Heterogeneous chloride homeostasis and GABA responses in the median preoptic nucleus of the rat

Abstract

The median preoptic nucleus (MnPO) is an integrative structure of the hypothalamus receiving periphery-derived information pertinent to hydromineral and cardiovascular homeostasis. In this context, excitability of MnPO neurones is controlled by fast GABAergic, glutamatergic and angiotensinergic projection from the subfornical organ (SFO). Taking advantage of a brain slice preparation preserving synaptic connection between the SFO and the MnPO, and appropriate bicarbonate-free artificial cerebrospinal fluid (CSF), we investigated a possible implication of an active outward Cl⁻ transport in regulating efficacy of the GABA_A receptor-mediated inhibitory response at the SFO–MnPO synapse. When somata of the MnPO neurones was loaded with 18 mm chloride, stimulation of the SFO evoked outward inhibitory postsynaptic currents (IPSCs) in 81% of the MnPO neurones held at −60 mV. Accordingly, E_IPSC was found 25 mV hyperpolarized from the theoretical value calculated from the Nernst equation, indicating that IPSC polarity and amplitude were driven by an active Cl⁻ extrusion system in these neurones. E_IPSC estimated with gramicidin-based perforated-patch recordings amounted −89.2 ± 4.3 mV. Furosemide (100 μm), a pharmacological compound known to block the activity of the neurone-specific K⁺–Cl⁻ cotransporter, KCC2, reversed IPSC polarity and shifted E_IPSC towards its theoretical value. Presence of the KCC2 protein in the MnPO was further detected with immunohistochemistry, revealing a dense network of KCC2-positive intermingled fibres. In the presence of a GABA_B receptor antagonist, high-frequency stimulation (5 Hz) of the SFO evoked a train of IPSCs or inhibitory postsynaptic potentials (IPSPs), whose amplitude was maintained throughout the sustained stimulation. Contrastingly, similar 5 Hz stimulation carried out in the presence of furosemide (50 μm) evoked IPSCs/IPSPs, whose amplitude collapsed during the high-frequency stimulation. Similar reduction in inhibitory neurotransmission was also observed in MnPO neurones lacking the functional Cl⁻ extrusion mechanism. We conclude that a majority of MnPO neurones were characterized by a functional Cl⁻ transporter that ensured an efficient activity-dependent Cl⁻ transport rate, allowing sustained synaptic inhibition of these neurones. Pharmacological and anatomical data strongly suggested the involvement of KCC2, as an essential postsynaptic determinant of the inhibitory neurotransmission afferent to the MnPO, a key-structure in the physiology of the hydromineral and cardiovascular homeostasis.

The median preoptic nucleus (MnPO) is the median structure of the lamina terminalis (LT) and it is considered a pivotal brain site for integrating periphery-derived information pertinent to hydromineral balance (Mangiapane et al. 1983; Gardiner et al. 1986; Aradachi et al. 1996; McKinley et al. 1999). Central integration of this physiological information is supported by neuronal networks interconnecting the MnPO with other regions of the LT, especially the subfornical organ (SFO). Part of the SFO–MnPO connection probably conveys osmotic information from the plasma, as well as information required for long-term regulation of the hydromineral balance, like the angiotensin-induced dipsogenic response. Indeed, osmoreceptors and Na⁺ sensors have been identified in the SFO (Sibbald et al. 1988; Anderson et al. 2000; Hiyama et al. 2002) and angiotensin II (ANG II) alters the excitability of SFO neurones (Okuya et al. 1987; Li & Ferguson, 1993; Tanaka et al. 1995). GABA is a major neurotransmitter in the hypothalamus, particularly in the neuronal network involved in the osmoregulatory response (Tappaz et al. 1982; Randle et al. 1986; Decavel & Hatton, 1995; Di & Tasker, 2004). Neuroanatomical data reported the presence of GABAergic neurones in the SFO (Grob et al. 2003), some of them projecting to the MnPO as demonstrated with inhibitory synaptic events evoked by electrical stimulation of the SFO (Kolaj et al. 2004). Therefore, continuous adjustment in the strength of fast GABAergic neurotransmission at the SFO–MnPO synapse might represent a crucial aspect in the central process serving the hydromineral homeostasis.

Efficacy of the fast inhibitory synaptic transmission has been reported to be dependent on intrinsic properties of the postsynaptic neurone, such as the functional expression of cation–Cl⁻ cotransporters (for a review see Payne et al. 2003). In particular, the neuronal isoform of the K⁺–Cl⁻ cotransporter, KCC2, has been identified as an important regulator of neuronal Cl⁻ homeostasis, thereby modulating fast GABAergic synaptic communication in various brain structures (Payne et al. 1996; Rivera et al. 1999; Martina et al. 2001; Gulacsi et al. 2003; Payne et al. 2003; Woodin et al. 2003).

In the present study, we used an electrophysiological approach applied to an in vitro brain slice preparation preserving synaptic connectivity between the SFO and MnPO to investigate the cellular determinant of the GABA_A receptor-mediated synaptic transmission afferent to the MnPO. In particular, we took advantage of a bicarbonate-free extracellular solution and high intracellular chloride concentration to investigate the presence of a functional Cl⁻ transporter in MnPO neurones. Besides the role of a powerful Cl⁻ extrusion mechanism in adjusting the gain of the fast GABA_A receptor-mediated inhibitory synapse, we investigated the ability of such a cellular mechanism to maintain a powerful Cl⁻ gradient during sustained activity evoked at this central inhibitory synapse. This study will therefore document the influence of both constitutive and activity-dependent activity of a neuronal Cl⁻ transporter in regulating and maintaining strength of the synaptic GABA_A response at an identified neuronal connection of the LT, an essential structure in the physiology of the hypothalamus.

Methods

The experiments described in the present study were performed in accordance with the guidelines established by the Canadian Council on Animal Care and were duly approved by the Animal Care Committee of the Centre Hospitalier de l'Université Laval.

