Abstract
Activation of GABAA receptors (GABAARs) produces two forms of inhibition: phasic inhibition generated by the rapid, transient activation of synaptic GABAARs by presynaptic GABA release, and tonic inhibition generated by the persistent activation of perisynaptic or extrasynaptic GABAARs, which can detect extracellular GABA. Such tonic GABAAR-mediated currents are particularly evident in dentate granule cells in which they play a major role in regulating cell excitability. Here we show that in rat dentate granule cells in ex vivo hippocampal slices, tonic currents are predominantly generated by GABA-independent GABAA receptor openings. This tonic GABAAR conductance is resistant to the competitive GABAAR antagonist SR95531 (gabazine), which at high concentrations acts as a partial agonist, but can be blocked by an open channel blocker, picrotoxin. When slices are perfused with 200 nm GABA, a concentration that is comparable to CSF concentrations but is twice that measured by us in the hippocampus in vivo using zero-net-flux microdialysis, negligible GABA is detected by dentate granule cells. Spontaneously opening GABAARs, therefore, maintain dentate granule cell tonic currents in the face of low extracellular GABA concentrations.
Introduction
In addition to fast synaptic GABAA receptor (GABAAR)-mediated signaling, there is a slower form of signaling resulting from the tonic activation of GABAARs (Semyanov et al., 2004; Farrant and Nusser, 2005; Glykys and Mody, 2007a). Tonically active GABAARs can have profound effects on neuronal excitability, synaptic plasticity, network oscillations, and neurogenesis (Ge et al., 2006; Pavlov et al., 2009; Holter et al., 2010; Mann and Mody, 2010; Martin et al., 2010; Duveau et al., 2011) and have been implicated in neuronal development, information processing, cognition, and memory (Semyanov et al., 2004; Farrant and Nusser, 2005; Brickley and Mody, 2012). The conventional view is that tonic currents are mediated by high-affinity extrasynaptic GABAARs that can detect low concentrations of ambient GABA, and therefore the magnitude of tonic currents is regulated by the expression of these receptors and the availability of extracellular GABA (Semyanov et al., 2004; Farrant and Nusser, 2005; Glykys and Mody, 2007a).
The sources and concentration of extracellular GABA ([GABA]e) are, however, still debated. Vesicular release, “leak” through bestrophin channels, and reversal of GABA transporters have all been suggested to contribute to the extracellular GABA pool (Attwell et al., 1993; Gaspary et al., 1998; Lee et al., 2010). In vivo estimates indicate that [GABA]e is in the micromolar or submicromolar range (Lerma et al., 1986; Kuntz et al., 2004; Nyitrai et al., 2006). In vitro studies have suggested even lower concentrations of GABA, in the nanomolar range; indeed, active uptake maintains GABA at sufficiently low concentrations to prevent tonic GABAB receptor activation (Isaacson et al., 1993). Thus, tonic currents in hippocampal neurons have often been measured in conditions that artificially raise [GABA]e by inhibiting GABA uptake or metabolism, or by adding exogenous GABA to the perfusate (Overstreet and Westbrook, 2001; Nusser and Mody, 2002; Stell and Mody, 2002; Holter et al., 2010). When GABA is not increased by these means, various magnitudes of tonic currents have been obtained using different GABAAR antagonists (Bai et al., 2001; Mtchedlishvili and Kapur, 2006; Zhan and Nadler, 2009; Mtchedlishvili et al., 2010). Here we report a novel form of tonic inhibition mediated by GABA-independent openings of GABAARs, which can explain these discrepancies.
We focus on dentate gyrus granule cells (DGCs), the excitability of which critically depends upon the presence of tonic GABAAR-mediated conductances (Overstreet and Westbrook, 2001; Nusser and Mody, 2002; Stell and Mody, 2002; Coulter and Carlson, 2007; Holter et al., 2010). We demonstrate that under baseline conditions, or when the perfusate contains the same concentration of GABA that is found in CSF, the major contributors to the tonic current in DGCs are spontaneously opening GABAARs. This tonic GABAAR conductance is resistant to the competitive GABAAR antagonist gabazine (SR95531), but can be blocked by an open channel blocker, picrotoxin. On increasing [GABA]e, a SR95531-sensitive component of tonic current emerges. Together, these results indicate that the GABA-independent component of tonic current mediated by spontaneously opening GABAARs maintains tonic inhibition in the presence of low extracellular GABA concentrations measured in vivo.
