Gabapentin Increases a Tonic Inhibitory Conductance in Hippocampal Pyramidal Neurons

Experiments were approved by the Animal Care Committee of the University of Toronto (Toronto, Ontario, Canada). Primary cultures of hippocampal neurons were prepared from embryonic Swiss White mice as previously described.32Briefly, fetal pups (17 days in utero ) were removed from pregnant mice killed by cervical dislocation. The hippocampi were dissected from each fetus and placed in an ice-cooled culture dish. Neurons were then dissociated by mechanical trituration and plated on 35-mm culture dishes at a density of approximately 1 × 10⁶cells/ml. For the first 10 days in vitro , the cells were maintained in minimum essential medium supplemented with 10% fetal bovine serum and 10% horse serum (Life Technologies, Grand Island, NY). The neurons were cultured at 36.5°C in an environment of 5% CO²and 95% air. After the background cells had grown to confluence, 0.1 ml of a mixture of 5-fluorodeoxyuridine (4 mg) and uridine (10 mg) in 20 ml minimum essential medium was added to the extracellular solution to reduce the number of dividing cells. Subsequently, the medium was supplemented with 10% horse serum and changed every 3 or 4 days. Cells were maintained in culture for 14–21 days, after which a tonic current could readily be detected. On the day of the experiments, the cultures were removed from the incubator and the medium was exchanged with an extracellular recording solution that contained a chemical buffer system designed for atmospheric conditions.

Conventional whole cell currents were recorded under voltage clamp conditions (−60 mV) by means of an Axopatch 200 amplifier (Molecular Devices Corporation, Sunnyvale, CA) interfaced with a Digidata 1200 (Instrutech Corp., Elmont, NY). Records were filtered at 2 kHz and digitized at 10 kHz for off-line analysis with pCLAMP6 software (Molecular Devices Corporation). The extracellular fluid contained 120 mm NaCl, 1.3 mm CaCl², 5.4 mm KCl, 30 mm HEPES, and 28 mm glucose; the pH was adjusted to 7.4 by adding 1 m NaOH. Tetrodotoxin (300 nm) was added to the extracellular fluid to block voltage-sensitive sodium channels, and 6-cyano-2,3-dihydroxy-7-nitroquinoxaline (10 μm) and 2-amino-5-phosphonovalerate (40 μm) were added to inhibit ionotropic glutamate receptors. Recording electrodes were filled with a solution that contained 120 mm CsCl, 30 mm HEPES, 11 mm EGTA, 2 mm MgCl², 1 mm CaCl², 4 mm MgATP, and 2 mm tetraethylammonium chloride; the pH was adjusted to 7.3 by adding 2 m CsOH. Extracellular fluid and drug-containing solutions were delivered to the cultured neurons through glass perfusion barrels that were positioned close to the somata.

Whole cell recordings were obtained from the neurons cultured for 14–21 days. Two protocols for gabapentin application were selected. To determine whether gabapentin directly modified the function of the GABA^Areceptors that generated the tonic and synaptic conductances as well as current evoked by exogenous GABA, gabapentin was applied acutely to the neurons. To determine whether prolonged exposure to gabapentin would increase the tonic and synaptic conductances, neurons were preexposed to gabapentin for 36–48 h. The neurons were also treated with vigabatrin or a combination of gabapentin and vigabatrin for 36–48 h before the recordings. We previously observed that the maximum effect of vigabatrin on the tonic conductance occurred after 36–48 h of treatment. The same period for gabapentin treatment was selected on the basis of previous findings that suggested vigabatrin and gabapentin acted by similar mechanisms.20For experiments that examined the effects of chronic treatment, the media containing gabapentin and vigabatrin was washed away before the recordings were performed. This study was not designed to examine, in detail, the time- and dose-dependent actions of gabapentin or vigabatrin on the tonic inhibition.

