Circuit Mediating Estradiol Negative Feedback in Anestrous Ewes

Abstract

Seasonal anestrus in ewes is driven by an increase in response to estradiol (E₂) negative feedback. Compelling evidence indicates that inhibitory A15 dopaminergic (DA) neurons mediate the increased inhibitory actions of E₂ in anestrus, but these neurons do not contain estrogen receptors. Therefore, we have proposed that estrogen-responsive afferents to A15 neurons are part of the neural circuit mediating E₂ negative feedback in anestrus. This study examined the possible role of afferents containing γ-aminobutyric acid (GABA) and nitric oxide (NO) in modulating the activity of A15 neurons. Local administration of NO synthase inhibitors to the A15 had no effect on LH, but GABA receptor ligands produced dramatic changes. Administration of either a GABA_A or GABA_B receptor agonist to the A15 increased LH secretion in ovary-intact ewes, suggesting that GABA inhibits A15 neural activity. In ovariectomized anestrous ewes, the same doses of GABA receptor agonist had no effect, but combined administration of a GABA_A and GABA_B receptor antagonist to the A15 inhibited LH secretion. These data are consistent with the hypothesis that endogenous GABA release within the A15 is low in ovary-intact anestrous ewes and elevated after ovariectomy. Using dual immunocytochemistry, we observed that GABAergic varicosities make close contacts on to A15 neurons and that A15 neurons contain both the GABA_A-α1 and the GABA_B-R1 receptor subunits. Based on these data, we propose that in anestrous ewes, E₂ inhibits release of GABA from afferents to A15 DA neurons, increasing the activity of these DA neurons and thus suppressing episodic secretion of GnRH and LH.

REPRODUCTION IS UNIQUE among physiological systems in that it can be shut down for prolonged periods. In females, the most common instances of such suppression are the anovulation that occurs before puberty (1), during lactation (2), and annually in seasonally breeding species (3,4). Over the last two decades, considerable progress has been made in understanding the neuroendocrine mechanisms underlying seasonal breeding in sheep. It is now well established that a dramatic annual shift in the response to the negative feedback actions of estradiol (E₂) causes seasonal breeding in ewes (4,5). The nonbreeding (anestrous) season occurs because E₂ gains the capacity to potently inhibit GnRH and LH pulse frequency at this time of year (4,5) so that the low amounts of E₂ secreted by the ovary are responsible for the slow LH pulse frequency observed in ovary-intact anestrous ewes (6).

A group of dopaminergic (DA) neurons located at the base of the brain in the retrochiasmatic (RCh) area of the ovine hypothalamus plays a key role in mediating these seasonal changes in response to E₂ negative feedback (5,7). These DA neurons inhibit LH pulse frequency (8,9) in anestrus, but not the breeding season, and their activity is stimulated by E₂ only during anestrus (10,11,12). Because A15 DA neurons do not contain either estrogen receptor-α (ERα) or ERβ (13,14,15), they are most likely stimulated by afferent input from estrogen-responsive interneurons. Afferent neurons containing ERα have been identified in the ventromedial preoptic area (vmPOA) and RCh (16), and local administration of E₂ to these areas acts via a DA system to inhibit LH pulse frequency in anestrous, but not breeding-season, ewes (17,18,19).

Although two anatomical locations of estrogen-responsive afferents to A15 neurons have been identified, the phenotype of these neurons remains unknown. Likely candidate neurotransmitters, based on dual-label immunocytochemical studies (5,20,21,22,23), include γ-aminobutyric acid (GABA) (20), glutamate (21), nitric oxide (22), and dynorphin (23). The latter can be excluded because endogenous opioids are not involved in E₂ negative feedback in anestrus (24,25). In the present studies, we examined the possible roles of nitric oxide (NO) and GABA and obtained strong anatomical and pharmacological evidence that GABAergic neurons mediate, at least in part, the effects of E₂ on these DA neurons. In light of these data, we also determined whether there were seasonal changes in GABA input or GABA receptors that could account for seasonal differences in the ability of E₂ to stimulate A15 neurons.

Materials and Methods

Animal handling and surgical procedures

Mature, black-faced ewes with a history of normal reproductive cycles were moved indoors 3–7 d before the procedures and housed two per pen under artificial lighting with duration similar to that outdoors. They were fed a maintenance level of alfalfa pellets supplemented with grain and minerals and had free access to water. All pharmacological experiments were done during the anestrous season (from May through early August), and infertility was confirmed by undetectable progesterone concentrations and/or absence of corpora lutea at ovariectomy (OVX). For seasonal comparisons, we used tissue collected during anestrus and the breeding season (October through December) from E₂-treated OVX ewes (11,16) to avoid confounding effects of different circulating ovarian steroids in the two seasons.

All surgeries were done using sterile techniques with animals anesthetized with oxygen plus nitrous oxide, supplemented with 1–4% halothane as needed. Ovaries were removed via midventral laparotomy. For anatomical studies (experiments 6–8), OVX ewes were given a 3-cm-long sc SILASTIC brand (Dow Corning Corp., Midland, MI) implant containing E₂ (9,11) at the time ovaries were removed. Bilateral stainless steel guide tubes (18 gauge) aimed at the A15 area were stereotaxically implanted as previously described (17,19). Briefly, a 2-cm-diameter hole was drilled in the skull, the sagittal sinus ligated, and radioopaque dye injected into one lateral ventricle. Positive ventrilography in both sagittal and coronal planes was used to place the tips of the guide tubes at the posterior edge of the optic chiasm, 2.0 mm dorsal to the base of the brain and 3.0 mm lateral to midline. After protecting the surface of the brain with gel foam and a nylon mesh, guide tubes were cemented in place with cranial screws and dental acrylic, plugged with 22-gauge wire stylets, and protected with a plastic cap. All animals were treated pre- and postoperatively with dexamethasone and antibiotics, received analgesic just before surgery, and were allowed to recover for at least 2 wk before being used in an experiment. All procedures involving animals were approved by the West Virginia University Animal Care and Use Committee.

