GABAergic inhibition in the brain can be classified as either phasic or tonic. γ-Aminobutyric acid (GABA) uptake by GABA transporters (GATs) can limit the time course of phasic currents arising from endogenous and exogenous GABA, as well as decrease a tonically active GABA current. GABA transporter subtypes 1 and 3 (GAT-1 and GAT-3) are the most heavily expressed of the four known GAT subtypes. The role of GATs in shaping GABA currents in the neocortex has not been explored. We obtained patch-clamp recordings from layer II/III pyramidal cells and layer I interneurons in rat sensorimotor cortex. We found that selective GAT-1 inhibition with NO711 decreased the amplitude and increased the decay time of evoked inhibitory postsynaptic currents (IPSCs) but had no effect on the tonic current or spontaneous IPSCs (sIPSCs). GAT-2/3 inhibition with SNAP-5114 had no effect on IPSCs or the tonic current. Coapplication of NO711 and SNAP-5114 substantially increased tonic currents and synergistically decreased IPSC amplitudes and increased IPSC decay times. sIPSCs were not resolvable with coapplication of NO711 and SNAP-5114. The effects of the nonselective GAT antagonist nipecotic acid were similar to those of NO711 and SNAP-5114 together. We conclude that synaptic GABA levels in neocortical neurons are controlled primarily by GAT-1, but that GAT-1 and GAT-2/3 work together extrasynaptically to limit tonic currents. Inhibition of any one GAT subtype does not increase the tonic current, presumably as a result of increased activity of the remaining transporters. Thus neocortical GAT-1 and GAT-2/3 have distinct but overlapping roles in modulating GABA conductances.
The inhibitory actions of γ-aminobutyric acid (GABA) in the CNS can be divided into synaptic, phasic inhibition (i.e., inhibitory postsynaptic currents [IPSCs]), and persistent, extrasynaptic tonic inhibition, seen as a tonically active conductance or current (Farrant and Nusser 2005; Mody 2005; Semyanov et al. 2004). Whereas phasic inhibition communicates information locally from a specific presynaptic neuron onto a postsynaptic neuron, tonic inhibition is involved in maintaining a general inhibitory tone (Brickely et al. 2001; Chadderton 2004; Ulrich 2003). GABA transporters (GATs) have been shown to modulate both tonic and phasic GABAergic signaling, particularly in the cerebellum and hippocampus (Nusser and Mody 2002; Overstreet et al. 2000; Rossi et al. 2003; Semyanov et al. 2003). GATs can limit “spillover” of GABA from the synapse, promoting synapse independence (Isaacson et al. 1993; Overstreet and Westbrook 2003; Rossi and Hamann 1998). GAT antagonists are widely used as anticonvulsants (Adkins and Noble 1998; Genton et al. 2001; Sills 2003) and brain GAT expression is altered in epilepsy, brain malformations (Calcagnotto et al. 2002), after chemical injury (Zhu and Ong 2004), and in patients with schizophrenia (Schleimer et al. 2004).
There are four identified GAT subtypes, classified in the rat as GAT-1, GAT-2, GAT-3, and BGT-1 (Borden 1996). Initial studies indicated that GAT subtypes were differentially targeted to neurons and glial cells (Krogsgaard-Larsen et al. 1987), with GAT-1 restricted to neurons and GAT-3 restricted to glia. Recent studies show that the idea of the exclusive allocation of GAT-1 to neurons is simplistic and GAT-1 has now been described on glial cells (Conti et al. 2004; Kinney and Spain 2002; Ribak et al. 1996; Vitellaro-Zuccarello et al. 2003). In the mature neocortex, glial cell processes in all layers express (in decreasing abundance) GAT-3, GAT-1, and GAT-2 (Conti et al. 1998, 1999; Minelli et al. 1995, 1996, 2003) Neurons express (in decreasing amounts) GAT-1 and GAT-2 (Conti et al. 1999). However, neuronal GAT-1 is specifically and highly expressed at axon fibers and terminals (Chiu et al. 2002), whereas GAT-2 is not located at synapses (Conti et al. 1998; Minelli et al. 2003). In general, GAT-2 expression is low in mature animals compared with GAT-1 and GAT-3 (Conti et al. 2004).
The first experiments investigating the role of GATs were conducted primarily with the GABA analog nipecotic acid, a nonselective GAT inhibitor (Schousboe et al. 1979). Nipecotic acid prolongs the decay of iontophoretically applied GABA and the late phase of evoked IPSCs (Dingledine and Korn 1985; Hablitz and Lebeda 1985). Inhibitors of GAT-1 such as tiagabine, SKF89976, and 1-(2-(((diphenylmethylene)imino)oxy)ethyl)-1,2,5,6-tetrahydro-3-pyridinecarboxylic acid (NO711) were the first widely available GAT subtype-specific antagonists. Differences were observed in the effects of the GAT-1 specific antagonists compared with nipecotic acid (Roepstorff and Lambert 1992). The interpretation of these results was complicated because, as opposed to tiagabine and NO711, nipecotic acid is a GAT substrate. In addition, it is thought that nipecotic acid can result in heteroexchange for GABA by GATs (Solis and Nicoll 1992) possibly confounding the conclusions of the role of GATs in regulating GABA concentrations.