The amplitude of the tonic current was calculated as the difference between the holding current (I^hold), measured before and during an application of the competitive GABA^Areceptor antagonist bicuculline (100 μm), as previously described.24,33,34The noncompetitive antagonist picrotoxin (100 μm) was also applied to ensure that the conductance was mediated by GABA^Areceptors. Fifteen 100-ms-long segments of recorded current that lacked miniature inhibitory postsynaptic currents (mIPSCs) were analyzed. All-point histograms were constructed from the data and mean values determined from Gaussian fits (pStat; Molecular Devices Corporation). The concentration–response plot was fit by nonlinear regression analysis using GraphPad Prism software (San Diego, CA) to estimate the concentration that caused 50% of the maximum effect (EC⁵⁰): Y = Control + (I^max− Control)/(1 + 10^{((Log EC50− X) · Hill slope)}), where Y is the response, I^maxis the maximum response, and X is the logarithm of the concentration. Y goes from the bottom to the top (I^max) of the data points with a sigmoid shape.

Spontaneous mIPSCs were analyzed with MiniAnalysis Software (Synaptosoft, Leonia, NJ), the detection threshold set three times higher than the level of baseline noise. The peak amplitude, charge transfer (Q, determined by integrating the area under the mIPSCs), and time constant of current decay (τ^decay) were analyzed. The τ^decaywas determined with the biexponential equation I(t) = A¹exp (−t/τ¹) + A²exp (−t/τ²), where I(t) is the current amplitude at any given time (t), A¹and A²are the amplitudes of the fast and slow decay components, and τ¹and τ²are the respective decay time constants of A¹and A². The weighted time constant of current decay was determined as τ^decay=ΣAⁱτⁱ/Σ^Ai.

For some experiments, GABA^Areceptor function was tested by application of exogenous GABA at either a subsaturating concentration (3 μm, EC¹⁰) or a saturating concentration (600 μm) for 2 s, followed by a washout period of 1 or 2 min, respectively, to allow the GABA^Areceptors to recover from desensitization.35Gabapentin was applied for at least 30 s before the application of a test pulse of GABA plus gabapentin.

The GABA sensitivity of GABA^Areceptor in neurons chronically treated with gabapentin (300 μm) was studied by constructing a concentration–response plot. GABA (0.1–600 μm) was applied for 10 s with a 2-min washout period between each application. The peak GABA-activated current was recorded at each concentration. Concentration–response curves were generated by fitting the data using nonlinear regression analysis and the sigmoidal equation described above. The amplitude of the maximum current generated by a saturating concentration of GABA (600 μm) was also measured.

Membranes from cultured neurons were prepared as described elsewhere36,37with all of the reagents stored on ice. Briefly, the neurons were scraped from the culture dishes and washed with phosphate-buffered saline (Invitrogen, Carlsbad, CA). The cells were then disrupted through a 22-gauge needle and lysed on ice for 40 min in 1 ml radioimmunoprecipitation assay lysis buffer (50 mm Tris-HCl, pH 8.0; 150 mm NaCl; 0.5% sodium deoxycholate; 0.1% sodium dodecyl sulfate; and 5 mm EDTA supplemented with protease inhibitors: 25 mm benzamidine (Sigma-Aldrich, St. Louis, MO), 0.1 mm sodium orthovanadate (Sigma-Aldrich), 5 μg/ml aprotinin (Sigma-Aldrich), 4 mm Pefabloc (Roche Applied Science, Laval, Quebec, Canada), 20 μg/ml leupeptin (Roche), and 1 μg/ml pepstatin (Roche)) per culture dish. The lysates were centrifuged for 20 min (10,000g  at 4°C), and the supernatant was collected. The protein concentrations of the supernatant were determined with use of the bicinchoninic acid protein assay reagent (Pierce Biotechnology, Rockford, IL).

