INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the mammalian CNS, glial cell processes have long been known to surround nerve cells and often tightly encapsulate their synapses. This is particularly prominent in the cerebellum where most excitatory synapses onto Purkinje cells are ensheathed by astrocytic processes, with an average degree of ensheathment that ranges between 65 and 87% (Johnstone et al. 1986; Spacek 1985; Xu-Friedman et al. 2001). Astrocytes contain a variety of transport systems that serve important homeostatic roles (Kimelberg et al. 1993; Schousboe et al. 1988; Walz 1989). These include the Na⁺-dependent glutamate transporters (Danbolt et al. 1998), which remove glutamate from the extracellular space. Five such transporters (excitatory amino acid transporters 1-5; EAAT1-5) have been cloned and characterized (Arriza et al. 1997; Fairman et al. 1995; Kanai and Hediger 1992; Pines et al. 1992; Storck et al. 1992). Of these, the glutamate aspartate transporter (GLAST=EAAT1) and the glial-transporter (GLT=EAAT2) are predominantly found in astrocytes (Lehre et al. 1995; Torp et al. 1994), and these are believed to sequester the majority of the neuronally released glutamate (Lehre et al. 1995; Torp et al. 1994). After its uptake by astrocytes, glutamate is converted to glutamine by glutamine-synthase (Laake et al. 1995; Zielke et al. 1989), and glutamine is shuttled back to neurons for the re-synthesis of glutamate or GABA in neurons. Through this glutamate-glutamine shuttle (Magistretti et al. 1999), astrocytes are thought to play an important role in the disposition and recycling of neuronally released glutamate (Hertz 1979; Hertz and Schousboe 1997; Hertz et al. 1988; Hogstad et al. 1988; Schousboe et al. 1988, 1993).

It has long been assumed that glial glutamate transport serves primarily in a neuroprotective capacity to limit spillage of glutamate (Glu) from synapses into the extracellular space (Kempski et al. 1991; Kimelberg et al. 1990; Rosenberg et al. 1992; Sonnewald et al. 1997; Sugiyama et al. 1989; Takahashi et al. 1997). Indeed, brief and small increases in the extracellular concentration of Glu ([Glu]_o) are sufficient to induce widespread excitotoxicity (Choi 1987, 1988, 1995), and this is believed to be a common final pathway for neuronal cell death in acute nervous system insults (stroke, trauma) and chronic nervous system diseases (Choi 1988; Lipton and Rosenberg 1994; Rothstein et al. 1995). However, recent studies suggest that astrocytes may participate more directly in neurotransmission. For example, it has been demonstrated in vitro (Linden 1998; Mennerick and Zorumski 1994; Mennerick et al. 1996) and in situ (Clark and Barbour 1997; Diamond and Jahr 1997; Diamond et al. 1998; Linden 1997) that Glu spilling from the synaptic cleft can induce astrocytic transport currents. Moreover, pharmacological inhibition of astrocytic Glu transport alters synaptic transmission (Asztely et al. 1997; Barbour et al. 1994; Scanziani et al. 1997; Tong and Jahr 1994). In these studies, it was difficult to discriminate between glial and neuronal Glu transport as pharmacological inhibitors affected both glial and neuronal transporters. Hence, the specific contribution of glial cells at glutamatergic synapses in shaping the postsynaptic response is not known.

We set out to delineate the specific contribution of glial transport by recording from cell pairs consisting of a neuron and a glial cell to ask the question whether changes in Glu uptake into a single glial cell contacting glutamatergic synapses can alter the excitatory postsynaptic neuronal response. We studied the parallel fiber to Purkinje cell excitatory synapse in cerebellar slices, which allows obtaining simultaneous patch-clamp recordings from Bergmann glial cells (astrocyte-like cells in the cerebellum) and Purkinje cells. We show that spontaneous EPSCs (sEPSCs) can be reversibly potentiated by gradual depolarization of a single Bergmann glial cell.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Tissue slices

Sprague Dawley rats were anesthetized using pentobarbital (30 mg/kg) and decapitated. A rapid craniotomy was performed to remove the occipital bone and mastoid processes allowing the cerebellum to be detached and removed and placed in ice-cold (4°C) artificial cerebrospinal fluid (ACSF, see composition in the following text), oxygenated with 95% O₂-5% CO₂ at pH 7.4. The tissue was then glued with cyanoacrylate to the stage of a vibratome, and 150- to 200-µm parasagittal and coronal slices of cerebellum were cut in cold oxygenated ACSF. After a recovery period of >1 h in ACSF, slices were placed in a flow-through chamber continuously super-perfused with oxygenated ACSF at room temperature and held in position by a nylon mesh glued to an U-shaped platinum wire. ACSF contained (in mM) 125 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃, and 10 glucose.

