Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Mar 15.
Published in final edited form as: Neuron. 2016 Sep 1;91(6):1330–1341. doi: 10.1016/j.neuron.2016.08.011

Purkinje cells directly inhibit granule cells in specialized regions of the cerebellar cortex

Chong Guo 1,3, Laurens Witter 1,3, Stephanie Rudolph 1, Hunter L Elliott 2, Katelin A Ennis 1, Wade G Regehr 1,4
PMCID: PMC5853127  NIHMSID: NIHMS811045  PMID: 27593180

Summary

Inhibition of granule cells plays a key role in gating the flow of signals into the cerebellum, and it is thought that Golgi cells are the only interneurons that inhibit granule cells. Here we show that Purkinje cells, the sole output neurons of the cerebellar cortex, also directly inhibit granule cells via their axon collaterals. Anatomical and optogenetic studies indicate that this non-canonical feedback is region specific: it is most prominent in lobules that regulate eye movement and process vestibular information. Collaterals provide fast, slow and tonic inhibition to granule cells, and thus allow Purkinje cells to regulate granule cells excitability on multiple time scales. We propose that this feedback mechanism could regulate excitability of the input layer, contribute to sparse coding and mediate temporal integration.

Introduction

The cerebellum is involved in a multitude of behaviors. It controls balance and motor coordination, and plays a role in non-motor functions, such as cognition and emotion (Koziol et al., 2014; Schmahmann, 2004). Cerebellar dysfunction results in ataxia and dystonia, and more recently has been implicated in autism spectrum disorder (Fatemi et al., 2012; Wang et al., 2014). A single canonical circuit is thought to underlie all cerebellar computations (Eccles et al., 1967; Schmahmann, 2004), although regional differences have been observed (Cerminara et al., 2015; Dino et al., 1999; Heath et al., 2014; Witter and De Zeeuw, 2015).

The first stage of cerebellar computation takes place in the input layer where signals from diverse sources are processed. Signals enter the cerebellum via the mossy fibers (MFs), which excite granule cells (GrCs) and Golgi cells (GoCs), the sole inhibitory interneuron in the input layer. GABA-mediated inhibition is critical for information processing at this stage. GoC to GrC inhibition exhibits a fast phasic component that arises from transient activation of synaptic GABAA receptors (Farrant and Brickley, 2003; Nusser et al., 1998; Rossi and Hamann, 1998) and a slow component that reflects a long-lasting GABA signal that can activate both synaptic and extrasynaptic receptors. In addition, ambient GABA persistently activates extrasynaptic high-affinity GABAA receptors giving rise to tonic inhibition (Brickley et al., 1996; Hamann et al., 2002; Kaneda et al., 1995; Wall and Usowicz, 1997). It has been proposed that the GABA that produces tonic inhibition is primarily released from glia and is not reliant on neuronal activity (Landis et al., 1983; Lee et al., 2010; Rossi et al., 2003). These different components of inhibition work together to create a sparse representation of MF input by controlling GrC excitability (Mitchell and Silver, 2003) and can enhance the fidelity of information transmission in the input layer (Duguid et al., 2012; Marr, 1969).

It is thought that cerebellar outputs do not influence the input layer in the canonical circuit. This contrasts with other areas in the brain where principle neurons often recruit interneurons to regulate local excitability and gain (Bortone et al., 2014; Freund and Buzsaki, 1996; Kato et al., 2013; Olsen et al., 2012; Sik et al., 1994). As GrCs axons ascend into the molecular layer, they give rise to parallel fibers (PFs) and excite Purkinje cells (PCs). PFs also recruit molecular layer interneurons (MLIs) and provide feed-forward inhibition to PCs. The output of the cerebellar cortex is relayed exclusively through PCs, which project to the deep cerebellar nuclei (DCN) to control extracerebellar structures. Since neither MLIs nor PCs make direct connections to GoCs to provide feedback control of the input layer (Hull and Regehr, 2012; Witter et al., 2016), it is thought that cerebellar computations are carried out sequentially: inputs are first processed in the GrC layer and then interpreted by the PCs (Marr, 1969; Albus 1971).

The canonical circuit neglects PC collaterals, which are recurrent axons that branch off from the main axon. PC collaterals are known to form functional synapses onto PCs and most interneurons with the exception of GoCs (Hirono et al., 2012; Watt et al., 2009; Witter et al., 2016). Here we investigate the existence of a direct feedback from the output to the input layer in the cerebellum via the collateral axons. We discovered that PCs makes functional synapses onto GrCs, allowing PCs to provide fast, slow and tonic inhibition. The prevalence of PC to GrC synapse is highly heterogeneous between different regions of the cerebellar cortex, suggesting that it serves specialized functions in specific cerebellar behaviors.

Results

To evaluate the properties of axon collaterals in different regions of the cerebellar cortex, we reconstructed axon collaterals in three functionally distinct lobules: lobule V is involved in motor control of the limbs and tail (Provini et al., 1968), lobule VIb is part of the cerebro-cerebellum and has extensive connections with the forebrain (Paallysaho et al., 1991) and lobule X is involved in vestibular processing (Barmack et al., 1993) (Fig. 1A). To investigate PC collateral axons, mice expressing a synaptophysin-tdTomato fusion protein in PCs were injected with a virus to drive expression of GFP in a limited number of PCs. This allowed us to evaluate both the morphology of isolated PC axon collaterals and the location of their synaptic contacts (Supp. Fig. 1).

Figure 1. The location of Purkinje cell synaptic contacts is lobule specific.

Figure 1

Open in a new tab

A. Schematic representation of a midsagittal section of the mouse cerebellum.

B. Representative example reconstructions of PCs for each of the three cerebellar regions. Dashed lines indicate the borders between the PC layer (PCL) and the molecular layer (ML) and between the PC layer and the GrC layer (GCL). The PC cell bodies and main axons are shown in grey, axon collaterals in turquoise and synapses in red.

C. Summary of the properties of PC collaterals in the three different lobules. Bars represent averages ± SEM, grey dots represent individual reconstructions. * indicates significant difference at p < 0.05.

D. Histograms of synapse locations in relation to the bottom of the PC layer. There is a graded increase in the number of synapses in the GrC layer going from lobule V to VIb to X.

We found that the properties of PC collaterals differ between regions. In lobule V, PC collaterals have the least complex morphology and are confined to a narrow sagittal plane (Fig. 1B, C, Supp. Fig. 2A). Most of their synaptic contacts are near PC cell bodies, and there are very few synapses within the GrC layer (Fig. 1D). PC collateral arbors in lobule VIb are slightly more complex, with an increased thickness and some synaptic contacts within the GrC layer (Fig. 1B–D, Supp. Fig. 2B). Lobule X collaterals have much more complex morphologies. They are longer, they are not confined to a narrow sagittal plane and they make many synaptic contacts both within and outside the PC layer (Fig. 1B–D, Supp. Fig. 2C). It appears that a major target of PC collaterals in lobule X is located in the GrC layer.

