Skip to main content
NCBI home page
The Journal of Physiology logoLink to The Journal of Physiology
. 2014 Feb 13;592(Pt 7):1493–1503. doi: 10.1113/jphysiol.2013.267039

Ectopic release of glutamate contributes to spillover at parallel fibre synapses in the cerebellum

Saju Balakrishnan 1,3, Katharine L Dobson 2, Claire Jackson 1, Tomas C Bellamy 1,2,✉
PMCID: PMC3979607  PMID: 24421351

Abstract

In the rat cerebellar molecular layer, spillover of glutamate between parallel fibre synapses can lead to activation of perisynaptic receptors that mediate short-and long-term plasticity. This effect is greatest when clusters of fibres are stimulated at high frequencies, suggesting that glutamate clearance mechanisms must be overwhelmed before spillover can occur. However, parallel fibres can also release transmitter directly into the extracellular space, from ‘ectopic’ release sites. Ectopic transmission activates AMPA receptors on the Bergmann glial cell processes that envelop parallel fibre synapses, but the possible contribution of this extrasynaptic release to intersynaptic communication has not been explored. We exploited long-term depression of ectopic transmission, and selective pharmacology, to investigate the impact of these release sites on the time course of Purkinje neuron excitatory postsynaptic currents (EPSCs). Depletion of ectopic release pools by activity-dependent long-term depression decreased EPSC decay time, revealing a ‘late’ current that is present when fibres are stimulated at low frequencies. This effect was enhanced when glutamate transporters were inhibited, and reduced when extracellular diffusion was impeded. Blockade of N-type Ca2+ channels inhibited ectopic transmission to Bergmann glia and decreased EPSC decay time. Similarly, perfusion of the Ca2+ chelator EGTA-AM into the slice progressively eliminated ectopic transmission to glia and decreased EPSC decay time with closely similar time courses. Collectively, this evidence suggests that ectopically released glutamate contributes to spillover transmission, and that ectopic release therefore degrades the spatial precision of synapses that fire infrequently, and may make them more prone to exhibit plasticity.

Introduction

In the molecular layer of the cerebellum, spillover of glutamate from one synapse to another is thought to occur when spatial or temporal summation of release events leads to saturation of uptake mechanisms (Carter & Regehr, 2000; Marcaggi & Attwell, 2007; Szapiro & Barbour, 2009). It can be thought of as a paracrine signalling mechanism, and has been implicated in the activation of perisynaptic receptors on both presynaptic and postsynaptic cells during tetanic stimulation (Neale et al. 2001b; Casado et al. 2002; Marcaggi & Attwell, 2005). Commonly, activation of such receptors plays a role in plasticity by engaging second messenger pathways that modulate presynaptic release probability or postsynaptic receptor density.

Hundreds of thousands of parallel fibres form synapses with the dendritic arbour of each Purkinje neuron, forming an array of closely packed terminals, each of which is enveloped by processes from accessory Bergmann glial cells (Grosche et al. 2002). Both the glia and the Purkinje neurons express glutamate transporters (excitatory amino acid transporter 1 and 2 (EAAT1 and EAAT2) isoforms in glia, and EAAT4 in Purkinje cells) that rapidly remove glutamate from the extracellular space. Experimentally, spillover manifests as a prolongation of the Purkinje neuron excitatory postsynaptic current (EPSC) time course (Barbour et al. 1994), a process that is dramatically enhanced by pharmacological or genetic elimination of glutamate transporters from neurons and/or glia (Brasnjo & Otis, 2001; Marcaggi et al. 2003; Takayasu et al. 2005). To what extent, and under what circumstances, such spillover can occur under physiological conditions has been the subject of intense debate (Szapiro & Barbour, 2009).

Over the last decade, evidence has accumulated that parallel fibre terminals can also release glutamate from sites ‘ectopic’ to the active zone (Matsui & Jahr, 2003, 2004; Matsui et al. 2005). This route for release delivers glutamate directly into the extrasynaptic space, where it activates AMPA receptors and glutamate transporters on the enveloping Bergmann glia. As AMPA receptor activation is essential for maintaining glial–synapse interaction (Iino et al. 2001), it was proposed that ectopic release provided a route for neuron–glial transmission that promoted the juxtaposition of glial transporters with sites of glutamate release. Ectopic sites show greater sensitivity to the N-type Ca2+ channel blocker ω-conotoxin GVIA than synaptic sites, as well as greater sensitivity to Ca2+ chelators when loaded as esterified precursors (Matsui & Jahr, 2003, 2004). We have also reported that repetitive stimulation of parallel fibres at frequencies in the range 0.1–1 Hz causes long-term depression of neuron–glial transmission, without having any impact on the amplitude of neuronal EPSCs (Bellamy & Ogden, 2006). We recently reported that the mechanism for such depression is a failure to rapidly recycle presynaptic vesicles to ectopic release sites (Balakrishnan et al. 2011), allowing the depression of ectopic release under experimental conditions in which the strength of active zone transmission is unchanged, or even potentiated.

We used these activity-dependent and pharmacological means to suppress ectopic release, to investigate whether ectopic sites contribute to spillover of glutamate around the parallel fibre synapse.

Methods

Ethical approval

All experiments complied with the UK Animals (Scientific Procedures) Act 1986, and were approved by the Babraham Research Campus Animal Welfare, Experimentation, and Ethics Committee, and the University of Nottingham Animal Welfare and Ethical Review Body.

