Activation of A-Type γ-Aminobutyric Acid Receptors Excites Gonadotropin-Releasing Hormone Neurons

Abstract

γ-Aminobutyric acid (GABA), acting through GABA_A receptors (GABA_AR), is hypothesized to suppress reproduction by inhibiting GnRH secretion, but GABA actions directly on GnRH neurons are not well established. In green fluorescent protein-identified adult mouse GnRH neurons in brain slices, gramicidin-perforated-patch-clamp experiments revealed the reversal potential (E_GABA) for current through GABA_ARs was depolarized relative to the resting potential. Furthermore, rapid GABA application elicited action potentials in GnRH neurons but not controls. The consequence of GABA_AR activation depends on intracellular chloride levels, which are maintained by homeostatic mechanisms. Membrane proteins that typically extrude chloride (KCC-2 cotransporter, CLC-2 channel) were absent from the GT1–7 immortalized GnRH cell line and GnRH neurons in situ or were not localized to the proper cell compartment for function. In contrast, GT1–7 cells and some GnRH neurons expressed the chloride-accumulating cotransporter, NKCC-1. Patch-clamp experiments showed that blockade of NKCC hyperpolarized E_GABA by lowering intracellular chloride. Regardless of reproductive state, rapid GABA application excited GnRH neurons. In contrast, bath application of the GABA_AR agonist muscimol transiently increased then suppressed firing; suppression persisted 4–15 min. Rapid activation of GABA_AR thus excites GnRH neurons whereas prolonged activation reduces excitability, suggesting the physiological consequence of synaptic activation of GABA_AR in GnRH neurons is excitation.

SUPPRESSION OF FERTILITY in physiological and pathological situations involves inhibition of GnRH neurons, the final common pathway for reproductive function. Models addressing regulation of GnRH neurons by sex steroids and other factors often postulate changes in synaptic inputs to these cells. In many models, γ-aminobutyric acid (GABA) plays a primary role in the transsynaptic control of GnRH neurons (1, 2). GABA is the dominant inhibitory neurotransmitter in the hypothalamus (3), GABAergic neurons synapse on GnRH neurons (4), and there are functional GABA_A receptors on these cells (5). Although excitatory responses to GABA were reported in embryonic (6) and transformed GnRH neurons (7, 8), this was largely attributed to the immature nature of those cells, and an underlying assumption has been that activation of GABA_A receptors on GnRH neurons in adults inhibits these cells as it does in the vast majority of adult neurons (9).

GABA_A receptors are ligand-gated ion channels permeable primarily to chloride and, to a small extent, bicarbonate (9). The action of GABA at GABA_A receptors thus depends on membrane potential and intracellular chloride concentration. The latter is set by developmentally regulated chloride homeostatic mechanisms. In most mature neurons, low intracellular chloride (<10 mm) is maintained by a neuron-specific potassium-coupled chloride cotransporter, KCC-2, which is widely expressed in the adult brain, but present at low levels in the fetal/neonatal brain (10–13). Under physiological conditions, KCC-2 causes chloride extrusion, resulting in a reversal potential (E_GABA) that is near or below the resting potential so that activation of GABA_A receptors at the resting potential results in decreased excitability through shunting inhibition, or chloride influx and inhibition via membrane hyperpolarization, respectively (9). The hyperpolarization-activated chloride channel CLC-2 also depletes intracellular chloride (14). In contrast, the sodium-potassium-2-chloride cotransporter (NKCC-1) typically mediates accumulation of intracellular chloride, depolarizing E_GABA (15). In most brain regions, NKCC-1 is expressed at higher levels in the fetal/neonatal period than in adulthood (e.g. Ref. 13).

Given the emphasis that many hypotheses have placed on inhibition of GnRH neuron activity by activating GABA_A receptors, we examined the direct actions of GABA on GnRH neurons using brain slice electrophysiological techniques, and defined the expression of transporters and channels involved in chloride homeostasis using molecular and immunohistochemical approaches. Our results show that, in marked contrast to the vast majority of central nervous system neurons, GABA_A receptor activation depolarizes GnRH neurons beyond threshold regardless of the developmental or reproductive status of the animals. We conclude that synaptic activation of GABA_A receptors excites GnRH neurons due to homeostatic mechanisms that maintain elevated intracellular chloride in these cells.

RESULTS

Monitoring Recording Quality

The gramicidin-perforated patch-clamp method is a technique for assessing the native intracellular chloride concentration (16). After formation of the GΩ seal, seal/input resistance, access resistance, and capacitance were measured in a voltage clamp every 10 sec for 5–30 min to monitor insertion of the pores and every few minutes throughout the recordings to check quality (Fig. 1). Figure 1 shows changes in these parameters over time, with the gray area indicating when recordings were considered to be of sufficient quality for measurements to be made. Membrane potential (17) was not a reliable measure of the quality of perforated patch recordings, as stable membrane potentials were obtained at access resistances in excess of 500 MΩ, and there was little change in membrane potential over 5-fold decrement in access resistance to an acceptable quality of 100 MΩ.

Figure 1.

Monitoring of Perforated Patch Recording Quality A, Current in response to a 5-mV hyperpolarizing voltage step at different stages of pore insertion. B, Access resistance (top), seal/input resistance (middle), and reversal potential (bottom) during a gramicidin-perforated patch recording. The gray area indicates the time when recordings were of sufficient quality for measurement of reversal potential. Numbers (1, 2, 3, 4 ) indicate when traces in A were obtained. The pipette contained 1 mm chloride. See Results for details. WC, Whole cell.

E_GABA in GnRH Neurons Is Depolarized Relative to Rest and Action Potential Threshold

Gramicidin-perforated voltage-clamp recordings were used to measure the reversal potential of currents evoked by local pressure application of GABA to GnRH neurons (E_GABA) in the presence of tetrodotoxin, d-(−)-2-amino-5-phosphonopentanoic acid (APV), and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) (Fig. 2, A and C). GnRH neurons were identified in acutely prepared slices through the hypothalamus of transgenic mice in which green fluorescent protein (GFP) is targeted to GnRH neurons under the control of the GnRH promoter (GnRH-GFP mice). Previous characterization of these mice demonstrated that the fidelity of GFP identification of GnRH neurons was greater than 99% (18). E_GABA measured 100 msec after GABA application was −36.5 ± 1.2 mV (n = 16 cells from 13 adult diestrous females). This was positive relative to the resting potential (−50.7 ± 1.7 mV, from current-clamp measurements), indicating GABA_A receptor activation would depolarize GnRH neurons. We tested the effects of rapid GABA on membrane potential in current-clamp recordings using gramicidin. Membrane potential peaked at −35.3 ± 1.5 mV, a 15.4 ± 1.2 mV depolarization from rest (n = 15 in the presence of tetrodotoxin). In separate gramicidin current-clamp recordings in the absence of tetrodotoxin, rapid GABA consistently evoked action potentials (n = 6, Fig. 2G). In contrast to GnRH neurons, E_GABA was negative relative to the resting potential in non-GnRH neurons of the hypothalamus (E_GABA −75.2 ± 2.9 mV, V_rest −48.8 ± 5.3 mV, n = 4), and rapid GABA hyperpolarized these cells to an average membrane potential of −74.0 ± 3.0 mV (−25.2 ± 6.5 mV hyperpolarization from rest −48.8 ± 5.3 mV, n = 4).

Figure 2.

GABA Reversal Potential Is Depolarized in Adult GnRH Neurons Representative gramicidin-perforated patch (A) and whole-cell recording with 1 mm pipette chloride (B) of a GnRH neuron from an adult diestrous animal (age 83 d). The voltage protocol consisted of steps every 15 mV from −70 to +5 mV for gramicidin and from −120 to −80 mV for whole cell. Open circles indicate where baseline current was measured. GABA (1 mm) was briefly (5 msec) applied via a pressure pipette positioned near the cell at the time indicated by the arrow; filled circles indicate when current after GABA application was measured. C and D, Plot of current through the GABA_A receptor as a function as step potential for the GnRH neuron shown in A and B, respectively. E, Chloride reversal in GAD-GFP neurons is not depolarized, but rather is hyperpolarized relative to the resting potential of these neurons; voltage protocol was every 10 mV from −100 to −50 mV; details are as in panels A and B. F, Plot of current through the GABA_A receptor as a function as step potential for the GAD-GFP neuron in panel E. G, Current-clamp recording in the gramicidin-perforated patch configuration showing the depolarizing response and action potential firing of a GnRH neuron (top) and the hyperpolarization of a GAD-GFP neuron (bottom). All voltages reported are corrected for series resistance and liquid junction potential. Panels A–F were recorded in the presence of tetrodotoxin, APV, and CNQX; panel G was recorded in the presence of APV and CNQX only.

