Statistics of the local field potential (LFP) spectral power and Ca²⁺ recordings in the three experimental groups. (A) Comparison of the statistics of LFP and whole brain Ca²⁺ fluctuations recorded in the three experimental groups, wild-type (WT) as control zebrafish, PTZ (Pentylenetetrazol) after treatment with pentylenetetrazole and MO (kcnj10a splice-morpholino) morphants of kcnj10a. The LFP transients in the PTZ and MO groups are associated to generalized positive fluctuation of the Ca²⁺. (B) Magnification of the LFP events is indicated by the magenta bars in panel A. The magenta traces show the full band data, while the blue traces show the data band passed in the 30–95 Hz range. (C) Statistical analysis of the electrophysiological activity. (D) Gaussian-like distribution of the ΔF/F0 values in the three models. (E) Difference between the Secondary Mode (SM) mean and the mean of the Main Mode (MM) (ΔmeanSM - MM). (F) Comparison of the MM mean of the three different models. (G) Statistics of the Pearson’s coefficient of skewness of the distribution of the fluorescence fluctuations, ΔF/F0, integrated on the entire optical section. Statistics have been cumulated from 5 recordings for each larva for each condition (* p ≤ 0.05,** p ≤ 0.01, *** p ≤ 0.001).

Dynamic evolution of the Ca²⁺ domains. (A,B) The volume representation depicts the active pixels from the binary stacks, clustered in domains of different colors of PTZ and MO larvae. The frame on the left is the projection of active domains during the entire period. The LFP events in red have high Ca²⁺ recruitment (>700 voxels). (C) The shaded sections in A and B are the spatial patterns of activation during the interictal-like events (frame 1) and the generalized events (frame 2) in the PTZ (red) and MO (blue). (D) Difference of the spatio-temporal morphology of PTZ and MO clusters with respect to WT clusters. The maximum surface occupied by Ca²⁺ clusters is plotted in relation to their duration. The color bars represent the p-value of the difference with the WT model. We report only the significative differences (p-value ≤ 0.05). 4 recordings for each larva, n = 3 larvae for each condition. (E) Comparison of the amounts of detected voxels for the three models in different anatomical regions. (F) Anatomical localization of the origin of the recruited neuronal clusters with volume > 300 voxels. The data have been normalized to the total number of active domains measured in each recording session (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001).

Valproate treatment. (A) Valproate (2 mM) was administered to MO kcnj10a larvae after 30 min of baseline recording of LFP and two-photon imaging. After a diffusion time of 15 min, the double recording was resumed. (B) Alignment of LFP and ΔF/F0 calcium traces in MO VPA( Valproic Acid)-treated larvae. (C) The analysis of the statistical distribution of the electrophysiological signal showed a clear effect of VPA in decreasing the energy of the LFP events. (D) The skewness of statistical distributions of Ca²⁺ fluctuations decreased during VPA treatment. (E) During VPA treatment, movements of the larvae did not correlate with a generalized increase of calcium signal. Therefore, VPA treatment led to a physiological phenotype, where tail flicks are associated not to a convulsive seizure but to the activation of a limited neuronal circuitry. 2 recordings for each larva, n = 3 larvae for each condition (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001).

Localization of the sources of the LFP transients. (A) Maps obtained by computing the cross-correlation of the Ca²⁺ fluctuations of each pixel with the logarithm of the RMS (Root Mean Sqaure) power in the band 30–95 Hz in the three experimental models. (B) Example of recordings from a PTZ-treated larva where the correlation between the largest electrophysiological events (labelled by an asterisk) and Ca²⁺ activity measured in the OT (green trace) is apparent, and in the MOb (pink trace). (C) Ca²⁺ fluctuations measured in the Cb (orange trace, represented as an area to better appreciate details) correlate with both the large RMS events and the smaller interictal-like transients. (D) ΔF/F0 of the frames corresponding to baseline electrophysiological activity (labelled 1 in the above panels), to a small interictal event (2), in which Ca²⁺ activity is sparse and mostly localized in the Cb, and to a large electrophysiological event (3) associated to generalized Ca²⁺ activation. The overlay shows ΔF/F0 fluctuations larger than 10% over-imposed on the mean projection of the image stack.

Identification of the micro-circuitry associated with different epileptic-like events. (A) Cross-correlation maps computed during two different intervals in th presence of only interictal activity (left) or during large ictal events (right). The cross-correlation maps have been normalized to the respective maximum value. (B) Details of the Cb, as outlined by the yellow rectangle (300 μm high) in panel A, showing the different sources of LFP signal during these two different classes of events. The finder map on the right shows three areas belonging to the cerebellum that are differentially active during interictal and ictal-like events. (C) ΔF/F0 measured in the three areas outlined by the corresponding colors in B. While all areas respond to the three large ictal events, the neuronal territory involved in interictal activity (green area and trace) is far more limited.

Principal component analysis (PCA) analysis identifies two different neuronal populations. (A) Population activity profiles represented by the first two PCA components describing the Ca²⁺ dynamics of single neurons. The black trace represents the corresponding RMS power. (B) Representative ΔF/F0 activity selected from single neurons best described by either PC1 (magenta) or by PC2 (green) in a PTZ-treated larva. PC1 neurons’ activation precedes the LFP bursts, whereas PC2 neurons activate during the LFP peaks. The grey bar indicates the events magnified in panel A. (C) Mean cross-correlograms of the two selected PCs with the aligned RMS power (n = 15 events from 4 morphants and n = 16 events from 4 PTZ larvae). (D) Cumulative maps for the PTZ (n = 4) and the MO (n = 4) larvae. Each cell is represented by its normalized loading (see the Materials and Methods Section) with respect to the PC1 and PC2. The plots on the right represent the normalized signal (arbitrary units) of the two-cell population along the rostro-caudal axis. The distribution of the two functional populations are significantly different in both models (p = 0.002 for the morphant and p < 0.0001 for the PTZ, two-samples Kolmogorov–Smirnoff test). In order to standardize the images, we registered the maps into a reference brain mask using a non-rigid image registration algorithm (see the Materials and Methods Section).