Regulation of tonic GABA inhibitory function, presympathetic neuronal activity and sympathetic outflow from the paraventricular nucleus by astroglial GABA transporters

Abstract

Neuronal activity in the hypothalamic paraventricular nucleus (PVN), as well as sympathetic outflow from the PVN, is basally restrained by a GABAergic inhibitory tone. We recently showed that two complementary GABA_A receptor-mediated modalities underlie inhibition of PVN neuronal activity: a synaptic, quantal inhibitory modality (IPSCs, I_phasic) and a sustained, non-inactivating modality (I_tonic). Here, we investigated the role of neuronal and/or glial GABA transporters (GATs) in modulating these inhibitory modalities, and assessed their impact on the activity of RVLM-projecting PVN neurons (PVN-RVLM neurons), and on PVN influence of renal sympathetic nerve activity (RSNA). Patch-clamp recordings were obtained from retrogradely labelled PVN-RVLM neurons in a slice preparation. The non-selective GAT blocker nipecotic acid (100–300 μm) caused a large increase in GABA_AI_tonic, and reduced IPSC frequency. These effects were replicated by β-alanine (100 μm), but not by SKF 89976A (30 μm), relatively selective blockers of GAT3 and GAT1 isoforms, respectively. Similar effects were evoked by the gliotoxin l-α-aminodipic acid (2 mm). GAT blockade attenuated the firing activity of PVN-RVLM neurons. Moreover, PVN microinjections of nipecotic acid in the whole animal diminished ongoing RSNA. A robust GAT3 immunoreactivity was observed in the PVN, which partially colocalized with the glial marker GFAP. Altogether, our results indicate that by modulating ambient GABA levels and the efficacy of GABA_AI_tonic, PVN GATs, of a likely glial location, contribute to setting a basal tone of PVN-RVLM firing activity, and PVN-driven RSNA.

γ-Aminobutyric acid (GABA) is the major inhibitory neurotransmitter in the mammalian nervous system. Within the hypothalamus, GABA has been shown to play fundamental roles in reproduction, energy and fluid balance, and autonomic control (Li & Patel, 2003; van den Pol, 2003; Henderson, 2007).

A major hypothalamic centre mediating GABA actions on autonomic and neuroendocrine control is the paraventricular nucleus (PVN) (Herman et al. 2002; Li & Patel, 2003; Stern, 2004; Li et al. 2006). The PVN is recognized as a critical centre for the coordination of autonomic and neuroendocrine homeostatic responses (Swanson & Kuypers, 1980; Guyenet, 2006). PVN influence on autonomic outflow is largely mediated through descending projections to brainstem and spinal cord sympathetic centres, including the rostral ventrolateral medulla (RVLM) and the intermediolateral column of the spinal cord (Swanson & Kuypers, 1980; Shafton et al. 1998; Pyner & Coote, 2000). Accumulating evidence support an important role for PVN-RVLM projections in autonomic control of the cardiovascular system. For example, PVN-RVLM neurons are activated in response to fluid balance and cardiovascular challenges (Badoer & Merolli, 1998; Horiuchi et al. 1999; Stocker et al. 2006), and their activation elicits sympathoexcitatory and pressor responses (Coote et al. 1998; Yang & Coote, 1998; Tagawa & Dampney, 1999). Importantly, PVN-RVLM projections have been shown to contribute to enhanced sympathoexcitatory drive during hypertension (Allen, 2002).

Most GABA actions within the PVN are mediated by ionotropic GABA_A receptors. Both PVN-RVLM neurons and sympathetic outflow are tonically inhibited by GABA (Zhang & Patel, 1998; Chen et al. 2003; Li et al. 2003; Li & Pan, 2005), and an altered PVN GABAergic function has been demonstrated to contribute to enhanced sympathoexcitatory drive during hypertension (Martin & Haywood, 1998; Li & Pan, 2006) and heart failure (Zhang et al. 2002).

In a recent study we demonstrated that similar to other CNS regions (Brickley et al. 1996; Salin & Prince, 1996; Nusser & Mody, 2002; Semyanov et al. 2003), GABA mediates two different inhibitory modalities in PVN-RVLM neurons: a conventional quantal synaptic form of inhibition (also term phasic inhibition), and a slower, persistent form of inhibition (termed tonic inhibition) (Park et al. 2007). Tonic inhibition appears to result from the persistent activation of GABA_A receptors, often located remotely from synapses (Semyanov et al. 2004; Farrant & Nusser, 2005). Despite this wealth of information supporting a critical role for GABA-mediated inhibition in the PVN, the precise mechanisms influencing GABAergic inhibitory function in sympathetic-related neurons in this region are still poorly understood.

The efficacy of GABAergic inhibitory function is dependent on multiple factors, including presynaptic (e.g. number of presynaptic contacts and probability of neurotransmitter release) and postsynaptic ones (e.g. number and types of postsynaptic GABA receptors (Mody et al. 1994). Another key factor influencing synaptic efficacy is the topography of neuronal-glial microenvironments. For example, by tightly enwrapping synaptic sites, thin glial processes act as a diffusion barrier (Piet et al. 2004), limiting the ability of synaptically released GABA to spill over from the synaptic cleft (Isaacson et al. 1993). Furthermore, this particular neuronal–glial arrangement favours a strong and efficient removal of GABA from the cleft, through active uptake mechanisms (Conti et al. 2004). The presence and functional role of GABA transporters (GATs) has been well characterized in major CNS areas, including hippocampus and cerebellum (see review, Dalby, 2003), where GATs have been shown to play a critical role in modulating tonic GABA_A inhibition, as well as overall network excitability. Four GATs so far have been cloned and characterized (GAT1–4) (Guastella et al. 1990; Lopez-Corcuera et al. 1992; Liu et al. 1992, 1993). GAT1 is the most abundantly expressed in the brain, and is predominantly located in presynaptic neuronal terminals (Gadea & Lopez-Colome, 2001; Schousboe, 2000). On the other hand, the GAT3 and GAT4 transporters are predominately expressed in glia and other non-neuronal cells (Minelli et al. 1996, 2003; Ribak et al. 1996).

