Convergent extension by intercalation without mediolaterally fixed cell motion.

We construct and implement a stochastic model of convergent extension, using a minimal set of assumptions on cell behavior. In addition to the basic assumptions of volume conservation, random cell motion, and cell-cell and cell-ECM adhesion, and a non-standard assumption that cytoskeletal polymerization generates an internal pressure tending to keep cells convex, we find that we need only two conditions for convergent extension. (1) Each cell type has a particular aspect ratio towards which it regulates its geometry. We do not require that cells align in a specific orientation, e.g. to be oriented mediolaterally. (2) The elongating tissue is composed of cells that prefer to be elongated, and these cells must be accompanied by cells which prefer to be round. The latter effectively provide a boundary to capture. In simulations, our model tissue extends and converges to a stacked arrangement of elongated cells one cell wide, an arrangement which is seen in ascidian notochords, but which has not been observed in other models. This arrangement is achieved without any direct mediolateral bias other than that which is provided by the physical edge of the adjacent tissue.