The observation that PC collaterals make so many synapses within the GrC layer is surprising. To obtain a comprehensive view across all lobules, we used pcp2cre x synaptophsin-tdTomato mice to identify PC synapses within the GrC layer. Presynaptic boutons of PCs are brightly labeled in these mice, but there is also a low level of background labeling of PC somata and dendrites(Witter et al., 2016). Quantifying synapses in the molecular and the PC layer is therefore challenging, but quantifying GrC layer synapses is straightforward and can be automated. We cut parasagittal slices to visualize lobules II–X and coronal slices to visualize the flocculus (FL) and the paraflocculus (PFL) (Fig. 2A). An example of the synaptophsin-tdTomato labeling is shown for a confocal stack acquired from lobule VIb (Fig. 2B left). The faint extrasynaptic labeling of the PC soma is sufficient to permit visualization and segmentation of the PC layer and the molecular layer. The intense fluorescence of the presynaptic puncta allows the detection of each presynaptic bouton (see Methods and Supp. Movies 1–3). An example 3D analysis is shown with the segmented PC layer in grey and the detected presynaptic puncta in yellow for clarity (Fig. 2B right). We imaged each lobule of the vermis, the FL and the PFL (Fig. 2D). For each presynaptic bouton the distance to the PC layer was determined. Plots of the density of PC synaptic contacts in the GrC layer revealed that there are prominent synaptic contacts near the PC layer for all regions, but few synaptic contacts within the GrC layer in lobules II – VIa, VII – IXa and dorsal PFL (dPFL). There is a moderate number of synaptic contacts in the GrC layer in lobules VIb, IXb, IXc and a high density of synaptic contacts in the ventral PFL (vPFL), FL and lobule X. PC synapses are observed throughout the GrC layer in lobule X, but the density is higher near the PC layer. We observed very similar lobule-specific density distributions in the two mice analyzed (Fig. 2D red vs. blue). To obtain an overview of a whole midsagittal and floccular section of the cerebellum we acquired high-resolution image stacks to obtain a heat map of PC synaptic contacts (Fig. 2C). This map reveals a dense clustering of PC contacts near the PC layer throughout all regions of the cerebellar cortex, but also prominent connections in the GrC layer in lobules VIb, IXb, IXc, X, vPFL and FL.

Figure 2. Purkinje cells synapses in the granule cell layer are prominent in some cerebellar lobules.

Figure 2

Open in a new tab

A. Schematic showing different regions of the cerebellar cortex. Left: midsagittal section, right: flocculus (Fl) and paraflocculus (PFl).

B. Fluorescence from a pcp2cre x synaptophysin-tdTomato mouse (left). Processed image in which Purkinje cell somata are indicated in grey and presynaptic puncta are indicated in yellow (right)

C. Heat map providing and overview of the distribution of synaptophysin-labeled puncta within the cerebellar cortex of a pcp2cre x synaptophysin-tdTomato mouse. Overview as in A, color scale bar in volume percent.

D. Examples of analyzed images for determining synaptic density for the indicated regions (left columns). Quantification of synaptic density as a function of distance from the bottom of the PC layer (right columns, red and blue traces for two different animals)

The prevalence of PC synapses within the GrC layer raises the possibility that PCs might form inhibitory synapse onto GrCs. To test for functional PC to GrC synapses we selectively expressed ChR2 in PCs using pcp2cre x ChR2-EYFP mice. We made whole cell recordings of GrCs in voltage clamp configuration and stimulated nearby PC axons with a blue laser (50 Hz, 160 mW/mm2, 473 nm). This stimulation often evoked large IPSCs in lobule X (27/29 cells, 89±17 pA, Fig. 3A bottom, 3C right). The GABAAR antagonist GABAzine (SR95531) (5 µM) eliminated these synaptic responses (reduced to 2.3±0.5% of control, n=7, Fig. 3B). In comparison, optically evoked IPSCs are smaller and less frequently observed in lobule VIb GrCs (15/19 cells, 27±5 pA, Fig. 3A middle, 3C middle). In lobule V, very few GrCs responded to stimulation at all (3/15 cells, 6.9±1.0 pA, Fig. 3A top, 3C left). The electrophysiology is in good agreement with our anatomical results in which PC collateral boutons in the GrC layer are rare in lobule V, sparse in VIb but dense in lobule X (Fig. 1C, 2A,C).

Figure 3. Optogenetic activation of ChR2 in Purkinje cells evokes lobule-specific inhibition of granule cells.

Figure 3

Open in a new tab

A. Light-evoked synaptic responses in lobules V, VIb and X.

B. Light-evoked synaptic response in lobule X in the absence of (control, black) or with (grey) gabazine (SR95531, 10µM) to block GABAA receptors.

C. Summary of evoked current across three lobules.

D. Schematic of optical and electrical stimulation

E. A representative cell showing the effect of the mGluR2 agonist LY354740 on electrically (top) and optically (bottom) evoked IPSCs.

F. Summary time course of normalized total charge transfer from optically (turquoise) or electrically (black) evoked IPSC during LY354740 wash-in (n = 5).

The optically evoked inhibition of GrCs is consistent with direct inhibition by PCs, but it is important to examine alternative explanations. GoCs can directly inhibit GrCs, and PCs are known to inhibit LCs (Crook et al., 2007; Hirono et al., 2012; Witter et al., 2016), and LCs can in turn inhibit GoCs (Dieudonne and Dumoulin, 2000). This indirect pathway (PC to LC to GoC to GrC) is unlikely to explain our results because the latencies observed after light stimulation were generally short (1.9 ± 0.1 ms, n=6) and in agreement with a direct synaptic connection, rather than a trisynaptic pathway. A second possibility is that leaky expression of ChR2 in GoCs allows GoCs to contribute directly to light evoked responses. We took advantage of the fact that GoCs but not PCs express mGluR2 (Geurts et al., 2001). Activation of mGluR2 can suppress both evoked and spontaneous release from most GoCs (Watanabe and Nakanishi, 2003) (Supp. Fig. 3). To test whether optically evoked inhibitory currents in GrCs in part originate from GoCs we alternated optical and electrical stimulation while recording from GrCs in the GrC layer of lobule X (Fig. 3D). Application of the mGluR2 agonist LY354740 (200 nM) robustly reduced electrically evoked IPSCs in all cells tested (n=5) (Fig. 3E, top, 3F black) but had no effect on the optically evoked IPSCs (Fig. 3E bottom, 3F turquoise). The strong effect of LY354740 on only the electrically evoked IPSCs suggests that GoC IPSCs are the dominant source of electrically evoked inhibition, but they do not contribute to optically evoked IPSCs.

To further examine whether PC collaterals form direct synaptic contacts onto GrCs, and to assess the properties of synaptic inputs made by a single PC, we performed paired recordings from PCs and GrCs. Experiments were conducted in lobule X where collateral boutons are dense within the GrC layer (Fig. 2C), and where light activation of PCs leads to prominent short-latency IPSCs in GrCs (Fig. 3A). GrCs were recorded in whole-cell voltage clamp and spontaneous spiking of PCs was monitored with an on-cell electrode. PC to GrC connections were rare for randomly chosen pairs (1.4%, 4/280). However, there are many PCs within lobule X and they have extensive collaterals. Therefore, even though it is unlikely that a given PC/GrC pair will be connected, it is conceivable that a large fraction of GrCs in lobule X will directly contacted by at least one PC.

The typical properties of single PC to GrC synapses are shown for a representative pair (Fig. 4A–C). The on-cell recording revealed spontaneous PC firing at 21 Hz (Fig. 4A top). Voltage clamp recordings from a postsynaptic GrC revealed spontaneous IPSCs (sIPSCs) with an average amplitude of 33 pA at a rate of 4.3 Hz, which was significantly less frequent than PC firing rates (Fig. 4A bottom). Most sIPSCs (53%) occurred between 1 to 2 ms after a PC spike with an average latency of 1.6 ms (Fig. 4B). The average latency for all cells tested was 1.58±0.07 ms (Fig. 4D, n=4). A fraction of sIPSCs was not time-locked to spikes in the presynaptic PC (Fig. 4A bottom asterisk). These could come from either GoCs or other PCs. Conversely, many PC spikes were not followed by an IPSC (Fig. 4B grey traces and Fig. 4C grey histogram), indicating that the action potential often failed to evoke neurotransmitter release. A low success rate was observed in all cell pairs (0.13±0.02, n=4, Fig. 4D). The amplitude histogram of AP-evoked IPSCs is well fit by a Gaussian function (Fig. 4C, red trace). The average potency was 45±13 pA (Fig. 4D, n=4). These recordings establish that PCs form direct synaptic contacts onto GrCs and mediate fast inhibition.

Figure 4. Single Purkinje cells directly inhibit granule cells.

Figure 4

Open in a new tab

A. Representative traces of PC spiking monitored with an on-cell electrode and GrC synaptic currents monitored with whole-cell voltage clamp. Three sIPSC time locked to PC spike can be seen, one spurious sIPSC is labeled with an asterisk.

