GABAergic activation of an inwardly rectifying K⁺ current in mouse cerebellar Purkinje cells

Abstract

Cerebellar Purkinje cells integrate motor information conveyed by excitatory synaptic inputs from parallel and climbing fibres. Purkinje cells abundantly express B-type G-protein-coupled γ-aminobutyric acid receptors (GABA_BR) that are assumed to mediate major responses, including postsynaptic modulation of the synaptic inputs. However, the identity and function of effectors operated by GABA_BR are not fully elucidated. Here we characterized an inwardly rectifying current activated by baclofen (I_bacl), a GABA_BR agonist, in cultured mouse Purkinje cells using a ruptured-patch whole-cell technique. I_bacl is operated by GABA_BR via G_i/o-proteins, as it is not inducible in pertussis-toxin-pretreated cells. I_bacl is carried by K⁺ because its reversal potential shifts with the equilibrium potential of K⁺. I_bacl is blocked by 10⁻³m Ba²⁺ or Cs⁺, and 10⁻⁸m tertiapin-Q. Upon the onset and offset of a hyperpolarizing step, I_bacl is activated and deactivated, respectively, with double-exponential time courses (time constants, <1 ms and 30–80 ms). Based on similarities in the above properties, G-protein-coupled inwardly rectifying K⁺ (GIRK) channels are thought to be responsible for I_bacl. Perforated-patch recordings from cultured Purkinje cells demonstrate that I_bacl hyperpolarizes the resting potential and the peak level achieved by glutamate-evoked potentials initiated in the dendrites. Moreover, cell-attached recordings from Purkinje cells in cerebellar slices demonstrate that I_bacl impedes spontaneous firing. Therefore, I_bacl may reduce the postsynaptic and intrinsic excitability of Purkinje cells under physiological conditions. These findings give a new insight into the role of GABA_BR signalling in cerebellar information processing.

The cerebellar cortex plays central roles in motor coordination and learning (Thach et al. 1992; Llinas & Welsh, 1993; Kawato et al. 2003; Ohyama et al. 2003). Purkinje cells are the sole output neurones of the cerebellar cortex, and they integrate motor information conveyed by excitatory synaptic inputs from parallel and climbing fibres (Palay & Chan-Palay, 1974; Llinas et al. 2003). Purkinje cells express a very high density of B-type G_i/o-protein-coupled γ-aminobutyric acid receptor (GABA_BR) (Jones et al. 1998; Kaupmann et al. 1998; Kuner et al. 1999). GABA_BR in Purkinje cells may be activated by not only GABA synaptically released from the innervating interneurones (Llinas et al. 2003), but also by GABA spillover from synapses between neighbouring neurones (Hirono et al. 2001). Thus, GABA_BR is assumed to mediate major cellular functions in Purkinje cells. A previous study (Kawaguchi & Hirano, 2000) reported that GABA_BR activation leads to suppression of rebound potentiation of inhibitory synaptic inputs to Purkinje cells (Kano et al. 1992). Moreover, GABA_BR is thought to play a role in the postsynaptic modulation of the excitatory synaptic inputs because the peak subcellular density of GABA_BR is found at the excitatory synapses (Fritschy et al. 1999; Ige et al. 2000; Kulik et al. 2002). For example, GABA_BR activation may lead to the enhancement of the postsynaptic responses mediated by type-1 metabotropic glutamate receptors (Hirono et al. 2001). However, the identity and physiological actions of effectors operated by GABA_BR are not fully elucidated in Purkinje cells.

In this study, we explored a GABA_BR-operated ion current presumably produced by G-protein-coupled inwardly rectifying K⁺ (GIRK) channels (Bichet et al. 2003) in Purkinje cells. GIRK channels are gated by the βγ subunit complex released from activated G_i/o-proteins (Sadja et al. 2003), and they form a major class of GABA_BR-operated effectors in various central neurones (Jan & Jan, 1997). GIRK channels can produce a large inward K⁺ current at membrane potentials (E_m) more negative than the equilibrium potential of K⁺ (E_K, ∼−100 mV), while producing only a small outward K⁺ current at more positive membrane potentials (Bichet et al. 2003). Thus, K⁺ conductance through GIRK channels at physiological membrane potentials is much smaller than the maximum. Nevertheless, this conductance is often capable of hyperpolarizing the resting potential (E_rest) and attenuating excitatory postsynaptic potentials (Jan & Jan, 1997; Mark & Herlitze, 2000; Seeger & Alzheimer, 2001). A functional GIRK channel is a tetramer consisting of various combinations of subunits termed GIRK1–4 (Kir3.1–3.4)(Wischmeyer et al. 1997; Bichet et al. 2003). Some in situ hybridization and immunohistochemical studies detect the mRNAs and proteins of GIRK subunits (Karschin et al. 1996; Iizuka et al. 1997; Lauritzen et al. 1997; Murer et al. 1997) except GIRK1 (Liao et al. 1996; Ponce et al. 1996; Miyashita & Kubo, 1997) in Purkinje cells. In certain heterologous expression systems, GIRK2–4 can form functional homo- and heteromeric channels (Wischmeyer et al. 1997). However, macroscopic currents produced by these channels are much smaller than that of GIRK1-containing channels (Wischmeyer et al. 1997). Therefore, it is important to functionally determine at the whole-cell level whether GABA_BR activation induces a significant inwardly rectifying K⁺ current in Purkinje cells.

We used a ruptured-patch whole-cell voltage-clamp technique to extract an inwardly rectifying current activated by baclofen (I_bacl), a GABA_BR-selective agonist (Bowery, 1993). We characterized the basic properties of I_bacl, including the G_i/o-protein-dependence, carrier ion species, pharmacological profile and kinetics. These results demonstrate that Purkinje cells are equipped with an inwardly rectifying K⁺ current most likely produced by GIRK channels. These results also offer a useful tool for studying G-protein signalling in Purkinje cells because the amplitude of GIRK currents directly reflects the activity of G-proteins (Sadja et al. 2003). Moreover, we assessed the possible physiological contribution of I_bacl to regulation of the postsynaptic and intrinsic excitability of Purkinje cells using a perforated-patch whole-cell technique in cultured cell preparations, and a cell-attached technique in cerebellar slice preparations. These findings give a new insight into the role of GABA_BR signalling in cerebellar information processing.

Methods

Cultured cell preparation

Cerebellar Purkinje cells from C57BL/6 mice were cultured as described elsewhere (Tabata et al. 2000). Briefly, perinatal embryos were removed by Caesarian section from pregnant mice, which were deeply anaesthetized, and killed with diethylether. The embryos were deeply anaesthetized by cooling in chilled phosphate-buffered saline, and then killed by decapitation. The cerebella from these embryos were dissociated with trypsin, and plated onto plastic dishes (diameter, 35 mm; Falcon 3001). The neurones were cultured in an ultra-low-serum, hormone/nutrient-supplemented Dulbecco's modified Eagle's medium for 10 days to 3 weeks. Purkinje cells were identified by their large somata (15–30 μm) and thick primary dendrites.

