Journal: Molecular Biology of the Cell

Article Title: Local cellular neighborhood controls proliferation in cell competition

doi: 10.1091/mbc.E17-06-0368

Figure Lengend Snippet: High-throughput image analysis pipeline for long-term analysis of single-cell dynamics. (A) Representative image containing H2b-GFP MDCK WT (green) and H2b-RFP scribble kd (magenta) cells mixed in an initial 90:10 ratio. The image corresponds to one field of view (1600 × 1200 pixels, 530 µm × 400 µm) acquired by wide-field epifluorescence using 20× magnification. The region of interest in the white rectangle is shown in B. (B) Time series from the competition assay in the region boxed in A. The white arrow indicates a scribble kd cell surrounded by MDCK WT neighbors that undergoes apoptosis. The acquisition of both transmission and fluorescence images enables detection of the apoptotic fragmentation of the loser’s nucleus happening before extrusion of the cell body from the monolayer. Timings are indicated in the top left corner in hours and minutes. (C) Segmentation of the final image in B. MDCK WT cells are outlined in green and scribble kd cells in magenta. (D) Flowchart of the computational pipeline implemented for the study of competition dynamics. The strategy is based on segmentation of individual cells (cell detector), automatic annotation of morphological classes related to cell cycle state and apoptosis (track compiler), and postprocessing analysis of single-cell tracking data. (E) CNN for object classification. The CNN inputs are single-object patches, both in the transmission (BF) and fluorescence channels (left). The CNN stacks together four types of layers: convolutional/ReLU/max-pooling, and fully connected layers (middle). The CNN transforms the input image layer by layer from the original pixel values to the final class scores with the highest score reflecting the most probable classification of the image data (right). (F) Confusion matrix showing the matching of human annotations vs. the annotation of the CNN system. (G) HMM used for modeling progression through the cell cycle. The figure depicts the permitted directional transitions between five classes (interphase, prometaphase, metaphase, anaphase, and apoptosis). (H) Automated annotation of cell trajectories over time. A random selection of 100 trajectories (rows) is aligned and shown over a 40-min period. Colors refer to state labels as defined in F. Left, tracks following division start with anaphase before proceeding to interphase. Middle, tracks terminating in a division are preceded by interphase before going through prometaphase and metaphase. Right, tracks terminating with apoptosis are often preceded by interphase but can arise through failed division events (highlighted with arrows).

Article Snippet: A custom-built automated epifluorescence microscope was built inside a standard CO 2 incubator (Thermo Scientific Heraeus BL20) that maintained the temperature at 37°C and in a 5% CO 2 atmosphere.

Techniques: High Throughput Screening Assay, Competitive Binding Assay, Transmission Assay, Fluorescence, Single Cell Tracking, Selection