The depolarized reversal potential observed in GnRH neurons was not due to accidental rupture of the patch and subsequent dialysis of the cell with the pipette solution as results with 140 mm or 1 mm chloride in the pipette solution were indistinguishable (E_GABA 1 mm Cl⁻, −35.2 ± 1.7 mV, n = 9; 140 mm Cl⁻, −38.2 ± 1.8 mV, n = 7, P > 0.2 two-tailed t test). When the patch was intentionally ruptured and the GnRH neuron dialyzed with the 1 mm chloride solution, E_GABA was hyperpolarized to −89.0 ± 4.1 mV (n = 4, Fig. 2B). Although GnRH neurons have been reported to be hyperpolarized after activation of GABA_B receptors (19), the time course of that G protein-coupled response is too slow to contribute to the currents at the time when E_GABA was measured; consistent with this notion, inclusion of the GABA_B antagonist SCH50911 (100 μm, Tocris, Ellisville, MO) did not alter E_GABA in GnRH neurons (n = 4). Another possible explanation for a depolarized E_GABA is that bicarbonate ions can pass through the GABA_A receptor channel (9). This only happens, however, after strong activation of the GABA_A receptor has neutralized the driving force on chloride ions, revealing depolarizing, outward bicarbonate flux (20). Biphasic responses to GABA were not observed in any GnRH neuron studied in voltage clamp; furthermore, replacement of extracellular bicarbonate with a HEPES-based buffer, which reduces the influence of bicarbonate, did not alter the depolarization observed (n = 2). The intracellular chloride levels estimated from E_GABA are shown in Table 1.

Table 1.

Estimation of Intracellular Chloride from E_GABA Using the Nernst Equation and a Chloride Activity Coefficient of 0.76

Cell type (n)	[Cl−]i (mm)
GnRH (untreated) (16 )	25.6 ± 0.3
GnRH + bumetanide (4 )	11.5 ± 0.9a
Non-GnRH (4 )	5.8 ± 0.4a
GAD-GFP (4 )	7.7 ± 0.1a

^a

P < 0.001 vs. untreated GnRH neurons.

Although GFP is widely used as a probe to investigate living systems without affecting their function, it is conceivable that its presence in GnRH neurons could be affecting our measures of E_GABA. The normal number of GnRH neurons in the GnRH-GFP mice (18) argues against chemotoxic effects observed in some systems (21); furthermore, a recent report (22) demonstrated normal chloride homeostasis in the presence of two spectral variants of GFP. Phototoxic effects stemming from illuminating the tissue to reveal GFP are also important to consider, as exposure of any tissue to either room light or fluorescent illumination will generate free radicals regardless of the presence of fluorescent markers [i.e. nonchemotoxic probes still contribute to phototoxicity (23)]. We used gramicidin-perforated patch to determine E_GABA for GAD-GFP neurons, which express the same form of GFP under the control of the glutamic acid decarboxylase (GAD)67 promoter, targeting GABAergic neurons of the cortex and hippocampus (24). E_GABA was −67.3 ± 0.7 mV in hippocampal GAD-GFP neurons, significantly hyperpolarized to the resting potential (−51.3 ± 2.9 mV, n = 4 adults, Fig. 2, E and F) as previously reported for this cell type in the absence of GFP (25). Consistent with this, GAD-GFP neurons were hyperpolarized by rapid GABA, rather than depolarized as GnRH neurons were (Fig. 2G). These data indicate that chloride homeostasis is not altered by either the expression or fluorescent illumination of the enhanced form of GFP present in these transgenic mice. The depolarized reversal potential and elevated intracellular chloride levels observed in GnRH neurons are thus normal physiological properties of these cells.

Expression of KCC-2, NKCC-1, and CLC-2 in GT1–7 Cells

The finding that GnRH neurons maintain an elevated intracellular chloride concentration led us to investigate which of the transporters and channels typically involved in chloride homeostasis are expressed in these cells. Initially these studies were performed in GT1–7 cells, an immortalized murine GnRH neuronal cell line that allows molecular analyses not possible on GnRH neurons in situ (26). GT1–7 cells do not express detectable levels of KCC-2 mRNA (Fig. 3A) or protein (Fig. 3F, left panels). Although GT1–7 cells contain both CLC-2 mRNA (Fig. 3B) and a CLC-2 protein species of correct molecular size (Fig. 3D), confocal microscopy revealed that this protein is not appropriately targeted to the cell membrane (Fig. 3F, upper middle and right panels), assuming instead a perinuclear location that would preclude its participation in cytoplasmic chloride homeostasis. In contrast to KCC-2, GT1–7 cells showed readily detectable levels of NKCC-1 mRNA (Fig. 3C), NKCC-1 protein species of the appropriate size (27) (Fig. 3E), and cellular expression of the NKCC-1 protein (Fig. 3F, lower middle and right panels). Thus, the depolarizing response of GT1–7 cells to GABA_A receptor stimulation (8) appears to be related to both the absence of chloride cotransporter/channels able to extrude intracellular chloride, and the presence of a cotransporter that accumulates intracellular chloride against its electrochemical equilibrium.

Figure 3.

Detection of KCC-2, CLC-2, and NKCC-1 mRNAs and Their Corresponding Protein Products in GT1–7 Cells by RNase Protection Assay, Immunoblotting, and Immunohistofluorescence Confocal Microscopy A–C, RNAse protection assays: KCC-2 mRNA is not detectable in GT1–7 cells (A), which instead express the mRNAs encoding CLC-2 (B) and NKCC-1 (C). D and E, Western blots: GT1–7 cells contain both CLC-2 (D) and NKCC-1 (E) proteins. In panel D, preadsorption of the CLC-2 antiserum with the CLC-2 peptide used as an antigen eliminates the CLC-2 band detected in both GT1–7 cells and liver. In panel E, the NKCC-1 protein migrates as an approximately 150-kDa species in GT1–7 cells, whereas in mouse brain there is an additional species of about 140 kDa, in agreement with published results (69 ). F, Immunohistofluorescence-confocal images: notice that the cells contain immunoreactive CLC-2, but that the protein is mostly localized to the perinuclear region, whereas NKCC-1 is unambiguously targeted to the cell membrane. Bars, 10 μm. M, RNA molecular markers; UP, undigested probe; DP, digested probe; GT, GT1–7 cells; GTn, GT1–7 RNA not hybridized to the cyclophilin probe; HC, hippocampus; CC, cerebral cortex; LV, liver; KD, kidney; PO, preoptic area; cyclo, cyclophilin; GT 25 and GT 5, 25 and 5 μg, respectively, of GT1–7 total RNA; r, rat; m, mouse.

Expression of KCC-2, NKCC-1, and CLC-2 in GnRH Neurons in Situ

To determine whether native GnRH neurons have a similar pattern of expression, we employed two complementary approaches, dual immunohistofluorescence-confocal microscopy and combined immunohistochemistry/in situ hybridization. The latter technique was performed in both rats and mice. This allowed us to 1) compare our data with the majority of studies describing the localization of these transporters in brain that have been performed in rats (e.g., Refs. 12–14 and 27), 2) extend our observations to an additional species, and 3) verify the rat and GT1–7 findings in mouse neurons in situ. Using dual immunohistofluorescence-confocal microscopy we examined the hypothalamic-preoptic region of 3-, 28-, and 60-d-old female rats and found that GnRH neurons were surrounded by CLC-2 positive cells, but they themselves were CLC-2 negative (Fig. 4A, GnRH neurons in red, surrounding CLC-2 positive neurons in green, denoted by arrows). Likewise, we did not detect KCC-2 in GnRH neurons but found KCC-2 immunostaining in adjacent non-GnRH cells (Fig. 4B, arrow). In contrast, some GnRH neurons, as well as some non-GnRH neurons, had low levels of NKCC-1 immunoreactivity (Fig. 4C, arrowhead and arrow, respectively). Many more cells, including some GnRH neurons, were NKCC-1 positive in 3-d-old brains (not shown). Omission of the first antibody to CLC-2, KCC-2, or NKCC-1 in the presence of GnRH antibodies abolished all chloride channel/cotransporter immunoreactivity (Fig. 4D).

Figure 4.