Protein samples were resolved on 10% sodium dodecyl sulfate–polyacrylamide gels and transferred to nitrocellulose membranes. Blots were incubated in blocking buffer containing 5% nonfat milk in Tween-Tris–buffered saline (500 mm NaCl, 20 mm Tris Base, and 0.1% Tween-20, pH 7.5) for 1 h at room temperature with α¹, α², or α⁵rabbit polyclonal primary antibodies at dilutions of 1:1,000, 1:200, and 1:1,000, respectively. The α¹antibody was purchased from Upstate Biotechnology Inc. (Waltham, MA), and the α²antibody was purchased from Alomone Labs (Jerusalem, Israel). The α⁵antibody, which was previously described,38was donated by Werner Sieghart, Ph.D. (Professor, Division of Biochemistry and Molecular Biology, Vienna University Medical School, Vienna, Austria). The blots were then washed for 10 min four times with Tween-Tris–buffered saline and incubated for 1 h at room temperature in horse radish peroxidase–conjugated goat anti-rabbit antibody (Amersham Biosciences, Little Chalfont, Buckinghamshire, England) at a dilution of 1:10,000 in blocking buffer. The blots were washed again for 10 min four times in Tween-Tris–buffered saline, developed with SuperSignal West Pico for detection, stripped with Restore Western Blot Stripping Buffer, and reprobed with 1:1,000 anti–α-actinin antibody (Sigma-Aldrich, Oakville, Ontario, Canada) and horseradish peroxidase–conjugated donkey anti-mouse antibody (Amersham). The immunoblots were scanned, and the optical density of the positive bands was measured with a calibrated gray scale. The density of α-subunit staining was normalized to that of α-actinin. The experiments were repeated three times for each experimental condition with the use of homogenates from three animals. The values for protein from the drug-treated cultures were normalized to those for vehicle controls.

Bicuculline methiodide, picrotoxin, and gabapentin were purchased from Sigma-Aldrich. Vigabatrin, baclofen, and CGP55845 were obtained from Tocris Cookson (Ellisville, MO). All stock solutions were prepared by dissolving the appropriate amount of drug in distilled water, except for picrotoxin and CGP55845, which were dissolved in ethanol and dimethyl sulfoxide, respectively. The solutions were then filtered and stored at 4°C until used. The vehicle solutions did not influence the tonic conductance at the concentrations used.

Results are presented as mean ± SEM. Data were scrutinized for deviations from normality and appropriately transformed. Statistical significance was determined with a paired or unpaired Student t  test, as appropriate. When multiple groups were compared, an analysis of variance and Dunnett multiple comparison post hoc  test were used. Differences between groups were considered significant at P < 0.05.

The effects of gabapentin were assessed with the use of several electrophysiologic protocols. First, to examine direct effects of gabapentin on the function of GABA^Areceptors, current was evoked by applying exogenous GABA at a subsaturating or a saturating concentration. The GABA concentrations, 3 and 600 μm, were selected to mimic those that activate tonic and synaptic conductances, respectively. As shown in figure 2, gabapentin (300 μm) had no effect on the amplitude or time course of the current evoked by 3 μm GABA (control, 754.8 ± 54.1 pA; gabapentin, 732.6 ± 66.1 pA; n = 6) or 600 μm GABA (8.39 ± 0.61 and 8.12 ± 0.52 nA, respectively; n = 4). Similarly, gabapentin (1 mm) did not influence the current evoked by 3 μm GABA (265.3 ± 69.6 and 256.0 ± 67.4 pA, respectively; n = 3). These results confirm that gabapentin does not directly increase or reduce the function of GABA^Areceptors in pyramidal neurons.

Next, the short-term effect of gabapentin on a tonic inhibitory conductance generated by endogenous neurotransmitter was studied. In keeping with our previous findings,23,24pyramidal neurons grown in primary cultures generated a tonic inhibitory conductance that was revealed by applications of the GABA^Areceptor antagonists bicuculline and picrotoxin. The antagonists decreased I^holdand reduced the baseline noise, consistent with a reduction in the number of open receptors. Brief application of gabapentin, at concentrations up to 1 mm, did not influence I^holdeven when cells were patched and observed for up to 30 min (ΔI^holdwith 300 μm gabapentin, 0.1 ± 0.9 pA; n = 5). Therefore, gabapentin had no direct effect on the GABA^Areceptors that generated the tonic inhibitory conductance.