Hypothalamic slices

Acute hypothalamus slices were prepared from male Wistar rats (4–5 weeks old). The animals were anaesthetized with a ketamine–xylasine solution (43.7 and 1.25 mg kg⁻¹, respectively) injected intraperitoneally, and decapitated. Brains were quickly removed from the skull and submerged in ice-cold (2°C) artificial cerebrospinal fluid (aCSF) continuously bubbled with a gas mixture (95% O₂–5% CO₂) containing (mm): 123 NaCl, 3.1 KCl, 2.4 CaCl₂, 1.3 MgCl₂, 26 NaHCO₃, 1.2 NaHPO₄, 10 d-glucose, pH 7.4. Osmolarity was adjusted to 298–300 mosmol l⁻¹ with mannitol. One or two sagittal hypothalamus slices (350-μm-thick) containing both the SFO and MnPO were obtained with a vibratome (VT1000S; Leica, Nusloch, Germany), and then directly transferred to a submersion-type recording chamber (Warner Instruments Inc.; Hamden, CT, USA) mounted on a Gibraltar plateform (Burleigh Instruments, Inc.; Fishers, NY, USA). The slice was continuously bathed at 2–3 ml min⁻¹ with oxygenated aCSF containing (mm): 123 NaCl, 3.1 KCl, 2.9 CaCl₂, 1.3 MgCl₂, 22 Na-gluconate, 10 Hepes, 5 d-glucose, 0.4 ascorbic acid, 0.8 thiourea. pH 7.4; osmolarity 298–300 mosmol l⁻¹. Bath temperature was first maintained at 30°C for approximately 20 min, and then stabilized at 25°C for the patch-clamp recordings, using a heater controller (TC-344B, Warner Instruments).

Electrophysiology

Whole-cell patch recordings were mainly performed in neurones located in the ventral part of the MnPO, in a region immediately adjacent to the anterior commissure. A tight gigaohm seal was obtained on individual neurones under visual control using the near-infrared differential interference contrast principle. Patch pipettes were made from borosilicate glass capillaries (G75150T-4, Warner Instruments) with a resistance of about 4–5.5 MΩ. Pipettes were filled with a solution containing (mm): 124 K-gluconate, 12 KCl, 6 NaCl, 2 Na⁺-ATP, 0.1 Na⁺-GTP, 10 Hepes. pH was adjusted to 7.2 with KOH, and osmolarity to 295 mosmol l⁻¹ with sorbitol. For whole-cell perforated patch recordings, gramicidin was used for membrane perforation. For these experiments, the tip of the pipette was filled with the internal solution described above, and back-filled with the same internal solution containing gramicidin (5 μg ml⁻¹, Sigma).

Recordings were performed with an EPC9 amplifier (Heka Electronics, Inc., Mahone Bay, NS, Canada). Liquid junction potential was evaluated at 12.6 mV and membrane potential was corrected accordingly. The fast capacitance electrode was first compensated, and appropriate whole-cell and series-resistance compensation was applied after rupture of the cell membrane (series resistance 10.7 ± 1.1 MΩ). For perforated patch, adequate access to the cell was attained 5–15 min after seal formation (series resistance 10.8 ± 2.4 MΩ). Electrophysiological signals were filtered at 3 kHz, digitalized at 2 kHz, and stored on the computer hard drive for further analysis. Data analysis was performed using the Pulse/Pulsefit software (Heka Electronics).

Synaptic currents or potentials were evoked with a concentric bipolar tungsten electrode placed within the SFO or, sometimes, in the fibre track descending from the SFO and coursing along the fornix. To record pharmacologically isolated inhibitory postsynaptic currents (IPSCs) or potentials (IPSPs), electrical stimulation of the SFO was carried out in the presence of 1 mm kynurenic acid, a large spectrum blocker of ionotropic excitatory amino acid receptors. Furthermore, contribution of HCO₃⁻ anions to the GABA_A receptor-mediated anion currents was almost null, as the pH of the extracellular solution was balanced with Hepes. Therefore, under nominally HCO₃⁻-free conditions, the reversal potential of IPSCs (E_IPSC) approached the reversal potential of the chloride ions (E_Cl). Theoretical E_Cl was calculated from the Nernst equation. Under our experimental conditions:

with RT/F= 25.69 (t = 25°C), [Cl⁻]_out= 134.5 mm, [Cl⁻]_in= 18 mm.

In most recordings, MnPO neurones were voltage clamped at −60 mV, resulting in inward evoked IPSCs. A current-to-voltage relationship of IPSCs was built by varying the holding potential in 10 mV increments from −110 to −50 mV, and measuring the resulting IPSC amplitude (I). In a single neurone, the reversal potential of IPSCs (E_IPSC) was determined by using linear regression to calculate the best-fit line for I/I_max. I_max represented the amplitude of the IPSC measured at −110 mV and normalized to −1. The intercept of this line with the abscissa was taken as E_IPSC. The muscimol-induced GABA_A/Cl⁻ current was evaluated by subtracting the current resulting from a depolarizing voltage ramp (−90 to −60 or −50 mV; 16 mV s⁻¹) applied at the peak of the muscimol response (5 μm, 30 s) from a similar current elicited before muscimol application (Δcurrent). The reversal potential of the GABA_A/Cl⁻ current (E_GABAA) was then determined with the intercept of the Δcurrent with the 0 current line.

Drugs and application

Muscimol (Tocris Cokson, Inc., Ellisville, MO, USA) was applied on the ventral region of the MnPO using a fast solution changer and manifold (model RSC-160; Bio-Logic, Grenoble, France). Furosemide and bicuculline (RBI/Sigma, Natik, MA, USA) were added to the extracellular solution at the concentration indicated in the text. These drugs were bath applied at least 4 min for steady state conditions.

Statistical analysis

Raw data are expressed as means ± s.e.m. Statistical comparisons were performed using paired or unpaired Student's t test and P < 0.05 was considered significant.

Immunohistochemistry (IHC) protocol

Rats matching the age of the animals used for the electrophysiological study were deeply anaesthetized (ketamine, 80 mg kg⁻¹ and xylazine 10 mg kg⁻¹ solution, i.p.) and killed by a transcardiac perfusion of cold saline (4°C), immediately followed by a fixative solution (4% paraformaldehyde in PBS, 4°C). The brains were quickly removed, postfixed for 4 h and immersed in a sucrose solution (20%) diluted in 4% paraformaldehyde for 18 h at 4°C. Frozen brains were mounted on a microtome and sliced in 30 μm coronal sections that were collected in a cryoprotectant solution and then stored at −20°C.