Cerebellar slice preparation

C57BL/6 mice (4–5 weeks old) were deeply anaesthetized with diethylether, and killed by cervical dislocation. Parasagittal cerebellar slices (250 μm thick) were prepared, using a vibrating slicer (VT-1000S, Leica, Wetzlar, Germany), as described elsewhere (Kano et al. 1995).

Electrophysiology

We measured whole-cell currents from the somata of cultured Purkinje cells using a ruptured-patch voltage-clamp technique (Marty & Neher, 1995) (25–26°C). The pipette solution consisted of (mm): 130 d-gluconate potassium salt, 10 NaCl, 10 Hepes, 0.5 ethyleneglycol-bis-(β-aminoethylether)N,N,N′,N′-tetraacetic acid, 4 Mg-ATP and 0.4 Na₂-GTP; pH was adjusted to 7.3 with KOH or d-gluconic acid; the total concentrations of K⁺ and Mg²⁺ were adjusted to 150.6 and 5.2 mm with KCl and MgCl₂, respectively. The recording chamber (culture dish) was perfused at a rate of 1–2 ml min⁻¹ with a saline whose standard composition was (mm): 116 NaCl, 16 KCl, 1.1 NaH₂PO₄, 23.8 NaHCO₃, 2 CaCl₂, 0.3 MgCl₂, 5.5 d-glucose and 5 Hepes; pH was adjusted to 7.3 with HCl. Voltage-gated Na⁺ channels and ionotropic receptors for glutamate and GABA were always blocked by supplementing the saline with (μm): 0.5 tetrodotoxin (TTX), 1000 kynurenic acid (dissolved with 1 eq. NaOH) and 10 (–)-bicuculline methochloride. Current signals were low-pass filtered at 0.5–1 kHz, and sampled at 2–20 kHz using a voltage-clamp amplifier (EPC 8 or 9/2, HEKA, Lambrecht, Germany) driven by PULSE software (version 8.31 or 8.53, HEKA). Command potentials were corrected for a liquid junction potential between the pipette solution and the saline, and given to the examined cell, employing electronic compensation (by 60%) of the series resistance (typically 5–20 MΩ). The capacitance cancellation circuitry was adjusted to erase the slowest component of the capacitive currents (monitored with 5 mV voltage steps; typically 20–70 pF; Tabata et al. 2000), which is thought to reflect the capacitance of the soma and proximal dendrites (Llano et al. 1991). Upon membrane rupture, we repeatedly monitored the amplitude of current responses at an interval of 30–60 s until it became stable. We then consecutively obtained basal and test records, applying the control and drug-containing saline to the cell, respectively. When a drug had a detectable effect on a current response, we checked the recovery of the response, washing out the drug with the control saline (30–60 s). Only if the effect of the drug was at least partially reversible, the case was adopted for the following analyses.

We measured the somatic E_m of cultured Purkinje cells using a perforated-patch current-clamp technique (Marty & Neher, 1995) (25–26°C). The pipette solution consisted of (mm): 140 d-gluconate potassium salt, 10 NaOH, 10 Hepes and 8 MgCl₂; pH was adjusted to 7.3 with HCl; 0.005 vol. dimethylsulfoxide solution of amphotericin B (0.2 mg μl⁻¹) was added before recordings. The ionic composition of the chamber-perfusing saline was the same as that used for the voltage-clamp measurements, except that the total concentration of K⁺ was reduced to 3 mm. Toxins included in the saline are specified in the corresponding figure legends. Current stimulation and E_m signal acquisition were performed using the fast current-clamp circuitry of the EPC-9/2 amplifier driven by the Pulse software. Signals were low-pass filtered at 5–10 kHz, and sampled at 10–20 kHz. We started recording after the series resistance decreased below 90 MΩ. The measured E_m was corrected for a liquid junction potential between the pipette solution and the saline.

We monitored the spontaneous firing of Purkinje cells in cerebellar slices by measuring transient inward currents corresponding to action potentials under voltage clamp (we hereafter term this current a spike current) in a cell-attached mode (Hamill et al. 1981) (24–26°C). The pipette solution consisted of (mm): 125 NaCl, 2.5 KCl, 10 Hepes, 2 CaCl₂ and 1 MgSO₄; pH was adjusted to 7.3 with NaOH. The recording chamber (volume, 1 ml) was perfused at a rate of 0.8 ml min⁻¹ with a saline whose standard composition was (mm): 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, 2 CaCl₂, 1 MgSO₄, and 20 d-glucose (bubbled with 95% O₂ and 5% CO₂). Toxins included in the saline are specified in the corresponding figure legend. The command potential was set to 0 mV. Current signals were low-pass filtered at 2 kHz, and sampled at 20 kHz using a voltage-clamp amplifier (Axopatch-1D, Axon Instruments) driven by PULSE software (version 8.54, HEKA).

Drug preparation and application

Baclofen (0417, Tocris, Avonmouth, Bristol, UK), (–)-N⁶-(2-phenylisopropyl)adenosine (R-PIA; P-4532, Sigma-Aldrich), BaCl₂, CsCl, l-glutamate sodium salt (G-1626, Sigma), (–)-bicuculline methochloride (0131, Tocris) and CGP55845A (gift from Novartis, Basel, Switzerland) were dissolved into water to concentrations 1000–10 000 higher than the final levels, and kept at ≤4°C until use. Tertiapin-Q (TQ; 1316, Tocris) was dissolved into water to concentrations 300–1000 times higher than the final level, and kept at ≤−20°C until use. (RS)-α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA; 0169, Tocris) was dissolved into the saline to a concentration of 10 mm, and kept at −20°C until use. 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX; 0190, Tocris) was dissolved into dimethylsulfoxide to a concentration of 100 mm, and kept at 4°C until use. Pertussis toxin (PTX; 516560, Calbiochem, San Diego, CA, USA) was reconstituted into the culture medium to a concentration of 50 μg·ml⁻¹, and kept at 4°C until use.

In the ruptured-patch recordings, saline containing the drugs or the control agent (0.001 vol. of water) was applied locally to the whole of the examined Purkinje cell through a theta-tube (BT150-10, Sutter, Novato, CA, USA) under the control of gravity. In some experiments, baclofen and TQ were bath-applied with the chamber-perfusing saline. BaCl₂ and CsCl were added to both the chamber-perfusing and locally applied saline. For pretreatment with PTX, the drug (0.5 μg ml⁻¹) was added to the culture dishes ≥20 h before recordings.