Tissue-slice physiology

Whole cell patch-clamp recordings followed standard methods (Hamill et al. 1981) and were essentially identical to methods that we used in previous studies (Bordey and Sontheimer 1997; Ransom and Sontheimer 1995; Roy and Sontheimer 1995; Roy et al. 1996). Patch pipettes of 2-3 M resistance for Purkinje cells and 4-6 M for Bergmann glial cells were made from thin-walled borosilicate glass (WPI, TW150F-40). Patch pipettes for Purkinje cell recordings were filled with a solution containing (in mM) 140 KCl or 140 Kgluconate, 2 MgCl₂, 1 CaCl₂, 10 EGTA, 4 Na₂ATP, and 10 HEPES and Alexa 568 0.1%. For Bergmann cell recordings, patch pipettes contained (in mM) 140 KCl or KNO₃ (to potentiate Glu transporter currents), 2 MgCl₂, 1 CaCl₂, 10 EGTA, 4 Na₂ATP, and 10 HEPES plus Lucifer yellow 0.1%. In both cases, pH was adjusted to 7.25 using Tris. Patch-slice recordings were made on a Nikon Optiphot 2, which had been modified to allow placement of two patch-clamp electrodes at a distance from each other through use of a XY translation stage under the microscope and fixed stage chamber. This arrangement allowed placement of electrodes at 10 µm distance from each other. The microscope was equipped with water-immersion Nomarski phase-contrast and fluorescence optics (×40, 1.8-mm working distance). Whole cell currents and potentials were measured at room temperature.

Paired recordings

Whole cell recordings were first obtained from a Purkinje cell using an Axopatch-200A (Axon Instrument). After 4-5 min of recording to obtain baseline synaptic activity at a holding potential of -70 mV, a whole cell recording of an adjacent Bergmann glial cell (not more than 30 µm away) was established using an Axopatch-1D amplifier (Axon Instruments). Current signals were low-pass-filtered at 1-5 kHz and digitized at 25-100 kHz using a Digidata 1200 digitizing board controlled by PClamp 7.0 or a Labmaster TL-125 digitizing board (Axon Instruments) controlled by PClamp 6.0 (Axon Instrument).

Only stable recordings from cells with membrane potentials more negative than 55 mV for Purkinje cells and 65 mV for Bergmann glial cells (with either KCl or KNO₃-based intracellular solution) were included. A stable recording was defined as a recording in which neither the series (access) resistance (R_s) nor the holding current varied by >10% from their initial values throughout the experiment. R_s was compensated to 60%.

Electrical synaptic stimulation

Stimulation of synaptic currents (50-500 µA for 300 µs) was achieved with a bipolar saline-filled stimulation electrode made from theta borosilicate glass (~4- to 10-µm tip diameter). For eliciting EPSCs in Purkinje cells or synaptically induced inward currents in Bergmann glial cells, stimulation electrodes were placed either in the granule cell layer at the edge of the molecular layer or in the parallel fiber pathway.

Contribution of inhibitory currents to Purkinje cell sEPSCs

In addition to excitatory inputs, Purkinje cells receive inhibitory inputs from interneurons. To limit their contribution to postsynaptic currents and to clearly distinguish them from EPSCs, the chloride concentration in the patch pipette was 6 mM, yielding a reversal potential for GABA-mediated inhibitory postsynaptic currents (IPSCs) near 80 mV as opposed to ~0 mV for Glu-mediated EPSCs. In some experiments, IPSCs were inhibited using picrotoxin (100-200 µM) added in the bath solution.

Drug applications

Drugs were applied either by addition to the bath perfusate or applied locally via pressure. The latter used a computer-controlled pressure ejection system (Picospritzer II). Pressure ejection pipettes were standard unpolished patch-electrodes with resistances of 5-7 M, positioned just above the slice at a distance of 40-50 µm from the recorded cell. The pressure applied ranged between 4-8 and 2-4 psi when directly applied onto the cell soma. Pressure ejection of control ACSF (without any drug) was used as control.