B. Spike triggered sample traces of cell attached PC (top), whole cell GrC recordings (middle) with successes (black), failures (grey) and the average success (red), and the latency histogram (bottom).

C. Amplitude histogram showing failures (grey) and successes (black) overlaid by its Gaussian fit with mean = 33 pA and SD = 12 pA (red).

D. Summary of properties of single PC-GrC synapses.

E. Representative traces of PC spiking and GrC-IPSCs during reversible shutdown.

F. Abolishing PC activity (top) reduced the frequency of IPSCs (middle) and decreased the amplitude of the tonic current (bottom).

G. Summary statistics of single PC's inhibition onto a GrC.

We also assessed the possibility that single PCs can contribute to tonic inhibition of GrCs. As shown for another PC-GrC pair (Fig. 4E, F), we silenced the PC by setting the voltage of the on-cell electrode to a negative potential (Fig. 4E top, 4F top). This reduced the frequency of spontaneous IPSCs recorded in the postsynaptic GrC to 62% of baseline levels (Fig. 4E bottom, 4F middle, 4G left). We also analyzed the tonic current for the GrC, which revealed a slowly decaying and slowly recovering change in the tonic current of 1.88 pA when the firing of a single PC was suppressed (Fig. 4F bottom). Similar effects on tonic inhibition were seen in all connected pairs (Fig. 4G right), with the magnitude of the effect on tonic inhibition linearly related to the initial PC firing frequency (R2=0.99, Figure 4G middle). These results show that spontaneous activity from a single PC can provide tonic inhibition to GrCs.

The firing rates of large populations of PCs are modulated during many behaviors (Armstrong and Edgley, 1984; Kase et al., 1980; Lisberger and Fuchs, 1978a). We therefore used optogenetics to determine how firing rate modulation of many PCs influences inhibition of the GrC layer. We either reduced or increased PC firing rates using mice that either expressed halorhodopsin (Halo) or ChR2 selectively in PCs (pcp2cre x Halo-EYFP and pcp2cre x ChR2-EYFP, respectively). PC spiking was monitored with an on-cell electrode and inhibition in GrCs was monitored in voltage clamp. Experiments were conducted in lobule X and a step illumination over a broad area (spot size, 1 mm2) was used to regulate the firing of most PCs within lobule X of the brain slice (Fig. 5B).

Figure 5. Optogenetic control of spontaneous Purkinje cell firing regulates phasic and tonic inhibition of granule cells.

Figure 5

Open in a new tab

A. PC firing was monitored with an on cell electrode and low or moderate light levels were used to either decrease or increase PC firing (top). The effect of light-induced changes in PC firing on GrC IPSCs was monitored with a whole-cell electrode (bottom).

B. Experiments were performed using slices from animals in which PCs expressed either halorhodopsin (Halo) or channelrhodopsin2 (ChR2).

C. Summary average (c3) and time course of PC firing rate during Halo (c1) and ChR2 (c2) manipulation.

D. Summary average (d3) and time course of sIPSC frequency during Halo (d1) and ChR2 (d2) manipulation.

E. Summary average (e3) and time course of tonic inhibition during Halo (e1) and ChR2 (e2) manipulation.

Suppressing PC firing in Halo-expressing mice reduced phasic and tonic inhibition of GrCs. As shown for an example cell, a step of low intensity illumination at 0.3 mW/mm2 or 0.4 mW/mm2 in the Halo-expressing animal reversibly slowed down or abolished spontaneous PC firing, respectively (Fig. 5A1 top). The average PC firing rate was reduced from 34.1±2.5 Hz, to 15.2±1.9 Hz (0.3 mW/mm2), or to 0.14±0.04 Hz (0.4 mW/mm2) (n=20) (Fig. 5C1). sIPSCs were present in 17 of 18 GrCs. These experiments were conducted in the presence of the mGuR2 agonist LY354740 to suppress GoC inputs. The persisting spontaneous events observed in GrCs were consistent with the presence of PC-mediated direct phasic IPSCs as observed in paired recordings (Fig. 4). As shown for an example cell, illumination with yellow light decreased the frequency of sIPSCs in GrCs (Fig. 5A1 bottom). The average sIPSC rate in GrCs was reduced from 4.15±0.51 Hz to 2.29±0.35 Hz (0.3 mW/mm2) and to 1.37±0.29 Hz (0.4 mW/mm2) (n=18) (Fig. 5D1). The robust reduction in the frequency of sIPSCs that persisted in the presence of mGluR2 agonist again suggests that many GrCs in lobule X are directly inhibited by PCs. Illumination also decreased the tonic current mediated charge transfer by 4.19±0.93 pC (0.3 mW/mm2) and 6.7±1.7 pC (0.4 mW/mm2) (n=18) (Fig. 5E1). Thus, the baseline activity of PCs can be a significant source of tonic inhibition in GrCs.

In an analogous set of experiments, elevating PC firing in mice expressing ChR2 increased sIPSCs and tonic inhibition of GrCs. As shown for example cells, low intensity illumination in animals expressing ChR2 reversibly increased spontaneous PC firing (Fig. 5A2 top) and the rate of sIPSCs in GrCs (Fig. 5A2 bottom). Step illumination of 10 µW/mm2 and 25 µW/mm2 increased the average PC firing rate from 34.1±2.5 Hz, to 62.7±4.4 Hz and 107.0±9.7 Hz (n=20) respectively (Fig. 5C2). It also elevated the average sIPSC rate in GrCs from 4.2±0.5 Hz baseline, to 6.5±1.5 Hz and 10.2±2.2 Hz (n=14; Fig. 5D2), and increased the tonic current mediated charge transfer by 3.8±0.6 pC and 9.3±1.5 pC (n=14; Fig. 5E2). The summaries of the Halo and ChR2 experiments illustrate the linear relationships between PC firing and sIPSC frequency or tonic inhibition in GrCs (R2 = 0.99, p < 0.001 in both cases; Fig. 5 C3–E3). It is apparent that both sIPSC frequency and tonic components maintained their sensitivity to changes in the cerebellar output across a large physiological range of PC firing rate.

Lastly, we examined the relative contribution as well as the time course of sIPSCs and tonic inhibition during a step change in the activity of nearby PCs. When we compared the stimulus-evoked charge transfer mediated by sIPSCs and tonic inhibition during the one-second stimulation. Tonic inhibition accounted for over 90% of the total charge transfer. This is consistent for both Halo and ChR2 manipulations (91±3% and 90±2 % respectively, two-sample t-test: p = 0.65, Fig. 6B) suggesting that the tonic component likely plays a larger role in controlling GrC excitability. Interestingly, the time courses of the tonic current after normalizing to the steady state amplitude (Fig. 6A bottom) are nearly identical regardless of the direction and amplitude of PC firing rate modulation (R2 > 0.99, p<0.001 for all pairwise correlation). When fitted with a single exponential curve, the tonic current has a slow and characteristic time constant of 312±22 ms (Fig. 6A bottom, Fig. 6C). In contrast, both PC firing rate and sIPSC frequency changed rapidly (Fig. 6A top, middle) with a time constant of 9±6 ms and 35±28 ms respectively (Fig. 6C). This indicates that the slow time constant of the tonic component is not a result of our failure to rapidly switch PC firing rate or delayed/asynchronous transmitter release at the PC-GrC synapse.

Figure 6. The time course of Purkinje cell-mediated tonic inhibition of granule cells.

Figure 6

Open in a new tab

A. Normalized time course of PC firing rate, sIPSC frequency and tonic inhibition during Halo and ChR2 manipulations (see legend) and average time course across all four conditions (black).

B. Fraction of change in total charge transfer mediated by tonic inhibition during PC firing rate decrease (Halo) or increase (ChR2)

C. Time constant (τONSET) of PC firing rate, sIPSC frequency and tonic inhibition change at the onset of light stimulation.