Materials and Methods
Hippocampal slice preparation
Transverse hippocampal slices (350 μm thick) were used for in vitro electrophysiological recordings. Slices were prepared from 3- to 4-week-old male Sprague Dawley rats and δ−/− knockouts or δ+/+ littermate control mice on a C57B6 background (Mihalek et al., 1999; Herd et al., 2008). Animals were killed by an overdose of isoflurane according to the United Kingdom Animals (Scientific Procedures) Act of 1986. After decapitation, brains were rapidly removed and dissected, and hippocampi were sliced with a Leica VT1200S vibratome in ice-cold sucrose-based solution containing the following (in mm): 70 sucrose, 80 NaCl, 2.5 KCl, 7 MgCl2, 0.5 CaCl2, 25 NaHCO3, 1.25 NaH2PO4, 22 glucose, equilibrated with 95% O2 plus 5% CO2, pH 7.4, 315–330 mOsm. Slices were maintained in continuously oxygenated sucrose-free storage solution at 33°C for 15 min, equilibrated to a room temperature for 15 min and then placed to recover in continuously oxygenated humid interface holding chamber at room temperature for at least 1 h before recording. In experiments with concanamycin, slices were prepared in exactly the same way, but before placing them in a holding chamber for the recovery they were incubated for 2 h in a submerged chamber with continuously oxygenated storage solution and 0.5 μm concanamycin. After that, slices were placed in an interface holding chamber and allowed to rest for about half an hour before recording. After recovering slices were transferred into recording chamber. The perfusion and storage medium contained the following (in mm): 119 NaCl, 2.5 KCl, 1.3 MgSO4, 2.5 CaCl2, 26.2 NaHCO3, 1 NaH2PO4, 22 glucose and was gassed with 95% O2 and 5% CO2, pH 7.4; 290–298 mOsm.
In vitro electrophysiology
Whole-cell recordings.
Visualized patch-clamp recordings from mature dentate granule cells (input resistance, Rin = 310 ± 50 MΩ; membrane capacitance, Cm = 48 ± 6 pF) were performed using an infrared differential interference contrast imaging system. Tonic GABAAR-mediated currents were measured in voltage-clamp mode (at holding potential Vhold = −70 mV) in the presence of ionotropic glutamate receptor blockers dl-APV (50 μm) and NBQX (20 μm), metabotropic glutamate receptors blocker (S)-α-methyl-4-carboxyphenylglycine (MCPG; 250 μm), and GABAB receptor blocker (2S)-3-[[(1S)-1-(3,4-dichlorophenyl)ethyl]amino-2-hydroxypropyl](phenylmethyl)phosphinic acid (CGP55845; 1 μm). The intracellular pipette solution contained the following (mm): 120.5 CsCl, 10 KOH-HEPES, 2 EGTA, 8 NaCl, 5 QX-314 Br− salt, 2 Mg-ATP, 0.3 Na-GTP, 10 Na-phosphocreatine, pH adjusted to 7.2 and osmolarity adjusted to 295 mOsm. The tonic GABAAR-mediated current was measured as an outward shift in holding current following application of picrotoxin (100 μm). Changes in root mean square (RMS) noise have also been proposed to reflect changes in tonic GABAAR-mediated conductances and have been used because they are unaffected by current drift. Although RMS noise decreased in experiments in which tonic currents were blocked, this measurement is confounded by the presence of synaptic currents. Moreover, RMS noise is nonlinearly related to current and can paradoxically decrease when tonic currents increase (Traynelis and Jaramillo, 1998; Glykys and Mody, 2007a). We, therefore, only used RMS noise as a measure in experiments in which we were trying to block the tonic current. Three other GABAAR antagonists, bicuculline methiodide (BMI) (10 μm), SR95531 (0.5, 25, and 125 μm), and pentylenetetrazol (1.5 mm) were used in experiments. Gabazine (25 μm) and bicuculline (10 μm) were used at concentrations that would be predicted to give >90% occupancy of GABAARs in dentate granule cells (Jones et al., 2001). The affinity of gabazine for GABAARs in dentate granule cells is an order of magnitude greater than that of bicuculline (Jones et al., 2001), so that equimolar concentrations of gabazine displace bicuculline from its binding site. Glycine receptors were blocked with 1 μm strychnine, GABAC receptors—with 50 μm (1,2,5,6-tetrahydropyridin-4-yl)methylphosphinic acid (TPMPA). Recordings were performed at 32–34°C. The whole-cell pipette resistance was 4–5.5 MΩ. Series resistance was monitored throughout experiments using a −5 mV step command. Cells showing a >20% change in a series resistance, a resistance >20 MΩ, or unstable holding current were rejected.
Acquisition and analysis.
