Dendritic excitation-inhibition balance shapes cerebellar output during motor behaviour

Abstract

Feedforward excitatory and inhibitory circuits regulate cerebellar output, but how these circuits interact to shape the somatodendritic excitability of Purkinje cells during motor behaviour remains unresolved. Here we perform dendritic and somatic patch-clamp recordings in vivo combined with optogenetic silencing of interneurons to investigate how dendritic excitation and inhibition generates bidirectional (that is, increased or decreased) Purkinje cell output during self-paced locomotion. We find that granule cells generate a sustained depolarization of Purkinje cell dendrites during movement, which is counterbalanced by variable levels of feedforward inhibition from local interneurons. Subtle differences in the dendritic excitation-inhibition balance generate robust, bidirectional changes in simple spike (SSp) output. Disrupting this balance by selectively silencing molecular layer interneurons results in unidirectional firing rate changes, increased SSp regularity and disrupted locomotor behaviour. Our findings provide a mechanistic understanding of how feedforward excitatory and inhibitory circuits shape Purkinje cell output during motor behaviour.

Figure 1. Bidirectional Purkinje cell SSp modulation during locomotion.

(a) Schematic showing feedforward circuitry of the cerebellum. DCN, deep cerebellar nuclei; GCs, granule cells; PC, Purkinje cell; PFs, parallel fibres; MFs, mossy fibres. (b) Feedforward input models underpinning locomotion-dependent bidirectional PC simple spike (SSp) modulation. exc, excitation; freq., frequency; inh, inhibition. (c) Recording configuration in awake mice where locomotion was captured using digital video. (d) Intracellular biocytin labelling of a PC via the recording electrode. Scale bar, 30 μm. GCL, granule cell layer; PCL, Purkinje cell layer; ML, molecular layer. (e) Somatic voltage recordings, spike rate histograms and associated motion index values (MI; dark grey) from two PCs during quiet wakefulness and voluntary locomotion (blue). Dashed red line denotes average firing rate during quiet wakefulness. (f) Purkinje cell quiet wakefulness SSp firing rate (Qw) as a function of locomotion-related SSp firing rate (Loc, n=38 cells, N=33 mice). Symbols represent individual PCs and dotted line represents unity. (g) Average change in PC SSp firing rate during quiet wakefulness (Qw) and locomotion (Loc; n=38 cells, N=33 mice). (h) Average change in PC SSp firing rate as a function of increasing movement. Grey lines represent exponential fits to the data in individual cells, solid purple lines represent PCs that increase/decrease their firing rates relative to the magnitude of movement and dashed purple lines represent PCs that are highly sensitive only to small movements (n=38 cells, N=33 mice). a.u., arbitrary units. (i) Distribution of Tau_MI values taken from the exponential fits shown in h. (j) Representative example of a climbing fibre-mediated complex spike (CS). (k) Purkinje cell quiet wakefulness (Qw) CS firing rate as a function of locomotion-related CS firing rate (n=38 cells, N=33 mice). Symbols represent individual PCs and dotted line represents unity. (l) Average change in PC CS firing rate during quiet wakefulness (Qw) and locomotion (Loc; n=38 cells, N=33 mice). (m) Distribution of CS versus SSp firing rate correlation coefficients during quiet wakefulness (Qw) and locomotion (Loc). Black circles denote PCs that displayed reciprocal firing rates during quiet wakefulness or locomotion (n=18/38 cells, N=33 mice).

Figure 2. Purkinje cell dV_m changes during locomotion.

(a) Schematic showing PC recording configurations (GC, granule cell; MLI, molecular layer interneuron; PC, Purkinje cell; PF, parallel fibre) and representative somatic (lower trace), proximal dendritic (middle trace, ∼60 μm from soma) and distal dendritic (upper trace, ∼100 μm from soma) voltage recordings from three PCs in vivo. Asterisks denote complex spikes. Right—dendritic calcium spike waveforms (50 consecutive traces overlaid (grey) and average V_m (black)). (b) Representative dendritic voltage recordings from two different PCs that depolarize (upper trace) or hyperpolarize (lower trace) during self-paced locomotion (blue shading). Motion index (MI, grey) defines the magnitude and duration of each locomotion bout. Note dendritic calcium spikes have been truncated to improve visualization of dendritic V_m (dV_m) changes. Red lines depict smoothed average V_m. (c) Average PC dV_m changes (ΔdV_m) during self-paced locomotion (n=19 cells, N=17 mice). Black filled circles depict dV_m changes in distal dendritic recordings (n=6 cells) and grey symbols depict dV_m changes in proximal dendritic recordings (n=13 cells). (d) Average change in dV_m (ΔdV_m) as a function of increasing movement (MI). Grey lines represent exponential fits to the data in individual cells, solid purple lines represent examples of dV_m changes that are relative to the magnitude of movement and dashed purple lines represent cells that are highly sensitive to small movements associated with movement preparation or initiation (n=19 cells, N=17 mice). a.u., arbitrary units. (e) Distribution of Tau_MI values taken from the exponential fits shown in d. (f) Relationship between ΔdV_m and percentage change in dSSp firing rate during self-paced, locomotion (n=13 cells, N=11 mice). Filled symbols represent the data from individual PCs and the solid line is a linear fit to the data r=0.71, P<0.01. (g) Average dendritic calcium spike frequency during quiet wakefulness and locomotion (n=19 cells, N=17 mice). Filled circles represent individual cells and dotted line represents unity.