Brain sections were washed in sterile potassium-buffered saline (KPBS) and incubated at 4°C overnight with primary antibody (rabbit anti-KCC2 1 : 1000, Upstate, Charlottesville, VA, USA) mixed in sterile KPBS, Triton X-100 (0.04%), bovine serum albumin (1%) and goat serum (1%). Brain sections were thereafter rinsed in sterile KPBS and incubated with a mixture of KPBS-heparin-biotinylated secondary antibody (goat antirabbit IgG 1 : 1500, BA 1000, Vector Laboratories, Burlingham, CA, USA) for 120 min. The sections were then rinsed with KPBS and incubated at room temperature for 60 min with avidin–biotin–peroxidase complex (Elite, Vector Laboratories). After several sterile rinses in sterile KPBS, brain sections were reacted in a mixture containing sterile KPBS, the chromagen 3,3′-diaminobenzidine tetrahydrochloride (DAB, 0.3%), and hydrogen peroxide (0.05%). Thereafter, tissues were rinsed in sterile KPBS, mounted onto poly l-lysine-coated slides, desiccated under vacuum overnight, dehydrated through graded concentrations of alcohol, cleared in xylene and coverslipped with DPX.

Specificity of the KCC2 antiserum was verified by preabsorbing the sera with the recombinant fusion protein containing residues 932–1043 of the rat KCC2 (Upstate). Alternate sections of the same brains (n = 3) were exposed to KCC2 antibody preabsorbed with 10–25 μm of the immunogen and to KCC2 antibody without preabsorption, as positive control. Experimental conditions for this test were similar to those described above. No KCC2 staining was observed either in the MnPO, cortex, or hippocampus of brain sections that were exposed to KCC2 antiserum preabsorbed with the immunogen.

Results

Characteristics of the postsynaptic GABA_A response in the MnPO neurones

Basic characteristics of the fast inhibitory synaptic transmission between neurones of the subfornical organ and neurones located in the ventral region of the median preoptic nucleus were first examined. Whole-cell patch-clamp recordings were made from MnPO neurones held at −60 mV and electrical stimulation of the SFO evoked pharmacologically isolated inhibitory postsynaptic currents (IPSCs) showing opposite direction. In 6 out of 32 cells tested, SFO stimulation evoked IPSCs of −11.6 ± 4.2 pA (Fig. 1Aa). The inward direction of the IPSCs was in accordance with the chloride equilibrium potential (E_Cl) predicted from the Nernst equation (E_Cl, −51.7 mV; see Methods for details). Contrastingly, in the remaining 26 neurones, similar SFO stimulation led to IPSCs of +14.7 ± 2.3 pA (Fig. 1Ab). These outward currents were considered as atypical inhibitory synaptic responses according to E_Cl. Inward and outward IPSCs did both result from the activation of postsynaptic GABA_A receptors, since bath application of bicuculline (50 μm) abolished the synaptic currents (Fig. 1Aa and 1Ab, lower traces). The opposite direction of the IPSCs recorded at −60 mV strongly suggested that the reversal potential of the synaptic GABA_A current (E_IPSC) could be different among neuronal populations of the MnPO. E_IPSC estimated from neurones showing inward IPSCs at −60 mV was −50.1 ± 1.9 mV, a value very close to E_Cl predicted from the Nernst equation (n = 6; Fig. 1Ba). Interestingly, E_IPSC estimated from a neuronal population showing outward IPSCs at −60 mV amounted to −75.7 ± 1.6 mV, a value hyperpolarized by about 20 mV compared to E_Cl (n = 26; Fig. 1Bb). The difference in E_IPSC was statistically significant (unpaired Student's t test, P < 0.01).

Figure 1. Heterogeneous synaptic GABA_A response in neurones of the MnPO.

In 19% of the MnPO neurones maintained at a membrane potential of −60 mV, electrical stimulation of the SFO evokes an inward inhibitory postsynaptic current that is blocked by bicuculline (Aa). However, in a majority of MnPO neurones recorded under identical conditions, stimulation of the SFO evokes an outward inhibitory postsynaptic current that is also blocked by bicuculline (Ab). The upper panel in Ba illustrates a typical intensity–voltage relationship of evoked IPSCs displaying an inward direction at −60 mV. Normalized intensity–voltage relationship obtained from six different neurones shows that the reversal potential of IPSCs matched E_Cl predicted from the Nernst equation (Ba, lower panel). I_max represents the amplitude of the IPSC measured at −110 mV under control conditions and normalized to −1. Contrastingly, the intensity–voltage relationship of evoked IPSCs displaying an outward direction at −60 mV is correlated with a reversal potential that is hyperpolarized by about 20 mV, compared to E_Cl (n = 26) (Bb).

It was conceivable that variety in cell volume may cause a slow and incomplete dialysis of MnPO neurones by the pipette solution, thereby affecting E_Cl in the neurones showing outward IPSCs. We therefore investigated the synaptic GABA_A response under an unaltered intracellular Cl⁻ concentration ([Cl⁻]_i). This was achieved with a perforated-patch recording configuration, using gramicidin as the monovalent cation-selective ionophore (Ebihara et al. 1995; Kyrozis & Reichling, 1995). In all the neurones recorded at −60 mV with the gramicidin-containing intracellular solution, electrical stimulation of the SFO evoked outward IPCSs amounting +9.9 ± 1.8 pA (n = 7, Fig. 2A). Outward polarity of the IPSCs was correlated with hyperpolarized E_IPSC that was estimated at −89.2 ± 4.3 mV (Fig. 2A). Rupturing the membrane after recording outward IPSCs and hyperpolarized E_IPSC was possible in some cells (n = 3). In these neurones held at −60 mV, rupture of the membrane patch produced a change in the polarity of the IPSCs from +10 ± 4 pA to −14.8 ± 5.7 pA. The change in polarity was correlated with a shift in E_IPSC toward a more depolarized value (−46.7 ± 5.8 mV; Fig. 2B). This latter observation was in good agreement with a rapid dialysis of the intracellular content with the high-chloride solution (18 mm) contained in the pipette. Taken together, these results indicated that MnPO neurones maintained a low [Cl⁻]_i (4.5 mm calculated from the Nernst equation) under preserved intracellular integrity.

Figure 2. Reversal potential of evoked IPSCs is estimated with gramicidin-perforated-patch recording.

When the whole-cell configuration is achieved using the monovalent cation-selective ionophore, gramicidin, the electrical stimulation of the SFO evokes an outward IPSC in MnPO neurones held at −60 mV (A, left panel). The outward direction of the evoked IPSC is associated with a hyperpolarized value of the reversal potential (E_IPSC), compared to E_Cl (n = 7) (A, right panel). I_max represents the amplitude of the IPSC measured at −110 mV under control conditions and normalized to −1. In a cell displaying outward IPSCs, rupture of the patch membrane reversed polarity of the evoked IPSC (B, left panel), which was correlated with a shift of E_IPSC toward a depolarized value (n = 3) (B, right panel).