The anatomical and functional properties of neuronal–glial microenvironment has been well studied and characterized in the hypothalamic magnocellular neurosecretory system (Hatton, 2004; Oliet et al. 2004), in which neurotransmitter transporters were reported to play a major role in the modulation of neuronal activity and neurosecretory function, both in control and stimulated conditions (Oliet et al. 2004; Piet et al. 2004). Conversely, whether and how GABA transporters shape GABA inhibitory function in hypothalamic neurons, and whether this process efficiently influences overall sympathetic outflow from this region, is at present unknown.

To address these important questions, we combined in the present study immunohistochemistry, in vitro patch clamp recordings, and whole animal nerve recordings. We determined the role of GABA transporters in modulating phasic and tonic GABAergic inhibitory efficacy in PVN-RVLM neurons, as well as their influence on sympathetic outflow from the PVN.

Methods

Male Wistar rats (180–220, n= 55, Harlan Laboratories, Indianapolis, IN, USA) were housed with a 12/12 h light–dark schedule and allowed free access to food and water. All animal experiments adhered to the policies of the Medical College of Georgia, as well as the University of Nebraska Medical Center, regarding the use and care of animals.

Retrograde labelling

PVN-RVLM neurons were retrogradely labelled as previously described (Park et al. 2007). Briefly, rats were anaesthetized with an intraperitoneal injection of a ketamine–xylazine mixture (90 mg kg⁻¹ and 5 mg kg⁻¹, respectively). The depth of anaesthesia achieved was monitored using a positive toe and tail pinch, the respiration rate and the degree of muscle relaxation. Rhodamine-labelled microspheres (Lumaflor, Naples, FL, USA) were microinjected unilaterally (200 nl) in the RVLM at bregma (B) −11.96, L 2.0, D 8.0, in anaesthetized rats. Postoperatively, analgesic treatment was provided (a single local s.c. infusion of 0.1 ml lidocaine) and a heat pad was used to provide supplemental heating until rats fully recovered from the anaesthesia. Animals were allowed to recover, and electrophysiological experiments were performed 5–10 days following the microinjections. The injection sites and extensions were confirmed histologically, as previously described (Li et al. 2003; Sonner & Stern, 2007). Animals showing a misplaced injection were discarded from this study.

Electrophysiology and data analysis

Patch-clamp recordings from identified PVN-RVLM neurons were obtained in hypothalamic slices (200 μm) as previously described (Park et al. 2007). Briefly, rats were anaesthetized with pentobarbital sodium (50 mg kg⁻¹i.p.) and decapitated, and brains were rapidly extracted. Slices were perfused with artificial cerebrospinal fluid (aCSF) containing (in mm): NaCl 126; KCl 2.5; MgSO₄ 1; NaHCO₃ 26; NaH₂PO₄ 1.25; glucose 20; ascorbic acid 0.4; CaCl₂ 1; pyruvic acid 2; pH 7.3–7.4, saturated with 95% O₂–5% CO₂. Recordings were obtained either at room temperature or at 35°C, as indicated, using a Multiclamp 700A amplifier (Axon Instruments, Union City, CA, USA). Current and voltage output were filtered at 2 kHz and digitized at 10 kHz (Digidata 1322A, pCLAMP 9 software, Axon Instruments). For voltage-clamp experiments, patch pipettes were filled with a high Cl⁻ containing solution (in mm): 140 KCl, 10 Hepes, 0.9 Mg-ATP, 20 sodium phosphocreatine, 0.3 Na-GTP and 10 EGTA. For current-clamp experiments, 140 KCl was replaced with 130 potassium gluconate + 10 KCl.

The holding current (I_holding) and root mean square (RMS) noise were measured in 50 ms epochs of traces lacking PSCs, separated by ∼800 ms, in periods of control aCSF and in the presence of the GABA_A blocker bicucculline (BIC, 20 μm) (n= 40 epochs in each case). The mean I_holding within the control and treated period was calculated, and the GABA_A receptor-mediated tonic current (I_tonic) was defined as the difference in I_holding before and after application of the GABA_A receptor blockers. RMS noise was measured in the same epochs using MiniAnalysis (Synaptosoft, Decatur, GA, USA). Spontaneous inhibitory postsynaptic currents (sIPSCs, recorded at −70 mV), were detected and analysed using MiniAnalysis as previously described (Park et al. 2006). Previous studies from our laboratory indicated that under basal conditions, spontaneous synaptic activity and GABA_AI_tonic were independent of action potential activity in our slice preparation (Park et al. 2006, 2007). Thus, all recordings in the present study were performed in the absence of the Na⁺ channel blocker tetrodotoxin (TTX).

Immunohistochemistry

Single and double immunohistochemical fluorescence reactions were used to study the expression of the GABA transporters GAT1 and GAT3 in PVN-RVLM neurons, and their possible colocalization with astroglial cell/processes (glial fibrillary acid protein (GFAP) immunoreactive) (Park et al. 2006). For these studies, a group of rats (n= 4) was deeply anaesthetized with sodium pentobarbital (100 mg kg⁻¹i.p.) and perfused transcardially with 0.01 m phosphate-buffered saline (PBS; 150 ml) followed by 4% paraformaldehyde (500 ml). Brains were post-fixed in 4% paraformaldehyde for 4 h at 4°C. Fixed brains were cryoprotected at 4°C with 0.1 m phosphate buffered saline (PBS) containing 30% sucrose for a minimum of 48 h. Sections (25 μm) were then cut with a cryostat, and incubated in a solution of 0.01 m PBS with 0.01% Triton X-100 and 10% normal goat serum for 1 h.

For single immunofluorescence reactions, sections were incubated for 48 h in the presence of a polyclonal rabbit anti-GAT1 or –GAT3 primary antibodies (Chemicon International, Temecula, CA, USA, 1: 100 dilution). For double immunofluorescence reactions, sections were incubated for 48 h in a mix of primary antibodies that included the polyclonal rabbit anti-GAT1 or GAT3 antibody (as described above) in conjunction with a monoclonal mouse anti-GFAP (Chemicon International, Temecula CA, 1: 10 000 dilution) antibody. Incubation in primary antibodies was followed by a 4 h incubation in secondary antibodies (donkey anti-rabbit fluorescein isothiocyanate (FITC) labelled, together with a donkey anti-mouse Cy5 labelled, 1: 200 and 1: 50 dilutions, respectively). All antibodies were diluted with PBS containing 0.1% Triton X-100. All secondary antibodies were obtained from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA, USA). Control experiments were performed by omitting primary or secondary antibodies.