Reagents

The GABA_A receptor agonist muscimol and antagonist (−)-bicuculline methochloride were purchased from Tocris (Ellisville MO). A GABA_B receptor agonist [R-(+)-baclofen] and antagonist (phaclofen) were obtained from Sigma Chemical Co. (St. Louis MO), and a second GABA_B antagonist (CGP52432) came from Tocris. These drugs have been extensively used to examine the role of different GABA receptors in control of GnRH secretion in rodents and sheep (26,27). Nitric oxide synthesis (NOS) inhibitors S-methyl thiocitrulline (SMTC) and NG-nitro-l-arginine methyl ester (l-NAME) (28) were purchased from Sigma. All drugs were stored in crystalline form according to the directions of the manufacturer. When used for microinjections, drugs were dissolved in sterile water and used within 2 wk. Primary antisera used for immunocytochemical (ICC) studies were mouse anti-tyrosine hydroxylase (anti-TH), rabbit antibodies against glutamic acid decarboxylase 67 (GAD67) and GABA_A receptor α (GABA_A-Rα1) (all from Chemicon International, Temecula, CA), and guinea pig antisera against GABA_B receptor-1 (GABA_B-R1) (Research Diagnostics, Concord, MA). Antisera against GAD67 and GABA receptor subunits were chosen after screening a number of commercially available antibodies for use in sheep (data not shown).

General experimental protocols

Blood samples were collected by jugular venipuncture every 12 min from 36 min before to 4 h after insertion of microimplants or microinjections (ovary-intact animals) or for 2–3 h before and 3 h after microinjections (OVX ewes). Blood was allowed to clot overnight, and serum was harvested and stored at −20 C until assayed. For microimplants (experiment 1), drugs were tamped into the lumen of 22-gauge stainless steel tubing cut to extend 1–2 mm beyond the end of the guide tube. Microimplants were inserted and left in place for the 4 h of frequent blood collection and then replaced with stylets. For microinjections, drugs were dissolved in sterile water and 300 nl (or 400 nl for experiment 5b) rapidly injected bilaterally using a 1-μl Hamilton syringe with fixed needle that extended 1–2 mm beyond the end of the guide tubes. These volumes were selected, based on the spread of tract tracers (16), to encompass the A15 without encroaching on the third ventricle (estimated injection radius of 1.5–2 mm). The doses of drugs (range of 1.5–3 μg/injection) were based on previous work in sheep, which delivered doses of GABA receptor ligands of 10 μg (29,30), 4.5–8 μg (31), or 15–45 μg (32). Ewes were treated with gentamicin (250 mg im) prophylactically after each blood collection period. Experiments used a replicate design so that all animals received each experimental and control treatment, in random order, with 4–5 d between treatments.

Tissue collection

At the end of the experiment, fixed tissue was collected from drug-treated ewes for histological verification of microinjection/implant sites and for ICC analysis. Ewes were injected before perfusion twice (10 min apart) with 25,000 U heparin iv, followed by sodium pentobarbital (about 4–5 g iv). When the animal had stopped breathing, we quickly removed its head and perfused the brain via both internal carotid arteries with 6 liters of 4% paraformaldehyde in 0.1 m phosphate buffer (PB) containing 10 U heparin/ml and 0.1% sodium nitrite (pH 7.4). A block of tissue containing preoptic and hypothalamic tissue was dissected out, incubated in the same fixative overnight (4 C), and then infiltrated with 20% sucrose in PB. Thick coronal sections (50 μm) were cut using a freezing microtome and either stored at −20 C in cryopreservative until processed for dual ICC, or every fifth section was mounted on slides and stained with cresyl violet for identification of treatment sites.

Hormonal and immunocytochemical procedures

LH concentrations were measured in duplicate 100- or 200-μl aliquots of serum by RIA as previously described (23); assay sensitivity averaged 0.6 ng/ml (NIH S24), and inter- and intraassay coefficients of variation of a pool that produced 60% displacement of iodinated LH were 11 and 21%, respectively. Progesterone was measured in selected samples to confirm ewes were in anestrus using a commercially available kit (Diagnostic Systems Laboratories, Inc., Webster, TX) validated for use in sheep (23), with a sensitivity of 0.05 ng/ml.