Immunohistochemistry

Tissue sections were stained using standard immunohistochemical techniques as we have previously described (Bordey and Sontheimer 1997; Sontheimer and Waxman 1993; Sontheimer et al. 1991). After fixation, slices were washed three times in PBS and permeabilized with 1% Triton X-100 for 10 min and incubated for 24 h with primary antibodies to glial fibrillary acidic protein (GFAP, 1:100, rabbit anti-mouse monoclonal antibody, INCStar, Stillwater, MN) in the presence of 1% normal goat serum (Vector) and 0.2% Triton X-100. Slices were then washed with PBS and incubated with secondary antibodies (goat anti-rabbit IgG conjugated to rhodamine 1:100) for 2 h at room temperature, and, following a final wash, slices were mounted on glass slides with a fluorescent microscopy mounting solution and sealed with nail polish.

Data evaluation and statistics

Spontaneous EPSCs were analyzed off-line using an event-detection routine (Minianalysis 5, Synaptosoft). Detection threshold was adjusted to approximately three times RMS noise, i.e., between 6 and 10 pA. To examine drug effects on sEPSCs, membrane currents were recorded over a control period of 3-4 min (except otherwise mentioned), and then the bath solution was exchanged to the test solution. After a period of 3-4 min to allow complete bath exchange, synaptic currents were recorded and later analyzed. Analyzed recordings were between 2 and 4 min under each experimental condition. Only one cell per slice was tested per bath-applied drug. Evoked currents and drug-induced currents were averaged and measured with Fetchan and Clampfit (Axon). Then these measures were exported to spreadsheets (Excel) for computation of statistical values (mean ± SD and mean ± SE) and for plotting. Data are given as means ± SD Where applicable, significance testing was done using either t-test or Mann-Whitney (Wilcoxon) test depending on distribution of data. Student's t-test for independent, unpaired variables was used whenever the data follow a Gaussian distribution. A two-sided Mann-Whitney (Wilcoxon) test was used for data sets with non-Gaussian distribution.

Chemicals were purchased from Sigma, unless otherwise noted.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Whole cell recordings were obtained from 34 Bergmann glial cells and 55 Purkinje cells in cerebellar slices from 9- to 22-day-old rats. Bergmann glial cells were initially selected by morphological criteria but were subsequently identified antigenically as Bergmann glial cells. Specifically, we pursued cells that had a small fusiform soma of ~8-12 µm diam. These cells were much smaller than Purkinje cells, which were ~40 µm in diameter but were larger than granule cells (~6-8 µm diam). As is typical, Bergman glial cell bodies were in the vicinity of Purkinje cells. Whole cell recordings from Bergmann glial cells revealed a characteristic low input resistance [48.2 ± 28.5 (SD) M, n = 34], hyperpolarized resting membrane potentials (-82.4 ± 7.0 mV, n = 34), and lack of current-induced action potential under current clamp (data not shown). Membrane capacitance averaged 53.5 ± 16.7 pF (n = 34). Purkinje cells had a more depolarized mean resting membrane potential of -64.4 ± 3.1 mV (n = 55) and a larger mean membrane capacitance of 77.6 ± 23.0 pF (n = 55) than those of Bergmann glial cells. Figure 1A shows a representative example of a Lucifer yellow-filled Bergmann glial cell. Lucifer yellow filled the cell body and processes; this allowed for subsequent immunohistochemical identification of the recorded cell using antibodies to the astrocyte specific protein GFAP (Eng 1985) (Fig. 1B). A morphological feature that is very typical and unique to Bergmann glial cells in the cerebellum is the termination of their processes as endfeet on the pial surface (Fig. 1A, ).

View larger version (63K): [in this window] [in a new window]   Fig. 1. Glutamate transporters in Bergmann glial cells. A: photograph of a Lucifer-yellow-filled Bergmann glial cell in a 250-µm cerebellar slice. B: anti-glial fibrillary acidic protein (GFAP) antibody labeling of the cell shown in A. C: D,L-threo-beta-hydroxyaspartate (THA)-induced currents recorded at different holding potentials from 80 to +60 mV. THA was pressure applied at 1 mM for 100 ms. D: mean I-V curves of 1 mM THA and 500 µM D-asparate-induced currents ( and , respectively) in Bergmann glial cells. The recordings were obtained with a KNO3-based intracellular solution.