Histological sections were examined with a Zeiss LSM 510 confocal microscope system. The argon–krypton and He–Ne lasers were used to excite the FITC (488 nm) and the Cy5 fluorophores (633 nm), respectively. Fluorescence signal cross-talk between the two channels was avoided by setting image-acquisition parameters with individually labelled sections. To determine co-localization of various fluorescent markers, a stack of confocal images (5 consecutive sections, 1 μm Z-step interval) were obtained from the different fluorophores, and merged. A positive co-localization was considered by the appearance of yellow (red + green) profiles in the merged image. Figures were composed using Adobe Photoshop (Adobe Systems Inc.).

Haemodynamic and renal sympathetic nerve activity measurements

On the day of the experiment, rats were anaesthetized with urethane (0.75 g kg⁻¹i.p.) and α-chloralose (70 mg kg⁻¹i.p.) and the left femoral vein was cannulated and connected to a computer-driven data recording and analysing system (PowerLab, ADInstruments) via a pressure transducer (Gould P23 1D) for recording arterial blood pressure (BP) and heart rate (HR). The anaesthetized rat was placed in a stereotaxic apparatus (Davis Kopf Instruments; Tujanga, CA, USA). A longitudinal incision was made on the head and the bregma was exposed. The coordinates for the PVN were determined from the Paxinos and Watson Atlas (Paxinos & Watson, 1986). They were 1.5 mm posterior to the bregma, 0.4 mm lateral to the midline, and 7.8 mm ventral to the dura. A small burr hole was made in the skull. For the microinjections, a thin needle (0.5 mm o.d. and 0.1 mm i.d.) connected to a microsyringe (0.5 ml; model 7000.5 Hamilton microsyringe) was lowered into the PVN. RSNA was recorded as described previously (Li et al. 2006). Briefly, the left kidney was exposed through a retroperitoneal flank incision. A branch of the renal nerve was isolated from the fat and connective tissue and was placed on a pair of thin bipolar platinum electrodes. The nerve–electrode junction was insulated electrically from the surrounding tissue with a silicone gel (Wacker Sil-Gel, 604 A B). The electrical signal was amplified (10 000 times) with a Grass amplifier (P55, Grass Technologies/Astro-Med Inc., West Warwick, RI, USA) with a high- and low-frequency cutoff of 1000 and 100 Hz, respectively. The output signal from the Grass amplifier was directed to a computer-run data acquisition system (PowerLab) to record and integrate the raw nerve discharge. The signal recorded at the end of the experiment (after the rat was dead) was deemed as background noise. The basal value of the nerve activity was defined by subtracting the background noise from the actual nerve activity value before the administration of drugs into the PVN. The peak response of RSNA to the administration of drugs into the PVN during the experiment (averaged over a period of 20–30 s) was subsequently expressed as a percent change from baseline, and values compared to those evoked by a microinjection of the same volume of cerebrospinal fluid (CSF).

Statistical analysis

Numerical data are presented as means ±s.e.m. Student's t test for paired data was used to compare the effects of a drug treatment vs. baseline. One-way analysis of variance (one-way ANOVA) followed by Tukey's post hoc test (OriginPro 8.0, OriginLab Corp., Northampton, MA, USA) was used to compare the effects among drug treatments. Cumulative histograms were compared using the Kolmogorov–Smirnov test. Dunnett's multiple comparison test was used to compare the effects of PVN microinjections of different GAT blockers against control CSF microinjections.

Results

Electrophysiological recordings were obtained from a total of 61 PVN-RVLM neurons, which had a mean input resistance of 971.7 ± 69.7 MΩ. To isolate and study GABA_A-mediated currents, recordings were performed in the presence of the AMPA and NMDA glutamate receptor antagonists CNQX (10 μm) and AP-5 (100 μm). As previously described (Park et al. 2007), PVN-RVLM neurons were found to be under the influence of two distinct GABA_A receptor-mediated inhibitory modalities: I_phasic (fast postsynaptic GABA_A IPSCs) and a sustained, non-inactivating I_tonic, which became evident as a shift in the holding current (I_holding) in the presence of the GABA_A receptor antagonist BIC (Fig. 1).

Figure 1. Blockade of GABA transporters (GATs) increased a GABA_A receptor-mediated tonic current (I_tonic) in PVN-RVLM neurons.

A, bath application of the non-selective GAT blocker nipecotic acid (NA, 300 μm) induced a large inward shift in I_holding, which was blocked by the GABA_A receptor blocker BIC (20 μm). Note that transient GABA_A-mediated sIPSCs (arrow) are blocked by BIC. B, the increase in I_holding evoked by NA was dose dependent and transient, with I_holding returning to baseline after drug wash-out. C, bath application of the GAT1 selective blocker SKF 89976A (SKF, 30 μm) caused minimal changes in I_holding. D, the GAT3 blocker β-alanine (β-Ala, 100 μm) induced a robust and significant inward shift in I_holding, which was partially blocked by BIC. E, β-Ala-induced shift in I_holding was completely blocked by combined application of BIC and the glycine receptor blocker strychnine (STR, 10 μm). Each panel represents data from a different neuron. IPSCs in panels B and C were truncated.

Modulation of GABA_AI_tonic by neuronal and glial GABA transporters

In various neuronal populations, including hippocampal interneurons, cerebellar granule cells and magnocellular neurosecretory neurons, I_tonic was reported to be influenced by ambient GABA concentration, known to be tightly control by the activity of GABA transporters (GATs) (Wall & Usowicz, 1997; Nusser & Mody, 2002; Rossi et al. 2003; Semyanov et al. 2003; Park et al. 2006). To determine if this was the case in PVN-RVLM neurons, the effects of GAT blockers on the magnitude of I_tonic were tested (Fig. 1).

