Abstract
Gabapentin (Neurontin, Pfizer Global R & D) is a novel anticonvulsant, antihyperalgesic, and antinociceptive agent with a poorly understood mechanism of action. In this study, we show that gabapentin (EC50 2 μM) inhibited up to 70 to 80% of the total K+-evoked Ca2+ influx via voltage-dependent calcium channels (VD-CCs) in a mouse pituitary intermediate melanotrope clonal mIL-tsA58 (mIL) cell line. mIL cells endogenously express only γ-aminobutyric acid type B (GABAB) gb1a-gb2 receptors. Moreover, activity of the agonist gabapentin was dose dependently and completely blocked with the GABAB antagonist CGP55845 and was nearly identical to the prototypic GABAB agonist baclofen in both extent and potency. Antisense knockdown of gb1a also completely blocked gabapentin activity, while gb1b antisense and control oligonucleotides had no effect, indicating that gabapentin inhibition of membrane Ca2+ mobilization in mIL cells was dependent on a functional GABAB (gb1a-gb2) heterodimer receptor. In addition, during combined whole cell recording and multiphoton Ca2+ imaging in hippocampal neurons in situ, gabapentin significantly inhibited in a dose-dependent manner subthreshold soma depolarizations and Ca2+ responses evoked by somatic current injection. Furthermore, gabapentin almost completely blocked Ca2+ action potentials and Ca2+ responses elicited by suprathreshold current injection. However, larger current injection overcame this inhibition of Ca2+ action potentials suggesting that gabapentin did not predominantly affect L-type Ca2+ channels. The depressant effect of gabapentin on Ca2+ responses was coupled to the activation of neuronal GABAB receptors since they were blocked by CGP55845, and baclofen produced similar effects. Thus gabapentin activation of neuronal GABAB gb1a-gb2 receptors negatively coupled to VD-CCs can be a potentially important therapeutic mechanism of action of gabapentin that may be linked to inhibition of neurotransmitter release in some systems.
Gabapentin [Neurontin, 1-(aminomethyl)cyclohexaneacetic acid, Pfizer Global R & D] was developed as a brain penetrant 3-alkyl-substituted analog of γ-amino-butyric acid (GABA) (reviewed by Bryans and Wustrow, 1999). Gabapentin is approved for clinical use in the treatment of refractory partial seizures and secondary generalized tonic-clonic seizures and is being investigated as treatment for a number of disorders including bipolar, social phobias, neuropathic pain, dental pain, osteoarthritis, and migraine. Gabapentin has been reported to bind with nanomolar affinity to the auxiliary α2δ subunit of voltage-dependent calcium channels (VD-CCs) (Gee et al., 1996). However, no direct functional correlation to this binding has been reported to date, and it is unknown whether this accounts for the anticonvulsant, antihyperalgesic, and antinociceptive actions of gabapentin (Taylor et al., 1998).
The principal physiological role of GABA in the neural axis is synaptic inhibition. Ionotropic GABAA multisubunit chloride channel receptors mediate the fast synaptic inhibitory actions of GABA, whereas metabotropic GABAB G protein-coupled receptors mediate the slower, longer lasting synaptic inhibitory actions implicated in hippocampal long-term potentiation, slow-wave sleep, absence epilepsy, muscle relaxation, and antinociception (see Bowery and Enna, 2000 and references therein). Recent studies have suggested that the functional and high-affinity agonist binding neuronal GABAB receptor is a heterodimer of individually inactive gb1 and gb2 seven-transmembrane spanning subunits (reviewed by Bowery and Enna, 2000). Three molecularly and pharmacologically distinct human GABAB receptor subtypes termed gb1a-gb2, gb1b-gb2, and gb1c-gb2 have been identified that could account at least in part for the diverse biological functions of GABAB receptors (Ng et al., 2001). Many of the physiological roles of GABAB receptors can be attributed to the modulation of P/Q- (α2δ, β1, α1A subunits) and N-type (α2δ, β1, α1B subunits) VD-CCs by presynaptic receptors and modulation of inwardly rectifying K+ channels (GIRKs) by postsynaptic GABAB receptors (Bowery and Enna, 2000 and references therein). GABABreceptor regulation of VD-CC function is thought to be mediated by G protein βγ subunits via a membrane-delimited mechanism (Herlitze et al., 1996; Ikeda, 1996) resulting in the inhibition of membrane Ca2+ conductance and a decrease in neurotransmitter release (Doze et al., 1995; Wu and Saggau, 1997). Activation of presynaptic GABAB receptors negatively coupled to VD-CCs is likely the mechanism underlying the antinociceptive effects of GABA and the prototypic nonselective GABAB agonist baclofen, which have been reported to inhibit the release of pain transmitters such as calcitonin gene-related peptide and substance P in spinal cord slices (Malcangio and Bowery, 1993, 1996). Baclofen has also been reported to be efficacious when given intrathecally for the treatment of central pain following stroke or spinal cord injury (Loubser and Akman, 1996), however its wider clinical use has been limited because doses (p.o.) needed for efficacy are associated with flaccidity and hypotonia. Compelling evidence that GABAB receptors are attractive therapeutic targets has been lacking because of the absence of efficacious and selective GABAB ligands with few side effects.
