Postsynaptic GABA_B receptor signalling enhances LTD in mouse cerebellar Purkinje cells

Abstract

Long-term depression (LTD) of excitatory transmission at cerebellar parallel fibre–Purkinje cell synapses is a form of synaptic plasticity crucial for cerebellar motor learning. Around the postsynaptic membrane of these synapses, B-type γ-aminobutyric acid receptor (GABA_BR), a G_i/o protein-coupled receptor for the inhibitory transmitter GABA is concentrated and closely associated with type-1 metabotropic glutamate receptors (mGluR1) whose signalling is a key factor for inducing LTD. We found that in cultured Purkinje cells, GABA_BR activation enhanced LTD of a glutamate-evoked current (LTD_glu), increasing the magnitude of depression. It has been reported that parallel fibre–Purkinje cell synapses receive a micromolar level of GABA spilt over from the synaptic terminals of the neighbouring GABAergic interneurons. This level of GABA was able to enhance LTD_glu. Our pharmacological analyses revealed that the βγ subunits but not the α subunit of G_i/o protein mediated GABA_BR-mediated LTD_glu enhancement. G_i/o protein activation was sufficient to enhance LTD_glu. In this respect, LTD_glu enhancement is clearly distinguished from the previously reported GABA_BR-mediated augmentation of an mGluR1-coupled slow excitatory postsynaptic potential. Baclofen application for only the induction period of LTD_glu was sufficient to enhance LTD_glu, suggesting that GABA_BR signalling may modulate mechanisms underlying LTD_glu induction. Baclofen augmented mGluR1-coupled Ca²⁺ release from the intracellular stores in a G_i/o protein-dependent manner. Therefore, GABA_BR-mediated LTD_glu enhancement is likely to result from augmentation of mGluR1 signalling. Furthermore, pharmacological inhibition of GABA_BR reduced the magnitude of LTD at parallel fibre–Purkinje cell synapses in cerebellar slices. These findings demonstrate a novel mechanism that would facilitate cerebellar motor learning.

In the cerebellar cortex, individual Purkinje cells integrate excitatory synaptic inputs from numerous parallel fibres, the axons of granule cells which receive mossy fibre inputs that convey sensory information arising from various parts of the body and motor command signals from the upper centres (Thach et al. 1992; Llinas et al. 2003). The efficacy of transmission at certain parallel fibre–Purkinje cell synapses undergoes long-term depression (LTD) following correlated transmission at these synapses and climbing fibre–Purkinje cell synapses (Ito, 2002). Cerebellar LTD modifies information flow along the cerebellum and is therefore crucial for cerebellar motor learning (Ito, 2002).

Purkinje cells express a high density of B-type γ-aminobutyric acid receptor (GABA_BR), a G_i/o protein-coupled receptor for the inhibitory neurotransmitter GABA (Jones et al. 1998; Kaupmann et al. 1998; Kuner et al. 1999). Interestingly, GABA_BR is concentrated on the annuli of parallel fibre-innervated dendritic spines (Ige et al. 2000; Kulik et al. 2002) where LTD occurs. When the neighbouring interneurons are stimulated at a relatively high frequency, GABA_BR in the dendritic spines may receive 5–10 μm GABA spilt over from the synapses of the interneurons (Dittman & Regehr, 1997). GABA_BR has a high enough affinity (EC₅₀, ∼1 μm) to sense this level of GABA (Sodickson & Bean, 1996). GABA_BR activated by spilt-over GABA could possibly influence the signalling of type-1 metabotropic glutamate receptor (mGluR1), a G_q/11 protein-coupled receptor, because these receptors colocalize at the annuli of Purkinje cell dendritic spines (Lujan et al. 1997; Kulik et al. 2002) and are likely to form complexes (Tabata et al. 2004). mGluR1 signalling is an essential factor for inducing LTD (Conquet et al. 1994; Shigemoto et al. 1994; Ichise et al. 2000; Ito, 2002); this signalling leads to the facilitated endocytosis of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-type glutamate receptor at parallel fibre–Purkinje cell synapses (Matsuda et al. 2000; Wang & Linden, 2000; Chung et al. 2003). Therefore, possible influence of GABA_BR on mGluR1 could modulate the induction of LTD.

In this study, we tested this possibility and explored the underlying mechanism. In cerebellar slice preparations, pharmacological manipulation of GABA_BR should affect glutamate release by the parallel fibres and climbing fibres through modulating the presynaptic mechanisms (Dittman & Regehr, 1996). This masks the postsynaptic action of GABA_BR on LTD. Thus, we employed an in vitro experimental system using cultured Purkinje cells, in which parallel and climbing fibre inputs are replaced with local application of glutamate and somatic depolarization, respectively (Linden et al. 1991). In this system, a glutamate-evoked current corresponds to parallel fibre-mediated excitatory postsynaptic current and undergoes LTD following conjunctive glutamate/depolarization stimuli (LTD_glu).

We show here that postsynaptic GABA_BR activation enhances LTD_glu, increasing the magnitude of depression. Our pharmacological analyses indicate that this LTD_glu enhancement is likely to result from G_i/o protein-mediated augmentation of mGluR1 signalling. Furthermore, we show that GABA_BR indeed influences LTD of parallel fibre–Purkinje cell excitatory postsynaptic currents (EPSCs) in cerebellar slices.

These findings together demonstrate a novel mechanism that would facilitate cerebellar motor learning.

Methods

Cell Culture

Cerebellar cell culture was prepared as described elsewhere (Tabata et al. 2000). Briefly, perinatal C57BL/6 embryos were caesarean-sectioned from pregnant mice deeply anaesthetized and killed with diethylether or isoflurane. The embryos were deeply anaesthetized by cooling in chilled phosphate-buffered saline and then killed by decapitation. Cerebellar neurons were obtained from these embryos, dissociated with trypsin, and plated on plastic dishes or low-fluorescence plastic films (Sumilon, MS-92132, Sumitomo, Tokyo, Japan). The cells were cultured in a low-serum medium for 9–22 days. Purkinje cells were identified by their large somata (18 μm or greater) and/or thick primary dendrites. This procedure fully conforms to the guidelines administered by the committee on animal experiments of Osaka University.