Bergmann glial cells in situ show Glu transporter currents

Pressure application of 1 mM D,L-threo-beta-hydroxyaspartate (THA) for 100 ms induced a transient, inward transport current in all the recorded Bergmann glial cells (Fig. 1C). THA is a substrate agonist of Glu transporters and is therefore routinely used to induce transporter currents (Arriza et al. 1994). In an effort to increase the size of the transport current facilitating its detection, a pipette solution containing KNO₃ was used to increase the anion conductance mediated by Glu transporters (Wadiche et al. 1995). Under these conditions, 1 mM THA-induced Glu transporter currents averaged 206.7 ± 112.3 pA (n = 6) at 70 mV. We also examined the effects of D-aspartate, another substrate agonist of Glu transporters that is a weak agonist of AMPA/kainate receptors. L-Glu induced large inward currents largely mediated by AMPA/kainate receptors. These were not observed with D-aspartate, which is equally potent in activating Glu transport currents, making it the preferred agonist for these studies. In the presence of 2.5 mM kynurenic acid to block Glu receptor activation, D-asparate (500 µM, 100 ms) induced a transient Glu transporter current that averaged -173.5 ± 77.2 pA (n = 4, data not shown).

When we gradually changed the holding potential of the Bergmann glial cell from 80 to +70 mV, THA (Fig. 1C) and D-aspartate-induced Glu transporter currents decreased progressively in amplitude. This is readily visible in the current-voltage (I-V) plot in Fig. 1D, which shows the voltage dependence of the THA- and D-aspartate-induced currents ( and , n = 6 and 3, respectively). Transport currents were completely absent at +70 mV. This is a characteristic voltage dependence for glial Na⁺-Glu transporters (Brew and Attwell 1987).

Glu transporter currents in Bergmann glial cells in situ can be induced by electrical stimulation of parallel fibers/granule cells

Transient, inward currents similar to those in Fig. 1 could also be induced by presynaptic stimulation in the presence of 100 µM picrotoxin to block GABA_A receptor activation and 10 µM CNQX and 20 µM AP5 to block ionotropic Glu receptor activation (Fig. 2). For these experiments, a bipolar stimulating electrode was positioned in the molecular layer or in the granule cell layer, ~50 µm away from the voltage-clamped Bergmann glial cell. The stimulating electrode was moved along the molecular layer until the largest response was observed. We employed the minimal stimulation required to get a maximal response (typically 200 µA/300 µs). The resulting inward currents resembled those recorded in response to THA, although their mean amplitude was significantly smaller (-39.2 ± 26.5 pA, n = 4, when recorded from a holding potential of -70 mV with a KNO₃-based intracellular solution). However, the currents showed the voltage dependence typical of Glu transporter currents (Fig. 2B). We obtained similar recordings from Purkinje cells in the vicinity of the Bergmann glial cells and have superimposed two representative recordings of currents induced by electrical or synaptic stimulation (average of 10 and 20 events in the neuron and the glial cell, respectively) in Fig. 2D. The electrically induced currents in the Bergmann glial cells display a fast rise time (4.3 ± 1.1 ms, n = 4) but a relatively slow decay compared with that of synaptic currents electrically induced in Purkinje cells [monoexponential decay with a mean time constant of 8.3 ms (Llano et al. 1991)]. These currents in the Bergmann glial cells could be fit by the sum of two exponentials with mean decay time constants of 17.5 ± 3.2 and 162.5 ± 79.1 ms (n = 4). In addition, for a similar stimulus intensity and length, the currents induced in the glial cells were much smaller in amplitude than the synaptic currents induced in the Purkinje cells (-60 pA in the glial cell vs. -740 pA in the Purkinje cell).

View larger version (31K): [in this window] [in a new window]   Fig. 2. Electrically induced currents in Bergmann glial cells. A: electrically induced inward currents recorded at different holding potentials in a Bergmann glial cell. The stimulation electrode was placed in the molecular layer in the path of parallel fibers. The stimulation intensity was set at 200 µA for 300 µs. The recordings were obtained with a KNO3-based intracellular solution. B: I-V curves of the electrically induced currents. C: photograph of the Bergmann glial cell whose recordings are shown in A. D: superimposed and scaled electrically induced inward current in a Bergmann glial cell and excitatory postsynaptic currents (EPSCs) in a Purkinje cell recorded at -70 mV. Inset: expanded time scale of the currents in D to illustrate the similar rise time of both the glial and neuronal evoked currents. The recordings were obtained in the presence of 100 µM picrotoxin to block GABAA receptor activation.

