Imaging calcium events in living IO from the ventral side. (A1-3) Transfection of IO and CN neurons using AAV9-based viral vectors results in GCaMP6s expression in IO neurons and ChrimsonR in nucleo-olivary axons. (A1) Schematic of viral transfection targeting CN and IO. (A2) Confocal images of transfected CN (top) and IO (bottom) regions (10x maximum projection). (A3) Higher-resolution confocal image (20x) highlights transfected axons in the medial accessory olive (MAO), principal olive (PO), and dorsal accessory olive (DAO) with mediolateral imaging areas shaded (pink and cyan). (B1-3) Surgical access to the ventral medulla enables imaging of the IO. (B1) Ventral view during surgery. (B2) Schematic shows the ventral IO location, with key features like the blood vessel dividing PO and DAO marked. (B3) Schematic CCTX and CN regions targeted by PO and DAO climbing fibers. (C1-2) Example calcium recording from the IO. (C1) Standard deviation (STD) projection of a recording showing somata in PO (pink) and DAO (cyan), with manually marked regions of interest (ROIs). The red dashed line indicates the dividing blood vessel. (C2) Baseline-normalized (DF/F) intensity traces of fluorescence fluctuations in selected somata from PO (P1-P4) and DAO (D1-D7). Abbreviations: CN, cerebellar nuclei; IO, inferior olive; MAO, medial accessory olive; PT, pyramidal tract; PO, principal olive; DAO, dorsal accessory olive; CCTX, cerebellar cortex; med, intP, intA, lat: medial, posterior interpositus, anterior interpositus, lateral CN. Scale bars: 500 μ m (A2); 200 μ m (A3); 1 mm (B1-2); 20 μ m (C1); 20 s (C2).

Spontaneous event properties in IO subnuclei display synchronicity-related modulation. A, example fluorescence imaging traces from a PO (pink) and DAO (cyan) neurons. Top traces display unprocessed DF/F values; bottom traces display the same data after background subtraction. Black dots denote events detected from the background-subtracted trace. See Methods for description of the processing. Regions of the background-subtracted traces indicated with dashed rectangles are shown with expanded scale insets to the right. B, summary of basic event shape characteristics. B1, average event shapes from PO and DAO neurons; shaded region denotes std. Events are vertically aligned at rise onset. B2, cumulative histogram of event amplitude distributions. Amplitudes are shown normalized to the baseline “noise” (see Methods for details). B3, empirical cumulative distribution of event widths (measured at half-amplitude). B4, empirical cumulative distribution plot of event rise times. C, spike raster visualization of the events detected in the recording shown in Figure 1. Each spike is color-coded by its amplitude, normalized to the largest event in a given cell. White arrowheads point to some instances when several large-amplitude events occur synchronously. Scale bar, 25 s. D, cumulative histogram of mean event frequency (D1) and the inter-event interval (D2). E, clustered event amplitude distributions show shift to larger values compared to asynchronous events both in PO and DAO recordings. E1, empirical cumulative distribution of clustered (bright colors) and single (dark colors) events in PO and DAO. Clustered spikes are defined with at least 2 neurons spiking within one frame (50 ms) of each other. For results with other definitions, see Figure 2 - Supplement 1. E2, cluster sizes (shown with empirical cumulative distribution plots relative to number of cells visible in a recording) are larger in PO than in DAO. E3, comparison of clustering between PO and DAO. Left, fraction of events recorded that were classified as clustered; right, normalized by expected value. Each dot represents a single field-of-view recording. Abbreviations: DF/F, fluorescence normalized by baseline intensity; PO, principal olive; DAO, dorsal accessory olive; N refers to number of cells, n to number of events. *, p < 0.05; **, p < 0.01.

Comparison of event amplitudes and widths (left) and rise times (right), all shown normalized to the largest values within each cell. Amplitudes do not grow linearly throughout the range of event width, possibly due to GCamP6s saturation with largest events. However, with the low sampling frequency, amplitude is more reliable metric than duration.

Parameter sweeps for clustering effect on event amplitude for PO and DAO. A, effect sizes on event amplitudes with clustering window (epsilon in DBSCAN algorithm) increasing from 0 300 ms (0-7 frames), and minimum number of events needed to form a cluster increasing from 2 to 7. Color brightness depicts effect size of the change in event amplitude distribution; effect sizes also indicated in squares. Gray squares indicate no available data. Highests effects are seen with short windows and moderate numbers of events in a cluster, as less data is available for large clusters. B, event amplitude distributions for 3-spike clusters are modulated by varying acceptable window lengths for clustering in PO (left) and DAO (right). C, event amplitude distributions for 100-ms windows are affected by varying cluster sizes in PO (left) and DAO (right).