Slice preparation

Cerebellar slices were prepared as described elsewhere (Kakizawa et al. 2000; Kakizawa et al. 2005; Kakizawa et al. 2007). Briefly, 3-week-old C57BL/6 mice were deeply anaesthetized with diethylether and then killed by decapitation. Cerebella obtained from these mice were sliced along the parasagittal planes to a thickness of 250 μm. This procedure fully conforms to the guidelines established by the Animal Welfare Committee of the University of Tokyo.

Electrophysiology

Somatic whole-cell voltage-clamp recordings were made from cultured Purkinje cells in dishes, using a perforated-patch technique (holding potential after the correction of a liquid junction potential between the pipette and bath solutions, −70 mV). The recording pipette contained (mm): 95 Cs₂SO₄, 15 CsCl, 0.4 CsOH, 8 MgCl₂, 10 Hepes and 500 μg ml⁻¹ amphotericin B (pH 7.35). The bath was perfused at a rate of 1–2 ml min⁻¹ with a saline consisting of (mm): 140 NaCl, 5 KCl, 2 CaCl₂, 0.8 MgCl₂, 10 Hepes, 0.5 μm tetrodotoxin, and 10 μm (−)-bicuculline methochloride (pH 7.3, 25°C). Signals were sampled at 5 kHz, using an EPC-8 amplifier (HEKA, Lambreht, Germany) driven by PULSE software (versions 8.77, HEKA).

In the measurements using cerebellar slices, somatic whole-cell voltage-clamp recordings (holding potential after the correction of the liquid junction potential, −80 to −90 mV) were made from Purkinje cells, using pipettes (2.5–3.5 MΩ) containing (mm): 60 CsCl, 40 caesium d-gluconate, 20 TEA-Cl, 1 EGTA, 4 MgCl₂, 4 ATP, 0.4 GTP, and 30 Hepes (pH 7.3). The bath solution contained (mm): 125 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgSO₄, 1.25 NaH₂PO₄, 26 NaHCO₃ and 20 glucose (24–25°C), which was bubbled continuously with a mixture of 95% O₂ and 5% CO₂. Bicuculline (10 μm) was always present in the saline to block spontaneous inhibitory postsynaptic currents. Parallel fibre–Purkinje cell EPSCs were monitored delivering electrical pulses (duration, 0.1 ms) to the parallel fibres through a glass pipette filled with the bath solution at 0.1 Hz except for the period of LTD induction. The intensity of the pulses was adjusted as to evoke parallel fibre–Purkinje cell EPSCs with a basal amplitude of 100–150 pA. Signals were sampled at 20 kHz, using an EPC-9 amplifier (HEKA). After stable basal recording was obtained for at least 10 min, LTD was induced with a conjunctive stimulation protocol stated in the corresponding figure legend.

Ca²⁺ imaging

Cerebellar neurons on a film were incubated with fura-2 acetoxymethyl ester (5 μm) at 37°C for 15 min. Then, the film was placed on a glass-based recording chamber and perfused at a rate of ∼0.8 ml min⁻¹ with saline (see above) supplemented with 6-nitro-7-sulfamoylbenzo(f)quinoxaline-2,3-dione (NBQX) and (3-[(R)-2-carboxypiperazin-4-yl]propyl-1-phosphonic acid (R-CPP). Intracellular free calcium ion concentration ([Ca²⁺]_i)-dependent fluorescence signals were captured at 2–5 Hz, using an imaging system (Polychrome II, TILL, Planegg, Germany) attached to an inverted microscope (IX70 with a ×20 objective lens of NA 0.75, Olympus, Tokyo, Japan). The amplitude of a [Ca²⁺]_i rise was expressed as a change in the ratio of somatic fluorescence signals excited at 340 and 380 nm (exposure duration, 40 and 20 ms, respectively) (F₃₄₀/F₃₈₀).

Drug application

Test drugs were dissolved into water to concentrations 1000 times higher than the final levels, kept at 4 or −20°C until use, and diluted into saline on the day of recordings unless otherwise stated. Forskolin and thapsigargin were dissolved into DMSO to concentrations of 5000 and 10 000 times higher than the final levels, respectively, and kept at −20°C until use. Glutamate was applied iontophoretically (30 ms, 0.05 Hz) through a glass pipette filled with 10 mm l-glutamate and 10 mm Hepes (pH adjusted to 7.1 with NaOH, kept at −20°C until use) and located about 20 μm from the branching point of a primary dendrite of the examined cell. The level of ejection current was adjusted (200–800 nA, 30 ms) so that the peak amplitude of basal glutamate-evoked currents were in a range of 100–300 pA. Bath application of baclofen, GABA, mastoparan, and forskolin was done by perfusing the recording chamber at rate of 1–2 ml min⁻¹ with the drug-containing saline. Local application of (R,S) - 3,5-dihydroxyphenylglycine (DHPG), baclofen, and K⁺-rich saline was done by delivering the drug-containing saline through a wide-tipped pipette located near the examined cell under the control of gravity. In some experiments, pertussis toxin (PTX, reconstituted in the culture medium at a concentration 100 times higher than the final level and kept at 4°C until use) was added to the culture medium 16–24 h before recordings.

Data analysis

The peak amplitude of a response was measured as a difference from the prestimulus level to the maximal deflection throughout the record (glutamate-evoked current) or during a 10 s (DHPG-evoked [Ca²⁺]_i rises) or 1 s (K⁺-evoked [Ca²⁺]_i rises) agonist application.

