INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Presynaptic inhibition of an inhibitory synapse is an intriguing method for creating local negative feedback or gain control mechanism, and it appears to play a significant role in the cerebellar cortex. Here inhibitory interneurons (molecular layer interneurons or MLIs) and the principal neurons (Purkinje cells, PC) are driven by a common input, the parallel fibers. In such a circuit, presynaptic inhibition between MLIs can theoretically switch their effect on Purkinje cells from significant inhibition to robust excitation by a mechanism of disinhibition.

The axons of the MLIs are all organized parasagittally. Due to this spatial arrangement, stimulating a beam of parallel fibers generates a lateral inhibitory response at both sides of the activated beam. Using optical imaging of electrical activity in an isolated cerebellum, we have previously demonstrated that, in a paired-pulse paradigm, the second lateral inhibitory response is markedly attenuated (Cohen and Yarom 2000). Moreover, when two spatially separated beams of parallel fibers are activated sequentially, the second lateral inhibitory response in the area between the beams is also attenuated even though it is induced by a different population of MLIs. This led us to propose that the MLI axons exert presynaptic inhibition on each other.

Presynaptic inhibition is most commonly attributed to the activation of GABA_B receptors. This member of the GABA receptor family is a metabotropic receptor that either blocks Ca²⁺ channels or activates K⁺ channels (Misgeld et al. 1995). The Ca²⁺ blocking type appears to act presynaptically, whereas the receptor that activates K⁺ conductance can be found at postsynaptic sites (Yamada et al. 1999). Two types of GABA_B isoforms were found, and it has been suggested that presynaptic GABA_B receptors contain GABA_BR1a subunits while the postsynaptic receptors contain GABA_BR1b subunits (Billinton et al. 1999; Bischoff et al. 1999; Kaupmann et al. 1998b).

An extensive presence of GABA_B receptors in the molecular layer of the cerebellar cortex has been demonstrated morphologically using autoradiography, in situ hybridization, and immunohistochemistry, mainly on PC dendrites and on parallel fiber terminals (Albin and Gilman 1989; Billinton et al. 1999; Durkin et al. 1999; Fritschy et al. 1999; Kaupmann et al. 1998a; Martinelli et al. 1992; Turgeon and Albin 1993). Electrophysiological recording have shown that the GABA_B receptors located on Purkinje cell activate K⁺ channels (Vigot and Batini 1997), while those on the parallel fibers decrease synaptic release either by blocking Ca²⁺ channels or by blocking processes downstream of Ca²⁺ entry (Dittman and Regehr 1997). The existence of GABA_B receptors on MLIs is still a matter of debate. While Fritschy and colleagues (1999) reported that MLIs are devoid of either GABA_BR1a or GABA_BR1b subunits, Bischoff et al. (1999), using in situ hybridization, have reported that cerebellar stellate cells do express GABA_BR1a subunit.

Here we show that GABA_B agonists have a profound effect on MLI both on membrane properties and synaptic transmission, indicating the presence of both pre- and postsynaptic GABA_B receptors. These effects are manifested within the normal range of activity of these neurons and appear to be mediated by blocking Ca²⁺ channels. The functional implications of these receptors in shaping MLI activity are discussed.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Slice preparation

Sagittal slices (300 µm thick) were prepared from the vermis of a guinea pig cerebellum. Guinea pigs (80-120 g; 7-17 days) were anesthetized intraperitoneally with pentobarbital sodium (60 mg/kg) and perfused through the heart with 100 ml of cold (0-1°C) oxygenated physiological solution (containing in mM: 124 NaCl, 5 KCl, 1.3 MgSO₄, 1.2 KH₂PO₄, 26 NaHCO₃, 10 glucose, and 2.4 CaCl₂; pH 7.4; aerated with 95% O₂-5% CO₂). Following decapitation, the cerebellum was quickly removed and sliced (Campden Instruments LTD 752 M vibroslice) in cold sucrose solution (containing in mM: 5 KCl, 1.3 MgSO₄, 1.2 KH₂PO₄, 26 NaHCO₃, 10 glucose, 2.4 CaCl₂, and 124 sucrose). The slices were incubated in the sucrose solution at room temperature for 20 min, after which the sucrose solution was slowly replaced by normal physiological solution over a period of 1 h. This procedure was found to be crucial for the viability of the stellate and basket cells. Slices were kept at room temperature in oxygenated physiological solution until transferred to the recording chamber.

