Tracking ripple events.

(A) Example LFP averages aligned to ripple peaks during slow-wave sleep (left), and the mean ripple amplitude for all channels on a single shank (right). The ripple amplitude was obtained from the 100 Hz to 250 Hz bandpass-filtered signal. For each shank, the electrode with the maximum amplitude (red dot) was selected for subsequent analyses. (B) Representative LFP traces during a sharp-wave ripple event. The raw signal from an electrode in the pyramidal layer is shown at the top, with the ripple-filtered signal in the middle. The bottom trace displays a raw LFP signal from the stratum radiatum layer, where the sharp wave manifests as a negative deflection associated with the ripple. (C) The spectrogram of a ripple event was obtained by wavelet transform (left). The center frequency (red dot) was defined as the frequency of maximum amplitude after averaging across the time window (right trace). (D) Top panel: Histology showing shanks placement scheme across dorsal CA1. Adapted from Paxinos and Watson (2006). Middle and bottom panels: Representative ripple-filtered traces from several consecutive shanks. In this example, ripple events detected in shank 1 (marked by green dashed lines) were used as reference time points.

Ripple features do not differ between the left and right hippocampal hemispheres.

(A) Ripple abundance. Left panel: Raincloud plot showing ripple events per second (abundance) in the left (blue) and right (red) hippocampus (side, Nleft = 50, Nright = 40, p – value = 0.15, LMMR). Right panel: Mean effect size (Hedges’ g) for ripple abundance difference (right – left, p – value = 0.7, permutation t-test). (B) Ripple abundance across increasing detection thresholds. Boxplots and half-violin plots illustrate ripple abundance in the left (blue) and right (red) hippocampus at multiple thresholds (LMMR, Nleft = 50, Nright = 40, repeated measures across thresholds; side: p – value = 0.155; threshold: p-value = 8.33 10−177; side vs. threshold: p-value = 0.27). (C) Ripple frequency. Left panel: Raincloud plots of ripple frequency (Hz) based on maximum amplitude in the left (blue) and right (red) hippocampus (LMMR, side, Nleft = 50, Nright = 40, p – value = 0.42). Right panel: Mean effect size (Hedges’ g) for inter-ripple interval difference (permutation t-test, right − left, p – value = 0.73). (D) Distribution of the pooled ripple frequency in the left (blue) and right (red) hippocampus.(E) Inter-ripple interval. Left panel: Raincloud plots showing the time intervals (in seconds) between consecutive ripple events in the left (blue) and right (red) hippocampus (LMMR, side, Nleft = 50, Nright = 40, p – value = 0.13). Right panel: Mean effect size (Hedges’ g) for the inter-ripple interval difference (right – left, p – value = 0.6, permutation t-test). (F) Distribution of the pooled inter-ripple intervals in the left (blue) and right (red) hippocampus. (G) Number of ripple cycles. Left panel: Raincloud plots displaying the average number of cycles per ripple event in the left (blue) and right (red) hippocampus (LMMR, side, Nleft = 50, Nright = 40, p – value = 0.31). Right panel: Mean effect size (Hedges’ g) for ripple cycles difference (permutation t-test, right – left, p – value = 0.28). (H) Distribution of the pooled mean ripple cycle count in the left (blue) and right (red) hippocampus. In the left panels of A, C, D and E, the dots represent the mean value per shank pooled across all animals. In the right panels, the filled black circle indicates the mean difference, the purple half-violin plot displays the distribution of 5000 bootstrapped mean differences, and the vertical line around the mean shows the bootstrap 95% confidence interval.

Ripple filtered signals are phase-locked within but not between hemispheres.

