|Home | About | Journals | Submit | Contact Us | Français|
Cerebellar granule cell to Purkinje cell synapses have been reported to show plasticity when stimulating the parallel fibres, but not when granule cell axons are stimulated in the granular layer. The latter absence of plasticity has been attributed either to the synapses made by ascending granule cell axons lacking some feature needed to evoke plasticity, such as metabotropic glutamate receptors, or to spillover of glutamate between adjacent active synapses being essential for plasticity to occur and having a greater effect for parallel fibre stimulation than for granular layer stimulation. Here we show that both long-term depression (LTD) and endocannabinoid plasticity can depend on interaction between adjacent synapses. These results focus attention on the need to characterise the spatial pattern of parallel fibre activity evoked by physiological stimuli, in order to assess the conditions under which synaptic plasticity will occur in vivo.
Alterations in the strength of cerebellar synapses are thought to underlie the storage of motor patterns by the cerebellar cortex. Changes of synaptic strength mediated by retrograde endocannabinoid signalling and during long-term depression (LTD) at the cerebellar granule cell to Purkinje cell synapse are triggered by activation of postsynaptic mGluR1 metabotropic glutamate receptors (Linden et al., 1991; Hartell, 1994; Conquet et al., 1994; Ichise et al., 2000; Neale et al., 2001; Brown et al., 2003; Safo & Regehr, 2005). Both types of plasticity occur when trains of stimuli are applied to the parallel fibres (the axons of granule cells) in the molecular layer, which is the commonest way to study granule cell to Purkinje cell synaptic transmission. However, neither type of plasticity is evoked when granule cell axons are stimulated near the recorded Purkinje cell in the granular layer (Fig. 1A; Sims & Hartell, 2005; Marcaggi & Attwell, 2005). Two explanations have been proposed for the lack of plasticity evoked by granular layer stimulation.
First, it has been debated (Gundappa-Sulur et al., 1999; Isope & Barbour, 2002) whether the synapses made by ascending granule cell axons, which may have been preferentially activated by granular layer stimulation near a recorded Purkinje cell (Fig. 1A) in the studies of Sims & Hartell (2005) and Marcaggi & Attwell (2005), may differ from parallel fibre synapses, perhaps lacking mGluR1 receptors or (for LTD) being unaffected by climbing fibre stimulation (Sims & Hartell, 2005). Alternatively, we have shown that, in contrast to stimulation in the granular layer, stimulation of the parallel fibres in the molecular layer leads to glutamate diffusion between adjacent activated synapses. This enhances activation of AMPA receptors (Marcaggi et al., 2003), presumably because the effect on surrounding receptors of glutamate diffusing out of a synapse is larger when glutamate is also released at other synapses nearby because of non-linear summation of glutamate’s effects, either via the receptors’ dose-response curve or via local saturation of glutamate uptake. Activation of the mGluR1 receptors which initiate retrograde cannabinoid signalling and LTD is also enhanced by crosstalk between adjacent active synapses evoked by parallel fibre stimulation (Marcaggi & Attwell, 2005). Thus, activation of spatially dispersed synapses by granular layer stimulation (Fig. 1), leading to less activation of mGluR1 than for parallel fibre stimulation, might also explain the lack of plasticity evoked by granular layer stimulation.
Resolving this controversy is important for understanding the conditions under which synaptic plasticity and information storage can occur in the cerebellar cortex. Studies of the plasticity which is evoked by stimulating in the molecular layer to activate nearby parallel fibre synapses are widely assumed to provide an understanding of cerebellar learning occurring in vivo yet, because of the massive number of parallel fibres contacting a Purkinje cell (~150,000), one might imagine that physiological input to the granular layer of the cerebellar cortex would result in the activation of synapses that are more spatially dispersed on the dendritic tree.
We therefore took advantage of the geometrical architecture of the cerebellum to compare the effects of activating the same number of parallel fibre synapses with either a sparse or a dense spatial distribution on the Purkinje cell dendritic tree, while ruling out contaminant activation of synapses made by ascending granule cell axons when stimulating in the granular layer (Fig. 1B). Our data suggest that the lack of endocannabinoid-mediated plasticity (Marcaggi & Attwell, 2005) and LTD (Sims & Hartell, 2005) observed for granular layer stimulation is due to a more spatially dispersed pattern of synapse activation, rather than recruitment of ascending axon synapses with properties differing from those of the parallel fibre synapses.