Solutions and drugs

During the various experiments, the following drugs were added to the physiological solution to reach a final concentration of 0.1 µM TTX (Molecular Probes); 1, 10, or 100 µM baclofen (Sigma); 50 µM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; RBI); 50 µM bicuculline (Sigma); and 1 µM CGP 55845A and 10 µM CGP 44533 (kindly provided by Novartis Pharma AG, Basel). Slices were perfused for 10 min with the solution containing the drug prior to the examination of its effect. No Ca²⁺ solution contained (in mM) 134 NaCl, 6 KCl, 1.3 MgSO₄, 26 NaHCO₃, 10 glucose, and 5 CoCl₂. The pipette solution contained (in mM) 140 K-gluconate, 4 NaCl, 0.5 CaCl₂, 5 EGTA, 3 Mg-ATP, 0.4 GTP, and 10 HEPES (pH 7.2). In the high Cl pipette solution, K-gluconate was partially replaced by KCl to reach the final desired Cl concentration, usually of 30 mM. When QX-314 was used, it was added to the pipette solution at a concentration of 0.1 mM.

Electrophysiological recordings

The recording chamber, mounted on an upright microscope stage (Zeiss Axioskop), was maintained at a constant temperature of 30°C by a temperature control unit and was continuously perfused with aerated physiological solution. The molecular layer interneurons in the slice were readily identified using infrared differential interference contrast optics, and whole cell patch recordings were quickly achieved. The patch pipettes were pulled on a Narishige pp-83 puller and had a DC resistance of 10-12 M. Recordings were made with an Axoclamp 2B for intracellular current-clamp experiments and extracellular recordings and with Axopatch 1D for voltage clamp (Axon Instruments). Two amplifiers were used for simultaneous recordings from two cells. For extracellular recordings, the patch pipette, which was placed near the cell membrane without generating a seal, was filled with physiological solution. Data were stored on videocassette (Neurocorder DR-484) for off-line analysis. Data were sampled at 5-10 kHz, using National Instruments A/D board derived by software program written in LabVIEW.

Optical imaging and the isolated cerebellum

Optical measurement from the isolated guinea pig cerebellar preparation has been described in detail elsewhere (Cohen and Yarom 1999). Briefly, the intact cerebellum and the brain stem are removed from the animal, and a cannula is inserted into one of the vertebral arteries. Physiological solution is perfused via the vertebral artery at a rate of 0.5 ml/min using a peristaltic pump. The intravascular solution consists of (in mM) 124 NaCl, 5 KCl, 1.3 MgSO₄, 1.2 KH₂PO₄, 26 NaHCO₃, 10 glucose, 2.4 CaCl₂, and 5% dextran. A similar solution without dextran is used for the external solution. The preparation was maintained at 28°C. CGP 55845A was added to the external solution in a final concentration of 2 µM. The voltage-sensitive dye RH-414 (2 µg/µl) was injected into one of the cerebellar folia using a glass micropipette. Optical signals were monitored by 464 photodiodes organized in a hexagonal array. Each element of the array detects light from a surface of 25 µm × 25 µm using a ×40 objective. The signals were amplified in two AC-coupled stages with a time constant of 200 ms and then sampled and digitized with 12-bit accuracy at maximal resolution of 180 µs (Microstar, DAP 3400a).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Baclofen decreases spontaneous activity in MLI