Electrophysiology

Sixteen-to 20-day-old Wistar rats of either sex were killed by cervical dislocation, and transverse cerebellar slices (300 μm) were prepared as previously described (Balakrishnan & Bellamy, 2009). For recording, slices were perfused in an immersion chamber with a bath solution containing (mm): NaCl (126), KCl (3), NaH2PO4 (1.2), NaHCO3 (25), glucose (15), MgSO4 (2) and CaCl2 (2), which was continuously bubbled with 95% O2/5% CO2. For Purkinje neuron experiments, the bath solution was supplemented with 20 μm picrotoxin to inhibit GABAA receptors. For recordings at higher temperatures, the perfusion solution was gassed at 40°C in a water bath, and perfused through an in-line heater (Warner Instruments, Hamden, CT, USA) to maintain a temperature of 34 ± 1°C in the recording chamber.

Recording electrodes were manufactured as previously described (Balakrishnan & Bellamy, 2009). Internal solution consisted of (mm): potassium gluconate (110), KCl (5), Hepes (50), EGTA (0.05), MgSO4 (4), ATP (4), GTP (0.2) and phosphocreatine (9), pH to 7.4 with 1 m KOH.

Whole cell voltage clamp recordings were made from Bergmann glia (holding potential = −80 mV) and Purkinje neuron (holding potential = −70 mV) somata in the Purkinje cell layer with an Axopatch 200B or Multiclamp 700B amplifier (Axon Instruments, Foster City, CA, USA). Currents were low pass filtered at 4–5 kHz, and sampled at 25 kHz, using Spike2 software (CED, Cambridge, UK). Series resistances ranged from ∼3 to 15 MΩ, were monitored throughout the experiment and were compensated by 85–90% in Purkinje neuron recordings, but uncompensated in glial recordings. If series resistance increased above 15 MΩ, recordings were discarded.

Parallel fibres were stimulated with a patch electrode (∼1–2 MΩ) filled with bath solution and positioned in the lower third of the molecular layer or in the granular layer (approximately 50–100 μm from the Purkinje cell layer), and connected to an isolated constant current stimulator (5–40 μA, 80 μs; Digitimer, Welwyn Garden City, UK).

Data analysis

Analysis of Purkinje neuron EPSCs and Bergmann glial extrasynaptic currents (ESCs) was carried out in Spike2 software. Amplitude was measured as the peak inward current after both pulses, relative to the current before stimulation. Decay time was measured as the time for an ESPC to decline from 90 to 10% of the peak current. ESC and EPSC traces shown are the average of five sequential recordings at the indicated frequency. Stimulus artefacts are truncated or blanked for clarity. Aggregate data are the mean ± SEM of multiple cells as indicated in the figure legends. Statistical significance was tested for by single sample Student's t test or paired t test using Origin 7.5 (OriginLab, Northampton, MA, USA) and taken to be significant if P < 0.05 (P values are given in the figure legends or main text).

Results

Depression of ectopic transmission inhibits a late current in Purkinje neurons

Parallel fibres were stimulated with a pair of pulses (10 ms interval) at a baseline frequency of 0.033 Hz. Under these control conditions, an ESC due to AMPA receptor and EAAT1 transporter activation is detected in Bergmann glia (Fig. 1A; Bellamy & Ogden, 2005), and an EPSC in Purkinje neurons (Fig. 1B). As we have previously reported, the glial ESC is persistently depressed by repetitive stimulation at frequencies >0.033 Hz, apparently due to a failure to recycle vesicles to ectopic release sites (Balakrishnan et al. 2011). In this way, ectopic transmission can be persistently suppressed, without lasting impact on the strength of transmission at the synaptic active zone.

Figure 1.

Figure 1

Open in a new tab

A, representative whole-cell recordings of glial ESCs generated by paired pulse stimulation (10 ms interval) of parallel fibres in transverse cerebellar slices at 0.033 Hz (first panel). Raising stimulation frequency to 0.2 Hz (second panel) and 1 Hz (third panel) for 10 min leads to a progressive depression of ESCs. After returning baseline frequency to 0.033 Hz for 10 min (fourth panel), the persistent depression is evident in comparison with initial recordings (grey trace). B, representative recordings of EPSCs from Purkinje neurons under the same stimulation conditions as in A. Note persistent decrease in decay time. C, aggregate data of EPSC amplitude (Amp) and decay time (Dec) from Purkinje neurons (n = 17) and ESC amplitude in Bergmann glia (BGC; n = 5). Data are mean ± SEM normalized to initial values at 0.033 Hz. *P < 0.0001 (single sample t test). D, plot of change in decay time after recovery against initial ESPC amplitude for all Purkinje neurons (n = 17) with linear regression (dashed line). E, subtraction of ESCs (left panel) and EPSCs (right panel) recorded after recovery from initial recordings at 0.033 Hz shows the current sensitive to depression. Grey traces are from all recorded cells, black traces are mean currents.

Increasing baseline frequency to 0.2 Hz for 10 min decreased glial ESC amplitude, but had no impact on neuronal EPSC amplitude, as expected (Fig. 1A and B). Raising baseline again to 1 Hz almost eliminated glial ESC, but only reduced EPSC amplitude by ∼25%. Returning baseline frequency to 0.033 Hz for 10 min resulted in only marginal recovery of ESC amplitude, but EPSC amplitude fully recovered to control levels and in some cells showed potentiation (7/17 cells with >10% increase in mean amplitude). In contrast to the minimal impact on amplitude, the decay time of Purkinje neuron EPSCs was significantly reduced during both 0.2 Hz and 1 Hz stimulation. Furthermore, this decrease in decay time persisted after returning to a 0.033 Hz baseline for 10 min, mirroring long-term depression of glial currents (Fig. 1A–C).

Subtraction of neuronal EPSC and glial ESC after 10 min of recovery at 0.033 Hz from the responses recorded during the initial period of stimulation reveals the current component sensitive to depression of ectopic release (Fig. 1E). For glia, this represents the majority of total current, but for Purkinje neurons, it is a delayed slow-rising current with amplitude 325 ± 66 pA (n = 17; mean ± SEM) and time to peak of 31.5 ± 3.8 ms after the second pulse in the pair. We term this a ‘late’ current to distinguish it from known fast and slow currents.