In measurements using cerebellar slices, the amplitudes of parallel fibre–Purkinje cell EPSCs were normalized to mean over a 10 min basal recording. For each Purkinje cell, the magnitude of LTD was evaluated as the mean of parallel fibre–Purkinje cell EPSCs over 26–30 min after the conjunctive stimuli. Data were discarded in the following cases: (i) either the series resistance or membrane resistance changed by more than 10% during a recording session, (ii) the slope of parallel fibre–Purkinje cell EPSC amplitude changed by more than 2% during the 10 min basal recording, and/or (iii) the amplitude continued to change over 30 min after the formation of whole-cell configuration (Namiki et al. 2005; Kakizawa et al. 2007).

Groups of numerical data are presented as means ± s.e.m. Statistical differences were examined by two-sided Mann–Whitney's U test unless otherwise stated.

Immunohistochemistry

Cerebellar cells cultured on films were fixed sequentially with 0.1 m sodium phosphate buffer containing 4% paraformaldehyde at room temperature for 30 min and with 100% methanol at room temperature for 10 min. The fixed cells were rinsed with phosphate-buffered saline (PBS) for three times and then treated with PBS containing 0.1% Triton X-100 and 10% normal donkey serum at room temperature for 30 min (Kamikubo et al. 2006). The treated cells were incubated with a primary antibody against phospholipase C (PLC) β3 and PLCβ4 (raised in guinea pigs and rabbits, respectively; Nakamura et al. 2004; Nomura et al. 2007), GABA_BR1 subunit (GBR1; raised in rabbits; Kulik et al. 2002), mGluR1 (raised in guinea pigs; Tanaka et al. 2000), or calbindin (raised in mice; C9848, Sigma, St Louis, MO, USA) (1 μg ml⁻¹ in the Triton X-100/serum-containing PBS) at 4°C overnight and then treated with a secondary antibody conjugated with indocorbocyanine (Cy3; Jackson ImmunoResearch, West Grove, PA, USA) or Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA) 1: 200 in the Triton X-100/serum-containing PBS at room temperature for 2.5 h. The film was mounted on a glass slide with Perma Fluor Aquaus mounting medium (434980, Thermo, Pittsburgh, PA, USA), and the cells were examined using a confocal laser microscope (LSM510, Carl Zeiss Göttingen, Germany). The fluorescent signals of Cy3 (red) and Alexa Fluor 488 (green) were excited at 543 and 488 nm, respectively.

Results

GABA_BR activation enhances LTD_glu

We first assessed whether and how GABA_BR activation affects LTD of a glutamate-evoked current (LTD_glu) in cultured Purkinje cells. We iontophoretically applied glutamate to the dendrite of the examined cell for every 20 s and monitored glutamate-evoked inward currents in a perforated-patch whole-cell mode (holding potential, −70 mV) (Fig. 1A). After the basal recording, we gave six sets of depolarizing voltage steps (0 mV, 3 s) through the recording pipette in conjunction with glutamate iontophoresis (Fig. 1A). In the normal saline (control), the glutamate-evoked current underwent depression lasting over 50 min following the conjunctive stimuli (Fig. 1B and C; ‘Control’). The peak amplitude of the glutamate-evoked current was reduced to 76.2 ± 2.4% of the basal level (n = 7) at the end of the recording sessions (48–52 min after the conjunctive stimuli) (Fig. 1C). We found that GABA_BR activation enhanced LTD_glu, increasing the magnitude of depression (Fig. 1B and C). In the continuous presence of baclofen, a GABA_BR-selective agonist (3 μm), the peak amplitude was reduced to 56.2 ± 6.2% (n = 7) at the end of the recording sessions (Fig. 1C). This value was significantly smaller than the control (Fig. 1C). LTD_glu enhancement emerged immediately after the conjunctive stimuli. As early as 5 min after the conjunctive stimuli, the peak amplitude with baclofen (74.6 ± 4.8%, n = 7) was already significantly smaller than that with the normal saline (91.4 ± 3.0%, n = 7) (Fig. 1C). The early emergence of enhancement indicates that GABA_BR signalling may modulate the induction process of LTD_glu. Moreover, baclofen (3 μm, 15 min) did not reduce the peak amplitude of a glutamate-evoked inward current as its direct effect (102.7 ± 1.6%, n = 7; data not illustrated). Thus, LTD_glu enhancement cannot be ascribed to modulation of AMPA-type glutamate receptor channels by GABA_BR.

Figure 1. GABA_BR activation enhances LTD_glu in cerebellar Purkinje cells.

A, setting of the measurement of long-term depression of glutamate-evoked current (LTD_glu). Glutamate was iontophoretically applied to the first branching point of a dendrite of the examined cell (100–300 pA, 30 ms) for every 20 s. Glutamate-evoked inward currents were monitored under voltage clamp at −70 mV. To induce LTD_glu, six sets of depolarizing voltage steps (0 mV, 3 s) were given through the recording pipette in conjunction with glutamate iontophoresis. B and C, baclofen, a GABA_BR-selective agonist enhanced LTD_glu. B, each pair of traces indicates sample glutamate-evoked currents of a cell before and after the conjunctive depolarization/glutamate stimuli in the absence (‘Control’) or continuous presence of baclofen (3 μm). The time when the traces were recorded are indicated by the corresponding integers in the plots (C). C, each plot indicates the time course of the mean peak amplitude of the glutamate-evoked current in the absence or continuous presence of baclofen. Thick bar, timing of the conjunctive stimuli. Dot, data binned for each 5 min period. Error bar, ± s.e.m. *P < 0.05; **P < 0.01, Mann–Whitney's U test. These conventions of data acquisition and representation also apply to the following figures. D and E, GABA, a native GABA_BR agonist enhanced LTD_glu. D, sample responses of a cell before and after the conjunctive stimuli in the continuous presence of GABA (3 μm). E, time courses of the mean peak amplitudes of glutamate-evoked currents in the absence or continuous presence of GABA.

A previous study (Dittman & Regehr, 1997) suggests that the parallel fibre-innervated dendritic spines of Purkinje cells may receive 5–10 μm GABA spilt over from neighbouring inhibitory neurons' synapses in situ. In the continuous presence of a comparable dose (3 μm) of GABA, the peak amplitude of the glutamate-evoked current (66.9 ± 1.7%, n = 7 at 48–52 min after the conjunctive stimuli) was significantly smaller than the control (Fig. 1D and E). Thus, GABA_BR-mediated LTD_glu enhancement could occur under physiological conditions.