Normalized amplitude of calcium events. GLMM analysis

FWHM of calcium events (s). GLMM analysis

RiseDuration of calcium events (s). GLMM analysis

Airpuff-evoked events in the principal olive. Periocular airpuff stimulation (A1) reliably evoked events in PO neurons. (A2) Top: example traces from a single cell over 10 stimulations (dark trace: average). Bottom: heat-map of the full trial, with traces sorted by event amplitude. (A3) Peri-stimulus time histogram (mean ± SEM, 1-s bins) for all trials. B: AP-evoked event waveforms (B1) and amplitudes (B2) compared to the spontaneous events in the same cells and recordings. Waveforms are shown as mean ± SEM, vertically aligned at event onset. C, event clustering as defined for events shown in Figure 2. C1,(C) Event clustering. (C1) Clustering increased for AP-evoked events relative to spontaneous ones (left) but normalized to event rates (right). Note that for visual clarity markers are not shown on the right; see source data. C2, clustered and single AP-event amplitudes compared with spontaneous events from the same cells and trials. The difference between clustered and single AP-events does not reach significance with GLMM analysis due to low number of events per cell. D, AP event timing. D1, timing histogram of AP-evoked events with respect to stimulus. D2, timing histogram of AP-evoked events (blue) and preceding spontaneous events with respect to stimulus. Cyan: spontaneous events within trials with AP-evoked events, showing a refractory period > 1 s (dashed rectangle). (E) Intervals before and after AP-events. E1, comparison of intervals preceding (cyan) and following (gray) 42 AP-evoked events with a preceding and following spontaneous event within 10 s. E2, post/pre-interval ratios for clustered and single AP-events, with each marker representing a single stimulation. Abbreviations: IO, inferior olive; AP, airpuff; PO, principal olive; DAO, dorsal accessory olive; spont, spontaneous events; GLMM, generalized linear mixed model. **, p < 0.01; ***, p < 0.001; ****, p < 0.0001

Frequency of spontaneous calcium events (Hz). GLMM analysis

Interval of spontaneous calcium events (s). GLMM analysis

Cluster fold PO vs. DAO. Mann-Whitney U test

Peristimulus (AP) event frequency in PO neurons (Hz). BootStrap analysis

Cluster fraction spon vs AP. Mann-Whitney U test

Cluster fold spon vs AP. Mann-Whitney U test

Pre and post intervals of AP events (s). Wilcoxon paired rank sum

Pre and post ratios of clustered and single AP events. Mann-Whitney U test

Peristimulus (N-O) event frequency in PO neurons (Hz). BootStrap analysis

Peristimulus (N-O) event frequency in DAO neurons (Hz). BootStrap analysis

Changes in IO activity evoked by stimulation of N-O axons. (A) Experimental arrangement schematic (A1) and spike classification based on timing (A2). The first second of the stimulation period (gray box) is excluded from analysis; the yellow box denotes the rebound activity window. (B) Examples of NO stimulation effects on IO spiking in single cells (B1) and across all cells and stimulations in a single recording (B2), sorted by the time of the first spike after stimulus ends. In B2, color indicates event amplitude normalized to each cell’s maximum; shading corresponds to A2. Shading as in A2. (C) Summary of N-O stimulation effects on spike frequency in PO (C1, top) and DAO (C1, bottom) cells; box plots show median±SEM values. (C2) Mean±SEM of GCaMP6s fluorescence traces from all stimulation trials. (D) Example traces from PO (top) and DAO (bottom) cells with events during NO stimulation (shaded area) (D1). (D2) Cumulative distributions of event amplitudes during N-O stimulation (red) versus outside stimulation (black) in the same cells and trials; PO and DAO data are pooled due to low event numbers. (D3) Comparison of clustering extent, shown as the ratio of observed to expected clustered event fractions, for events during N-O stimulation (red) and outside stimulation (black). (D4) Cumulative distributions of cluster sizes during and outside N-O stimulation, relative to the total available cell pool. Abbreviations: CN, cerebellar nuclei; IO, inferior olive; N-O, nucleo-olivary; stims, stimulations; BSL, baseline (periods outside stimulation). Statistical significance indicted with asterisks as follows : *, p < 0.05.

A, example traces from a single cell (4 stimulations) where optogenetic activation of N-O axons robustly leads to a spike at the onset of the stimulation. Gray box indicates the time window during which spikes were categorized as N-O-evoked. B, Cumulative histograms of NO-evoked spike amplitudes vs. the spontaneous events in the same trials in PO and DAO. C, timing of N-O-evoked events with respect to onset of optogenetic stimulation. D event frequency histogram time-aligned at the onset of N-O stimulation from trials where a N-O evoked spike was seen (white bars).