GAT-1 antagonists increase the decay of evoked IPSCs (Engel et al. 1998; Rossi and Hamann 1998; Overstreet and Westbrook 2003; Roepstorff and Lambert 1992; Thompson and Gahwiler 1992) while generally having no effect on spontaneous IPSCs (sIPSCs). GAT-1 inhibitors also increase a tonic conductance in both the cerebellum (Rossi et al. 2003) and the hippocampus (Nusser and Mody 2002; Petrini et al. 2004; Semyanov et al. 2003). Evidence is emerging that nonGAT-1 transporters are also actively involved in GABA regulation. Tonic currents in cerebellar granule cells are increased by β-alanine, a transportable GAT-2/3 antagonist (Rossi et al. 2003), and β-alanine inhibits synaptically evoked transporter currents in neocortical glial cells (Kinney and Spain 2002). (s)-(−)-1-[2-[tris-(4-methoxyphenyl)methoxy]ethyl]-3-piperidinecarboxylic acid (SNAP-5114) is a nontransportable antagonist with selectivity for GAT-2 and GAT-3 (Borden 1996; Borden et al. 1994; Soudijn and Wingaarden 2000). Few studies exist that have examined the effects of SNAP-5114 in the brain (Bolteus and Bordey 2004; Dalby 2000; Galvan et al. 2005).
In the present study, we used the whole cell patch-clamp technique to explore the role of specific and nonspecific GAT antagonists in modulating phasic and tonic currents in layer I interneurons and layer II/III pyramidal cells of the rat neocortex.
Postnatal days 17–22 Sprague–Dawley rats were deeply anesthetized with ketamine (100 mg/kg) and decapitated. All experiments were conducted in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals using a protocol approved by the University of Alabama at Birmingham Institutional Animal Care and Use Committee. The brain was rapidly removed and submerged in ice-cold, oxygenated (95% O2 and 5% CO2) cutting solution containing (in mM): 125 NaCl, 3.5 KCl, 26 NaHCO3, 10 d-glucose, 3 MgCl2, and 1 CaCl2. Coronal slices (300 μm) were cut from the right sensorimotor cortex with a Vibratome (Ted Pella, Riverside, CA). Slices were stored in the cutting solution for 45–60 min at 37°C and then maintained at room temperature (22–24°C) until used for recordings.
Whole cell voltage-clamp recordings were made from visually identified layer I interneurons and layer II/III pyramidal cells using a Zeiss Axioscop FS (Carl Zeiss, Thornwood, NY) microscope equipped with infrared differential contrast optics. Layer I interneurons were selected based on their location in the cell-sparse area within 80 μm from the pia. Layer II/III pyramidal cells were identified by their pyramidal shape, the presence of a prominent apical dendrite, and distance from the pia (200–300 μm). Roughly equal numbers of layer I and layer II/III pyramidal cells were used in this study. No significant differences were found between the two groups of cells and thus the results were pooled.
Electrodes (KG-33 glass, Garner Glass, Claremont, CA) had resistances of 2–6 MΩ when filled with intracellular solution. Series resistance (Rs) and input resistance (Rin) was carefully monitored during each experiment with a 2- to 5-mV hyperpolarizing voltage step. Rs was estimated by measuring the peak of the transient current according to the formula Rs = Vstep/Itransient. Experiments were excluded from analysis if Rs exceeded 20 MΩ or changed by >25% during the experiment. Series resistance was not compensated.
The extracellular recording solution contained (in mM): 125 NaCl, 3.5 KCl, 26 NaHCO3, 10 d-glucose, 1.3 MgCl2, 2.5 CaCl2, 0.01 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), 0.02 d-(−)-2-amino-5-phosphonopentanoic acid (APV), and 0.01 SCH50911. Slices were placed in a 1-mL recording chamber, submerged in the recording solution, and perfused by a peristaltic pump at a flow rate of 4 mL/min. The solution was heated to 32–35°C by an in-line heater (Warner Instruments, Hamden, CT) and the temperature was monitored by a thermistor placed in the recording chamber. All drugs were bath applied.
The intracellular pipette solution for most evoked IPSC and GABA pressure application (“puff”) experiments contained (in mM): 125 KCl, 10 HEPES, 0.5 EGTA, 2 Mg-ATP, and 0.2 Na-GTP. The liquid junction potential was experimentally measured to be 5 mV. Intracellular solution for tonic current and reversal potential measurements contained (in mM): 120 Cs-MeSO4, 10 HEPES, 11 EGTA, 1 CaCl2, 2 MgCl2, and 5 lidocaine N-ethyl bromide (QX314). Liquid junction potential for this solution was 15 mV. The pH of the intracellular solutions was adjusted to 7.3 with either KOH or CsOH, and the osmolarity adjusted to 295–300 mOsm with sucrose. Reported potentials have been corrected for the liquid junction potential. Some evoked IPSC and pressure application experiments were also performed using the Cs-MeSO4 internal solution. Our observations were not affected by the choice of internal solution used and thus all data were pooled. The estimated chloride reversal potential for the KCl-based solution was 0 mV and the experimentally measured reversal potential for the Cs-MeSO4 solution was −30 mV. Neurons were voltage clamped at −65 mV (KCl solution) or +45 mV (Cs-MeSO4 solution) and thus the chloride-mediated currents were inward (negative) with KCl and outward (positive) with Cs-MeSO4. In the figures all currents are shown as outward currents, with the axis labeled “−pA” or “−nA” to designate the experiments recorded with KCl.