The βγ subunits of G_i/o protein may mediate GABA_BR-mediated LTD_glu enhancement

In the following sections, we explored mechanisms underlying GABA_BR-mediated LTD_glu enhancement. To test the involvement of G_i/o protein, the primary messenger coupled to GABA_BR, we assessed LTD_glu in Purkinje cells pretreated with pertussis toxin (PTX), a G_i/o protein inhibitor (500 ng ml⁻¹, > 16 h). This pretreatment is reported to eliminate a G_i/o protein-coupled inwardly rectifying K⁺ current in Purkinje cells (Tabata et al. 2005). In the normal saline, the conjunctive stimuli induced LTD_glu in the PTX-pretreated cells while the magnitude of LTD_glu (peak amplitude at 48–52 min after the conjunctive stimuli, 90.1 ± 4.6%, n = 5) was small compared with that in the untreated cells (Fig. 2A and B; cf. Fig. 1C). PTX pretreatment by itself might impede LTD_glu (see Discussion). Addition of baclofen (3 μm) to the saline did not increase the magnitude of LTD_glu in the PTX-treated cells (87.5 ± 4.3%, n = 6) (Fig. 2A and B). This result suggests the involvement of G_i/o protein in GABA_BR-mediated LTD enhancement.

Figure 2. G_i/o protein mediates GABA_BR-mediated LTD_glu enhancement.

A and B, PTX, a G_i/o protein inhibitor abolished baclofen-induced LTD_glu enhancement. A, each pair of traces indicates sample glutamate-evoked currents of a PTX (500 ng ml⁻¹, > 16 h)-pretreated cell before and after the conjunctive depolarization/glutamate stimuli in the absence (‘Control’) or continuous presence (‘Baclofen’) of baclofen (3 μm). B, time courses of the mean peak amplitudes of glutamate-evoked currents in the absence and continuous presence of baclofen. C and D, mastoparan, a G_i/o protein agonist, mimicked GABA_BR-mediated LTD_glu enhancement. C, each pair of traces indicates sample glutamate-evoked currents of a cell before and after the conjunctive depolarization/glutamate stimuli in the absence or continuous presence of mastoparan (1 μm). D, time courses of the mean peak amplitudes of glutamate-evoked currents in the absence (‘Control’) and continuous presence of mastoparan.

In supporting the above result, addition of mastoparan, a G_i/o protein agonist (1 μm) to the saline significantly increased the magnitude of LTD_glu (peak amplitude at 48–52 min after the conjunctive stimuli, 64.6 ± 2.4%, n = 5). This drug directly activates G_i/o protein by facilitating GTP binding to G_i/o protein without the aid of G protein-coupled receptors (Higashijima et al. 1988; Shpakov & Pertseva, 2006). Thus, activation of G_i/o protein is sufficient to enhance LTD_glu.

Upon activation, G_i/o protein is cleaved into the α and βγ subunits (G_αi/o and G_βγi/o, respectively). We determined the relative contribution of these subunits to LTD_glu enhancement. The main action of active G_αi/o is inhibition of adenylyl cyclase, which results in a reduction in cAMP level. We interfered with this action by adding forskolin (20 μm), a potent activator of adenylyl cyclase to the saline. This dose of forskolin is shown to indeed raise cAMP level in cultured Purkinje cells (Tabata et al. 2007). It is reported that in a neuron, a similar dose (10 μm) of forskolin activates adenylyl cyclase regardless of the presence or absence of micromolar levels of baclofen (Onali et al. 2003). Despite such an effect of forskolin, baclofen enhanced LTD_glu (peak amplitude at 48–52 min after the conjunctive stimuli, 55.7 ± 4.3%, n = 7) (Fig. 3A and B). This result suggests that G_βγi/o may primarily mediate LTD_glu enhancement.

Figure 3. G_αi/o does not mediate GABA_BR-mediated LTD_glu enhancement.

A and B, forskolin, an adenylyl cyclase activator did not abolish baclofen-induced LTD_glu enhancement. A, sample glutamate-evoked currents of a cell before and after the conjunctive depolarization/glutamate stimuli in the continuous presence of forskolin (20 μm) and baclofen (3 μm). B, time courses of the mean peak amplitudes of glutamate-evoked currents in the absence (‘Control’) and the continuous presence (‘Forskolin & Baclofen’) of forskolin (20 μm) and baclofen (3 μm).

GABA_BR-mediated LTD_glu enhancement is attributable to augmentation of mGluR1 signalling

The immediate emergence of GABA_BR-mediated LTD_glu enhancement (see above, Fig. 1A and B) indicates that GABA_BR signalling may modulate mechanisms underlying LTD induction. This possibility is further supported by experiments with pin-point application of baclofen at either the induction phase or expression/maintenance phase of LTD_glu. When applied only during the conjunctive stimuli, baclofen enhanced LTD_glu (Fig. 4A and B). By contrast, when applied after the conjunctive stimuli, baclofen failed to enhance LTD_glu (Fig. 4C and D).

Figure 4. GABA_BR activation during the induction phase of LTD_glu is sufficient for LTD_glu enhancement.

A and B, baclofen applied during the conjunctive depolarization/glutamate stimuli enhanced LTD_glu. A, sample glutamate-evoked currents of a cell before and after the conjunctive depolarization/glutamate stimuli with baclofen (3 μm) application during the stimuli. B, time courses of the mean peak amplitudes of glutamate-evoked currents with or without (‘Control’) baclofen application during the conjunction stimuli. C and D, baclofen applied after the conjunctive depolarization/glutamate stimuli did not enhance LTD_glu. C, sample glutamate-evoked currents of a cell before and after the conjunctive depolarization/glutamate stimuli with baclofen (3 μm) application after the stimuli. D, time courses of the mean peak amplitudes of glutamate-evoked currents with or without (‘Control’) baclofen application after the conjunction stimuli.