Gabapentin has been reported to inhibit K+-evoked Ca2+ rises in neocortical synaptosomes via inhibition of VD-CCs (Fink et al., 2000) and reduce K+-evoked glutamate release from neocortical and hippocampal slices (Dooley et al., 2000). Moreover, gabapentin has been reported to inhibit excitatory neurotransmitter release in the spinal cord dorsal horn (Patel et al., 2000; Shimoyama et al., 2000). We have recently reported that gabapentin is a selective agonist at the recombinant gb1a-gb2 heterodimer and neuronal GABAB receptor coupled to GIRKs with no partial agonist or antagonist activity at gb1b-gb2 or gb1c-gb2 subtypes (Ng et al., 2001). This led us to the hypothesis that the inhibitory pharmacological effects of gabapentin on excitatory neurotransmitter release in the studies referenced above were attributed to selective activation of neuronal GABAB receptors coupled to VD-CCs. In this study, we show for the first time that gabapentin is an agonist at endogenously expressed GABABgb1a-gb2 heteromers coupled to inhibition of VD-CCs in mouse pituitary intermediate melanotrope clonal mIL-tsA58 (mIL) cells and hippocampal neurons in situ. We propose that gabapentin activation of neuronal GABAB gb1a-gb2 heteromers and inhibition of voltage-dependent calcium channels may represent a novel mechanism accounting for the anticonvulsant, antihyperalgesic, and antinociceptive properties of gabapentin.
Experimental Procedures
Materials.
(R)-Baclofen and CGP55845 were purchased from Sigma (St. Louis, MO) and Tocris Cookson (Ballwin, MO), respectively. Gabapentin was obtained commercially (Sigma), stored at −80°C, and freshly prepared and used immediately in the functional assays. Indo-1/AM, indo-1 pentapotassium salt, carboxy SNARF-1/AM, carboxy SNARF-1, Pluronic F-127, and Ca2+calibration kits were purchased from Molecular Probes, Inc. (Eugene, OR). Pertussis toxin, penicillin/streptomycin, and dimethyl sulfoxide were obtained from Sigma. The DMEM and the Gibco BRL trypsin-free buffer were purchased from Life Technologies (Grand Island, NY). Other chemicals and reagents were purchased from Fisher Scientific (St. Louis, MO).
mIL-tsA58 Cells and Culture Conditions.
The mIL-tsA58 cells were isolated from a mouse intermediate lobe tumor (M. J. Low, manuscript in preparation). Briefly, a strain of transgenic mice was generated that developed pituitary tumors because of the melanotrope-specific expression of a pro-opiomelanocortin promoter-simian virus 40 large T antigen (temperature-sensitive A58 mutant) fusion gene. The phenotype of these mice was similar to those described previously (Low et al., 1993) despite the substitution of the tsA58 mutant T antigen for wild type. The mIL-tsA58 cells were isolated from a single tumor using procedures analogous to those reported for another melanotrope cell line (Hnasko et al., 1997). They express the POMC gene and dopamine D2 receptors. Growth rate and morphology of the cells were similar at 33°C, the permissive temperature for tsA58, and at 37°C. These cells display a normal mouse karyotype after more than 60 passages. They either grow as free floating spheres of tightly associated cells or can be coaxed to adhere to plastic, although they remain clustered under this condition as well. Few individual cells are found in cultures, and the clusters are difficult to dissociate enzymatically without destroying the cells.