Depression of late current does not depend on current amplitude

In postnatal day (P)16–20  rats, Purkinje neuron dendrites significantly filter the EPSC when measured at the cell soma (Roth & Hausser, 2001). Dendrites also exhibit active conductances, which could in principle be engaged if voltage control was inadequate due to poor space clamp. Thus, if the cable properties of Purkinje neurons are altered by stimulation of parallel fibres at 0.2–1 Hz, the changes in EPSC decay time may be attributable to altered recording conditions, rather than plasticity of ectopic release.

To explore this, we investigated the effect of EPSC amplitude on decay time. Problems due to poor space clamp would be expected to worsen as EPSC amplitude increased. As such, errors in decay time due to clamping artefacts would be expected to increase with EPSC amplitude. In the 17 cells tested in Fig. 1, there was no obvious correlation between EPSC amplitude at 0.033 Hz and the subsequent decrease in decay time evoked by raising stimulation frequency to 0.2 Hz (Fig. 1D; correlation coefficient of linear fit = 0.178, SD = 0.138). To test the influence of ESPC amplitude on EPSC time course more directly, we incubated slices with a submaximal concentration of the high affinity AMPAR antagonist 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX) during continuous stimulation at 0.033 Hz. After 5–10 min incubation with 100 nm NBQX, EPSC amplitude was significantly decreased to 0.65 ± 0.08 of control (P = 0.016, single-sample t test; n = 5) whereas decay time was not significantly affected (mean 0.90 ± 0.05 of control; P = 0.130, single-sample t test; n = 5).

These results suggest that the failure of EPSC decay time to recover from 0.2–1 Hz stimulation, despite full recovery of EPSC amplitude (Fig. 1C), was unlikely to be explained by changes in the effectiveness of the voltage clamp.

Inhibition of glutamate transporters increases late current amplitude

To exaggerate the extent of spillover, we incubated slices with the glutamate transporter inhibitor dl-threo-β-benzyloxyaspartic acid (TBOA; 200 μm) during paired pulse stimulation at 0.033 Hz (Fig. 2A). As reported by Takayasu et al. (2004), TBOA marginally increased EPSC amplitude, but substantially increased the decay time. Increasing baseline frequency to 0.2 Hz significantly decreased both EPSC amplitude and decay time (Fig. 2A), indicating that approximately 50% of the increase in spillover when transporters are blocked is attributable to sources that are depressed by 0.2 Hz stimulation.

Figure 2.

Figure 2

Open in a new tab

A, first panel: representative EPSCs recorded from Purkinje neurons before (grey trace) and after (black trace) addition of 200 μm TBOA for 10 min, during stimulation at 0.033 Hz. Second panel: raising baseline frequency to 0.2 Hz for 10 min after TBOA treatment decreased both ESPC amplitude and decay time (black trace). Third panel: mean ± SEM aggregate data from n = 7 cells for amplitude (Amp) and decay time (Dec) at 0.033 Hz (grey columns) and 0.2 Hz (black columns) after addition of TBOA, normalized to values before addition of TBOA. *P = 0.009 and 0.001 (paired t test) for change after 0.2 Hz stimulation in mean amplitude and decay time, respectively. B, representative EPSCs recorded from Purkinje neurons before (grey trace) and after (black trace) bath incubation with 2.5% dextran for 5 min. Second panel shows traces normalized to amplitude. Third panel shows mean ± SEM amplitude (Amp) and decay time (Dec) for six cells, normalized to values before addition of dextran. *P = 0.043 and 0.023 (single sample t test) for change in mean amplitude and decay time after dextran, respectively. C, representative EPSCs recorded from Purkinje neurons in slices incubated at 34°C stimulated at 0.033 Hz (grey trace) and 0.2 Hz for 10 min (black trace). Second panel shows traces normalized to amplitude. Third panel shows mean ± SEM amplitude (Amp) and decay time (Dec) for eight cells at 0.2 Hz, normalized to values at 0.033 Hz. *P = 0.034 (single sample t test).

Impeding diffusion decreases late current amplitude

The prolongation of EPSC decay time can be explained by mechanisms other than spillover of transmitter through the extracellular space. Delayed release (Atluri & Regehr, 1998) and repetitive firing (Isope et al. 2004) are known to occur at parallel fibre synapses, which may be alternative explanations for the late current that arise from release at the active zone. Impeding diffusion in the extracellular space would select between these hypotheses: fast synaptic release of glutamate should show increased EPSC amplitude due to the prolonged local transient, whereas sources of glutamate that must diffuse through the extrasynaptic space to reach receptors should result in slower rise times and reduced amplitude (Nielsen et al. 2004; Satake et al. 2006).

Impairment of diffusion in this manner can be achieved with bath perfusion of dextran (Nielsen et al. 2004). We added 2.5% dextran to the extracellular medium, during stimulation of parallel fibres at 0.033 Hz (Fig. 2B). Dextran decreased the decay time of the EPSC, reaching a mean of 0.887 ± 0.043 of control after 5 min incubation (Fig. 2B, P = 0.023). The mean EPSC amplitude increased over the same time course, reaching 1.21 ± 0.092 of control (Fig. 2B, P = 0.043). These results are consistent with the amplitude of fast currents being potentiated by blocking extracellular diffusion, but the amplitude of late currents being reduced.

Spillover occurs at physiological temperatures

Glutamate transporters are highly temperature sensitive, and so we also tested the effect of 0.2 Hz stimulation on EPSC time course at a more physiological temperature (34°C). As with stimulation at room temperature, increasing baseline frequency to 0.2 Hz had no effect on EPSC amplitude, but significantly reduced decay rate (Fig. 2C), although to a lesser extent than at room temperature. This is consistent with the reduced depression of Bergmann glial ESC amplitude by 0.2 Hz stimulation at 34°C (Balakrishnan et al. 2011), suggesting that ectopic transmission is depressed to a lesser extent at the more physiological temperature.

