Date: 20 April, 2025 ------------------------------------------------------------------------------------------------ Dataset Title: Dataset on cell areas and nuclear densities in differentiating stem cell colonies ------------------------------------------------------------------------------------------------ Dataset Contact: Xufeng Xue xufeng.xue@cchmc.org ------------------------------------------------------------------------------------------------ Dataset Creators: Name: Xufeng Xue Email: xufeng.xue@cchmc.org Institution: Cincinnati Children's Hospital Medical Center Division of Developmental Biology and UC Department of Pediatrics ORCID: https://orcid.org/0000-0002-9379-8589 Name: Hayden Nunley Email: nunley@umich.edu Institution: University of Michigan Biophysics Program ORCID: https://orcid.org/0000-0002-4634-9422 Name: Jianping Fu Email: jpfu@umich.edu Institution: University of Michigan Department of Mechanical Engineering, Department of Biomedical Engineering, Department of Cell and Developmental Biology ORCID: https://orcid.org/0000-0001-9629-6739 Name: David K. Lubensky Email: dkluben@umich.edu Institution: University of Michigan Department of Physics and Biophysics Program ORCID: https://orcid.org/0000-0002-4619-116X ------------------------------------------------------------------------------------------------ Funding: CMMI 1917304 (NSF), DGE 1256260 (NSF), DMR 2243624 (NSF), 597491-RWC and 1764421 (Simons Foundation/SFARI, NSF), CMMI 1129611 (NSF), CBET 1149401 (NSF), CMMI 1662835 (NSF), 12SDG12180025 (American Heart Association) ------------------------------------------------------------------------------------------------ Key Points: - After optimizing culture conditions, we performed experiments with stem cell colonies on micro-patterned substrates as an in vitro model for an important cell fate decision in vivo. - We imaged stem cell colonies stained for DAPI (staining nuclei), E-Cadherin & N-Cadherin (both at adherens junctions) at different experimental timepoints and in different media. - We developed a method and code to estimate the area per nucleus as a function of distance from the colony center. Because it is difficult to segment cells in these images, as a proxy we calculate the number of nuclei (in concentric rings about the colony center) and divided by the total cellular area (estimated by semantic segmentation based on the E-Cad signal) in each ring. - We concluded that cells at the colony edge are significantly more spread than those in the colony bulk. - These spread cells at the colony edge form the basis of an important assumption in our mathematical model of the cell fate decision. The cells at the very edge of the colony are mechanically different from other cells (more spread and more contractile) and mechanically deform their neighbors, thus initiating the cell fate patterning. ------------------------------------------------------------------------------------------------ Research Overview: In an earlier publication (Xue et al. Nature Materials 2018), the authors reported a set of in vitro experiments in which uniformly supplied chemical media induced spatially patterned fates in cell colony in a disc geometry. They provided significant evidence that inter-cellular mechanical interactions, as well as mechanical interactions between cells and the substrate, play an important role in this in vitro differentiation process. In this subsequent publication, we propose a mathematical model for this fate patterning process and explore how the fate pattern depends on substrate stiffness. One ingredient of this mathematical model is that the cells at the very edge of the colony (lacking adherens junctions on one side) are geometrically different than the rest (by occupying a larger area on the micropattern). These images of DAPI (staining nuclei) and ECad (at adherens junctions) for colonies during early cell differentiation demonstrate this difference. Corresponding code for analysis is included. ------------------------------------------------------------------------------------------------ Methodology: Embryonic stem cells were seeded on micro-patterned substrates and were supplied with chemical media. The chemical media EITHER consisted of only conditioned medium (to support feeder-free growth of these cells without differentiation) OR of conditioned medium followed by Dual SMAD ("Suppressor of Mothers Against Decapentaplegic") Inhibitor media to induce differentiation into neural plate and neural plate border. At a couple of experimental time points, colonies were fixed and stained for DAPI (nuclei), E-Cadherin and N-Cadherin (both important proteins involved in cell-cell adhesions). The former two signals were used to estimate the number of nuclei per colony area. We wrote code to detect nuclear centroids, to compute a semantic segmentation of the colony, and to compute the number of nuclei per colony area in concentric rings about the colony center. ------------------------------------------------------------------------------------------------ Date Coverage: 2018-2024 (Date range include dates of experimental data acquisition (2018-2019), and subsequent development and use of analysis code for manuscript preparation (2019-2024)). Instrument and/or Software specifications: See Methods for experimental instruments, code written in MATLAB ------------------------------------------------------------------------------------------------ There are two zipped folders: 20180112_colony_thickness and my_code_height_profile. The former contains all the experimental data. The latter contains the code for image analysis. 