During early embryonic development in vertebrates, a round clump of cells reshapes into an elongated structure with the future head at the anterior (A) end and the rump or tail at the opposite, posterior (P) end. This process, which is known as convergent extension, involves the narrowing and lengthening of a group of cells, the chordamesoderm cells, parallel to the A–P axis. A report in Nature now sheds light on how the cell movements that underlie convergent extension are aligned with the A–P axis in Xenopus laevis embryos.

By dissociating chordamesoderm cells, mixing them up and letting them reaggregate, Ninomiya et al. found that the cells rearranged themselves according to their original A or P position. In support of this intrinsic chordamesodermal A–P pattern, the X. laevis brachyury (Xbra) and chordin (chd) genes show a typical reverse expression pattern, with low A and high P Xbra expression, and the opposite pattern for chd.

So, what's the relevance of chordamesodermal A–P patterning? To probe this problem, the authors dissociated chordamesoderm cells from the A and P regions, reaggregated them separately, and then combined these aggregates. Combinations of identical aggregates remained round, whereas combined A and P aggregates became elongated along the A–P axis. A–P polarity therefore seems to be required for convergent extension.

At certain concentrations, the signalling molecule activin induces chd expression and reduces that of Xbra. Treating explants in vitro with graded doses of activin caused them to elongate, whereas the uniform exposure to high or low doses of activin meant that the explants remained spherical. Expression of chd and Xbra was more evenly distributed in uniformly exposed explants when compared with the graded explants, which showed an Xbra–chd expression gradient. So, graded activin signalling can establish A–P polarity and trigger convergent extension. According to the authors, a different signalling molecule, Nodal, rather than activin, is likely to be responsible for controlling A–P patterning in vivo.

In a final set of experiments, the impairment of the Wnt/planar cell polarity (PCP) pathway prevented the elongation of explants, but did not disturb the graded gene-expression pattern. So, Wnt/PCP signalling seems to control convergent extension independently of A–P polarity, which indicates that the two pathways probably function in parallel.

Together, these new findings pave the way for investigating the specific cellular properties that make polarized chordamesoderm cells intercalate, and converge and extend along the A–P axis, thereby separating head from tail.