To functionally verify the presence of GABA_B receptors in molecular layer interneurons (MLIs), we first examined the effect of the specific GABA_B agonist, baclofen, on MLI activity. Bath application of baclofen greatly decreased the spontaneous firing rate of these interneurons, as shown by extracellular recordings in parasagittal slices, using a loose cell-attached configuration (Fig. 1) (see also Mann-Metzer and Yarom 1999). In the example depicted in Fig. 1A, the application of baclofen (10 µM) increased the average interspike interval (ISI) from 101 ± 15 to 182 ± 41 (SD) ms, an increase of 79%, which is clearly seen in the ISI histogram (Fig. 1A, middle panel). This effect of baclofen was blocked by 1 µM CGP 55845A (Fig. 1A, bottom panel). Moreover, the average ISI in the presence of the blocker was lower than under control conditions (74 ± 14 ms). This overshoot of the firing rate suggests that the GABA_B receptors are tonically active due to endogenous GABA released by the spontaneously active neurons. This decrease in ISI was also observed when CGP 55845A alone was added to the bath solution (see Table 1).

View larger version (24K): [in this window] [in a new window]   Fig. 1. Baclofen decreases spontaneous firing rate and synaptic activity in molecular layer interneurons (MLIs). A: extracellular recording of spontaneous action potentials (left) and their corresponding interspike interval (ISI) histograms (right) under control conditions (top), in the presence of 10 µM baclofen (middle) and after the addition of 1 µM CGP 55845A (bottom). The same cell was recorded in all 3 conditions, and the analysis was performed on 60 s of continues recordings. Note the different scales of the Y-axis. B: intracellular recordings of spontaneous synaptic potentials (left) and their corresponding inter-postsynaptic potential (PSP) interval histograms (right) under control conditions (top), in the presence of 10 µM baclofen (middle) and after the addition of 2 µM CGP 55845A (bottom). Membrane potential was hyperpolarized, and a high Cl intracellular solution was used to reverse and amplify the inhibitory synaptic potentials. The same cell was recorded in all 3 conditions, and the analysis was performed on 60 s of continuous recordings. Note the different scales of the Y-axis.

Intracellular recordings in the whole cell configuration showed a high frequency of spontaneous synaptic potentials (Fig. 1B, top) (see also Häusser and Clark 1997; Kondo and Marty 1998; Mann-Metzer and Yarom 1999). As there is very little excitatory input from the parallel fibers in this preparation (Häusser and Clark 1997), the vast majority of the synaptic potentials are GABAergic. A modified intracellular solution containing 30 mM Cl was used to amplify these potentials. Figure 1B shows the profound decrease in synaptic activity caused by baclofen (10 µM). In this example, the average inter-postsynaptic potential (PSP) interval increased by 158% from 97 ± 85 ms (top panel) to 250 ± 220 ms (middle panel). The addition of CGP 55845A (2 µM) increased the synaptic activity to 257% of the control rate. An increase in PSP frequency was also seen when CGP 55845A alone was added to the bath solution (see Table 1).

Table 1 summarizes the average change in firing rate and synaptic potentials rates induced by GABA_B receptors agonists and antagonists. Both effects of baclofen showed dose dependence; 1 µM of baclofen was sufficient to induce 37 ± 15% reduction in firing rate and 23 ± 17% reduction in PSP frequency, while 10 µM baclofen induced a reduction of 55 ± 19% and 82 ± 15%, respectively. Adding CGP 55845A, in addition to baclofen, increased the firing rate by 229 ± 71% and the PSP rate by 1,102 ± 686% in comparison to the activity in the presence of baclofen. Relative to the control values CGP 55845A, either alone or in the presence of baclofen, induced an increase of 26 ± 18% and 39 ± 38% in firing and the PSP rates, respectively.