P18 rats were killed by cervical dislocation followed by decapitation in accord with the UK Animals (Scientific Procedures) Act (1986). For most experiments, the cerebellum was rapidly removed and sliced in the coronal plane (using a Leica VT1000 S slicer) in ~4°C dissection medium containing (mM) NaCl 124, KCl 2.5, NaH2PO4 1, NaHCO3 26, CaCl2 3, MgCl2 1, glucose 10, kynurenic acid 1, bubbled with 95% O2/5% CO2.. The slices (200μm thick) were then stored in that medium for up to 5 hours. Purkinje cells were whole-cell clamped at 33-35°C. The external solution contained (mM) NaCl 124, KCl 2.5, NaH2PO4 1, NaHCO3 26, CaCl2 3, MgCl2 1, glucose 10, GABAzine 0.01 (to block GABAA receptors), bubbled with 95% O2/5% CO2. For the LTD experiments in Fig. 4A, CaCl2 was 2mM and 5μM DPCPX and 1μM CGP55845 were present to block adenosine (A1) and GABAB receptors (as used previously to block possible K+ currents evoked by GABA or adenosine release, by Casado et al. (2002), who also studied LTD in coronal slices). The pipette solution contained (mM) K-gluconate 140, NaCl 4, HEPES 10, MgATP 4, Na3GTP 0.5, K2.5EGTA 0.5, CaCl2 0.1, pH adjusted to 7.3 with KOH. The electrode junction potential was compensated. During LTD experiments the series resistance was monitored every 10 sec; the effect on the EPSC amplitude of moderate variations in series resistance was compensated for (by measuring the effect on EPSC amplitude of spontaneous rapid changes of series resistance that occurred) but if the series resistance changed by more than 50% then the data were discarded. In voltage-clamp mode, cells were clamped at -70mV. In current-clamp mode current was injected to set the resting potential to -70mV.
For the LTD experiments in Fig. 4B the experimental procedure mimicked that used by Sims and Hartell (2006) (the main differences are that they used a different brain slicing solution with NaCl replaced by sucrose, they recorded at room temperature, and had stronger [H+] buffering in their internal solution; in addition, their internal solution recipe did not specify the GTP salt used (so we used Tris3-GTP) or pH (we adjusted the pH to 7.3), and had less Mg than ATP (but we added the same Mg as ATP)). Purkinje cells were whole-cell clamped at room temperature (22°C). The dissection and slicing medium contained (mM) sucrose 250, KCl 2.5, NaH2PO4 1.25, NaHCO3 26, CaCl2 2, MgCl2 1, glucose 10. After slicing, the slices were stored for up to 3 hours in the external solution used for recording which contained (mM) NaCl 120, KCl 2.7, NaH2PO4 1.2, NaHCO3 25, CaCl2 2.5, MgCl2 1.2, glucose 11, picrotoxin 0.02 (to block GABAA receptors), bubbled with 95% O2/5% CO2. The pipette solution contained (mM) K-gluconate 132, NaCl 8, HEPES 30, MgATP 4, Tris3GTP 0.3, K2.5EGTA 0.5, pH adjusted to 7.3 with KOH.
Stimuli were applied from a glass pipette placed at a distance of 175μm from the recorded Purkinje cell (Fig. 1B), either in the molecular layer to activate parallel fibres directly, or in the granular layer (which will also activate parallel fibres: Coutinho et al., 2004). For studying mGluR1 mediated currents or endocannabinoid plasticity, stimulus trains were applied every minute. To evoke LTD, climbing fibre and granule cell stimulation were paired as described in the legend to Fig. 4, using either a protocol similar to that used by Sims & Hartell (2005) and Safo & Regehr (2005) (Fig. 4A), or a protocol identical to that of Sims & Hartell (2006) (Fig. 4B). The size of the AMPA receptor mediated EPSC was monitored by stimulating once per 10 seconds.
Data are presented as mean ± s.e.m. Statistical comparisons were by Student’s t-test.