Recordings were obtained using a MultiClamp 700B or Axopatch 200B amplifier (Molecular Devices), filtered at 4 kHz, digitized at 10 kHz, and stored on a PC. WinEDR (Strathclyde Electrophysiology Software) was used for data acquisition and pClamp (Molecular Devices), and OriginPro 8.1 (OriginLab) for off-line analysis.
For analysis of tonic currents, mean values of holding current during 200 ms epochs free of synaptic currents were measured every 30 s. The amplitude of the tonic current was calculated as the difference between the holding current (ΔIhold) measured 6 min before and 10 min after the application of an antagonist. The speed of perfusion in our system of 3.5 ml/min allowed full equilibration of the solution in the recording chamber at this time point. The values of RMS noise were calculated for 200 ms epochs free of synaptic events. The change of RMS noise (Δrms) was calculated as the difference between the values before and after the application of an antagonist.
Outside-out and nucleated patch recordings.
GABAAR-mediated currents were recorded in the presence of 0.1 μm CGP55845, 100 μm d-APV, 10 μm NBQX, 200 μm S-MCPG, and 1 μm strychnine. The intracellular pipette solution contained the following (in mm): 120.5 CsCl, 10 KOH-HEPES, 10 BAPTA, 8 NaCl, 5 QX-314 Br− salt, 2 Mg-ATP, 0.3 Na-GTP, pH 7.2, osmolarity 295 mOsm. Patches were pulled from dentate granule cells, and recordings were performed in voltage-clamp mode at 33–34°C (Vhold = −70 mV) using MultiClamp 700B amplifier. Signals were digitized at 10 kHz. The patch pipette resistance was 5–7 MΩ.
For the rapid solution exchange we used a θ-glass application pipette with ∼200 μm tip diameter attached to the micromanipulator. The position of the pipette was controlled by piezoelectric element (the speed of switch was 50–100 μs). One pipette channel was filled with the bath aCSF solution or bath solution plus 10 μm GABA; another channel had 10 μm GABA, 25 or 125 μm SR95531, 10 μm bicuculline, or 20 μm picrotoxin. Pressure was regulated by a PDES-02DX pneumatic micro ejector (npi) using compressed nitrogen separately in each of two channels. Solutions with SR95531, bicuculline, and picrotoxin were exchanged in a pipette channel (7–12 s) during the exposure of nucleated patch to the bath solution channel (Bennett and Kearns, 2000; Sylantyev et al., 2008).
Analysis of the single-channel recordings
All analyses were performed on stretches of data that were longer than 3 min.
The opening frequency of GABAAR-mediated channels was calculated as N/Δt, where N is the number of openings and Δt is the time of recording. N was counted using a detection threshold of 1.5 pA more negative than mean baseline and a minimum opening time of 0.2 ms. To calculate single-channel conductance and average open time, we first built all-points amplitude histograms and fitted them with a two-Gaussian function: where m1 and m2 are the mode values of Gaussians, σ1 and σ2 are the SDs of corresponding modes, n is the value of electrical current, and p1 and p2 are the fitting constants. The channel conductance was calculated as G = (m2 − m1)/Vm.
The Newton–Raphson iteration method was used to obtain the value of nmin, which is the value of n at the minimum point of the function F in the interval m1 < n < m2. This was taken as the point of channel closure. Since the signal was digitized at 10 kHz (i.e., at intervals of 0.1 ms), the average open time (in milliseconds) was calculated as follows (the factor of 10 in the denominator is to account for the 0.1 ms interval): where nmax = 2m2 − nmin.
All calculations were performed using the Wolfram Mathematica 6.2 software; the general algorithm of histograms construction and interpretation was adapted from Bennett and Kearns (2000). We compared results obtained using this method against those obtained by averaging all the threshold-detected openings for the data shown in Figure 3. There was no difference between these two methods (Pearson's correlation coefficient of 0.992 for open time, and 0.979 for conductance).
Drugs and reagents
GABA receptor antagonists and concanamycin were purchased from Tocris Bioscience. Other reagents were from either Ascent Scientific or Sigma-Aldrich.
In vivo microdialysis
Animals.
Male Sprague Dawley rats (Harlan) were housed three per cage under standard housing conditions (lights on between 5:00 A.M. and 7:00 P.M.; 21–22°C; 50–60% relative humidity) with free access to food pellets and drinking water. Animals were handled daily (∼5 min/rat) starting 1 week before surgery and continuing until the day of the insertion of the microdialysis probe. At the time of surgery, rats weighed ∼240 g. Surgical procedures were performed under isoflurane (Merial Animal Health) anesthesia. Carprofen (Rimadyl, 4 mg/kg, s.c.; Pfizer) was administered for postoperative pain relief. All procedures were conducted in accordance with the United Kingdom's Animals (Scientific Procedures) Act 1986, and all efforts were made to minimize animal numbers and suffering.
