ES cells, derived from the pluripotent epiblast, can host transgenes and be reintroduced back into the embryo to generate a chimeric animal and a pure breeding line in future generations. A stunning application of ES cells in recent years has been their use to generate embryo-like structures in vitro. Several approaches have advanced our quest to recapitulate embryogenesis.

A 2D method using exclusively ES cells cultured as micropatterns offered a powerful route toward understanding how different cell types are established and signal between themselves. A second model, in which large aggregates of ES cells are treated with chemicals and growth factors, generated 3D structures developing many aspects of the segmental body plan, although still lacking body regions, particularly those required for forebrain development.

The importance of extraembryonic signalling was recognized through a series of whole-embryo models. The first such model, built from ES cells alone, pointed to the role of signals normally provided by the extraembryonic primitive endoderm, which can be replaced by the extracellular matrix to polarize ES cells to form a rosette-like structure that undertakes lumenogenesis. The second model, built from ES cells and trophectoderm stem cells, taught us that this interaction alone is sufficient to establish amniotic cavity and posterior embryo identity to induce mesoderm and germ cells. By incorporating a third stem cell type, extraembryonic endoderm cells, we achieved the formation of the anterior signalling centre and anterior–posterior patterning. Recently, additional approaches we and others undertook led to the generation of embryo models that were capable of developing much further to establish brain and heart structures and initiate organogenesis. Such whole-embryo-like models have brought insight into the biophysical and biochemical factors mediating stem cell self-organization and defining the cellular constituents, the chemical environment and the physical context required for embryo assembly.

Despite this progress, challenges remain. Cell fate specification relies on chemical cross-talk within and between lineages. Cell fate decisions must be spatiotemporally coordinated by establishing and interpreting gradients of numerous diffusible signalling proteins. We have much to learn about these combinatorial effects and about how to improve the efficiency with which different cell types combine to form embryo-like structures. A deep understanding of the components of cellular, biochemical and biophysical networks will be crucial to reaching this goal. Computational modelling will allow us to predict and guide self-organizational outcomes through exploitation of the capacity of cell communication to promote self-organization in vivo. It would also be powerful to advance our abilities to culture model embryos and replicate the maternal environment by delivering suitable nutrients to the circulatory system of the developing structure. These problems are also inherent to the assembly of synthetic organs, and I am certain that we will see a cross-talk between these different disciplines of synthetic biology for mutual benefit.