Discussion

Here we identify an unexpected source of feedback inhibition from PCs to GrCs in the cerebellar cortex that is highly lobule dependent. We find that PC to GrC synapses contribute to rapid and slow components of inhibition, and to tonic inhibition. This direct modulation of the input layer by the output neurons of the cerebellar cortex requires a revision of the canonical circuit and has important functional implications.

PC synapse onto GrCs via recurrent axon collaterals

Both anatomical and functional data indicate that PCs synapse onto GrCs via axon collaterals (schematized in Fig. 7A). By expressing a fluorescently labeled presynaptic marker in PCs, we found that PC collaterals make synapses within the GrC layer in a lobule-specific manner. Furthermore, by expressing ChR2 in PCs, we found that optical stimulation of PC collaterals evoked large IPSCs in GrCs. The agreement between the average evoked currents and the density of collateral boutons in three representative lobules (V, VIb and X) suggests that the density of collateral boutons in the GrC layer strongly predicts the presence of functional PC to GrC connections. Previously, GoCs were thought to be the only neurons providing inhibition to GrCs. It was therefore important that we ruled out a contribution of GoCs to this optically evoked inhibition by selectively suppressing GoC inhibition with an mGluR2 agonist. Furthermore, paired recordings provided direct evidence that PCs form inhibitory synapses onto GrCs that have a low probability of release.

Figure 7. Hypothesized functions of Purkinje cell to granule cell feedback.

Figure 7

Open in a new tab

A. Circuitry of the cerebellar cortex showing the PC collateral and PC to GrC feedback.

B. PC to GrC feedback changes the input/output transformation of GrCs in a manner that makes it more difficult for basal MF activity to evoked spikes in GrCs.

C. A transient MF input (black, top) inhibits PC output (turquoise, bottom left) through the inhibitory feedforward circuit (middle left). With inhibitory feedback (middle right), MF input is integrated in time by PC output (turquoise, bottom right).

Spontaneous PC activity is a major source of ambient GABA

Our results indicate that in select lobules PCs can be an important source of the ambient GABA that controls tonic inhibition. GrCs express numerous types of GABA receptors composed of different subunits. Receptors containing α6- and δ-subunits are well suited to sensing low levels of ambient GABA and mediating tonic inhibition because they have high affinity for GABA (Brickley et al., 1996; Hamann et al., 2002; Rossi and Hamann, 1998), they do not desensitize and they are located on extrasynaptic sites throughout the dendrites and soma (Nusser et al., 1998). The source of ambient GABA is thought to be developmentally regulated. In young animals (p15) tonic inhibition is controlled by activity-dependent GABA release from GoCs (Hamann et al., 2002; Rossi and Hamann, 1998). In older animals (p30) it is thought that activity-dependent GABA release from neurons is not a major source of ambient GABA, but instead GABA release from astrocytes (Lee et al., 2010; Rossi et al., 2003) and the activity of GABA transporters (Attwell et al., 1993; Chiu et al., 2005; Rossi et al., 2003) control ambient GABA levels and tonic inhibition. Our paired recordings established that single PCs accounted for several picoamps of tonic current. Optogenetic suppression of many spontaneously firing PCs resulted in a substantially larger reduction in tonic current, suggesting that GrCs receive feedback from multiple PCs. Moreover, it is likely that the effect of PC activity on tonic currents and GrC excitability is even larger in vivo, because many collaterals are severed during slicing, which can lead to an underestimate of the contribution of PC activity to tonic inhibition in slice experiments. Thus, in select lobules, PCs contribute to ambient GABA and tonic current and thereby allow PC activity to control the excitability of the GrC layer.

Lobule-specificity of PC-GrC feedback reflects functional compartmentalization

The density of PC collateral boutons in the GrC layer exhibits striking regional heterogeneity (Fig 2C, D). Moreover, the agreement between the average evoked currents and the density of collateral boutons in three representative lobules (V, VIb and X) suggests that the density of collateral boutons in the GrC layer predicts the prominence of functional PC to GrC connections. Based on our analysis using a genetically expressed presynaptic marker in PCs, we identified several regions where PC synapses are prominent in the GrC layer: lobules IX–X (also known as the uvula/nodulus), the FL/vPFL and lobule VIb. All of these regions receive vestibular and visual inputs and play diverse roles in visuomotor control (Barmack et al., 1993; Hoddevik, 1977; Voogd and Barmack, 2006). The FL and lobules IX–X are involved in optokinetic and vestibulo-ocular reflexes, while the vPF and lobule VIb are involved in pursuit eye movements in foveate animals, although their role in mice is less clear(Voogd and Barmack, 2006). In general, regions where PC to GrC feedback is prominent are involved in slow processes where slow feedback inhibition could be useful. In contrast, PC to GrC feedback is not prominent in lobules II–V, which are involved in the control of limb movements (Provini et al., 1968) and in the dPF, which is involved in fast eye movements such as saccades (Nagao et al., 1997). For such rapid movements slow feedback on the scale of hundreds of milliseconds provided by the PC to GrC synapse, might not be useful. Thus, the observed distribution of PC to GrC synapses suggests that PC to GrC synapses may help to solve a unique set of control challenges encountered in slow visuomotor control.

Functional roles of PC to GrC inhibition

Previous studies established that fast and slow GoC to GrC inhibition and tonic inhibition reduce GrC excitability (Crowley et al., 2009; Mitchell and Silver, 2003; Rothman et al., 2009). GoCs are present in the GrC layer throughout the cerebellar cortex and do not exhibit the profound regional specificity of PC to GrC synapses. GoCs are contacted by both MFs and GrCs and are thought to mediate both feedforward and feedback inhibition (Eccles et al., 1967; Palay and Chan-Palay, 1974), but numerous studies suggest that feedback GoC to GrC inhibition is weak (Dieudonne, 1998; Duguid et al., 2015). Our findings indicate that the circuitry of the cerebellar cortex must be revised to include a prominent source of feedback inhibition from PC to GrCs in some lobules (Fig. 7A), and that PC feedback can provide fast, slow and tonic inhibition to control sensory transformation in the GrC layer in a diverse and flexible manner.

The tonic inhibition from basal PC activity will likely reduce the fraction of activated GrCs in the input layer to create sparse activity patterns (Hamann et al., 2002) (Fig. 7B). Classical Marr-Albus theories and computational models of the cerebellum indicate that sparse coding in the input layer expands the number of patterns that can be stored at the output layer (Albus, 1971; Marr, 1969; Tyrrell and Willshaw, 1992). This additional source of tonic inhibition may be important in regions where PC to GrC feedback is observed, because vestibular inputs to these regions are excitatory, fire rapidly (Arenz et al., 2008; Barmack and Yakhnitsa, 2008; Lisberger and Fuchs, 1978b) and could evoke spontaneous firing in a large fraction of GrCs in the absence of robust inhibition.

The PC to GrC synapse could also allow the cerebellar cortex to perform temporal integration. Many cerebellar dependent behaviors such as maintaining balance, controlling eye movement, and vestibular ocular responses require temporal integration to create sustained responses (Green and Angelaki, 2003; Robinson, 1974). However, the neural mechanism responsible for such integration and the extent to which this operation is intrinsic to the cerebellum is unclear. The canonical cerebellar circuit can sustain activity in the presence of constant excitation (Medina and Mauk, 2000); yet many vestibular functions require activity that persists for seconds to tens of seconds beyond a transient input (Robinson, 1974), which is challenging for a feed-forward circuit. One intriguing possibility is that the prevalence of collateral feedback to the GrCs reflects a circuit specialization required for the cerebellum (lobule IX, X, FL and PFL) to carry out temporal integration. Temporal integration can be realized with recurrent excitation or with two mutually inhibiting populations that effectively create a positive feedback. This can be achieved with the known circuitry of the cerebellar cortex combined with PC to GrC inhibition. MFs excite GrCs, which in turn directly excite PCs; they also inhibit PCs disynaptically by recruiting MLIs (Fig. 7A). This can allow GrC layer activation to produce a net inhibition of PCs (Dizon and Khodakhah, 2011; Gao et al., 2016). Moreover, PC firing in the vestibulocerebellum responds out-of-phase with the ipsilateral vestibular afferents (Barmack and Yakhnitsa, 2003), which is consistent with GrC activity reducing PC firing. In the absence of feedback, transient MF activation transiently reduces PC output (Fig. 7C left). But with the addition of an inhibitory feedback from PC to GrC, this circuit is reminiscent of the classical neural integrator (Cannon et al., 1983) and a transient increase in MF activity produces a sustained reduction in PC activity (Fig. 7C right). This neural integrator has two advantages. First, the tonic component of PC to GrC synapses has a slow time constant (τ~300 ms), which greatly improves the stability of the feedback loop compared to neural integrators that rely on rapid positive feedback (Seung et al., 2000). Second, because neural integrators become more resilient to perturbation as they increase in size (Cannon et al., 1983), the massive number of GrCs and PCs promise to make a cerebellar neural integrator robust to perturbation.

