A multiscale model via single-cell transcriptomics reveals robust patterning mechanisms during early mammalian embryo development
Fig 2
Data-informed selective adhesion model leads to successful cell arrangement at 128 cell stage.
(a) Histogram of z-score of summation of log([EphA4] + 1) and log([EphrinB2] + 1) at E4.5 stage. The green curve shows the distribution of z-score. The red and blue curves show two components of Gaussian Mixture to fit the distribution. (b) Percentage of high/low adhesive gene expression levels in Nanog+/Gata6+/DP cells at E4.5 stage. (c) An EphA4/EphrinB2 driven selective adhesion mechanism between Nanog+/Gata6+/DP cells, where higher adhesion score (AS) indicates stronger adhesion, and a positive AS means strengthened adhesion. (d) Success rate for Nanog+/Gata6+ cells arrangement in simulations with different selective adhesion hypotheses: (H1) no selective adhesion; (H2) symmetric selective adhesion where adhesion between Nanog+/Nanog+, Gata6+/Gata6+ and DP/DP cells are the same; (H3) asymmetric selective adhesion where DP cells have same adhesion with Nanog+ and with Gata6+; (H4) asymmetric selective adhesion where DP cells have stronger adhesion with Nanog+ cells than with Gata6+ cells; (H5) asymmetric selective adhesion where DP cells have stronger adhesion with both Nanog+ and Gata6+ cells; (H6) the EphA4/EphrinB2 driven selective adhesion; (H7) the EphB2/EphrinB2 driven selective adhesion. (e) Pattern loss score of the simulations. Each data point corresponds to one simulation. A loss score of 0 indicates a perfect pattern and random cell type assignments have an expected loss score of 1. (f) Representative terminal Nanog+/Gata6+ cell arrangements for successful, partially successful, and failed cases.