Pharmacological inhibition of ectopic release reduces late current

The foregoing activity-dependent methods for suppressing ectopic transmission suggest that the late current in Purkinje neurons depends on this glutamate source. We next tested the effect on EPSC decay time of known pharmacological inhibitors that selectively target ectopic sites over synaptic sites.

The N-type Ca2+ channel blocker ω-conotoxin GVIA has been reported to inhibit ectopic release more effectively than active zone release (Matsui & Jahr, 2004). Incubation of slices with increasing concentrations of ω-conotoxin (0.1–3 μm) caused progressive inhibition of glial ESC amplitude with an IC50 of 0.25 μm (Fig. 3). In contrast to another report (Mintz et al. 1995), we found no significant effect of ω-conotoxin on EPSC amplitude at Purkinje neurons at any concentration up to 3 μm, a discrepancy that may be attributable to our use of paired pulse stimulation, or the difference in age (Mintz and colleagues used 9-to 14-day-old rats). However, ω-conotoxin did decrease EPSC decay time, with an IC50 of 0.28 μm (Fig. 3).

Figure 3.

Figure 3

Open in a new tab

A, representative traces of glial ESCs (BGC, first panel) and Purkinje neuron EPSCs (PN, second panel) before (grey trace) and 10 min after (black trace) addition of 3 μm of the N-type Ca2+ channel blocker ω-conotoxin GVIA to the bath. B, concentration–response relationship for glial ESC amplitude (open circles; IC50 = 0.25 μm) and Purkinje neurons EPSC decay time (filled circles; IC50 = 0.28 μm) against conotoxin concentration. Lines are fits to the Hill equation with slope 1.

Ectopic release sites have also been shown to have greater sensitivity to the membrane-permeable Ca2+ chelator EGTA-AM than active zone sites (Matsui & Jahr, 2003). The suppression of vesicular exocytosis by EGTA-AM should develop progressively in slice preparations, given the time taken for diffusion into the tissue parenchyma, de-esterification and accumulation of EGTA in the various cellular compartments. We therefore reasoned that if the late current depends on ectopic release, we would observe a similar time course for inhibition of the amplitude of glial ESCs and the decay time of Purkinje neuron EPSCs following bath application of the chelator to slices. To test this, we simultaneously recorded from anatomically adjacent pairs of Purkinje neurons and Bergmann glial cells (to ensure that diffusion time into the slice was as evenly matched as possible), whilst stimulating in the molecular layer at 0.033 Hz. As predicted, 3–4 min after application of 20 μm EGTA-AM, glial AMPAR current and Purkinje neuron EPSC decay rate began to decrease, and thereafter declined progressively over 30 min with a closely similar time course (Fig. 4). In contrast, EPSC amplitude was relatively stable throughout the experiment (Fig. 4).

Figure 4.

Figure 4

Open in a new tab

A, paired whole cell recordings from anatomically adjacent Bergmann glia (BGC, first panel) and Purkinje neurons (PN, second panel) before (grey traces) and 30 min after (black traces) incubation with 20 μm EGTA-AM. Stimulus artefacts are blanked for clarity. B, time course of EGTA-AM effect on glial ESC amplitude (filled squares) and Purkinje neuron EPSC amplitude (filled circles) and decay time (open circles) after addition to bath at t = 0. Data are mean ± SEM from four pairs of cells.

Late current does not require activation of glial cell AMPA receptors

Collectively, these results suggest that the late current observed in Purkinje neurons depends on ectopic release of glutamate. It is possible, however, that the late current may be activated indirectly – for example by the release of a ‘gliotransmitter’ from Bergmann glia in response to ectopic activation of glial AMPARs. To test this, we stimulated parallel fibres at 0.033 Hz, and applied the pore blocker of GluA2-deficient AMPARs, 1-naphthyl acetyl spermine (NASP; Koike et al. 1997), which would selectively inhibit glial AMPARs over Purkinje neuron AMPARs (and also block the metabotropic glutamate receptor (mGluR)-linked slow ESPC; Canepari et al. 2004). NASP at 50 μm effectively depressed glial AMPAR currents over 20–30 min, but had no significant effect on the amplitude or decay time of Purkinje neuron EPSCs (Fig. 5). This result suggests that activation of glial AMPARs is not required for generation of the late current in Purkinje neurons.

Figure 5.

Figure 5

Open in a new tab

A, representative recordings of glial ESCs (BGC, first panel) and Purkinje neuron EPSCs (PN, second panel) before (grey trace) and 30 min after (black trace) addition of the GluA2-deficient AMPAR pore blocker NASP (50 μm). B, time course of NASP's effect on glial ESCs (squares), Purkinje neuron EPSC amplitude (filled circles) and EPSC decay time (open circles). Data are mean ± SEM from six cells.

Late current depends on the distribution of parallel fibre inputs

Previous investigations into spillover in the cerebellar cortex have suggested that the pattern of parallel fibre stimulation is crucial to observing the phenomenon. Release from a cluster of fibres by stimulation in the molecular layer results in glutamate transporters being overwhelmed, and spillover becoming more pronounced (Marcaggi & Attwell, 2007). In contrast, stimulation in the granular layer, which is expected to result in glutamate release from a more distributed array of parallel fibres, leads to markedly less crosstalk. Ectopic release is observed during granular layer stimulation, as measured by neuron–glial transmission (Balakrishnan & Bellamy, 2009), and so we investigated the impact of the spatial distribution of presynaptic release on EPSC decay time.