Detection of CLC-2, KCC-2, and NKCC-1 Immunoreactivity in GnRH Neurons of 60-d-Old Female Rats A and B GnRH neurons (red), located in the immediate vicinity of CLC-2 (A) or KCC-2 (B) positive non-GnRH cells (green, denoted by arrows), are devoid of CLC-2 or KCC-2 immunoreactive material. C, Example of a GnRH neuron containing NKCC-1 immunoreactivity (arrowhead), near a NKCC-1 expressing non-GnRH cell (arrow). D, Absence of chloride transporter immunostaining in a section incubated with GnRH antibody, but in the absence of either KCC-2 or NKCC-1 antibodies. Bar, 10 μm.

Using combined immunohistochemistry/in situ hybridization, we detected 223 GnRH-positive cells in one rat and 325 in the other, thus accounting for approximately 50% of the total GnRH neuronal population. Of these GnRH neurons, 304 were examined for KCC-2 expression (122 neurons in rat 1 and 182 in rat 2) and 244 for NKCC-1 expression (101 neurons in rat 1 and 143 in rat 2). Consistent with the pattern of expression previously observed in the adult rat brain (12, 27), KCC-2 transcripts were abundantly expressed in neurons throughout the entire region examined, whereas NKCC-1 transcripts were much less abundant. Figure 5 (middle) demonstrates that this differential mRNA expression persists in the anterior region of the preoptic area, which contains most of the neuroendocrine GnRH neurons (28) (Fig. 5, upper panels). Examination of individual GnRH neurons revealed that most were devoid of KCC-2 mRNA (Fig. 5, lower left panel, arrowheads), despite being surrounded by KCC-2 mRNA expressing cells (Fig. 5, lower left panel, examples denoted by short arrows). In contrast, GnRH neurons expressing low but discernible NKCC-1 hybridization were detected (Fig. 5, lower right panels, positive cells denoted by long arrows), despite being surrounded by non-GnRH neurons lacking NKCC-1 mRNA. Some GnRH neurons did not meet the criterion to be considered as NKCC-1 mRNA positive (arrowhead); some non-GnRH neurons were clearly positive (examples denoted by short arrows). Using criteria described in Materials and Methods, we estimated that 15 of a total of 304 cells from two animals (4.9%) contained KCC-2 mRNA. A significantly larger fraction of adult GnRH neurons (13.5%) contained NKCC-1 mRNA (33 of 244 neurons, P < 0.01 by χ² analysis). In no instance were GnRH neurons containing CLC-2 mRNA detected (not shown). The vast majority of KCC-2- and NKCC-1-expressing GnRH neurons were in the anterior region of the preoptic area where the greatest concentration of neuroendocrine GnRH neurons is normally observed (Fig. 5, upper panels) (28). Sections incubated with sense RNA probes did not exhibit specific cellular hybridization (not shown).

Figure 5.

Detection of KCC-2 and NKCC-1 mRNA in GnRH Neurons of the Adult (60-d-old) Rat Hypothalamus by Combined Immunohistochemistry (GnRH Neurons, Brown Color)/in Situ Hybridization (Silver Grains, White Color) Upper panels, Low magnification, bright-field view of GnRH neurons in the anterior preoptic region of the rat brain. Bars, 100 μm. Middle panels, Dark-field view of the same images depicted in the upper panels contrasting the striking abundance of KCC-2-containing cells with low level of NKCC-1 mRNA expression in this brain region. Lower panels, Higher magnification images of GnRH neurons illustrating that most of them have no detectable KCC-2 transcripts, whereas some are NKCC-1 mRNA positive. Arrowheads denote GnRH neurons lacking KCC-2 or NKCC-1 mRNA. Short arrows point to non-GnRH cells containing either KCC-2 or NKCC-1 transcripts; long arrows identify GnRH-NKCC-1-positive cells. Bars, 20 μm.

To verify that the mice have a pattern of chloride transporter expression similar to that detected in rats, we examined the preoptic region of one GnRH-GFP and one wild-type adult female mouse for the presence of KCC-2 and NKCC-1 mRNA in GnRH neurons. As seen in rats, most murine GnRH neurons were devoid of KCC-2 (Fig. 6upper left panels, arrowheads), despite being surrounded by non-GnRH cells containing an abundance of KCC-2 transcripts (short arrows). In contrast, and again in harmony with our findings in the rat brain, NKCC-1 mRNA was readily detectable in a subset of GnRH neurons (Fig. 6, upper right panels). A similar pattern of expression was observed in wild-type and GnRH-GFP transgenic mice. As in rats, the mouse preoptic region was found to be extremely rich in KCC-2 mRNA positive cells (Fig. 6, middle panels); adjacent sections incubated with a sense KCC-2 probe of identical base composition as the cRNA probe used to detect KCC-2 mRNA showed no hybridization signal (Fig. 6, lower panels), thus demonstrating the specificity of the hybridization reaction. A similar pattern of cotransporter expression was observed in the wild-type mouse brain examined (not shown). Thus, GnRH neurons expressing GFP have a pattern of KCC-2 and NKCC-1 expression indistinguishable from that of rats or wild-type mice.

Figure 6.

GnRH Neurons in Wild-Type and GnRH-GFP Mice Show a Pattern of Chloride Transporter Expression Indistinguishable from That of Rats Images from an adult (50-d-old) wild-type and a GnRH-GFP mouse of the same age are shown. Upper panels, Arrowheads denote GnRH neurons lacking KCC-2 mRNA, short arrows identify non-GnRH neurons expressing KCC-2 or NKCC-1 mRNA, and long arrows point to GnRH-NKCC-1 positive neurons. Notice the presence of a non-GnRH cell and GnRH neuron next to each other containing NKCC-1 mRNA transcripts (upper right panel, arrowhead, and long arrow, respectively). Bars, 20 μm. Middle and low panels, Bright-field and dark-field images demonstrating the abundance of cells containing KCC-2 mRNA in the preoptic region of a GnRH-GFP mouse and the absence of hybridization signal in an adjacent section incubated with a sense KCC-2 probe. Bars, 200 μm.

Function of NKCC in GnRH Neurons

The relative lack of expression of the transporter/channels that lower intracellular chloride, KCC-2 and CLC-2, in GnRH neurons and GT1 cells compared with the chloride accumulator NKCC-1 is consistent with our observation of a depolarized E_GABA in GnRH neurons. To determine the role that transporters of the NKCC family play in maintaining the high intracellular chloride concentration, we tested the effect of bumetanide (50 μm), a blocker of Na-K-2Cl cotransport, on E_GABA. Slices were incubated in bumetanide for 10 min before initiating recordings, a 25–35 min exposure before reversals were measured (including time required for attainment of satisfactory access resistance). Resting potential of control and bumetanide-treated GnRH neurons was not different (P > 0.2), but E_GABA of GnRH neurons treated with bumetanide was significantly hyperpolarized (−57.8 ± 4.2 mV, n = 4, Fig. 7, A and B) compared with GnRH neurons from untreated slices (−36.5 ± 1.2 mV, P < 0.02). In current-clamp mode, there was no significant change in membrane potential after rapid GABA (peak membrane potential after GABA −52.3 ± 4.8 mV; resting potential −58.4 ± 5.0 mV, difference 6.1 ± 1.1 mV, n = 5, Fig. 7C) because bumetanide lowered E_GABA to near the resting potential, markedly reducing the driving force (Fig. 7D). Acute actions of bumetanide could not be detected as these effects took more than 20 min, a duration similar to that of most gramicidin recordings after 100 MΩ access resistance was attained. Although bumetanide has been reported to alter intracellular chloride in the absence of NKCC-1 (29), suggesting it can act on KCC family members as well, the hyperpolarization of E_GABA observed in the present study is consistent with the primary effect of bumetanide being inhibition of NKCC-1. Together with the molecular and immunohistochemical studies presented above, these data suggest that a member of the NKCC family is expressed in GnRH neurons and actively maintains an elevated intracellular chloride concentration in these cells. Because E_GABA was hyperpolarized in all bumetanide-treated GnRH neurons relative to untreated GnRH neurons, it appears that the expression of functional NKCC family members in GnRH neurons is more widespread than indicated by the in situ hybridization detection of NKCC-1 mRNA. This underestimation may be due to both a low prevalence of NKCC-1 mRNA per neuron and the stringent criteria used for identifying a cell as positive.

Figure 7.