We used dual-ICC and confocal microscopy to determine whether GABAergic terminals make close contacts onto A15 neurons and whether A15 neurons contained GABA_A and/or GABA_B receptors. Tyramide amplification (33) was used to amplify the primary antibodies against GAD67, GABA_A-Rα1, GABA_B-R1, as previously described (34). Briefly, two to three free-floating sections per area per ewe from the RCh were washed in 0.1 m PB and then incubated sequentially in 1% H₂O₂ for 10 min at room temperature (RT), 4% blocking sera (normal goat or donkey serum; Jackson ImmunoResearch Laboratories, West Grove PA) for 1 h in PBS or PBS containing 0.4% Triton X-100 (TxBS), and primary antisera for 24–48 h (Table 1). Sections were then washed in PB and incubated sequentially at RT in 1) 1:200 biotinylated second antibody (Jackson) for 1 h, 2) 1:200 Vectastain ABC-elite (Vector Laboratories, Burlingame, CA) for 1 h, 3) 1:250 TSA Biotin System (tyramide signal amplification; PerkinElmer Life Sciences, Norwalk, CT) in PBS containing 0.003% H₂0₂ for 10 min, and 4) streptavidin conjugated to Cy3 (Jackson) or Alexa 555 (Molecular Probes, Eugene, OR) for 30–60 min (Table 1). To visualize A15 DA neurons, sections were next incubated in 4% blocking sera (normal donkey serum for GAD-stained sections; normal goat serum for others) for 1 h, followed by 1:5000 mouse anti-TH for 24 h at RT and then 1:200 donkey or goat antimouse conjugated to Alexa 488 (Jackson). Sections were washed in PB, mounted on slides, and coverslipped with gelvatol to prevent fading. Primary antisera used have been already characterized (34,35,36,37). Omission of one primary antigen had no effect on fluorescent signal from the other antigen but eliminated fluorescent signal of the antigen targeted by the primary antibody.

Table 1.

Primary reagents used for immunocytochemistry for GABA-related antigens

Reagent	GAD67	GABAA-Rα1	GABAB-R1
Blocking sera	4% NGS in PBS	4% NDS in PBS	4% NDS in 0.4% TxBS
Primary antisera	Rabbit anti-GAD 1:5000; 48 h at 4 C	Mouse anti-GABAA 1:55,000; 24 h at RT	Guinea-pig anti GABAB 1:100; 24 h at RT
Biotinylated secondary antisera	Goat antirabbit	Donkey antimouse	Donkey anti-guinea pig
Tyramide amplification	Vector ABC-Elite, TSA Biotin System	Vector ABC-Elite, TSA Biotin System	Vector ABC-Elite, TSA Biotin System
Fluor	CY3-strepavidin 1:100; 30 min at RT	CY3-strepavidin 1:400; 1 h at RT	Alexa 555-strepavidin 1:100; 1 h at RT

Reagents were from Chemicon (anti-GAD and anti-GABA_A-Rα 1), Research Diagnostics (GABA_B-R1), Jackson ImmunoResearch Laboratories (second antibodies, Cy3-steptavidin), Vector Laboratories (Vector ABC-Elite), PerkinElmer Life Sciences (TSA Biotin System), Molecular Probes (Alexa 555-stepavidin). NDS, Normal donkey serum; NGS, normal goat serum; TxBS, Triton X-100 in buffered saline. 

Data analysis

LH pulses were identified using the following well established (38) criteria: 1) the peak of the pulse occurred within two samples of the previous nadir, 2) the amplitude of the peak was greater than the sensitivity of the assay, and 3) the LH concentration at the peak exceeded the 95% confidence limits (based on the variability of the standard curve) of the concentrations at both the preceding and subsequent nadirs. For OVX animals, LH pulse frequency, interpulse interval, and mean LH concentration were calculated for pre- and posttreatment periods. A within-subjects design was used to analyze effects of drug treatments. Specifically, mean LH and interpulse interval were analyzed using two-way ANOVA with repeated measures (interpulse interval was used to assess episodic secretion because it can be analyzed by parametric statistics, whereas pulse frequency cannot) using SigmaStat 3.11 software. The infrequent pulses in control ovary-intact anestrous ewes precluded use of interpulse interval. Therefore, LH pulse frequency and mean LH concentration were determined for each ewe during the 4-h posttreatment period of ovary-intact ewes. Significant effects of receptor ligands (compared with controls) on LH pulse frequency in ovary-intact ewes were determined by the Friedman two-way ANOVA by ranks, a nonparametric within-subject analysis, and the effects on mean LH concentrations were determined by ANOVA, with repeated measures.

Images of at least 12 TH-immunoreactive neurons per section per ewe and either GAD67 (A15 only), GABA_A-Rα1 (A15, A14, and A12 regions), or GABA_B-R1 (A15, A14, and A12 regions) were acquired using a Carl Zeiss Axiovert 100M Laser Scanning Microscope (Carl Zeiss International, Thornwood, NY) operated by LSM 510 software (confocal microscopy; Carl Zeiss). Scans at each appropriate wavelength were done independently to avoid bleed-through between channels. Two sets of images were taken along the z-plane (z-stack) for each neuron, consisting of 18–22 images in each channel that sliced the neuron from top to bottom at intervals of 1–1.5 μm determined by the confocal software based on the size of the neuron. These z-stacks were then converted to a series of tif files. For GAD67-containing varicosities, close contacts were identified on individual z-sections, and the percentage of DA neurons with at least one close contact and the number of close contacts per neuron were calculated. Receptor data are expressed as percentage of DA neurons that coexpressed the receptor subunit. For percentages, data were transformed (arcsine of square root) and then compared between season and hypothalamic area by two-way ANOVA.