To further evaluate a possible gradient for Cl⁻ in MnPO neurones, we compared E_IPSC with the reversal potential of muscimol-mediated Cl⁻ current (E_GABAA). MnPO neurones were held at −60 mV and local application of muscimol (5 μm, 30 s) over the MnPO region elicited an inward GABA_A receptor-mediated Cl⁻ current (GABA_A/Cl⁻ current) of −36.5 ± 5.1 pA in seven neurones tested (Fig. 3A). Contrastingly, in 19 other neurones, muscimol triggered an outward GABA_A/Cl⁻ current of +18.7 ± 3 pA (Fig. 3B). Both GABA_A responses were abolished by 50 μm bicuculline (data not shown). The intensity-to-voltage relationship of the GABA_A/Cl⁻ current was determined from slow depolarizing ramps (16 mV s⁻¹) triggered before and at the peak of the GABA_A/Cl⁻ current. In four neurones displaying inward GABA_A response at −60 mV, E_GABAA was estimated at −56.7 ± 0.6 mV (Fig. 3Ca and Da). Contrastingly, in 11 neurones displaying outward GABA_A response at −60 mV, E_GABAA was estimated at −69.8 ± 1.1 mV (Fig. 3Cb and Db).

Figure 3. Heterogeneous muscimol-activated GABA_A response in neurones of the MnPO.

A, typical example of a muscimol-activated inward GABA_A/Cl⁻ current recorded from MnPO neurones held at a membrane potential of −60 mV. B, similarly to the inhibitory postsynaptic current, a majority of MnPO neurones display an outward GABA_A/Cl⁻ current when neurones are held at −60 mV. A slow depolarizing ramp ranging from −90 to −40 mV (16 mV s⁻¹) is used to estimate the reversal potential of the GABA_A/Cl⁻ current (E_GABAA) in neurones showing an inward current at a membrane potential of −60 mV. The isolated GABA_A/Cl⁻ current (Δcurrent) is obtained by subtracting the ramp current elicited before and at the peak of the muscimol-mediated response (Ca). In neurones displaying an inward current at −60 mV, E_GABAA estimated from Δcurrent is in agreement with E_Cl predicted from the Nernst equation (n = 4) (Da). I_max is the amplitude of Δcurrent measured at −90 mV and normalized to −1. A similar ramp protocol is used to estimate E_GABAA in neurones displaying an outward GABA_A/Cl⁻ current when held at −60 mV (Cb). In these cells, E_GABAA estimated from Δcurrent shows a hyperpolarized value, compared to E_Cl (n = 11) (Db).

Our results indicated that MnPO contained a large population of neurones displaying a prominent GABA_A response at −60 mV. This response was driven by a hyperpolarized value of E_IPSC and/or E_GABAA, suggesting a powerful regulation of Cl⁻ homeostasis in these neurones.

A furosemide-sensitive chloride transporter contributed to the hyperpolarized E_IPSC in MnPO neurones

Among the diversity of the cellular parameters that affect the reversal potential of the postsynaptic GABA_A receptor-mediated response, regulation of the chloride and bicarbonate (HCO₃⁻) homeostasis has a major influence. The fact that a hyperpolarized value of E_IPSC was reported in the presence of a bicarbonate-free extracellular solution (Hepes buffer) suggested that a powerful Cl⁻ extrusion system was effective in these neurones to maintain an inward chloride electrochemical gradient. One possibility was the presence of a cation–chloride cotransporter insuring a low [Cl⁻]_i in MnPO neurones, as shown in hippocampal and midbrain neurones (Misgeld et al. 1986; Jarolimek et al. 1999). With reference to this possibility, we used furosemide to test the implication of such a cellular mechanism in regulating Cl⁻ homeostasis in MnPO neurones. MnPO neurones that putatively displayed an active Cl⁻ extrusion system were identified by the outward direction of the IPSCs recorded at −60 mV. In seven neurones, bath application of furosemide (100 μm) reversed the polarity of IPSCs from +18.7 ± 6.2 pA to −6.2 ± 1.3 pA (Fig. 4Aa). In these neurones, furosemide induced a large shift in E_IPSC from −76.7 ± 2.7 mV to −51.6 ± 2.9 mV (paired t test; P < 0.01), the latter being similar to E_Cl predicted from the Nernst equation applied to our experimental conditions (Fig. 4Ab). In agreement with these results, the amplitude of the muscimol-induced GABA_A/Cl⁻ current recorded at −60 mV was +21.1 ± 5.7 pA under control and −5 ± 4.4 pA in the presence of furosemide (n = 9; Fig. 4Ba). In these cells, E_GABAA shifted from −69 ± 0.9 mV under control conditions to −58.6 ± 2.7 mV in the presence of furosemide (paired t test; P < 0.02; n = 4; Fig. 4Bb).

Figure 4. A furosemide-sensitive cation–chloride cotransporter maintains a low intracellular Cl⁻ concentration in MnPO neurones.

Bath application of furosemide (100 μm), a pharmacological blocker of cation-chloride cotransporters, reverses polarity of the SFO evoked outward IPSC recorded at −60 mV (Aa). The intensity–voltage relationship of IPSCs reveals that furosemide produces a parallel shift of E_IPSC toward a more depolarized value that matches E_Cl predicted from the Nernst equation (n = 7) (Ab). Polarity of the muscimol-activated outward GABA_A/Cl⁻ current recorded at −60 mV is also reversed by bath application of furosemide (100 μm, Ba). The intensity–voltage relationship of the isolated GABA_A/Cl⁻ current highlights the furosemide-induced parallel shift of E_GABAA toward a more depolarized value (n = 4) (Bb). Note that I_max in Ab and Bb refers to the amplitude of the inward GABA_A/Cl⁻ currents measured at −90 mV under control conditions and normalized to −1.

To validate constitutive expression of the neuronal isoform of the K⁺–Cl⁻ cotransporter within the MnPO, immunohistochemical detection of KCC2 was performed with a specific polyclonal antibody directed against the N-terminal His-tag fusion protein of the rat KCC2. Specificity of the antibody was verified by preabsorbing KCC2 antiserum with the immunogen (see Methods). Immunohistochemistry was performed on rats that matched the age of those used for electrophysiology (n = 4). As shown in Fig. 5A, KCC2 immunoreactivity was densely present in the ventral region of the MnPO (left panel), as well as in the anteroventral periventricular nucleus of the hypothalamus (not shown). Higher magnification clearly indicated that KCC2 staining formed a network of intermingled fibres in this region of the MnPO where all the electrophysiological recordings have been performed (right panel). No obvious staining of the cell body was detected. Interestingly, KCC2 was apparently not expressed, or expressed at an undetectable level in the SFO, as no organized labelling was found in this structure (Fig. 5B), confirming the selective expression of the KCC2 gene in restricted regions of the preoptic region.