Bath application of the non-selective GAT blocker nipecotic acid (NA, 100 and 300 μm) (Schousboe et al. 1979) caused a relatively large inward shift in I_holding (12.16 ± 3.35 pA, n= 6 and 32.79 ± 5.04 pA, n= 11, 100 and 300 μm NA, respectively; P < 0.01), an effect that was blocked by the GABA_A receptor blocker BIC (Fig. 1B). These results suggest that blockade of GATs leads to persistent accumulation of GABA in the extracellular space and further activation of extrasynaptic GABA_A receptors. RMS increased from 3.05 ± 0.30 pA to 4.29 ± 0.96 pA (n= 76, P= 0.06) and from 2.85 ± 0.24 pA to 6.42 ± 0.56 pA (n= 11, P < 0.01) by 100 and 300 μm NA, respectively.

Bath application of the GAT1 selective blocker SKF 89976A (30 μm) (Larsson et al. 1988; Solis & Nicoll, 1992; Rossi & Hamann, 1998) caused minimal changes in I_holding or RMS (I_holding: 4.14 ± 0.80 pA, n= 5, P < 0.05, RMS change: from 3.40 ± 0.45 to 4.04 ± 0.72, n= 5, P > 0.16) (Fig. 1C). On the other hand, the potent GAT3 blocker β-alanine (β-Ala, 100 μm) (Borden, 1996; Rossi et al. 2003) induced a large inward shift in I_holding (89.25 ± 22.66 pA, P < 0.01) as well as a significant increase in RMS (from 2.69 ± 0.15 pA to 8.59 ± 1.31 pA, n= 8, P < 0.01) (Fig. 1D). β-Ala effects, however, were incompletely blocked by BIC (59.1 ± 25.8%, n= 6) (Fig. 1D). Since β-Ala has also been shown to activate glycine receptors (Rajendra et al. 1997), we tested whether the BIC-insensitive, β-Ala-induced inward current was blocked by further addition of the glycine receptor blocker strychnine (STR). As shown in Fig. 1E, the sequential addition of BIC and STRY almost completely blocked the β-Ala-induced inward current in PVN-RVLM neurons (94.2 ± 2.5%, n= 3). These results suggest that in addition to blockade of GAT3 and subsequent activation of tonic GABA_A receptors, β-Ala also activates STRY-sensitive receptors in PVN-RVLM neurons. As summarized in Fig. 2, the BIC-sensitive component (i.e. GABA_A tonic current) of the β-Ala-induced current was of similar magnitude to that evoked by NA, and both were significantly larger than that evoked by SKF (P < 0.01).

Figure 2. Effects of GABA transporter blockade on I_tonic.

A, mean data summarizing changes in I_holding induced by BIC (i.e, GABA_AI_tonic), in the absence (control, CTL, n= 4) and in the presence of the non-selective GAT blocker nipecotic acid (NA 300 μm, n= 7), the GAT1 selective blocker SKF89976A (SKF, 30 μm, n= 3), and the GAT3 selective blocker β-alanine (β-Ala, 100 μm, n= 8). Note that while both NA and β-Ala induced a robust and significant increase in I_tonic, a smaller and statistically insignificant effect was evoked by the GAT1 blocker SKF. B, NA and β-Ala also induced a significant increase in RMS, an effect that was not evoked by the GAT1 blocker SKF. Data shown are means ±s.e.m. **P < 0.01 when compared to CTL), #P < 0.05 and ##P < 0.01 when compared to NA (one-way ANOVA, Tukey's post hoc test).

Altogether, these results suggest that the activity of GABA transporters, in particular the GAT3 isoform, strongly influences the magnitude of I_tonic in PVN-RVLM neurons. Due to the complexity of the β-Ala-induced effects in PVN-RVLM neurons, and the lack of an effect of the GAT1 selective SKF, most of the remaining studies in this work, unless otherwise specified, were based on the use of NA.

Given that GAT activity is known to be temperature dependent, we explored whether the effects of GAT blockade persisted at more physiological temperatures. Increasing the recording temperature from ∼23°C to ∼35°C (n= 5) did not alter the effects of GAT blockade on I_tonic or RMS. Thus, the non-selective GAT blocker NA (300 μm, n= 5) still induced a significant shift in I_holding and RMS, effects that were not statistically different from those evoked at room temperature (I_holding shift: control: 32.79 ± 5.04 pA, n= 11; high temperature: −40.81 ± 13.30 pA, n= 5, P > 0.4; RMS change: control: 3.57 ± 0.45; high temperature: 4.18 ± 0.88, P > 0.5). This is in agreement with our recent findings in supraoptic neurosecretory neurons (Park et al. 2006), as well previous reports in cortical interneurons (Semyanov et al. 2003).

Modulation of GABA_AI_phasic by neuronal and glial GABA transporters

The concentration and time course of GABA in the synaptic cleft are an important factor modulating the efficacy and fidelity of GABA_A-mediated synaptic transmission (I_phasic) (Mody et al. 1994). To determine whether I_phasic in PVN-RVLM neurons is under the tonic influence of GAT activity, we investigated the effects of GAT blockers on GABA_A-sIPSCs in PVN-RVLM neurons. Bath application of nipecotic acid (NA, 300 μm) significantly reduced sIPSC frequency by 34.8 ± 6.0% (n= 7, P < 0.05). On the other hand, neither IPSC amplitude nor IPSC decay time kinetics was affected by NA (P > 0.3, n= 7 in both cases). Results are summarized in Fig. 3.

Figure 3. Effect of GABA transporter blockade on I_phasic.

A, representative examples of GABA_A-mediated IPSCs recorded from an identified PVN-RVLM neuron, before and during bath application of the non-selective GAT blocker nipecotic acid (NA, 300 μm). Note the decreased IPSC frequency in the presence of NA. B, averaged sIPSCs (n= 150–200 events) obtained from the same neuron as in A, before (control) and in the presence of NA. C–E, cumulative plots for IPSC amplitude (C), decay time constant (D) and inter-event intervals (E), recorded from the same PVN-RVLM neuron as in A, before and during bath application of NA. The inter-event interval cumulative fraction plot (but not the amplitude nor the decay time constant plots) was significantly changed by NA (P < 0.0001, Kolmogorov–Smirnov test). F, summary data showing NA (300 μm) effects on IPSC properties in PVN-RVLM neurons (n= 8). Data shown are means ±s.e.m. *P < 0.05 when compared to its respective control.

GAT1 blockade with SKF 89976A (30 μm) did not affect IPSC frequency (control: 0.93 ± 0.10 Hz, SKF: 0.90 ± 0.32 Hz), amplitude (control: 66.8 ± 7.1 pA, SKF: 69.6 ± 10.9 pA) or weighted decay τ (control: 24.8 ± 1.5 ms, SKF: 24.9 ± 1.8 ms) (P > 0.6, n= 5 in all cases).