Experimental Procedure

Viral injection

All viral injections were done in p0 pups of a pcp2-cre Jdhu x Synaptophysin-tdTomato litter. Pups were anaesthetized using ice and held in place by hand for intraventricular injections. Virus (AAV1.CAG.FLEX.EGFP.WPRE.bGH, UPenn Vector Core) was injected bilaterally into the lateral ventricles(Kim et al., 2013). To determine the success of virus injection, the virus was diluted 1:1 (to a final titer of 5e11) with a filtered, saturated solution of Fast Green in saline. Upon successful injection of 200 nl virus, fast-green mixture the needle was retracted and the ventricles were colored green from the dye. Following the injection, pups were warmed on a heating pad to body temperature and returned to the mother.

Confocal imaging

Mice were deeply anaesthetized and transcardially perfused with 4% PFA in PBS (pH=7.4). The brain was then removed and post-fixed over night at 4 degrees in the same solution. Parasagittal vermal slices of the cerebellum were cut at 50 µm thickness on a Leica1000 vibratome. Slices were then mounted on superfrost slides (VWR), covered by Prolong Diamond mounting medium (Invitrogen) and a glass coverslip. After >24 h sections were acquired with an Olympus FV1200 confocal microscope.

Collateral reconstruction

Mice were anaesthetized and perfused as described above. 150 µm thick slices were cut, mounted on superfrost slides, covered with Mowiol 4–88, and imaged on an Olympus FV1200 confocal microscope. Whole collaterals were imaged by acquiring multiple stacks of images in grid like fashion through each section. Stacks were stitched together to obtain large-area stacks, and cells were traced using ImageJ (Longair et al., 2011; Preibisch et al., 2009). Reconstructed neurons were further processed in Imaris by converting the reconstruction into a 3d surface object and masking the tdTomato signal within this reconstruction to ultimately create a second 3d surface object of the tdTomato signal contained within the reconstruction. The molecular and Purkinje cell layers were segmented (Segmentation editor, ImageJ) and for each punctum the position relative to the PC layer was calculated in Matlab: For each punctum we determined the shortest distance to the bottom of the molecular layer. To this distance we added the thickness of the Purkinje cell layer to obtain for each punctum the distance to the bottom of the Purkinje cell layer.

Quantification of Purkinje cell synaptic contacts

Image stacks of the granule and Purkinje cell layers of the different lobules of the cerebellar vermis and the floccular complex were processed in Imaris as follows. First, an initial puncta segmentation was created using the surface masking module using an automatic threshold and local background subtraction radius of 1.5 µm and smoothing of 0.265 µm. This initial segmentation was then subtracted from the raw fluorescence to produce an image containing only dimmer, diffuse fluorescence. Then, using this image, PC somata and dendrites were segmented with a manual threshold using the surfaces module (background subtraction of 10 µm and smoothing of 4 µm). The threshold was set in the “elbow” of the intensity histogram (where the histogram most rapidly changes). Finally, this fluorescence within the segmented PC somata and dendrites was masked out, allowing true synaptic contacts made by PCs to be detected via an automatic threshold to ensure reproducibility between lobules (background subtraction of 1.5 µm and smoothing of 0.265 µm). Finally, for each punctum the distance to the PC somata was then determined by creating a distance transformation from the PC somata segmentation.

Overview of Purkinje cell synaptic contacts

To quantify synaptic contact location and density on whole cerebellar sections in an automated fashion we used a three-stage approach: PC/molecular layer masking, synapse detection and then local density calculation. In the first step the Weka Fiji plugin was used to train a 2D Random Forest (RF) pixel classifier with four classes corresponding to the molecular layer, PC layer, synapses and the GrC layer. Training was performed using 10 randomly selected z-planes, 5 from the midsagittal section and 5 from the flocculus. A segmentation was then produced from a weighted combination of the per-class probability maps, followed by thresholding and post-processing with morphological operations. Because the original whole-section images were ~30GB, RF training and segmentation were performed on versions of these images that had been down-sampled by 4× in X and Y. This mask was then applied to the full-resolution synaptophysin-tdTomato fluorescence, and synapse detection was performed in a distributed parallel fashion using Matlab in an HPC environment. Synapse detection was based on a local background subtraction, followed by PSF-matched Gaussian pre-filtering and then thresholding at 3× the STD of intensities within the adjacent masked molecular layer/PC layer, followed by a morphological watershed to split adjacent synapses. Finally, local synaptic density determined by calculating the volume fraction occupied by segmented synapses within a cubic moving window of width 18 microns with a step size of 9 microns.

Slice preparation for in vitro physiology

Sagittal slices were obtained from the cerebellar vermis of p30–40 C57BL/6 wild type or pcp2cre mice (Jackson labs, stock number 010536) crossed to ChR2-EYFP (Ai 32, Jackson labs, 024109) or halorhodopsin (eNpHR3.0)/EYFP mice (Jackson labs, 014539). Animals were anesthetized with an intraperitoneal injection of ketamine/xylazine/acepromazine mixture (100/10/3 mg/kg) and perfused intracardially with an ice-cold solution containing (in mM) 110 Choline Cl, 7 MgSO4, 2.5 KCl, 1.2 NaH2PO4, 0.5 CaCl2, 11.6 Na-ascorbate, 2.4 Na-pyruvate, 25 NaHCO3 and 25 glucose equilibrated with 95% O2 and 5% CO2. The cerebellum was dissected out and sliced at 250–270 µm thickness with a Leica VT 1200s vibratome while in ice-cold cutting solution as above. Slices were transferred to a submerged chamber with artificial cerebral spinal fluid (ACSF) containing (in mM) 125 NaCl, 26 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 1 MgCl2, 1.5 CaCl2, and 25 glucose (pH 7.4, osmolarity 315) equilibrated with 95% O2 and 5% CO2, incubated for 30 minutes at 32°C. Slices were then kept at room temperature for up to 6 hours.

Electrophysiology

Slices were superfused at 32°C in ACSF containing 5 µM NBQX, 2 µM R-CPP, 1 µM CGP and 1 µM strychnine to block AMPA receptors, NMDA receptors, GABAB receptors and glycine receptors, respectively. Visually guided whole cell recordings of GrCs were obtained using an Olympus BX51WI microscope equipped with differential interference contrast (DIC). Patch-pipettes were (4–8 MΩ) pulled from borosilicate capillary glass (World Precision Instruments) with a Stutter P-97 horizontal puller. The chloride-based internal for voltage-clamp GrC recordings contained (in mM) 135 CsCl, 4 NaCl, 0.5 CaCl2, 10 HEPES, 5 EGTA, 2 Mg-ATP, 0.5 Na-GTP and 2 QX-314 (pH 7.2 with CsOH, osmolarity 315). GrCs were identified based on a size (cell body diameter 7–8 µm), membrane capacitance (3–5 pF) and high input resistance (~ 0.8–1.2 GOhm). Access resistance was monitored online with a 10 mV hyperpolarizing pulse. Cells with >20% changes in access resistance were rejected from analysis. IPSCs were recorded at a holding potential of −70 mV without junction potential correction.

