Problems and paradigms: Is segmentation generic?

When two populations of cells within a tissue mass differ from one another in magnitude or type of intercellular adhesions, a boundary can form within the tissue, across which cells will fail to mix. This phenomenon may occur regardless of the identity of the molecules that mediate cell adhesion. If, in addition, a choice between the two adhesive states is regulated by a molecule the concentration of which is periodic in space, or in time, then alternating bands of non-mixing tissue, or segments, can form. But temporal or spatial periodicities in concentration will tend to arise for any molecule that is positively autoregu-latory. It is therefore proposed that segmentation is a ‘generic’ property of metazoan organisms, and that metamerism would be expected to have emerged numerous times during evolution. A simple model of segmentation, based solely on differential adhesion and periodic regulation of adhesion, can account for segment properties as disparate as those seen in long and short germ band insects, and for diverse experimental results on boundary regeneration in the chick hind brain and the insect cuticle. It is suggested that the complex, multicom-ponent segment-forming systems found in contemporary organisms (e.g., Drosophila) are the products of evolutionary recruitment of molecular cues such as homeobox gene products, that increase the reliability and stability of metameric patterns originally templated by generic self-organizing properties of tissues.