In cultured Purkinje cell preparations, the key factors for triggering LTD induction are glutamate-evoked mGluR1 signalling and depolarization-evoked Ca²⁺ influx (Linden et al. 1991; Ito, 2002). GABA_BR could enhance LTD_glu by augmenting either or both of these factors. We tested this possibility, using Ca²⁺ imaging. First, we examined the effect of baclofen on an mGluR1-mediated [Ca²⁺]_i rise evoked by DHPG, a group I mGluR agonist (5 μm, 10 s) (Fig. 5). In the recording conditions used here (the excitability of the cells was pharmacologically attenuated, see Methods), this response mainly reflects Ca²⁺ release from the intracellular stores (Sato et al. 2004) (see Fig. 7C). mGluR1 promotes the opening of Ca²⁺-permeable inositol trisphosphate receptor channel (IP₃R) on these stores via the G_q/11 protein-phospholipase C (PLC)-IP₃ cascade (Ito, 2002). Baclofen (3 μm) but not the normal saline augmented the peak amplitude (expressed in a change in F₃₄₀/F₃₈₀) of the [Ca²⁺]_i rise (154.9 ± 19.4%, n = 12 and 98.8 ± 8.0%, n = 11, respectively; Fig. 5A and D). Pretreatment with PTX (500 ng ml⁻¹, > 16 h) abolished this augmentation (Fig. 5B and D). These results suggest that GABA_BR potentiates mGluR1 signalling via G_i/o protein. The resting [Ca²⁺]_i level (mean F₃₄₀/F₃₈₀ over 5 s prior to DHPG application) changed little during application of the normal saline or baclofen (99.2 ± 1.1%, n = 11 and 92.0 ± 3.2%, n = 12, respectively; data not illustrated). Thus, the augmentation cannot be ascribed to [Ca²⁺]_i-dependent facilitation of Ca²⁺ release (Llano et al. 1994). Moreover, baclofen augmented the [Ca²⁺]_i rise in the presence of SKF96365, an antagonist against receptor-operated Ca²⁺ channels including TRPC1 (30 μm, Fig. 5C). The extent of this augmentation with SKF96365 (184.2 ± 36.0%, n = 8; Fig. 5D) was comparable with that in the normal saline (Fig. 5A and D). This result suggests that this augmentation is due largely to an increase of Ca²⁺ release from the intracellular stores; the contribution of an increase of Ca²⁺ influx through receptor-operated channels might be small although we do not exclude that the [Ca²⁺]_i rise includes such Ca²⁺ influx. These results suggest that GABA_BR signalling potentiates mGluR1 signalling via G_i/o protein.

Figure 5. GABA_BR activation augments mGluR1 signalling in a G_i/o protein-dependent manner.

A–D, baclofen induced a G_i/o protein-dependent augmentation of a DHPG-evoked [Ca²⁺]_i rise that reflects mGluR1-coupled intracellular Ca²⁺ store release. A–C, sample responses of cells untreated (A), pretreated with PTX (500 ng ml⁻¹, > 16 h; B), and perfused with SKF96365 (30 μm, C). Each pair of traces indicates sample DHPG (5 μm, 10 s)-evoked responses of a cell before (‘Basal’) and after (‘Test’) a 12 min application of the normal saline (‘Control’) or baclofen (3 μm). Arrowhead, DHPG onset. Vertical scale bar, change in F₃₄₀/F₃₈₀. D, mean peak amplitudes of the DHPG-evoked [Ca²⁺]_i rise after application of the normal saline or baclofen (expressed as percentage of the basal levels).

Figure 7. mGluR1-mediated Ca²⁺ release from the intracellular stores is important for inducing LTD_glu.

A and B, thapsigargin (endoplasmic reticular Ca²⁺-ATPase inhibitor), which was expected to deplete Ca²⁺ in the intracellular stores, abolished LTD_glu. A, sample glutamate-evoked currents of a cell before and after the conjunctive depolarization/glutamate stimuli in the continuous presence of thapsigargin (1 μm). B, time courses of the mean peak amplitudes of glutamate-evoked currents in the absence and continuous presence of thapsigargin. C, thapsigargin indeed abolished mGluR1-mediated Ca²⁺ release from the intracellular stores. Traces indicate sample DHPG (5 μm, 10 s)-evoked [Ca²⁺]_i rises of a cell before and after application of thapsigargin (1 μm, 12 min). Similar results were obtained from 4 cells. D, tharpsigargin did not abolish depolarization-evoked Ca²⁺ influx. Traces indicate sample K⁺ (75 mm, 1 s)-evoked [Ca²⁺]_i rises of a cell before and after application of thapsigargin (1 μm, 12 min). Similar results were obtained from 9 cells.

Second, we examined the effect of baclofen on a [Ca²⁺]_i rise evoked by application of K⁺-rich saline ([K⁺] = 75 mm, 1 s) (Fig. 6). This response was susceptible to ω-agatoxin IVA (100 nm), an antagonist against the dominant type (P/Q-type) of voltage-gated Ca²⁺ channels in Purkinje cells (n = 6, data not illustrated), suggesting that this response reflects Ca²⁺ influx through voltage-gated channels. Baclofen did not augment the K⁺-evoked [Ca²⁺]_i rise but rather reduced it (peak amplitude, 79.0 ± 2.8%, n = 15) (Fig. 6A and C). Pretreatment with PTX (500 ng ml⁻¹, > 16 h) abolished the baclofen-induced reduction of the K⁺-evoked [Ca²⁺]_i rise (Fig. 6B and C).

Figure 6. GABA_BR activation does not augment depolarization-evoked Ca²⁺ influx.