Results

Experiment 1: effects of A15 microimplants of GABA receptor antagonists and NOS inhibitors

As a first test of the role of GABA and NO in the control of A15 neural activity, we examined the effects of microimplants of GABA receptor antagonists and NOS inhibitors, containing about 250 μg drug, in the A15 on episodic LH secretion in ovary-intact anestrous ewes. Seven of 11 ewes had correct placements of microimplants in the A15. For this and subsequent experiments, data from misplaced treatments were excluded from the analysis. As expected, when ewes received control (empty) microimplants LH pulse frequency was less than one pulse/4 h (Fig. 1), which is typical of ovary-intact anestrous ewes. Phaclofen (GABA_B receptor antagonist) microimplants appeared to induce LH pulses in five animals and significantly (P < 0.05) increased LH pulse frequency, an effect that might be due to stimulation of endogenous GABA release via GABA_B autoreceptors (27,39). However, this stimulatory action of phaclofen was modest because it did not alter mean LH concentration (Fig. 1). Microimplants containing bicuculline (GABA_A receptor antagonist) or the nonspecific (l-NAME) or neuron-specific (SMTC) NOS inhibitors had no affect on LH pulse frequency or LH concentrations in these ewes. Microimplants containing NOS inhibitors were empty when removed from the ewes, whereas crystalline GABA receptor antagonists were still visible in the lumen at the end of treatments. Most of the animals treated with bicuculline had increased motor activity while the microimplants were in place; no behavioral effects were observed with phaclofen or the NOS inhibitors. Misplaced sites were rostral, caudal, or through the base of the brain; treatments in these areas produced no alterations in LH secretion (data not shown).

Figure 1.

Effects of A15 microimplants containing antagonists (ANTAG) to GABA_A (shaded bars) receptors (bicuculline), GABA_B (solid bars) receptors (phaclofen), and the NOS inhibitors (INHIBIT) (striped bars), l-NAME and SMTC, on LH pulse frequency (top panel) and mean LH concentrations (bottom panel) in ovary-intact anestrous ewes. *, P < 0.05 vs. control (Cont) treatment; there were no significant differences in mean LH (F = 1.2; P > 0.1).

Experiment 2: effect of A15 microinjections of GABA receptor agonists in ovary-intact ewes

Because the GABA_B receptor can sometimes function as an autoreceptor (27,39), the effects of phaclofen in experiment 1 could have been due to pharmacological stimulation of endogenous GABA release. Therefore, we next tested the effects of local administration of GABA receptor agonists into the A15 region on LH secretion using another group of ovary-intact anestrous ewes. Preliminary work indicated that microimplants of GABA agonists produced partial sedation. Therefore, drugs were administered as bilateral microinjections of 300 nl containing 10 μg drug/μl sterile water (n = 9), because this concentration of GABA receptor agonists altered LH secretion when injected into the POA of anestrous ewes (30). Specifically, water (control), GABA (3 μg; 28.6 nmol), the GABA_A receptor agonist musimol (3 μg; 24.7 nmol), or the GABA_B receptor agonist baclofen (3 μg; 12.0 nmol) were injected into all ewes in random order, with 4 d between treatments. No sedation or other behavioral effects were observed with any microinjection.

All seven ewes with microinjections in the A15 area (Fig. 2) responded to either the GABA_A or GABA_B receptor agonist with an increase in episodic LH secretion above that seen with control treatments (Fig. 2). Both agonists significantly increased both LH pulse frequency (P < 0.05) and mean LH concentrations (F = 5.85; P < 0.01). The first LH pulse occurred with a considerable delay after microinjection of either the GABA_A (68.6 ± 12.2 min) or the GABA_B (46.2 ± 7.6 min) receptor agonist. However, once episodic LH secretion had started, these agonists produced LH pulse frequencies (GABA_A: 2.7 ± 0.2 pulses/3 h; GABA_B: 2.3 ± 0.3 pulses/3 h) similar to those seen in OVX ewes (see Table 2). Microinjection of GABA increased episodic LH release in only three of seven ewes and did not significantly increase LH concentrations or pulse frequency (Fig. 2). Misplaced microinjections (n = 2, both rostral to the A15, open circles in Fig. 2) had no effect on LH release (data not shown).

Figure 2.

Left, Location of microinjection sites in experiments 2 and 3 shown on schematics of coronal sections though the ovine RCh area. Pairs of circles depict bilateral location in a single ewe. Missed microinjections, represented by open circles, were rostral to the A15 (A). Filled circles depict microinjections in either the central region (B, n = 5) or caudal region (C, n = 2) of the A15 cell group. Right, Effect of A15 microinjections (arrows) of water, GABA, the GABA_A agonist muscimol, and the GABA_B agonist baclofen on episodic LH secretion in a representative ovary-intact anestrous ewe. Solid circles depict peaks of LH pulses. Bars on the right depict mean (±sem, n = 7) LH pulse frequency (FREQ, top panel) and LH concentration (bottom panel) after treatment of ovary-intact ewes with A15 microinjections of water (C), GABA (G), muscimol (A), or baclofen (B). *, P < 0.05 vs. control (water) treatments. f, Fornix; OCh, optic chiasm; ot, optic tracts; PVN, paraventricular nucleus; V, third ventricle; VMH, ventromedial hypothalamus.

Table 2.

Effect of microinjections of GABA receptor agonists and antagonists into the A15 area on LH secretion in OVX anestrous ewes

Treatment	Pulse frequency (no./3 h)		Mean LH (ng/ml)
	Pre	Post	Pre	Post
Experiment 3
Control	2.9 ± 0.3	2.7 ± 0.3	8.7 ± 1.0	9.0 ± 1.1
Muscimol	2.9 ± 0.3	3.1 ± 0.3	11.5 ± 1.9	10.2 ± 1.3
Baclofen	3.1 ± 0.3	3.3 ± 0.3	11.1 ± 2.6	11.5 ± 1.9
Bicuculline	2.4 ± 0.2	2.1 ± 0.3	9.6 ± 1.6	7.2 ± 1.3
Phaclofen	2.3 ± 0.2	2.6 ± 0.2	8.6 ± 2.0	9.1 ± 1.8
Experiment 4
Control	2.4 ± 0.3	2.4 ± 0.4	9.5 ± 1.5	12.8 ± 1.6
CGP52432	2.4 ± 0.2	2.4 ± 0.2	9.2 ± 2.2	10.4 ± 1.7

Results are shown as mean ± sem. Pre, Before microinjection; Post, after microinjection. 

