Positional information revisited.

Positional information has been suggested to play a central role in pattern formation during development. The strong version of positional information states that there is a cell parameter, positional value, which is related to position as in a coordinate system and which determines cell differentiation. A weaker version merely emphasises position as a key determinant in cell development and differentiation. There is evidence for boundaries and orthogonal axes playing an important role in positional systems. A positional signal is distinguished from an inductive interaction because the former specifies multiple states, confers polarity, and can act over a long range. A gradient in a diffusible morphogen is just one way of specifying position. There is now good evidence in several systems for substances which may be the morphogen for positional signalling. The product of the bicoid gene in early Drosophila development is the best prospect. Retinoic acid is unique in its ability to alter positional value and may also be a morphogen. The best evidence for positional value, a concept fundamental to positional information, remains a biological assay based on grafting. The idea of positional value uncouples differentiation and position, and allows considerable freedom for patterning. It is not clear whether positional value or differentiation involves a combinatorial mechanism. Interpretation of positional information remains a central problem. There is good evidence that cells can respond differentially to less than a two-fold change in concentration of a chemical signal. It may be that interpretation involves listing the sites at which a particular class of cell differentiation will occur. The problem is made less severe when blocks of cells are specified together as in mechanisms based on an isomorphic prepattern. Isomorphic prepatterns could establish repeated structures which are equivalent and which are then made non-equivalent by positional information. This would enable local differences to develop. The combination of these two mechanisms may be wide-spread. There is evidence that positional signals within a single animal and in related animals are conserved. It is not clear just how wide this conservation is, but it is at phylotypic stages, rather than in eggs, that similarity might be expected. It is nevertheless impressive that the polar coordinate model can be applied to regulation in systems as diverse as insects, vertebrates and protozoa. The molecular basis of positional signalling is just becoming accessible; the molecular basis of positional value is still awaited. A brief personal history of positional information is provided in an appendix.