To determine whether prolonged exposure to gabapentin modified the tonic conductance, neurons were incubated for 36–48 h with various concentrations of gabapentin or a vehicle control. Gabapentin caused a concentration-dependent increase in the amplitude of the tonic conductance, as shown in figure 3A. The EC⁵⁰of gabapentin, determined by fitting the concentration–response plot, was 209 μm (95% confidence interval, 172–254 μm; figs. 3B and C). The maximum increase in tonic current, 2.7-fold, was observed at a gabapentin concentration of 300 μm, (control, 7.5 ± 1.1 pA; gabapentin, 23.6 ± 4.0 pA; n = 29, P < 0.05 by analysis of variance). Increasing the concentration of gabapentin to 1 mm did not further increase the tonic conductance (control, 8.9 ± 1.8 pA; gabapentin, 21.5 ± 4.6 pA; n = 10; P < 0.05).

The amplitude of the tonic current was studied by applying bicuculline and measuring the change in I^hold. However, bicuculline also inhibits several subtypes of potassium channels that might be influenced by gabapentin.39To ensure that the bicuculline-sensitive conductance was generated by GABA^Areceptors, bicuculline was washed away, and picrotoxin (1 mm) was applied. Picrotoxin and bicuculline caused a similar reduction in I^hold, which confirmed that GABA^Areceptors generated the tonic conductance (data not shown). Furthermore, ion channels regulated by GABA^Breceptors did not contribute to the tonic conductance, because application of a GABA^Breceptor antagonist (CPG 55845; 1 μm) and a GABA^Breceptor agonist (baclofen 100 μm) did not influence I^hold(ΔI^hold=−0.4 ± 0.5 pA, n = 11 and 0.0 ± 0.6 pA, n = 12, respectively; fig. 4).

In contrast to the nearly threefold increase in tonic conductance, prolonged exposure to gabapentin (300 μm) did not affect the frequency, amplitude, rise time, or time course of mIPSCs (table 1).

Gabapentin is reported to be a weak inhibitor of GABA transaminase,40whereas vigabatrin is a potent inhibitor with irreversible effects. Several laboratories, including ours, have reported that vigabatrin increases the tonic conductance in hippocampal pyramidal neurons, presumably by increasing the concentration of GABA in the extracellular fluid.29,30Vigabatrin caused a concentration-dependent increase in the tonic current, as shown in figures 5A and B. It was then postulated that if vigabatrin and gabapentin act through identical mechanisms, a maximally effective concentration of vigabatrin should occlude the actions of gabapentin. Neurons were pretreated with vigabatrin (100 μm or 1 mm) or a combination of gabapentin (300 μm) and vigabatrin (100 μm) for 36–48 h. The concentrations selected for coapplication caused a maximal increase in the tonic current when applied alone. As shown in figure 5C, coapplication had an additive effect: The increase was greater than that caused by gabapentin or vigabatrin alone (P < 0.05 by analysis of variance). Therefore, vigabatrin and gabapentin seem to act, at least in part, through different mechanisms.

An increase in tonic conductance results from an increase in either the expression of GABA^Areceptors or the probability of channel opening. The most plausible explanation is an increase in the ambient concentration of a ligand that activates GABA^Areceptors, which in turn increases the probability of channel opening. Nevertheless, gabapentin could increase the total number of high-affinity receptor subtypes expressed on the surface of neurons. In hippocampal pyramidal neurons, tonic conductance is primarily generated by α⁵subunit–containing GABA^Areceptors, whereas lower-affinity synaptic GABA^Areceptors in hippocampal pyramidal neurons likely contain α¹or α²subunits.23To determine whether gabapentin increased the number of synaptic or extrasynaptic GABA^Areceptors, we performed semiquantitative Western blot analyses using subunit-selective antibodies.41Homogenates were prepared from cultured neurons treated with placebo control, gabapentin (300 μm), or vigabatrin (100 μm) for 36–48 h. Experiments were performed three times with three sets of cultures, and the results were averaged. As shown in figures 6A and B, no differences were detected in the expression of subunit proteins after chronic treatment with gabapentin or vigabatrin. Therefore, the increase in tonic conductance by gabapentin or vigabatrin could not be attributed to increased expression of GABA^Areceptors that generate the tonic conductance in hippocampal neurons.