A–C, baclofen did not augment but rather reduced a [Ca²⁺]_i rise evoked by the K⁺-rich saline ([K⁺] = 75 mm, 1 s). A and B, sample responses of untreated or PTX (500 ng ml⁻¹, > 16 h)-pretreated cells. Each pair of traces indicates sample K⁺-evoked [Ca²⁺]_i rises of a cell before and after a 12 min application of the normal saline (‘Control’) or baclofen (3 μm). Arrowhead, K⁺-rich saline onset. C, mean peak amplitudes of the DHPG-evoked [Ca²⁺]_i rise after application of the normal saline or baclofen (expressed as percentage of the basal levels).

The results in Fig. 5 imply that GABA_BR-mediated LTD_glu enhancement could be due partly to augmentation of mGluR1-coupled Ca²⁺ release from the intracellular stores. This notion would be in a good agreement with previous reports that mGluR1-coupled Ca²⁺ release is essential for inducing LTD in cerebellar slices (e.g. Inoue et al. 1998). However, there is a report (Narasimhan et al. 1998) that this is not the case in certain cultured Purkinje cell preparations. We checked the dependence of LTD_glu upon mGluR1-coupled Ca²⁺ release in our cultured Purkinje cell preparations (Fig. 7). We depleted Ca²⁺ from the intracellular stores by adding thapsigargin, an endoplasmic reticular Ca²⁺-ATPase inhibitor, to the saline. Under this condition, the conjunctive stimuli failed to induce LTD_glu (105.5 ± 7.0% at 48–52 min after the conjunctive stimuli, n = 5; Fig. 7A and B). Thapsigargin (1 μm, 12 min) suppressed the DHPG (5 μm, 10 s)-evoked [Ca²⁺]_i rise as its direct effect (peak amplitude, 13.6 ± 3.6%, n = 4; Fig. 7C), confirming that thapsigargin indeed abolished mGluR1-mediated Ca²⁺ release from the intracellular stores. On the other hand, thapsigargin (1 μm, 12 min) little affected the K⁺ (75 mm, 1 s)-evoked Ca²⁺ influx (98.8 ± 8.0%, n = 9, Fig. 7D). These results suggest that mGluR1-coupled Ca²⁺ release from the intracellular stores is important for inducing LTD_glu at least in our preparations.

The result in Fig. 3 suggests that G_βγi/o mediates GABA_BR-mediated LTD_glu enhancement. In various cell types, G_βγi/o augments G_q/11 protein-coupled Ca²⁺ release from the intracellular stores by priming some PLC isoforms including PLCβ3 (for review, Park et al. 1993; Selbie & Hill, 1998). Such priming could underlie GABA_BR-mediated augmentation of mGluR1-coupled Ca²⁺ release because Purkinje cells more or less express PLCβ3 in vivo (Kano et al. 1998; Nomura et al. 2007). We checked whether cultured Purkinje cells express PLCβ3 by immunohistochemistry. Cultured Purkinje cells were identified by immunoreactivity against calbindin (green in Fig. 8A1 and red in Fig. 8B1). For all the calbindin-positive cells, PLCβ3 immunoreactivity was found throughout the somata, axons and dendrites (n = 34, Fig. 8A). A similar subcellular distribution is seen in PLCβ3-positive Purkinje cells in vivo (Nomura et al. 2007). It is reported that a subpopulation of Purkinje cells in vivo express a low level of PLCβ3 while complementarily expressing a high level of PLCβ4 (Sarna et al. 2006) that is insensitive to G_βγi/o (Park et al. 1993). Such heterogeneity of PLC isoforms was absent in the culture system. All of the calbindin-positive cells displayed PLCβ4 immunoreactivity (n = 42, Fig. 8B). PLCβ4 immunoreactivity was confined to the somata and dendrites as shown in vivo (Nakamura et al. 2004). Furthermore, all of the cells showing PLCβ3 immunoreactivity also displayed PLCβ4 immunoreactivity in the culture system (n = 28, Fig. 8C). These results suggest that priming of PLCβ3 by G_βγi/o may occur in most cultured Purkinje cells.

Figure 8. Cultured Purkinje cells express both PLCβ3 and β4.

A and B, Purkinje cells (14 days old in vitro) identified by calbindin (‘CB’) immunoreactivity (green in A1 and red in B1) are immunostained with anti-PLCβ3 (red in A2) and anti-PLCβ4 (green in B2) antibodies. Arrows and arrowheads, somata and axons, respectively. Scale bar, 20 μm. The almost complete overlap of PLCβ3 and calbindin signals of the cell (superimposed images in A3) suggests that PLCβ3 is expressed throughout the soma, dendrites and axon of the Purkinje cell. By contrast, PLCβ4 immunoreactivity is found in the soma and dendrites but not in the axon (B3). C, a Purkinje cell (14 days old in vitro) displays both PLCβ3 and PLCβ4 immunoreactivities.

Co-localization of GABA_BR and mGluR1 in cultured Purkinje cells

To explore the possible site for the functional interaction of GABA_BR and mGluR1 signalling cascades, we immunostained cultured Purkinje cells with antibodies against these receptors. Calbindin-positive cells had clear immunoreactivities for GBR1 (Fig. 9A) and mGluR1 (Fig. 9B). Higher levels of immunoreactivities were found at the dendritic spines of these cells (Fig. 9A4–6 and B4–6). Co-immunostaining with the anti-GBR1 and anti-mGluR1 antibodies shows that the subcellular distributions of GABA_BR and mGluR1 overlap each other (Fig. 9C). These results suggest that co-localization of these receptors at the dendritic spines seen in situ (Lujan et al. 1997; Kulik et al. 2002) is preserved in our cultured cell preparations.

Figure 9. Co-localization of GABA_BR and mGluR1 in cultured Purkinje cells.

A and B, Purkinje cells (20 days old in vitro) identified by calbindin (‘CB’) immunoreactivity (A1 and B1) are immunostained with anti-GBR1 (A2) and anti-mGluR1 (B2) antibodies (A3 and B3, superimposed images). Arrows and arrowheads, somata and axons, respectively. GBR1 and mGluR1 immunoreactivities are detected at most dendritic spines (double arrow heads in the close-up images, A5 and 6 and B5 and 6). C, co-immunostaining of Purkinje cells (20 days old in vitro) with the anti-GBR1 (C1) and anti-mGluR1 (C2) antibodies shows the overlapping subcellular distribution of these antigens (C3, superimposed image). Co-localization of these antigens is found at dendritic spines (double arrow heads in the close-up images, C4–C6). Scale bars, 10 μm.