Figure 5. Selective expression of KCC2 cotransporter in the MnPO.

A, representative photomicrographs of the ventral region of the MnPO obtained from a 5-week-old rat. Picture on the right represents an enlargement of the framed region (broken line). Arrows point KCC2-containing fibres that are widely distributed in this area of the lamina terminalis. Immunohistochemistry was performed with a selective polyclonal antibody directed against residues 932–1043 of the rat KCC2 (dilution 1/1000). B, representative photomicrograph of the SFO obtained from the same animal. Note the absence of KCC2 immunoreactivity in this circumventricular organ.

Altogether, these results support the presence of a furosemide-sensitive Cl⁻ extrusion mechanism, probably the KCC2 cotransporter, in a large population of MnPO neurones, thereby contributing to maintenance of a low Cl⁻ concentration within these neurones.

Intrinsic properties of the neuronal populations of the MnPO

In an attempt to provide physiological relevance of KCC2 expression in MnPO neurones, we compared basic properties of 60 MnPO cells that did, or did not express functional Cl⁻ transporter. In this context, examination of the resting membrane potential (no DC current injected), time constant and membrane resistance did not reveal any significant difference between the two neuronal populations (Table 1). Two intrinsic properties have been also used to distinguish between the two populations. MnPO neurones have been previously classified into three categories, based on the presence, or the absence of rectifying properties in response to a series of hyperpolarizing current pulses (Bai & Renaud, 1998). A similar experimental protocol consisting of 5–10 pA increment hyperpolarizing current pulses elicited from the holding potential (−60 mV) allowed similar categories of neurones to be distinguished during our random recordings of ventral MnPO neurones (Fig. 6A). Our results showed that a majority of neurones that did not express the Cl⁻ transporter (71%) did not display any rectification in response to hyperpolarizing pulses, as illustrated in Fig. 6B (type 1 category). However, a large population of MnPO neurones displaying the Cl⁻ transporter (54%) behaved similarly in response to hyperpolarizing pulses (Fig. 6B). Neurones characterized by the presence of a time-dependent rectification (type 2 category) were found to display outward IPSCs (36%), or inward IPSCs (23%). The last category of neurones, i.e. those displaying a strong rectification without time dependence was rarely represented in MnPO neurones displaying outward IPSCs (10%), and in neurones displaying inward IPSCs (6%). The presence or absence of rectifying properties in response to hyperpolarization did not represent a specific electrophysiological fingerprint to categorize neurones expressing functional Cl⁻ transporter.

Table 1.

Intrinsic properties of MnPO neurones

	Resting potential (mV)	Time constant (ms)	Input resistance (MΩ)
KCC2-containing cell	−56 ± 1.6 (n = 10)	116 ± 4.5 (n = 40)	1.1 ± 0.08 (n = 40)
No KCC2	−53.7 ± 4.7 (n = 5)	117.2 ± 8.4 (n = 11)	1.2 ± 0.15 (n = 11)
Unpaired t test (P value)	0.55	0.9	0.67

Resting potential was evaluated with null injected DC current in the absence of TTX. Input resistance and time constant were evaluated with a hyperpolarized current pulse (800 ms), an amplitude that did not trigger hyperpolarization-induced rectification.

Figure 6. Rectifying property, firing pattern of ventral MnPO neurones and relationship with the expression of a functional Cl⁻ extrusion mechanism.

A, superimposed voltage responses to a series of incremented current pulses (illustrated in the insert) highlight the lack of membrane rectification in a subpopulation of neurones recorded from the ventral part of the MnPO (upper traces). Similar experimental protocol allows to distinction of two additional categories of neurones based on the presence of a hyperpolarization-activated inward rectification displaying either time dependency (middle traces), or time independency (lower traces). B, population histogram illustrates the percentage of the three categories of ventral MnPO neurones that displayed either outward IPSCs, or inward IPSCs evoked by electrical stimulation of the SFO. C, superimposed subthreshold voltage responses to depolarizing current pulses (illustrated in the insert) highlight distinct categories of ventral MnPO neurones based on their firing pattern. D, population histogram illustrates the percentage of each category of ventral MnPO neurones that displayed either outward IPSCs, or inward IPSCs evoked by electrical stimulation of the SFO.

In a next step, we analysed the discharge characteristics of the 60 MnPO neurones in response to transient suprathreshold current pulses (800 ms duration) applied through the recording pipette. As illustrated in Fig. 6C, four categories of neurones were distinguished by characteristics of their spike discharge pattern in response to stepwise depolarizing currents elicited from a membrane potential fixed around −60 mV. Two categories of neurones were frequently recorded in the MnPO. The first category was characterized by regular spike discharges showing light adaptation (n = 19; regular spiking with light adaptation; type A neurones). The second category of neurones was characterized by a short burst of action potentials superimposed on a robust calcium spike (n = 18; calcium spikes; type B neurones). A third category of neurones was characterized by a strong spike frequency adaptation (n = 14; phase spiking neurones; type C neurones). Finally, some MnPO neurones displayed regular spike discharge without adaptation (n = 9; regular spiking neurones; type D neurones). Fifty-one out of the 60 MnPO neurones displayed outward IPSCs at −60 mV, as reported in the previous sections. These majority of these neurones were in the type A (31%), or type B (35%) category (Fig. 6D). The other neurones (9 out of 60; 15%) displayed inward IPSCs at −60 mV in response to electrical stimulation of the SFO. These neurones were essentially of type C category (67%), with some cells (33%) belonging to type A category (Fig. 6D). These results seem to indicate that a distinct firing pattern of MnPO neurones may serve as an electrophysiological criterion to distinguish neurones with active Cl⁻ extrusion mechanism from the others. Neurones displaying calcium spikes or regular spiking without adaptation did unequivocally express the Cl⁻ transporter. In contrast, neurones showing a strong spike adaptation rarely expressed this protein. Furthermore, in the 60 neurones tested, no significant correlations between the presence, or the absence of rectification in response to hyperpolarization and the patterns of spike discharges were noticed in the neurones showing outward or inward IPCSs evoked by electrical stimulation of the SFO.

