Microscopy images from: A cadherin-integrin-ECM code for presomitic mesoderm fluidity
Data files
Oct 22, 2025 version files 22.72 GB
-
cellposemodels.zip
49.17 MB
-
ECMfibers.zip
121.04 MB
-
README.md
2.44 KB
-
secretedGFP_zstack.zip
5.03 GB
-
timelapses.zip
17.52 GB
Abstract
Animal tissues exist within a continuum of fluid to solid states, and transitions between states are important for embryonic development, wound healing and cancer metastasis. Fluid to solid transitions are governed by the ratio of adhesive energy to kinetic energy. Here, we find that presomitic mesoderm solidification is driven by an intrinsic decline in cell speed along with an increase in adhesion mediated by Cadherin 2 in parallel with Fibronectin and its receptor Integrin α5. A computational model of cell-cell adhesion in the central tissue mesenchyme and cell-ECM adhesion on the tissue surface explains the observed phenotypes. Further, we identify negative feedback within the ECM as Fibronectin supports the formation of a separate layer of Fibrillin 2b matrix that inhibits solidification. These data reveal a tissue fluidity code in which solidification is promoted by cadherins in parallel with Integrin α5 and Fibronectin, whereas negative feedback through Fibrillin 2b promotes fluidization.
Dataset DOI: 10.5061/dryad.wstqjq30g
Description of the data and file structure
This dataset contains microscopy images of the zebrafish PSM obtained for the purpose of calculating cell shape and packing fraction (Figure 2), cell velocity (Figure 3), and ECM structure (Figure 5) in A cadherin-integrin–ECM code for presomitic mesoderm fluidity.
Files and variables
File: cellposemodels.zip
Description: Custom trained Cellpose 2.0 models for segmenting cells in coronal and transverse orientations using membrane-RFP and secreted-GFP. See Figure 2C and S4.
Two models: one for XY and one for XZ orientation for use with cellpose 2.0
XY model: itgcdh_xy_gray
Trained on images of MZ-itga5-; cdh2- expressing memRFP, secreted GFP,
Segmentation done in grayscale ie channels=[0, 0]
Presumably cellposes's grayscale conversion is 1/2 red intensity + 1/2 green intensity.
XZ model: XZ_quarterweight
Trained on a variety of wt, cdh2- and MZ-itga5-, cdh2- embryos; memRFP, secreted GFP orthogonal reslice images
Segmentation done on custom grayscale image calculated as red intensity + 1/4 green intensity, saved as 16 bit image.
Import model with command: model = models.CellposeModel(pretrained_model=file path to model, model_type=model name)
Run segmentation with: model.eval(image, channels, diameter)
File: secretedGFP_zstack.zip
Description: Confocal z-stacks of embryos expressing membrane RFP and secreted GFP. File name pattern is genotype _ embryoID _ PSM ID.czi See Figure 2A.
File: timelapses.zip
Description: Confocal z-stack, time-lapse of embryos expressing histone 2A-GFP. Frame rate is 3 minutes. File name pattern is date _ genotype _id.czi See Figure 3.
File: ECMfibers.zip
Description: Max projection of Airyscan images of dorsal PSM surface of knock-in embryos with fb2nb-mScarlet1 and/or fn1a-mNG. Each sub-folder contains embryos with a specific genotype (wild-type, fbn2b-, fn1a-; fn1b-). Channel order in wild-type is mScarlet then mNG. File name pattern is generally date-fluorescent alleles-genotype-10som-embryo id-PSM id. File format is multipage tiff. Pixel size is 0.1 micron. See Figure 5.
Code/software
Images can be viewed with ImageJ.
Cell segmentation is performed with Cellpose v2.0.