GABA_BR enhances LTD in situ

Lastly, we tested whether GABA_BR indeed enhances LTD of parallel fibre–Purkinje cell synaptic transmission, using cerebellar slices. The extracellular fluid in the cerebellar slices might contain GABA because the cerebrospinal fluid contains GABA and cerebellar interneurons synaptically release GABA (see Discussion). We examined the effect of CGP55845, a GABA_BR-selective antagonist that is expected to interfere with GABA action on GABA_BR. We induced LTD with conjunctive stimuli (30 sets at 0.5 Hz) each of which consisted of a train of electrical pulses to the parallel fibres and a depolarizing voltage step to the Purkinje cell (Fig. 10A). The electrical pulses were given in relatively high-frequency bursts (100 Hz) to facilitate GABA release from interneurons activated by the parallel fibre stimuli (Dittman & Regehr, 1997; Hirono et al. 2001). Under the control condition, parallel fibre–Purkinje cell EPSC was depressed over 30 min after the conjunctive stimuli (Fig. 10B). The continuous presence of CGP55845 (2 μm) abrogated initial depression after the conjunctive stimuli and slowed down the development of LTD (Fig. 10B). For both conditions, LTD appeared to develop to the maximal extents at 17–21 min after the conjunctive stimuli (mean amplitudes of parallel fibre–Purkinje cell EPSC at this period were not significantly different those at 27–31 min after the conjunctive stimuli, P > 0.1, paired Student's t test). Over 30–31 min after the conjunctive stimuli, the mean relative amplitude of parallel fibre–Purkinje cell EPSC with CGP55845 (81.1 ± 5.9% of the basal level, n = 5) was significantly larger than that of the control (61.1 ± 4.4%, n = 5) (Fig. 10B). This result indicates that GABA_BR-mediated enhancement may occur for LTD in situ.

Figure 10. GABA_BR-mediated LTD enhancement may occur in situ.

A, as schematically shown, we measured LTD of parallel fibre–Purkinje cell (PF-PC) EPSC in cerebellar slices. Whole-cell voltage-clamp recording was made from a Purkinje cell in the lamina VII or VIII. Parallel fibres were stimulated by electrical pulses delivered through a glass pipette positioned within a 1/3-part amid the molecular layer. LTD was induced with 30 sets of conjunctive stimuli (repeated at 0.5 Hz) each of which consisted of 10 pulses to the parallel fibres (100 Hz) and a depolarizing voltage step (+20 mV, 100 ms) to the Purkinje cell through the recording pipette. B, relative amplitude of parallel fibre–Purkinje cell (PF-PC) EPSC in the absence (control, n = 5) or continuous presence (n = 5) of CGP55845, a GABA_BR-selective antagonist (2 μm) plotted against time after the conjunctive stimuli (‘Conj.’). CGP55845 was included in the bath solution from at least 30 min before the conjunctive stimuli. Dot and error bar, the mean and s.e.m. of the data for each 2 min period. Mean value over 29–30 min after the conjunctive stimuli significantly differs between the two conditions (*P < 0.05, Mann–Whitney's U test).

Discussion

We found that GABA_BR activation enhanced LTD_glu, increasing the magnitude of depression (Fig. 1B and C). GABA_BR-mediated LTD_glu enhancement emerged immediately after the induction phase of LTD_glu (Fig. 1C). GABA_BR agonist application within the period of the induction stimuli was sufficient to enhance LTD_glu (Fig. 4). These results suggest that GABA_BR signalling modulates mechanisms underlying LTD_glu induction (Fig. 11). In cultured Purkinje cell preparations, glutamate-evoked mGluR1 signalling and depolarization-evoked Ca²⁺ influx are known as the key factors for triggering LTD induction (Fig. 11A) (Linden et al. 1991; Ito, 2002). GABA_BR activation augmented DHPG-evoked mGluR1 signalling (Fig. 5A and D; for details, see below) while reduced K⁺-evoked Ca²⁺ influx (Fig. 6A and C). A previous study (Mintz & Bean, 1993) shows that GABA_BR negatively regulates the P/Q-type channels in Purkinje cells. These results suggest that GABA_BR signalling augments mGluR1 signalling in Purkinje cells and thereby facilitate LTD_glu induction (Fig. 11B).

Figure 11. Possible mechanisms underlying GABA_BR-mediated LTD_glu enhancement.

A, in the cultured Purkinje cells, glutamate-evoked mGluR1 signalling including Ca²⁺ release from the intracellular stores and depolarization-evoked Ca²⁺ influx are key factors for inducing LTD_glu. B, GABA and baclofen promotes activation of G_i/o protein by GABA_BR. Activated G_i/o protein is cleaved into the α and βγ subunits; the latter may be more important for LTD_glu enhancement than the former. The βγ subunits may augment mGluR1-mediated signalling presumably by acting on PLC, and this may result in LTD_glu enhancement. For further explanation, see Discussion. AMPAR, AMPA-type ionotoropic glutamate receptor; DAG, diacylglycerol; ER, endoplasmic reticulum; IP₃, inositol trisphosphate; IP₃R, IP₃ receptor; PKC, protein kinase C; PLC, phospholipase C; VGCC, voltage-gated calcium channel.