Undoubtedly, in a network of interconnected inhibitory neurons, a decrease in firing rate is bound to decrease the synaptic activity. However, the decrease in firing rate is smaller than the decrease in synaptic activity and therefore appears to be insufficient to fully account for it. Moreover, the reduction in synaptic activity cannot explain the decrease in firing rate, since these synapses are inhibitory, and, as will be shown below, their blockade increase the firing rate. We must therefore infer the existence of two mechanisms for GABA_B operation: a postsynaptic mechanism that reduces the firing rate and a presynaptic mechanism that decreases synaptic transmission.

Postsynaptic effect of GABA_B

The postsynaptic effect of baclofen was confirmed and characterized in the presence of bicuculline. Figure 2A shows the ISI histogram of MLI under three experimental conditions. Under control conditions, the average ISI was 78 ms with a rather skewed scatter, resulting in a standard deviation of 47 ms and the coefficient of variation (CV) of 0.6. When bicuculline was added to the bath, blocking the inhibitory GABA_A-mediated synaptic potentials, the ISI decreased to 59 ms with less dispersion, as indicated by the lower standard deviation of 18 ms and CV of 0.3. On co-application of baclofen, ISI increased to 99 ms but kept its regularity as indicated by the intermediate standard deviation of 33 ms and identical CV of 0.3 (the average reduction in firing rate compared with the rate in the presence of bicuculline was 28%, n = 3). Thus the baclofen-induced decrease in firing rate is independent of GABA_A-mediated synaptic transmission. As this is by far the major input to these cells in this preparation, and since baclofen induces similar decrease in firing rate (44%) in the presence of CNQX (results not shown), we conclude that baclofen has a postsynaptic effect on MLI, which is manifested as a decrease in spontaneous firing rate.

View larger version (15K): [in this window] [in a new window]   Fig. 2. The postsynaptic effect of baclofen. A: ISI histograms under control conditions (top), in the presence of bicuculline (100 µM; middle), and after the addition of baclofen (100 µM; bottom). The same cell was recorded in all 3 conditions, and the analysis was performed on 300 s of continuous recordings. Note the different scales of the Y-axis. B: the changes in membrane potential (top) and membrane current (middle and bottom) on application of 100 µM baclofen (open arrow), 1 µM CGP 55845A (closed arrow) or 2 mM Ba2+ (gray arrow at the bottom trace). Each dot represents the value averaged over 1 s. Once a drug was applied, it remained in the bath throughout the experiment. C: ISI histograms under control conditions (top) and in the presence of 10 µM CGP 44533 (bottom). The same cell was recorded in the 2 conditions, and the analysis was performed on 60 s of continuous recordings. Note the different scales of the Y-axis.

The ionic mechanism that underlies the postsynaptic effect of GABA was studied using current- and voltage-clamp techniques. The experiments were carried out in the presence of TTX to stabilize the otherwise fluctuating membrane potential and to avoid presynaptic influences. As shown in Fig. 2B (top trace), baclofen hyperpolarized the membrane potential from 40 to 48 mV (100 µM, open arrow). A similar hyperpolarization was seen in 80% of the cells (n = 15). The average change in membrane potential in those cells showing a significant response was 6 ± 2 mV. On adding the specific GABA_B antagonist CGP 55845A (1 µM, closed arrow), the membrane potential depolarized to almost the control level. Similar results were obtained using voltage-clamp methodology (Fig. 2B; middle trace). Baclofen induced a net outward current of about 6 pA (open arrow), which was blocked by application of CGP 55845A (1 µM, closed arrow). An average outward current of 7.8 ± 3 pA was calculated for four cells. These results show that the postsynaptic effect of baclofen is associated with either the induction of an outward current or reduction in a persistent inward current.

Further experiments were carried out to determine whether the effects were caused by K⁺ channels activation or blocking of Ca²⁺ channels as is known for GABA_B receptors (Misgeld et al. 1995). Given the relatively large driving force of Ca²⁺, it would be expected that a rather small reduction in calcium conductance would be sufficient to induce the observed hyperpolarization. If a K⁺ current is involved, one would expect to see a larger increase in conductance, given its lower driving force. We measured the conductance using either current pulses in current-clamp or voltage steps with voltage-clamp techniques, but we detected no conductance change during the application of baclofen. Nor could we detect changes in the shape of the inhibitory postsynaptic potentials (IPSPs; a measure of remote conductance changes) following baclofen application (not shown).