Treatment with the NKCC Blocker Bumetanide Hyperpolarizes E_GABA in GnRH Neurons A, Representative gramicidin-perforated patch recording of chloride reversal in a GnRH neurons from a slice treated with bumetanide. Voltage protocol consisted of steps every 15 mV from −85 to −40 mV. B, Plot of current through the GABA_A receptor as a function as step potential in the bumetanide-treated GnRH neuron shown in panel A. C, Comparison of membrane potential at rest and E_GABA in GnRH, GnRH + bumetanide, and two control groups: non-GnRH hypothalamic neurons (non-GnRH) and GAD-GFP neurons. D, Driving force (E_GABA − resting potential) on chloride ions in the same cell types. All voltages reported are corrected for series resistance and liquid junction potential. All recordings were done in the presence of tetrodotoxin, APV, and CNQX.

On-Cell Measurements of Membrane Potential Reveal GABA Depolarizes GnRH Neurons

As an independent confirmation of the results obtained with the gramicidin-perforated patch, we measured the change in membrane potential induced by rapid GABA application in the on-cell configuration, using a recently reported method that takes advantage of voltage-gated potassium channels within the patch (Fig. 8A) (25, 30). Using this technique, rapid GABA depolarized GnRH neurons relative to rest (n = 18, Fig. 8, B and F), whereas non-GnRH neurons (n = 10, Fig. 8, C and F) and GAD-GFP neurons (n = 7, Fig. 8, D and F) were hyperpolarized. Pretreatment with bumetanide markedly reduced the depolarization in response to GABA (n = 8, Fig. 8, E and F). Absolute membrane potential values from which the differences were calculated are shown in Table 2. Note that the difference in current profiles between the examples shown likely reflects different complements of voltage-gated potassium channels expressed by the various cell types as well as variation in the number of potassium channels within the on-cell patch.

Figure 8.

On-Cell Measurements of the Change in Membrane Potential in Response to Rapid Application of GABA A, Command potential protocol for monitoring E_K for channels within the patch pipette. B–D, Plots of current vs. membrane potential in representative GnRH (B), bumetanide-treated GnRH (C), GAD-GFP (D), and non-GnRH, non-GFP neurons (E) at rest (white circles) and after (black circles) rapid GABA application. Membrane potential is the command potential at zero current (25, 30 ). F, Group mean ± sem for the change in membrane potential in response to GABA. All recordings were done in the presence of tetrodotoxin, APV, and CNQX.

Table 2.

Absolute Membrane Potentials (V_m) Calculated From On-Cell Measures at Rest and After Brief GABA Application

Cell Type (n)	Vm at Rest (mV)	Vm After GABA (mV)
GnRH (untreated) (18 )	−72.7 ± 2.3	−56.5 ± 1.6
GnRH + bumetanide (8 )	−74.4 ± 6.2	−70.2 ± 3.7
GAD-GFP (7 )	−64.6 ± 2.2	−77.8 ± 2.1
Non-GnRH (10 )	−61.3 ± 3.0	−75.5 ± 4.6

GnRH Neurons Are Excited by Activation of the GABA_A Receptor in a Variety of Animal Models

On-cell recordings were also used to characterize the physiological response to rapid GABA application. This recording method allows the monitoring of action currents from identified cells without disturbing the natural chloride concentration and with greater stability than perforated patch experiments (31). Rapid application of GABA inhibits firing of non-GnRH neurons (n = 13, Fig. 9A) and GAD-GFP neurons (n = 2, not shown) but induces action currents in GnRH neurons (Fig. 9B). Action currents in GnRH neurons in response to GABA are confined to the time when current is flowing through the GABA_A receptor as measured in the whole-cell configuration (Fig. 9C). An important point is the brevity of GABA treatment in the present study, illustrated by the rapid recovery of firing after GABA in the control neuron shown in Fig. 9A, and rapid decay of the excitatory response in the GnRH neuron in Fig. 9, B and C.

Figure 9.

Activation of GABA_A Receptors Excites GnRH Neurons, but Inhibits Other Adult Neurons Representative on-cell recordings of action currents in a non-GnRH (A) and a GnRH neuron (B). Rapid application of GABA inhibits firing in non-GnRH neurons of the hypothalamus and GAD-GFP neurons, but elicits action current firing in GnRH neurons. C, Whole-cell recording of GnRH neuron shown in the on-cell configuration in panel B illustrating the time course of the current elicited by the rapid GABA application. All recordings were done in the presence of APV and CNQX.

One possible explanation for the reversible suppression of fertility observed in numerous natural and pathological conditions is that there is a change in the physiological response of the GnRH neuron to GABA such that under some conditions, the response is excitation and under others, inhibition (32). We tested the response of GnRH neurons to rapid GABA in a variety of animal models. Almost invariably (n = 49 of 52), rapid GABA evoked action potentials from GnRH neurons. There was no difference among the ages, sex, steroid milieux, or estrous cycle stages tested (see Materials and Methods for details). There was also no effect of time of day, an important finding as the neuroendocrine signal for ovulation that is provided by GnRH neurons occurs on a daily basis in rodents at a specific circadian time (33). Although the excitation of GnRH neurons by activation of the GABA_A receptor is remarkably consistent, we did observe three GnRH neurons in which GABA did not elicit action currents [1 of 16 diestrus, 2 of 14 ovariectomy + estradiol (OVX+E)]. These cells represent 5.2% of the total GnRH neurons examined for action currents; of interest, this percentage is similar to that of detected GnRH neurons expressing message for KCC-2 (4.9%).

Long-Term Activation of the GABA_A Receptor Inhibits GnRH Neurons

The present results showing that activation of the GABA_A receptor excites GnRH neurons is in conflict with numerous studies that have approached this question in vivo or in hypothalamic explants using GABA or GABA agonists (reviewed in Refs. 1 and 2 ; also Refs. 34 and 35) and with the one published report examining GnRH neurons directly in brain slices (17). These studies examined the response of the hypothalamic-pituitary axis to prolonged exposure (≥1 min) to GABA. We thus examined the effects of bath application of muscimol, a GABA_A receptor agonist, on action currents in GnRH neurons recorded in on-cell mode. Muscimol (50 μm) had a biphasic effect on action current firing in GnRH neurons (Fig. 10, A and B, n = 19 cells). An example from a cell that was quiescent is shown in Fig. 10A (10 of 19 cells), whereas an example from a cell that was spontaneously active at the time of the study is shown in Fig. 10B (9 of 19 cells, minimum of three spontaneous action currents). There was an increase in the rate of action current firing within approximately 60 sec from adding the muscimol to the perifusion system (there is a delay of ∼45–60 sec to initiate solution exchange at the slice) regardless of whether the cell was quiescent or active. This excitation gave rise to inhibition within 10–20 sec. During this muscimol-induced inhibition, the excitatory response to rapidly applied GABA was lost. Upon washout of muscimol, the lack of response to rapid GABA application persisted for 4–15 min (10.2 ± 0.9 min, n = 13 recordings of adequate duration to observe washout). At least four phenomena could contribute to this: collapse of the chloride gradient, desensitization of the GABA_A receptor, depolarization of the cell toward E_GABA, and shunting inhibition secondary to changes in input resistance. On-cell measurements of resting membrane potential and the response to rapid GABA application (as above) made before and after muscimol application (Fig. 10C) demonstrated membrane potential was not different but the depolarizing response to GABA was eliminated (Table 3). This indicates that inhibition of firing due to prolonged depolarization of the membrane potential cannot account for the muscimol effect. Measurement of GABA-evoked currents in whole-cell voltage clamp mode with 20 mm pipette chloride revealed the response to GABA persisted in muscimol but the amplitude of the currents was diminished (n = 3 cells). Furthermore, the chloride reversal hyperpolarized during muscimol application and input resistance was markedly reduced. These observations in whole-cell mode suggest that a combination of collapse of the chloride gradient, receptor desensitization, and shunting inhibition contribute to the inhibition of firing after strong activation of the GABA_A receptor by muscimol. Future studies will distinguish between these possibilities.

Figure 10.

Long-Term Activation of GABA_A Receptors Leads to Shunting Inhibition of GnRH Neurons A and B, Recordings of GnRH neuron activity over time. Solid arrows indicate rapid GABA applications, which elicited action currents before treatment with muscimol (as in Fig. 9B). Bath application of muscimol initially increases firing rate, illustrated by the barrage of action currents (open arrows), and then inhibits firing. Examples are shown from a GnRH neuron that was spontaneously active before muscimol treatment (A) and one that was quiescent before muscimol treatment (B). Dashed lines in panel A indicate action currents induced by rapid GABA. During muscimol treatment and for several minutes thereafter, there is a loss of response to rapid application of GABA indicated; asterisks indicate failure of response to rapid GABA. C, On-cell measurement of membrane potential at rest and after rapid GABA in a GnRH neuron reveals depolarization before treatment with muscimol. D, On-cell measurement of membrane potential before and after rapid GABA in the same GnRH neuron after muscimol treatment demonstrates rapid GABA no longer depolarizes GnRH neurons. All recordings were done in the presence of APV and CNQX.