Cl⁻ gradient was maintained during high-frequency stimulation of the SFO efferent fibres in neurones displaying the Cl⁻ transporter

The presence of putative KCC2 cotransporter in a majority of MnPO neurones was functionally correlated with the maintenance of a powerful gradient for Cl⁻ ions, as demonstrated by the outward direction of evoked IPSCs recorded at −60 mV. Here, we investigated the ability of the Cl⁻ transporter to maintain powerful extrusion of Cl⁻ ions during sustained activation of the postsynaptic GABA_A receptors. Electrical stimulation of the SFO was achieved in the presence of CGP52432 (10 μm) in the extracellular solution, to rule out possible modulation of GABA release by the activation of presynaptic GABA_B receptors (Kolaj et al. 2004). Neurones were held at −60 mV, and a series of 15 consecutive IPSCs evoked at 5 Hz was sampled after verification of the outward direction of the IPSCs and stabilization of the IPSC amplitude. Furosemide was then bath-applied (6–10 min) at a concentration of 50 μm in order to strongly reduce, but not to block, the activity of the Cl⁻ transporter. Under this steady-state pharmacological condition, the polarity of single evoked IPSCs was not reversed, but their amplitude was significantly reduced, as illustrated in Fig. 7A (+23 ± 4.2 pA under control and +8.6 ± 2.1 pA under furosemide; paired t test, P < 0.01; n = 7). A similar series of 15 IPSCs elicited at 5 Hz was then repeated in the presence of furosemide and the amplitude of the first and last IPSC within the train was measured and compared to that obtained before furosemide application. Under control conditions, the strength of the synaptic inhibitory transmission was slightly attenuated during high-frequency stimulation, as illustrated in Fig. 7Ba and C. The amplitude of the last IPSC in the train was inhibited by 16.3 ± 6.9%, when compared to the first IPSC (paired t test, P = 0.08; n = 5; range 0–40.5% of inhibition). As expected for neurones displaying outward IPSCs, bath application of furosemide (50 μm) reduced the amplitude of IPSCs in the train. Indeed, the amplitude of the first IPSC was inhibited by 55.3 ± 6.1%, when compared to the first IPSC under control conditions (paired t test, P < 0.01; n = 5). Moreover, under this pharmacological condition, where the activity of the Cl⁻ transporter was strongly reduced, the strength of the inhibitory synaptic currents collapsed with high-frequency stimulation (Fig. 7Bb and C). Indeed, the amplitude of the last IPSC in the train was inhibited by 72.8 ± 12.2%, when compared to the first IPSC (paired t test, P < 0.01; n = 5; range 56.5–100% of inhibition).

Figure 7. Partial block of the Cl⁻ transporter activity with furosemide impaired sustained inhibitory synaptic neurotransmission.

A, bath application of 50 μm furosemide, a concentration that did not completely abolish Cl⁻ transporter activity, did not reverse polarity of the evoked IPSCs recorded at −60 mV, but strongly reduced their amplitude. Ba, under control condition, the amplitude of the evoked IPSCs was maintained during high-frequency stimulation (5 Hz) of the SFO. Bb, similar high-frequency stimulation carried out in the presence of furosemide greatly reduced IPSC amplitude within the train. C, histogram representing the normalized amplitude of the first (*) and the last IPSC (**) evoked at 5 Hz under control condition (open bars) and in the presence of furosemide (filled bars).

In the next series of experiments, we evaluated the influence of the Cl⁻ transporter in shaping sustained inhibition when intracellular chloride concentration was lowered from 18 mm to 8 mm, a consensus concentration established for mammalian neurones (McCormick, 1990). Under this experimental condition, electrical stimulation of the SFO evoked hyperpolarizing inhibitory postsynaptic potentials (IPSPs) in MnPO neurones held at −60 mV (−7.4 ± 1 mV; n = 10). The polarity of the IPSPs recorded at −60 mV could not support the presence of a functional Cl⁻ extrusion mechanism since theoretical E_Cl obtained with 8 mm[Cl⁻]_i was close to E_Cl reported in the previous sections of the present study (−72.1 mV versus−75.7 ± 1.6 mV and −76.7 ± 2.7 mV, respectively). Partial block of the Cl⁻ transporter activity with furosemide was expected, however, to reduce the size of the IPSPs, as well as the sustained inhibition produced by high-frequency stimulation of the SFO. In the presence of 10 μm CGP52432 in the extracellular solution, a train of 15 high-frequency stimulation (5 Hz) pulses induced a sustained inhibition of the neurone (Fig. 8Aa and B), where the amplitude of the last IPSP was slightly reduced when compared to the amplitude of the first IPSC (19.2 ± 5.4% of inhibition, paired t test, P = 0.012; n = 7; range 0–36.4% of inhibition). In these cells, bath application of 50 μm furosemide significantly reduced the amplitude of the first IPSP in the train by 42.4 ± 4.8%, when compared to the first IPSP in control conditions (paired t test, P = 0.0001; Fig. 8A and B). In addition, the amplitude of the last IPSP in the train was also inhibited by 48.5 ± 10.2%, when compared to the first IPSP (paired t test, P = 0.003; range 0–81.3% of inhibition; Fig. 8Ab and B). These results indicated that the strength of the inhibitory events was not maintained in the presence of furosemide. Interestingly, the situation was different in three other MnPO neurones tested for high-frequency stimulation. In these cells, evoked inhibition collapsed during the high-frequency stimulation, as shown by the strong reduction in the amplitude of the last IPSP, when compared to that of the first IPSP (52.8 ± 2.8% of inhibition; paired t test, P = 0.027; range 47.4–56.4% of inhibition; Fig. 9Aa and B). Bath application of furosemide (50 μm) did not affect the amplitude of the first IPSP (10.3 ± 4% of inhibition; paired t test, P = 0.124) and the amplitude of the last IPSP was reduced by 52.2 ± 11.7%, when compared to the first IPSP (paired t test, P = 0.046; range 29.6–68.6% of inhibition; Fig. 9Ab and B). These results indicated that MnPO neurones that did not express active furosemide-sensitive Cl⁻ transporter were not able to follow high-frequency inhibition.

Figure 8. The furosemide-sensitive Cl⁻ transporter was an essential component of the inhibitory synaptic neurotransmission in the MnPO.