Granule cell somata and parallel fibre bundles were stimulated with paired pulses (10 ms interval) at 0.033 Hz, during recording from individual Purkinje neurons (Fig. 6A and B). Stimulation intensity was adjusted to ensure that the EPSC amplitude was consistent for each input (from 4 to 80 μA; granular layer stimulation typically required higher stimulation intensities). The aim was to match input strength from each stimulation site, with the assumption that the number of terminals and extent of release would be similar in each case. The input that was stimulated first was alternated from cell to cell, but no difference in outcome was observed, and so results were pooled.

Figure 6.

Figure 6

Open in a new tab

A, diagram of electrode positioning for stimulation of closely associated parallel fibres in the molecular layer (ML, left panel), and more distributed parallel fibres from stimulating somata in the granular layer (GL, right panel), during recording from a single Purkinje neuron (grey). B, representative recordings of EPSCs from a single Purkinje neuron when stimulated in the molecular layer (ML, first panel) or the granular layer (GL, second panel). The third panel shows an overlay of the traces (molecular layer in grey). C, representative recordings of EPSCs from a Purkinje neuron stimulated in the granular layer at 0.033 Hz (first panel) and 0.2 Hz (second panel) for 10 min. The third panel shows an overlay of the traces (0.033 Hz in grey). D, mean ± SEM of n = 9 cells for amplitude (left panel) and decay time (right panel) when stimulating in the molecular layer (grey, ML) or granular layer (black, GL). *P = 0.028 (paired t test). E, mean ± SEM of n = 6 cells for amplitude (left panel) and decay time (right panel) when stimulating in the granular layer at 0.033 Hz (grey) and 0.2 Hz (black). No statistically significant differences were observed.

Stimulation in the granular layer resulted in a faster decay time than stimulation in the molecular layer (Fig. 6B and D; P = 0.028). This suggests that the late current generated by ectopic release is greater when bundles of adjacent synapses are stimulated synchronously. Accordingly, raising stimulation frequency to 0.2 Hz during granular layer stimulation had no effect on EPSC decay time (Fig. 6C and E). Stimulation at 0.2 Hz in the molecular layer brought the mean decay time down to a value closely similar to that observed during granule cell layer stimulation at 0.033 Hz (mean = 34.73 ± 3.1 ms for molecular layer stimulation at 0.2 Hz, n = 17; vs. mean = 36.97 ± 2.39 ms for granular layer stimulation at 0.033 Hz, n = 6). Collectively, these results are consistent with earlier reports that detectable spillover depends on clustered stimulation patterns, but further show that ectopic glutamate is the primary source of this extrasynaptic accumulation at low stimulation frequencies.

Discussion

Suppression of ectopic release reduces decay time of Purkinje neuron EPSCs

Depression of ectopic transmission by activity-dependent depletion of vesicles or pharmacological inhibition of release persistently decreases the decay time of Purkinje neuron EPSCs. Current subtraction suggests that a late current, rising with a mean time to peak of ∼30 ms, is lost after depression of ectopic release.

Several hypotheses could explain these results. It has long been known that parallel fibre–Purkinje cell EPSCs have a prolonged time course relative to parallel fibre–interneuron EPSCs. Several explanations have been proposed for this phenomenon, such as dendritic filtering, imperfect space clamp giving rise to dendritic voltage-activated conductances, crosstalk between synapses and repetitive action potential generation in granule neurons (Llinas & Sugimori, 1980; Barbour et al. 1994; Roth & Hausser, 2001; Isope et al. 2004). Another hypothesis could be that the late current arises from an alternative source of delayed-release transmitter, such as release of glutamate from Purkinje neuron dendrites (Duguid et al. 2007). Changes in these processes could therefore result in the observed decrease in EPSC decay time, providing an alternative explanation for the origin of the late current to depression of ectopic transmission.

Our results indicating no correlation between EPSC amplitude and the magnitude of the decreased decay time argue against the idea that alterations in dendritic filtering or efficiency of space clamp can account for the decreased EPSC decay time. In the case of parallel fibre stimulation generating ‘rebound’ action potentials that can prolong the EPSC (Isope et al. 2004), this phenomenon appears to be negligible in the case of paired pulse stimulation (as used here).

Both rebound and delayed release of transmitter (Atluri & Regehr, 1998; Isope et al. 2004) are thought to be localized to the synaptic cleft, as they show similar time courses to miniature or evoked EPSCs. Slowing extracellular diffusion with large macromolecules, such as dextran, would help to distinguish between glutamate released within the cleft and glutamate released into the extracellular space. For glutamate released within the cleft, dextran should delay the diffusion out of the cleft into the extracellular space. Modelling studies suggest this would increase EPSC amplitude, as diffusion would not limit the maximum concentration reached within the synaptic cleft (Nielsen et al. 2004). In contrast, as others have demonstrated (Satake et al. 2006), the efficiency of spillover through the extracellular space is reduced by dextran, meaning that spillover currents should be reduced in amplitude. Our results show that dextran increases the amplitude of Purkinje neuron EPSCs, but decreases decay time (Fig. 2B). Although interpreting this outcome is not straightforward (as late current amplitude cannot be directly measured), it is consistent with the late current depending on diffusion of glutamate in regions outside of the synaptic cleft. This would be consistent with ectopic release, but not delayed or rebound release.

Autocrine release of glutamate from Purkinje neuron dendrites depends on depolarization-evoked Ca2+ influx, and has so far only been demonstrated in response to climbing fibre stimulation in the range 2 Hz (Duguid et al. 2007). It therefore seems unlikely that this source could underlie the late currents generated by parallel fibre stimulation, which are depressed by stimulation at frequencies >0.033 Hz.

More generally, it seems an unlikely coincidence that these alternative explanations for changes in decay time would be sensitive to the same frequency-dependent plasticity and pharmacological inhibitors as ectopic release, especially given the unusual pattern of long-term depression of ectopic transmission to glia. A more parsimonious interpretation of the foregoing results is that changes in decay time arise from a loss of ectopic release of glutamate from the parallel fibre terminal.