Voltage-clamp recordings were made with a Multiclamp 700A amplifier (Axon Instruments, Riverside, CA). Cells were held for ≥10 min after breakthrough before beginning recording data. IPSCs were evoked using a bipolar tungsten electrode placed on the surface of the slice within layer I, 100 μm lateral to recorded neuron. Stimulus durations and intensities ranged from 80 to 150 μs and 40 to 120 μA, respectively, and were adjusted to reliably yield the smallest possible amplitude IPSCs. Except where noted, IPSCs were evoked every 15 s and alternated in each individual neuron between a single stimulus and a “train” stimulation (three stimuli at 166 Hz).
GABAA agonist puffs were controlled with a picospritzer (General Valve, Fairfield, NJ) and delivered by a patch pipette at intervals of 15 s. The pipette was placed 10–20 μm from the cell soma and contained either GABA (10 μM) or muscimol (3 μM) dissolved in (in mM): 125 NaCl, 3.5 KCl, 20 HEPES, and 10 glucose; the pH was adjusted to 7.3 with NaOH. Delivery pressure was fixed at 6 psi and the duration (8–25 ms) was adjusted to yield initial currents of between 500 and 3,000 pA. In the absence of drugs, GABA-evoked currents had stable amplitudes and kinetics for >45 min (n = 3). Pressure application of the puff solution without GABA yielded no detectible current (n = 2; see also DeFazio et al. 2000). Increasing the duration of pressure application in the same neuron predictably increased the recorded GABA currents, but had negligible effects on response kinetics, over a wide range of test current amplitudes (500–3,000 pA).
Data were filtered at 2–5 kHz (IPSCs) or 500 Hz (pressure application) and digitized at 10 kHz using Clampex 8 software (Axon Instruments). Data were analyzed with Clampfit 9 software (Axon Instruments). IPSCs were analyzed to measure peak amplitudes relative to baseline holding current, and decay times are listed as the time between 80 and 20% of the peak amplitude. Unless otherwise mentioned, the traces shown in the figures and reported values were obtained by averaging the last 3 min of data for both IPSCs (six total trials) and pressure application experiments (12 total trials) during the control, drug, and wash periods. sIPSCs were detected and analyzed using the template searching method in Clampfit 9. The baseline holding current for tonic current measurements was determined from the average holding current of three 100-ms epochs free of sIPSCs. Tonic currents were calculated as the change in holding current after 3 min of treatment with 20 μM bicuculline. Reversal potential calculations were mathematically corrected for series resistance error and analyzed as described previously (DeFazio et al. 2000).
Origin 7.5 (Microcal, Northampton, MA) was used for graphing and additional analysis. Error bars represent the SE. A Wilcoxon signed-rank test for matched pairs or a Wilcoxon rank-sum test was used to calculate P values. A value of P < 0.05 was considered statistically significant.
Ionic salts were obtained from Fisher Chemicals (Fairlawn, NJ), and all other compounds were purchased from Sigma–Aldrich (St. Louis, MO) except for HEPES (Calbiochem, San Diego, CA), and CNQX, APV, and SCH50911 (Tocris Cookson, Ellisville, MO).
To evaluate the possible role of GATs in modulating upper-layer neocortical GABAA currents, we pharmacologically isolated GABAA receptors by recording in the presence of the excitatory amino acid antagonists APV (20 μM) and CNQX (10 μM) and the GABAB antagonist SCH50911 (10 μM).
Effect of nipecotic acid on IPSCs and GABA-induced currents
Extracellular stimulation in layer I resulted in an evoked IPSC in the recorded neuron (Fig. 1A; thick line). Bath application of the nonselective GAT antagonist nipecotic acid (1 mM; thin line) caused a significant decrease in the IPSC amplitude and an increase in the decay time. Similar results were obtained in this neuron using a train of three stimuli at 166 Hz, a protocol designed to increase presynaptic GABA release compared with the single-stimulus protocol (Fig. 1B). The inset shows the rising phase of the train IPSCs on an expanded timescale. Figure 1, C and D, depicts the time course of the effects of nipecotic acid on IPSC amplitude and decay time. It can be seen that the GABAA antagonist bicuculline (20 μM) completely abolished the IPSCs. In addition to its effects on IPSCs, nipecotic acid also induced significant increases in the holding current (Fig. 1E). The change in the holding current on application of bicuculline was used as the measure of the tonic current. The average tonic current in nipecotic acid–treated cells was 440 pA (SD56; n = 5). An increase in the tonic current is consistent with increased activation of GABAA receptors resulting from persistently elevated extracellular GABA levels stemming from uptake inhibition and possible nipecotic acid–induced heteroexchange of GABA.
