Computational protocol: The Physical Basis of Coordinated Tissue Spreading in Zebrafish Gastrulation

Similar protocols

Protocol publication

[…] Laser scanning two-photon microscopy images were processed using Fiji (NIH) as follows: images were converted from original 16-bit type to 8-bit type at a certain threshold level. Because the red (mCherry and RFP) and far-red (Alexa 647) fluorescence slightly bled through into the other channel's detector (i.e. red fluorescence into the far-red detector and vice versa), images from these channels were subtracted from each other between the same x, y and z pixel positions based on the signal intensities. Images were subsequently processed by image registration using the Fiji plugin Correct 3D drift () and then used for further analyses. To average results from different embryos, embryo images were temporally aligned by using the beginning of the last nuclei division within external yolk syncytial layer (YSL), which is a known characteristic feature of embryos at the onset of doming (). For BYI surface area (3D) measurements, dextran signals of BYI were processed by filtering using Gaussian blur and binarization, and detected at their closest position to the interface with the blastoderm using a custom-built Fiji script. The BYI surface area was then calculated by connecting those detected dextran signal points as a surface using Matlab. Deep cell radial speed was measured by processing two-photon microscopy-obtained images of blastoderm nuclei using the Spots Object function in Imaris (version 7.4, Bitplane), which allows to track the movement of each nucleus over time and obtain x-, y- and z-coordinates of their movements. To measure the radial speed of deep cells from the tracking data, the three-dimensional prospective center of the embryo was first determined at each time point (A and S7B). To this end, embryo side view image stacks were made by the Fiji ‘Reslice’ function along x and y axes, respectively (A). These image stacks were processed with maximum intensity projection and then binarized with a threshold in gray scale value, which includes about 30% darker pixels of total number of pixels. The arc contours of the binary images corresponding to the embryo outer surface were detected and fitted with the Fiji ‘Fit Circle’ function (B). These circles were then used to determine the x-, y- and (mean) z-coordinates of the embryo center. To obtain the radial speed, the distance between each tracked nucleus and the embryo center was used to calculate the radial displacement of deep cells using Matlab. In embryos containing transplanted EVL/surface cells, the deep cell speed was measured only under the transplanted cells. For EVL surface area measurements, two-photon microscopy images of embryos expressing mem-GFP or mem-RFP were used. Because the images from two-photon microscopy contained signals from both the EVL/surface cells and the underlying deep cells, images were first processed by deleting signals from deep cells in each z slice after Gaussian filtering, binerization and determining the area which was created by reducing the outer contour of binerized image by the assumed thickness of the surface layer (C–S7K). The processed images were assembled as a maximum z-projection, followed by semiautomatic segmentation using Packing Analyzer (v6.5; ), 3D correction of the segmented images and measurement of the surface area using a custom-built Matlab script. The surface area was measured for cells (and daughter cells after division) that remained within the image frame throughout the analyzed time points. To measure the fluorescent intensity of Lifeact-EGFP and Myl12.1-EGFP at the BYI, optical single z sections were obtained using two-photon microscopy in a ∼100 μm depth from the embryo surface. The BYI was manually spotted, and the mean gray values of the fluorescent signal were measured within a ∼100 μm region of the YSL below the BYI. The measured values were divided by the ratio of the mean gray values of the whole image at each time point over those at the initial time point (0 min) to correct for bleaching. For the aspect ratio and angles of the deep cells and yolk granules, mem-RFP and dextran Alexa 647 signals were used to detect their membrane and cytoplasm, respectively. The images were segmented with Packing Analyzer and then measured with a custom-built Fiji script that fits an ellipse to each cell/granule shape. To measure the angle of the cell's major axis, the center of the embryo section was determined by fitting a circle to the contour of the section image and the connected to the center of the fitted ellipse of the cell by a line. To determine the alignment of the cell's major axis with the radial axis of the embryo, the angle between this line and the cell's major axis was measured. To determine the preferred orientation of the yolk granules, the angle between the major axis of yolk granules and the horizontal line of the image was measured after tilting the whole image so that the BYI before doming was oriented parallel to the horizontal line of the image. To quantify the subcellular localization of actin in deep cells, the image was first segmented with Packing Analyzer using mem-RFP expression for outlining individual cells. The segmented area was then further subdivided into 12 sectors with the center of those sectors being aligned with the center of a fitted ellipse and also divided by a circumferential line which surrounds about 3 μm inside from the plasma membrane (B, orange area), creating small compartments at the edge of each cell shape. Finally, the mean gray value of each small compartment was calculated and normalized by the value from 0-30 degree region from the major axis in order to localize signal intensity relative to the major axis of the fitted ellipse. To analyze the density and speed of deep cells near the surface of the embryo, the position of the outer surface of the embryo was first determined from confocal microscopy images of blastoderm cells expressing mem-GFP by using a custom-built Matlab script. In short, thresholded binary images of embryo side views were used to identify the outer surface of the blastoderm in a single pixel resolution, followed by treating the extracted surface with an averaging filter with 50 × 50 pixel size. The distance of deep cells from the embryo surface was determined by calculating the distance between each deep cell nucleus, which was detected by using the Spot Object function of Imaris, and its nearest point on the embryo surface. The density and radial speed of deep cells were measured in a region around the animal pole with an area of 100 × 100 μm in the equatorial plane and 50 μm away from the surface. […]

Pipeline specifications

Software tools Correct 3D Drift, Imaris
Application Laser scanning microscopy
Organisms Danio rerio, Caenorhabditis elegans