Experiment 3: effects of A15 microinjections of GABA agonists and antagonists in OVX ewes

The ability of GABA agonists to stimulate LH secretion observed in experiment 2 raised the possibility that endogenous release of GABA inhibits A15 DA neurons. If this is the case, then E₂ should inhibit these GABAergic afferents because this steroid stimulates A15 neurons. Consequently, GABA release would be expected to be low in ovary-intact ewes and elevated in OVX animals. Thus, administration of GABA agonists that increased LH secretion when injected into the A15 of ovary-intact ewes should have little, or no, effect in OVX animals, whereas GABA antagonists should inhibit LH secretion. To test these predictions, we OVX the animals used in experiment 2. Starting 2 wk later, blood samples were collected for 3 h before and 3 h after bilateral microinjection of 300 nl sterile water containing muscimol (3 μg; 24.7 nmol), baclofen (3 μg; 12.0 nmol), bicuculline (1.5 μg; 3.4 nmol), or phaclofen (2.7 μg; 10.8 nmol) into the A15 (doses of antagonists were decreased due to low solubility in water) using a Latin-square design with 4–5 d between treatments.

LH pulse patterns with control treatments (and before drugs) showed the relatively slow frequency (about one pulse every 70–90 min) pattern typical of OVX ewes in the anestrous season (Table 2). In the seven ewes with properly placed injections, neither agonist nor antagonist significantly altered interpulse interval (F = 1.64; P > 0.1 for interaction of time × treatment) or mean LH concentrations (F = 0.28; P > 0.8 for interaction of time × treatment). Two animals given bicuculline were briefly hyperactive, but there were no obvious behavioral effects of the other drugs.

Experiment 4: effects of A15 microinjections of a more potent GABA_B receptor antagonist in OVX ewes

The lack of effects of GABA receptor agonists in OVX ewes (experiment 3) are consistent with the hypothesis that endogenous GABA release is high in these animals, but the lack of effects of the GABA antagonists is not. One possible explanation for this discrepancy is that an insufficient dose of GABA receptor antagonist was given. This seems unlikely with the GABA_A antagonist because it produced some behavioral effects but could account for the lack of effects of phaclofen, which is a fairly weak agonist (40). Therefore, we tested the effects of a more potent GABA_B antagonist, CGP52432 (40,41) using another group of ewes.

We first tested the effects of local administration of baclofen in ovary-intact animals (n = 5 with correct microinjections) to confirm earlier data. Microinjections of this GABA_B receptor agonist (3 μg; 12.0 nmol) again significantly increased LH pulse frequency (2.8 ± 0.2 pulse/4 h) above that observed with vehicle control (1.2 ± 0.4 pulse/4 h). Animals were then OVX and, starting 17 d later, the effects of CGP52432 (1.5 μg; 7.8 nmol) were compared with vehicle (sterile water) using a crossover design. Blood samples were collected for 2 h before and 3 h after injection of water or CGP52432. Similar to phaclofen, this GABA_B receptor antagonist had no effect (Table 2) on LH pulse frequency (F = 0.15; P > 0.6 for interaction of time × treatment), or mean LH concentrations (F = 1.6; P > 0.2 for interaction of time × treatment).

Experiment 5: effects of combined blockade of GABA_A and GABA_B receptors in the A15 region in OVX ewes

Because A15 neurons responded to both the GABA_A and the GABA_B receptor agonist (experiment 2) and appear to contain both receptors (see experiments 7 and 8), it may be necessary to block both GABA_A and GABA_B receptors simultaneously to relieve GABAergic inhibition and activate sufficient DA release to inhibit GnRH and LH secretion in OVX ewes. Therefore, we tested the effects of microinjection of a combined dose of bicuculline (2.5 μg/μl) and CGP52432 (2.5 μg/μl) into the A15 of OVX ewes in two replicates. In the first replicate, performed in summer of 2005, blood samples were collected every 12 min for 2 h before and 3 h after bilateral injection of 300 nl water or antagonists (1.7 nmol bicuculline plus 1.9 nmol CGP52432) into five OVX ewes, using a crossover design. The second replicate, done in eight OVX ewes during the summer of 2006, was similar in design except that, based on the results of the first experiment, the injection volume was increased to 400 nl (2.3 nmol bicuculline plus 2.7 nmol CGP52432).

In the first replicate, the antagonists tended to decrease LH concentrations (from 8.7 ± 0.7 to 5.7 ± 1.4 ng/ml; F = 3.5; P = 0.083 for interaction of time × treatment) and LH pulse frequency (from 2.6 ± 0.4 to 1.3 ± 0.7 pulses/3 h; F = 3.42; P = 0.088 for interaction of time × treatment by ANOVA of interpulse interval) in the four ewes with microinjections in the A15, whereas microinjections of vehicle produced no obvious changes in LH pulse patterns (Fig. 3). However, LH pulses were completely abolished in the two ewes with injections in the center of the A15 (Fig. 3) but were not obviously affected in the other two ewes, with injections in either the rostral or caudal portion of the A15. In the second replicate, the combined antagonists significantly reduced mean LH concentrations and decreased LH pulse frequency in the five ewes with correct placements in the A15 (Fig. 4). No effects of antagonists were observed in ewes with injections that missed the A15 (one ewe in the first replicate, two in the second). If data from the two replicates were combined, there was a significant effect of placement of the microinjection within the A15 (Fig. 5), with injections in the center of the A15 inhibiting LH pulse frequency (F = 17.1; P = 0.001 for time × treatment interaction) and LH concentrations (F = 5.55; P = 0.036 for time × treatment interaction), whereas those on the periphery did not (F = 0.97; P > 0.3 for time × treatment interaction on interpulse interval).