One explanation for the reduction in IPSC amplitude caused by nipecotic acid is the possible inhibition of presynaptic GABA release. Although GABAB receptors were blocked with SCH50911, release may have been inhibited by hyperpolarization or shunting as a result of the activation of tonic inhibition. Thus we examined the effects of nipecotic acid on currents elicited by GABA puffs (10 μM), where the amount of GABA release can be kept constant (Fig. 1F). Bath application of nipecotic acid caused a reduction of GABA current amplitude and increased the decay time of the GABA-evoked currents. The magnitude of these effects was similar to those seen with the IPSCs, implying that the actions of nipecotic acid on IPSCs are primarily postsynaptic, although we cannot rule out presynaptic contributions to the observed decreases in IPSC amplitudes.
Both GABA and nipecotic acid are substrates for GATs, and can compete for binding to GATs and subsequent transport. We used the GABAΑ agonist muscimol, which is not a substrate for GATs, to see how nipecotic acid affects an agonist that is not transported. Nipecotic acid reduced the amplitude of muscimol (3 μM) evoked currents (Fig. 1G). Nipecotic acid had no effect on the kinetics of the muscimol currents, as seen when the trace is scaled to the size of the control. The lack of an effect on the decay time of muscimol currents is the expected result for an agonist that has no transporter to remove it from the extracellular space. The decay time of the muscimol-evoked current provides an approximation of the theoretical time course of GABA current activation in the absence of GABA uptake.
Figure 1H summarizes the results of our experiments with nipecotic acid on IPSCs and agonist puffs. Absolute measurements for this and other figures are listed in Table 1. It can be seen that single and train-evoked IPSC amplitudes were significantly decreased by nipecotic acid, whereas decay times were increased. We found that these effects were reversible on wash, which contrasts to the general observations of others (Dingledine and Korn 1985; Roepstorff and Lambert 1992). Both GABA and muscimol current amplitudes were decreased to a similar extent as the evoked IPSCs. This points to a postsynaptic mechanism of action to account for the decrease in amplitude.
Because nipecotic acid induces a large tonic current, the uptake-dependent decrease in the amplitude of GABAA currents could be explained by either a reduction in the chloride driving force as a result of persistent chloride flux through GABAA receptors (DeFazio and Hablitz 2001; Ling and Benardo 1995), desensitization of GABAA receptors, or reduced number of unbound GABAA receptors arising from elevated extracellular GABA levels. It has been shown that chloride equilibrium potentials can be changed during whole cell patch-clamp recordings despite the dialysis of neurons with the intracellular solution through the patch pipette (DeFazio and Hablitz 2001; DeFazio et al. 2000). To test whether an elevation of the tonic current altered the driving force for chloride despite the dialysis of the patched neurons using the whole cell patch-clamp technique, we measured the reversal potential of GABA puffs by measuring the zero current level during a series of voltage steps before and after a 10-min application of nipecotic acid. Using the Cs-MeSO4 solution, the chloride equilibrium potential shifted from −30 (SD12) to −5 mV (SD10; n = 4, data not shown). Based on our holding potential of +45 mV when using Cs-MeSO4, this represents a decrease in the chloride driving force of 33% after treatment with nipecotic acid (i.e., [Vm − Ecl] decreased from 75 to 50 mV). Although this decrease in driving force likely contributes to the decreased GABA currents, it cannot entirely account for the 60 to 80% reduction in our observed IPSC and GABA puff amplitudes. In addition, the effects of nipecotic acid on GABA current amplitudes were similar when recording with the KCl-based intracellular solution, where chloride flux would be less likely to cause elevations in intracellular chloride concentrations sufficient to significantly alter the chloride equilibrium potential. From these observations we conclude that GABAA receptor desensitization or receptor occupancy is a major cause of the nipecotic acid–induced reduction of GABA-dependent current amplitudes.
Synergistic effects of SNAP-5114 and NO711
The role that individual GAT subtypes play in modulating the effects of GABA in the neocortex has not been investigated. Using the GAT-1–specific antagonist NO711 and the GAT-2/3–selective antagonist SNAP-5114 we quantified the relative contribution of GAT-1 and GAT-2/3 to the effects seen with nipecotic acid. Figure 2A shows a plot of holding current as a function of time. The average holding current in the presence of bicuculline was set to zero for each condition. The change in the holding current after the application of bicuculline represents the tonically active GABAA conductance. The tonic currents were larger in the presence of nipecotic acid or a combination of SNAP-5114 and NO711 than the control case or with either SNAP-5114 or NO711 alone. The tonic currents (as measured by the difference in holding currents between the arrows) for each condition are shown in Fig. 2B. Neither 20–100 μM NO711 nor 100 μN SNAP-5114 increased the tonic current from control levels. However, lower doses of SNAP-5114 and NO711 (40 and 10 μM, respectively) significantly increased the tonic current when used together. At higher concentrations, coapplication of NO711 and SNAP-5114 matched the effect of nipecotic acid. These results show that both GAT-1 and GAT-2/3 are involved in regulating tonic GABA currents. The synergistic effects of SNAP-5114 and NO711 argue against nonspecificity of either antagonist, and suggest that GAT-1 and GAT-2/3 each function at less than maximal capacity under control conditions.