(A) Top: A representative reference ripple event (100-250 Hz; black). Middle and bottom: The ripple event on an ipsilateral shank (left hemisphere, blue) exhibits constant phase difference (green), whereas the ripple event on the contralateral hippocampus (right hemisphere, red) shows a non-constant phase difference (pink) to the reference event. (B) Polar histograms displaying phase differences between ripples in the reference shank and ipsilateral (top) or contralateral (bottom) shank. PLV denotes the inter-regional phase coupling metric, defined as the length of the mean resultant from unitary vectors (red arrows). (C) Raincloud plots showing PLV for ipsilateral and contralateral shanks (LMMR, side, Nipsi = 220, Ncontra = 245, p-value ≈ 0). (D) Same as (C), with data sorted by inter-shank distances. Regression line slope for ipsilateral is –0.076 for 200 μm inter-shank distance (LMMR, Nipsi = 220, p – value = 4.94 10−78). Regression line slope for contralateral is –0.001 for 200 μm inter-shank distance (LMMR, Ncontra = 245, p−value = 0.053). (E) Thin traces show a representative cross-correlation between the ripple-filtered signal from the reference shank and the ripple-filtered signal from ipsilateral (green) or contralateral (pink) shanks. The thick traces depict the amplitude envelopes of these cross-correlations; the peak amplitude is taken as a phase coupling metric. (F) Raincloud plots of cross-correlation maximum peak of ripple filtered signal for ipsilateral and contralateral shanks (LMMR, side, Nipsi = 220, Ncontra = 245, p−value ≈ 0). (G) Same as (F), but data sorted by inter-shank distances. Regression line slope for ipsilateral is −0.41 × 106 (cross-correlation peak) for 200 μm inter-shank distance (LMMR, Nipsi = 220, p – value = 4.03 × 10−46). Regression line slope for contralateral is −0.094 × 106 (cross-correlation peak) for 200 μm inter-shank distance (LMMR, Ncontra = 245, p – value = 0.006).

The amplitude of ripple events correlates within and between hemispheres.

(A) Representative ripple-filtered signals (thin traces) and their instantaneous amplitude envelopes (thick traces). Dots represent detected ripple events at the reference shank. (B) Representative scatter plot of ripple amplitude during ripple events for the reference vs. ipsilateral shank (instantaneous amplitude values were averaged within 100 ms bins, 50 ms overlap). The Pearson’s correlation coefficient (r) is used as a metric of amplitude coupling. (C) Raincloud plots showing correlation coefficients between the amplitude of ripple events for ipsilateral and contralateral shanks (LMMR, side, Nipsi = 220, Ncontra = 245, p-value = 9.5 10−44). (D) Same as (C), but data sorted by inter-shank distances. Regression line slope for ipsilateral is –0.028 for 200 μm inter-shank distance (LMMR, Nipsi = 220, p – value = 1.7 10−14). Regression line slope for contralateral is –0.010 (Pearson’s correlation coefficient) for 200 μm inter-shank distance (LMMR, Ncontra = 245, p – value = 8.6 10−5). (E) Representative cross-correlation between the instantaneous ripple amplitude in the reference shank and in the ipsilateral (green) or contralateral (pink) shanks (the whole time series – and not only ripple events – was used in this analysis). (F) Raincloud plots of cross-correlation maximum peak of instantaneous ripple amplitude for ipsilateral and contralateral shanks (LMMR, side,Nipsi = 220, Ncontra = 245, p−value = 9.9 ×10−181). (G) Same as (F), but data sorted by inter-shank distances. Regression line slope for ipsilateral is −0.22 × 106 (cross-correlation peak) for 200 μm inter-shank distance (LMMR, Nipsi = 220, p – value = 2.6 × 10−37). Regression line slope for contralateral is −0.031×106 (cross-correlation peak) for 200 μm inter-shank distance (LMMR, Ncontra = 245, p – value = 3.5 × 10−4).

Ripple synchronizes between hemispheres at the amplitude but not phase level.