The mIL cells were maintained in DMEM supplemented with 10% horse serum, 2.5% fetal bovine serum, 100 U/ml penicillin/streptomycin (10,000 units of penicillin and 10 mg of streptomycin per ml), 2 mM glutamine, and 0.1 mM nonglutamine essential amino acids under 5% CO2 at 37°C.
Measurement of Intracellular Calcium [Ca2+]i Kinetics.
[Ca2+]i and intracellular pH (pHi) were measured simultaneously using a custom-built ultra low light multi-imaging video microscope as described previously (Beatty et al., 1993; Morris et al., 1994). Cells grown on number 00 coverslips (Corning, Corning, NY) were simultaneously loaded with 5 μM indo-1/AM and 5 μM SNARF-1/AM (cell permeant acetoxymethyl (AM) esters), in DMEM, 12.5% dimethyl sulfoxide, and 0.04% (w/w) Pluronic F-127 for 30 min at 37°C in a humidified incubator under 10% CO2. After incubation, the cells were washed and left in complete medium at 37°C under 10% CO2 for a 30-min recovery period to allow the esterase to cleave the dyes to their active, impermeant forms. Cells were examined within 90 min following the recovery period.
Coverslips with dye-loaded cells were placed in 1.0 ml of standard balanced salt solution (138 mM NaCl, 2 mM KCl, 2 mM MgCl2, 10 mM HEPES, 5.5 mM glucose, 2 mM CaCl2, and 50 μM EGTA, pH 7.4) in a microscope stage perfusion chamber maintained at 37°C. Phase contrast and fluorescence images of the cells were obtained simultaneously at 405, 475, 575, and 640 nm. Using these images as a guide, up to eight regions of interest, each representing a single cell, were defined for Ca2+ (405 and 475 nm images), along with eight corresponding regions for pH (575 and 640 nm images). After 30 to 60 s of video recording of the [Ca2+]i and pHi baseline activity, 1.0 ml of an iso-osmotic, high K+ depolarizing solution (10 mM NaCl, 130 mM KCl, 2 mM MgCl2, 10 mM HEPES, 5.5 mM glucose, and 2 mM CaCl2, pH 7.4) was added for a final extracellular [K+] of 66 mM. During recordings, the addition of other chemicals or drugs was made directly to the bath in 100-fold excess to achieve the final concentrations indicated. In all cases, osmolarity changed less than 1% by the addition of these agents. Ionomycin (1 μM), a Ca2+ ionophore, was added at the end of some experiments to ensure viability of the cells as well as the responsiveness of the fluorescent dyes. In certain cases, baclofen, gabapentin, or the GABABreceptor agonist CGP55845 was added 5 min prior to, or pertussis toxin 10 to 14 h prior to, the exposure of the cells to K+ and the measurement of [Ca2+]i and pHi.
The effects of GABAB agonists and antagonists were tested as follows: 5 min before the start of the baseline recording, drugs were added to the bath to the final concentrations indicated. If both a GABAB agonist and an antagonist were being tested, then the antagonist was added first. After baseline recording, the cells were depolarized with high K+ as described above.
Antisense Oligodeoxynucleotide Synthesis and Administration.
Two antisense deoxynucleotides (ADNs) directed against either GABABR1a or GABABR1b isoforms of the receptor were designed and synthesized. The gb1a probe is antisense to bases 5′-CAC CAG CAG CAG CAG CAG-3′ of GABABR1a (bases 4–22, GenBank accession number AJ102185) and the gb1b probe is antisense to bases 5′-ACA GGG TCC CCC CGG GCC-3′ of GABABR1b (bases 4–22, GenBank accession number AJ02186). Using the National Institutes of Health BLAST search engine (http://www.ncbi.nlm.nih.gov/BLAST/), these antisense sequences showed extensive overlaps with GABABR1a and GABABR1b receptor sequences from other species but had no significant overlaps with sequences of other cDNAs as of December, 2000. A missense probe containing the same nucleotide base content as the gb1a ADN, but with bases randomly assigned, was used as a control. This missense probe, 5′-CCA GCA GAC ACG CAG CAG-3′ has eight overlaps with the gb1a antisense probe and no known complementarity with other sequences. All nucleotide sequences were synthesized by Integrated DNA Technologies (Coralville, IA) as phosphorothioated derivatives.