NO711 (20 μM) reduced the amplitude and increased the decay time of both single and train-evoked IPSCs (Fig. 3A, left and right, respectively). In contrast, SNAP-5114 (100 μM) had little effect on evoked IPSCs (Fig. 3B). Coapplication of NO711 and SNAP-5114 (10 and 40 μM) reduced the amplitude of evoked IPSCs, and markedly increased the decay time (Fig. 3C). The scaled responses highlight that the initial decay closely matches that of the control IPSC, but is followed by a long, slow second decay component. Higher concentrations of NO711 and SNAP-5114 (20 and 100 μM, respectively; see Fig. 4) increased the decay time of both single and train-evoked IPSCs to as much as 4,000 ms.
Roughly half of the experiments (43%, 7/16) involving coapplication of NO711 and SNAP-5114 resulted in IPSCs with two clearly visible, distinct components to the decay, as seen in Fig. 3C. However, NO711 plus SNAP-5114 also resulted in IPSCs in which the initial fast decay was very small or appeared to be absent (Fig. 3D). In this example, both nipecotic acid and the coapplication of NO711 and SNAP-5114 resulted in an IPSC with similar kinetics. Nipecotic acid treatment resulted in cells with one decay component (as seen in Figs. 1, A and B and 2D) in 64% of experiments (7/11) and IPSCs with two clear decay components were observed in the other 36% (4/11) of experiments. Whether the fast component is observed in a given neuron may depend on the extent of desensitization at synaptic GABA receptors (see discussion).
Despite its effects on evoked IPSCs, NO711 (20 μM) had no effect on the decays [9.4 (SD1.6) vs. 9.7 ms (SD1.6)], amplitudes [55 (SD18) vs. 54 pA (SD20)], or frequencies [3.4 (SD1.7) vs. 3.1 Hz (SD1.9)] of sIPSCs (n = 7). Averaged sIPSCs before and after NO711 treatment are shown in Fig. 3E. We did not observe any spontaneous events after treatment with nipecotic acid or coapplication of NO711 and SNAP-5114. This may have been the result of an inability to resolve sIPSC, possibly stemming from a reduction in sIPSC amplitude together with an increase in baseline noise arising from an increase in the tonic current.
Our results with the effects of various combinations of GAT antagonists on IPSCs are summarized in Fig. 4. SNAP-5114 had no effects on its own, but worked synergistically with NO711 to mimic the effects of nipecotic acid. Unlike nipecotic acid, neither NO711 nor SNAP-5114 serves as a substrate for GATs. The fact that substrate and nonsubstrate antagonists yield similar results emphasizes that our observed effects on IPSCs are independent of the mechanism of transporter inhibition.
GAT inhibition and GABA-induced currents
We next tested the effects of GAT-subtype–specific antagonists on responses to GABA puffs. NO711 increased the amplitude and increased the half-width of GABA-induced currents (Fig. 5A). The effect on the amplitude was in contrast to the decrease seen with NO711 treatment on IPSCs. SNAP-5114 had little effect on GABA responses (Fig. 5B). Coapplication of NO711 (10 μM) and SNAP-5114 (40 μM) reduced the amplitude and increased the decay time (Fig. 5C). These results, a summary of which is shown in Fig. 5D, show that the effects of the uptake antagonists on GABA-evoked currents generally mimic those observed with evoked IPSCs, with the exception of NO711 by itself. This observation implies that the GABAA receptors activated by GABA puffs are not the same as those activated by synaptic GABA release (see discussion).
Our findings indicate that neocortical interneurons and pyramidal cells exhibit a tonic GABAA conductance and that GAT-1 and GAT-2/3 act together in neocortex to modulate both tonic and phasic GABAA currents. We also found that the combination of GAT-1 and GAT-2/3 antagonists closely mimics the effects of nipecotic acid. This suggests that some previously reported effects of nipecotic acid, attributed to nonspecific actions, may instead be evidence of GAT-2/3 activity.
To investigate the possible GAT subtypes involved in modulating neocortical GABA currents, we used two selective GABA uptake inhibitors, NO711 and SNAP-5114. NO711 is a potent inhibitor of GAT-1, which has an IC50 of 0.38 μM and is three orders of magnitude more selective for GAT-1 than other GAT subtypes (Borden et al. 1995). SNAP-5114 is moderately selective for GAT-3 over GAT-2, with an IC50 of 5 and 21 μM, respectively, and has an IC50 of 388 μM at GAT-1. The concentrations of SNAP-5114 used in our experiments (40–100 μM) would likely inhibit both GAT-3 and GAT-2. Despite evidence that GAT-2 is mainly localized to the leptomeninges and blood vessels (Durkin et al. 1995), there is evidence for some expression on neurons and glia (Conti et al. 1999). Our results suggest that functional GAT-2 and GAT-3 transporters are present on neurons and glia and regulate tonic and phasic GABAA conductances.