Figure 3. Recruitment of feedforward excitatory and inhibitory circuits during locomotion.

(a) Schematic showing GC current (I)-clamp recording configuration. GC, granule cell; MLI, molecular layer interneuron; PC, Purkinje cell; PF, parallel fibre. (b) Representative voltage trace recorded from a GC during quiet wakefulness and locomotion (blue shading). (c) Average change in firing rate during quiet wakefulness (Qw) and locomotion (Loc). Grey symbols represent the data from individual GCs and black symbols represent mean±s.e.m., **P<0.001 Wilcoxon matched-pairs signed-rank test (n=13 cells, N=13 mice). (d) Raster plot (upper panels) and average firing rate histogram (lower panel, bin size=200 ms) of a population of GCs during quiet wakefulness (Qw) and locomotion (Loc; blue shading; n=13 cells, N=13 mice). (e) Schematic showing MLI voltage (V)-clamp recording configuration. (f) Representative current trace recorded at −70 mV (upper panel) from a MLI during quiet wakefulness and locomotion (blue shading). (g) Average EPSC charge transfer recorded during quiet wakefulness (Qw) and locomotion (Loc). Grey symbols represent data from individual MLIs, black symbols represent mean±s.e.m., **P<0.01 two-tailed t-test (n=7 cells, N=6 mice). (h) Average charge transfer as a function of increasing movement (motion index). Thin grey lines represent exponential fits to the data in individual cells, black line represents the average and pink shading the s.d. of the mean (n=7 cells, N=6 mice). (i) Schematic showing MLI I-clamp recording configuration. a.u., arbitrary units. (j) Representative voltage trace recorded from a MLI during quiet wakefulness and locomotion (blue shading). (k) Average firing rate of MLIs during quiet wakefulness (Qw) and locomotion (Loc; n=13 cells, N=13 mice). Grey symbols and connecting lines represent the data from individual MLIs, and black symbols represent mean±s.e.m. **P<0.01 two-tailed t-tests. (l) Average change in MLI firing rate as a function of increasing movement (motion index). Thin grey lines represent exponential fits to the data in individual cells, black line represents the average across all cells and pink shading the s.d. of the mean (n=13 cells, N=13 mice). a.u., arbitrary units.

Figure 4. Arch 3.0-mediated silencing of molecular layer interneurons.

(a) Low- and higher-magnification confocal images of parvalbumin (PV, red) and eArch3.0-eYFP labelling (green) along the apex of lobule V of the cerebellar vermis. Parasagittal sections of lobule V (upper panel, scale bar, 100 μm) were cut 9 days post virus injection. Two injections (white electrodes, middle panel) were performed at anterior and posterior locations in the craniotomy to ensure maximal viral infection across lobule V. Lower panels—higher-magnification images of lobule V highlighting the cell-selective expression of Arch 3.0 in MLIs (middle panels, scale bar, 50 μm; bottom panels, scale bar, 20 μm). (b) Example of light-evoked silencing of a MLI (green bar, 2 s pulse of 532 nm light) during locomotion (blue shading). Note spikes have been truncated to improve visualization of the photo-induced hyperpolarization. (c) Average firing rate of MLIs during locomotion (Loc), locomotion plus light activation (Loc+Arch) and after cessation of the light stimulus (Loc). Grey and green symbols and connecting lines represent the data from individual MLIs, and black symbols represent mean±s.e.m.**P<0.01, ns, not significant, two-tailed t-tests (n=6 cells, N=6 mice). (d) Normalized MLI firing frequency histogram (bin size=100 ms) aligned to the onset and offset of 532 nm light stimulation (green shading) during quiet wakefulness (grey left hand panels, n=9 cells, N=8 mice) and locomotion (blue right hand panels, n=6 cells, N=6 mice). Solid red line depicts the mean frequency before light stimulation and dashed lines indicate 2 × s.d. of the mean. (e) Mean percentage suppression of MLI firing frequency after Arch 3.0 stimulation during quiet wakefulness (Qw+Arch, n=9, N=8 mice) and locomotion (Loc+Arch, n=6, N=6 mice). Open circles represent the data from individual MLIs and bars represent mean±s.e.m., *P<0.05, two-tailed t-test. freq., frequency; Norm., normalization; supp., suppression.

Figure 5. Excitation–inhibition balance regulates Purkinje cell dV_m.