We studied the mGluR1-mediated postsynaptic current (Tempia et al., 1998), and the plasticity of AMPA receptor mediated synaptic signalling, evoked by stimulating either the granular layer or the molecular layer at a distance of 175μm perpendicular to the plane of the dendritic tree of the recorded Purkinje cell in coronal cerebellar slices (Fig. 1B). This is far larger than the thickness of the planar dendritic tree of the Purkinje cell (<15μm) or the separation between Purkinje cells (34μm) (Takacs & Hamori, 1994), ensuring that granular layer stimulation evokes parallel fibre activity without activity at synapses from ascending granule cell axons onto the recorded Purkinje cell (Fig. 1B). Furthermore, when stimulating in the granular layer, because the stimulated granule cell axons rise to different levels in the molecular layer (Cajal, 1911), the synapses that are activated will be spatially dispersed across the Purkinje cell dendritic tree (Fig. 1B) and no interaction between active synapses is likely. By contrast, when stimulating the parallel fibres in the molecular layer, adjacent parallel fibres are activated (Fig. 1B) and interaction between active synapses mediated by glutamate spillover is likely.
The stimuli for each location were adjusted to produce the same size of fast AMPA receptor mediated EPSC (441±51 and 448±54 pA, for stimulation in the molecular layer and in the granular layer respectively, at -70mV for a single stimulus in 29 cells). Previous work has shown that glutamate spillover between adjacent synapses produced by parallel fibre stimulation has little effect on the peak EPSC produced (Marcaggi et al., 2003), so matching the EPSC amplitudes implies matching the number of synapses activated.
With the Purkinje cell recorded in current clamp mode, a 10 pulse train of stimuli at 200Hz transiently depressed the AMPA receptor mediated fast EPSP produced by single stimulation of the parallel fibres, but potentiated the EPSP produced by granular layer stimulation (Fig. 2A, B). This difference in plasticity in response to stimulation of the molecular and granular layers is similar to that reported for stimulation close to the Purkinje cell (Marcaggi & Attwell, 2005), for which the mGluR1- and endocannabinoid-mediated depression (Neale et al., 2001; Brown et al., 2003) that is evoked by parallel fibre stimulation is abolished by blocking CB1 receptors (while the potentiation seen for granular layer stimulation is unaffected: Marcaggi & Attwell, 2005).
Applying a short burst of stimuli (4 pulses at 200Hz) with the Purkinje cell voltage-clamped revealed that parallel fibre stimulation evoked an AMPA receptor-mediated fast EPSC followed by a robust mGluR1-mediated slow EPSC that was abolished by the mGluR1 specific antagonist CPCCOEt, while granular layer stimulation evoked the fast EPSC followed by a much smaller or absent slow EPSC (Fig. 3A-C). Thus, there is minimal activation of the mGluR1 receptors which trigger endocannabinoid release when stimulating in the granular layer.
When stimulating in the granular layer 175μm from the plane of the dendritic tree of the Purkinje cell, the only synapses activated onto the Purkinje cell will be parallel fibre synapses (Fig. 1B), so the lack of a slow EPSC and lack of endocannabinoid-mediated synaptic plasticity seen for granular layer stimulation cannot reflect a preferential recruitment of ascending synapses which might lack mGluR1 receptors. Instead, it presumably reflects the fact that mGluR1 receptors are only activated when many nearby synapses release glutamate (Marcaggi & Attwell, 2005); a condition which occurs for parallel fibre stimulation (Fig. 3A), but not for the more spatially dispersed activation of synapses produced by granular layer stimulation (Fig. 3B). For parallel fibre stimulation, the post-tetanic depression (~40%) seen in Fig. 2B is smaller than when stimulating near the Purkinje cell (~80%: Marcaggi & Attwell, 2005), probably because the parallel fibres are not in fact completely parallel, and the activated fibres impinge on a more spatially spread out area of the Purkinje cell than is the case when stimulating near the cell, resulting in less interaction between activated synapses.
Pairing parallel fibre stimulation with climbing fibre stimulation produces a long term depression of the parallel fibre fast EPSC, that is dependent on mGluR1 activation (Linden et al., 1991; Hartell, 1994; Conquet et al., 1994; Ichise et al., 2000) and also on endocannabinoid release (Safo & Regehr, 2005), but this LTD is not seen when pairing climbing fibre stimulation with stimulation of the granular layer under the recorded Purkinje cell (Sims & Hartell, 2005). We stimulated 175μm away from the plane of the denditic tree of the Purkinje cell, so that granular layer stimulation activates only parallel fibre synapses and not ascending synapses (Fig. 1B). In this case, although pairing parallel fibre with climbing fibre stimulation evoked LTD, pairing granular layer stimulation with climbing fibre stimulation did not (Fig. 4A).