(A) Left : Effect sizes for ipsi – contra, measured as Hedges’ g, observed for phase coupling (Figure 3C) and amplitude coupling (Figure 4C). Positive values indicate that the mean coupling value is larger for ipsilateral shanks (Hedges’ g, Nipsi = 220, Ncontra = 245; phase difference = 3.99, p – value ≈ 0, Hedges’ g; amplitude difference = 0.859, p – value ≈ 0). Right : difference between the two effect size distributions shown on the left (Deltas’ g, phase–amplitudedifference = 3.2, p – value ≈ 0). A value higher than zero indicates that the effect size for phase coupling is larger than for amplitude coupling. Data is presented as a half-violin plot for 5000 bootstrap distribution of Hedges’ g along with the mean and a bootstrap 95 % confidence interval.(B) Mean distribution of the Hedges’ g difference (phase–amplitude). Open purple circles represent the mean difference of shanks from the same recording session, i.e., average of all combinations of ipsilateral and contralateral shanks (Paired T-test, Nphase = 8, Namplitude = 8, difference = 1.765, p – value = 0.00254). Values closer to zero indicate that phase and amplitude coupling metrics are approximately the same; one the other hand, values above zero indicate that the effect size of phase coupling is higher than amplitude coupling. Filled circle and vertical bar represent the mean and the bootstrap 95 % confidence interval, respectively. (C) Same as in (A) but for cross-correlation analyses. Left : Effect sizes for ipsi – contra, measured as Hedges’ g (Nipsi = 220, Ncontra = 245; phase difference = 4.31, p – value ≈ 0; amplitude difference = 2.07, p – value ≈ 0). The phase label represents the cross-correlation maximum peak of ripple filtered signal (Figure 3F); amplitude label, the cross-correlation maximum peak of instantaneous ripple amplitude (Figure 4F). Right : difference between the two effect size distributions shown on the left (Deltas’ g, phase – amplitude difference = 3.28, p – value ≈ 0). (D) Same as (B), but for cross-correlation analyses (Paired T-test, Nphase = 8, Namplitude = 8, difference = 2.757, p – value = 0.0056).

Ripple events occur in synchrony between hemispheres at the decisecond timescale.

(A) Box and half-violin plots for ripple coincidence, defined as the percentage of overlapped ripple events in the given time windows (5 ms, 50 ms or 100 ms). LMMR (Nipsi = 220, Ncontra = 245) revealed a main effect on the side variable (p-value = 1.7 10−102) and bin size (p-value = 9.4 10−157), and also a interaction effect on side vs. bin size (p – value = 1.2 10−18). Tukey HSD (FWER = 0.05) post-hoc pairwise comparisons show significant differences between all conditions. Green represents ipsilateral events and pink contralateral ones. (B) Mean effect size (Hedges’ g) for ripple coincidence differences across time bins (ipsi contra; 5 ms difference = 1.7, p – value ≈ 0; 50 ms difference = 0.769, p – value ≈ 0; 100 ms difference = 0.706, p – value ≈ 0). (C) Box and half-violin plots of peak values from cross-correlation of ripple events across 5, 50 or 100 ms time bins. In this analysis, ripples were coded as one if a single event occurred in that time bin, or zero otherwise. The cross-correlation value at zero lag was taken as the metric of synchrony. LMMR (Nipsi = 220, Ncontra = 245) revealed a main effect on the side variable (p – value = 2.3 10−135) and bin size (p-value = 1.4 10−108), and also a interaction effect on side vs. bin size (p-value = 9 10−20). Tukey HSD (FWER = 0.05) post-hoc pairwise comparisons show significant differences between all conditions, except “50 ms bin–contra” vs. “5 ms bin–ipsi”. (D) Mean effect size (Hedges’ g) for cross-correlation peak of ripple events differences across time bins (ipsi contra; 5 ms difference = 1.92, p – value ≈ 0; 50 ms difference = 1.07, p – value ≈ 0; 100 ms difference = 0.957, p-value ≈ 0). For effect size figures B and D, filled black circle indicates the mean difference, the purple half-violin plot displays the distribution of 5000 bootstrapped mean differences, and the vertical line around the mean shows the bootstrap 95 % confidence interval. Red dashed line indicates the threshold where the effect size difference is not statistically significant.

Spiking activity is globally coupled to ripple events.