In the perforated-patch recordings, the glutamate- or AMPA-containing saline was applied locally to the dendrites of the examined Purkinje cell through a pipette (tip diameter, < 1 μm) attached to a pressure ejection system (∼0.35 kg cm⁻², PicoSpritzer III, Parker, Fairfield, NJ, USA).

In the cell-attached recordings, CNQX, bicuculline, TQ and CGP55845A were bath-applied with the chamber-perfusing saline.

Data analysis

A baclofen- or R-PIA-induced current was extracted as a difference between the basal and test records (see above) of the whole-cell currents. The amplitude of the extracted component was expressed by the mean current level over a range of test potentials from −129.5 to −131.5 mV (referred to as the amplitude at −130 mV). The reversal potential (E_rev) was estimated from the null-current point of a line fitted to the local region (5–10 mV wide) of a I–V plot using Igor software (versions 4.00–5.01, WaveMetrics, Lake Oswego, OR, USA). The equilibrium potentials of K⁺ and Na⁺ (E_K and E_Na, respectively) were estimated using Nernst equation, assuming the equality of ion activity coefficients between the intra- and extracellular sides. For analysing the kinetics of a voltage step-activated current response, a double-exponential rise function (I(t) =a₁[1 − exp(−t/τ₁)]+a₂[1 − exp(−t/τ₂)]+c, where I(t), a, τ and c are current level at time t, amplitude, time constant and basal current level, respectively), and a double-exponential decay function [I(t) =a₁ exp(−t/τ₁) +a₂ exp(−t/τ₂) +c)] were fitted to the activation phase (50–100 ms) and deactivation phase (15–100 ms), respectively, using the Igor software.

The E_rest and peak amplitude of action potentials are expressed by the mean values of five records obtained during a 75 s period. The peak level and amplitude of glutamate- or AMPA-evoked potentials (GEPs and AEPs, respectively) are expressed by the mean values of six records obtained during a 100 s period.

Spike currents were analysed, using Mini Analysis Program (version 5.1.1, Synaptosoft, Decatur, GA, USA). Inter-spike current interval (ISCI) was measured as an interval between the maximal inward deflections of consecutive spike currents. The peak amplitude of a spike current is measured as a difference from the mean prespike level (length, 2 ms) to the maximal inward deflection. These measures are expressed by the mean values of all of the events in a 9 s record collected during a 3 min period.

Groups of numerical data are presented as means ± s.e.m. Differences between raw values were tested by Student's paired or unpaired t test (with analysis of variance (ANOVA) for more than two groups). Differences between percentage-scored values were tested by rank sum test.

Results

Extraction of a GABA_BR-operated inwardly rectifying current

To explore a possible GABA_BR-operated inwardly rectifying K⁺ current in cerebellar Purkinje cells, we measured the whole-cell currents activated by a voltage ramp (Fig. 1). The E_m was first held at −31 mV for 100 ms to facilitate the inactivation of depolarization-activated currents, and then ramped to −131 mV at a rate of −100 mV s⁻¹. In the control saline, the total currents consisted of an apparently inwardly rectifying component at test potentials more negative than −45 mV, and an apparently outwardly rectifying component at more positive test potentials (Fig. 1A and C, Basal). A major part of the apparently inwardly rectifying component is attributable to the slow activation of the hyperpolarization-activated mixed-cation current (I_h) (Crepel & Penit-Soria, 1986; Li et al. 1993) because of its pharmacological profile (analysed in Fig. 4A and B). The remaining part may include constitutively active inwardly rectifying K⁺ current (Liesi et al. 2000) presumably produced by IRK channels (Falk et al. 1995; Miyashita & Kubo, 1997). The apparently outwardly rectifying component may include the ‘tail’ currents of the depolarization-activated currents. Continuous application of the control saline (2 min) did not change the overall I–V relation of the total currents (Fig. 1A, Control; Fig. 1B). By contrast, baclofen, a GABA_BR-selective agonist (3 μm) augmented the inwardly rectifying component of the total current (Fig. 1C). We extracted the I_bacl as a difference between the total currents recorded before and at the end of a 2 min application of baclofen (3 μm) (Fig. 1D). I_bacl displayed a E_rev (−60.4 ± 3.6 mV, n = 11) close to the E_K (−57.6 mV, Fig. 1D, arrow) and inward rectification (more clearly discernible in Fig. 3B). Induction of I_bacl with 3 μm baclofen was partially or completely reversible (Fig. 1C, ‘Wash’). When compared at a test potential of −130 mV, the amplitude of I_bacl (256.4 ± 54.2 pA, n = 12) was significantly larger than a change in the amplitude of the total currents following a 2 min application of the control saline (−14.3 ± 12.9 pA, n = 10; Fig. 1E). These results exclude that I_bacl is an artefact. In experiments described in the following sections, we did not use higher concentrations of baclofen because their effect was often irreversible, and thus not readily distinguished from artefacts. The results in Fig. 1 clearly demonstrate that GABA_BR activation induces an inwardly rectifying K⁺ current in cerebellar Purkinje cells.

Figure 1. GABA_BR-operated inwardly rectifying current in cerebellar Purkinje cells.

A and B, local application of the control saline does not induce an inwardly rectifying current in cultured Purkinje cells. A, sample voltage ramp-activated currents recorded from a cell consecutively before (Basal) and at the end of (Control) a 2 min application of the control saline. The equilibrium potential of K⁺(E_K) =−57.6 mV. In this and the following figures, the test potential was ramped from −31 to −131 mV at a rate of −100 mV s⁻¹, and the resultant current was converted to a I–V plot. B, mean difference of the basal and control currents; grey lines, ± s.e.m.; n = 10. C and D, baclofen, a GABA_BR-selective agonist, induces an inwardly rectifying current (I_bacl) in cultured Purkinje cells. C, sample voltage-ramp-activated currents recorded from a cell consecutively before (Basal), at the end of (Baclofen) and after (Wash) a 2 min application of 3 μm baclofen. E_K =−57.6 mV. D, mean I_bacl (grey lines, ± s.e.m.) extracted by subtracting the basal currents from the baclofen currents. This extraction procedure is used throughout this study. n = 12. Note that the reversal potential (E_rev) of I_bacl is close to the E_K (arrow). E, mean amplitudes at −130 mV of the currents induced by the control saline (2 min, n = 10) and I_bacl (baclofen, 3 μm, 2 min, n = 12). Inward deflection is taken as positive. Error bars, ± s.e.m.; **P < 0.01, Student's unpaired t test.

Figure 4. Pharmacology of I_bacl.