We next determined whether the sensitivity of GABA^Areceptors to GABA was changed by chronic treatment with gabapentin (300 μm). Concentration–response plots for current evoked by exogenous GABA revealed no difference in the EC⁵⁰values for gabapentin-treated and control neurons (12.5 ± 0.9 μm, n = 6 vs.  11.7 ± 1.0 μm, n = 6, respectively; P > 0.05; fig. 6C). Also, the maximum amplitude of the current evoked by a saturating concentration of GABA did not change after chronic treatment with gabapentin (300 μm) compared with control (6.8 ± 1.0 nA, n = 6 vs.  7.6 nA ± 1.5 pA, n = 6, respectively; P > 0.05). Therefore, the increase in tonic current by gabapentin was not due to an increase in the number or expression of high-affinity GABA^Areceptors on the cell surface.

This study has provided the first evidence that gabapentin has different actions on extrasynaptic and synaptic GABA^Areceptors in hippocampal pyramidal neurons. Gabapentin increased tonic inhibitory conductance but did not change the frequency, amplitude, or time course of mIPSCs. The increase in tonic conductance could not be attributed to direct positive allosteric modulation of GABA^Areceptors or to an increase in receptor expression. The most plausible explanation is that gabapentin indirectly increased tonic conductance by increasing the ambient level of neurotransmitter. Extrasynaptic GABA^Areceptors have a higher affinity for GABA than do postsynaptic receptors. Hence, extrasynaptic receptors may act as biosensors, detecting small increases in neurotransmitter that do not modulate postsynaptic receptors. Alternatively, gabapentin may have increased the release of transmitter at sites that reside in close proximity to extrasynaptic but not synaptic GABA^Areceptors. Our results also suggest that neuronal cultures can be used for future studies of the functional and biochemical consequences of long-term exposure to gabapentin.

These in vitro  results certainly do not preclude actions on high-affinity sites such as voltage-gated calcium channels containing the α²δ subunit in vivo . Nevertheless, they do suggest that gabapentin (as with most analgesics) may exert multiple effects, particularly at higher doses. The enhancement of the tonic current that we observed was consistent with the previous observation from electrophysiologic recordings in rat optic nerves that gabapentin (100 μm) increased responses caused by the nonvesicular release of GABA.42Another study showed that gabapentin (100 μm) increased a GABAergic conductance activated by nipecotic acid, a drug that inhibited the reuptake of GABA in rat hippocampal slices.43Gabapentin increased the release of [³H]GABA in slices from rat neostriatum incubated in vitro .44The gabapentin concentration–response plot was bell-shaped, with 0.1 μm nearly doubling the release of [³H]GABA, whereas 1 and 10 μm were without effect.44In studies of the human brain using ex vivo  magnetic resonance spectroscopy analysis of neocortical slices, treatment with gabapentin (100 μm) for 1 h increased the concentration of GABA.45,In vivo  imaging of human occipital cortex with [¹H] nuclear magnetic resonance spectroscopy also showed a modest increase in GABA levels. Interestingly, a neurochemical study showed that gabapentin did not increase GABA concentrations in mouse brains.46 

Compared with the previous in vitro  studies, our results suggest that high concentrations of gabapentin (EC⁵⁰209 μm) are required to increase the level of ambient transmitter. The reasons underlying the apparent low sensitivity of the tonic conductance to enhancement by gabapentin are unknown. Gabapentin likely indirectly activates the tonic conductance by increasing the concentration of agonist at the GABA^Areceptor. The low density of neurons and glia in neuronal cultures compared with brain slices may release lower amounts of transmitter, and therefore, a higher gabapentin concentration is required to detect an effect. Clearly, in vitro  models, such as cultured neurons, do not reflect the complexity of GABA transport and metabolism or GABA^Areceptor activation in vivo . Nevertheless, such models have been used by numerous laboratories to gain fundamental insights into the pharmacologic features of GABA^Areceptors.