Table 3.

On-Cell Measurements of Membrane Potential (V_m) Showing the Effect of Bath Application of Muscimol on the Response to Rapid Application of GABA

	Vm at rest (mV)	Vm after GABA (mV)	ΔVm (mV)
Control	−66.8 ± 3.8	−46.1 ± 5.0	22.6 ± 5.9
Muscimol	−63.3 ± 1.0	−62.6 ± 2.91	0.7 ± 2.01

¹

P < 0.04 vs. control.

DISCUSSION

The present study combined electrophysiological and molecular approaches to demonstrate that activation of GABA_A receptors excites GnRH neurons. Molecular characterization of the transporters important in chloride homeostasis revealed a mechanistic basis for the high intracellular chloride and the resulting depolarized E_GABA in GnRH neurons.

These observations appear to contradict a body of work that largely suggests GABA_A receptor activation inhibits GnRH neurons. In vivo, there is an inverse relationship between preoptic GABA levels and LH release in rodents and sheep (35–37), and hypothalamic application of GABA_A antagonists (38–40) increased LH release. Furthermore, puberty in primates was advanced by administration of GABA_A antagonists or blockade of GABA synthesizing enzymes (41) due to premature activation of GnRH neurons. Measurement of GnRH release from hypothalamic explants has given mixed results. In some studies, using tissue fragments that include the entire medial basal hypothalamus and the preoptic region, blockade of GABA_A receptors was shown to increase GnRH release from immature hypothalami but decreased it from mature hypothalami (34). In others, activation of GABA_A receptors from hypothalamic fragments comprising only the arcuate-median eminence region stimulated GnRH release (42). Because in vivo manipulation of the GABA_A receptor system has been consistently shown to inhibit LH release (38–40), it would appear that activation of GABA_A receptors may lead to both inhibition and stimulation of GnRH release, but that the latter response is masked when the entire neuronal circuitry controlling GnRH release is subjected to GABA_A receptor-mediated activation. Additionally, gonadotropes express the GABA_A receptor, so direct modulation of LH release by GABA may be possible (43). It then follows that the aforementioned multineuronal systems cannot be used to resolve the nature of direct effects on GnRH neurons mediated by GABA_A receptors.

The present approach addresses this issue by directly analyzing the response of GnRH neurons to GABA_A receptor stimulation using electrophysiological approaches. E_GABA is an intrinsic property of neurons and thus determines the responsiveness of the GnRH neuron to activation of the GABA_A receptor independent of other cell types. The observations in the present study that E_GABA is depolarized in GnRH neurons and that brief local application of GABA excites these cells suggest that the inhibition of LH and GnRH release described in previous studies may be due to GABA modulating neurons afferent to GnRH neurons. Consistent with this, a temporally controlled, local increase in GABA availability near GnRH nerve terminals in the median eminence resulted in stimulation of the hypothalamic-pituitary-ovarian axis, instead of the expected inhibition (44).

Another important difference between our study and those using whole animals or tissue fragments is the method of GABA application, which was brief and near the GnRH neuron being recorded in the present study as opposed to long-term and widespread in most previous reports. Although even brief focal application of GABA may influence surrounding cells, the treated area is markedly smaller; furthermore, the inclusion of tetrodotoxin, APV, and CNQX would eliminate most neurotransmission. More importantly, this treatment is substantially more acute as illustrated by the rapid recovery of action currents in the control cell in Fig. 9A (onset of inhibition about 10 msec post-GABA, recovery within ∼200 msec) and equally rapid cessation of GABA-induced action current firing in GnRH neurons (Fig. 9B). Activation of GABA_A receptors can change the intracellular chloride concentration (e.g. Ref. 45), diminishing response to subsequent activation of these receptors. In the present measures of E_GABA, this pitfall was avoided by allowing sufficient time between brief GABA applications for recovery of the chloride gradient. This diminished response was evident when the GABA_A agonist muscimol was bath applied to mimic previous studies in which the GABA_A receptor was strongly activated over a prolonged time. Muscimol caused an initial increase in firing rate followed by a prolonged suppression. During muscimol and for several minutes after, brief GABA did not excite GnRH neurons. On-cell measures of membrane potential and whole-cell recordings indicated this lack of response to GABA had several components including collapse of the chloride gradient, partial desensitization of GABA_A receptors, and shunting inhibition. The question remains: what is the physiological GABA input to GnRH neurons? It is clear that GnRH neurons experience brief synaptically mediated GABA_A receptor currents (46, 47). It remains to be determined whether tonic GABA signals that have been reported in other brain regions (48) are part of the physiological repertoire of GABA’s influence on GnRH neurons.

The present findings disagree with a recently published report in which a gramicidin-perforated patch was used to examine the response of GnRH neurons identified by 5-chlormethyl-fluorescein di-β-d-galactopyranosidase cleavage in mouse brain slices from GnRH-β-galactosidase transgenics (17). In that report, bath-applied GABA had different effects among immature, peripubertal, and adult mice, with GnRH neurons from adults typically hyperpolarized by GABA. The same mechanisms that underlie the inhibition of GnRH neurons in the presence of muscimol could explain the inhibition of adult GnRH neurons observed in that study. Another possibility is that the β-galactosidase transgenic reveals a different subpopulation of GnRH neurons than GFP does; approximately 30% of GnRH neurons are revealed by the former and about 70–90% by the latter. The molecular characterization of chloride cotransporters in the present study suggests that subpopulations of GnRH neurons may exist with regard to chloride homeostatic mechanisms, but also that a putatively chloride-extruding (i.e. KCC-2-expressing) subpopulation is quite small (∼5% of cells). Although this percentage was estimated from a relatively small sample size, it is remarkably similar to that of GnRH neurons in which rapid GABA failed to elicit action currents. Whether or not such a subpopulation, which would be expected to have a very different response to GABA, has a separate physiological function is an interesting question for future studies.

Unlike the previous study (17), we found no shift in the response of GnRH neurons to activation of the GABA_A receptor. Instead, GABA excited GnRH neurons from a variety of reproductive models and at different circadian times. Furthermore, there were no developmental changes in chloride cotransporter expression detected in GnRH neurons. KCC-2 was not only absent from GT1–7 cells, but also from a majority of mature GnRH neurons. Conversely, NKCC-1 was expressed in both the immature cells and mature neurons. Importantly, this differential expression was observed in both rats and mice, indicating that instead of being a peculiarity of GT1–7 cells, the difference is preserved in native neurons of two different rodent species, and thus it has a much broader physiological significance. Although the levels of NKCC-1 mRNA in adult GnRH neurons may not be sufficiently high to be detected by in situ hybridization in all neurons, the electrophysiological response of these cells to bumetanide, an inhibitor of Na-K-2Cl transport, was remarkably consistent, suggesting that NKCC-dependent responses to GABA_A receptor activation are integral features of GnRH neuron physiology throughout development. This is different from many neuronal systems in which expression of chloride cotransporters is developmentally regulated. Specifically, in rat hippocampal and neocortical pyramidal cells and in the hypothalamus, NKCC-1 dominates in early postnatal life whereas KCC-2 expression is low during the first postnatal week but soon reaches mature levels (10). The early neonatal cotransporter milieu generates elevated intracellular chloride levels; the response to GABA_A receptor activation in many immature neurons is thus depolarizing (e.g. Ref. 49). As KCC-2 expression increases during early postnatal life, neuronal chloride levels are reduced and the excitatory response dissipates in most systems (10, 11).

Based on this developmental change in other systems, we and others (17, 34) hypothesized that there would be a shift in chloride homeostasis and response to GABA_A receptor activation in GnRH neurons at puberty and other reproductive transitions. The relative lack of KCC-2 expression and persistence of NKCC-1 expression in GnRH neurons, in combination with the elevated intracellular chloride levels and consistent excitatory response of these cells to GABA, forces us to reject this hypothesis. Shifts in response of GnRH neurons to activation of the GABA_A receptor thus cannot explain changes in GnRH secretion between different reproductive states.