The majority of GAT-1 nears full expression levels in the neocortex by P10 (Yan et al. 1999), whereas GAT-2 and GAT-3 are expressed at adult levels by about P14 (Minelli et al. 2003). Our experiments were conducted on P17–22 animals, and thus our results are likely to be similar to those obtained in adult animals.
Both GAT-1 and GAT-2/3 must be inhibited to increase tonic currents
GABAA antagonists block a tonically active current in hippocampus (Bai et al. 2001; Nusser and Mody 2002; Semyanov et al. 2003), cerebellum (Brickley et al. 1996; Kaneda et al. 1995), and neocortex (Salin and Prince 1996). In hippocampus and cerebellum, the tonic current is increased when GABA uptake by GAT-1 is inhibited (Nusser and Mody 2002; Rossi et al. 2003; Semyanov et al. 2003). In contrast to the observations in other brain regions, our results indicate that GAT-1 antagonists alone do not increase the tonic current in neocortical pyramidal cells and interneurons. The tonic current was dramatically increased, however, when uptake was broadly inhibited with the nonselective GAT antagonist nipecotic acid or coapplication of NO711 and SNAP-5114. This observation implies that the tonic current is limited by both GAT-1 and GAT-2/3. The effect on the tonic current of inhibiting one subtype alone is masked by uptake by the remaining subtypes. This suggests that under control conditions neither GAT-1 nor GAT-2/3 is working at full capacity. A similar conclusion was reached by Dalby (2000), based on in vivo observations with selective and nonselective GAT antagonists in the thalamus. It was hypothesized that GAT-3 activity limits the effect of GAT-1 inhibition, and thus co-localized GAT-1 and GAT-3 act as “connected sinks” to limit the effect of selective blockade (Dalby 2003).
The tonic current activated by nipecotic acid and the coapplication of SNAP-5114 and NO711 persisted for over 10 min despite evidence that GABAA receptors can undergo both slow (Bianchi et al. 2002; Overstreet and Westbrook 2000) and fast desensitization (Galeretta and Hestrin 1997; Jones and Westbrook 1995, 1996). It may be that increases in GABA levels were insufficient to significantly desensitize GABAA receptors. Our tonic currents reached a peak at roughly 5 min with a slight decline over the next 2 min before stabilizing (Fig. 1E). This decline may represent a slow accumulation of intracellular chloride as suggested by our measurements of a change in GABAA reversal potential during this time. A gradual onset of desensitization may also contribute to the observed decline in the tonic current from its peak.
It is believed that the GABAA receptors mediating the tonic current are primarily located outside the synapse (Mody 2001). Synaptic and extrasynaptic GABAA receptors vary in their subunit composition (Brunig et al. 2002; Crestani et al. 2002; Farrant and Nusser 2005; Nusser et al. 1995; Wei et al. 2003). It is thought that synaptic receptors are composed predominantly of a γ2 subunit in association with α1, α2, or α3, whereas extrasynaptic receptors contain α4, α5, or α6 together with either δ or γ2 (Farrant and Nusser 2005). Deletion of the α5 subunit reduces the tonic current in the dentate gyrus (Caraiscos et al. 2004) as does loss of δ and α4 subunits (Peng et al. 2004). The GABAA receptor subunits found extrasynaptically tend to desensitize less rapidly and have a higher affinity for GABA (Bianchi et al. 2002; Saxena and McDonald 1996; Tia et al. 1996; Yeung et al. 2003). Although there is yet no specific evidence in neocortex linking the tonic current to extrasynaptic receptors, it is reasonable based on data from hippocampus and cerebellum to postulate that extrasynaptic GABAA receptors with distinct biophysical properties are primarily responsible for sensing the ambient GABA levels that give rise to the tonic current.
Decreased uptake can desensitize synaptic GABAA receptors
Nonselective inhibition of GATs resulted in a decrease in the amplitude of evoked IPSCs. This decrease could arise from either presynaptic or postsynaptic effects. Presynaptic GABAB receptors, which are known to decrease GABA release (Cobb et al. 1999; Davies and Collingridge 1993; Deisz 1999; Deisz and Prince 1989), were blocked in our experiments. A decrease in synaptic release could still be expected because the increased tonic current can suppress excitability (Brickely et al. 2001; Chadderton 2004; Ulrich 2003) or could activate putative presynaptic GABAA receptors (Kullmann et al. 2005; Ruiz et al. 2003). However, the amplitudes of GABA puffs were reduced to an extent similar to that of the evoked IPSCs. Because the output of GABA from the puffer pipette was kept constant, this reduction in amplitude likely represents postsynaptic effects. The reduction in both IPSC amplitudes and GABA puffs attributed to nipecotic acid or SNAP-5114/NO711 treatment was measured concomitantly with an increased tonic current. In this environment, postsynaptic receptors are possibly open or desensitized because of bound GABA as a result of elevated background GABA levels. There would thus be fewer unbound GABAA receptors available to be activated by synaptically released GABA, resulting in smaller evoked IPSCs.