Aa, in the presence of a lowered [Cl⁻]_i (8 mm), electrical stimulation of SFO efferents evoked consistent IPSPs in MnPO neurones held at −60 mV, whose amplitude was slightly reduced during high-frequency stimulation. Ab, bath application of furosemide (50 μm) dramatically reduced the amplitude of the evoked IPSPs and almost abolished sustained inhibition. B, histogram representing the normalized amplitude of the first (*) and the last IPSP (**) evoked at 5 Hz under control condition (open bars) and in the presence of furosemide (filled bars).

Figure 9. The lack of furosemide-sensitive Cl⁻ transporter impaired sustained inhibition in a minority of MnPO neurones.

Aa, in the presence of a lowered [Cl⁻]_i (8 mm), high-frequency (5 Hz) electrical stimulation of the SFO efferents evoked a train of IPSPs whose amplitude collapsed during the stimulation in a minority of MnPO neurones held at −60 mV. Ab, bath application of furosemide (50 μm) was without effect on the amplitude of the evoked IPSPs. B, histogram representing the normalized amplitude of the first (*) and the last IPSP (**) evoked at 5 Hz under control condition (open bars) and in the presence of furosemide (filled bars).

Discussion

The present study has provided solid evidence indicating that the strength of the GABA_A response evoked at an identified synapse of the hypothalamus was dependent on an intrinsic cellular mechanism regulating Cl⁻ homeostasis. Indeed, we demonstrated the functional expression of a furosemide-sensitive Cl⁻ transporter, probably the K⁺–Cl⁻ cotransporter, KCC2, in a majority of MnPO neurones. This Cl⁻ extrusion mechanism ensured an efficient inhibitory neurotransmission, even at hyperpolarized membrane potential, like the resting membrane potential of these MnPO neurones. In addition, we showed that the Cl⁻ transporter was tightly involved in the maintenance of the efficacy of sustained inhibition afferent to these neurones. These results indicate that postsynaptic outward Cl⁻ transport was a crucial cellular mechanism regulating the efficacy of the GABAergic neurotransmission in the MnPO.

Strength of the synaptic GABA_A response is controlled by an active regulation of the Cl⁻ homeostasis in neurones of the MnPO

Fast synaptic inhibition in the adult brain is largely mediated by GABA type A receptors, which are permeable to Cl⁻ ions. The efficacy of the GABA_A receptor-mediated inhibition is thus, dependent on the transmembrane Cl⁻ gradient, but also on the bicarbonate anion permeability of the channel pore (Bormann et al. 1987; Kaila et al. 1993; for review see Kaila, 1994). It is now accepted that transport proteins can efficiently regulate intracellular Cl⁻ concentration (Kaila, 1994), and among these proteins, the neuronal-specific isoform of the K⁺–Cl⁻ cotransporter, KCC2 (Payne et al. 1996), regulates Cl⁻ homeostasis in a large variety of mature neurones of the central nervous system (DeFazio et al. 2000; Kakazu et al. 2000; Martina et al. 2001; Gulacsi et al. 2003; Woodin et al. 2003). Therefore, to validate the hypothesis of a synaptic GABA_A response regulated by an efficient postsynaptic Cl⁻ transport system in the MnPO neurones, we combined three different strategies. First, we manipulated both the intra- and extracellular Cl⁻ concentration under nominally HCO₃⁻-free conditions, an appropriate condition to characterize the presence of a cation/Cl⁻ cotransport. Under this experimental condition, we showed that evoked IPSCs, or muscimol-activated GABA_A/Cl⁻ current, recorded at a membrane potential of −60 mV were both characterized by an outward direction and a hyperpolarized reversal potential, compared to E_Cl predicted from the Nernst equation, in a majority of MnPO neurones (81%). Furthermore, the observation of outward IPSCs and hyperpolarized E_IPSC under gramicidin-perforated-patch recordings, a whole-cell configuration permitting reliable recordings of the IPSCs with a preserved cellular integrity (Ebihara et al. 1995; Kyrozis & Reichling, 1995), emphasized the presence of an active outward Cl⁻ transport in a large population of MnPO neurones. Second, we used a pharmacological approach to identify the Cl⁻ transporter in the MnPO neurones. The change in both the polarity of IPSCs, or GABA_A/Cl⁻ current, and the reversal potential of the inhibitory response by furosemide strongly suggested the presence of the neuronal isoform of the K⁺–Cl⁻ cotransporter KCC2 (Misgeld et al. 1986; Thompson & Gahwiler, 1989; Payne, 1997; Jarolimek et al. 1999; Martina et al. 2001). Finally, we performed an immunohistochemical study using a specific antibody directed against the N-terminal His–tag fusion protein of rat KCC2 to demonstrate the participation of this K⁺–Cl⁻ cotransporter in generating an inwardly directed Cl⁻ electrochemical gradient in MnPO neurones. Combination of the anatomical observations with the electrophysiological and pharmacological data clearly validated the functional expression of a cation/Cl⁻ cotransporter in the MnPO, probably the neurone-specific isoform, KCC2.