Ba²⁺ is known to block various K⁺ currents including G_irK and to enhance inward Ca²⁺ currents. Adding 2 mM Ba²⁺ to the physiological solution induced an inward current (Fig. 2B; bottom trace, gray arrow). Baclofen still induced a 9 pA net outward current that was blocked by CGP 55845A (open and closed black arrows, respectively). In addition, baclofen failed to induce hyperpolarization in the absence of extracellular calcium (n = 2; results not shown).

Finally, we studied the effect of CGP 44533, a specific GABA_B agonist suggested to act on receptors that inhibit Ca²⁺ channels (Yamada et al. 1999). All experiments (n = 3) showed a decrease in firing rate, resembling the effect of baclofen (Fig. 2C).

The results of these three experiments (the inability to detect any conductance change during baclofen application, the maintained action of baclofen in the presence of Ba²⁺, its elimination in the absence of external Ca²⁺, and the ability of CGP 44533 to mimic baclofen application) all point to the conclusion that baclofen decreases a persistent inward Ca²⁺ current. This decreased inward current generates a membrane hyperpolarization that can explain the observed reduction in firing rate. Further studying the effect of Ca²⁺ on the firing pattern of MLIs showed that the removal of Ca²⁺ from the physiological solution caused a profound decrease in firing rate (to 3% of the initial rate, n = 2), confirming that Ca²⁺ currents participate in shaping the firing pattern of MLIs.

Presynaptic effect of GABA_B

The profound decrease in synaptic activity caused by the addition of baclofen suggests a presynaptic effect also, as the reduction in synaptic activity was more than would be expected from the accompanying decrease in spontaneous firing rate. To confirm such a presynaptic effect, we studied the effect of baclofen in the presence of TTX, which blocks the generation of action potentials. As shown in Fig. 3A, the spontaneous synaptic activity in the presence of TTX was reduced to 0.3 ± 0.09 Hz (n = 4). Baclofen (10 µM) further decreased the activity to 0.08 ± 0.05 Hz, a reduction of 75 ± 13%. The average amplitude of the spontaneous IPSPs in the presence of TTX was unaffected by baclofen (5.6 mV in both conditions).

View larger version (21K): [in this window] [in a new window]   Fig. 3. The presynaptic effect of baclofen. A: histograms of the intervals between spontaneous synaptic potentials in the presence of TTX (0.1 µM) under control conditions (left) and in the presence of baclofen (10 µM; right). The same cell was recorded in the 2 conditions and the analysis was performed on 600 s of continuous recordings. Note the different scales of the Y-axis. B, left: simultaneous recording from 2 synaptically connected MLIs. Extracellular recording of the presynaptic cell (top) and intracellular recording of the postsynaptic cell (bottom) with a high intracellular Cl solution. Traces were aligned by the peak of the presynaptic spike. Right: histogram of the amplitudes of the postsynaptic potentials. The column at zero denotes failures in synaptic transmission. C: same as B but in the presence of baclofen (10 µM).

These results suggest that baclofen directly interferes with the probability of synaptic release but does not effect quantal size. To directly investigate this possibility, we recorded simultaneously from two synaptically connected MLIs to examine the synaptic transmission between them. The presynaptic cell was extracellularly monitored (Fig. 3, B and C, top traces), while postsynaptic activity was recorded intracellularly (Fig. 3, B and C, bottom traces). This configuration was chosen because prolonged whole cell recordings from presynaptic neurons lead to a reduction in synaptic transmission due to a wash out effect. The traces were aligned according to the peak of the spontaneously occurring presynaptic action potentials. The postsynaptic responses were measured and plotted as amplitude histograms (right panels). Under control conditions the average amplitude was 2.38 mV, and 23% of the action potentials failed to evoke synaptic release (Fig. 3B). In the presence of baclofen (10 µM), the average amplitude was reduced to 1.73 mV and the failure rate increased to 67% (Fig. 3C). Similar results were obtained in three additional experiments. Baclofen (10 µM) increased the overall failure rate by a factor of 3.3 ± 2.1 (n = 4).