(A-B) Ripple triggered firing rate of from ipsilateral (green) and contralateral (pink) CA1 pyramidal cells (A) or interneurons (B). Ripple peak amplitude was used as the trigger. Notice the presence of a ripple oscillatory component modulating firing rate in ipsilateral shanks.(C-D) Spike-phase distributions for ipsilateral and contralateral ripple events for pyramidal (C) or interneurons (D). Top trace shows a ripple oscillation phase reference. In the 2D histogram, each line shows a single neuron (z-scored firing rate) sorted according to the peak ripple phase. The same neurons are plotted in the contralateral histogram using their ipsilateral sorted positions. Red trace at the bottom shows the average firing rate for ripple phase across all neurons. (E-F) Histogram of the mean spike ripplephase distribution from panels C-D for ipsi (green) and contra (pink) ripple events. (G-H) Spike-ripple coupling metrics for pyramidal (G) or interneurons (H). Left panel : Spike coupling to ripple phase by means of PLV. The PLV is calculated for each neuron individually. Right panel : Neuronal firing rate difference between inside and outside ripple events for ripples detected in ipsilateral and contralateral shanks. Positive values mean that the firing rate is higher during ripple. LMMR revealed an effect for side variable of pyramidal neurons phase coupling (G left panel, Npyr = 559; p−value ≈ 0), but not for firing rate difference (G right panel, Npyr = 559; p – value = 0.218). Likewise, LMMR revealed an effect for side variable in interneurons phase coupling (H left panel, Nint = 128; p – value = 6.77 × 10−85), but not for firing rate difference (H right panel, Nint = 128; p – value = 0.61). (I-J) Left panel : Proportion of pyramidal neurons (I) or interneurons (J) significantly coupled to ipsilateral and contralateral ripple phase. Right panel : proportion difference of ripple phase coupled neurons between contralateral and ipsilateral (pyramidal neurons, Npyr = 559, contra–ipsi difference = −0.805, p – value ≈ 0; interneurons, Nint = 128, contra–ipsi difference = 0.281, p – value ≈ 0). The proportion difference measures the effect size between ipsi and contra. The filled black circle indicates the proportion difference, the grey half-violin plot displays the distribution of 5000 bootstrapped mean proportion differences, and the vertical line around the mean shows the bootstrap 95 % confidence interval. Neuronal phase coupling was assessed by creating a PLV control distribution of 300 random time shifts of ±10 s in neurons timestamps relative to ripple phase series; neurons whose PLV was higher than the 95th percentile were deemed as significant.

Ripple events are phase-locked within but not between hemispheres.

(A) Raincloud plots showing PLV during ripple events for ipsilateral and contralateral shanks (LMMR, side, Nipsi = 220, Ncontra = 245, coefficient = −0.568, p – value ≈ 0). (B) Same as (A), with data sorted by inter-shank distances. Regression line slope for ipsilateral is −0.091 for 200 μm inter-shank distance (LMMR, Nipsi = 220, p – value = 3.17 × 10−47). Regression line slope for contralateral is −0.006 for 200 μm inter-shank distance (LMMR, Ncontra = 245, p−value = 9 × 10−8).

Spikes are locally coupled to ripple signal but globally coupled to ripple amplitude across the septo-temporal axis.

(A-B) Mean spike-triggered ripple-filtered signal (left) or ripple amplitude (right) for ipsilateral and contralateral shanks for pyramidal (A) or interneurons (B). Dashed vertical lines represents the spike event. (C-D) Spike-triggered ripple-filtered signals for pyramidal neurons (C) or interneurons (D) separately. Colors show the z-scored normalized signal (red denotes positive signal deflections; and blue, negative ones). (E-F) Spike-triggered ripple-amplitude for pyramidal neurons (E) or interneurons (F) separately. Colors show the z-scored normalized amplitude (redness level represents the amplitude).

Spikes are locally phase-coupled to ripples and globally coupled to ripple events across the septo-temporal axis.

(A-B) PLV for pyramidal (A) or interneuron (B) spike ripple phase. This is the same as shown in left panels from Figure 7G and 7H, respectively, but across shank distances for ipsilateral (green) and contralateral shanks (pink). Regression line slope for PLV of pyramidal neurons in ipsilateral shanks is −0.07 for 200 μm inter-shank distance (LMMR, Nipsi = 3026, p – value ≈ 0); regression line slope for contralateral is −0.0002 for 200 μm inter-shank distance (LMMR, Ncontra = 2927, p−value = 0.38). Likewise, regression line slope for PLV of interneurons in ipsilateral shanks is −0.044 for 200 μm inter-shank distance (LMMR, Nipsi = 686, p – value = 2.9 × 10−83); regression line slope for contralateral is –0.0003 for 200 μm inter-shank distance (LMMR, Ncontra = 662, p – value = 0.328).(C-D) Firing rate difference (Hz) for pyramidal (C) or interneurons (D) between inside and outside detected ripple events. This is the same as shown in right panels from Figure 7G and 7H, respectively, but across shank distances for ipsilateral (green) and contralateral shanks (pink). In all cases, we did not find statistically significant effects on side, distance nor interaction variables from both pyramidal and interneurons.