A, examples of DAO cell stimulation trials where “rebound events” (RB) were seen in the post-stimulation window (indicated in yellow color). Circles denote detected events. B, histogram of the frequency of the event during N-O stimulation and 5 seconds following it in trials with RB events (yellow bars). The inset shows comparison of the amplitudes of RB events to spontaneous events. C, mean calcium traces from N-O stimulation trials in all DAO cells, grouped by whether an event occurred within the rebound window (yellow) or not (blue), Thich trace depicts average trace, shading±SSEM. No difference in the amplitude of calcium suppression during N-O stimulation is seen between these groups. D, timing of events preceding (blue) and following (brown) the RB event. Due to low number of events, data from trials where N-O stimulation led to an early spike are included (light-colored bars); data from trials after N-O driven spikes are excluded are shown in dark colors. E, F: RB events do not show difference with respect to event clustering probability (E) or cluster sizes (F). E3, comparison of synchronous and asynchronous amplitudes for post-stimulation events. E4, distribution of delays of post-stim spikes with respect to the offset of optogenetic stimulus (“OGoff”), compared to the distribution of inter-event intervals among spontaneous events.

Normalized amplitude of events during N-O stimulation versus outside stimulation. GLMM analysis

Cluster fold of spon and spon during N-O stimulation. Mann-Whitney U test

Cluster fraction of spon and spon during N-O stimulation. Mann-Whitney U test

Nucleo-olivary stimulation does not suppress airpuff-evoked spikes. (A) Periocular stimulation evokes IO spikes during N-O stimulation. (A1) Schematic of the experiment. (A2) Example calcium traces from a single cell. N-O stimulation suppresses spontaneous spiking (top), but not AP-evoked spikes (bottom). (A3) Raster plot of the trial (6 cells, 4 stimulations); sorted by event fluorescence intensity. (A4) Stimulus-triggered histogram of event frequencies. (B) N-O stimulation does not decrease AP-evoked event size. (B1) Mean event waveforms (±SEM) for AP-evoked events with (purple) and without (blue) N-O stimulation. (B2) Cumulative distributions and violin plots of event amplitudes with and without N-O stimulation. Inset shows violin plots of the same data. (C) Effect of N-O stimulation on AP-event clustering. (C1) AP-event clustering fraction with and without N-O stimulation; markers represent single recordings. (C2) Clustered and single AP-event amplitudes. Data from AP-events without N-O stimulation shown as a reference. (D) Timing of AP-evoked events. (D1) Purple bars denote events occurring within the 1-second window after stimulation; black bars are later spontaneous events. Circles represent a single spike; vertical position indicates event amplitude, and size indicates cluster size. Inset shows event timing compared to AP-events without N-O stimulation. (D2) Event times preceding and following AP-events (dashed line) without (blue) and with (purple) N-O stimulation. (E) N-O stimulation does not decrease the probability of evoking an IO spike with AP stimulation. (E1) Probability distribution for an AP-event in cells that responded at least once. Inset shows probability of evoking at least one spike in a field of view with a single AP is not affected by N-O stimulation; data from recordings with > 3 AP-responding cells. (E2) Summary of spike frequency with N-O alone (pink), AP alone (cyan), and AP combined with N-O (purple), shown in absolute frequencies (left) and normalised to baseline (right) (E3) Relative increase in event frequency with AP stimulation without and with N-O stimulation, shown in absolute frequency ratio (left) and baseline normalised (right). Abbreviations: CN, cerebellar nuclei; IO, inferior olive; AP, airpuff stimulation; PO, principal olive; DAO, dorsal accessory olive; N-O+AP, airpuff delivery during nucleo-olivary stimulation; Clust, clustered spike; Single, single spike; N-O early, early bin of N-O stimulation block. *, p < 0.05; **, p < 0.01; ***, p < 0.001

Delay of AP and N-O+AP events in PO neurons (s). GLMM analysis

Event probability in a PO cell. AP vs. N-O+AP. Mann-Whitney U test

Event probability in PO subnuclei. AP vs. N-O+AP. Mann-Whitney U test

Peristimulus event frequency (s). Bootstrap analysis

Peristimulus event frequency [Normalized to baseline]. Bootstrap analysis

Peristimulus event frequency fold-change. Bootstrap analysis

Peristimulus event frequency fold-change [Normalized to baseline]. Bootstrap analysis