Relative contribution of Glu transporter currents and ionotropic Glu receptor currents to the synaptically induced currents in Bergmann glia

To study the relative contribution of Glu transporter currents versus ionotropic Glu receptor currents in Bergmann glial cells, we recorded currents in response to presynaptic stimulation in the presence and absence of the Glu transporter inhibitor THA (Nakamura et al. 1993) and the ionotropic Glu receptors inhibitors, CNQX and AP5. These experiments were performed in the presence of 100 µM picrotoxin to inhibit GABA_A receptors, and AP5 was included to eliminate any possible activation of N-methyl-D-aspartate (NMDA) receptor. As shown in Fig. 3A, 300 µM THA reversibly reduced the inward current by ~70%. The residual current was sensitive to 10 µM CNQX +20 µM AP5, which, in combination with THA, completely eliminated the electrically induced currents (Fig. 3B). Mean values of inhibition obtained by either transport inhibition with THA or ionotropic receptor inhibition with CNQX and AP5 are illustrated in Fig. 3C and demonstrate that Bergmann glial cells show a composite response to parallel fibers/granule cell stimulation consisting of ~70% THA-sensitive transporter current and ~30% Glu receptor current.

View larger version (28K): [in this window] [in a new window]   Fig. 3. Pharmacology of the electrically induced currents in Bergmann glial cells. A, left: bath application of 100 µM THA reversibly reduced the electrically induced current by ~70% in Bergmann glial cell. Right: wash-out of THA. B: bath application of 10 µM 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX) and 20 µM 2-amino-5-phosphonopentanoic acid (AP5) reversibly reduced the electrically induced current. The remaining current is fully blocked by bath addition of bath-applied THA. These effects were partially reversible on wash out of THA, CNQX, and AP5. The recordings were obtained at -70 mV with a Kgluconate-based intracellular solution. C: mean ± SD values of the inhibition observed with either 100 µM THA to block transport currents or 10 µM CNQX in combination with 20 µM AP5 to inhibit ionotropic Glu receptors (n = 3).

Depolarization of a single Bergmann glial cell enhances the frequency of sEPSCs in Purkinje cells

Because we show above that depolarization of Bergmann glial cells to positive potentials eliminates inward Glu transporter currents, we set out to use this approach as a biophysical means to inhibit Glu transport in a single glial cell while also recording sEPSCs from a Purkinje cell. We therefore obtained double electrode patch-clamp data in which we simultaneously voltage-clamped a Bergmann glial cell and an adjacent Purkinje cell. For later identification, these cell pairs were filled with LY and either cascade blue or an Alexa dye (Fig. 4C). sEPSCs were recorded at either 50 mV to distinguish EPSCs from IPSCs or -70 mV where only sEPSCs could be observed (Fig. 4A). Nineteen such cell pairs were obtained. In 5/19 pairs, a significant increase in sEPSC frequency could be observed without any significant change in the amplitude of sEPSCs. Two distinct protocols were applied to the Bergmann glial cells: short successive depolarizations (1, 5, or 10 trains of 200-250 ms to either +40 or +80 mV) of the Bergmann glial cells or 30-s depolarization to 0 mV to provide a more prolonged blockade of Glu transporters in the Bergmann glial cells. In 4/5 cell pairs, a 38% increase in the frequency was observed without any detectable change in the amplitude of sEPSCs (measured at -70 mV) during a prolonged glial depolarization (Fig. 4A). Figure 4B shows the corresponding frequency histogram illustrating the increase in the frequency of the sEPSCs displayed in Fig. 4A. The mean frequency increased from 3.8 ± 4.2 to 5.2 ± 4.7 Hz during glial depolarization and returned to 3.7 ± 4.2 Hz after the glial depolarization (n = 4 cells, 70-1100 events in control, 85-725 events during glial depolarization and 70-980 events after the glial depolarization). The mean amplitude was 22.6 ± 11.4 pA in control and 23.9 ± 9.2 pA during glial depolarization. Similarly the mean 10-90% rise time of sEPSCs did not change during glial depolarization and during the length of the recordings (2.5 ± 0.3 ms in control and during the glial depolarization). These data are illustrated in the plots of the mean frequency (Fig. 5A), amplitude (Fig. 5B), and 10-90% rise time (Fig. 5C) as a function of the recording time. In the remaining cell 1/4, a train of 10 × 200-ms voltage steps to +80 mV resulted in transient 360 and 83% increases in sEPSC frequency and amplitude, respectively, recorded at -50 mV (data not shown). The sEPSC frequency and amplitude increased from 1.9 to 9.0 Hz and 14.8 ± 4.6 to 27.1 ± 12.5 pA, respectively (80 events in control and 140 events just following the glial depolarization). In this cell, recordings were obtained in the absence of picrotoxin, thus allowing the detection of spontaneous IPSCs (sIPSCs). Due to the difference in reversal potential of the GABA-mediated IPSCs (-78 mV) and Glu-mediated EPSCs (~0 mV) under the imposed ionic gradients, sIPSCs were outward (upward), whereas sEPSCs were inward (downward). Although glial depolarization increased sEPSC frequency, it did not alter sIPSC frequency. This is important as it provides evidence that this effect was most likely not mediated by unspecific K⁺ release from the depolarized Bergmann glial cell. The latter would have depolarized both GABAergic and glutamatergic presynaptic terminals, and hence should have resulted in increased frequency of both sIPSCs and sEPSCs. For all the recordings, we did not find any correlation between the sEPSC amplitude and 10-90% rise time (Fig. 5D). However, analysis was performed on sEPSCs with 10-90% rise time from 0 to 5 ms and with more restricted 10-90% rise times (1.5-2.5, 2.5-3.5, and 4-5 ms). Similar results regarding the change in EPSC frequency was obtained with each group of sEPSCs. Finally, in 3/19 pairs, short successive glial depolarizations induced either large calcium spikes in the Purkinje cells or a small transient inward current. Note that the degree of success to record from glial-neuronal cell pairs that showed physiological "pairing" (5/19) is consistent with the ratio of Bergmann glial cells to Purkinje cells on proximal synapses that is between 8:1 and 4:1.