Figure 3.

Left, LH pulse patterns before and after microinjection (arrows) of water (left panels) or the combination of bicuculline and CGP52432 (middle panels) into the central portion of the A15 of representative OVX anestrous ewes. Right panels, LH pulse patterns in representative ewes receiving the combination of antagonists into either the caudal (Caud) or rostral (Rost) portion of the A15. Top panels present data from the first replicate, whereas data from the second replicate of experiment 5 are shown in the bottom panels. Solid circles are peaks of LH pulses. Ant, Combined antagonists; Mid, injection in middle portion of A15; Veh, vehicle.

Figure 4.

A combination of GABA_A and GABA_B antagonists inhibits LH pulse frequency in OVX anestrous ewes. Bars depict mean (±sem, n = 5) LH pulse frequency (left panel) and LH concentration (right panel) before (open bars) and after (solid bars) microinjection of water (Con) or bicuculline plus CGP52432 (A+B) into the A15 of OVX anestrous ewes. *, P < 0.05 vs. pretreatment values.

Figure 5.

Left, Distribution of microinjection sites within the A15 shown on schematic coronal sections through the A15 region (experiment 5). Microinjections in the rostral (A.5) and caudal (C) portions of the A15 are depicted by shaded circles, whereas microinjections in the central portion of the A15 (B) are shown by solid circles. Note that the rostral panel is posterior to A in Fig. 2. Location of missed sites is not illustrated. Right, Only microinjections in center of A15 inhibit LH pulses. Bars depict mean (±sem) pulse frequency and LH concentration before (open bars) and after (shaded and solid bars) microinjection of water (Con) or bicuculline plus CGP52432 (A+B) into either the peripheral (PERI) or central region of the A15 of OVX anestrous ewes. *, P < 0.05 vs. pretreatment values.

Experiment 6: GAD67-immunoreactive close contacts on A15 neurons

Every DA (i.e. TH-immunopositive) neuron examined had at least one GAD-positive close contact (Fig. 6), with more close contacts on dendritic processes (9.0 ± 2.0, n = 6 ewes) than on cell soma (5.1 ± 0.2). Not surprisingly, there was also a relatively widespread distribution of GAD67-containing varicosities that were not associated with TH immunoreactivity (Fig. 6). The number of GAD67-containing close contacts on A15 DA neurons in OVX, E₂-treated breeding-season ewes (dendrites: 7.8 ± 0.8; cell soma: 4.6 ± 0.6 close contacts, n = 5 ewes) was not significantly different from those observed during anestrus.

Figure 6.

GABAergic neurons make close contacts (arrows) onto A15 neurons. Confocal images of a section through the A15 of an OVX, E₂-treated anestrous ewe processed using dual ICC for TH (green) and GAD67 (red). Scale bar, 20 μm.

Experiment 7: GABA_A receptors on DA neurons

When observed, GABA_A receptor immunoreactivity was clearly evident in the cell soma and sometimes extended to the primary dendrites of TH-positive neurons (Fig. 7). All DA cell groups (60–70 cells per group per ewe) examined exhibited coexpression of GABA_A receptors, albeit to varying degrees (Table 3). More than 80% of TH-immunoreactive neurons in the A12 coexpressed GABA_A receptors, whereas 60% of TH-immunoreactive neurons in the A14 coexpressed GABA_A receptors. In contrast, approximately 30% of A15 DA cells contained the GABA_A receptor. There was no difference in the percentage of TH-immunoreactive neurons coexpressing GABA_A receptors between the breeding season and anestrus (Table 3) in any DA cell population investigated (n = 6 ewes per season).

Figure 7.

A15 neurons contain GABA_A (top panels) and GABA_B (bottom panels) receptors. Confocal images of a section from the A15 region of an OVX, E₂-treated anestrous ewe processed using dual ICC for TH (left panels) and GABA_A-Rα1 (top middle panel) or GABA_B-R1 (bottom middle panel). Right panels depict computer-generated overlay of other two images illustrating localization of GABA_A (top) or GABA_B (bottom) receptors in A15 neurons (yellow). Scale bar (top left panel), 20 μm.

Table 3.

Percentage of perikarya containing GABA_A-Rα 1 in DA cell groups in anestrus and breeding season

	Anestrus	Breeding season
A12 group	84.7 ± 0.9%	86.5 ± 1.2%
A14 group	64.2 ± 9.1%	64.6 ± 5.3%
A15 group	19.9 ± 3.1%	27.0 ± 3.1%

Results are shown as mean ± se. 

Experiment 8: GABA_B receptors on DA neurons

We examined GABA_B receptor expression in 53 DA neurons in the A15 group per ewe, 20 DA neurons per ewe in the A14 group, and 27 DA neurons per ewe in the A12 group. Like the GABA_A receptor, the GABA_B-R1 receptor was evident primarily in TH-positive cell somata (Fig. 7). However, in contrast to GABA_A receptors, expression of the GABA_B-R1 receptor was observed in almost all (>95%) of the A15 DA neurons (n = 2 ewes). Similar, but slightly lower, expression of the GABA_B receptor was observed in A14 (90%) and A12 (79%) DA neurons.