To determine whether the effects of NA on GABA_A-sIPSC frequency could involve activation of presynaptic GABA_B receptors following an increase of ambient GABA levels, a set of recordings were obtained in slices preincubated in the presence of the GABA_B receptor antagonist CGP52432 (10 μm). Under this condition, bath application of NA (300 μm) failed to change GABA_A IPSC frequency (control: 2.53 ± 1.20 Hz, NA: 3.22 ± 1.04 Hz, n= 6, P > 0.9).

GAT1 and GAT3 immunoreactivity in the PVN

Given the differential contribution of GAT1 and GAT3 in the regulation of I_tonic and I_phasic in PVN-RVLM neurons, we also investigated the distribution pattern of these transporters in the PVN, using immunohistochemical approaches. As illustrated in the representative examples in Fig. 4A and B, GAT3 immunoreactivity was predominantly localized in the neuropil, in astroglial-like processes surrounding PVN somata. Differently from GAT3, PVN GAT1 immunoreactivity displayed a punctate pattern, resembling axonal terminals (Fig. 4B). As previously reported, we found a similar GAT-3 and GAT-1 immunoreactive pattern in the thalamic paraventricular nucleus (Ikegaki et al. 1994; De Biasi et al. 1998) (not shown), whereas the opposite staining pattern was found in the lateral amygdaloid nucleus, as previously described (Ikegaki et al. 1994) (see Fig. 4C and D). While we did not attempt to perform a quantitative study, we found GAT3 immunoreactivity to colocalize to a large extent with the glial marker GFAP (Fig. 4E). In a subset of experiments (n= 3), GAT3 immunoreactivity was combined with retrograde labelling of PVN-RVLM neurons (Methods). Images from these experiments revealed large GAT3 immunoreactive patches surrounding identified PVN-RVLM neurons (Fig. 4F).

Figure 4. GAT1 and GAT3 immunoreactivities in the PVN.

A–B, representative confocal photomicrographs of two different sections through the PVN, displaying GAT3 (A) and GAT1 (B) immunoreactivities. Note the high and the low abundance of GAT3ir and GAT1ir, respectively. C–D, the opposite staining pattern for GAT3 (C) and GAT1 (D) staining was observed in the lateral amygdaloid nucleus. E, high power confocal photomicrograph displaying GAT3 (green) and GFAP (red) immunoreactivities surrounding two representative PVN neurons (asterisks). Yellow colour represents colocalized immunoreactivity. F, confocal photomicrograph of a representative soma of a retrogradely labelled PVN-RVLM neuron (blue), surrounded by GAT3 immunoreactive patches (green, arrows). Vertical and horizontal arrows in B point to dorsal and medial aspects of the PVN, respectively. Scale bars: A and B= 20 μm; C and D= 5 μm.

Effects of the gliotoxin l-α-aminoadipic acid on GABA_AI_tonic and I_phasic

GAT1 and GAT3 have been shown to be predominantly (but not exclusively) located in neurons and astrocytes, respectively (Borden, 1996; Gadea & Lopez-Colome, 2001; Conti et al. 2004). Our electrophysiological and immunohistochemical experiments suggest that both I_tonic and I_phasic in PVN-RVLM neurons are under the predominant influence of glial GATs. To further confirm this, we evaluated the effects of the gliotoxin-selective compound l-α-aminoadipic acid (l-αAA) (Huck et al. 1984) on the properties of I_tonic and I_phasic in PVN-RVLM neurons. As shown in Fig. 5A and B, bath application of l-αAA (2 mm) induced an inward shift in I_holding (from 20.3 ± 5.6 to 25.01 ± 5.4, n= 6, P < 0.05), an effect that was blocked by BIC (20 μm). In addition, l-αAA (2 mm) significantly reduced IPSC frequency by ∼50% (P < 0.05, n= 6), without affecting IPSC amplitude or decay time course (see Fig. 5C).

Figure 5. Effect of the gliotoxin l-α-aminoadipic acid (l-αAA) on I_tonic and I_phasic.

A, representative example of a recorded PVN-RVLM neurone showing that bath application of the gliotoxin l-αAA (2 mm) in the presence of the broad glutamate transporter blocker kynurenic acid (1 mm), caused a significant inward increase in I_holding. This effect was blocked by the GABA_A receptor antagonist BIC (20 μm). B, summary data showing the effects of l-αAA on I_holding (n= 6, *P < 0.05 when compared to baseline). C, summary data showing the effects of l-αAA on GABA_A IPSCs properties (n= 6). l-αAA significantly decreased IPSC frequency, without affecting IPSC amplitude or decay time course. Data shown are means ±s.e.m. *P < 0.05 when compared to baseline. IPSCs in panel A were truncated.

GABA transporters regulate ongoing firing activity in PVN-RVLM neurons

To determine whether GAT modulatory effects on GABAergic efficacy influenced ongoing firing activity of PVN-RVLM neurons, recordings were obtained in the current-clamp mode. Bath application of the GAT blocker NA (100 μm) decreased the firing activity of PVN neurons by 58.4 ± 5.8% (P < 0.01 n= 7). This effect was prevented when slices were preincubated in the presence of the GABA_A receptor blocker BIC (n= 3). Results are summarized in Fig. 6.

Figure 6. Blockade of GABA transporters decreased the firing activity of PVN neurons.

A, representative trace from a PVN-RVLM neuron showing that bath application of the non-selective GAT blocker nipecotic acid (NA, 100 μm) transiently diminished the spontaneous ongoing firing activity of the recorded neuron. The lower panel shows a running histogram of the number of action potential as a function of time (binning size = 10 s) for the same neuron. B, representative trace from a different neuron showing that the NA effect was prevented in the presence of the GABA_A receptor antagonist BIC (20 μm). The lower panel shows a running histogram of the number of action potential as a function of time (binning size = 10 s) for the same neuron.