For paired recordings and minimal light stimulation experiments, PCs were monitored using an on-cell glass electrode (1–2 MΩ) filled with ACSF. PC-GrC pairs were recorded by obtaining a whole cell recording of a GrC first. PCs were recorded in cell attached mode for GrCs with a spontaneous IPSC frequency > 5Hz. To transiently abolish spontaneous firing of PCs using an extracellular electrode, a loose seal (> 500MΩ) was first obtained and the recording electrode was hyperpolarized to -80 mV under voltage clamp for 2 seconds (Barbour, Isope, 2000). Firing was continuously monitored online. The command voltage was adjusted accordingly if the firing resumed prior to end of the voltage step.

Optical and electrical stimulation

Slices from pcp2cre x ChR2-EYFP mice were kept in the dark. IPSCs were optically evoked in GrCs by delivering brief (0.5 ms) high-intensity (160 mW/mm2) pulses of blue light at 50 Hz focused to a 60 µm diameter spot near the recorded cell. The light stimulus was delivered with a laser (MBL-III-473-50 mW, Optoengine, Midvale, UT) coupled to the excitation pathway of a BX50WI upright microscope (Olympus) and focused onto the slice with a 60× water immersion objective. To record electrically evoked IPSCs in GrCs, nearby Golgi cell axons were stimulated with an ACSF-filled monopolar glass electrode (1–2 MΩ) with trains of stimuli (1 ms pulses at 50 Hz) via a linear stimulus isolator (A360, WPI). For simultaneous optical and electrical stimulation experiments, the mGluR2 agonist LY354740 (2 µM) was bath-applied to selectively abolish GABA release from Golgi cells.

For minimal intensity optical stimulation of PC firing, slices from pcp2cre x eNpHR3.0/EYFP or pcp2cre x eNpHR3.0/EYFP mice were used for optical inhibition or excitation of spontaneous activity, respectively. For optogenetic inhibition of PCs in halorhodopsin expressing animals, the mGluR2 agonist LY354740 was included in the bath to isolate PC-mediated inhibition. Full-field illumination of lobule X (1 mm diameter spot) with yellow light (590 nm) was delivered at low (0.3 mW/mm2) and high (0.4 mW/mm2) intensities to either decrease or abolish spontaneous activity for 1 s. Stimulation was delivered using a 160 mW Amber LED (590 nm, ThorLabs), coupled to an optical fiber in the excitation pathway of the microscope and focused onto lobule X through a 10× water immersion objective. For optogenetic excitation of PCs in ChR2 expressing animals, full-field (1 mm diameter spot) blue light was delivered at low (10 µW/mm2) and high (25 µW/mm2) intensities from a blue laser source (see above). The mGluR2 agonist was not included in the bath in this set of experiments. Neither the basal firing rate of PCs nor frequency of sIPSC in GrCs was significantly different (p = 0.76 and 0.72, respectively) from the Halo experiments, likely a reflection of the low survival rate of Golgi cells in slices obtained from adult mice.

Data Acquisition and Analysis

All electrophysiology recordings were performed using a Molecular Device MultiClamp 700B amplifier, an InstruTECH ITC-18 (Heka Instrument Inc. Bellmore, NY) and custom software native to IgorPro 6 (courtesy of Matthew Xu-Friedman, SUNY Buffalo, Buffalo, NY). Signals were filtered at 4–8 kHz (4-Pole analog Bessel filter), digitized at 20kHz and digitally filtered again at 1.5 kHz (equiripple FIR filter) in Matlab for analysis and display. sIPSC detection was performed in Matlab via peak detection and manually inspected. Tonic current was extracted by removing the detected phasic IPSC by setting a 30 ms segment during the IPSC to the baseline just prior to the IPSC. In paired PC-grC experiment (Fig. 4), IPSC latencies in paired recordings were calculated from the time difference between the peak of the first derivative of the PC AP waveform and the 5% rise time of the IPSC waveform. Single exponential fits were performed on the normalized average PC firing frequency, phasic IPSC frequency and tonic current change under each of the four stimulation conditions to obtain a time constant for their dynamics during PC firing rate modulation. Data are reported as mean ± SEM. For experiments involving LY354740, the percentage of IPSC reduction is measured relative to the average of control conditions.

Supplementary Material

1
2
3
Download video file (34.6MB, mp4)
4
Download video file (40MB, mp4)
5
Download video file (26.2MB, mp4)

Acknowledgments

This work was supported by National Institutes of Health Grants NIH R01NS032405 and R01NS092707, and grants from the Nancy Lurie Marks, the Goldenson, the Lefler and the Khodadad Foundations to W.G.R., a Leonard and Isabelle Goldenson Postdoctoral Fellowship and a Mahoney Postdoctoral fellowship to L.W., an Alice and Joseph Brooks fellowship and a Ruth L. Kirschstein Award F32NS087708 to S.R. We thank the Neurobiology Imaging Facility (supported by NINDS P30 Core Center Grant NS072030) for consultation and instrument availability. We thank the Image and Data Analysis Core (IDAC) for image analysis. We thank S. Jackman, J. Turecek, C. Weyrer and C. Chen for comments on the manuscripts. We thank K. McDaniels for genotyping and B. Harrison and A. Stefano for perfusion and slice preparation.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author Contributions

C.G. and L.W. contributed equally to this work. C.G., L.W., S.R. and W.G.R. conceived the experiments. C.G., L.W. and K.E. conducted the anatomical investigations. C.G., L.W and S.R. conducted the electrophysiology experiments. L.W., H.E. and K.E. conducted the anatomical analysis. C.G. and S.R. conducted the electrophysiology analysis. C.G., L.W. and W.G.R wrote the manuscript with contributions from S.R. and H.E.