Evoked IPSC kinetics in the presence of uptake antagonists suggest two distinct populations of GABAA receptors
Both nipecotic acid and coapplication of SNAP-5114 and NO711 increased the decay time of evoked IPSCs. In some cases, IPSCs required several seconds to decay to baseline. This is ample time for GABA to diffuse many microns away from synaptic GABA receptors (Clements 1996; Destexhe and Sejnowski 1995; Wahl et al. 1996). The GABAA current observed during these long decays is thus likely a result primarily of “spillover” activation of extrasynaptic receptors or activation of synaptic receptors at nearby synapses (Overstreet and Westbrook 2003; Rossi and Hamann 1998).
Half of our experiments with nonselective inhibition of GATs resulted in IPSCs with only one clear decay component (Figs. 1, A and B and 3D), whereas the rest had two distinct decay components (Fig. 3C). Scaled responses show that in IPSCs with two decay components the initial decay was fast and closely matched the decay of the control IPSCs, whereas the second component was extremely slow. The slow components were not well represented by an exponential function (data not shown), and baseline noise prevented accurate measurement near the tails of the IPSCs. We thus used the time to fall from 80 to 20% of the peak as our decay measurement, which is largely a measure of the slow component.
The presence of two decay components would correlate well with the existence of two distinct populations of GABAA receptors, distinguished by location, biophysical properties, or both. We propose that the fast component represents activation of unbound synaptic GABAA receptors, whereas the slow component reflects spillover out of the synapse induced by reduced uptake. If synaptic receptor availability is sufficiently reduced, perhaps by desensitization, then the fast component may become very small and give rise to the events with only a slow component, as seen in Figs. 1, A and B and 3D. This also implies that extrasynaptic receptors are less sensitive to desensitization.
GAT limits desensitization at synapses
In neocortical neurons, GAT-1 is highly localized to axon terminals (Chiu et al. 2002; Conti et al. 2004; Pow et al. 2005). Selective inhibition of GAT-1 with NO711 resulted in an increase in the decay time and decrease in the amplitude of evoked IPSCs. NO711 did not increase the tonic current, suggesting that the overall increase in extracellular GABA was small or otherwise limited largely to synapses. Evidence from the hippocampus also suggests that NO711-induced GABA elevations are primarily restricted to the synapse (Overstreet et al. 2000).
NO711 had no effect on either the amplitude or decay of sIPSCs. It may be that the effects of GAT-1 inhibition manifest only when multiple fibers are stimulated, as is likely with evoked IPSCs, or when fibers have multiple closely spaced release sites that may result in transmitter pooling (Arnth-Jensen 2002; Scanziani 2000). It has also been shown that unitary IPSCs are preferentially prolonged when synaptic density is high (Destexhe and Sejnowski 1995; Overstreet and Westbrook 2003). An alternate explanation for the lack of effect on sIPSCs may be that desensitization of postsynaptic receptors is promoted by activity-dependent release of GABA. If we assume that there are roughly 600 inhibitory synapses per neocortical neuron (Beaulieu et al. 1992) and that they are functionally equivalent, then an sIPSC frequency of 3 Hz would mean that release from any given synapse would occur only every several minutes. Even with synaptic uptake inhibited, this would likely be sufficient time for GABA to completely diffuse out of the synapse, thus limiting desensitization before the next spontaneous event. Evoking IPSCs every 15 s, as in our experiments, may have been sufficiently frequent to allow synaptic GABA levels to slowly build and desensitize GABAA receptors at those synapses, possibly by a slow desensitization (Overstreet et al. 2000). The effects of uptake inhibition were not dependent on the initial amplitude of evoked IPSCs because GAT antagonists had similar effects on both single evoked IPSCs and the larger train-evoked IPSCs.
In contrast to evoked IPSCs, NO711 increased the amplitude and half-width of responses to GABA puffs. It is unclear which populations of GABAA receptors are activated by the GABA puffs and thus comparisons with evoked IPSCs can be difficult. It may be that puff application of GABA preferentially targets extrasynaptic receptors that do not desensitize. In that case, the main effects of uptake inhibition would be to effectively increase GABA concentrations as seen by GABAA receptors. In addition, the relatively slow kinetics of puff currents would likely mask the effects of any desensitization because the reduced uptake allows the GABA from the puffs to spread over a larger area and activate larger numbers of receptors compared with the relatively limited spread of GABA released from presynaptic terminals. When SNAP-5114 was applied together with NO711 the amplitude of GABA puffs was instead decreased, whereas the half-width was significantly increased (Fig. 5D), despite the reduction in amplitude. Because SNAP-5114 plus NO711 also induces a large tonic current, the reduction in GABA puff amplitude could be attributable either to receptor desensitization or to the reduced availability of unbound receptors. The half-widths of GABA puffs in the presence of SNAP-5114 and NO711, as well as nipecotic acid, were similar to those seen with muscimol puffs under control conditions (Table 1). Because the binding affinity of GABA and muscimol are similar for many GABAA receptors (Ebert et al. 1997), and muscimol is not subject to any transport mechanisms, this suggests that GABA uptake was completely inhibited in the presence of those drugs.