Surgical and microdialysis procedures.
Nine days before the start of the experiment, rats were surgically prepared for microdialysis of the hippocampus by stereotaxic implantation of a guide cannula (MAB 6.14.IC, Microbiotech/se AB) essentially as described in detail before (Droste et al., 2008). After surgery, rats were housed individually in Plexiglas cages (length × width × height = 27 × 27 × 35 cm) under similar housing conditions as described above.
Seven days after surgery, a microdialysis probe (polyethersulfone membrane, length 4 mm, 15 kDa cutoff, and outer diameter 0.6 mm; MAB 6.14.4, Microbiotech/se AB) was inserted via the guide cannula into the CA3–dentate gyrus region of the hippocampus (Linthorst et al., 1994) under short-lasting isoflurane anesthesia (see Fig. 3C, schematics). Rats were connected to a liquid swivel and a counterbalance arm (Microbiotech/se AB), allowing free movement in all directions. Fluorethylene polymer tubing with a dead volume of 1.2 μl per 100 mm length (Microbiotech) was used for all connections. Dead volumes were accounted for during the experiment. Microdialysis probes were perfused with sterile, pyrogen-free Ringer solution (Delta Pharma) at a flow rate of 2 μl/min using a micro-infusion pump (KDS220, KD Scientific).
The zero-net-flux experiment was started at 9:00 A.M. on the second day after the insertion of the microdialysis probe. Ten minute samples were collected throughout the complete experiment. After 1 h of baseline sampling, microdialysis probes were subsequently perfused with increasing concentrations of GABA (10, 20, 40, 80, 160 and 240 nm; Sigma-Aldrich) dissolved in Ringer solution (reverse microdialysis). Each GABA perfusion lasted 40 min during which four 10 min samples were collected. Microdialysis samples were collected using an automated, refrigerated fraction collector (CMA470, CMA Microdialysis AB) and were stored at −80°C for later HPLC analysis.
Histology.
At the end of the experiment, rats were killed using an overdose of pentobarbital (Euthatal, 200 mg/kg body weight, i.p.; Merial Animal Health), and the brains were removed and stored in a 4% buffered paraformaldehyde solution. Histological examination was performed as described previously (Linthorst et al., 1994; Droste et al., 2008). Only data from rats with correctly placed microdialysis probes were included in the analyses.
Measurement of GABA.
HPLC with electrochemical detection was used to measure GABA in the microdialysates essentially as described previously with small modifications (de Groote and Linthorst, 2007). Briefly, GABA was separated on a TARGA C18 10 cm × 1 mm column (particle size 3 μm; Higgins Analytical) using filtered and degassed mobile phase (18% methanol, 0.1 m NaH2PO4, 0.2 mm EDTA, pH 4.72) pumped at 50 μl/min using an Alexys LC-100 pump (Antec Leyden BV). Standards and samples were injected onto the column using a thermostatically controlled (8°C) Alexys AS-100 autosampler (Antec Leyden BV). Before injection, 13 μl of standard or sample was derivatized with 2 μl of o-phthaldialdehyde (OPA)/sulfite solution (1.6 mm OPA, 0.5% methanol, 1.88 mm Na2SO3, 11.25 mm Na2B4O7) for 4 min to make GABA electrochemically active. Next, 10 μl of the derivatized mixture was injected onto the column and GABA was detected using a VT-03 electrochemical flow cell (Antec Leyden BV) set at +850 mV against an Ag/AgCl reference electrode. Both column and detector were housed in a Faraday-shielded oven (DECADE II, Antec Leyden BV) thermostatically controlled at 38°C. Chromatograms were recorded and analyzed using Alexys chromatography software (Antec Leyden BV). The detection limit for GABA at a signal-to-noise ratio of 3:1 was 11–15 fmol per injection on column.
Calculations.
Perfusion of each concentration of GABA (Cin; nm) was performed for 40 min. However, to ensure that dialysis across the membrane had reached a steady state, the first 10 min sample was discarded and only the three subsequent samples were used for the calculations described below.
After measurement of the concentration of GABA in the collected samples (Cout; in nm), the difference between Cout and Cin (i.e., the net loss or gain of GABA in the dialysate) was calculated (Cout − Cin; in nm) for each concentration of GABA perfused. The baseline condition represented Cin = 0. Next, for each individual animal, Cout − Cin was plotted against Cin and, after regression analysis (GraphPad 5.0), the concentration of zero-net-flux was determined as the concentration of Cin at which Cout − Cin = 0. Next, the mean zero-net-flux concentration (±SEM; n = 5) was calculated. For graphical purposes, the mean ± SEM values for Cout − Cin were calculated for each concentration of GABA perfused and were plotted against Cin (see Fig. 3D).