Discussion

The results of these studies provide strong pharmacological and anatomical evidence that GABAergic neurons play an important role in controlling the activity of the A15 DA neurons that mediate E₂ negative feedback in anestrous ewes. In contrast, we found no evidence that NO plays a role, because the local administration of NOS inhibitors to the A15 had no effect on LH secretion in the first experiment. It is unlikely that a lack of effect is due solely to an inadequate dose of this inhibitor because the same dose of l-NAME stimulated LH secretion when given to the vmPOA of anestrous ewes in experiments run in parallel with this work (42). Thus, although NO apparently does not act within the RCh to mediate the effects of E₂ on A15 neurons, it may play a role elsewhere. In light of these considerations and the effects of GABA receptor agonists, subsequent experiments focused on the possible role of GABA.

Both the GABA_A and the GABA_B receptor agonist produced a dramatic increase in episodic LH secretion when injected into the A15 of ovary-intact anestrous ewes. These results indicate that GABA can, directly or indirectly, suppress the activity of A15 neurons, which are known to inhibit episodic GnRH secretion during anestrus. It is interesting to note that each agonist increased LH pulse frequency to levels seen in OVX ewes, suggesting that this local treatment completely disrupted E₂ negative feedback in ovary-intact ewes. The stimulatory actions of the GABA_B antagonist phaclofen and the lack of effect of GABA itself appear in conflict with this interpretation. The former could be due to a nonspecific effect caused by a high concentration of phaclofen produced by the crystalline microimplants used in the first experiment. Alternatively, based on previous work in rams (27,31,43,44), the stimulatory effects of phaclofen could be due to the presence of GABA_B autoreceptors on GABA terminals in this region. Because these autoreceptors are thought to be inhibitory, the GABA_B receptor blocker may stimulate endogenous GABA release, which would then inhibit A15 neurons via GABA_A receptors. The lack of effect of GABA could be due to an insufficient dose or to endogenous systems designed to maintain low extracellular levels of this neurotransmitter. GABA is actively taken up by both neurons and glia, and the presence of GABA autoreceptors would also tend to produce constant extracellular GABA concentrations in response to exogenous GABA. This explanation is supported by reports that exogenously induced increases in GABA concentrations returned to baseline within a few minutes after termination of treatment (45,46); in contrast, the half-life of baclofen (47) and muscimol (48) in the central nervous system are 5 and 2 h, respectively. Thus, despite the negative data with GABA, the potent stimulation of LH secretion produced by the selective GABA receptor agonists, together with the anatomical evidence for GABA receptors on A15 neurons, supports the hypothesis that inhibitory GABA input contributes to the control of A15 neuronal activity.

If GABAergic input mediates the stimulatory actions of E₂ on A15 neurons, then E₂ must inhibit the release of GABA from these afferents. One would thus predict that endogenous GABA release is low in ovary-intact ewes and that GABA levels rise after OVX; this OVX-induced rise in GABAergic tone would then inhibit A15 neural activity and allow episodic GnRH and LH secretion to increase. The dramatic increase in LH pulses produced by the GABA agonists in ovary-intact ewes is consistent with low endogenous GABA concentrations when E₂ is present because it indicates the presence of free receptors available to interact with the exogenous GABA agonists. Based on similar logic, one can infer that GABA release is elevated after OVX because the agonists failed to increase LH pulse frequency in these animals. One could argue that agonists were ineffective because pulse frequency is already maximal in these OVX ewes, but this is unlikely because LH pulse frequency in OVX anestrous animals can be further increased by photoperiodic (49) or other pharmacological (50,51) manipulations. Thus, the changes in response to GABA agonists with endocrine condition provide indirect evidence that endogenous GABA release is inhibited by E₂ during anestrus.

The inhibitory effects of the combination of GABA_A and GABA_B receptor antagonists further support the hypothesis that endogenous GABA is inhibiting A15 neural activity in OVX ewes. However, this system appears to be quite resistant to GABAergic blockade. Neither GABA receptor antagonist alone produced any inhibitory effect, and the combination produced a strong inhibition of LH only when injected in the middle of the A15. The former is consistent with the presence of both GABA_A and GABA_B receptors on A15 neurons, and the latter suggests that it is necessary to block GABA inhibition of a majority of A15 neurons to release sufficient DA to inhibit GnRH secretion. This resistance to GABA blockade could reflect either a high level of endogenous GABA release in the region of the A15 neurons or a decrease in responsiveness of GnRH neurons to inhibitory inputs that is known to occur in chronically OVX animals (52,53). Taken together, these pharmacological studies provide strong support for the hypothesis that in anestrus, E₂ negative feedback is mediated, at least in part, by the suppression of GABA release from afferents to the A15 DA neurons.

One potential problem in interpreting studies involving local administration of receptor agonists and antagonists to a ubiquitous neurotransmitter, such as GABA, is the possibility that any effects observed may be due to diffusion from the injection site. Although this is difficult to directly address without a detailed knowledge of local concentrations of endogenous GABA, levels of receptor ligands produced by microinjections and receptor affinities, there are several lines of evidence indicating that the drugs acted locally in the A15 in these experiments. First, missed injections of agonists produced no stimulatory effect in ovary-intact ewes even though they were within 1–1.5 mm of the A15. Similarly, the effectiveness of combined GABA_A and GABA_B receptor antagonists when administered to the middle, but not the rostral or caudal portion, of the A15 argues strongly for minimal effective spread of these drugs. Third, local administration to the ventromedial nucleus of GABA_A (54) or GABA_B (55) receptor agonists either suppressed (54) or had no effect (55) on GnRH/LH secretion in anestrous ewes. Because neither of these results were observed with A15 microinjections of GABA receptor agonists, these drugs are unlikely to have diffused from the A15 to the ventromedial nucleus. Finally, the anatomical observations demonstrate that the receptors are present on A15 DA neurons for local actions of both GABA_A and GABA_B receptor agonists and antagonists.

