Mechanism of anteroposterior axis specification in vertebrates. Lessons from the amphibians

Interest in the problem of anteroposterior specification has quickened because of our near understanding of the mechanism in Drosophila and because of the homology of Antennapedia-like homeobox gene expression patterns in Drosophila and vertebrates. But vertebrates differ from Drosophila because of morphogenetic movements and interactions between tissue layers, both intimately associated with anteroposterior specification. The purpose of this article is to review classical findings and to enquire how far these have been confirmed, refuted or extended by modern work. The "pre-molecular" work suggests that there are several steps to the process: (i) Formation of anteroposterior pattern in mesoderm during gastrulation with posterior dominance. (ii) Regional specific induction of ectoderm to form neural plate. (iii) Reciprocal interactions from neural plate to mesoderm. (iv) Interactions within neural plate with posterior dominance. Unfortunately, almost all the observable markers are in the CNS rather than in the mesoderm where the initial specification is thought to occur. This has meant that the specification of the mesoderm has been assayed indirectly by transplantation methods such as the Einsteckung. New molecular markers now supplement morphological ones but they are still mainly in the CNS and not the mesoderm. A particular interest attaches to the genes of the Antp-like HOX clusters since these may not only be markers but actual coding factors for anteroposterior levels. We have a new understanding of mesoderm induction based on the discovery of activins and fibroblast growth factors (FGFs) as candidate inducing factors. These factors have later consequences for anteroposterior pattern with activin tending to induce anterior, and FGF posterior structures. Recent work on neural induction has implicated cAMP and protein kinase C (PKC) as elements of the signal transduction pathway and has provided new evidence for the importance of tangential neural induction. The regional specificity of neural induction has been reinvestigated using molecular markers and provides conclusions rather similar to the classical work. Defects in the axial pattern may be produced by retinoic acid but it remains unclear whether its effects are truly coordinate ones or are concentrated in certain regions of high sensitivity. In general the molecular studies have supported and reinforced the "pre-molecular ones". Important questions still remain: (i) How much pattern is there in the mesoderm (how many states?) (ii) How is this pattern generated by the invaginating organizer? (iii) Is there one-to-one transmission of codings to the neural plate? (iv) What is the nature of the interactions within the neural plate? (v) Are the HOX cluster genes really the anteroposterior codings?