Cells with strong β-catenin signaling (red) are sorted out of the notochord with other somitic cells (green nuclear staining).

Cells of a feather stick together, but not because of common adhesion levels, new evidence suggests. The findings from Reintsch et al. (page 675) put into question an old theory of adhesion-based cell sorting.

Cells sort, this theory proposes, according to their adhesiveness. In a mixed population, stickier cells aggregate in the middle, and less adhesive cells are driven to the periphery. Adhesion-based differences are thus used to explain why cells do not cross boundaries separating two tissues.

Such boundaries exist between the notochord and somite in the developing frog embryo. This boundary depends on β-catenin, which induces somite formation and inhibits notochord development. β-Catenin connects the the cytoskeleton to adhesion protein cadherin, suggesting that the old theory might be at play. But the new results reveal that sorting is not affected by the strength of cadherin-mediated adhesion, although cells still sort according to β-catenin signaling activity.During the widespread migration of notochord and somite cells that narrows and lengthens their domain, cells overexpressing β-catenin were sorted away from the notochord side of the boundary. When these β-catenin–expressing cells touched the boundary coming from the notochord side, they were quickly incorporated into the somite side.

The signaling effects downstream of β-catenin leading to LEF1 activation were at the base of this sorting behavior. Strong LEF1 activity sorted cells to somites, whereas the inhibition of β-catenin signaling sent cells to the notochord. Adhesion was not necessary; interference with cadherins did not disturb sorting.

The authors believe that each cell type might express a particular sensor that distinguishes self and somehow repels nonself cells. When on the wrong side of the boundary, cells that expressed β-catenin were misshapen, extending unstable protrusions and failing to make appropriate contacts with their neighbors. This behavior would be expected to favor maintenance of cells on the correct side of any boundary, where they could grip other cells securely. It is not clear what these self-sensors are, but transmembrane signaling proteins such as ephrins or protocadherins might fit the bill.