The effects of GAT-2/3 are primarily extrasynaptic
The GAT-2/3 antagonist SNAP-5114 had no effect on evoked IPSCs. Yet when coapplied with NO711 it acted synergistically to increase decays and reduce amplitudes. Our data are consistent with the picture derived from immunohistochemical evidence in the rat neocortex where GAT-1 on neurons and glia is primarily concentrated at synapses, whereas GAT-2 and GAT-3 are expressed diffusely at extrasynaptic locations on glia with possible extrasynaptic expression of GAT-2 on neurons (Fig. 6). Glial cells also express nonsynaptic GAT-1. Thus GAT-1 inhibition elevates synaptic GABA levels, which can then spill out of the synapse. GABA spread and extracellular GABAA activation, however, are limited by GAT-2/3 located nonsynaptically on neurons and glia. Inhibition of GAT-1 allows GABA to leave the synapse, but it is not until uptake by GAT-2/3 is also blocked that GABA levels rise enough to result in a measurable tonic current. Because GAT-2/3 inhibition has no effects on its own, GAT-1 activity seems sufficient to constrain GABA mostly within the synapse, whereas GAT-2/3 works together with GAT-1 to maintain low levels of extracellular GABA. Evoked GABA release results in both GAT-1 and GAT-2/3 currents in neocortical glial cells (Kinney and Spain 2002), confirming immunohistochemical data and supporting our theory that GAT-2/3 is involved in extrasynaptic GABA regulation.
There is recent evidence that GAT4 (the mouse analog of the rat GAT-3) is potently inhibited by zinc (Cohen-Kfir et al. 2005), which is released together with glutamate at some glutamatergic terminals. Zinc has also been shown to increase extracellular GABA levels in the hippocampus (Takeda et al. 2004). This points to a role for GAT-3 in increasing GABA-mediated inhibition in response to increased excitatory activity. The extrasynaptic location of GAT-3 makes this transporter well suited for this task, but it remains to be seen whether this mechanism exists in the neocortex.
Nipecotic acid as a nonselective GAT antagonist
GATs are highly influenced by substrate concentrations on both sides of the membrane (Wu et al. 2003) and are easily reversible (Richerson and Wu 2003). Nipecotic acid is a substrate for GATs (Krogsgaard-Larsen et al. 1987) and can result in heteroexchange of GABA (Honmou 1995; Solis and Nicoll 1992; Szerb 1982) where one molecule of GABA is transported out of the cell as nipecotic acid is transported in. In addition, nipecotic acid has been reported to act as a GABAA receptor agonist (Barrett-Jolley 2001; Draguhn and Heinemann 1996). Activation of a tonic current by nipecotic acid as seen by us and others (Dingledine and Korn 1985; Draguhn and Heinemann 1996; Hablitz and Lebeda 1985; Wu et al. 2001) could be explained by both GABA heteroexchange and the GABAA agonist properties of nipecotic acid. Because GAT-1 was considered the predominant neuronal GABA transporter, differences in the results obtained with nipecotic acid and GAT-1–specific transporters have been attributed to these nonspecific effects of nipecotic acid (Roepstorff and Lambert 1992). Nonspecific effects of SNAP-5114 and NO711 have not been reported, and their synergistic action in increasing the tonic current suggests a direct effect on GAT-mediated uptake. A contribution of heteroexchange or agonist properties of nipecotic acid to our results cannot be ruled out. However, the fact that the tonic current was equally enhanced by nipecotic acid and the combination of SNAP-5114 and NO711 indicate that it is primarily inhibition of GAT-2/3 by nipecotic acid which accounts for the differences in effect between nipecotic acid and selective GAT-1 antagonists. It is likely that GAT-2 or GAT-3 plays a larger role in previously studied systems than originally thought.
GAT antagonists are known anticonvulsants. Seizure disorders such as epilepsy can originate and spread in specific areas of the brain (see Timofeev and Steriade 2004). Because the GAT subtypes are differentially distributed in various brain regions (Chiu et al. 2002; Engle et al. 1998; Vitellero-Zuccarello et al. 2003), the ability to inhibit specific GAT subtypes could become a powerful tool in treating epilepsy and other seizure disorders. Specifically, the ratio of selective GAT antagonist dosages could be best chosen for a given disorder phenotype and possibly maximize beneficial effects while minimizing unwanted side effects.
This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-22373.
We thank A. Margolies for excellent technical support and M. W. Quick for helpful critique of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 by the American Physiological Society