Conversely, we found that the increased firing activity induced by the GABA_A receptor blocker BIC was enhanced when GAT activity was pharmacologically blocked. Thus, BIC increased PVN-RVLM firing discharge by ∼160% (from 0.62 ± 0.19 Hz to 0.91 ± 0.27 Hz, n= 4, P < 0.05) in control conditions, and ∼240% (from 0.65 ± 0.24 Hz to 1.49 ± 0.54 Hz, n= 6, P < 0.05) in presence of NA (100 μm) (P < 0.05 control vs. NA). Altogether, these studies support a physiological role for GATs in controlling the magnitude of GABA_AI_tonic, as well as GABAergic influence on PVN-RVLM firing activity.

GABA transporters regulate sympathetic outflow from the PVN

Numerous studies support a key role for GABA within the PVN in the control of tonic and reflex sympathetic outflow (Zhang & Patel, 1998; Chen et al. 2003; Li & Pan, 2005). To determine whether GAT regulation of GABA ambient levels, and subsequent effects on I_tonic and I_phasic affected sympathoexcitatory outflow from the PVN, we evaluated the actions of acute PVN microinjections of β-Ala and NA (both at 100 pmol) on renal sympathetic nerve activity (RSNA), blood pressure (BP) and heart rate (HR) (Methods). As summarized in Fig. 7, microinjections of β-Ala and NA (n= 8 in each case) significantly decreased RSNA when compared to microinjections of CSF (P < 0.01 in both cases). On the other hand, no significant changes in BP were evoked by either of the tested compounds (P > 0.5). Finally, whereas both treatments decreased HR, this effect reached statistical significance only in the presence of NA (P < 0.05).

Figure 7. Blockade of GABA transporters within the PVN decreases renal sympathetic nerve activity.

A, representative segments of original recordings showing changes in heart rate (HR, beats min⁻¹; bpm), arterial pressure (BP), and renal sympathetic nerve activity (RSNA) following administration of nipecotic acid (NA) into the PVN. B, summary data showing the effects of PVN microinjections of CSF (50 nl, n= 5), the non-selective GAT blocker nipecotic acid (NA, 100 pmol, n= 8), or the GAT3-selective blocker β-alanine (β-Ala, 100 pmol, n= 8), on renal sympathetic nerve activity (RSNA), blood pressure (BP) and heart rate (HR). Data shown are means ±s.e.m. *P < 0.05 and **P < 0.01 when compared to control CSF.

Discussion

The main findings of this work could be summarized as followed: (1) PVN GAT activity regulates the magnitude of GABA_A tonic inhibition (I_tonic), and to a lesser extent the degree of GABA_A phasic activity (I_phasic, IPSCs); (2) by restraining the magnitude of basal GABA_AI_tonic, GAT activity contributes to ongoing firing discharge of PVN-RVLM projecting neurons, as well as PVN drive of renal sympathetic nerve activity (RSNA); (3) our pharmacological and immunohistochemical studies support that these effects are mediated by GATs of a glial (GAT3), rather than a neuronal (GAT1) location. Altogether, these data indicate that by influencing the availability of extracellular ambient GABA levels, PVN astroglial cells modulate neuronal activity and sympathetic outflow from the PVN. Thus, as shown in other CNS regions in relation to other brain functions (Araque et al. 2001; Haydon & Carmignoto, 2006), our results support neuronal–glial interactions within the PVN as a novel and important mechanism contributing to the maintenance of basal autonomic homeostasis.

GABA_A tonic (I_tonic) and phasic (I_phasic) inhibitory modalities in the PVN

It has become increasingly clear in recent years that in addition to mediating a transient, temporally and spatially restricted type inhibition (i.e. IPSCs or I_phasic), GABA_A receptors also mediate a much slower form of inhibition, which results from the persistent activation of GABA_A receptors (also termed ‘tonic’ inhibition, I_tonic) (Semyanov et al. 2004; Farrant & Nusser, 2005). There is general consensus that GABA_A receptors mediating phasic inhibition are clustered at postsynaptic sites, have a relatively low affinity for GABA, and are activated by brief exposure to high concentration of neurotransmitter (Mody et al. 1994). On the other hand, GABA_A receptors mediating tonic inhibition appear to be located at peri- and/or extrasynaptic sites, display high affinity for GABA, and are activated by low ambient concentration of transmitter in the extracellular space (Yeung et al. 2003). GABA_AI_tonic inhibition has been recently described in major CNS areas, including the cerebellum (Brickley et al. 1996), hippocampus (Nusser & Mody, 2002) and cortex (Salin & Prince, 1996; Semyanov et al. 2003), where it has been shown to play a critical role in modulating network excitability. In a recent study, we demonstrated that GABA_A phasic and tonic inhibitory modalities are both present in PVN-RVLM neurons, and provided a detailed pharmacological and functional characterization of GABA_AI_tonic in this neuronal population (Park et al. 2007). In this sense, we found that I_tonic accounted for ∼70% of the total GABA_A-receptor mediated current in PVN-RVLM neurons, playing thus a key role in regulating their ongoing firing discharge (Park et al. 2007). Given the importance of GABA_AI_tonic in these neurons, we aimed in this study to elucidate mechanisms influencing the availability and efficacy of this inhibitory modality, as well as its overall influence on sympathoexcitatory outflow from the PVN.

Glial GABA transporters modulate I_tonic in PVN-RVLM neurons

GABA transporters (GATs) are important molecules responsible for removing GABA from the extracellular space. Whereas four GABA transporters (GATs) have so far been cloned and characterized, the GAT1 and GAT3 isoforms are the most likely ones expressed in the brain (Guastella et al. 1990; Liu et al. 1992; Lopez-Corcuera et al. 1992; Liu et al. 1993; Dalby, 2003). Given that GATs efficiently control local ambient extracellular GABA concentration, they stand as major candidates influencing the magnitude of GABA_AI_tonic. This has been previously demonstrated in other CNS regions, including the hippocampus, cerebellum and supraoptic nucleus (Wall & Usowicz, 1997; Nusser & Mody, 2002; Rossi et al. 2003; Semyanov et al. 2003; Park et al. 2006). In agreement with these previous reports, we found that blockade of GAT activity increased the strength of I_tonic in PVN-RVLM neurons, an effect likely to be mediated by the resulting increase in extracellular ambient GABA levels. The absolute magnitude of the GAT effect on GABA_AI_tonic should be taken cautiously, given that numerous factors associated with our experimental approach likely to contribute to its over- and/or underestimation. For example, the use of a non-physiological, high Cl⁻ intracellular solution, conventionally used to study GABA_AI_tonic (Brickley et al. 1996, 2001; Rossi & Hamann, 1998; Nusser et al. 2001; Hamann et al. 2002; Semyanov et al. 2003; Wei et al. 2004) is likely to overestimate the magnitude of GABA_AI_tonic and its modulation by GATs. Conversely, it is likely that the concentration of extracellular GABA in our superfused slice preparation is lower than that observed in the whole brain (Glykys & Mody, 2006), a factor that may partially mask GABA_AI_tonic and its modulation by GATs.