Our data from GnRH neurons in brain slices agree with earlier reports of GABA action on the GT1 GnRH cell lines (7, 8). Bath application of GABA elicited a biphasic response in GnRH release from GT1–1 cells, with an initial stimulation due to activation of GABA_A receptors and a prolonged inhibition through GABA_B receptors (7). Because GT1 cells are clonal, action through non-GnRH afferents cannot explain this response. Patch-clamp analysis of GT1–7 cells revealed excitation in response to GABA (8). One interpretation of this excitatory response is that GT1 cells were likely transformed early in development and thus may represent immature neurons with regard to chloride cotransporter milieu. Our observations in adult GnRH neurons, however, suggest that the response in GT1 cells is physiological and that GnRH neurons do not undergo the typical postnatal shift in cotransporter expression that characterizes neurons that are inhibited by GABA_A receptor activation.

A depolarizing response to GABA_A receptor activation has been described during development (49), after neural trauma (50), in Rohon-Beard cells of the spinal cord (51), and in many adult sensory systems including dorsal root ganglion cells (29), horizontal cells of the retina (52), and primary olfactory sensory neurons (53). The latter is of particular interest with regard to GnRH neurons as their embryonic origin is the olfactory placode (54). The response to GABA in many neural systems, including spinal cord cells (51), fast-spiking interneurons of the cortex and amygdala (55), and thalamus (56), has been traced to elevated intracellular chloride concentrations resulting from reduced KCC-2 or persistent NKCC-1 activity (29, 57). The present finding in GnRH neurons supports and extends these observations to centrally located neuroendocrine cell type, reemphasizing the importance of chloride homeostasis in regulating the response to activation of the GABA_A receptor.

Even in cells in which GABA_A receptor activation is depolarizing, the final response is rarely excitatory in contrast to what we have consistently observed in GnRH neurons. This begs the question of why GnRH neurons are different from other cell types. One attractive possibility is that this permits GnRH neurons a degree of autonomy in responding to GABAergic inputs. That is, the excitatory response to GABA_A receptors can be balanced with inhibition through GABA_B receptors within the GnRH neuron itself by shifts in GABA receptor subtype expression. In this regard, the work of Lagrange et al. (19) demonstrated that activation of the GABA_B receptor on GnRH neurons in guinea pigs inhibits these cells though gating of a G protein-coupled inwardly rectified potassium current. Our preliminary results suggest GABA_B receptors also inhibit GnRH neurons in the mouse (DeFazio, R. A., S. D. Sullivan, and S. M. Moenter, unpublished). A more speculative possibility is that an opposite response of the reproductive axis to a prevalent neurotransmitter such as GABA would be one mechanism for achieving regulation that is out of phase with other neuroendocrine systems such as those governing stress (58) and lactation (59), which must essentially be turned off for reproduction to proceed. Although resolution of this issue awaits further investigation, the data in the present study redefine the role of GABAergic inputs to GnRH neurons by demonstrating that the response of the majority of GnRH neurons to rapid activation of the GABA_A receptor is excitatory.

MATERIALS AND METHODS

Animals

GnRH neurons were recorded from GnRH-GFP mice (18) that had ad libitum access to food and water and held on a 14-h light, 10-h dark cycle with lights on at 0200 h Eastern Standard Time (EST). Estrous cycle stage was monitored by vaginal lavage. Most characterizations were done on reproductively competent adult females (>42 d) during the diestrous stage of the estrous cycle. Animal models representing different reproductive states were used to determine whether changes in reproductive status are accompanied by modifications in the response of GnRH neurons to activation of GABA_A receptors. Developmental changes were assessed by comparing action current firing in response to GABA (see below) in gonadal-intact neonatal (4 d, n = 3 cells), juvenile (15–38 d, n = 2 cells), and adult (>42 d) mice. Sex differences were assessed by comparing male (n = 4 cells) and female mice (in diestrous and estrous phases of the estrous cycle, n = 16 and 2 cells, respectively). The effects of steroid negative feedback were tested in female mice that were bilaterally ovariectomized under Metofane anesthesia (Janssen Pharmaceuticals, Toronto, Ontario, Canada) and were either implanted with a SILASTIC (Dow Corning Corp., Midland, MI) capsule containing 0.625 μg estradiol (OVX+E, n = 14 cells) or left untreated (OVX, n = 11 cells). This estradiol treatment lengthens the interval between episodes of increased firing rate in GnRH neurons and the effects of estradiol are conveyed, at least in part, by GABAergic networks (60). Experiments were begun in the late morning with recording typically spanning 1300–1800 h EST (Toronto, Ontario, Canada). A shift in response to GABA has recently been suggested to account for day/night differences in activity of neurons of the circadian clock in the suprachiasmatic nucleus (32). To control for circadian changes in reproduction, such as the late afternoon events associated with central neural signals for ovulation in rodents, experimental time was varied in a subset of trials so that recordings were done at 0700–1000 h (early morning) and 2100–0100 h (late evening). To control for effects of GFP expression, adult gonad-intact GAD-GFP mice (The Jackson Laboratory, Bar Harbor, ME) were used (24). The Animal Care and Use Committees of the University of Virginia approved all procedures used in these experiments.

Female Sprague Dawley rats (B & K Universal, Fremont, CA) were used for immunohistofluorescence and combined immunohistochemistry/in situ hybridization studies. They were housed under 14-h light, 10-h dark photoperiod (lights on at 0500 h) and temperature (23–25 C) and were given free access to tap water and pelleted rodent chow. Transgenic GnRH-GFP mice (18) and wild-type mice of the same strain were also employed for combined immunohistochemistry/in situ hybridization experiments. All animals were used in accordance with the NIH guide for the Care and Use of Laboratory Animals, and the experimental protocols were approved by the Oregon National Regional Primate Research Center Institutional Research Animal Committee.

Brain Slice Preparation

All reagents were purchased from Sigma(St. Louis, MO). All solutions were bubbled with 95% O₂/5% CO₂ throughout the experiments and for at least 30 min before exposure to tissue. Brain slices were prepared through the preoptic area and hypothalamus as previously described (60). The brain was rapidly removed and placed in ice-cold sucrose saline solution containing (in millimolar concentration): 250 sucrose, 3.5 KCl, 26 NaHCO₃, 10 glucose, 1.25 NaHPO₄, 1.2 MgSO₄, and 3.8 MgCl₂. Coronal 200-μm brain slices were cut with a vibratome (Ted Pella, Inc., Redding, CA) and incubated for 15 min at room temperature in a solution of 50% sucrose-saline and 50% normal saline (NS). NS contained (in millimolar concentration): 125 NaCl, 3.5 KCl, 26 NaHCO₃, 10 glucose, 1.25 NaHPO₄, 1.2 MgSO₄, and 2.5 CaCl₂. Slices were then transferred to a solution of 100% NS at 30−32 C for at least 90 min before recording.

Recording Solutions and Data Acquisition

The bath solution for recording was 100% NS plus 20 μm APV and 10 μm CNQX to block ionotropic glutamatergic receptors. Tetrodotoxin (0.5 μm) was included except when action currents or action potentials were monitored, as detailed below. During recording, slices were continuously superfused at 5 ml/min with oxygenated bath solution kept at 32 C with an inline-heating unit (Warner Instruments, Hamden, CT). GFP-positive cells were visualized with a combination of infrared differential interference contrast and fluorescence microscopy. Recordings were made with an EPC-8 patch-clamp amplifier (HEKA, Lambrecht, Germany). The data acquisition system consisted of an ITC-18 acquisition interface (Instrutech, Port Washington, NY) and Pulse control software (Instrutech) running within Igor Pro (Wavemetrics, Lake Oswego, OR) on a Macintosh G4 (Apple, Cupertino, CA).

GABA Application

Local pressure application of GABA was accomplished as previously described (11). Patch pipettes were filled with a HEPES-buffered saline (containing in millimolar concentration: 150 NaCl; 10 HEPES; 10 glucose; 3.5 KCl; 2.5 CaCl₂; 1.3 MgCl₂, pH 7.2; with NaOH) similar to the extracellular solution with the addition of 1 mm GABA. The pressure pipette was placed adjacent to the soma or proximal process and a 5- to 7-msec pulse of 3−10 psi was delivered using a Picospritzer II (General Valve Corp., East Hanover, NJ). Pressure application of the HEPES-buffered solution alone had no effect.