A–C, I_bacl is not inducible in the continuous presence of extracellular Ba²⁺ or Cs⁺. A and B, sample voltage-ramp-activated currents recorded from single cultured Purkinje cells consecutively before (Basal) and at the end of (Baclofen) a 2 min application of 3 μm baclofen in the saline containing 1 mm BaCl₂ (A) or 3 mm CsCl (B). In addition to the effect on I_bacl, Cs⁺ also blocks the inwardly rectifying component of basal current; note that this component is relatively small in the CsCl-containing saline (B) compared with that in the CsCl-free saline (cf. Figs 1A and C, Basal; see Results for more detailed explanation). C, mean amplitudes of I_bacl at −130 mV (baclofen, 3 μm, 2 min) measured in the normal (n = 12, reproduced from Fig. 1E), 1 mm BaCl₂-containing (n = 9) and 3 mm CsCl-containing (n = 16) saline. Error bars, ± s.e.m.**P < 0.01, ANOVA and Student's unpaired t test. D–E, the GIRK channel antagonist TQ reduces I_bacl. D, sample voltage-ramp-activated currents recorded from a cultured Purkinje cell consecutively before (grey line, Basal) and during application of the labelled drugs. In this experiment, we observed the induction of I_bacl at the end of a 2 min application of 3 μm baclofen alone (continuous black line) and then repeatedly monitored at an interval of 30 s the reduction of I_bacl, coapplying 3 μm baclofen and 30 nm TQ (dashed black line). The effect of TQ reached a maximum within 30–90 s of the coapplication onset. E, mean reduction in the amplitude of I_bacl at −130 mV caused by coapplication of baclofen and TQ (3 μm and 30 nm, respectively, n = 10, ‘+’). For each cell, the maximal effect was expressed as a percentage of the amplitude of I_bacl induced by baclofen alone (3 μm, 2 min). For control, mean reduction of I_bacl was measured during prolonged application of baclofen alone (3 μm, 30–120 s, n = 6, −). Error bars, ± s.e.m.; n.s., P > 0.05; **P < 0.01; Student's paired t test.

Figure 3. Carrier ion species of I_bacl.

A–C, the E_rev of I_bacl shifts with the E_K. A, sample voltage-ramp-activated currents recorded from a cultured Purkinje cell consecutively before (grey line, Basal), at the end of (continuous line, Baclofen) and after (dashed line, Wash) a 2 min application of 3 μm baclofen in the saline with reduced KCl (5.4 mm). E_K =−85.5 mV. B, mean I_bacl (baclofen, 3 μm, 2 min) in the 5.4 mm KCl-containing saline. Grey lines, ± s.e.m., n = 10. Note that the E_rev of I_bacl is close to E_K (arrow). C, mean E_rev of I_bacl (baclofen, 3 μm, 2 min) measured in the saline containing 5.4 mm KCl (n = 10) and 16 mm KCl (n = 11, data taken from Fig. 1D) plotted against the corresponding E_K. Error bars, ± s.e.m.; **P < 0.01, Student's unpaired t test. D–F, the E_rev of I_bacl does not shift with equilibrium potential of Na⁺ (E_Na). D, sample voltage-ramp-activated currents recorded from a cultured Purkinje cell consecutively before (grey line, Basal), at the end of (continuous line, Baclofen) and after (dashed line, Wash) a 2 min application of 3 μm baclofen in the saline with reduced NaCl (25.9 mm; by equimolar replacement with N-methyl-d-glucamine). E_Na =+22.5 mV. E, mean I_bacl (baclofen, 3 μm, 2 min) in the 25.9 mm NaCl-containing saline. Grey lines, ± s.e.m.; n = 7. Note that the E_rev of I_bacl is close to the E_K (−57.6 mV, arrow) but not E_Na. F, mean E_rev of I_bacl (baclofen, 3 μm, 2 min) measured in the saline containing 25.9 mm NaCl (n = 7) and 141.9 mm NaCl (n = 11, data taken from Fig. 1D) plotted against the corresponding E_Na. Error bars, ± s.e.m.; n.s., P > 0.05, Student's unpaired t test.

G_i/o-protein-dependence of I_bacl

Cerebellar Purkinje cell may express GIRK channels (see Introduction) which can be operated by GABA_BR via G_i/o-proteins (Sadja et al. 2003). We examined the G_i/o-protein-dependence of I_bacl (Fig. 2). To this end, we pretreated cultured Purkinje cells for ≥20 h prior to the recordings with PTX, which uncouples G_i/o-proteins from G-protein-coupled receptors. In these cells, baclofen (3 μm) failed to induce I_bacl (Fig. 2A). A change in the amplitude of the total currents following a 2 min application of baclofen (3 μm) in the PTX-pretreated cells (3.6 ± 20.8 pA, n = 7) was significantly smaller than the amplitude of I_bacl in the untreated cells (254.6 ± 54.2 pA, n = 12; Fig. 2B). This result clearly demonstrates that GABA_BR operates I_bacl via G_i/o-proteins.

Figure 2. G_i/o-protein dependence of I_bacl.

I_bacl is not inducible in cultured Purkinje cells pretreated with pertussis toxin (PTX), a G_i/o-protein inhibitor (0.5 μg ml⁻¹, ≥20 h). A, sample voltage-ramp-activated currents recorded from a cell consecutively before (Basal) and at the end of (Baclofen) a 2 min application of 3 μm baclofen. E_K =−57.6 mV. B, mean amplitude at −130 mV of I_bacl (baclofen, 3 μm, 2 min) measured in the untreated cells (n = 12, reproduced from Fig 1E, −) and the PTX-pretreated cells (n = 7, +). Inward deflection is taken as positive. Error bars, ± s.e.m.**P < 0.01, Student's unpaired t test.

Ion-selectivity of I_bacl

In general, inwardly rectifying currents are classified into K⁺-selective currents and I_h which is carried by both K⁺ and Na⁺ (for review see Tabata & Ishida, 1996). To determine the ion species carrying I_bacl, we measured the E_rev of I_bacl with varying E_K and E_Na (Fig. 3). In the standard saline containing 16 mm KCl and 141.9 mm NaCl, the E_rev (−60.4 ± 3.6 mV, n = 11) was close to the E_K (−57.6 mV), but not the E_Na (+ 66.2 mV) (Figs 1D and 3C). When the E_K was set to −85.5 mV by reducing the concentration of KCl in the saline to 5.4 mm, the E_rev shifted to −80.7 ± 3.9 mV (n = 10, Fig. 3A–C). By contrast, the E_rev did not shift (−60.3 ± 4.1 mV, n = 7) when the E_Na was set to +22.5 mV by replacing 120 mm Na⁺ in the saline with N-methyl-d-glucamine, a channel-impermeant monovalent cation (Fig. 3D–F). These results clearly demonstrate that I_bacl is carried by K⁺ but not Na⁺, and is different from I_h.