View larger version (59K): [in this window] [in a new window]   Fig. 4. Long depolarizations of a Bergmann glial cell during a paired neuron-glial recording. A: spontaneous EPSCs (sEPSCs) recorded at -70 mV in a Purkinje cell. A simultaneous whole cell patch-clamp recording (not shown) was performed in an adjacent Bergmann glial cell. The Bergmann glial cell was depolarized for 30 s from -80 to 0 mV as indicated by the bar above the sEPSC (above C). C: photograph of the recorded pair consisting in a Lucifer-yellow-filled Bergmann glial cell (yellow) and a Alexa dye-filled Purkinje cell (red). Inset: photograph of a Bergmann glial cell-Purkinje cell pair where a 30-s glial depolarization did not affect the sEPSC properties.

View larger version (36K): [in this window] [in a new window]   Fig. 5. Effect of a 30-s-long glial depolarization on sEPSC properties. A-C: plots of the mean sEPSC frequency (A), amplitude (B), and 10-90% rise time (C) as a function of the recording time (n = 4 cells). Glial depolarization increased the sEPSC frequency but did not affect their amplitude and rise time. D: plot of the amplitude as a function of the 10-90% rise time. No correlation was found.

Pharmacological inhibition of glial Na⁺-Glu transport potentiates spontaneous Glu postsynaptic currents:

To further study the contribution of Glu transport to the modulation of glutamatergic synaptic transmission, we recorded from Purkinje cells in cerebellar slices and monitored spontaneous Glu postsynaptic currents (sEPSCs) in the presence and absence of the Glu transporter inhibitor THA (Fig. 6). These sEPSCs were mediated by AMPA receptor activation because they were totally and reversibly blocked by 10 µM CNQX. Bath application of 100 µM THA (Fig. 6B) led to a reversible increase in the frequency and the mean amplitude of sEPSCs as shown by the frequency and cumulative amplitude plots for this representative example (Fig. 6, C and D, respectively). The mean sEPSC frequency and amplitude were potentiated by 410 ± 313 and 51.1 ± 31.2%, respectively (3.7 ± 0.99 Hz and 25.1 ± 9.9 pA in control for 600-1,500 events per cell, and 17.4 ± 10.0 Hz and 37.7 ± 14.4 pA in the presence of THA for 650-1,900 events per cells, n = 5 cells). Similarly, pressure application of 1 mM THA induced a 2.6 ± 0.17-fold increase in sEPSC frequency and a 45.2 ± 21.5% increase in sEPSC amplitude (n = 4 cells, 30-140 events in control and 200-650 events per cell with THA). THA was pressure applied for 10 s. These data suggest a modulatory contribution of THA-sensitive Glu uptake on spontaneous EPSCs. These effects on sEPSCs by either bath or pressure applied THA were accompanied by the induction of an inward transport current (Fig. 6E, mean of -188.3 ± 120.2 pA, n = 5 and -271.4 ± 245 pA, n = 5 with THA bath and pressure applied, respectively, at -70 mV and with a Kgluconate-based recording pipette solution) in Purkinje cells. Because THA has been reported to be an agonist of NMDA receptors (Tong and Jahr 1994), we tested whether a pressure application of NMDA would mimic the effects of THA. A pressure application of 1 mM NMDA for 1 s induced a 2.6 ± 1.1-fold increase in sEPSC frequency and a 134.6 ± 114.8% increase in sEPSC amplitude, respectively (n = 4 cells, 200-300 events in control and 270-650 events per cell with NMDA). Figure 6F illustrates a typical effect of NMDA on sEPSCs.