Inhibition of G_i/o protein by PTX abolished GABA_BR-mediated LTD_glu enhancement (Fig. 2A and B). PTX did not appear to exert an inhibitory effect directly on glutamate-evoked currents in Purkinje cells (see online supplemental material, Supplemental Fig. 1). Thus, the effect of PTX on LTD_glu enhancement cannot be ascribed to a PTX-induced reduction of glutamate-evoked currents which would occlude LTD_glu. Activation of G_i/o protein by mastoparan mimicked LTD_glu enhancement (Fig. 2C and D). Activation of adenylyl cyclase by forskolin did not abolish GABA_BR-mediated LTD_glu enhancement (Fig. 3). Forskolin at the dose used is thought to increase cAMP level regardless of the presence or absence of micromolar levels of baclofen (Onali et al. 2003). Thus, this manipulation should cancel the inhibitory action of G_αi/o on adenylyl cyclase. These results raise the possibility that G_βγi/o primarily mediates GABA_BR-mediated LTD_glu enhancement (Fig. 11B). There are several possible targets for G_βγi/o. In various cell types, G_βγi/o are known to exert a synergistic action with G_q/11 protein on some PLC isoforms including PLCβ3 (Park et al. 1993). All of the cultured Purkinje cells expressed PLCβ3 (Fig. 8). Thus, in most of the cases in this study, such a synergistic action on PLC should occur, and this might augment downstream mGluR1 signalling mediated by PLC (Fig. 11B). Moreover, a study (Zeng et al. 2003) has reported a possibility that G_βγi binds to and exerts a direct agonistic action on type-1 IP₃R, which is important for cerebellar LTD induction (Inoue et al. 1998).

Depletion of Ca²⁺ in the intracellular stores by thapsigargin abolished mGluR1-mediated [Ca²⁺]_i rises (Fig. 7C). In the presence of this agent, the conjunctive depolarization/glutamate stimuli did not induce LTD_glu (Fig. 7A and B). These results suggest that mGluR1-coupled Ca²⁺ release from the intracellular stores is essential for inducing LTD_glu in our cultured cell preparations (Fig. 11A) as reported in cerebellar slice preparations (Inoue et al. 1998). Thus, the GABA_BR-mediated augmentation of mGluR1-coupled Ca²⁺ release may contribute to LTD_glu enhancement (Fig. 11B). A resultant elevation of the cytoplasmic [Ca²⁺] may promote activation of PKC (Ito, 2002), which in turn facilitates endocytosis of AMPA-type glutamate receptors (Matsuda et al. 2000; Chung et al. 2003). A recent study (Hansel et al. 2006) shows that genetic or pharmacological suppression of α-calcium/calmodulin-dependent kinase II, which is activated at a high [Ca²⁺]_i, switches the polarity of plasticity from depression to potentiation at parallel fibre–Purkinje cell synapses. mGluR1-coupled Ca²⁺ release might ensure induction of LTD_glu by boosting the activity of this kinase.

PTX pretreatment by itself attenuated the magnitude of LTD_glu (Fig. 2A and B) as previously reported (Ito & Karachot, 1990). A small amount of G proteins may be active in a constitutive manner due to the ligand-independent basal activity of G protein-coupled receptors (Smit et al. 2007). Thus, G_i/o protein could always enhance LTD_glu slightly unless inhibited by PTX.

Previous studies (Hirono et al. 2001; Tabata et al. 2004) show that in Purkinje cells, GABA_BR activation augments an mGluR1-mediated inward cation current (slow EPSC at parallel fibre–Purkinje cell synapses) by raising the ligand sensitivity of mGluR1 in a G_i/o protein-independent manner and by potentiating the mGluR1 signalling cascade in a G_i/o protein-dependent manner. GABA_BR-mediated LTD_glu enhancement resembles the latter action of GABA_BR in terms of G_i/o protein dependence. However, activation of G_i/o-coupled receptors other than GABA_BR (e.g. A1 adenosine receptor; Tabata et al. 2007) fails to augment the slow EPSC (Hirono et al. 2001), suggesting that G_i/o protein activation is not sufficient to enhance this response. This contrasts with G_i/o protein activation being sufficient to enhance LTD_glu (Fig. 2C and D). LTD_glu enhancement and cation current augmentation may involve different mechanisms.

Co-localization of GABA_BR and mGluR1 at the dendritic spines of Purkinje cells is seen commonly in culture (Fig. 9) and in situ (Lujan et al. 1997; Kulik et al. 2002). Therefore, GABA_BR-mediated enhancement described above could possibly occur for mGluR1-mediated LTD at real parallel fibre–Purkinje cell synapses. We confirmed this possibility, using cerebellar slices. A GABA_BR-selective antagonist reduced the magnitude of LTD induced by conjunctive parallel fibre stimulation and postsynaptic depolarization (Fig. 10). In mammals, the cerebrospinal fluid contains a few tens of nanomolar of GABA in vivo (Bohlen et al. 1979). Some fraction of this GABA could remain in the cerebellar slices. Parallel fibre stimulation may further increase ambient GABA concentration in the cerebellar slices by facilitating GABA release from interneurons activated by the parallel fibre inputs. Released GABA may spill over from the synaptic clefts and act on neighbouring parallel fibre–Purkinje cell synapses (Dittman & Regehr, 1997; Hirono et al. 2001). CGP55845 might interfere with GABA action on GABA_BR in the Purkinje cells (Fig. 11). The effect of CGP55845 on LTD is not attributable to its action on presynaptic GABA_BR. GABA_BR mediates presynaptic inhibition of parallel fibre–Purkinje cell synapses (Dittman & Regehr, 1996). A GABA_BR antagonist is shown to relieve this inhibition (Hirono et al. 2001), and this would rather enhance LTD.

The concentration of GABA contained in the cerebrospinal fluid of healthy animals (see above) is high enough to activate GABA_BR considerably (GABA_BR's affinity for GABA, ∼1 μm; Sodickson & Bean, 1996). GABA spilt over from the interneurons' synapses might further facilitate GABA_BR activation in Purkinje cells. Therefore, GABA_BR-mediated enhancement of cerebellar LTD could occur under physiological conditions in vivo. Our findings demonstrate a novel mechanism that would facilitate cerebellar motor learning.

Supplemental material

Online supplemental material for this paper can be accessed at:

http://jp.physoc.org/cgi/content/full/jphysiol.2007.141010/DC1 and http://www.blackwell-synergy.com/doi/suppl/10.1113/jphysiol.2007.141010