Pharmacology of I_bacl

We analysed the pharmacological profile of I_bacl (Fig. 4). Sensitivities to millimolar levels of Ba²⁺ and Cs⁺ are widely used criteria to distinguish certain classes of inwardly rectifying K⁺ currents from I_h (for review see Tabata & Ishida, 1996). Inwardly rectifying K⁺ currents can be blocked by both cations. I_h is blocked by Cs⁺ while resistant to Ba²⁺. When 1 mm BaCl₂ (n = 9) or 3 mm CsCl (n = 16) was included in both the chamber-perfusate and the locally delivered saline, baclofen (3 μm) failed to induce I_bacl (Fig. 4A and B). Changes in the amplitude of the total currents induced by a 2 min application of baclofen under these conditions (−12.0 ± 6.6 pA, n = 9 and −2.5 ± 6.2 pA, n = 16, respectively) were significantly smaller than the amplitude of I_bacl in the absence of these cations (254.6 ± 54.2 pA, n = 12; Fig. 4C). This result indicates that I_bacl belongs to inwardly rectifying K⁺ currents, but not to I_h. In addition, the apparently inward rectifying component of the total currents (e.g. Fig. 1A and C, Basal) was persistent in the Ba²⁺-containing saline (Fig. 4A, Basal) while obscured in the Cs⁺-containing saline (Fig. 4B, Basal). This result suggests that this component largely consists of I_h.

We next examined the sensitivity of I_bacl to tertiapin-Q (TQ), a synthetic peptide derived from a bee venom toxin which potently blocks certain classes of inwardly rectifying K⁺ channels, including GIRK channels (Jin & Lu, 1999). For the reason of availability, it was difficult to continuously include TQ in the chamber-perfusing saline. Instead, we sequentially applied baclofen alone (3 μm, 2 min; Fig. 4D, Baclofen) and baclofen together with TQ (3 μm and 30 nm, respectively; Fig. 4D, Baclofen & TQ) locally to the Purkinje cell, and evaluated the reduction of I_bacl following the coapplication. The effect of TQ reached the maximum within 30–90 s of the onset of the coapplication. At the maximum, the amplitude at −130 mV of I_bacl was significantly reduced by 74.7 ± 21.4% (n = 10, Fig. 4E). By contrast, prolonged application of baclofen alone (3 μm, 30–120 s) did not significantly reduce the amplitude (by 2.8 ± 8.1%, n = 6; Fig. 4E), indicating that the effect of TQ is not an artefact due to the run-down or inactivation of I_bacl. This result shows that TQ-sensitive classes of inwardly rectifying K⁺ channels are responsible for I_bacl.

Kinetics of I_bacl

We examined the activation and deactivation kinetics of I_bacl to abrupt voltage changes (Fig. 5). We recorded the total current responses to a hyperpolarizing voltage step (to −91 mV from a holding potential of −61 mV; duration, 100 ms) before and during baclofen application (3 μm) (Fig. 5A). I_bacl was extracted as a difference of these currents (Fig. 5B, continuous black line). Upon the step onset, a major part of I_bacl was activated within a millisecond period, and the remaining part continued to develop onward. These rapid and slow components of the activation phase can be described by a double-exponential rise function (see Methods for definition; Fig. 5B, continuous grey line) with two much different time constants (Table 1). I_bacl displayed no inactivation during hyperpolarization (Fig. 5B). Upon the step offset, a major part of I_bacl was deactivated within a millisecond period, and the remaining part more slowly. These rapid and slow components of the deactivation phase can be described by a double-exponential decay function (see Methods for definition; Fig. 5B, dashed grey line) with two much different time constants (Table 1). These results show that I_bacl can rapidly respond to voltage changes, and once activated it continuously functions for a prolonged time.

Figure 5. Kinetics of I_bacl.

I_bacl is rapidly activated and deactivated upon membrane potential (E_m) jumps, and sustains without inactivation during hyperpolarization. A, sample voltage-step-activated currents recorded from a cultured Purkinje cell consecutively before (Basal), at the end of (Baclofen) and after (Wash) a 1 min application of 3 μm baclofen. The schematic above the left trace indicates the time course of the test potential. Dashed line, null-current level. E_K =−56.7 mV. B, I_bacl extracted as a difference between the basal and baclofen currents in A. The activation and deactivation phase of I_bacl (continuous black line) can be described by a double-exponential rise function (continuous grey line) and a double-exponential decay function (dashed grey line), respectively (see Methods for the definition of these functions). The parameters of the fitted functions are presented in Table 1.

Table 1.

Parameters of double-exponential functions fitted to the activation and deactivation phases of the inwardly rectifying current activated by baclofen

Component	Relative amplitude	Time constant (ms)
Activation, rapid	0.754 ± 0.039	0.81 ± 0.15
Activation, slow	0.246 ± 0.039	82.4 ± 43.1
Deactivation, rapid	0.930 ± 0.028	0.60 ± 0.11
Deactivation, slow	0.070 ± 0.028	30.5 ± 16.4

Data are collected from 8 cells in the experiment shown in Fig. 5, and presented as means ± s.e.m. See Methods for the definition of the functions.

Possible functional contribution of I_bacl

We assessed how I_bacl influences the E_rest and action potentials of cerebellar Purkinje cells using a perforated-patch current-clamp technique in cultured cell preparations (Fig. 6). To minimize spontaneous firing, we injected a constant background current (typically −15 to +15 pA) as to set the mean E_rest to ∼−65 mV at the beginning of a recording session (Fig. 6A, Basal), and continued injection at the same intensity for the rest of the recording session. This level of the E_rest was naturally observed for Purkinje cells without vigorous spontaneous firing (data not illustrated). When an additional current stimulus was not imposed, the momentary E_m fluctuated by only a few millivolts around the mean E_rest (cf. Fig. 6D). Baclofen (3 μm, 1–3 min) hyperpolarized the E_rest by 3.46 ± 1.15 mV (n = 14; Fig. 6A and C). There was a significant difference between the mean E_rest before and during baclofen application (Fig. 6C). This effect can be ascribed to I_bacl because it was not reproduced with a statistical significance in the continuous presence of 0.1–1 mm BaCl₂ in the saline (n = 12; Fig. 6B and C).

Figure 6. Actions of I_bacl on the E_rest and action potentials.