Our experiments with NASP block of GluA2-deficient AMPAR currents suggest that activation of glial AMPARs is not a necessary step in the generation of the late current by, for example, triggering release of transmitters from the glia or (conceivably, although the connection is weak) gap-junctional charge transfer between the cells (Pakhotin & Verkhratsky, 2005). We therefore propose that the simplest explanation of our results is that ectopically released glutamate diffuses to Purkinje neurons, generating a late current that prolongs the EPSC.

The receptors targeted by ectopic release

Although whole cell recordings can give useful information on the amplitude of dendritic currents, the useful spatial information is limited. Furthermore, the anatomical details of ectopic sites are not currently well defined, although ultramicroscopic evidence suggests that at least some of the sites are located on the presynaptic bouton (Matsui et al. 2005). Consequently, it is not clear which receptors would be engaged by ectopically released glutamate.

The greatest density of receptors will of course be in the synaptic cleft, and ectopic glutamate that diffuses from the extracellular space into the cleft would engage these receptors. In such a scenario, the late current would result from spillover from one presynaptic terminal to either adjacent postsynaptic spines or the juxtaposed spine (which would not strictly be spillover as usually defined). A counterargument to this notion, however, is that glutamate transporter density appears to be greatest in the perisynaptic regions of the synapse (Tanaka et al. 1997). This makes intuitive sense, in that this distribution pattern would clear glutamate that has escaped from the cleft, but it would also act as a sink that limits entry into the cleft for ectopically released glutamate. Ectopic release may instead result in the preferential activation of extrasynaptic receptors on neurons, giving an input of decidedly different character to that of synaptic junctions – a weak, distributed signal delivered to extrasynaptic receptors on dendrites rather than clustered receptors at postsynaptic spines. Ectopically released glutamate may also be better positioned for activation of perisynaptic and presynaptic mGluRs associated with synaptic plasticity than glutamate that must escape the synaptic cleft. Indeed, we previously found that activation of perisynaptic mGluR-dependent slow currents was detected with a pair of pulses under conditions when ectopic release was optimal, but required a greater number of pulses when ectopic release had been depressed (Balakrishnan & Bellamy, 2009).

A further consideration is how the pattern of input affects spillover. Our results are consistent with others (Marcaggi & Attwell, 2007), in that distributed inputs are less effective at generating detectible late currents than clustered inputs. This is consistent with the idea that local saturation of transporters is necessary for ectopic glutamate to reach extrasynaptic receptors at appreciable concentrations. How and when such patterns of input are observed in vivo remains an open question.

Physiological roles of late currents

The consequences of late currents for Purkinje cell function are not clear. In general, spillover will prolong the depolarization of the cell soma, increasing the probability of reaching action potential threshold, and extending the time window for temporal summation. This could increase the number of single spikes evoked in Purkinje neurons by granule neuron connections, thereby altering the computational algorithm for the cell (Walter & Khodakhah, 2006).

The other major role proposed for spillover is the activation of perisynaptic receptors. These receptors are linked to induction of plasticity, and many forms are known to be present at the parallel fibre synapse. Postsynaptic mGluR1 receptors mediate long-term depression (Conquet et al. 1994) and post-tetanic depression (Neale et al. 2001b). Furthermore, presynaptic mGluR4 receptors are linked to depression of presynaptic release (Neale et al. 2001a). Accordingly, activation of these receptors during spillover would tend to promote weakening of the connection. Terminals competent for ectopic release may therefore contribute more to total dendritic charge transfer, and be more susceptible to activity-dependent changes in strength, than terminals in which ectopic release has been suppressed.

A noteworthy feature of spillover due to ectopic release is that it would be associated with a very different pattern of activity from conventional spillover. Ectopic release is optimal when fibres fire at low frequencies and in short bursts (Balakrishnan et al. 2011), conditions consistent with weak sensory input to the cortex. This scenario is in striking contrast to conventional spillover, which would occur in regions of the cortex where clearance mechanisms have been overwhelmed by high-frequency bursting.

Under physiological conditions, parallel fibres fire in short bursts with variable baseline frequencies. In the case of anaesthetized mice, the mean baseline frequency was ∼0.5 Hz (Chadderton et al. 2004), whereas awake cats showed more complex patterns of activity: those granule neurons receiving strong sensory input were essentially silent under resting conditions, whereas those cells with weaker input showed spontaneous firing in the range 0.4–10 Hz (Jorntell & Ekerot, 2006). As we have previously argued (Balakrishnan et al. 2011), this pattern of activity would suggest that for most fibres in vivo, ectopic release will be limited (assuming the same frequency-dependence of depression is observed in vivo as ex vivo in slice preparations), as there is an inverse relationship between baseline frequency and strength of ectopic transmission. Given that engagement of perisynaptic mGluR1 or mGluR4 receptors leads to depression of parallel fibre synaptic strength, ectopic spillover may thus be a mechanism for weakening those connections that fire at atypically low rates.

Key points

  • Release of neurotransmitter can sometimes occur outside of the synaptic cleft, a process known as ectopic release.

  • Spillover of the excitatory transmitter glutamate between synapses occurs when parallel fibres in the cerebellum are stimulated at high frequencies.

  • We investigated the effect of activity-dependent and pharmacological reduction of ectopic release on the time course of postsynaptic currents and found that the decay time is reduced, suggesting that ectopic release contributes to spillover at the synapses.

  • This finding suggests that ectopic transmission can cause activation of extrasynaptic receptors even at low frequencies, and so may play a significant role in synaptic plasticity.

  • The results help us understand how signalling in and around the synapse can alter network activity in the cerebellum, a brain region essential for fine motor coordination.

Acknowledgments

The authors thank Julie March for technical assistance.