Gramicidin-Perforated Patch Recordings

Gramicidin-perforated patch recordings were performed using a modification of previously described methods (16). Patch pipettes (2−5 mΩ) were first filled with 1.25–1.75 μl of filtered, gramicidin-free internal solution and then backfilled with freshly sonicated, unfiltered pipette solution containing 10–50 μg/ml gramicidin D. Gramicidin stock solution was prepared every 1–3 h at 50 mg/ml in DMSO. The pipette solution consisted of (in millimolar concentration): K gluconate, 139; KCl, 1; HEPES, 10; EGTA, 5; CaCl₂, 0.1; MgCl₂, 4, pH 7.2; with NaOH; for high chloride, K gluconate was replaced with KCl). Recording pipettes containing either 140 mm or 1 mm chloride were used with similar results after liquid junction correction of 3 and 10 mV, respectively [JPCalc (61)]. All voltages reported are corrected.

Gramicidin-perforated patch recordings were monitored as detailed in Results (Fig. 1). Current-clamp recordings were initiated when the access resistance fell below 200 mΩ. Recordings were made in I = 0, zero current mode to eliminate voltage offsets due to the high access resistance of perforated patch mode. Voltage clamp recordings were initiated when the access resistance fell below 100 mΩ (range 27–100 mΩ). Rupture of perforated patch recordings to whole-cell mode was detected by an abrupt decrease in access resistance associated with a rapid hyperpolarization of the reversal potential when 1 mm chloride pipette solution was used (as in Fig. 2) or a reversal near 0 mV with 140 mm pipette chloride.

Voltage steps every 10–15 mV bracketing the reversal potential were used; GABA was delivered as described above 250 msec into the voltage step to allow the membrane current to stabilize (56). Voltage steps and accompanying GABA puffs were delivered at an interval of 30 sec; this prevented decrement in the GABA response due to changes in intracellular chloride. Baseline current was measured just before the GABA pulse; GABA-induced current was measured 100 msec after rapid GABA application. The step potential was series resistance corrected independently for each measurement based on the absolute current, the series resistance and Ohm’s law (V_corrected = V_step − V_error; V_error = I * Rseries). Because the currents are different for the baseline and the GABA response, the step potential is different after correction (56). The reversal potential was determined from the crossing of the current-voltage relationship of the baseline current and the GABA response using a custom-written computer program running under IGOR Pro and verified by eye for each reversal. Only reversals that were stable (<5-mV variation) for two or more measurements (5–10 min) were accepted for this study. Intracellular chloride was estimated from reversal potentials using the Nernst equation (E_GABA = 60.15 log [Cl⁻]_i/[Cl⁻]_o) assuming 100% of current through the GABA_A receptor was carried by Cl⁻ and a chloride activity coefficient of 0.76.

On-Cell Measurement of Membrane Potential

During a recording in the on-cell configuration, the patch of membrane within the pipette is exposed to a potential difference equal to the membrane potential (V_cell) minus the pipette command potential (V_patch = V_cell − V_pipette). Potassium channels within the pipette can be manipulated by varying the pipette potential and the reversal of current through these channels (E_K) used to estimate membrane potential (25, 30). This method assumes that the concentration of potassium in the cell is similar to that in the pipette solution, resulting in a reversal potential for potassium (E_K) near 0. After establishing the on-cell configuration in the presence of 0.5 μm tetrodotoxin, inactivation of potassium channels is removed by setting V_pipette to +100 mV for 60 msec (V_patch ∼−150 mV, assuming −50 mV V_cell). Channels are then activated by ramping the pipette voltage from +100 mV to −150 mV (V_patch ∼ −150 to +200 mV) over 30 msec. During the voltage ramp, potassium channels are opened and result in an initial inward current followed by an outward current. Leak correction was applied by subtracting a linear fit of the current during the ramp before the activation of potassium currents. The ramp potential where the leak corrected current crosses 0 pA reflects the membrane potential of the cell. Accurate assessment of V_cell relies on knowing the concentration of intracellular potassium, estimated to be approximately 150 mm (62). Although the concentration of intracellular potassium in GnRH neurons has not yet been determined, a difference of 15 mm results in an offset of only 5 mV (30). Such an offset would not affect the relative difference in membrane potential due to GABA. To assess the effect of rapid GABA, we delivered a control ramp to estimate the resting membrane potential, applied GABA, and then immediately delivered another ramp to estimate the membrane potential response to GABA (total duration of voltage protocol including both ramps was 300 msec, see Fig. 8A). As reported previously (25, 30), we found that the absolute potentials (Table 2) were on average more hyperpolarized than measurements in whole-cell or gramicidin methods, possibly due to Donnan potentials that exist during the latter recording modalities (25).

On-Cell Action Currents

After formation of the GΩ seal using any of the internal solutions, large amplitude fast current spikes were detected in voltage clamp at 0 mV and the response to rapid GABA application was evaluated (e.g. Ref. 63).

Response to Muscimol

To test the effects of long-term activation of the GABA_A receptor, the GABA_A receptor agonist muscimol (50 μm) was bath applied for 1–2 min. Action currents and membrane potential in response to rapid GABA application were determined as detailed above in the on-cell mode. Whole-cell recordings with 20 mm pipette chloride were used to assess GABA_A receptor desensitization and input resistance changes in response to muscimol. Cells were held at −68 mV and stepped to a variable potential of −53 mV, −38 mV, or −15 mV to estimate the reversal potential. This was followed by a step to −38 mV (near the empirical reversal potential for 20 mm pipette chloride) to assess changes in [Cl⁻]_in.

RT-PCR Cloning and Sequencing

Total RNA was isolated from GT1–7 cells and mouse cerebral cortex using TriReagent (Molecular Research Center, Inc., Cincinnati, OH) according to the manufacturer’s instructions. Two hundred nanograms were reverse transcribed for 1 h at 37 C using Omniscript Reverse Transcriptase (QIAGEN, Valencia, CA) in a 20-μl volume. Two microliters of each reaction were then subjected to PCR amplification [95 C for 15 min followed by 30 cycles of denaturing at 94 C for 30 sec, annealing at 57 C (for KCC-2 and CLC-2) or 52 C (for NKCC-1) for 1 min, and extension at 72 C for 1 min, followed by a final extension at 72 C for 10 min, using HotStarTaq Polymerase (QIAGEN)].

A 720-bp cDNA fragment of the mouse CLC-2 gene was amplified using oligodeoxynucleotides corresponding to nucleotide (nt) 520–540 (5′-primer) and 1220–1240 (3′-primer) in the mouse CLC-2 mRNA sequence (64). A 260-bp cDNA fragment of the mouse KCC-2 gene was amplified using primers corresponding to nt 3168–3188 (5′-primer) and 3407–3427 (3′-primer) in the mouse KCC-2 mRNA sequence (65). In the case of NKCC-1, a 321-bp cDNA fragment was amplified using oligodeoxynucleotides corresponding to nt 3251–3271 (5′-primer) and 3551–3571 (3′-primer) in the coding region of mouse NKCC-1 mRNA (66). PCR products were cloned into the vector pGEM-T (Promega Corp., Madison, WI) and their identities were verified by sequencing.

Ribonuclease (RNase) Protection Assay

The procedure used has been described in detail elsewhere (67). In brief, each RNA sample was separately hybridized to 500,000 cpm of gel-purified ³²P-UTP-labeled CLC-2, KCC-2, or NKCC-1 cRNA probes, each in combination with 5000 cpm of a ³²P-UTP-labeled cyclophilin cRNA probe to correct for procedural variability (67). The cRNA probes were generated by SP6 or T7 polymerase-directed transcription using as templates each of the chloride channel cDNAs previously PCR cloned from cellular RNA. The cyclophilin probe was transcribed from a 158-bp PCR-generated cDNA fragment corresponding to nt 264–421 in mouse cyclophilin mRNA (68).

Western Blots

For protein extraction of GT1–7 cells, cell monolayers were washed with ice-cold PBS and lysed with 1 ml radioimmunoprecipitation assay (RIPA) buffer. Protein extraction from frozen rodent tissue was performed using the same extraction solution in a glass homogenizer. The lysates were centrifuged at 13,000 × g, at 4 C for 15 min, and the supernatants were frozen at −85 C until use. Protein concentrations were estimated using the Bio-Rad protein assay (Bio-Rad Laboratories, Inc., Hercules, CA). Size fractionation was achieved by SDS-PAGE, using 8% polyacrylamide gel for CLC-2 and 6% gel for NKCC-1. After electrophoresis, proteins were transferred for 2 h at 1 Amp to polyvinylidene membranes (Pierce Chemical Co., Rockford, IL). Membranes were blocked with 3% milk in 0.05%Tween 20/Tris buffered saline (TBS, pH 7.4) for 1 h and incubated with the first antibody overnight at 4 C. For CLC-2 detection we used a rabbit polyclonal anti-CLC-2 serum (1:300, Alomone Labs, Jerusalem, Israel). The specificity of this antiserum was determined by preadsorbing it with the CLC-2 peptide used as antigen (1 μg CLC-2 peptide per 2 μg antibody in 0.05%Tween 20/TBS) for 2 h at room temperature before final dilution.