View larger version (38K): [in this window] [in a new window]   Fig. 6. THA and N-methyl-D-aspartate (NMDA) effect on s- and miniature EPSCs. A and B: bath application of 100 µM THA induced an increase in the frequency and amplitude of sEPSCs. C: plot of the frequency of sEPSCs shown in (A and B) as a function of the recording time. D: cumulative probability of the amplitude of sEPSCs shown in (A and B). E and F: pressure application of 1 mM THA (E) and 1 mM NMDA (F) induced an increase in the frequency and amplitude of sEPSCs. G: miniature EPSCs (mEPSCs) recorded under control conditions (left) and in the presence of 300 µM bath applied THA (right). - - -, the baseline current. All the recordings were performed at -70 mV.

Because NMDA and THA appeared to have a similar effect, we wondered whether both drugs mediated their effects on sEPSCs by activation of NMDA receptors on either the presynaptic excitatory terminals and/or on the soma/dendrites of excitatory presynaptic neurons. This latter action would result in an increase firing rate of excitatory neurons and an increase in the frequency and amplitude of sEPSC. Therefore we examined the effects of bath application of THA in the presence of 1 µM tetrodotoxin (TTX) to block action potential-mediated synaptic release (Fig. 6G). With TTX, bath application of THA (300 µM) did not significantly increase the frequency (+1.7 ± 4.1%), amplitude (+4.7 ± 7.3%), and decay time-constant (2.5 ± 6.5%) of mEPSCs, suggesting that THA and NMDA modified sEPSC properties by activation of NMDA receptors on the soma/dendrites of presynaptic excitatory neurons. The mean frequency and amplitude of mEPSCs were 2.92 ± 0.77 Hz and 19.6 ± 8.3 pA in control and 2.99 ± 0.87 Hz and 21.0 ± 10.6 pA in the presence of THA (200-1000 events, n = 5 cells, in each condition). mEPSCs could be fit by a single exponential with a mean decay time constants of 5.0 ± 0.9 ms in control and 5.1 ± 0.7 ms with THA. The 10-90% rise time of mEPSCs did not change during the course of the experiment (mean of 2.4 ± 0.9 ms).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Using an acute cerebellar slice preparation, we show that Glu transporter currents in Bergmann glial cells can be induced by Parallel fiber stimulation and account for ~70% of the inward current. Importantly, we show that transient depolarization of Bergman glial cells is sufficient to inhibit Glu transporter currents and to enhance the frequency of sEPSCs in Purkinje cells. Previous studies have demonstrated the almost exclusive expression and activation of GLAST in Bergman glial cells (Bergles et al. 1997; Clark and Barbour 1997). Indeed, activation of a composite currents in astrocytes or Bergmann glial cells that is mediated by ionotropic Glu receptors and Glu transporters has been demonstrated in hippocampus (Bergles and Jahr 1998; Diamond and Jahr 1997; Diamond et al. 1998) and cerebellum (Dzubay and Jahr 1999). The concentration of Glu that spills from synapses and reaches astrocytic processes has been experimentally determined to be ~180 µM (Dzubay and Jahr 1999), and in the cerebellum, the time course of the glial Glu transporter current closely follows the time course of Glu released from presynaptic terminals (Bergles et al. 1997). Thus little doubt remains that Glu rapidly leaves the synaptic cleft and reaches perisynaptic glial processes.

