Electroconvulsive stimulation (ECS) triggers similar slow neuronal oscillations in mice and humans.

(A)Top: Schematic of mouse ECS setup. We performed concurrent widefield calcium imaging and EEG in the mouse dorsal cortex during biauricular ECS. Bottom: Layout of the EEG probe. (B)Spectrograms of two widefield calcium imaging ROIs surrounding EEG electrodes L12 and R12 (as highlighted in A). Duration of the ECS is marked by a black bar. (C)Spectrograms of the EEG recorded concurrently with the widefield calcium data in C. Time during which EEG is not recorded due to a stimulation artifact is marked by gray shading. (D)Schematic of patient ECS setup. EEG recordings were performed using two frontal electrodes after bifrontal ECS in patients. (E)Spectrograms of EEG recorded following ECT in a patient. Patient EEG recordings are started following ECS and terminated when the oscillation stops. Gray shading marks times without EEG recording. (F)Top: Oscillations observed in early, mid, and late phases after ECS in an example mouse widefield calcium ROI (blue) and in the concurrently recorded EEG (green). Bottom: Oscillations observed in an example patient EEG recording. (G)Oscillation duration for mouse and patient EEG recordings. The central box represents the interquartile range (IQR), spanning from the first (Q1) to the third (Q3) quartiles. Whiskers extend to data points not considered outliers (Methods). A central horizontal line indicates the median and an open circle the mean of the data distribution. (H)Power-weighted mean frequency at peak and offset of oscillations for mouse widefield, mouse EEG and patient EEG recordings. Here and elsewhere: n.s.: not significant, *: p < 0.05, **: p < 0.01, ***: p < 0.001. See Table S1 for all information on statistical testing.

ECS can trigger a travelling calcium event.

(A)Left: Schematic of mouse ECS setup without concurrent EEG recordings. Right: Widefield recordings of mouse dorsal cortex were parcellated into 818 ROIs, each of which was assigned to a cortical region according to the Allen Brain Atlas taxonomy. The ROIs highlighted correspond to the calcium activity shown in B and E. (B)Example of calcium activity in ROIs highlighted in A during the traversal of a calcium event. (C)Propagation speed of calcium events (Methods). Each data point is the median speed across one calcium event. (D)Maximum calcium activity throughout the dorsal cortex during pre-ECS, direct ECS response and calcium event periods. Two direct ECS responses are outside the plot range and not shown (1707 % and 2374 % ΔF/F0). (E)Left: Low-passed (< 0.5 Hz) calcium activity during an example ECS session triggering calcium event over both hemispheres. Each panel shows the average calcium activity over a 10 ms window starting at the time indicated above the panel. Right: Isochrones (spaced at two seconds increments) indicating the time at which each region reaches 200% of its baseline fluorescence. (F)As in E, but for a unilateral calcium event over the right hemisphere following a 150 s delay post-ECS. (G)As in E, but for a unilateral calcium event over the left hemisphere.

Calcium events are triggered preferentially by higher total ECS charge.

(A)Probability distribution of the laterality of calcium events. (B)Spatial distribution of calcium events detection sites. For bilateral calcium events, only the hemisphere where the calcium event was detected earliest is considered as the detection site. (C)Fraction of stimulation responses, oscillations and calcium events that were bilateral. (D)Probability of oscillations being observed if a calcium event is triggered (P(osc. | calcium event)), of a calcium event being observed if an oscillation is triggered (P(calcium event | osc.)), and of neither being observed (P(neither)). (E)ECS parameter map of the likelihood of triggering a calcium event (top) or an oscillation (bottom). (F)Probability of triggering a calcium event as a function of total ECS stimulation charge. The solid line represents the linear fit of the data; the shaded region represents 95% confidence interval on the fit. (G)As in E, but for the probability of triggering an oscillation.

Calcium events first spread in supragranular layers and elicit long-lasting calcium increase in cells.