For ADN experiments, the mIL cells were exposed to nucleotide for a total of 4 days and then tested for Ca2+ channel activity by fluorescence video microscopy. The cells were first cultured in T25 flasks for 1 to 2 days. The cultures were then placed in 5 ml of serum-free medium and treated with 5 μl of 1.0 mM of the test ADN or missense oligodeoxynucleotide solution (10 μM final concentration). Following a 2-h incubation at 37°C under 5% CO2, 500 μl of fetal horse serum and 125 μl of fetal bovine serum were added to each flask. Five microliters of nucleotide solution were added to each flask at 2 and 3 days of culture. Cells were harvested on day 3 in serum-free medium and then plated onto cover slips in 12-well plates. The cells were serum-deprived for 30 min at 37°C under 5% CO2to facilitate adherence to the cover slip, then nucleotide was added to 10 μM final concentration. Serum was added after an additional 30 min, and Ca2+ channel activity was tested 24 h later.
Data Analysis.
Results were analyzed either in real time or from video tape recordings. Twenty-five consecutive frames were averaged in real time, then Ca2+ and pH ratio images of the microscope field (uncorrected for background or shading error) displayed on the computer monitor display at one image per second. At the same time, the integrated gray levels of up to eight regions of interest were extracted from the 25-frame average image, and the data were stored on an ASCII file for further analysis. In addition, the uncorrected ratio values for Ca2+and pH were plotted on the computer monitor screen, permitting immediate evaluation of cell viability and the effects of treatments on [Ca2+]i and pHi. The experiments were also recorded on ¾-inch U-matic video tape as a backup and to allow analysis of other cells in the video field since the same data display and data extraction procedures could be applied offline. Correction of [Ca2+]i using the prevailing intracellular pH, standardization, data reduction analysis, and statistical methods has been previously described (Morris et al., 1994).
We have previously shown that K+ depolarization of mIL cells results in a two-phase increase in [Ca2+]i. The fast phase, which peaks within about 10 s, is due to influx through high voltage-activated Ca2+ channels. A second slower phase, with a lower amplitude that peaks much later (30–60 s), is due to release from intracellular stores (Chronwall et al., 2001). Therefore, the change in Ca2+ level due to membrane channel activity was measured as: percent change in [Ca2+]i = (Max Ca2+ − Min Ca2+)/(Min Ca2+ × 100), where Max Ca2+ = maximal [Ca2+]i value achieved within 10 s following depolarization and Min Ca2+ = initial resting [Ca2+]i.
Electrophysiology and Calcium Imaging of Hippocampal Neurons in Brain Slices.
Experiments were performed on CA1 pyramidal neurons in 300-μm-thick hippocampal slices from 25- to 28-day-old male Sprague-Dawley rats (Nurse and Lacaille, 1999). Slices were allowed to recover for at least 1 h before use. The recording chamber was continuously perfused with oxygenated (95% O2, 5% CO2) artificial cerebrospinal fluid containing 124 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl2, 26 mM NaHCO3, 1.25 mM NaH2PO4, 2 mM MgSO4, and 10 mM glucose, pH 7.35 to 7.4. Experiments were conducted in the presence of 0.5 μM tetrodotoxin (TTX) to block voltage-dependent Na+ channels. To block K+ channels in the recorded neuron, patch pipettes (4–8 MΩ) were filled with a cesium-based solution containing 140 mM CsMeSO3, 1 mM MgCl2, 5 mM NaCl, 2 mM ATP, 0.4 mM GTP, 10 mM HEPES, and 100 μM of the Ca2+ indicator Oregon Green BAPTA-I (Molecular Probes, Inc.) titrated with CsOH to pH 7.25 to 7.28. Combined whole cell current clamp recordings and confocal calcium imaging were performed from CA1 pyramidal neurons using an Axopatch 200B amplifier (Axon Instrument, Foster City, CA) and a multiphoton confocal laser scanning microscope LSM 510 (Carl Zeiss, Kirkland, QC) equipped with a 40× long-range water-immersion objective (numerical aperture 0.8).