The observation that every TH-immunoreactive neuron within the A15 group received at least one close contact containing GAD67 provides strong evidence that the activity of these neurons is under the direct control of GABAergic afferents. Comparison of the mean number of GAD-positive close contacts with previous data on the total number of synapsin-containing close contacts (37) on to A15 neurons indicates that GABAergic input represents 25–30% of afferents to these neurons. Similarly, the presence of GABA_B-R1 receptors in 99% of A15 neurons indicates that these neurons can respond to GABA released from these afferents. Somewhat surprisingly, a much smaller percentage of A15 neurons expressed the GABA_A-Rα1 subunit. This low level is unlikely to be due to technical difficulties because receptor expression was evident in some, but not other, A15 neurons in the same section (Fig. 7), and a much higher level of expression was observed in both A14 and A12 DA neurons. The latter is consistent with evidence that local administration of GABAergic ligands in the vicinity of A12 neurons alters prolactin secretion in sheep (56). These data, of course, do not rule out the possibility that A15 neurons contain a different isoform of the α-subunit; the robust effects of the GABA_A receptor agonist suggest either that this is the case or that the subset of A15 DA neurons containing GABA_A-Rα1 are dedicated to inhibition of GnRH neurons.

Although these data provide anatomical support for direct effects of GABA on A15 neurons, they do not rule out the possibility of indirect actions of this neurotransmitter. There are clearly many GAD-positive varicosities within the A15 area that do not synapse on TH-containing neurons (Fig. 6). Thus, the GABA agonist could be acting via local interneurons. The relatively long delay (46–69 min) between microinjection of the GABA agonists and the first LH pulse in ovary-intact ewes is consistent with this possibility.

In addition to mediating E₂ negative feedback during anestrus, one other key property of A15 neurons is their seasonal nature; E₂ stimulates them in anestrus but not the breeding season (10,11,12). This characteristic largely accounts for the seasonal changes in response to E₂ negative feedback in the ewe (4,5,6,7) and may reflect seasonal differences in afferent input to these neurons (37). The lack of a seasonal change in GAD-positive close contacts and expression of GABA_A receptors on A15 neurons and the high percentage of these neurons containing GABA_B receptors in anestrus suggest that an increase in GABAergic afferents or increased responsiveness of A15 neurons to GABA may not play a major role in the suppression of A15 activity in the breeding season. This, of course, does not preclude an important functional change in this inhibitory input. For example, a simple explanation of the seasonal changes in response to E₂ negative feedback would be the loss of ERα in GABAergic afferents during the breeding season. If this occurred, E₂ would no longer be able to inhibit GABAergic tone so that A15 neurons would be inactive regardless of circulating E₂ concentrations, thus producing the loss of response to E₂ negative feedback in the breeding season. A test of this attractive hypothesis awaits procedures for identifying the GABAergic cell bodies afferent to the A15 neurons, for example, by combining retrograde tract tracing from the A15 with in situ hybridization for GAD. It is also possible that seasonal changes in other A15 afferents are more critical than changes in GABAergic tone. Glutamatergic neurons also synapse on to A15 neurons and appear to participate in E₂ negative feedback in anestrus (57). Moreover, preliminary work indicates that an increase in stimulatory glutamatergic tone may occur in anestrus because the number of glutamatergic close contacts on to A15 neurons is higher at this time of year (58).

An important role for GABA in steroid negative feedback has previously been proposed for rats (59,60,61), sheep (27,29,30,44,62), and primates (63,64,65). In most other systems, however, GABA has been proposed to exert its effects directly on GnRH neurons, not indirectly via another set of inhibitory neurons. Although there is strong evidence for such a role in male rodents (60,61), the data in many other models are difficult to interpret because both receptor agonists and antagonists inhibit LH secretion (29,66,67), paradoxical stimulatory effects of GABA_B receptor agonists are sometimes observed (27,55,56), and there is considerable controversy as to whether GABA inhibits (68) or stimulates (69) GnRH neurons in adult rodents. Perhaps the closest parallel to the anestrous ewe is the rhesus monkey in which it has been proposed that GABA mediates the enhanced negative feedback action of E₂ before puberty, with a decrease in GABAergic tone contributing to the onset of the menarche (63,64,65). Interestingly, as may be the case in the ewe, reciprocal changes in glutamatergic and GABAergic input may both be involved in the onset of reproductive function in female rhesus monkeys. It is important to point out, however, that because GABA and glutamate appear to act indirectly via inhibitory A15 neurons in the ewe, any changes would be the inverse of those in the primate (i.e. low GABAergic and high glutamatergic tone induces infertility in sheep).

In conclusion, the results of these studies provide strong evidence that GABAergic afferents to A15 DA neurons participate in E₂ negative feedback in anestrous ewes. The pharmacological data suggest that GABA inhibits A15 activity and that GABAergic tone is in turn suppressed by E₂ to stimulate A15 neurons, which then inhibit GnRH secretion. The anatomical data demonstrate that A15 neurons receive GABA input and have the receptors to respond to this neurotransmitter. However, this anatomical input does not appear to change with season. Whether seasonal changes in GABAergic tone are important to the seasonal variation in A15 responsiveness to E₂ awaits further study.