(A)Schematic of the experimental setup. We performed two-photon (2p) calcium imaging in the right primary visual cortex during ECS. (B)2p calcium activity during an example ECS session triggering a calcium event. Time of maximum (TOM) is shown on the right. Each panel shows the average calcium fluorescence over a 500 ms window starting at the time indicated above the panel. Gray inset highlights two example somata whose calcium signals are shown in G. (C)Top: Schematic of the micro-prism implanted below the imaging window to allow for simultaneous recording in all cortical layers (see Methods). Bottom: Widefield imaging of the implant 18 seconds after ECS, confirming the prism does not obstruct the calcium event propagation. A gray arrow indicates the orientation of the prism. (D)As in B, but imaged through a prism implant. Cortical layers 3, 4, and 5a (L3, L4, L5a) are defined based on soma densities (Methods). (E)Calcium fluorescence of superficial (blue) and deep (green) layers somata during calcium events, aligned to the time of maximum of all neuropil (Methods). Lines represent the hierarchical bootstrap estimates of the mean value for each time bin. Here and elsewhere, shading around the mean is one standard deviation of the bootstrap distribution at each time bin. Response curves are compared for each time bin: in the horizontal bars above the plot, black marks time bins where p<0.05 and gray mark time bins where p>0.05. (F)Time of 50% of maximum activity for superficial and deep layers somata during calcium events. (G)Top: Calcium fluorescence from the two example somata in B. Bottom: Calcium fluorescence from the neuropil surrounding these somata. Fitted slopes during the 60 seconds following calcium event onset are shown as dashed lines. (H)Calcium fluorescence of somata (dark gray) and neuropil (purple) in all layers during calcium events. Each soma is aligned to the time of maximum of the neuropil. (I)Distribution of the fitted slopes for cells and neuropil during the 60 seconds following calcium events. (J)Scatter plot of fitted slopes and correlation of cells with their neuropil. Top and right histograms are marginal distributions over slopes and correlations, respectively.

The ECS triggered increase in Fos expression in cortex is driven by the calcium event.

(A)Example of coronal brain sections stained for Fos from three sham ECS (top) and three ECS-treated (bottom) mice. (B)Maximum calcium fluorescence during direct ECS response (first row), the amplitude of the ECS induced oscillation (second row), and the peak fluorescence to illustrate the presence or absence of a calcium event (third row). From left to right, data from three example mice that had unilateral left, unilateral right, and bilateral calcium events, respectively. The bottom row shows corresponding coronal sections from the same brains, stained for Fos expression. (C)Quantification of the Fos throughout the entire cortex, expressed as a fold change for an entire hemisphere normalized to sham mice (see Methods). Paired hemispheres (experiments in which ECS triggered a bilateral direct ECS response and oscillation followed by a unilateral calcium event) are linked by a line. Thus, the calcium event, not the direct ECS response or the phase III oscillation, is the correlate of Fos expression.

Calcium events are associated with ictal EEG metrics that predict clinical ECT efficacy.

(A)Example of an ECS session triggering a direct ECS response followed by a slow oscillation. Data is shown for two example electrodes. A gray transparent square indicates the duration of the oscillation. First row: Calcium imaging spectrogram of the ROI surrounding the electrode. Second row: EEG signal spectrogram of the electrode. Third row: z-scored EEG signal in the 1-5 Hz band to visualize the oscillation amplitude. Fourth row: EEG Hilbert amplitude in the 1-30Hz band, with dashed lines indicating the mean amplitudes during and after the oscillations used to compute the postictal suppression index (PSI). Fifth row: Calcium imaging fluorescence, low-passed below 0.5 Hz. (B)As in A, with identical ECS stimulation parameters, but for an ECS session triggering a direct ECS response followed by a bilateral calcium event. (C)Mid-oscillation amplitude for each electrode during ECS sessions triggering only direct ECS response (gray), ECS sessions triggering a unilateral calcium event over the contralateral hemisphere (blue) and ECS sessions triggering a calcium event on the ipsilateral hemisphere (red, either unilateral or bilateral calcium event). See Methods for the details on the ictal EEG metrics definitions. (D)As in C, but for the oscillation duration. (E)As in C, but for the postictal suppression index. (F)As in C, but for the inter-hemispheric coherence of the EEG signal.

Statistics table.