References

  1. Albus JS. A theory of cerebellar function. Mathematical Biosciences. 1971;10:25–61. [Google Scholar]
  2. Arenz A, Silver RA, Schaefer AT, Margrie TW. The contribution of single synapses to sensory representation in vivo. Science. 2008;321:977–980. doi: 10.1126/science.1158391. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Armstrong DM, Edgley SA. Discharges of Purkinje cells in the paravermal part of the cerebellar anterior lobe during locomotion in the cat. J Physiol. 1984;352:403–424. doi: 10.1113/jphysiol.1984.sp015300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Attwell D, Barbour B, Szatkowski M. Nonvesicular release of neurotransmitter. Neuron. 1993;11:401–407. doi: 10.1016/0896-6273(93)90145-h. [DOI] [PubMed] [Google Scholar]
  5. Barmack NH, Baughman RW, Errico P, Shojaku H. Vestibular primary afferent projection to the cerebellum of the rabbit. J Comp Neurol. 1993;327:521–534. doi: 10.1002/cne.903270405. [DOI] [PubMed] [Google Scholar]
  6. Barmack NH, Yakhnitsa V. Cerebellar climbing fibers modulate simple spikes in Purkinje cells. J Neurosci. 2003;23:7904–7916. doi: 10.1523/JNEUROSCI.23-21-07904.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Barmack NH, Yakhnitsa V. Functions of interneurons in mouse cerebellum. J Neurosci. 2008;28:1140–1152. doi: 10.1523/JNEUROSCI.3942-07.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bortone DS, Olsen SR, Scanziani M. Translaminar inhibitory cells recruited by layer 6 corticothalamic neurons suppress visual cortex. Neuron. 2014;82:474–485. doi: 10.1016/j.neuron.2014.02.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brickley SG, Cull-Candy SG, Farrant M. Development of a tonic form of synaptic inhibition in rat cerebellar granule cells resulting from persistent activation of GABAA receptors. J Physiol. 1996;497:753–759. doi: 10.1113/jphysiol.1996.sp021806. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cannon SC, Robinson DA, Shamma S. A proposed neural network for the integrator of the oculomotor system. Biol Cybern. 1983;49:127–136. doi: 10.1007/BF00320393. [DOI] [PubMed] [Google Scholar]
  11. Cerminara NL, Lang EJ, Sillitoe RV, Apps R. Redefining the cerebellar cortex as an assembly of non-uniform Purkinje cell microcircuits. Nat Rev Neurosci. 2015;16:79–93. doi: 10.1038/nrn3886. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chiu CS, Brickley S, Jensen K, Southwell A, McKinney S, Cull-Candy S, Mody I, Lester HA. GABA transporter deficiency causes tremor, ataxia, nervousness, and increased GABA-induced tonic conductance in cerebellum. J Neurosci. 2005;25:3234–3245. doi: 10.1523/JNEUROSCI.3364-04.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Crook JD, Hendrickson A, Erickson A, Possin D, Robinson FR. Purkinje cell axon collaterals terminate on Cat-301+ neurons in Macaca monkey cerebellum. Neuroscience. 2007;149:834–844. doi: 10.1016/j.neuroscience.2007.08.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Crowley JJ, Fioravante D, Regehr WG. Dynamics of fast and slow inhibition from cerebellar golgi cells allow flexible control of synaptic integration. Neuron. 2009;63:843–853. doi: 10.1016/j.neuron.2009.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Dieudonne S. Submillisecond kinetics and low efficacy of parallel fibre-Golgi cell synaptic currents in the rat cerebellum. J Physiol. 1998;510(Pt 3):845–866. doi: 10.1111/j.1469-7793.1998.845bj.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Dieudonne S, Dumoulin A. Serotonin-driven long-range inhibitory connections in the cerebellar cortex. J Neurosci. 2000;20:1837–1848. doi: 10.1523/JNEUROSCI.20-05-01837.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Dino MR, Willard FH, Mugnaini E. Distribution of unipolar brush cells and other calretinin immunoreactive components in the mammalian cerebellar cortex. J Neurocytol. 1999;28:99–123. doi: 10.1023/a:1007072105919. [DOI] [PubMed] [Google Scholar]
  18. Dizon MJ, Khodakhah K. The role of interneurons in shaping Purkinje cell responses in the cerebellar cortex. J Neurosci. 2011;31:10463–10473. doi: 10.1523/JNEUROSCI.1350-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Duguid I, Branco T, Chadderton P, Arlt C, Powell K, Hausser M. Control of cerebellar granule cell output by sensory-evoked Golgi cell inhibition. Proc Natl Acad Sci U S A. 2015;112:13099–13104. doi: 10.1073/pnas.1510249112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Duguid I, Branco T, London M, Chadderton P, Hausser M. Tonic inhibition enhances fidelity of sensory information transmission in the cerebellar cortex. J Neurosci. 2012;32:11132–11143. doi: 10.1523/JNEUROSCI.0460-12.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Eccles JC, Ito M, Szentágothai Jn. The cerebellum as a neuronal machine. Berlin, New York etc: Springer-Verlag; 1967. [Google Scholar]
  22. Farrant M, Brickley SG. Properties of GABA(A) receptor-mediated transmission at newly formed Golgi-granule cell synapses in the cerebellum. Neuropharmacology. 2003;44:181–189. doi: 10.1016/s0028-3908(02)00363-5. [DOI] [PubMed] [Google Scholar]
  23. Fatemi SH, Aldinger KA, Ashwood P, Bauman ML, Blaha CD, Blatt GJ, Chauhan A, Chauhan V, Dager SR, Dickson PE, et al. Consensus paper: pathological role of the cerebellum in autism. Cerebellum. 2012;11:777–807. doi: 10.1007/s12311-012-0355-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Freund TF, Buzsaki G. Interneurons of the hippocampus. Hippocampus. 1996;6:347–470. doi: 10.1002/(SICI)1098-1063(1996)6:4<347::AID-HIPO1>3.0.CO;2-I. [DOI] [PubMed] [Google Scholar]
  25. Gao Z, Proietti-Onori M, Lin Z, Ten Brinke MM, Boele HJ, Potters JW, Ruigrok TJ, Hoebeek FE, De Zeeuw CI. Excitatory Cerebellar Nucleocortical Circuit Provides Internal Amplification during Associative Conditioning. Neuron. 2016;89:645–657. doi: 10.1016/j.neuron.2016.01.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Geurts FJ, Timmermans J, Shigemoto R, De Schutter E. Morphological and neurochemical differentiation of large granular layer interneurons in the adult rat cerebellum. Neuroscience. 2001;104:499–512. doi: 10.1016/s0306-4522(01)00058-6. [DOI] [PubMed] [Google Scholar]
  27. Green AM, Angelaki DE. Resolution of sensory ambiguities for gaze stabilization requires a second neural integrator. J Neurosci. 2003;23:9265–9275. doi: 10.1523/JNEUROSCI.23-28-09265.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Hamann M, Rossi DJ, Attwell D. Tonic and spillover inhibition of granule cells control information flow through cerebellar cortex. Neuron. 2002;33:625–633. doi: 10.1016/s0896-6273(02)00593-7. [DOI] [PubMed] [Google Scholar]
  29. Heath NC, Rizwan AP, Engbers JD, Anderson D, Zamponi GW, Turner RW. The expression pattern of a Cav3-Kv4 complex differentially regulates spike output in cerebellar granule cells. J Neurosci. 2014;34:8800–8812. doi: 10.1523/JNEUROSCI.0981-14.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Hirono M, Saitow F, Kudo M, Suzuki H, Yanagawa Y, Yamada M, Nagao S, Konishi S, Obata K. Cerebellar globular cells receive monoaminergic excitation and monosynaptic inhibition from Purkinje cells. PLoS One. 2012;7:e29663. doi: 10.1371/journal.pone.0029663. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Hoddevik GH. The pontine projection to the flocculonodular lobe and the paraflocculus studied by means of retrograde axonal transport of horseradish peroxidase in the rabbit. Exp Brain Res. 1977;30:511–526. doi: 10.1007/BF00237641. [DOI] [PubMed] [Google Scholar]