Using relatively specific pharmacological agents, we found that these actions were most likely mediated by the GAT3, but not the GAT1, transporter isoform. This is in part supported by the fact that both the non-selective GAT blocker nipecotic acid and the relatively GAT3 selective blocker β-alanine induced an increase in I_tonic of similar magnitude. On the other hand, the relatively specific GAT1 blocker SKF 899976-A failed to induce a significant change in GABA_AI_tonic. These results are in agreement with our recent finding in hypothalamic magnocellular neurosecretory neurons (Park et al. 2006) as well as those in neocortical neurons (Keros & Hablitz, 2005), but differed from those previously reported in the hippocampus and cerebellum, where I_tonic was found to be influenced by GAT1 activity (Nusser & Mody, 2002; Rossi et al. 2003; Semyanov et al. 2003).

The predominant influence of GAT3 over GAT1 in controlling GABA_AI_tonic in PVN-RVLM neurons is also supported by our immunohistochemical studies, showing a predominance of GAT3 over GAT1 immunoreactivity in the PVN. We found GAT3 immunoreactivity to display a characteristic glial-like neuropile staining (which colocalized to a large extent with the glial marker GFAP), surrounding the somata of PVN neurons, including retrogradely labelled PVN-RVLM ones. On the other hand, the weaker, less dense GAT1 immunoreactivity displayed a more discrete, punctate staining. These findings are in general agreement with previous studies supporting the notion that these two GAT isoforms are differentially distributed within specific cellular compartments in the CNS, with GAT1 and GAT3 predominantly (but not exclusively) located in neurons and astrocytes, respectively (Borden, 1996; Gadea & Lopez-Colome, 2001; Conti et al. 2004). Thus, based on the electrophysiological and immunohistochemical evidence obtained in our studies, we propose that GATs of a likely glial origin are responsible for regulating the degree of GABA_AI_tonic in PVN-RVLM neurons. In further support of this notion, we found the selective gliotoxin l-α-aminoadipic acid (l-αAA), commonly used to disable astrocytic function (McBean, 1994; Pannicke et al. 1994; Baudoux & Parker, 2008), to evoke a similar increase in GABA_AI_tonic to that observed with the GAT blockers. It is important to recognize however, that l-αAA is a glutamate analogue known to interfere with electrogenic glutamate transport (Alvarez-Maubecin et al. 2000). Thus, while all our recordings were performed in the presence of ionotropic glutamate receptor blockers, we cannot rule out the possibility that l-αAA-mediated effects could be, at least in part, mediated by activation of metabotropic glutamate receptors.

Modulation of phasic synaptic activity by glial GATs in PVN-RVLM neurons

Under particular physiological conditions, in addition to modulating GABA_A tonic currents, GATs have been shown to shape the properties of phasic synaptic activity. For example, GAT blockade was reported to prolong the decay time course and/or to diminish the amplitude of evoked IPSCs, while generally having no effect on spontaneous IPSCs (Thompson & Gahwiler, 1992; Engel et al. 1998; Rossi & Hamann, 1998; Isaacson & Vitten, 2003; Overstreet & Westbrook, 2003; Keros & Hablitz, 2005). These studies suggest that GAT influence on synaptic activity occurs under conditions of high activity-dependent release of GABA and/or following multiple fibre stimulation. Proposed underlying mechanisms include buildup of background GABA levels near the synaptic cleft during these conditions, followed by desensitization of postsynaptic GABA_A receptors (Keros & Hablitz, 2005), or activation of putative presynaptic receptors (Jensen et al. 2003; Kirmse & Kirischuk, 2006).

In our study, we found no effects of GAT blockade on sIPSC amplitude or decay time course. These results are in agreement with the above referenced work, and suggest (a) that following GAT blockade, GABA levels around GABAergic inhibitory synapse in PVN-RVLM neurons do not increase enough to desensitize postsynaptic GABA_A receptors, and (b) that these GATs (likely glial GAT3) are located spatially distant from the GABAergic synapse.

On the other hand, we found GAT blockade to result in a diminished frequency of sIPSCs in PVN-RVLM neurons. Such an effect was also observed following GAT1 blockade in developing visual cortex neurons (Kirmse & Kirischuk, 2006), as well as in hippocampal neurons in a GAT1 deficient mouse (Jensen et al. 2003). A decreased sIPSC frequency following GAT blockade in PVN-RVLM neurons could be due to activation of presynaptic GABA_B receptors, known to be present in the PVN (Richards et al. 2005; Li et al. 2008). This is in fact supported by our studies showing that the effect of NA on sIPSC frequency was prevented by CGP52432, a selective GABA_B receptor blocker.

Our results thus indicate that PVN GATs restrain GABA_A tonic inhibition, facilitating at the same time the activity of incoming inhibitory synaptic inputs in PVN-RVLM neurons. Whereas the I_phasic modality is known to mediate specific, point-to-point inhibition, playing a critical role in fine-tuning of spike timing (Bevan et al. 2002; Jonas et al. 2004; Kononenko & Dudek, 2004), I_tonic on the other hand influences overall neuronal and network excitability level. Based on the well-established disparate roles of these two inhibitory modalities, along with their differential modulation by GATs reported in this study, it is reasonable to speculate that varying the degree of GAT activity in the PVN could be an efficient mechanism by which the contrast between these two inhibitory modalities is dynamically altered in PVN-RVLM neurons, facilitating either global or point-to point inhibition. Elucidating thus whether, how and under what conditions PVN GAT activity is modulated will be critical to obtaining a more comprehensive understanding of the different mechanisms and modalities by which GABA influences PVN neuronal activity.