All values are rounded to two significant figures. The tests used were two-tailed hierarchical bootstrap, two-tailed bootstrap, one-tailed bootstrap, F-test, and t-test. Benjamini-Hochberg correction was used for false discovery rate (FDR) adjustment in multiple comparisons, in which case the adjusted p-value padj is the one reported in the associated figure and text. Sample sizes were estimated based on previous experiments performed in mouse behavioral paradigms (Attinger et al., 2017; Heindorf et al., 2018).

Number of subjects in each analysis.

ECS triggers a direct calcium response during the EEG disconnection. Related to Figure 1.

(A) Left: Widefield recordings (without concurrent EEG) were parcellated into 818 ROIs, each of which was assigned to a cortical region according to the Allen Brain Atlas taxonomy. Right: Direct ECS calcium activity triggered by an example ECS. Each panel shows the average calcium activity over a 10 ms window starting at the time indicated above the panel. (B)Direct ECS calcium response estimated using hierarchical bootstrap across all ECS. Response curves are compared against baseline distribution for each time bin: in the horizontal bars above the plot, black marks time bins where p<0.05 and gray mark time bins where p>0.05. Duration of ECS is marked as a dark gray shaded area. (C)Left: during ECS (dark gray shaded area), the EEG electrode is temporarily disconnected to prevent arcing (Methods), leading to a period in which no EEG data is recorded (light gray shaded area). To eliminate the capacitive discharge artifact upon reconnection, the first 2 s of data are excluded from analysis (dashed box). Right: same signal with a longer time scale, following artifact removal and 0.5–10 Hz band-pass filtering.

ECS triggered oscillations are synchronized between widefield calcium and EEG in mice. Related to Figure 1.

(A) Activity observed early after ECS in an example mouse widefield calcium imaging ROI (blue) and in the concurrently recorded EEG (green). Oscillations peaks were detected using Automatic Multi-scale Peak Detection on the EEG channel (Methods). (B)As in A, in a period of time 112 s after ECS. (C)As in A, in a period of time 197 s after ECS.

Influence of stimulation parameters on calcium events. Related to Figure 3.

(A) Probability of triggering a calcium event as a function of stimulation frequency (left) or current (right). The solid line represents the linear fit of the data; the shaded region represents 95% confidence interval on the fit. Error bars on individual data points indicate standard error. (B)As in A, but for oscillations. (C)Left: Distribution of calcium events speed across triplets of ROIs (see Methods). Right: Median calcium event speed as a function of stimulation charge. Dashed line indicates median calcium event speed across all charges. (D)Left: Distribution of the duration of calcium events in each widefield ROI (see Methods). Right: Median duration of calcium event as a function of stimulation charge. Dashed line indicates median duration across all charges. (E)Left: Delay from stimulation to calcium event onset. Right: Delay from stimulation to calcium event onset as a function of stimulation charge.

Direct ECS response in calcium imaging predicts the occurrence of calcium events. Related to Figure 3.

(A) Left: Maximum calcium activity during the direct ECS response as a function of stimulation charge. Red values are measured during ECS triggering a calcium event (either unilateral or bilateral) and gray values from ECS that did not trigger a calcium event. (B)Maximum calcium activity in both hemispheres fluorescence during the direct ECS response, before a unilateral calcium event was triggered.

ECS triggers hemodynamic responses. Related to Figure 3.

(A) Left: Direct ECS fluorescence measured with GCaMP6s (as shown in Figure S1B) and EGFP, estimated across the population using hierarchical bootstrap. In the horizontal bars above the plot, black marks time bins where p<0.05 and gray mark time bins where p>0.05. The top bar represents the comparison between GCaMP6s and EGFP fluorescence, while the bottom bar compares the EGFP fluorescence to baseline distribution. Right: Direct EGFP fluorescence evoked by a single example ECS. Each panel shows the average calcium activity over a 10 ms window starting at the time indicated above the panel. (B)Low-passed EGFP fluorescence during an example ECS session with unilateral calcium event-like propagation over the right hemisphere. (C)Maximum ΔF/F0 during calcium and EGFP events. (D)Propagation speed of travelling events (Methods). Each data point is the median speed across one event.