After obtaining the whole cell configuration, at least 20 min were allowed for intracellular diffusion of the fluorophore. For two-photon confocal imaging, a tunable mode-locked Ti:Sapphire laser at 780 nm was used (5W Verdi argon ion laser and Mira 900, Coherent, Santa Clara, CA). Emission was detected through a long-pass filter (cut-off 505 nm) and recorded to a PC using the LSM 510 software (Carl Zeiss). The confocal aperture was opened fully. Linescans were taken from the soma at a rate of 3.8 ms per line for a total scan time of 12 s (Fig. 5A1). Pyramidal cells were activated by somatic depolarizing current injections via the recording pipette (Fig. 5A2). Series resistance was monitored and compensated throughout the experiments. For linescans, the fluorescence intensity (Fline) was averaged over the region of interest of the line across the cell soma. Changes in fluorescence were calculated for each line relative to the averaged baseline fluorescence prior to stimulation (Frest) and expressed as: %ΔF/F = [(Fline − Frest)/Frest] × 100. The values were then processed with a low-pass digital filter to remove fast transients (Igor Pro, Wavemetrics, Lake Oswego, OR), and the peak calcium response was determined for each linescan. Linescans and electrophysiological recordings were initiated manually. To compensate for small variations in the start time of the linescans, electrophysiological and Ca2+ responses were temporally aligned by eye (Fig. 5A2).
Statistical Analysis.
The level of significance for differences between means was measured by Fisher's test or analysis of variance followed by the Bonferroni post-test (InStat, GraphPad Software, Inc., San Diego, CA).
Results
Gabapentin Inhibition of Calcium Mobilization via Selective Activation of Endogenously Expressed GABAB gb1-gb2 Heterodimers in mIL Cell Lines.
It is well known that pituitary intermediate lobe melanotropes release adrenocorticotropic hormone, α-melanocyte-stimulating hormone, and β-endorphin by exocytosis. This process requires Ca2+ (Thomas et al., 1990) and is inhibited by GABAB agonists (Taraskevich and Douglas, 1990; Morris et al., 1998). We have recently reported that immortalized pituitary melanotrope mIL cells express endogenous functional GABAB gb1a-gb2 but not gb1b-gb2 heteromers coupled negatively to VD-CCs (Chronwall et al., 2001). Since presynaptic GABAB receptors inhibit VD-CCs, mIL cells represented a suitable model system to study the pharmacology of the wild-type brain gb1a-gb2 subtype and to test whether gabapentin inhibition of VD-CCs in isolated neurons (Stefani et al., 1998; Dooley et al., 2000; Fink et al., 2000) was mediated by selective activation of GABAB receptors.
Figure 1A shows typical changes in intracellular Ca2+ levels of individual mIL cells in response to depolarization by high extracellular K+ concentration. There is a sharp and major increase in [Ca2+] attributed to the depolarization and activation of the VD-CCs followed by a slower second peak or shoulder due to release from intracellular stores consistent with previous findings in melanotropes (Morris et al., 1998; Chronwall et al., 2001). Figure 1B shows that the prototypical nonselective GABAB agonist baclofen (1 μM) reduces the primary Ca2+ response. Figure 1C shows that baclofen effects are reversed by addition of the GABAB antagonist CPG55845 (3 μM) (compare Fig.1, B and C). CPG55845 has no significant effect on the response to depolarization (compare Fig. 1, A and C). Figure 1D shows that 1 μM gabapentin action is nearly identical to 1 μM baclofen (compare to Fig. 1B). The gabapentin effect is also completely blocked by 3 μM CGP55845 (compare Fig. 1, D and E). Figure2 shows the dose-response curve for gabapentin inhibition of K+-evoked calcium mobilization with an EC50 value of 2 μM and 70 to 80% of the total channel activity inhibited at 1 mM. Figure3 shows that the VD-CC-dependent rise in intracellular Ca2+ is blocked by 1 μM gabapentin and that the gabapentin activity could be blocked completely and in a dose-dependent manner (30 nM–3 μM) by the GABAB antagonist CGP55845. Gabapentin inhibition of calcium mobilization was similar in magnitude (70–80%) and potency to baclofen (EC50 1 μM) suggesting that gabapentin actions were mediated by binding to the native gb1a-gb2 heteromer.