NKCC-1 was detected with rabbit anti-NKCC-1 polyclonal antibodies (1:500, overnight, at 4 C, a generous gift form E. Delpire, Vanderbilt University Medical Center, Nashville, TN) (69). After extensive washing with 0.05% Tween 20/TBS, membranes were exposed to horseradish peroxidase-conjugated goat antirabbit IgG (1:40,000, Pierce Chemical Co.) for 1 h at room temperature. After three washes in 0.05% Tween 20/TBS, bound antibody was detected by enhanced chemiluminescence using the Super Signal Western Dura chemiluminescence substrate (Pierce Chemical Co.).

Immunohistofluorescence-Confocal Microscopy

GT1–7 GnRH neurons were seeded on coverslips precoated with poly-l-ornithine and cultured in DMEM containing 10% fetal bovine serum plus penicillin (100 U/ml) and streptomycin (100 μg/ml), as described previously (68). At approximately 60% confluence, cells were rinsed with PBS, fixed in Zamboni’s fixative for 30 min at room temperature, and kept at 4 C in PBS until processed for immunohistochemistry.

Rat brains were fixed by transcardiac perfusion of Zamboni’s fixative as described earlier (70). Immunohistochemical reactions were performed on either 14-μm cryostat sections obtained from frozen, 20% sucrose-cryoprotected, optimal cutting temperature-embedded tissue, or 40-μm floating vibratome sections obtained from brain tissue that had been stored at 4 C in PBS after fixation (70). GnRH was identified with monoclonal antibody 4H3 (71) in both GT1–7 cells and brain sections (1:2000, overnight at 4 C and 1:3000, 48 h at 4 C, respectively). The reaction was developed with Texas Red goat antimouse γ-globulin (1:200, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) as outlined elsewhere (70). Initial examination of different antibodies to CLC-2, KCC-2, and NKCC-1 showed that some antibodies worked better in GT1–7 cells than in brain sections, and vice versa. Therefore, in each instance we selected the most appropriate antibody for immunostaining. CLC-2 was identified in GT1–7 cells with rabbit polyclonal antibodies (1:500, 48 h, 4 C) (14) (a generous gift from Rod Smith, University of Colorado Health Sciences Center, Denver, CO). For brain sections, we used chicken polyclonal antibodies against rat CLC-2 (1:5000, 48 h, 4 C) (72) (kindly provided by C. Murray and P. L. Zeitlin, John Hopkins Medical Institutions, Baltimore, MD). KCC-2 and NKCC-1 were identified in GT1–7 cells with polyclonal rabbit antibodies (73, 74) (generously provided by J. Payne, University of California School of Medicine, Davis, CA; and E. Delpire, Vanderbilt University Medical Center, Nashville, TN) diluted 1:500. In brain sections, the KCC-2 antibody was used at 1:1000, and NKKC1 staining was performed using monoclonal antibody T4 (78) diluted to 1.5 μg/ml (a generous gift from C. Lytle, University of California, Riverside, CA).

All chloride channel and transporter immunoreactions in GT1–7 cells were developed with a fluorescein isothiocyanate-labeled goat antirabbit γ-globulin (1:200, Jackson ImmunoResearch Laboratories, Inc.) (70). For brain sections, the reactions were developed using the biotinylated tyramine enhancement method, and the appropriate biotinylated secondary antibodies, as previously described (76). Cell nuclei were detected by staining with Hoechst 33258 (Molecular Probes, Inc., Eugene, OR) at 0.1 μg/ml potassium PBS for 1 min after completion of the immunohistochemical reactions (44). The immunofluorescence images were acquired with a TCS-SP laser scanning confocal system (Leica Corp., Heidelberg, Germany) and a Leica Corp.IRBE microscope, as previously described (44).

Combined Immunohistochemistry/in Situ Hybridization

These experiments were performed to determine the presence of CLC-2, KCC-2, and NKCC-1 mRNA in GnRH neurons during postnatal development of the rodent hypothalamus. The brains of infantile (3-d-old), prepubertal (28-d-old), and adult (60-d-old) female rats, as well as the brains of adult 50-d-old GnRH-GFP and wild-type mice were fixed by transcardiac perfusion with 4% paraformaldehyde in borate buffer, pH 9.5 (77), and subjected to combined immunohistochemistry/in situ hybridization, using procedures described in detail elsewhere (70). The reactions were performed on 14-μm cryostat sections and every other section analyzed (i.e. a periodicity of every 28 μm for each probe). GnRH neurons were identified with monoclonal antibody 4H3 (71), diluted at 1:1000. Upon completion of the immunohistochemical reaction, the sections were dried in a vacuum oven, and the next day CLC-2, KCC-2, and NKCC-1 mRNAs were detected using the same riboprobe used in RNA protection assays, but labeled with ³⁵S-UTP instead of ³²P-UTP. The hybridization mixture (71, 72) contained 5 × 10⁷ cpm/ml of NICK column (Pharmacia Biotech)-purified labeled probe. The hybridization reaction was carried out for 20 h at 55 C; posthybridization washes in decreasing concentrations of sodium chloride-sodium citrate buffer (SSC) were as reported (71, 72) to a final stringency of 0.1 × sodium chloride-sodium citrate buffer at 65 C and were preceded by an RNase digestion step (30 min at 37C with 20 μl RNase/ml). Control sections were incubated with sense RNA probes transcribed from the same plasmids but linearized on their 3′-end to transcribe the coding strand of each cDNA. A more detailed description of the in situ hybridization procedure used can be found in Refs. 71 and 72 . After dehydration in graded alcohols, the slides were exposed to Hyperfilm (Amersham Biosciences, Piscataway, NJ) x-ray film for 3 d to verify the specificity and tissue pattern of hybridization; thereafter, the sections were defatted, and dipped in NTB-2 emulsion (Roche, Indianapolis, IN). The reaction was developed 3 wk later, and the sections were counterstained with 1% methyl green before analysis under both bright- and dark-field illumination. GnRH neurons were considered to express a cotransporter mRNA when they had 10 or more silver grains overlying the nucleus and perikaryon. In all cases, the density of silver grains overlying a positive cell was at least 3 times that of adjacent areas devoid of cells.

Statistics

Data are reported as mean ± sem. Comparisons between GnRH and non-GnRH neurons were made by two-tailed t test with P < 0.05 accepted as significant. The χ² goodness of fit analysis was employed to determine the statistical significance of differences in the number of GnRH neurons containing NKCC-1 and KCC-2 mRNA.

Acknowledgments

We thank Xu-Zhi Xu and Maria E. Costa for expert technical assistance, Dr. Anda Cornea for her invaluable advice with confocal microscopy, and Glenn Harris, Craig Nunemaker, and Shannon Sullivan for editorial comments.

This work was supported by NIH Grants HD-34860 and HD-41469 and Whitehall Foundation Grant 2000-12-43-A (S.M.M.); NIH Grant RR-00163 for the operation of the Oregon National Primate Research Center; and NICHD/NIH Grant through cooperative U54 HD-18185-16 (S.R.O.) and U54 HD-28934 (John C. Marshall, Principle Investigator) as part of the Specialized Cooperative Centers Program in Reproduction Research. S.H. was a Pediatric Endocrinology Fellow supported by a fellowship from the European Society for Pediatric Endocrinology (ESPE).

^*

R.A.D. and S.H. contributed equally to these studies.

Abbreviations:

 
• APV,

  d-(−)-2-Amino-5-phosphonopentanoic acid;

 
• CNQX,

  6-cyano-7-nitroquinoxaline-2,3-dione;

 
• GABA,

  γ-aminobutyric acid;

 
• GFP,

  green fluorescent protein;

 
• KCC,

  potassium-chloride contransporter;

 
• NKCC,

  sodium-potassium-2-chloride cotransporter;

 
• NS,

  normal saline;

 
• nt,

  nucleotide;

 
• OVX,

  ovariectomy;

 
• OVX+E,

  ovariectomy + estradiol;

 
• RNase,

  ribonuclease;

 
• TBS,

  Tris-buffered saline.