20180112 colony thickness.zip contains an E-Cadherin, N-Cadherin, and DAPI (staining nuclei) image for each imaged colony. There are two different experimental conditions (CM for conditioned medium; DSI for dual SMAD inhibitor media). See Methods, Fig. S1A of Xue et al. Nature Materials 2018 for explanation of different media. Files ending in '_w1405.TIF' are DAPI images. Files ending in '_w2488.TIF' are E-Cadherin images. Files ending in '_w3561.TIF' are N-Cadherin images. Files ending in '.nd' are the image format from the microscope with all three channels in one file. These .nd files contain the same information that is in the corresponding three .TIFs, so all data are available in a non-proprietary format! These essentially serve as duplicates. In any case, these .nd files can be opened using the bio-formats plugin in ImageJ. The beginning of each file name is '20180212 XX DayY Ecad-488 N-cad-555 20x_Z', where XX is either CM or DSI, Y is the number of days after cell seeding at which image was acquired, Z is an index for the colony. There are 4 colonies in CM at Day 3. There are 5 colonies in CM at Day 4. [one of which was rejected from analysis, see below] There are 5 colonies in Dsi at Day 3. [one of which was rejected from analysis, see below] There are 3 colonies in Dsi at Day 4. Pixel dimensions are [10/7 um, 10/7 um] The folder called 'analyzed_images' stores cropped colony images ('cropped_image.mat' which contains 'c1_mat' and 'c2_mat' for DAPI and E-Cad z-stacks, respectively) and images with detected nuclei ('segmented_nuclei.mat' and/or 'segmented_nuclei_edited.mat'). Each 'segmented_nuclei.mat' is a 2D image (corresponding to z-projected image of DAPI signal) where each nuclear centroid is labeled by a single pixel. All are in the '.mat' format. All of these files can be recreated by running the code here on the data here. If the folder also contains a 'segmented_nuclei_edited.mat', then use this 'segmented_nuclei_edited.mat' since it contains any further corrections that were necessary in the nuclear centroid image. Some 'segmented_nuclei.mat' did not require further corrections. Please note that '20180212 CM Day4 Ecad-488 N-cad-555 20x_19' and '20180212 Dsi Day3 Ecad-488 N-cad-555 20x_4' have neither a 'segmented_nuclei.mat' nor a 'segmented_nuclei_edited.mat' file. This is because there was very unequal cell spreading in these colonies, including possible example of cell death (for example, a region devoid of cells and nuclei at the colony center). The remaining folder is my_code_height_profile.zip, a zipped folder containing code for analyzing images from 20180112 colony thickness.zip. The main functions for analysis all begin with 'count_nuclei_', in particular count_nuclei_func_rad3.m. Code in folder 'identify_nuclei' was used to detect nuclei, then to correct those detections. See more detailed list below: - identify_nuclei/identify_nuclei_by_imregionalmax.m is used to generate initial centroid detection for nuclei based on z-projection of DAPI image - identify_nuclei/remove_and_add_nuclei.m is used to remove nuclei from or add nuclei to the centroid image (computed by identify_nuclei_by_imregionalmax.m) - threshold_matrices.m is used to threshold both the DAPI and E-Cad images - sum_matrices.m is used to sum-project both the DAPI and E-Cad images along z - sum_intensity_fcn_of_z.m is used to calculate the integrated intensity (over xy in the whole colony) as a function of z. This can be used to infer where adherens junctions are. - simple_slice_c1c2.m is used to extract a cut (/ cross-section) through the 3D image in each channel. - open_image_stacks.m is a helper function used to open the DAPI and E-Cad z-stacks in a specified directory. - find_centroid_and_radius_from_threshold_mat.m estimates the radius and centroid of the colony (based on radius of smallest circle which encompasses entire segmented colony) - crop_c1c2_images.m crops the DAPI and E-Cad images based on the segmentation of the colony - count_nuclei_func_rad_voronoi_plot.m estimates the number of nuclei per colony area in multiple concentric rings [r, r+delta_r] and plots a Voronoi tessellation of the colony based on nuclear centroids. - count_nuclei_func_rad3.m is similar to count_nuclei_func_rad_voronoi_plot.m, but directly takes as input indices for Day 3 and Day 4 images in both media -- then loading the pre-saved results of nuclear detection and colony segmentation. - count_nuclei_fixed_rad.m computes the number of nuclei in one concentric ring centered at the colony centroid (between two specified radii). The folder min_circle contains helper functions for fitting the smallest circle in which all some segmented region fits. - min_circle/IcosahedronMesh.m generates a triangular surface mesh of an icosahedron. - min_circle/TriQuad.m subdivides triangular mesh using generalized triangular quadrisection. - min_circle/FitCircle2Points.m fits a circle to a set of 2 or at most 3 points in 3D space. - min_circle/ExactMinBoundCircle.m computes exact minimum bounding circle of a 2D point cloud using Welzl's algorithm. See Welzl, E. (1991), 'Smallest enclosing disks (balls and ellipsoids)', Lecture Notes in Computer Science, Vol. 555, pp. 359-370 ------------------------------------------------------------------------------------------------ Related publication(s): Nunley H, Xue X, Fu, J, Lubensky, DK. Generation of fate patterns via intercellular forces. BioRxiv 442205 [Preprint]. April 30, 2021 [cited 2025 Feb 20]. Available from: doi: https://doi.org/10.1101/2021.04.30.442205 Xue X, Sun Y, Resto-Irizarry A.M. et al. Mechanics-guided embryonic patterning of neuroectoderm tissue from human pluripotent stem cells. Nature Mater 17, 633–641 (2018). https://doi.org/10.1038/s41563-018-0082-9 ------------------------------------------------------------------------------------------------ Use and Access: This data set is made available under a Creative Commons Public Domain license (CC BY-NC 4.0). ------------------------------------------------------------------------------------------------ To Cite Data: Nunley, H., Xue, X., Fu, J., Lubensky, D. K. Dataset on cell areas and nuclear densities in differentiating stem cell colonies [Data set], University of Michigan - Deep Blue Data. https://doi.org/10.7302/0n9v-6397