To confirm that the inhibition of calcium mobilization was the result of selective activation of the endogenous GABABg1a-gb2 heteromer and not an activity of gabapentin on the endogenous VD-CC (e.g., binding to the α2δ subunit), we tested whether ADN treatment could selectively abolish the effect of gabapentin on K+-evoked increase in intracellular Ca2+.
mIL cells were treated for 4 days with either gb1a or gb1b ADNs or a gb1a missense targeting sequence as reported under Experimental Procedures. These conditions were identical to the conditions used in previous studies in which we demonstrated that selective knockdown of either gb1 or gb2 subunits but not missense control antisense led to a selective reduction in protein expression of the targeted gene product and, in both cases, a complete loss of GABAB receptor-initiated reduction in VD-CC function (Chronwall et al., 2001). In this series of experiments, 10 and 30 μM gabapentin mediated 70 to 80% inhibition of the total K+-evoked VD-CC activity (Fig.4, compare A to B and C). Antisense knockdown of gb1a in mIL was accompanied by a complete block of 10 and 30 μM gabapentin activity (Fig. 4, compare columns B and C to E and F). In contrast, treatment with gb1b ADN or the gb1a missense nucleotide was without effect on both 10 and 30 μM concentrations of gabapentin (compare columns B and C to H and I, K and L, respectively). Taken together, this demonstrated that gabapentin inhibition of calcium mobilization in mIL cells was due to activation of gabapentin-sensitive endogenously expressed gb1a-gb2 heteromers.
Gabapentin Inhibition of VD-CCs via Neuronal GABABReceptors in Rat Hippocampal Neurons in Situ.
To confirm stimulation by gabapentin at neuronal GABABreceptors coupled to VD-CCs in situ, we combined whole cell current clamp recordings of pyramidal cells with multiphoton confocal calcium imaging (Fig. 5, A1 and A2) and examined the effects of gabapentin on calcium responses evoked by somatic current injections in CA1 pyramidal neurons of rat hippocampal slices. In the presence of TTX and a K+ channel blocker (intracellular cesium), positive current pulses were applied to the pyramidal cell soma via the recording electrode, and the evoked calcium responses were recorded electrophysiologically (membrane potential) and optically (fluorescence) (Fig. 5A2). The amplitude of the current pulse was varied to elicit Ca2+ responses by subthreshold stimulation (Fig. 5, B1 and C1) and Ca2+ spikes by suprathreshold stimulations (Fig. 5, B2 and C2). Subthreshold current injections induced Ca2+ responses of small amplitude and short duration (Fig. 5, B1 and C1), whereas suprathreshold current injections triggered larger and longer lasting Ca2+responses (Fig. 5, B2 and C2) at the cell soma. In our experimental conditions, these responses induced by somatic current injection were solely mediated by Ca2+ since they were totally blocked in Ca2+-free artificial cerebrospinal fluid (n = 2 cells, data not shown).
For a given subthreshold current injection, the membrane depolarization and its associated Ca2+ response (traces 1 in Fig. 5B1) were significantly depressed in the presence of 1 mM gabapentin (traces 2 in Fig. 5B1). These effects of gabapentin were reversible (data not shown). When the current pulse amplitude was adjusted to elicit a Ca2+ spike in the pyramidal cell (traces 1 in Fig. 5B2), the same current pulse in the presence of gabapentin (1 mM) failed to induce a Ca2+ spike (traces 2 in Fig. 5B2) and solely evoked a subthreshold depolarization and a very small Ca2+ response. However, cells were still capable of producing Ca2+ spikes in the presence of gabapentin since larger current injections induced a Ca2+ spike and its associated large and long-lasting Ca2+ response at the cell soma (traces 3 in Fig. 5B2). Both the peak amplitude of the Ca2+ spike and Ca2+response elicited by the larger stimulation in the presence of gabapentin were not significantly different from those in control conditions (76.3 ± 4.7 mV and 165.3 ± 27% ΔF/F in gabapentin versus 82.9 ± 3 mV and 218.7 ± 24% ΔF/F in control, respectively, n = 4, Fig. 5B2).
As observed for gabapentin, baclofen (40 μM) depressed in a reversible manner the membrane depolarizations and Ca2+ responses induced by both sub- and suprathreshold somatic current injections (Fig. 5C). As for gabapentin, increasing the somatic current injection in the presence of baclofen restored both Ca2+ spikes and Ca2+ responses to levels similar to those in control (71.9 ± 5.6 mV and 167.9 ± 61.4% ΔF/F in baclofen versus 76.9 ± 2.5 mV and 254.1 ± 28.6% ΔF/F in control conditions; n = 3, Figs. 5C2 and 6D).