I_bacl hyperpolarizes the resting potential (E_rest), while it does not much affect action potentials. A and B, sample current-clamp traces recorded from single cultured Purkinje cells consecutively before (grey line, Basal), during (1–2 min of baclofen onset, black line, Baclofen) and after (in A only; dashed line, Wash) baclofen application (3 μm) in the BaCl₂-free (A) and 0.1 mm BaCl₂-containing (B) saline. Note a change in the E_rest in A. Depolarizing pulses (300 pA, 5 ms, applied between arrow heads) evoke action potentials. Each trace indicates the average of three to five records. Insets, action potentials (single records) expanded in time scale and justified at the peaks. The saline contained 3 mm KCl, 10 μm bicuculline and 1 mm kynurenic acid, but not TTX. The recordings were made in perforated-patch mode. C, mean changes in the E_rest from the basal to baclofen records in the BaCl₂-free (n = 14, −) and BaCl₂-containing (0.1–1 mm, n = 12, +) saline. Hyperpolarizing shift is taken as positive. Error bars, ± s.e.m.**P < 0.01 and n.s. (P > 0.05) between the basal and baclofen records, Student's paired t test. D, peak levels of action potentials are plotted against the momentary E_m (mean over 10 ms) prior to the depolarizing pulses. Open and closed symbols indicate the data of the basal and baclofen records, respectively. Each peak level is expressed as a difference from the mean level of 5 basal records. The dashed line indicates a linear function fitted to the data by regression (slope, −0.84). Data were pooled from 3 cells.

Injection of a strong depolarizing pulse (300 pA, 5 ms) evoked an action potential (Fig. 6A and B). Baclofen (3 μm, 1 min) prolonged the peak latency of the action potential by ∼1 ms in five out of eight Purkinje cells (Fig. 6A, inset). Baclofen (3 μm, 1 min) also raised the peak level achieved by the action potential by ∼2 mV in five out of eight cells (Fig. 6A, inset). The peak level was higher when the E_m prior to depolarizing pulse was more hyperpolarized in both the absence and presence of baclofen (data pooled from three cells with the most evident effect of baclofen on the peak level, Fig. 6D). These effects can be ascribed to I_bacl because they were not seen in the continuous presence of 0.1–1 mm BaCl₂ in the saline (n = 7; Fig. 6B, inset). However, the effects of baclofen on the peak latency and peak level of action potentials were not obvious in the remaining cells. Thus, baclofen-induced changes in the mean peak latency (from 3.99 ± 0.56 to 4.77 ± 0.90 ms) and the mean peak level (from +24.8 ± 4.4 to +25.3 ± 4.5 mV) for all of the examined cells (n = 8) were not statistically significant (P > 0.05, Student's paired t test). Baclofen did not delay the recovery from after-hyperpolarization to the E_rest (Fig. 6A). The mean time constant of the exponential functions fitted to the main component of the recovery phase before baclofen application was 41.6 ± 13.5 ms, and that during baclofen application was 33.4 ± 5.4 ms (n = 6, data not illustrated). There was no significant difference between the time constants (P > 0.05, Student's paired t test). These results suggest that I_bacl may decrease the readiness of firing by hyperpolarizing the E_rest, while it may not much affect the waveform of an action potential itself.

We next assessed how I_bacl influences GEPs in Purkinje cells using a perforated-patch current-clamp technique (Fig. 7). We used glutamate puff instead of stimulation of glutamatergic neurones innervating Purkinje cells because baclofen is known to attenuate synaptic release via the presynaptic mechanism (Dittman … Regehr, 1996, 1997). We puffed glutamate (1 mm, 3–6 ms) to the middle of the dendritic arbor of a Purkinje cell repeatedly at an interval of 20 s, and recorded GEPs from the soma (Fig. 7A). The peak amplitude of the GEPs ranged from 5 to 20 mV (Fig. 7B and C, Basal). Baclofen (3 μm, 80–180) had little effect on their waveform and peak amplitude (n = 6, Fig. 7B). There was no significant difference between the peak amplitudes before (16.8 ± 3.3 mV) and during (16.7 ± 3.7 mV) baclofen application (n = 6; P > 0.05, Student's paired t test). However, baclofen (3 μm, 80–180 s) significantly lowered the peak level achieved by the GEPs by 6.29 ± 2.19 mV (n = 6), along with a hyperpolarizing shift in the E_rest (Fig. 7B and D). This effect of baclofen can be ascribed to I_bacl because a significant lowering in the peak level was not observed in the continuous presence of 1 mm BaCl₂ in the saline (n = 5, Fig. 7C and D). In Purkinje cells, glutamate may evoke not only ionotropic glutamate receptors but also metabotropic glutamate receptors, and both the types of receptors may contribute to GEPs (see Tabata et al. 2004 for review). GABA_BR activation is reported to lead to augmentation of mGluR1-mediated responses (Hirono et al. 2001). Thus, baclofen could augment the mGluR1-mediated component of GEPs, and this could mask the attenuation of the ionotropic glutamate receptor-mediated component. However, these possibilities are not the cases because baclofen did not reduce the peak amplitude of postsynaptic potentials evoked by AMPA (AEP), an ionotropic glutamate receptor agonist (0.1–1 mm, 10.1 ± 3.8 mV before baclofen application and 11.3 ± 4.2 mV after baclofen application, n = 6; Fig. 7E). Ionotropic glutamate receptors might largely mediate GEPs elicited by the glutamate application protocol used in this study. The results shown in Fig. 7 suggest that I_bacl may impede excitatory synaptic inputs from triggering action potentials.

Figure 7. Actions of I_bacl on GEPs.

I_bacl lowers the peak levels achieved by glutamate-evoked potentials (GEPs). A, schematic shows the recording configuration. l-Glutamate (1 mm) or AMPA (0.1–1 mm) was repeatedly pressure-applied (duration 3–8 ms; interstimulus interval, 20 s) to the middle of the dendritic arbor of a cultured Purkinje cell through a pipette. The agonist-evoked potentials were recorded from the soma using a perforated-patch current-clamp technique. The saline contained 3 mm KCl, 0.3–0.5 μm TTX, and 10 μm bicuculline but not kynureic acid. B and C, sample GEPs recorded from single cells consecutively before (Basal), during (80–180 s of baclofen onset, Baclofen) and after (Wash) baclofen application (3 μm) in the BaCl₂-free (B) or 1 mm BaCl₂-containing (C) saline. Baclofen was bath-applied. Each trace indicates the average of three to six records. D, mean change in the peak level achieved by GEPs from the basal to baclofen records in the BaCl₂-free (n = 6, −) and 1 mm BaCl₂-containing (n = 5, +) saline. Hyperpolarizing shift is taken as positive. Error bars, ± s.e.m.*P < 0.05; n.s., P > 0.05; between the basal and baclofen records, respectively, Student's paired t test. E, sample AMPA-evoked potentials (AEPs) recorded from a cell consecutively before (Basal), during (80–180 s of baclofen onset, Baclofen) and after (Wash) baclofen application (3 μm) in the BaCl₂-free saline. Each trace indicates the average of six records.