Glial GATs modulate PVN-RVLM neuronal activity and sympathoexitatory outflow from the PVN

In a recent study, we demonstrated that blockade of GABA_AI_tonic induced membrane depolarization, increased firing discharge, and enhanced the gain of the input-output function in PVN-RVLM neurons (Park et al. 2007), supporting GABA_AI_tonic as a major inhibitory modality restraining firing output from PVN-RVLM. Results from our present studies showing that GAT blockade (a) diminished PVN-RVLM firing discharge in a BIC-sensitive manner and (b) resulted in a small though significant inhibition of RSNA further underscore the functional relevance of GABA_AI_tonic in the regulation of PVN-RVLM neuronal activity as well as sympathetic outflow from the PVN.

The inhibitory effect of GAT blockade on PVN-RVLM neuronal activity was likely to have been the result of increased Cl⁻ influx and membrane hyperpolarization, following activation of GABA_AI_tonic in these neurons. Alternatively, shunting of excitatory synaptic inputs could also contributed to GAT-mediated inhibition. Our previous studies in magnocellular neurosecretory neurons, however, showed that the increased firing activity following GABA_AI_tonic blockade was still observed in the presence of ionotropic glutamate receptor antagonists (Park et al. 2006), arguing against this latter mechanism. Whether this is true also in PVN-RVLM neurons, however, remains to be determined.

While several previous studies demonstrated a role of GABA_AI_tonic or GATs in the regulation of neuronal firing activity (Cope et al. 2005; Park et al. 2006), the implication of this mechanism at the whole system level has not been adequately examined (see for example (Dalby, 2003). In this sense, and to the best of our knowledge, our present study is the first one to demonstrate an important role for GAT modulation of GABA_AI_tonic in the CNS control of sympathetic nerve activity.

Both PVN-RVLM neuronal activity (Li et al. 2003; Li & Pan, 2005; Park et al. 2007) and PVN control of RSNA are known to be tonically inhibited by a local GABAergic tone (Martin & Haywood, 1998; Kenney et al. 2001; Zhang et al. 2002; Chen et al. 2003; LaGrange et al. 2003; Yang & Coote, 2003; Li et al. 2006), such that PVN disinhibition typically results in an increased RSNA (Martin & Haywood, 1998; Kenney et al. 2001; Zhang et al. 2002; Chen et al. 2003; LaGrange et al. 2003; Yang & Coote, 2003; Li et al. 2006). Thus, it is likely that the inhibition of RSNA following PVN GAT blockade reported here was the consequence of the diminished PVN-RVLM neuronal activity that we also observed.

Changes in RSNA induced by PVN nipecotic acid-mediated GAT blockade occurred along with a modest decrease in HR, and no change in arterial blood pressure. It is worth mentioning that in a previous study, we found that PVN microinjections of the GABA_A receptor agonist muscimol (which is likely to have resulted in the activation of both synaptic and extrasynaptic GABA_A receptors) evoked a larger inhibition of RSNA (∼40%), accompanied by a decrease in arterial blood pressure and heart rate (Zhang & Patel, 1998). One possible explanation for the difference observed between GAT blockade and muscimol is that GABA_AI_tonic has a relatively lower contribution to the regulation of basal RSNA when compared to GABA_AI_phasic. Previous observations in our in vitro slice preparation showing that the GABA_AI_tonic modality accounted for ∼70% of the total GABA_A-mediated currents in PVN-RVLM neurons (Park et al. 2007) would argue against this possibility. It is worth noting, however, that most peripheral inputs that actively drive local GABAergic activity within the PVN are absent in vitro. Thus, whether a similar predominance of GABA_AI_tonic over GABA_AI_phasic is also present in the intact animal is at present unknown. Alternatively, it is possible that the dose of nipecotic acid used on our in vivo studies was insufficient to completely block all PVN GAT activity, resulting in a partial activation of GABA_A receptors mediating I_tonic, and consequently, a more modest increase in RSNA when compared to the effects previously reported with muscimol. Finally, it is important to consider that these studies were based on a unilateral injection within the PVN, which would be expected to have limited responses.

Accumulating evidence supports a blunted GABAergic inhibition within the PVN as a critical mechanism contributing to sympathoexcitation in cardiovascular-related disorder, including hypertension, diabetes and heart failure (Reynolds et al. 1996; Haywood et al. 2001; Allen, 2002; Zhang et al. 2002; Li & Pan, 2007). Given that changes in the expression level or activity of amino acid neurotransmitter transporters, including GAT, have been demonstrated to occur in various CNS disorders, including ischaemia, stroke, epilepsy and amotrophic lateral sclerosis (Allen et al. 2004), it would be important to determine in the future whether a change in PVN GAT expression and/or function contributes as an underlying mechanism to GABA-dependent sympathoexcitation in cardioavscular diseases.

In summary, our results support the view that GATs, of a likely glial location, contribute to setting a basal tone of PVN-RVLM firing activity and PVN-driven RSNA. Neuronal–glial communication is emerging as a novel and major player in CNS information processing and function (Araque et al. 1999, 2001), including functional hyperaemia, and neurosecretory function, among others. Our present study also supports neuronal–glial interaction as an important factor influencing CNS control of autonomic activity and homeostasis.

Glossary

Abbreviations

BP

blood pressure

GAT

GABA transporter

GFAP

glial fibrillary acid protein

HR

heart rate

NA

nipecotic acid

PSC

postsynaptic current

PVN

paraventricular nucleus

RMS

root mean square

RSNA

renal sympathetic nerve activity

RVLM

rostral ventrolateral medulla

Author contributions

J.B.P.: conception, design, analysis and interpretation of electrophysiological data, and drafting and final approval of the manuscript. J.Y.J. performed additional experiments, analysis and interpretation, as requested by reviewers for the revised manuscript. H.Z.: analysis and interpretation of electrophysiological data, and drafting and final approval of the manuscript. K.P.P.: conception, design, analysis and interpretation of whole animal data, and drafting and final approval of the manuscript. J.E.S.: conception, design, analysis and interpretation of electrophysiological and whole animal data, and drafting and final approval of the manuscript. In vitro experiments were performed partly at the University of Cincinnati and partly at the Medical College of Georgia (Dr Stern's laboratory was relocated from UC to MCG during the course of this work). In vivo experiments were performed at the University of Nebraska Medical Center.