Glossary

EAAT1

excitatory amino acid transporter 1

EAAT2

excitatory amino acid transporter 2

EAAT4

excitatory amino acid transporter 4

ESC

extrasynaptic current

NASP

1-naphthyl acetyl spermine

NBQX

2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide

TBOA

dl-threo-β-benzyloxyaspartic acid

Additional information

Competing interests

The authors have no competing interests to declare.

Author contributions

S.B., K.L.D., C.J. and T.C.B. conceived and designed the experiments. S.B., C.J. and T.C.B. collected, analysed and interpreted data at the Babraham Institute. K.L.D. collected, analysed and interpreted data at the University of Nottingham. T.C.B. drafted the manuscript, and S.B., K.L.D., C.J. and T.C.B. revised it. All authors have approved the final manuscript.

Funding

This work was funded by the Biotechnology and Biological Sciences Research Council, UK (grant numbers: BB/D018501/1, BB/B500958/1 and BB/J015660/1).

References

  1. Atluri PP, Regehr WG. Delayed release of neurotransmitter from cerebellar granule cells. J Neurosci. 1998;18:8214–8227. doi: 10.1523/JNEUROSCI.18-20-08214.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Balakrishnan S, Bellamy TC. Depression of parallel and climbing fibre transmission to Bergmann glia is input specific and correlates with increased precision of synaptic transmission. Glia. 2009;57:393–401. doi: 10.1002/glia.20768. [DOI] [PubMed] [Google Scholar]
  3. Balakrishnan S, Jackson C, Russell N, Bellamy TC. Ectopic release sites lack fast vesicle recycling mechanisms, causing long-term depression of neuron–glial transmission in rat cerebellum. Glia. 2011;59:82–93. doi: 10.1002/glia.21078. [DOI] [PubMed] [Google Scholar]
  4. Barbour B, Keller BU, Llano I, Marty A. Prolonged presence of glutamate during excitatory synaptic transmission to cerebellar Purkinje cells. Neuron. 1994;12:1331–1343. doi: 10.1016/0896-6273(94)90448-0. [DOI] [PubMed] [Google Scholar]
  5. Bellamy TC, Ogden D. Short-term plasticity of Bergmann glial cell extrasynaptic currents during parallel fibre stimulation in rat cerebellum. Glia. 2005;52:325–335. doi: 10.1002/glia.20248. [DOI] [PubMed] [Google Scholar]
  6. Bellamy TC, Ogden D. Long-term depression of neuron to glial signalling in rat cerebellar cortex. Eur J Neurosci. 2006;23:581–586. doi: 10.1111/j.1460-9568.2005.04588.x. [DOI] [PubMed] [Google Scholar]
  7. Brasnjo G, Otis TS. Neuronal glutamate transporters control activation of postsynaptic metabotropic glutamate receptors and influence cerebellar long-term depression. Neuron. 2001;31:607–616. doi: 10.1016/s0896-6273(01)00377-4. [DOI] [PubMed] [Google Scholar]
  8. Canepari M, Auger C, Ogden D. Ca2+ ion permeability and single-channel properties of the metabotropic slow EPSC of rat Purkinje neurons. J Neurosci. 2004;24:3563–3573. doi: 10.1523/JNEUROSCI.5374-03.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Carter AG, Regehr WG. Prolonged synaptic currents and glutamate spillover at the parallel fibre to stellate cell synapse. J Neurosci. 2000;20:4423–4434. doi: 10.1523/JNEUROSCI.20-12-04423.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Casado M, Isope P, Ascher P. Involvement of presynaptic N-methyl-D-aspartate receptors in cerebellar long-term depression. Neuron. 2002;33:123–130. doi: 10.1016/s0896-6273(01)00568-2. [DOI] [PubMed] [Google Scholar]
  11. Chadderton P, Margrie TW, Hausser M. Integration of quanta in cerebellar granule cells during sensory processing. Nature. 2004;428:856–860. doi: 10.1038/nature02442. [DOI] [PubMed] [Google Scholar]
  12. Conquet F, Bashir ZI, Davies CH, Daniel H, Ferraguti F, Bordi F, Franz-Bacon K, Reggiani A, Matarese V, Conde F. Motor deficit and impairment of synaptic plasticity in mice lacking mGluR1. Nature. 1994;372:237–243. doi: 10.1038/372237a0. [DOI] [PubMed] [Google Scholar]
  13. Duguid IC, Pankratov Y, Moss GW, Smart TG. Somatodendritic release of glutamate regulates synaptic inhibition in cerebellar Purkinje cells via autocrine mGluR1 activation. J Neurosci. 2007;27:12464–12474. doi: 10.1523/JNEUROSCI.0178-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Grosche J, Kettenmann H, Reichenbach A. Bergmann glial cells form distinct morphological structures to interact with cerebellar neurons. J Neurosci Res. 2002;68:138–149. doi: 10.1002/jnr.10197. [DOI] [PubMed] [Google Scholar]
  15. Iino M, Goto K, Kakegawa W, Okado H, Sudo M, Ishiuchi S, Miwa A, Takayasu Y, Saito I, Tsuzuki K, Ozawa S. Glia–synapse interaction through Ca2+-permeable AMPA receptors in Bergmann glia. Science. 2001;292:926–929. doi: 10.1126/science.1058827. [DOI] [PubMed] [Google Scholar]
  16. Isope P, Franconville R, Barbour B, Ascher P. Repetitive firing of rat cerebellar parallel fibres after a single stimulation. J Physiol. 2004;554:829–839. doi: 10.1113/jphysiol.2003.056523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Jorntell H, Ekerot CF. Properties of somatosensory synaptic integration in cerebellar granule cells in vivo. J Neurosci. 2006;26:11786–11797. doi: 10.1523/JNEUROSCI.2939-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Koike M, Iino M, Ozawa S. Blocking effect of 1-naphthyl acetyl spermine on Ca2+-permeable AMPA receptors in cultured rat hippocampal neurons. Neurosci Res. 1997;29:27–36. doi: 10.1016/s0168-0102(97)00067-9. [DOI] [PubMed] [Google Scholar]