The inhibition of Ca2+ responses by gabapentin was dose-dependent. The graphs on Fig. 6, A and B, show the effects of different concentrations of gabapentin (100 μM to 1 mM) on membrane depolarizations and Ca2+ responses evoked by sub- and suprathreshold current injections. To test if the inhibitory action of gabapentin on Ca2+ responses was mediated by activation of GABAB receptors, we investigated the effect of the GABAB antagonist CGP55845 on gabapentin actions. Gabapentin (2 mM) significantly reduced membrane depolarizations and Ca2+ responses evoked by both sub- and suprathreshold soma current injections (Fig.7, C and D). This depressant effect of gabapentin was blocked in the presence of 4 μM CGP55845 for both sub- (Fig. 7, A1 and C) and suprathreshold (Fig. 7, A2 and D) current injections. Similarly, the inhibition of Ca2+responses by baclofen was also blocked by CGP55845 (Fig. 7, B–D). These results indicate that gabapentin negatively couples to VD-CCs via GABAB receptors in hippocampal pyramidal neurons in situ. Taken together with the selective activation demonstrated for gabapentin at the endogenous brain GABAB receptor in mIL cells, our results suggest that one possible mechanism by which gabapentin exerts its central nervous system therapeutic actions is by a selective activation of neuronal gb1a-gb2 GABABreceptor heterodimers coupled to VD-CCs.
Discussion
It has been proposed that the auxiliary α2δ subunit of voltage-dependent calcium channels may be a molecular target for gabapentin, conceivably altering VD-CC function, but direct experimental proof is still lacking to link this with the anticonvulsant, antihyperalgesic, and antinociceptive actions of this drug (Gee et al., 1996; Taylor et al., 1998). Gabapentin has been reported to have no effect on VD-CCs in cultured rodent neurons (Rock et al., 1993) and in acutely dissociated human dentate gyrus granule cells from patients with temporal lobe epilepsy (Schumacher et al., 1998). Yet gabapentin has been reported to inhibit predominantly L-type calcium currents in isolated rat neocortical, striatal, and pallidal neurons (Stefani et al., 1998). More recently, gabapentin has been found to inhibit K+-evoked glutamate release from rat neocortical and hippocampal slices (Dooley et al., 2000; Fink et al., 2000). However, the mechanisms underlying these gabapentin actions were not elucidated.
We have reported recently that gabapentin is a selective agonist for the recombinant and neuronal GABAB gb1a-gb2 heteromer subtype coupled to GIRKs and that it is not a partial agonist and does not block GABA activity at gb1b-gb2 and gb1c-gb2 heteromers (Ng et al., 2001). Selective gabapentin activation of GABAB receptors negatively coupled to VD-CCs may account for gabapentin actions in the aforementioned K+-evoked Ca2+-dependent responses, and this notion is also consistent with the depressant action of gabapentin on voltage-sensitive calcium currents in some central neurons (Stefani et al., 1998). Indeed we show herein that gabapentin is an agonist at brain GABAB gb1a-gb2 heteromer receptors endogenously expressed in mIL cells mediating robust dose-dependent inhibition, similar to that of the prototypical GABAB agonist baclofen, of VD-CC function. This was attributed to selective activity at the GABABreceptor since it could be blocked with GABABantagonists and following selective antisense knockdown of the gb1 subunit in agreement with the selective activity reported at the recombinant GABAB receptors (Ng et al., 2001). Recombinant GABAB heteromers have been also shown to couple to calcium channels in cultured NG108-15 cells and sympathetic neurons (Easter and Spruce, 2000; Filippov et al., 2000), and activation of native receptors in rat pituitary melanotropes and dorsal root sensory neurons leads to inhibition of calcium currents (Morris et al., 1998; Chronwall et al., 2001; Hand et al., 2000). Our results also indicate that, in hippocampal neurons in situ, gabapentin activates GABAB receptors negatively coupled to N- and/or P/Q-type VD-CCs. But our data do not support a predominant action of gabapentin on L-type Ca2+ channels since gabapentin inhibited subthreshold Ca2+responses but did not prevent Ca2+ action potentials in the present experiments. Gabapentin actions on neuronal GABAB receptors coupled to VD-CCs is consistent with the previously reported actions of baclofen in hippocampal neurons (Scholz and Miller, 1991; Lambert and Wilson, 1996). These studies underscore that VD-CCs represent a major and physiologically important effector for neuronal GABAB receptors.