Finally, we examined whether I_bacl can function in situ using Purkinje cells in cerebellar slice preparations. We tried to measure the effects of baclofen on events observed under current clamp in a ruptured-patch whole-cell configuration. However, this attempt was hampered because the effects of baclofen and the intrinsic excitability of the examined cells severely ran down during a prolonged recording in this configuration presumably due to cytoplasmic disturbance (n = 3, data not illustrated). Instead, we used a noninvasive recording technique (cell-attached technique), and evaluated the effects of baclofen on spontaneous firing. In a cell-attached configuration, the depolarizing phase of an action potential can be observed as a transient inward current (spike current). In the normal saline, most of the examined cells displayed regular spontaneous firing with an ISCI of 21–93 ms (55.2 ± 8.0 ms, n = 7). Inclusion of CNQX and bicuculline, antagonists against ionotropic glutamate and GABA receptors in the saline, did not stop or slow down spontaneous firing (ISCI, 18–37 ms, n = 4; data not illustrated), suggesting that the intrinsic mechanisms but not excitatory or inhibitory synaptic inputs initiate spontaneous firing. In the regularly firing cells, baclofen significantly increased the ISCI by 221.8 ± 23.6% (n = 7, Fig. 8A and D). Subsequent application of 2 μm CGP55845A, a GABA_BR-selective antagonist, reversed this effect of baclofen (n = 7, Fig. 8A–C and E). Baclofen significantly increased the ISCI even in the continuous presence of 0.1 mm BaCl₂ (by 73.4 ± 19.7%, n = 4) or 150–300 nm TQ (by 74.6 ± 12.3, n = 7) in the saline (Fig. 8B–D). However, the extent of the effect with BaCl₂ or TQ was significantly smaller that that without these agents (Fig. 8D, # and ##). The baclofen-induced increase in the ISCI was accompanied with augmentation of spike currents (Fig. 8A and F). This effect was significant only in the absence of 0.1 mm BaCl₂ or 150–300 nm TQ (Fig. 8A–C and F), and reversed by 2 μm CGP55845 (Fig. 8A and G). The results shown in Fig. 8 demonstrate that I_bacl can function in situ and influence the intrinsic excitability of Purkinje cells.

Figure 8. Actions of I_bacl on the intrinsic excitability in situ.

I_bacl reduces the rate of spontaneous firing of Purkinje cells in cerebellar slices presumably by hyperpolarizing the E_rest. A–C, sample spike currents (see Methods) recorded from single Purkinje cells consecutively before (Basal) and during baclofen application (10 μm, at 5–8 min of the application onset), and during CGP55845A application (a GABA_BR-selective antagonist, 2 μm, at 5–8 min of the application onset) in the BaCl₂-/TQ-free (A), 0.1 mm BaCl₂-containing (B) and 300 nm TQ-containing (C) saline. The recordings were made in a cell-attached mode. D–G, mean changes in inter-spike current interval (ISCI; D and E) and the peak amplitude (F and G) of spike currents following baclofen and CGP55845A application without (n = 7) or with 0.1 mm BaCl₂ (n = 4) or 150–300 nm TQ (n = 7) measured as in A–C. Error bars, ± s.e.m. For each condition, the effects of baclofen (difference between the basal and baclofen records) and CGP55845A (difference between the basal and CGP55845A records) were tested by rank sum test (n.s., P > 0.05; *P < 0.05; **P < 0.01). When judged significant by the previous test, the effects were further compared between the conditions by rank sum test (N.S., P > 0.05; #P < 0.05; ##P < 0.01).

Discussion

Functional identification of a GABA_BR-operated inwardly rectifying current in cerebellar Purkinje cells

In this study, we characterized an inwardly rectifying current activated by the GABA_BR-selective agonist baclofen (I_bacl) in cultured cerebellar Purkinje cells, using a ruptured-patch whole-cell voltage-clamp technique (Figs 1–5). In central neurones, inwardly rectifying currents may be produced by Na⁺-/K⁺-permeable channels (HCN channels; Robinson & Siegelbaum, 2003) and/or the supergene family of K⁺-selective channels (Kir channels; Bichet et al. 2003). Cerebellar Purkinje cells possess I_h (Crepel & Penit-Soria, 1986; Li et al. 1993), which is produced by HCN channels (Robinson & Siegelbaum, 2003), and they appear to express GIRK2–4 (Kir3.2–3.4, see below), IRK1,3 (Kir2.1,2.3) (Falk et al. 1995; Miyashita & Kubo, 1997), and Kir7.1 (Krapivinsky et al. 1998). In this section, we compare the physiological properties of I_bacl (Figs 1, 2, 3, 4, 5) with these channels.

The E_rev of I_bacl was close to the E_K, and shifted depending on the extracellular concentration of K⁺ but not that of Na⁺ (Figs 1 and 3). These results demonstrate that I_bacl is produced by Kir channels but not HCN channels. Uncoupling G_i/o-proteins from GABA_BR almost completely suppressed I_bacl (Fig. 2). This result indicates that I_bacl is operated by GABA_BR via G_i/o-proteins and is produced by GIRK channels which are highly dependent on G_i/o-proteins compared with other Kir channels (Isomoto et al. 1997; Bichet et al. 2003; Sadja et al. 2003). Both Ba²⁺ and Cs⁺, at millimolar levels, completely blocked I_bacl (Fig. 4A–C). I_bacl shares these metal ion sensitivities with GIRK (Hommers et al. 2003) and IRK (Isomoto et al. 1997) channels. By contrast, HCN channels (I_h) are resistant to Ba²⁺ (Crepel & Penit-Soria, 1986; Tabata & Ishida, 1996; Robinson & Siegelbaum, 2003), and Kir7.1 channels are resistant to both Ba²⁺ and Cs⁺ (Krapivinsky et al. 1998). TQ at 30 nm blocked a large fraction of I_bacl (Fig. 4D and E). The degree of blockade is consistent with the reported affinity of TQ for GIRK channels (K_i, 13.3 nm) but not that for IRK channels (no significant effect even at 1 μm) (Jin & Lu, 1999). Upon the onset of a hyperpolarizing voltage step, I_bacl displayed the activation phase consisting of rapid and slow components with much different time constants (Fig. 5, Table 1). The activation phases of GIRK currents in cerebellar granule neurones (Slesinger et al. 1997) and GIRK-transfected heterologous cells (Wischmeyer et al. 1997) also consist of rapid and slow components. I_bacl is distinguished from Kir7.1 current, which is instantaneously activated to the maximal level (Krapivinsky et al. 1998), and I_h, which displays single-exponential activation phase with a time constant (50–100 ms; Crepel & Penit-Soria, 1986; Tabata & Ishida, 1996) much longer than that of the main activation component of I_bacl (∼0.8 ms, Table 1). During a 100 ms hyperpolarization, I_bacl sustained without inactivation (Fig. 5). I_bacl is distinguished from IRK1 currents, which display apparent inactivation due to open channel block by external Na⁺ (Kubo et al. 1993). Taken together, our voltage-clamp measurements (Figs 1–5) demonstrate that cerebellar Purkinje cells are equipped with a GABA_BR-operated inwardly rectifying current most likely produced by GIRK channels.