(a) Schematic showing MLI current (I)-clamp recording configuration during light activation of Arch 3.0 (green, 532 nm). (b) dV_m recordings from two PCs during quiet wakefulness, locomotion (blue) and locomotion during light stimulation (green). (c) Normalized dV_m distributions (from b) during quiet wakefulness (black), locomotion (blue) and locomotion plus light stimulation (green). (d) Normalized PC dV_m (bin size=100 ms) aligned to the onset and offset of light stimulation (green) during quiet wakefulness (grey, n=6 cells, N=6 mice) and locomotion (blue, n=7 cells, N=6 mice). Red line depicts mean dV_m during quiet wakefulness±2 × s.d. (e) Average ΔdV_m after light stimulation during quiet wakefulness (Qw+Arch, n=6, N=6 mice) and locomotion (Loc+Arch, n=7, N=6 mice). Circles represent the data from individual MLIs and bars represent mean±s.e.m., *P<0.05, two-tailed t-test. (f) Average locomotion-related changes in dV_m (ΔdV_m) before (grey, Loc), during (green, Loc+Arch) and after (grey, Loc) light stimulation. Connecting lines represent the data from individual PCs *P<0.05, ns, not significant, two-tailed t-tests (n=7 cells, N=6 mice). (g) Relationship between ΔdV_m and ΔdSSp frequency during locomotion in the presence (grey) and absence (green, n=4, N=3 mice) of feedforward inhibition. Grey connecting lines represent the data from individual PCs and thick line is a linear fit to the data (r=0.95, P<0.01). (h) Estimated dendritic excitation/inhibition balance. Inhibitory effect on dV_m was calculated using ΔdV_m Loc−(ΔdV_m Loc+Arch)+(ΔdV_m Qw+Arch) (n=7, N=6 mice), effects of excitation alone were calculated using (ΔdV_m Loc+Arch)−(ΔdV_m Qw+Arch). Dotted line represents unity. (i) Estimated ratio of effects of excitation ((ΔdV_m Loc+Arch)−(ΔdV_m Qw+Arch)) versus inhibition (ΔdV_m Loc−(ΔdV_m Loc+Arch)+(ΔdV_m Qw+Arch)) on ΔdV_m during locomotion. Ratio was calculated using absolute values of ΔdV_m change (n=7, N=6 mice). Norm., normalization.

Figure 6. Excitation–inhibition balance shapes Purkinje cell SSp output during locomotion.

(a) Schematic showing PC somatic recording configuration during light activation. (b) Cell-attached (upper) and whole-cell (lower) recordings from two PCs during quiet wakefulness, locomotion (blue) and locomotion plus light stimulation (green). Asterisks denote complex spikes. (c) Normalized PC SSp frequency histogram (bin size=100 ms) aligned to the onset and offset of 532 nm light stimulation (green) during quiet wakefulness (grey, n=16 cells, N=14 mice) and locomotion (blue, n=16 cells, N=14 mice). Solid red line depicts mean frequency during quiet wakefulness±2 × s.d. (d) Average change in PC SSp frequency after Arch 3.0 stimulation during quiet wakefulness (Qw+Arch) and locomotion (Loc+Arch). Circles represent the data from individual PCs and bars represent mean±s.e.m., **P<0.01, two-tailed t-test, (n=16 cells, N=14 mice). (e) Average ΔSSp frequency before (grey, Loc), during (green, Loc+Arch) and after (grey, Loc) light stimulation. Connecting lines represent the data from individual PCs, **P<0.01, ns, not significant, two-tailed t-tests (n=16 cells, N=14 mice). (f) Estimated effects of excitation and inhibition on PC SSp output. Inhibitory effect on SSp firing rate was calculated using ΔSSp Loc−(ΔSSp Loc+Arch)+(ΔSSp Qw+Arch) (n=16, N=14 mice), while the effects of excitation alone was calculated using (ΔSSp Loc+Arch)−(ΔSSp Qw+Arch). Red and blue circles represent PCs with high (>65 Hz) and low (<65 Hz) quiet wakefulness firing rates, respectively. Dotted line represents unity. (g) Ratio of effects of excitation ((ΔSSp Loc+Arch)−(ΔSSp Qw+Arch)) versus inhibition (ΔSSp Loc−(ΔSSp Loc+Arch)+(ΔSSp Qw+Arch)) on ΔSSp firing rate during locomotion. Ratio was calculated using absolute values of ΔSSp change and ΔdV_m ratios were taken from Fig. 5 for comparison. (h) Average change in the CV of PC SSp inter-event intervals (CV_{ISIs_SSp}) before (grey, Loc), during (green, Loc+Arch) and after (grey, Loc) light stimulation. Connecting lines represent data from individual PCs **P<0.01, ns, not significant, two-tailed t-tests (n=16 cells, N=14 mice). (i,j) Distribution of PC instantaneous firing regularity (LvR, i) and LvR s.d. (j) during quiet wakefulness (black, Qw, n=54 cells, N=47 mice) and during locomotion before (blue, Loc, n=54 cells, N=47 mice), and after light-evoked silencing of MLIs (green, Loc+Arch, n=16 cells, N=14 mice). *P<0.05, **P<0.01, two-tailed t-tests. (k) Normalized motion index (MI) aligned to the onset (left) and offset (right) of light stimulation (green). Grey traces represent the data from individual cells and black line represents smoothed average (n=11 and 14, respectively, N=21 mice). (l) Relative distributions of mice displaying no change, slowing or a complete halt in locomotion after onset (Light On) and offset (Light Off) of light stimulation (n=29, N=29 mice).