Estimation of E_IPSC in the neurones of the MnPO expressing putative KCC2

Previous data obtained from MnPO neurones recorded in a bicarbonate-containing aCSF, indicated that E_IPSC was estimated around −66 mV in these cells (Kolaj et al. 2004). Although our data present discrepancies at a first glance, they are complementary to the results reported by Kolaj and colleagues. Indeed, using experimental conditions particularly appropriate to assess cellular mechanisms regulating Cl⁻ homeostasis in neurones, we did present electrophysiological and pharmacological evidence in favour of two distinct populations in the ventral part of the MnPO, based on the functional expression of a Cl⁻ transporter, probably the KCC2 protein. The previous study reported a more depolarized value of E_IPSC obtained from neurones randomly recorded in the ventral and dorsal part of the MnPO. Therefore, mean E_IPSC reported in this study might reflect averaged E_IPSC sampled among the two different neuronal populations that were characterized in the present study. It has to be noted that E_IPSC in our recording conditions was artificially set to a depolarized value (−52 mV) for technical purposes, i.e. using IPSC polarity as an indicator of a functional Cl⁻ extrusion mechanism in the recorded neurones. Therefore, the real value, or physiological value of E_IPSC cannot be deduced from our recordings in the minority of MnPO neurones where E_IPSC matched E_Cl. In this neuronal population, regulation of Cl⁻ homeostasis might be achieved by cellular mechanisms other than a cation–Cl⁻ cotransporter (Payne et al. 2003), and a contribution of bicarbonate permeability of the GABA_A receptor anion pore might also influence E_IPSC. In the majority of MnPO cells showing an active Cl⁻ extrusion system, especially those recorded with the gramicidin-perforated patch, E_IPSC might reflect the reversal potential of the inhibitory synaptic current, as the presence of bicarbonate did not influence E_IPSC in neurones expressing functional KCC2 (Gulacsi et al. 2003). A marked difference in E_IPSC value (about 12 mV), was, however, noticed between recordings carried out with a perforated-patch and a conventional whole-cell configuration. The discrepancy between the two values might result from the experimental procedure itself. Indeed, perforated-patch recordings were known to preserve the integrity of the intracellular compartment, including regulatory mechanisms that might optimize the activity of the KCC2 cotransporter. Therefore, estimation of E_IPSC obtained with the gramicidin-perforated-patch configuration (−89.2 ± 4.3 mV) might represent real E_IPSC in MnPO neurones expressing putative KCC2. Such a hyperpolarized reversal potential for GABA_A synaptic events was encountered in other neurones of the CNS impaled with sharp electrodes, a recording technique that also preserves intracellular integrity (Mouginot & Gähwiler, 1995). A difference in the reversal potential of fast inhibitory synaptic events recorded with conventional whole-cell and perforated-patch configuration has already been reported (Martina et al. 2001), and this observation combined with ours raised the hypothesis that the active conformational state of the K⁺–Cl⁻ transporter might be altered by dialysis of the intracellular compartment with the pipette solution. Few data are available on the regulation of KCC2 activity in central neurones. KCC2 displays a tyrosine phosphorylation consensus site (Payne, 1997) and one study performed on cultured hippocampal neuronal cells has shown that KCC2 function was deactivated by protein tyrosine kinase inhibitors (Kelsch et al. 2001). Additional data need to be collected to demonstrate the regulation of KCC2 in central neurones, especially in the MnPO. However, it remains highly plausible that in MnPO neurones recorded with conventional whole-cell configuration, the value of E_IPSC might be underestimated, as suggested by the results obtained with the gramicidin-perforated-patch.

Efficient outward Cl⁻ transport was necessary to maintain inhibition of MnPO neurones during sustained GABA neurotransmission

The presence of putative KCC2 was essential to maintain a hyperpolarized E_IPSC in a large population of MnPO neurones, thereby ensuring a powerful inhibition of these cells during stimulation of inhibitory projection neurones from the SFO. Here, we showed that sustained stimulation of the SFO GABAergic efferents provided an efficient reduction of the excitability of the MnPO neurones. The role of the Cl⁻ extrusion mechanism was essential in the maintenance of the fast inhibition, since pharmacological reduction of the Cl⁻ transporter activity led to a dramatic reduction in the strength of the GABAergic input during high-frequency stimulation of the SFO efferent fibres. This important observation indicated that outward Cl⁻ transport was able to rapidly counteract massive entry of Cl⁻ ions, to maintain Cl⁻ homeostasis and, thereby, a strong gradient for Cl⁻ ions. This result indicated that constitutive activity of putative KCC2 in MnPO neurones was able to exhibit a higher transport rate when intracellular chloride concentration increased, as shown for K⁺–Cl⁻ cotransporters in a model of avian red blood cells (Lytle & McManus, 2002). This activity-dependent Cl⁻ transport rate has to be considered as an essential cellular mechanism for maintaining sustained GABAergic neurotransmission, not only in hypothalamic neurones, but also in all the neurones expressing functional KCC2 cotransporter. This was emphasized by the fact that high-frequency inhibition collapsed in MnPO neurones lacking functional Cl⁻ transporter (furosemide-insensitive IPSPs). Localization of the KCC2 cotransporter on dendritic shafts has been reported in various types of central neurones (Coull et al. 2003; Gulacsi et al. 2003; Woodin et al. 2003), especially in the vicinity of inhibitory synapses (Gulacsi et al. 2003). The pattern of expression of the KCC2 protein in the MnPO (mainly on fibres), as well as the weak reduction in IPSP amplitude during high-frequency stimulation, may suggest that the Cl⁻ transporter was located close to the postsynaptic GABA_A receptor, thereby ensuring an efficient extrusion of local Cl⁻ ions to maintain the driving force at the inhibitory GABA_A synapses.

Physiological implication

MnPO neurones are involved in the integrated autonomic regulation as an essential component of a wide preoptic neuronal network connecting circumventricular organs to neuroendocrine cells and preautonomic neurones in the parvocellular division of the paraventricular nucleus of the hypothalamus. The finding that a large population of the MnPO neurones expressed a Cl⁻ transporter, probably the KCC2 K⁺–Cl⁻ cotransporter, and maintained a powerful GABA_A response at hyperpolarized membrane potentials, even during sustained inhibitory neurotransmission, strongly suggested that fast inhibition might be a crucial aspect of the physiology of the MnPO. Indeed, MnPO neurones have been shown to act as genuine Na⁺ sensors (Grob et al. 2004), and hyponatremia-induced hyperpolarization of the neurones might therefore, summate with GABA_A-mediated inhibitory input to strongly reduce excitability of the MnPO neurones. This physiological aspect was supported by the fact that neurones expressing putative KCC2 were considered as Na⁺ sensors, since these neurones did respond to the two stimuli, at least on the subpopulation tested (M.G., personal communication). One critical aspect in the physiology of the SFO–MnPO inhibitory projection was the identification of the neurochemical content of the postsynaptic MnPO cells that expressed functional Cl⁻ transporter. Such an identification would give a more precise picture of the influence of the SFO projection on the activity of the preoptic neuronal network. Indeed, the SFO inhibitory synapses might turn into an inhibition of excitatory neurones, or an inhibition of inhibitory neurones with, therefore, opposite effect on the activity and/or ouput of the neuronal network. The firing pattern used as an obvious electrophysiological fingerprint to categorize MnPO neurones indicated that more than one-third of the putative KCC2-expressing neurones specifically displayed Ca²⁺ spikes and a bursting firing pattern. Interestingly, a recent electrophysiological study (Spanswick & Renaud, 2005) reported that a subpopulation of MnPO neurones displayed bursting electrical activity in response to exogenous angiotensin II, that shared similarities with the bursting neurones described in the present study. It is therefore tempting to speculate that inhibitory projection from the SFO may target these neurones to reduce bursting activity in response to angiotensin stimulation, or to ensure synchronization of neuronal activity of this subpopulation of MnPO neurones, as shown in pyramidal hippocampal cells (Traub et al. 1998). Further studies are required to give a clear view of the neuronal populations that compose the MnPO and how fast neurotransmitter, like GABA, and neuropeptides can act together to control the neuronal excitability of these populations.