Intercellular Coordination in Epithelial Morphogenesis

During the development of an embryo, tissue rearrangements are driven by the precisely coordinated activity of thousands of cells. How this high degree of organization is achieved and regulated is one of the most important questions in developmental biology. Genetics provides unidirectional information that determines cell fate and triggers downstream biochemical processes but is not flexible enough to interact on minute-to-minute time scales. Mechanical forces, on the other hand, constitute signals that travel fastest in an epithelial tissue and therefore can provide information for intercellular coordination and dynamic regulation on the fly. Studies that focus on mechano-signaling in morphogenesis, however, are still scarce. In this thesis, I investigate intercellular coordination on both the theoretical and experimental level and provide novel techniques for studying epithelial dynamics during morphogenesis. Two prototypical processes in Drosophila morphogenesis are Dorsal Closure in the amnioserosa and germband convergent extension, both being substantially driven by the coordinated intrinsic activity of individual cells. One key result of my work is that synchronization of cell activity in the amnioserosa is mediated by mechano-sensitive Tmc channels. A theoretical model of the amnioserosa combined with large scale analysis of experimental data shows that these ion channels transduce mechanical signals into intracellular response variables in neighboring cells. Furthermore, the synchronization of cell activity is directly linked to establishing the morphology and isotropic distribution of tension in the tissue. In the germband, convergent extension is driven by local neighbor exchanges, which are called T1 transitions. Cells in T1 quadruplets synchronize their contractions in a time-dependent manner that is lost in xit embryos, a gene that affects the E-Cadherin distribution at adherens junctions. Loss of synchronization in xit embryos further is associated with longer T1 transitions and higher failure rate, establishing a role of E-Cadherin for signal transduction during cell intercalation. On the technical level, two new methods are presented in this dissertation. First, an experimental technique for optochemically triggering contractions of epithelial cells in vivo that does not rely on transgenes and therefore can be easily and widely applied. Using this method, mechanical stimuli can be presented to the tissue with high temporal and single cell precision. Secondly, I developed a novel image segmentation pipeline based on deep neural networks. Using CycleGANs, training can be performed without paired ground truth data. Rule-based methods are substantially outperformed and the high accuracy enables complete segmentation of an ensemble of embryos in feasible time.2022-08-0