On the basis of these observations, we concluded that baclofen directly decreases the inhibitory synaptic transmission between two MLIs.

Activation of GABA_B receptors by endogenous GABA

We have shown so far that functional GABA_B receptors are present on cell bodies or dendrites and on axonal terminals of MLIs, but the question remains whether they are activated during normal physiological processes. This can be tested by examining whether GABA released from presynaptic terminals activates these receptors and causes a decrease in synaptic transmission or firing rate. To do this, we first measured the effect of CGP 55845A on spontaneous activity; second, we measured the probability of synaptic release in a pair pulse paradigm; and, finally, we activated a large population of stellate cells and measured the postsynaptic responses.

Adding the GABA_B blocker CGP 55845A increased the spontaneous rate of both action potentials and IPSPs [Fig. 4; see also Fig. 1 and Table 1, average increase 26 ± 18% (n = 5) and 39 ± 38% (n = 7), respectively]. These increases are due to tonic activation of the receptors by endogenous GABA. The large variation probably reflects the different amounts of endogenous GABA in different cerebellar slices. These results indicate a modulatory effect of endogenous GABA via GABA_B receptors, which determines the overall activity in the molecular layer.

View larger version (20K): [in this window] [in a new window]   Fig. 4. Endogenous GABA tonically activates the GABAB receptors. A: ISI histograms under control conditions (top) and in the presence of 1 µM CGP 55845A (bottom). The same cell was recorded in the 2 conditions, and the analysis was performed on 132 s of continuous recordings. B: histograms of the intervals between spontaneous synaptic potentials under control conditions (top) and in the presence of 1 µM CGP 55845A (bottom). The same cell was recorded in the 2 conditions, and the analysis was performed on 60 s of continuous recordings.

To study whether GABA_B receptors also play a temporal role in cerebellar information processing, we used the paired-pulse paradigm in two experimental systems. Based on previous observations, we have chosen an interval of 30 ms (Cohen and Yarom 2000). GABA_B is known to have a more prolonged and delayed effect; however, substantial effect was also reported at intervals of 50 ms (Gil et al. 1997). In the first set of experiments, we recorded simultaneously from two synaptically connected stellate cells while evoking two action potentials in the presynaptic cell (Fig. 5A, top trace). Each of the action potentials generated an IPSP in the postsynaptic cell (Fig. 5A, bottom trace). While the failure rate of the first IPSP was 17%, that of the second IPSP increased to 45%, almost three times higher. Similar results were obtained in two further pairs of neurons with an average ratio of 3.5, showing that there is a decrease in the probability of release in a paired-pulse paradigm. However, in any single trial, the probability of the second spike to evoke release was independent of the release evoked by the first spike (Fig. 5C). Thus if GABA does play a role in reducing the second response, we must infer multi-site connections between MLIs, i.e., MLI axons establish several synaptic contacts with their postsynaptic cell, each with an independent probability of release (Kondo and Marty 1998). Such an arrangement would mask the correlation between the release in the first and second pulse in each trial.

View larger version (13K): [in this window] [in a new window]   Fig. 5. Paired-pulse stimulation. A: simultaneous intracellular recording from 2 synaptically connected MLIs. Two action potentials, 30 ms apart, were evoked in the presynaptic cell (top trace), and inhibitory postsynaptic potentials (IPSPs) were recorded in the postsynaptic cell (bottom trace). Intracellular solution containing 0.1 mM QX-314 was used for the postsynaptic recording. B: amplitude histograms of the 1st (top) and 2nd (bottom) evoked IPSPs. The column at zero denotes failures in synaptic transmission. C: amplitude of the 2nd IPSP as a function of the 1st IPSP amplitude in the same trial. Failures in both occurred in 7 of 98 repetitions.