The role that glial Glu transport plays with regards to modifying the excitatory signal has been less clear albeit several recent studies have examined this question (Bergles and Jahr 1997; Bezzi et al. 2001; Mennerick and Zorumski 1994; Oliet et al. 2001; Parpura et al. 1994; Robitaille 1998; Zorumski et al. 1996). Overwhelmingly these studies suggest that pharmacological inhibition of Glu transporters using dihydrokainate (DHK), THA, or L-trans-pyrrolidine-2,4 dicarboxylic acid (PDC) modulates evoked EPSCs. However, pharmacological inhibition does not permit to discriminate between transporters expressed in Bergmann glia versus those expressed in cerebellar neurons. Moreover, these drugs, even if applied locally, can affect multiple cells in the vicinity of the application. We overcame these limitations by disabling Glu uptake in only a single glial cell biophysically. Depolarization of Bergmann glial cells to potentials more positive than 0 mV inhibit Glu uptake as previously described for retinal glial cells (Sarantis and Attwell 1990). This was sufficient to transiently elevate sEPSC frequency. Our interpretation is that Glu uptake was temporarily compromised leading to a build-up of perisynaptic Glu. Alternatively, however, glial depolarization may have lead to K⁺ efflux and thus may have depolarized pre- or postsynaptic elements. We find this less likely for two reasons. First, we obtained some of these recordings under conditions that permitted both EPSCs and IPSCs, and we set the ionic gradients such that GABAergic currents were outward and glutamatergic currents inward. Under these conditions, we saw rapid changes in the frequency and amplitude of glutamatergic synaptic currents, but little effect on the frequency and amplitude of GABAergic synaptic currents following glial depolarization. If K⁺ release from glia had depolarized the presynaptic cell, thereby altering the rate of transmitter release, one would have expected the frequency of IPSCs to change as well, which was not the case. Second, if the postsynaptic membrane would experience K⁺ release from glial cells, we should have detected a baseline shift in the holding current under voltage clamp, which we did not. We interpret these findings as suggestive that inhibition of Glu uptake into Bergmann glial cells was principally responsible for the observed modulation of EPSCs.

Our experiments in which we compared the effects of THA in the presence and absence of TTX suggest that the THA induced release of Glu acts primarily presynaptically. The lack of effect of THA on mEPSCs suggests that THA and NMDA modified EPSC properties by activating NMDA receptors not directly on the presynaptic terminals but on the soma/dendrites of excitatory granule cells resulting in an increasing in their firing rate. Thus rather than activating postsynaptic Glu receptors, the glial-released Glu appears to modulates the presynaptic release of neurotransmitter and may in fact do so by acting on presynaptic Glu receptors on granule cells. This would be consistent with the recent findings that in the cerebellum extrasynaptic Glu is much more likely to modulate presynaptic Glu receptors than postsynaptic ones (Lehre and Rusakov 2002). Two recent studies suggest that in the cerebellum, the amount of Glu present in extrasynaptic sites near the presynaptic release sites for Glu affects the subsequent recruitment of either NMDA receptors (Lehre and Rusakov 2002) or mGluRs (Clark and Cull-Candy 2002) to the postsynaptic current. Hence, in the cerebellum, glial Glu uptake may have an important role in regulating excitatory transmission through modulation of presynaptic Glu receptors.

While our studies suggest that any impairment of glial Glu uptake will modulate glutamatergic neurotransmission, they do not allow us to judge the relative importance for normal glutamatergic neurotransmission. It is possible that glial Glu uptake operates at a fairly constant rate and provides a static sink for Glu to be drawn away from the synaptic cleft. However, it is equally possible, almost likely, that neuronal activity will also regulate the rate of glial Glu uptake. Notable, as was demonstrated before (Clark and Barbour 1997), we observed the activation of ionotropic Glu receptors on glial cells in response to parallel fiber/granule cell stimulation. This would depolarize these glial cells, thereby reducing the electrochemical gradient for the uptake of Glu by Na⁺-dependent uptake. Moreover, glial cells may depolarize, at least locally as K⁺ is released from neurons, which again would tend to reduce Glu uptake. A family of regulatory proteins, which regulate Glu transporters and which are released by neurons, has recently been characterized (Jackson et al. 2001). These again may alter the rate of glial Glu uptake in an activity dependent fashion. This bidirectional communication may be very important in the cerebellum. Iino et al. (2001) showed that activation of Ca²⁺-permeable Glu receptors in Bergmann glial cells was necessary to allow the generation and maintenance of the appropriate structural and functional association between glutamatergic synapses onto Purkinje cells and Bergmann glial cell processes. Thus changes in extracellular Glu surrounding Bergmann glial processes would determine the level of occupancy of these Ca²⁺-permeable Glu receptors, allowing a dynamic regulation of the synapse ensheathment by Bergmann glial processes. Clearly further studies are necessary to examine the contribution of glial cells to neurotransmission at glutamatergic synapses.