  19. Llinas R, Sugimori M. Electrophysiological properties of in vitro Purkinje cell dendrites in mammalian cerebellar slices. J Physiol. 1980;305:197–213. doi: 10.1113/jphysiol.1980.sp013358. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Marcaggi P, Attwell D. Endocannabinoid signalling depends on the spatial pattern of synapse activation. Nat Neurosci. 2005;8:776–781. doi: 10.1038/nn1458. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Marcaggi P, Attwell D. Short-and long-term depression of rat cerebellar parallel fibre synaptic transmission mediated by synaptic crosstalk. J Physiol. 2007;578:545–550. doi: 10.1113/jphysiol.2006.115014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Marcaggi P, Billups D, Attwell D. The role of glial glutamate transporters in maintaining the independent operation of juvenile mouse cerebellar parallel fibre synapses. J Physiol. 2003;552:89–107. doi: 10.1113/jphysiol.2003.044263. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Matsui K, Jahr CE. Ectopic release of synaptic vesicles. Neuron. 2003;40:1173–1183. doi: 10.1016/s0896-6273(03)00788-8. [DOI] [PubMed] [Google Scholar]
  24. Matsui K, Jahr CE. Differential control of synaptic and ectopic vesicular release of glutamate. J Neurosci. 2004;24:8932–8939. doi: 10.1523/JNEUROSCI.2650-04.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Matsui K, Jahr CE, Rubio ME. High-concentration rapid transients of glutamate mediate neural–glial communication via ectopic release. J Neurosci. 2005;25:7538–7547. doi: 10.1523/JNEUROSCI.1927-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mintz IM, Sabatini BL, Regehr WG. Calcium control of transmitter release at a cerebellar synapse. Neuron. 1995;15:675–688. doi: 10.1016/0896-6273(95)90155-8. [DOI] [PubMed] [Google Scholar]
  27. Neale SA, Garthwaite J, Batchelor AM. Metabotropic glutamate receptor subtypes modulating neurotransmission at parallel fibre–Purkinje cell synapses in rat cerebellum. Neuropharmacology. 2001a;41:42–49. doi: 10.1016/s0028-3908(01)00046-6. [DOI] [PubMed] [Google Scholar]
  28. Neale SA, Garthwaite J, Batchelor AM. mGlu1 receptors mediate a post-tetanic depression at parallel fibre–Purkinje cell synapses in rat cerebellum. Eur J Neurosci. 2001b;14:1313–1319. doi: 10.1046/j.0953-816x.2001.01769.x. [DOI] [PubMed] [Google Scholar]
  29. Nielsen TA, DiGregorio DA, Silver RA. Modulation of glutamate mobility reveals the mechanism underlying slow-rising AMPAR EPSCs and the diffusion coefficient in the synaptic cleft. Neuron. 2004;42:757–771. doi: 10.1016/j.neuron.2004.04.003. [DOI] [PubMed] [Google Scholar]
  30. Pakhotin P, Verkhratsky A. Electrical synapses between Bergmann glial cells and Purkinje neurones in rat cerebellar slices. Mol Cell Neurosci. 2005;28:79–84. doi: 10.1016/j.mcn.2004.08.014. [DOI] [PubMed] [Google Scholar]
  31. Roth A, Hausser M. Compartmental models of rat cerebellar Purkinje cells based on simultaneous somatic and dendritic patch-clamp recordings. J Physiol. 2001;535:445–472. doi: 10.1111/j.1469-7793.2001.00445.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Satake S, Song SY, Cao Q, Satoh H, Rusakov DA, Yanagawa Y, Ling EA, Imoto K, Konishi S. Characterization of AMPA receptors targeted by the climbing fibre transmitter mediating presynaptic inhibition of GABAergic transmission at cerebellar interneuron–Purkinje cell synapses. J Neurosci. 2006;26:2278–2289. doi: 10.1523/JNEUROSCI.4894-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Szapiro G, Barbour B. Parasynaptic signalling by fast neurotransmitters: the cerebellar cortex. Neuroscience. 2009;162:644–655. doi: 10.1016/j.neuroscience.2009.03.077. [DOI] [PubMed] [Google Scholar]
  34. Takayasu Y, Iino M, Kakegawa W, Maeno H, Watase K, Wada K, Yanagihara D, Miyazaki T, Komine O, Watanabe M, Tanaka K, Ozawa S. Differential roles of glial and neuronal glutamate transporters in Purkinje cell synapses. J Neurosci. 2005;25:8788–8793. doi: 10.1523/JNEUROSCI.1020-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Takayasu Y, Iino M, Ozawa S. Roles of glutamate transporters in shaping excitatory synaptic currents in cerebellar Purkinje cells. Eur J Neurosci. 2004;19:1285–1295. doi: 10.1111/j.1460-9568.2004.03224.x. [DOI] [PubMed] [Google Scholar]
  36. Tanaka J, Ichikawa R, Watanabe M, Tanaka K, Inoue Y. Extra-junctional localization of glutamate transporter EAAT4 at excitatory Purkinje cell synapses. Neuroreport. 1997;8:2461–2464. doi: 10.1097/00001756-199707280-00010. [DOI] [PubMed] [Google Scholar]
  37. Walter JT, Khodakhah K. The linear computational algorithm of cerebellar Purkinje cells. J Neurosci. 2006;26:12861–12872. doi: 10.1523/JNEUROSCI.4507-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from The Journal of Physiology are provided here courtesy of The Physiological Society

RESOURCES