The prospects of deciphering the genetic program underlying embryonic development were recently boosted by the generation of large sets of precisely organized quantitative molecular data. In contrast, although the precise arrangement, interactions, and shapes of cells are crucial for the fulfilment of this program, their description remains coarse and qualitative. To bridge this gap, we developed a generic software, 3D Virtual Embryo, to quantify the geometry and interactions of cells in interactive three-dimensional embryo models. We applied this approach to early ascidian embryos, chosen because of their simplicity and their phylogenetic proximity to vertebrates.

We generated a collection of 19 interactive ascidian embryos between the 2- and 44-cell stages. We characterized the evolution with time, and in different cell lineages, of the volume of cells and of eight mathematical descriptors of their geometry, and we measured the surface of contact between neighboring blastomeres. These analyses first revealed that early embryonic blastomeres adopt a surprising variety of shapes, which appeared to be under strict and dynamic developmental control. Second, we found novel asymmetric cell divisions in the posterior vegetal lineages, which gave birth to sister cells with different fates. Third, during neural induction, differences in the area of contact between individual competent animal cells and inducing vegetal blastomeres appeared important to select the induced cells.

In addition to novel insight into both cell-autonomous and inductive processes controlling early ascidian development, we establish a generic conceptual framework for the quantitative analysis of embryo geometry that can be applied to other model organisms.