Statistics
Statistical comparisons were made using paired and unpaired (as indicated in the text/figure legends) Student's t test, and Wilcoxon signed ranks test (Fig. 1C, experiments). Differences were considered significant at p < 0.05. Data are presented in the text and figures as the mean ± SEM.
Results
Tonic activation of GABAARs in DGCs does not require synaptically released GABA
Since the accumulation of GABA in the extracellular space may result from synaptic release, we asked whether depleting vesicular GABA by incubating hippocampal slices in the vesicular H+-ATPase inhibitor concanamycin (0.5 μm) would affect tonic currents. We confirmed that concanamycin abolished synaptic currents (Rossi et al., 2003); however, it had no significant effect on the magnitude of the tonic current revealed by application of the GABAAR antagonist picrotoxin. Application of picrotoxin produced a 11.9 ± 1.5 pA outward shift in the Ihold in control experiments, and a 13.3 ± 3.0 pA shift when slices were preincubated in concanamycin (Fig. 1A). These results indicate that in acute hippocampal slices tonic currents in DGCs are not dependent on the concentration of GABA in vesicles or on GABA release into the synapse under baseline conditions. Glycine receptors can contribute to inhibition in the hippocampus and may also be blocked by picrotoxin in a use-dependent manner (Danglot et al., 2004; Yang et al., 2007). We tested their contribution by adding the glycine receptor antagonist strychnine (1 μm) and observed that it had no effect on Ihold and did not occlude the effect of picrotoxin (ΔIhold following application of strychnine: −0.8 ± 0.8 pA, n = 5, p = 0.40 compared with control; consecutive application of picrotoxin: 15.6 ± 3.3 pA, n = 4, p = 0.018 compared with strychnine). We further ruled out the potential contribution of GABACRs to picrotoxin-induced shift in Ihold by applying the GABAC antagonist TPMPA (50 μm), which also did not affect the amplitude of the picrotoxin-sensitive tonic current (ΔIhold following application of TPMPA: −0.14 ± 1.78 pA, p = 0.9 compared with control; picrotoxin: 11.7 ± 1.7 pA, p = 0.006 compared with TPMPA; n = 4). Thus, we conclude that neither glycine receptors nor GABACRs contribute to tonic conductance revealed by picrotoxin.
Tonic currents in DGCs are insensitive to SR95531 under baseline conditions
These results suggest either that [GABA]e is maintained at a constant concentration despite decreased vesicular GABA release (e.g., by nonvesicular release) and/or that there is another mechanism involved in generating tonic currents that is independent of [GABA]e.
To distinguish between these possibilities, we took advantage of the distinct pharmacologies of different GABAAR antagonists. Picrotoxin is an open channel blocker and acts as a noncompetitive GABAAR antagonist that has equivalent efficacy in blocking low-affinity synaptic and high-affinity extrasynaptic GABAARs (Stell and Mody, 2002). In contrast, SR95531 is a competitive GABAAR antagonist and displaces GABA from its binding site (Hamann et al., 1988). While low concentrations of SR95531 (0.5 μm) partially inhibited and high concentrations (125 μm) totally blocked sIPSCs, there was no significant increase in Ihold, which became more negative at high SR95531 concentrations (ΔIhold, 6.7 ± 1.6 pA, n = 5, p = 0.014 compared with 0.5 μm SR95531; Fig. 1B; Table 1), suggesting a partial agonist effect (this effect was not observed when SR95531 was applied after picrotoxin: ΔIhold, −1.2 ± 0.9 pA, n = 3, p = 0.3). Subsequent application of picrotoxin resulted in an outward shift of Ihold by 18.9 ± 3.4 pA (p = 0.0049, n = 5), and a decrease in baseline RMS noise by 0.39 ± 0.05 pA (p = 0.0012, n = 5) in the same cells (Fig. 1B). This effect of picrotoxin was occluded by prior application of pentylenetetrazol (Table 1), another noncompetitive GABAAR antagonist (Huang et al., 2001). ΔIhold following application of picrotoxin in the presence of pentylenetetrazol was 0.4 ± 0.7 pA (n = 5).
