Predictive Multicellular Assembly Design

During development, cells have the remarkable ability to self-organize into complex spatial arrangements, ultimately creating the body plan of a whole organism. Cell-cell adhesion molecules like cadherins play a critical role in sorting cell types into the correct tissue configuration. Harnessing the principles of adhesion to predictively design multicellular structures would thus allow us to engineer custom tissues and materials. In this work, we physically characterize a set of adhesion molecules, generate cell types with a variety of defined adhesion profiles, and predict their self-organization through differentiable simulations. We can then invert this model to design the cell types needed to achieve a target multicellular structure. The design of such systems holds significant promise in being able to build complex tissues like organoids or even organizer structures that can precisely direct developmental programs.