Time-lapse imaging

For the imaging of cells dynamics during apoptosis induced by 500 ng/mL AMD, HeLa cells stably expressing histone H2B tagged with GFP (H2B-GFP) were used to seed Ø21 mm uncoated glass-bottomed “Ibidi μ-Dish-500” Petri dishes (Ibidi GmbH, Germany). To study mitochondrial depolarization during apoptosis [15], we added tetramethylrhodamine ethyl ester (TMRE) at a final concentration of 40 nM and incubated the cells for 30 minutes before imaging. After adding 500 ng/mL AMD, dishes were placed on the stage of an LSM 710-NLO laser scanning confocal microscope (Zeiss Microsystems, Gennevilliers, France) enclosed in an XL-5 dark LS 2000 incubator (PeCon, Germany) maintained at 37°C with a heating unit and a temperature controller. For three-dimensional time-lapse imaging (4D and 5D), images were acquired with a 63x Plan-apochromat 1.4 NA oil objective (Zeiss Microsystems, Gennevilliers, France). Two-photon excitation at 860 nm, with a CHAMELEON femtosecond titanium-sapphire laser (Coherent, Santa Clara, CA) at a power of 1.5% was used to simultaneously generate GFP and TMRE fluorescence with limited phototoxicity and photobleaching [16]. Differential interference contrast (DIC) imaging was also performed simultaneously with a specific detector for transmitted light, to investigate changes in the morphology of the cells and their main compartments (organelles in the cytoplasm, nuclei and nucleoli).

Typically, z-stacks of 512 x 512 pixels (corresponding to a field of view of 134.69 μm x 134.69 μm) containing 62 to 85 slices were simultaneously acquired in three channels (GFP fluorescence, TMRE fluorescence and DIC), every 5 minutes for 7 h 15. In these conditions, 62 to 85 z-stacks were collected with an xyzt resolution of 0.26 x 0.26 x 0.3 μm x 5 minutes for each experiment. Each z-stack was then processed with Amira® 5.6 (FEI VSG, Mérignac, France) and Imaris® (Bitplane, Zurich, Switzerland) to perform 3D visualization of GFP and TMRE fluorescence at each time point. For this, we used two procedures. The first one, called surface rendering, computes a solid 3 D surface of an object boundary by using only pixels of the latter which are defined by a fixed grey level. The second one, called volume rendering, computes a transparent view of an object by using the different grey levels of all its voxels. To analyze the changes of nuclear volume and of TMRE intensity in each chosen cell, we performed a semi-quantitative analysis of GFP and TMRE signals. First, we used Imaris software to quantify the volume of all voxels containing H2B-GFP fluorescence which is approximately the volume of the nucleus. Second, we used ImageJ software (NIH, Bethesda) to quantify TMRE intensity. We defined a region of interest (ROI) surrounding a cell of interest. We then sum the intensity of all pixels containing TMRE fluorescence in this ROI in all optical sections of each z-stack (on a 32 bits scale). We calculated the mean pixel intensity of TMRE fluorescence in this ROI for each time point. Finally, as we evidenced that TMRE outside of mitochondria produced a faint fluorescence which disappeared after the first hour of the experiment, we decided to analyze the changes of TMRE intensity relatively to its value reached at time 1 hour.

We selected one optical section from each DIC z-stack for investigation of the morphological changes occurring within the whole cell or within the cytoplasm or the nucleus during AMD treatment. The 62 to 85 images for each channel (3D visualization of GFP fluorescence, of TMRE fluorescence and of DIC optical sections) were then assembled with Quick-time® to create movies showing the 3D changes in chromatin and in mitochondria over the course of the whole experiment.