Ca2+ measurement
- Cell population: live Ca2+ fluctuations were recorded at 37°C on ST-hmGlu5 GcaMP6s-P2A-Scarlet HEK293T cells, or when specified on GcaMP6s-P2A-Scarlet HEK293T cells transfected with pRK-SNAPTag-AT1 (25 ng/well), seeded at 5 x 104 cells per well in 96-wells plates with transparent bottom, with the FDSS/µCell plate reader (Hamamatsu®) – Acquisitions of 1Hz; Exc: 480 nm – Em: 540 nm; high-speed digital EM CCD camera. Each independent experiment was performed in triplicate, which average corresponds to n = 1. Data were normalized to the maximum response at Vrestin each independent experiment. Basic calculations from the raw data files were performed on Microsoft Excel (Office 2016), statistical analyses and graphics on GraphPad Prism 8.1.
- Single cell imaging: GcaMP6s fluorescence fluctuations were recorded on ST-hmGlu5 GcaMP6s-P2A-Scarlet HEK293T cells, using an Axio Observer 7 KMAT fluorescence microscope (Carl Zeiss), equipped with a Plan-Neofluar 40x/1.30 EC oil objective (M27, ZEISS®), Exc 470/40 nm – Em 525/50 nm filters and ORCA-Quest qCMOS camera (Hamamatsu); all controlled with Metamorph software. Images on HEK cells were acquired at 3Hz and analysed with Fiji using a custom code computing the average fluorescence intensity of the regions of interest. For the frequency analysis, MATLAB was used to automatically detect the oscillations occurrences and compute the instantaneous and global oscillation frequencies. First, baselines were computed as the 10th percentile of the trace using a sliding window of size 400 frames. Signal was then divided by the baseline. Oscillations were detected using the function findpeaks with a minimal peak prominence of 0.01. Instantaneous frequency was computed as the number of oscillations minus one divided by the time between first and last oscillation. Global frequency was computed as the number of oscillations divided by the total duration. Spontaneous Ca2+transients in neurons infected with the pWPT-CAMKIIa-GcaMP6s-P2A-Scarlet lentivirus were acquired at 10 Hz, for series of 2 min and 15 sec. Image stacks were then analyzed as follow: images were filtered with a 5*5 median kernel. Each pixel was divided by its average value over the whole stack. Each pixel was gaussian filtered along the time axis with a sigma of 2 frames. Time derivative was applied to the stack with a step of 6 frames. Images were thresholded with a value of image mean + 6 * image standard deviation. A filter was then applied to keep only voxels with at least 10 thresholded voxels in their 3*3*3 neighborhood, to remove isolated voxels. Thresholded voxels spatially connected were assembled as events. Maximal time projection of those events gave their spatial footprint: the list of pixels in the image. Events with lower than 16 pixels were rejected. Events were classified in different categories based on two criterions: a threshold of 1000 pixels separating big events from small events, a threshold of 0.8 px of propagation distance separating events propagating or not. Propagation distance was measured as the difference of position between the centroid of the pixels thresholded at the beginning and at the end of the onset. Both thresholds were applied to focus on “small, non-propagating calcium transients” at synapses. Projection images were obtained by accumulation of the events footprint. Fluorescence time traces were extracted for each event as the average fluorescence over the event pixels and normalized by the fluorescence just before the onset of the event. Area under curve was computed between the onset and 4 seconds later.