As previously reported (Semyanov et al., 2003), markedly increasing GABA in the perfusate to 5 μm revealed a SR95531-sensitive component of the tonic current (Fig. 1C). Under these conditions, application of a low concentration of SR95531 (0.5 μm) caused a significant outward shift of 29.4 ± 14.4 pA in the holding current (p = 0.016, n = 8) and a significant decrease in RMS noise by 3.0 ± 0.8 pA (p = 0.008, n = 8, Fig. 1C). Even in the presence of a high concentration of SR95531 (125 μm) under these conditions, application of picrotoxin caused a further outward shift in holding current by 9.2 ± 2.1 pA (p = 0.031, n = 6) and a reduction of RMS noise by 0.51 ± 0.09 pA (p = 0.003, n = 6; Fig. 1C).
The lack of efficacy of SR95531 under baseline conditions suggests that only negligible ambient GABA can be detected by dentate granule cells. Alternatively, SR95531 may not bind to the receptors mediating tonic current. The efficacy of SR95531 in expression systems in which α4δ subunit-containing receptors (the main contributor to tonic currents in dentate granule cells) (Glykys and Mody, 2007b) are expressed (Brown et al., 2002) argues against this. Nevertheless, we tested this using another competitive GABAAR antagonist, bicuculline. Application of bicuculline (10 μm) blocked all synaptic activity, induced a small outward shift of holding current by 5.9 ± 0.6 pA (p = 0.00019, n = 6), and decreased baseline RMS noise by 0.72 ± 0.11 pA (p = 0.0014, n = 6; Fig. 1D). This result can be explained by the inverse agonist activity of bicuculline (Ueno et al., 1997; Bai et al., 2001; McCartney et al., 2007). If SR95531 does not bind to the same receptors as bicuculline, then it should have no effect in the presence of bicuculline. Conversely, if SR95531 does bind then it should displace the bicuculline and paradoxically induce an inward current. The effect of bicuculline was indeed reversed by application of SR95531 (25 μm), which activated an inward current of 10.7 ± 1.7 pA (p = 0.0013, n = 6), and increased baseline noise by 0.27 ± 0.06 pA (p = 0.008, n = 6). This was completely blocked by subsequent application of picrotoxin (Fig. 1D). These results suggest that all three GABAAR antagonists bind to the same receptors and that these receptors are not detecting GABA under baseline conditions. Moreover, the lack of effect of SR95531 in baseline conditions indicates that other possible endogenous agonists (e.g., taurine) are also not mediating the tonic current (Jia et al., 2008).
SR95531-insensitive tonic currents in DGCs are mediated by not only δ subunit-containing GABAARs
Tonic GABAAR-mediated currents in DGCs are predominantly mediated by α4 and δ subunit-containing receptors (Stell and Mody, 2002; Stell et al., 2003; Caraiscos et al., 2004). We therefore tested whether δ-GABAARs contribute to the SR95531-resistant tonic current. We took advantage of knock-out mice lacking δ subunits of GABAARs. In these mice, in contrast to rats, there was a very small SR95531-sensitive tonic current. However, the majority of the tonic current was SR95531 insensitive (Fig. 2). In δ−/− mice in the presence of 25 μm SR95531, the effect of picrotoxin on holding current was ∼60% less than that in wild-type mice (littermate controls), implying that δ-GABAARs contribute to SR95531-resistant tonic currents (Fig. 2). However, there was a significant SR95531-insensitive tonic current in δ−/− mice, suggesting that other GABAAR subtypes can also contribute to the SR95531-insensitive tonic current in DGCs (Glykys et al., 2008).
Low [GABA]e is maintained both in vitro and in vivo
The fact that DGCs display SR95531-insensitive tonic currents suggests that [GABA]e in ex vivo tissue is maintained at concentrations that are not detectable by high-affinity GABAARs.
We used GABAAR openings in outside-out “sniffer” patches from dentate granule cells (Fig. 3A,B) to give a semiquantitative estimate of [GABA]e in slices. We first confirmed that channel openings could be recorded in outside-out patches in the presence of 10 μm GABA (channel openings disappeared in the absence of GABA) (Fig. 3A). These channels were blocked by picrotoxin and had an opening frequency of 20.1 ± 5.6 Hz, a conductance of 39.4 ± 7.2 pS, and an average open time of 32.1 ± 7.1 ms (n = 11). We were able to detect GABAAR openings when the patch was 5 μm above the slice (0.55 ± 0.05 Hz, n = 5), indicating that there is sufficient extracellular GABA to bind to GABAARs (Fig. 3B). These openings did not occur when the patch was moved 300 μm above the slice and recurred on lowering the patch again to the slice (Fig. 3B). Consequent application of 200 nm GABA to the perfusate, comparable to that found in CSF (Glaeser and Hare, 1975), almost doubled the frequency of GABAAR openings recorded above the slice surface. The effect of 200 nm GABA application was even more evident when patches were raised 300 μm above the slice. This indicates that [GABA]e is <200 nm in the slice (Fig. 3B) and is consistent with the predicted GABA transporter equilibrium [GABA]e of ∼100 nm (Wu et al., 2007). We further confirmed that similarly low [GABA]e is maintained in vivo by using the zero-net-flux microdialysis method. Microdialysis probes were inserted into the CA3-dentate gyrus region via guide cannula, and in vivo [GABA]e was calculated to be 92 ± 10 nm (n = 5; Fig. 3C,D).