The participation of GABA_B receptors in the process was examined by adding CGP 55845A to the bath. The failure rate of the second response now equaled that of the first (50 and 54%, respectively), implicating GABA in the paired-pulse depression mechanism. The overall increase in failure rate reflects the continuous decline in synaptic transmission, which is probably the result of a rapid wash out in these small cells.

To further examine the involvement of GABA_B receptors in paired-pulse depression, we activated a large population of MLIs in an isolated cerebellum preparation. Using optical imaging, we tested the responses to activation of a beam of parallel fibers. The characteristic response to surface stimulation was an excitatory beam flanked by lateral inhibition (Cohen and Yarom 2000). Figure 6 shows average responses in five adjacent diodes located laterally to the activated beam (see inset) under three conditions. The response to a single stimulus was characterized by a fast depolarizing wave followed by a prolonged hyperpolarization (Fig. 6A). In paired-pulse stimulation, the second hyperpolarizing phase was completely blocked, while the second excitatory response was enhanced (Fig. 6B). When CGP 55845A (2 µM) was applied, both stimuli elicited depolarizing waves of similar amplitudes, and the second hyperpolarization was recovered. The optical system most likely measures the activity generated in Purkinje cells (Cohen and Yarom 1999), while the data described so far were measured from MLIs. Since the MLIs axons participate in both phenomena, we can conclude that paired-pulse depression is at least partially mediated by GABA_B receptors activated by endogenous GABA release.

View larger version (21K): [in this window] [in a new window]   Fig. 6. Optical imaging demonstrating the involvement of GABAB receptors in depression of inhibitory response to a 2nd stimulus. An isolated cerebellar preparation was used. A beam of activity was evoked by surface stimulation. The response at 5 diodes, located laterally to the active beam (as marked in the inset) was averaged. A and B: the response to a single and pair stimulus, respectively. Note that the depolarizing response to the 2nd stimulus is larger than that of the 1st stimulus, whereas the hyperpolarizing response was completely eliminated. C: the same as in B after 10 min of washing with 2 µM CGP 55845A. Note the increase in the 2nd inhibitory response and the similarity between the 2 positive responses.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our results clearly demonstrate the existence of functional GABA_B receptors on the inhibitory interneurons of the cerebellar molecular layer. Application of the specific GABA_B agonist baclofen caused a substantial decrease in MLI spontaneous firing rate and synaptic activity. In the sagittal slice preparation the parallel fibers are truncated, thus excitatory synaptic input to MLIs is scarce. Therefore our results reflect the effect of baclofen directly on MLIs. The data presented here suggest that both pre- and postsynaptic mechanisms are involved. Simultaneous recordings from two synaptically connected neurons showed that baclofen directly decreased the probability of release both in evoked and TTX-resistant release. The presynaptic effects of baclofen are usually attributed to a blockade of Ca²⁺ channels by GABA_B receptor activation (Misgeld et al. 1995), and we assume that this applies to our findings with MLI also.

The effect of baclofen on firing rate was maintained when synaptic activity was blocked by specific GABA_A blockers, thus supporting the existence of a postsynaptic mechanism as well. Application of baclofen caused a net outward current through the postsynaptic membrane, and the resultant hyperpolarization was large enough to account for the decrease in firing rate. As no somatic or remote dendritic changes in conductance could be detected, the baclofen effect appears to be mediated via a rather small membrane conductance change. This observation favors the possibility that baclofen decreases Ca²⁺ conductance rather than increase K⁺ conductance. This is supported by the facts that the hyperpolarizing effect of baclofen is maintained in the presence of Ba²⁺ and eliminated when Ca²⁺ currents are blocked and by the decrease in firing rate induced by CGP 44533.