  32. Hull C, Regehr WG. Identification of an inhibitory circuit that regulates cerebellar Golgi cell activity. Neuron. 2012;73:149–158. doi: 10.1016/j.neuron.2011.10.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Kaneda M, Farrant M, Cull-Candy SG. Whole-cell and single-channel currents activated by GABA and glycine in granule cells of the rat cerebellum. J Physiol. 1995;485(Pt 2):419–435. doi: 10.1113/jphysiol.1995.sp020739. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kase M, Miller DC, Noda H. Discharges of Purkinje cells and mossy fibres in the cerebellar vermis of the monkey during saccadic eye movements and fixation. J Physiol. 1980;300:539–555. doi: 10.1113/jphysiol.1980.sp013178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kato HK, Gillet SN, Peters AJ, Isaacson JS, Komiyama T. Parvalbumin-expressing interneurons linearly control olfactory bulb output. Neuron. 2013;80:1218–1231. doi: 10.1016/j.neuron.2013.08.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Kim JY, Ash RT, Ceballos-Diaz C, Levites Y, Golde TE, Smirnakis SM, Jankowsky JL. Viral transduction of the neonatal brain delivers controllable genetic mosaicism for visualising and manipulating neuronal circuits in vivo. Eur J Neurosci. 2013;37:1203–1220. doi: 10.1111/ejn.12126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Koziol LF, Budding D, Andreasen N, D'Arrigo S, Bulgheroni S, Imamizu H, Ito M, Manto M, Marvel C, Parker K, et al. Consensus paper: the cerebellum's role in movement and cognition. Cerebellum. 2014;13:151–177. doi: 10.1007/s12311-013-0511-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Landis DM, Weinstein LA, Halperin JJ. Development of synaptic junctions in cerebellar glomeruli. Brain Res. 1983;284:231–245. doi: 10.1016/0165-3806(83)90008-1. [DOI] [PubMed] [Google Scholar]
  39. Lee S, Yoon BE, Berglund K, Oh SJ, Park H, Shin HS, Augustine GJ, Lee CJ. Channel-mediated tonic GABA release from glia. Science. 2010;330:790–796. doi: 10.1126/science.1184334. [DOI] [PubMed] [Google Scholar]
  40. Lisberger SG, Fuchs AF. Role of primate flocculus during rapid behavioral modification of vestibuloocular reflex I. Purkinje cell activity during visually guided horizontal smooth-pursuit eye movements and passive head rotation. J Neurophysiol. 1978a;41:733–763. doi: 10.1152/jn.1978.41.3.733. [DOI] [PubMed] [Google Scholar]
  41. Lisberger SG, Fuchs AF. Role of primate flocculus during rapid behavioral modification of vestibuloocular reflex. II. Mossy fiber firing patterns during horizontal head rotation and eye movement. J Neurophysiol. 1978b;41:764–777. doi: 10.1152/jn.1978.41.3.764. [DOI] [PubMed] [Google Scholar]
  42. Longair MH, Baker DA, Armstrong JD. Simple Neurite Tracer: open source software for reconstruction, visualization and analysis of neuronal processes. Bioinformatics. 2011;27:2453–2454. doi: 10.1093/bioinformatics/btr390. [DOI] [PubMed] [Google Scholar]
  43. Marr D. A theory of cerebellar cortex. J Physiol. 1969;202:437–470. doi: 10.1113/jphysiol.1969.sp008820. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Medina JF, Mauk MD. Computer simulation of cerebellar information processing. Nat Neurosci. 2000;3:1205–1211. doi: 10.1038/81486. [DOI] [PubMed] [Google Scholar]
  45. Mitchell SJ, Silver RA. Shunting inhibition modulates neuronal gain during synaptic excitation. Neuron. 2003;38:433–445. doi: 10.1016/s0896-6273(03)00200-9. [DOI] [PubMed] [Google Scholar]
  46. Nagao S, Kitamura T, Nakamura N, Hiramatsu T, Yamada J. Differences of the primate flocculus and ventral paraflocculus in the mossy and climbing fiber input organization. J Comp Neurol. 1997;382:480–498. doi: 10.1002/(sici)1096-9861(19970616)382:4<480::aid-cne5>3.0.co;2-z. [DOI] [PubMed] [Google Scholar]
  47. Nusser Z, Sieghart W, Somogyi P. Segregation of different GABAA receptors to synaptic and extrasynaptic membranes of cerebellar granule cells. J Neurosci. 1998;18:1693–1703. doi: 10.1523/JNEUROSCI.18-05-01693.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Olsen SR, Bortone DS, Adesnik H, Scanziani M. Gain control by layer six in cortical circuits of vision. Nature. 2012;483:47–52. doi: 10.1038/nature10835. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Paallysaho J, Sugita S, Noda H. Brainstem mossy fiber projections to lobules VIa, VIb,c, VII and VIII of the cerebellar vermis in the rat. Neurosci Res. 1991;12:217–231. doi: 10.1016/0168-0102(91)90112-c. [DOI] [PubMed] [Google Scholar]
  50. Palay SL, Chan-Palay V. Cerebellar cortex: cytology and organization. Berlin, Heidelberg, New York: Springer; 1974. [Google Scholar]
  51. Preibisch S, Saalfeld S, Tomancak P. Globally optimal stitching of tiled 3D microscopic image acquisitions. Bioinformatics. 2009;25:1463–1465. doi: 10.1093/bioinformatics/btp184. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Provini L, Redman S, Strata P. Mossy and climbing fibre organization on the anterior lobe of the cerebellum activated by forelimb and hindlimb areas of the sensorimotor cortex. Exp Brain Res. 1968;6:216–233. doi: 10.1007/BF00235125. [DOI] [PubMed] [Google Scholar]
  53. Robinson DA. The effect of cerebellectomy on the cat's vestibulo-ocular integrator. Brain Res. 1974;71:195–207. doi: 10.1016/0006-8993(74)90961-5. [DOI] [PubMed] [Google Scholar]
  54. Rossi DJ, Hamann M. Spillover-mediated transmission at inhibitory synapses promoted by high affinity alpha6 subunit GABA(A) receptors and glomerular geometry. Neuron. 1998;20:783–795. doi: 10.1016/s0896-6273(00)81016-8. [DOI] [PubMed] [Google Scholar]
  55. Rossi DJ, Hamann M, Attwell D. Multiple modes of GABAergic inhibition of rat cerebellar granule cells. J Physiol. 2003;548:97–110. doi: 10.1113/jphysiol.2002.036459. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Rothman JS, Cathala L, Steuber V, Silver RA. Synaptic depression enables neuronal gain control. Nature. 2009;457:1015–1018. doi: 10.1038/nature07604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Schmahmann JD. Disorders of the cerebellum: ataxia, dysmetria of thought, and the cerebellar cognitive affective syndrome. J Neuropsychiatry Clin Neurosci. 2004;16:367–378. doi: 10.1176/jnp.16.3.367. [DOI] [PubMed] [Google Scholar]
  58. Seung HS, Lee DD, Reis BY, Tank DW. Stability of the memory of eye position in a recurrent network of conductance-based model neurons. Neuron. 2000;26:259–271. doi: 10.1016/s0896-6273(00)81155-1. [DOI] [PubMed] [Google Scholar]
  59. Sik A, Ylinen A, Penttonen M, Buzsaki G. Inhibitory CA1-CA3-hilar region feedback in the hippocampus. Science. 1994;265:1722–1724. doi: 10.1126/science.8085161. [DOI] [PubMed] [Google Scholar]
  60. Tyrrell T, Willshaw D. Cerebellar cortex: its simulation and the relevance of Marr's theory. Philos Trans R Soc Lond B Biol Sci. 1992;336:239–257. doi: 10.1098/rstb.1992.0059. [DOI] [PubMed] [Google Scholar]
  61. Voogd J, Barmack NH. Oculomotor cerebellum. Prog Brain Res. 2006;151:231–268. doi: 10.1016/S0079-6123(05)51008-2. [DOI] [PubMed] [Google Scholar]
  62. Wall MJ, Usowicz MM. Development of action potential-dependent and independent spontaneous GABAA receptor-mediated currents in granule cells of postnatal rat cerebellum. Eur J Neurosci. 1997;9:533–548. doi: 10.1111/j.1460-9568.1997.tb01630.x. [DOI] [PubMed] [Google Scholar]
  63. Wang SS, Kloth AD, Badura A. The cerebellum, sensitive periods, and autism. Neuron. 2014;83:518–532. doi: 10.1016/j.neuron.2014.07.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Watanabe D, Nakanishi S. mGluR2 postsynaptically senses granule cell inputs at Golgi cell synapses. Neuron. 2003;39:821–829. doi: 10.1016/s0896-6273(03)00530-0. [DOI] [PubMed] [Google Scholar]
  65. Watt AJ, Cuntz H, Mori M, Nusser Z, Sjostrom PJ, Hausser M. Traveling waves in developing cerebellar cortex mediated by asymmetrical Purkinje cell connectivity. Nat Neurosci. 2009;12:463–473. doi: 10.1038/nn.2285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Witter L, De Zeeuw CI. Regional functionality of the cerebellum. Curr Opin Neurobiol. 2015;33:150–155. doi: 10.1016/j.conb.2015.03.017. [DOI] [PubMed] [Google Scholar]
  67. Witter L, Rudolph S, Pressler RT, Lahlaf SI, Regehr WG. Purkinje Cell Collaterals Enable Output Signals from the Cerebellar Cortex to Feed Back to Purkinje Cells and Interneurons. Neuron. 2016 doi: 10.1016/j.neuron.2016.05.037. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS811045-supplement-1.docx (127.4KB, docx)
2
NIHMS811045-supplement-2.pdf (703.1KB, pdf)
3
Download video file (34.6MB, mp4)
4
Download video file (40MB, mp4)
5
Download video file (26.2MB, mp4)

RESOURCES