Abstract

AVE cells show a stereotypic unidirectional migration essential for correct orientation of the anterior-posterior axis. They migrate within a simple epithelium, but it is unknown how they negotiate their way amongst the surrounding Visceral Endoderm (VE) cells while it remains a monolayer and retains epithelial integrity. It is unclear what the relative contributions of cell shape changes, regional differences in proliferation rates, oriented division etc. are to such migration. To address these questions, we used lightsheet microscopy to generate a multi-embryo, single-cell resolution, longitudinal dataset of cell behaviour. We developed a machine learning based computational pipeline to segment cells and extract morphological and behavioural parameters of AVE and surrounding VE cells. Unbiased clustering of this single-cell ‘phenomic’ dataset reveals considerable patterned phenotypic heterogeneity within the VE and AVE and a previously unknown behavioural sub-grouping within the AVE . It reveals that while migrating, AVE cells remain relatively constant in morphology, do not exchange neighbours and appear crowded, with a constant relatively low apical surface area. In contrast, VE cells ahead of them become highly elongated, undergo neighbour exchange and show bilateral polonaise-like rotational movements. The dataset shows that AVE cells are also characterised by higher levels of apical and junctional F-actin, suggesting they may be mechanical distinct from surrounding VE cells, which we verify in mouse embryos using a live tension sensor probe and Fluorescence Lifetime imaging. These data lead us to propose a model whereby AVE migration is facilitated by an unjamming transition of surrounding VE cells, while AVE cells themselves remain in a jammed state throughout migration.