The subunit composition of GIRK channels mediating I_bacl remains to be elucidated in future studies. Some insitu hybridization studies show that Purkinje cells express the mRNAs of GIRK2 (Lauritzen et al. 1997), GIRK3 (Karschin et al. 1996) and GIRK4 (Karschin et al. 1996; Iizuka et al. 1997). Some immunohistochemical studies show that GIRK2 protein is present in Purkinje cells (Lauritzen et al. 1997) (albeit there are opposing reports; Liao et al. 1996; Murer et al. 1997), and GIRK4 protein localizes to the proximal dendrites, somata and proximal axonal segments of Purkinje cells (Iizuka et al. 1997; Murer et al. 1997). Thus, Purkinje cells may express homo- and/or heteromeric channels consisting of GIRK2–4. In certain heterologous expression systems, GIRK4 form functional channels alone or together with GIRK3 (Iizuka et al. 1995; Wischmeyer et al. 1997). These channels could be candidates for the I_bacl-mediating channels. On the other hand, Purkinje cells do not have detectable levels of the mRNA (Karschin et al. 1996) and protein of GIRK1 (Ponce et al. 1996; Miyashita & Kubo, 1997). Thus, Purkinje cells express virtually no GIRK1/2 heteromeric channel, which is the predominant type in many other central neurones (Jan & Jan, 1997). In the heterologous expression system, GIRK1-free channels produce smaller macroscopic currents than GIRK1-containing channels (Wischmeyer et al. 1997). Lack of GIRK1 may explain that the amplitude of I_bacl in Purkinje cells is comparable with that in much smaller neurones (e.g. cerebellar granule cells; Slesinger et al. 1997).

Possible physiological roles of a GABA_BR-operated inwardly rectifying current in cerebellar Purkinje cells

Using a perforated-patch whole-cell current-clamp technique, we assessed how I_bacl influence the postsynaptic excitability of cultured cerebellar Purkinje cells (Figs 6 and 7). GABA_BR in Purkinje cells is reported to respond to not only GABA released from the innervating interneurones, but also to GABA spillover from neighbouring inhibitory synapses between other neurones in situ (Hirono et al. 2001). One study (Dittman & Regehr, 1997) estimates that the spillover elevates the extracellular GABA concentration around Purkinje cells to ≤10 μm. GABA and baclofen have similar binding affinities (∼60 nm) and potencies for GABA_BR (Bowery, 1993). Thus, the events observed with 3 μm baclofen in Figs 6 and 7 may occur under physiological conditions when cerebellar cortical GABAergic neurones are active.

Baclofen (3 μm) did not much affect the waveform of an action potential evoked by a suprathreshold stimulus (Fig. 6A). This result indicates that I_bacl may not limit the contribution of strong excitatory synaptic inputs (e.g. climbing fibre inputs) to the generation of action potentials, whereas I_bacl may affect weak excitatory inputs (e.g. parallel fibre inputs) in a manner mentioned below. In some cells, baclofen raised the peak level achieved by the action potentials (see Results; Fig. 6A, inset). In general, such augmentation could result from the increased activity of the voltage-gated Na⁺ channels producing the depolarizing phase of an action potential. One possibility is that the GABA_BR–G_i/o-protein signalling cascade suppresses activation of cyclic adenosine monophosphate-dependent kinase, which is reported to reduce the peak current of the voltage-gated Na⁺ channels (Catterall, 1999). Another possibility is that a hyperpolarizing shift in the E_rest caused by I_bacl (Figs 6A and C) facilitates disinactivation of the Na⁺ channels (i.e. recovery from depolarization-induced inactivation; for review see Hayashida & Ishida, 2004). The latter possibility is consistent with the observation that the peak level depended on the E_m prior to the action potential (Fig. 6D).

In hippocampal pyramidal neurones, activation of GIRK channels results in attenuation of GEPs initiated in the dendrites (Seeger & Alzheimer, 2001), presumably because GIRK channels effectively reduce the input resistance at the stimulated site (Koch, 1999). In cerebellar Purkinje cells, by contrast, such attenuation was not observed (Fig. 7B). Some immunohistochemical studies (Iizuka et al. 1997; Murer et al. 1997) suggest condensation of GIRK subunits around the soma. Thus, it could be possible that K⁺ conductance through the I_bacl-producing channels is relatively small in the dendrites, and does not effectively reduce the input resistance for GEPs. However, baclofen significantly hyperpolarized the E_rest (Fig. 6A and C) and the peak level achieved by GEPs in Purkinje cells (Fig. 7B and D). These effects are due largely to I_bacl because they were not observed in the continuous presence of extracellular Ba²⁺ (Figs 6B and C, and 7C and D). These results indicate that when activated, I_bacl effectively increases K⁺ conductance around the soma, and may impede excitatory synaptic inputs from driving the E_m towards the firing threshold. The coincidence of synaptic inputs from parallel fibres and a climbing fibre in the Purkinje cell is a key step to induce cerebellar long-term depression (LTD), a form of synaptic plasticity crucial for cerebellar motor learning (for review see Ito, 2002). GABAergic activation of I_bacl might influence induction of LTD by preferentially limiting the contribution of relatively weak parallel fibre inputs to postsynaptic excitation.

The cell-attached recordings from Purkinje cells in cerebellar slices showed that baclofen decreases the rate of spontaneous firing (Fig. 8A and D), which is initiated by the intrinsic mechanisms of Purkinje cells but not synaptic inputs (see Results). I_bacl activated via the GABA_BR is thought to be the primary mechanism enabling this effect of baclofen because the GABA_BR-selective antagonist CGP55845A reversed it (Fig. 8A and E), and extracellular Ba²⁺ and TQ largely eliminated it (Fig. 8B–D). I_bacl could exert the effect on the firing rate by hyperpolarizing the E_rest (Fig. 6A and C). This notion is consistent with the observation that baclofen augmented the peak amplitude of spike currents (Fig. 8A and F); such augmentation is expected to occur with the hyperpolarized E_rest (see above; Fig. 6A and D). These results suggest that I_bacl reduces the intrinsic excitability of Purkinje cells, and strongly support the notion I_bacl can function in Purkinje cells in situ.

In conclusion, we have functionally identified a GABA_BR-operated inwardly rectifying K⁺ current most likely produced by GIRK channels, and demonstrated its possible inhibitory actions on the postsynaptic and intrinsic excitability of cerebellar Purkinje cells. These findings give a new insight into the physiological roles of GABA_BR signalling in cerebellar cortex.