Since an increase in GABAAR openings occurs as the patch approaches the surface, yet there is no evidence of significant GABA-medicated tonic currents, it is likely that the activity of GABA transporters prevents extrasynaptic GABAA receptors from being exposed to even such low concentrations under baseline conditions. When GABA in the perfusate was increased to 200 nm, comparable to that found in CSF (Glaeser and Hare, 1975), Ihold did not change significantly and SR95531 (25 μm) had no significant effect on Ihold (ΔIhold following consequent application of GABA: −1.21 ± 2.03 pA, p = 0.6; ΔIhold following consequent application of SR95531: 1.7 ± 1.4 pA, p = 0.29; n = 5), indicating that there is no increase in the GABA detected by extrasynaptic GABAARs.
Spontaneously opening GABAARs in DGCs
The evidence thus far implies that tonic inhibition in DGCs in situ is predominantly GABA independent. Can tonic currents be mediated by spontaneously opening GABAARs? Such receptors have been reported in excised patch and cell-attached recordings from hippocampal neurons (Birnir et al., 2000a), but not in the DGCs (Birnir et al., 1994). Failure to detect spontaneously opening GABAARs may be due to the small area sampled using outside-out patches and/or the dependence of such openings upon the internal neuronal environment. Indeed, we noted that in very few excised patches (<10%) could we detect infrequent channel openings (less than one every 20 s). We therefore performed recordings from nucleated patches (i.e., whole-cell excisions containing intact nuclei). In this preparation, spontaneous GABAAR openings were recorded in the majority of patches (Fig. 4). These openings had a conductance (37.4 ± 6.1 pS) similar to that observed in excised outside-out patches with applied GABA (39 ± 7.2 pS; Fig. 3A), and comparable to that determined in other studies (Bright et al., 2011). As expected, spontaneous openings were not affected by SR95531 (25 μm) but were fully blocked by picrotoxin (Fig. 4A–D). Since a high concentration of SR 95521 (125 μm) had acted as a partial agonist (Fig. 1), we tested the effect of SR 95521 (125 μm) on spontaneous openings in nucleated patches and, as predicted, found that it increased the frequency of openings (Fig. 4E). Addition of 10 μm GABA to the nucleated patches revealed SR95531-sensitive channels with the same conductance (Fig. 5).
Discussion
Our results indicate that GABA-independent openings of GABAARs are the major contributor to tonic currents in dentate granule cells ex vivo, even when GABA concentrations in the perfusate are increased to levels comparable to those measured in vivo.
We took advantage and confirmed the different pharmacologies of GABAAR antagonists (Table 1), and have shown that in situ DGC tonic currents are resistant to SR95531, indicating that they are not being generated by GABA binding to the receptors. This is consistent with similar observations in cultured neurons, other brain areas and cell types (Birnir et al., 2000b; McCartney et al., 2007). Importantly, SR95531 has a paradoxical effect of inhibiting the action of the inverse agonist bicuculline. This result suggests that SR95531 is displacing bicuculline from the receptors and consequently that the lack of effect of SR95531 is not due to a failure to bind to the receptors that are mediating the tonic current. At high concentrations (125 μm), we also found that SR95531 can act as a partial agonist. A partial agonist effect of SR95531 (100 μm) and bicuculline (1 mm) has been described previously at GABAARs in which there is a point mutation in the γ subunit (Ueno et al., 1997). We did not test bicuculline at this concentration and so cannot exclude a partial agonist effect of very high concentrations of bicuculline at extrasynaptic receptors.
This lack of GABA binding explains why inhibiting vesicular GABA release in our studies has no effect on the tonic current. The low concentration of GABA despite vesicular release is almost certainly due to efficient GABA uptake, because, when GABA transporters are blocked, tonic currents become dependent upon vesicular GABA release (Glykys and Mody, 2007b). The effect of decreasing vesicular release has variable results depending upon the nature of the tissue studied, suggesting that GABA concentrations and the consequent detection of extracellular GABA by GABAARs vary between brain regions. For example, in thalamocortical neurons, 50 μm SR95531 has been demonstrated to reveal robust tonic GABAAR-mediated currents (Cope et al., 2005). There may even be local inhomogeneities of GABA concentrations due to regional differences in the distribution of GABA transporters (Semyanov et al., 2003). Indirect evidence of this in the hippocampus is supplied by the observation that tonic currents are reduced in CA1 interneurons, but not in pyramidal cells of GAD65-deficient mice (Song et al., 2011). However, whether the heterogeneity of extracellular GABA concentrations observed in vitro occurs in vivo is unclear.