Despite an increase in tonic conductance, gabapentin did not alter spontaneous postsynaptic inhibitory transmission. The later finding contrasts with a 37% reduction of evoked inhibitory postsynaptic current by gabapentin (300 μm) recorded in CA1 neurons in brain slices.47The reduction in evoked synaptic currents may be attributed to inhibition of presynaptic voltage-gated calcium channels. A reduction in the concentration of cytosolic calcium in the presynaptic terminal may reduce the synchronized release of vesicular GABA during evoked stimulation. Unlike evoked inhibitory postsynaptic current, tonic conductance is likely generated by GABA that is released by calcium– and soluble N-ethylmaleimide–sensitive factor attachment protein receptor (SNARE)–independent mechanisms.48Tonic conductance is insensitive to calcium channel blockers and botulinum toxin and occurs in neurons from Munc18-1–deficient mice, in which vesicular release is abolished.48Our results are consistent with the notion that gabapentin preferentially increases the nonvesicular release of GABA.49 

Our results showed that a maximally effective concentration of vigabatrin did not occlude the action of gabapentin on tonic conductance. These results can be interpreted in several ways. Vigabatrin and gabapentin may act by different mechanisms. Biochemical studies have shown that gabapentin increases the activity of glutamic acid decarboxylase (GAD), the principal enzyme for GABA synthesis,50and causes sevenfold and twofold increases in the activity of the two isoforms of GAD that occur in neurons, GAD⁶⁵and GAD⁶⁷, respectively.49,50Vigabatrin irreversibly inhibits GABA transaminase, the principal enzyme that degrades cytosolic GABA to succinic acid.51Gabapentin’s inhibition of GABA transaminase is partially reversible. Gabapentin may therefore act at a different regulatory site on the enzyme than vigabatrin.40Alternatively, gabapentin may exert actions independent of GABA transaminase and GAD activity. For example, gabapentin may increase the levels of other amino acids, such as β-alanine and taurine, which activate GABA^Areceptors.52Future studies will investigate these possibilities.

We observed that chronic treatment with vigabatrin did not influence the amplitude or frequency of mIPSCs in cultured hippocampal pyramidal neurons as previously reported.30This finding contrasts with a dramatic reduction in amplitude of mIPSCs and evoked inhibitory transmission in dentate granule neurons recorded in hippocampal slices that were treated with vigabatrin (100–400 μm) for 2–5 h.53The higher concentration of vigabatrin (400 μm) also reduced the frequency of mIPSCs without changing the time course of current decay.53The authors attributed the reduction in current amplitude by vigabatrin, in part, to the accumulation of extracellular GABA and desensitization of GABA^Areceptors. The reduction in frequency of mIPSCs frequency by vigabatrin was attributed to GABA activation of presynaptic GABA^Breceptors because the decrease in frequency was not observed in the presence of the GABA^Breceptor antagonist CGP55845. Reasons underlying differences in vigabatrin sensitivity of postsynaptic currents in hippocampal pyramidal neurons compared with dentate granule cells are unknown but may relate to the duration of drug exposure, differences in the subunit composition of the underlying GABA^Areceptors, or factors that regulate GABA metabolism in different populations of neurons.

Our results showed that the gabapentin EC⁵⁰value for enhancement of the tonic inhibition was approximately 209 μm. The concentration of gabapentin that increased tonic inhibitory conductance in vitro  is considerably higher than the therapeutic concentrations reported for the treatment of pain or seizure control.20,54The high concentrations of gabapentin required to affect the tonic current may be relevant to the toxic effects such as memory impairment. The tonic conductance in hippocampal pyramidal neurons is primarily generated by GABA^Areceptors containing the α⁵subunit.23This subpopulation of GABA receptors has been strongly implicated in learning and memory processes.23,55,56An increase in tonic inhibition is predicted to impair memory performance, as observed with general anesthetics.25,56Therefore, we speculate that increased tonic inhibition in the hippocampus might contribute to the memory impairment reported with gabapentin treatment. Consistent with this notion, mild cognitive deficits were reported in healthy patients after gabapentin treatment compared with drug-free conditions.57Any relevance of gabapentin action on the tonic current to pain control is highly speculative. Therefore, it will be of interest to determine in future studies whether lower concentrations of gabapentin increase the tonic conductance in experimental models where the architecture of neuronal networks remains intact.

The authors thank Marko Katic (Administrative Assistant, Department of Research Design and Biostatistics, Sunnybrook and Women’s College Health Science Centre, Toronto, Ontario, Canada) for advice regarding the data analyses. Werner Sieghart, Ph.D. (Professor, Medical University of Vienna, Vienna, Austria), generously provided the α⁵rabbit polyclonal antibody.