Our results further indicate that gabapentin may have multiple anticonvulsant actions linked to GABAB receptors. In addition to its selective activation of gb1a-gb2 GABAB receptors coupled to GIRKs that produce postsynaptic hyperpolarization (Ng et al. 2001), gabapentin may inhibit Ca2+ influx during burst discharges or seizures via its activation of postsynaptic gb1a-gb2 receptors negatively coupled to VD-CCs. It is interesting to note that gabapentin actions on hippocampal neurons are therefore dictated not only by its selective activity at the gb1a-gb2 heteromer subtype but also by the cellular domain where these receptors are found in the cell. Gabapentin-sensitive GABAB receptors present in the soma and dendritic regions couple to VD-CCs (Fig. 3) and GIRKs (Ng et al., 2001). In contrast, the GABAB gb1b/c-gb2 subtypes, which are likely present in glutamate and GABA axon terminals of hippocampal neurons and also are negatively coupled to VD-CCs, are insensitive to gabapentin. This is in agreement with the subtype-selective agonist activity defined using recombinant receptors and the lack of presynaptic effect of gabapentin on synaptic transmission in hippocampus (Ng et al., 2001).
Gabapentin has been recently reported to depress excitatory amino acid neurotransmission in spinal cord dorsal horn (Patel et al., 2000;Shimoyama et al., 2000), and the effect of the agonist gabapentin on GABAB receptors coupled to VD-CCs could account for these effects since a well established physiological role of presynaptic neuronal GABAB receptors is inhibition of P/Q- and N-type VD-CCs and transmitter release (Menon-Johansson et al., 1993; Wu and Saggau, 1997; Bowery and Enna, 2000). This conclusion is also consistent with the anatomical localization of the gb1a-gb2 heteromer to some presynaptic elements in the neural axis (Benke et al., 1999; Billinton et al., 1999; Towers et al., 2000). GABAB distribution studies in the lumbar spinal cord and dorsal root ganglia showed that the gb1a mRNA is the predominant species (accounting for ∼90%) of the total gb1 mRNA in the afferent fiber cell body. This suggests that gb1a subunits together with gb2 subunits, which exhibit equivalent density to gb1a, comprise presynaptic GABAB receptors on primary afferent terminals (Towers et al., 2000). Indeed in this report, immunocytochemical analysis showed denser labeling of gb1a in the superficial dorsal horn and presence in neuropil, whereas gb1b was more associated with cell bodies in this region.
The predominant expression of GABAB gb1a-gb2 receptors in the superficial laminae where nociceptive primary afferent fibers terminate, together with studies that suggest the antinociceptive effects of baclofen (Henry, 1982; Hammond and Drower, 1984; Sawynok and Dickson, 1985) and gabapentin (Xiao and Bennet, 1997; Patel et al., 2000; Shimoyama et al., 2000) are mediated presynaptically, lead us to suggest that, at least in part, the antihyperalgesic, antiallodynic, and antinociceptive effects of gabapentin can be attributed to selective activation of presynaptic GABAB gb1a-gb2 receptors coupled to VD-CCs in the spinal cord dorsal horn.
Footnotes
-
↵1 Co-first authors.
-
S.B. was supported by a postdoctoral fellowship from the Savoy Foundation and a Cordeau/Servier fellowship from the Center de Recherche en Sciences Neurologiques, Université de Montréal. The work in the laboratory of S.J.M. was supported by the Loeb Charitable Foundation, National Science Foundation Grant IBN 9907571, and University of Missouri Research Board (B.M.C.). The work in the laboratory of J.-C. L. was supported by the Canadian Institutes of Health Research, the Fonds de la Recherche en Santé du Québec (FRSQ), a Research Center grant from the Fonds pour la Formation de Chercheurs et l'Aide àla Recherche (FCAR) to the Groupe de Recherche sur le Système Nerveux Central (GRSNC) and an Équipe de Recherche grant from the FCAR.
- Abbreviations:
- GABA
- γ-amino-butyric acid
- gb1
- GABABR1 receptor subunit
- gb2
- GABABR2 receptor subunit
- VD-CC
- voltage-dependent calcium channel
- mIL
- mIL-tsA58
- GIRK
- inwardly rectifying K+ channel
- AM
- acetoxymethyl
- DMEM
- Dulbecco's modified Eagle's medium
- pHi
- intracellular pH
- [Ca2+]i
- intracellular calcium
- ADN
- antisense deoxynucleotide
- TTX
- tetrodotoxin
- Received January 19, 2001.
- Accepted March 20, 2001.
- The American Society for Pharmacology and Experimental Therapeutics