Machine learning directed organoid morphogenesis uncovers an excitable system driving human axial elongation

The human embryo breaks symmetry to form the anterior-posterior axis of the body. As the embryo elongates along this axis, progenitors in the tailbud give rise to axial tissues that generate the spinal cord, skeleton, and musculature. The mechanisms underlying human axial elongation are unknown. While ethics necessitate in vitro studies, the variability of human organoid systems has hindered mechanistic insights. Here we developed a bioengineering and machine learning framework that optimizes symmetry breaking by tuning the spatial coupling between human pluripotent stem cell-derived organoids. This framework enabled the reproducible generation of hundreds of axially elongating organoids, each possessing a tailbud and an epithelial neural tube with a single lumen. We discovered that an excitable system composed of WNT and FGF signaling drives axial elongation through the induction of a signaling center in the form of neuromesodermal progenitor (NMP)-like cells. The ability of NMP-like cells to function as a signaling center and drive elongation is independent of their potency to generate mesodermal cell types. We further discovered that the instability of the underlying excitable system is suppressed by secreted WNT inhibitors of the secreted frizzled-related protein (SFRP) family. Absence of these inhibitors led to the formation of ectopic tailbuds and branches. Our results identify mechanisms governing stable human axial elongation to achieve robust morphogenesis.