Sims and Hartell (2006) reported that a slowly developing LTD could still be observed when climbing fibre activation was paired with activation of dispersed parallel fibre inputs (activated by stimulating in the granular layer 150-300μm from the Purkinje cell dendritic plane). As this appears to be in conflict with our observations, we tried to reproduce the Sims and Hartell (2006) result mimicking their conditions (i.e. experiments at room temperature, and with a different composition for the dissecting, extracellular and intracellular media; also, the stimulation protocol differed slightly from the one used in Fig. 4A). However, when parallel fibre synapses were activated by granular layer stimulation 175μm lateral to the Purkinje cell dendritic plane, on average no significant LTD was observed (in 5 recordings lasting more than 55 min following the induction protocol: Fig. 4B).
The work reported here clarifies the pattern of activity in cerebellar granule cell axons that is needed to produce synaptic plasticity at the granule cell to Purkinje cell synapse. These experiments differ from those in the earlier work of Sims & Hartell (2005) and Marcaggi & Attwell (2005) because for granular layer stimulation they ensure that the granule cell synapses activated are those on the parallel fibres, and eliminate any possible activation of synapses from ascending granule cell axons to the recorded Purkinje cell (compare Fig. 1A and 1B). Our data show that the previously reported absence of both endocannabinoid plasticity and LTD when stimulating the granule cell axons in the granular layer is not a result of synapses on the ascending granule cell axon lacking mGluR1 receptors or being unaffected by climbing fibre stimulation (Sims & Hartell, 2005), since the same lack of plasticity is observed when activating only parallel fibre synapses from the granular layer. However, mGluR1 receptors are activated much more strongly when the parallel fibres are stimulated than when the granular layer is stimulated, reflecting these receptors acting as detectors of glutamate spillover between adjacent active synapses (Marcaggi & Attwell, 2005). Plasticity dependent on mGluR1 activation is expected, therefore, to be far stronger when stimulating the parallel fibres than when stimulating the granular layer, as is indeed observed, and in vivo one would expect plasticity to occur most readily when adjacent parallel fibres are active.
In contrast to our results, Sims & Hartell (2006) very recently reported that LTD could be evoked by stimulating in the granular layer at a large distance from the recorded Purkinje cell, even though this did not evoke an mGluR1-mediated current in the cell (Fig. 3). This discrepancy with our results could be due to some difference in the protocols used for the preparation of the slices, the recording of the cells, or the induction of LTD. In particular, Sims & Hartell (2006) performed experiments at room temperature which is expected to reduce glutamate transporter activity and hence promote glutamate spillover. However, even using the same protocols as in Sims & Hartell (2006), we did not observe LTD for stimulation in the granular layer at a large distance from the Purkinje cell (Fig. 4B). At present we have no explanation for this discrepancy.
In addition to the supralinear dependence of the mGluR1-mediated EPSCslow on the number of nearby activated synapses which, as mentioned above, is due to the effects of glutamate spillover (Marcaggi & Attwell, 2005), a nonlinear dependence of the postsynaptic rise in calcium or sodium concentration on the number of nearby activated synapses has also been reported (Eilers et al., 1995; Callaway & Ross, 1997). When paired with climbing fibre activation in order to evoke LTD, increasing the number of activated parallel fibre synapses produces an amplification of the postsynaptic calcium rise, mediated by voltage-gated calcium channels (Wang et al., 2000). We cannot rule out the possibility that such voltage amplification may contribute to favouring the induction of LTD when the activated synapses are in close proximity.
Since both LTD and endocannabinoid plasticity depend critically on the proximity of the activated synapses, the commonly used experimental protocol of stimulating the parallel fibres may evoke plasticity that would not occur in vivo with a more spatially dispersed pattern of synaptic inputs. However, some protocols may induce plasticity even when spatially dispersed synapses are activated (Casado et al., 2002). These data highlight the need for a better understanding of the spatial distribution of granule cell axon activity that impinges on Purkinje cells physiologically, to understand under exactly what conditions plasticity will occur in vivo.
Supported by the Wellcome Trust and a Wolfson-Royal Society award. We thank Angus Silver, Alasdair Gibb and Philippe Isope for comments on the paper.