Our results thus suggest that both the pre- and postsynaptic effects of baclofen are mediated by the same GABA_B receptors that decrease Ca²⁺ currents. This conclusion is supported by the recent findings that MLIs express the GABA_BR1a subunits (Bischoff et al. 1999). This type of subunit has been postulated to be exclusively associated with modulating Ca²⁺ currents (Bischoff et al. 1999). The pre- and postsynaptic effects of baclofen in hippocampus and granule cells and cells of the supraoptic nucleus are similarly due to a decrease in Ca²⁺ currents (Harayama et al. 1998; Huston et al. 1990; Wojcik et al. 1990).

Previous experiments with bicuculline suggest that the postsynaptic GABA_B receptors are extrasynaptic. Bicuculline, a specific antagonist of GABA_A receptors completely blocked the inhibitory synaptic response in MLIs (Häusser and Clark 1997; Pouzat and Marty 1998). However, if the GABA_B receptors were located at the synapse, these synapses should still be active. We therefore assume that the postsynaptic effect of baclofen is mediated via extrasynaptic receptors. A substantial number of extrasynaptic receptors have been reported in different cell types, and they most likely serve a regulatory role (Brickley et al. 1996; Fritschy et al. 1999; Somogyi et al. 1989). Extrasynaptic receptors can sense the concentration of transmitter in the extracellular space and can modulate the neuronal activity accordingly. There is a clear advantage to modulation via suppression of Ca²⁺ currents by activating GABA_B receptors rather than evoking Cl currents, as is the case with GABA_A receptors. The minimal conductance change associated with Ca²⁺ modulation causes no interference with the integration of synaptic inputs, whereas the large conductance changes associated with chloride modulation are bound to affect the computational processes taking place in the MLIs.

The increased activity in MLIs when GABA_B receptors were blocked (Fig. 4) demonstrates a modulatory effect of endogenous GABA, which determines the overall activity in the molecular layer. The temporal role that GABA_B receptors may play in the information processing was examined using the paired-pulse paradigm. This showed that the synapse between two MLIs is depressed at the second pulse (Fig. 5). We propose that the GABA released by the first stimulus can exert a temporally specific effect via GABA_B receptors that decreases the release evoked by the second stimulus. While the absence of one-to-one correlation between the probability of release of the first and second spike argues against this, this lack of correlation can also be explained by multiple releasing sites. The effect of CGP 55845A, although inconclusive since it is superimposed on a continuous decline of synaptic transmission, favors the possibility that GABA_B receptors participate in the process. If paired-pulse depression is mediated via GABA_B receptors, it is likely that they serve as auto-receptors, e.g., extrasynaptic receptors, which are activated by GABA released from the same terminal.

A temporally more precise effect that was clearly mediated by GABA_B receptors was demonstrated in our optical recordings. In this case the activation of many neuronal elements released enough GABA to fully activate the receptors possibly strengthening the effect. The depression seen in the optical recordings can be explained by GABA_B receptors serving either as autoreceptors or as axo-axonic synapses. Autoreceptors imply that the strength of a synaptic connection is modulated by its own activity. In contrast axo-axonic synapses suggest a network modulation of the synaptic strength. The possibility of axo-axonic synapses is supported by our previous study demonstrating a depression evoked by the activation of two spatially separated beams (Cohen and Yarom 2000). Even in this case, however, the massive release of GABA might cause spillover of transmitter and result in a nonspecific activation of GABA_B receptors.

In conclusion, GABA_B receptors in MLIs undoubtedly serve a modulatory role in several different ways. Presynaptic autoreceptors, in close proximity to GABA release sites, can be sensitive to small amounts of GABA and can exert a local effect. Extrasynaptic receptors located on the cell body and dendrites probably require larger amounts of GABA in the extrasynaptic space, and their activation will be manifested over a larger area. Both receptor types can affect the temporal interaction between consecutive inputs to the same area. Axo-axonic receptors, if present, can effect the temporal interaction between adjacent areas.