Spontaneous openings of GABAARs have been previously described in some preparations (e.g., in isolated hippocampal pyramidal cells and in outside-out patches from pyramidal neurons, but not in DGCs). However, the role for these spontaneous openings, and whether they can contribute to GABAAR-mediated signaling in situ, has remained unclear (Macdonald et al., 1989; Birnir et al., 2000b). Here, we recorded from nucleated patches to show directly that spontaneously opening receptors are present in dentate granule cells. As predicted from other studies (McCartney et al., 2007), they are not affected by the competitive GABAAR antagonist SR95531 and are completely blocked by the open channel blocker picrotoxin. Lack of evidence of such openings in outside-out patches from dentate granule cells in previous studies may be related both to the small area of membrane and also to the effects of the internal cellular milieu on GABAAR channel gating.
A vexed question is how GABA in a slice relates to the in vivo situation. Direct measurements of GABA in CSF have revealed concentrations on the order of 200 nm (Glaeser and Hare, 1975). Using the zero-net-flux microdialysis method, we have estimated the concentration in the extracellular fluid to be even lower. This is somewhat at odds with an earlier study that estimated the GABA concentration in the extracellular fluid of the hippocampus to be 800 nm (Lerma et al., 1986). This measurement, however, was confounded by the complex method used to retrieve absolute concentrations from dialysate concentrations and reliance on estimates of diffusion across the dialysis membrane. It should also be noted that these previous estimates were performed in rats under urethane anesthesia, and sampling was performed 1 h after probe implantation, when the blood–brain barrier could still be disrupted. The microdialysis method averages over a large area, and so we cannot exclude the possibility that the regional GABA concentrations may differ and, in particular, that the concentration detected by neurons may be greater than this depending upon the balance of local release and uptake; however, this was not observed by us (and others) using the sniffer patch technique. Indeed, 100 nm is close to the EC20 for extrasynaptic δ-subunit-containing GABAARs (Bright et al., 2011), and so the absence of a detectable GABA-mediated current under baseline conditions or when GABA in the perfusate is increased to 200 nm indicates that the GABA detected by neurons is less than that present in the extracellular space, perhaps due to efficient local GABA uptake. These results strongly argue against the widespread addition of high GABA concentrations (usually 5 μm) to the perfusate during ex vivo experiments. Indeed, these concentrations of GABA are 50-fold greater than those we detect in vivo.
Our findings have several important implications. First, our results indicate that studies that use SR95531 to measure tonic currents may significantly underestimate them. The results may also explain discrepancies in the measurement of tonic currents in DGCs, which in one study (and in contrast to others) was proposed not to be present in control tissue because of the lack of effect of SR95531 (Zhan and Nadler, 2009). Furthermore, the widespread use of SR95531 and on occasion high concentrations of SR95531, which could have a partial agonist effect, may significantly underestimate the magnitude of tonic currents present. Even when GABA is gating the receptor, displacement by SR99531 would not prevent spontaneous openings, and may therefore not accurately measure the tonic current. Moreover, our results indicate that there is always a tonic current present in dentate granule cells even when extracellular GABA concentrations are low (as they are most of the time). The importance of maintaining such an inhibitory tone in neurons is further underscored by the observation that knocking out receptors generating tonic current in cerebellar granule cells leads to an upregulation of a two-pore potassium channel to maintain the conductance (Brickley et al., 2001).
Footnotes
This work was supported by Wellcome Trust Grants WT083163 (A.I.W., A.S., F.K., A.C.E.L., and M.C.W.) and WT084311 (D.A.R.); Medical Research Council Grants G10000008 (D.B. and J.J.L.), and G0900613 and G0802216 (D.A.R.); Tenovus Scotland (J.J.L. and D.B.); Anonymous Trust (J.J.L. and D.B.); Epilepsy Research UK Grants F1001 (M.B.H.) and A0832 (I.P.); and Worshipful Company of Pewterers (I.P.).
- Correspondence should be addressed to either Ivan Pavlov or Matthew C. Walker, Department of Clinical and Experimental Epilepsy, UCL Institute of Neurology, London WC1N3BG, UK, i.pavlov{at}ucl.ac.uk or mwalker{at}ion.ucl.ac.uk