Developing organisms still hold many surprises for biologists. For instance, an entirely new 'organizer' — a clump of cells that tells other cells what to do — has been discovered at a very early stage of zebrafish development.
The vertebrate tail is a somewhat neglected organ. Protruding from the rear, it does not, at first glance, seem to serve much purpose other than to chase off flies on cows and horses. In humans, the rare atavistic occurrence of a rudimentary tail can even be an embarrassment. Yet the post-anal tail is one of the defining features of chordates. In primitive chordates and fish it promotes locomotion; lancelets use it to burrow in sand; in mammals it stabilizes posture and allows communication; and monkeys also use it as a third leg. All vertebrates develop a tail as embryos, and where the adults do without, it disappears during fetal development or metamorphosis. Despite its importance, however, we know little about the molecular mechanisms that underlie tail formation. On page 448 of this issue, Agathon et al.1 present an analysis of tail development in zebrafish. Unexpectedly, they have discovered a new 'tail-organizing centre' in early embryos, and have shown that its molecular components include some of the usual suspects.
The question of how the embryonic tail forms is part of the greater problem of how the main body axis — from head to trunk to tail — is generated during early vertebrate development. Of central importance for the establishment of this body axis in amphibians is the upper dorsal blastopore lip, better known as the Spemann organizer, which can, after being transplanted into a different embryo, cause a second body axis and hence a twin embryo to form. Structures equivalent to the amphibian organizer have been found in all other vertebrates. The organizer produces signals that control the specialization of nearby tissues; two of its main functions are to induce the production and patterning of neural tissue, and to cause ventral tissues of the middle embryonic layer known as the mesoderm to form dorsal tissues (such as muscle)2.
Early transplantation experiments by Otto Mangold3 showed that the Spemann organizer can be subdivided into head, trunk and tail organizers (Fig. 1a). So it comes as a complete surprise that Agathon et al.1 have discovered a tail-organizing centre that appears to be independent of the Spemann organizer. The authors carried out experiments with zebrafish blastulae, early embryos that consist of just a hollow ball of cells. They took cells from the ventral margin — a tissue that does not become part of the 'shield', considered to be the fish equivalent of the Spemann organizer — and transplanted the cells into host embryos. The result was extra tails, consisting of donor- as well as host-derived cells. Agathon et al. also showed that inactivating the Spemann organizer did not interfere with this 'ectopic' tail formation, implying that the two organizers are indeed independent. The authors concede, however, that the Spemann organizer is required for 'proper' tail formation, as the ectopic tails were incomplete: for instance, they lacked the stiff supporting rod known as the notochord, which is derived from the Spemann organizer.
The authors also established that this tail-organizer activity involved growth-factor proteins of the Wnt, bone morphogenetic protein (BMP) and Nodal families. These signalling molecules were good candidates for such a role because they are known to be present in the ventral margin, and interfering with the signalling pathways to which they contribute inhibits tail formation. Agathon et al. now show that misexpressing these growth factors in combinations, but not singly, leads to the generation of extra tails — which are, again, incomplete. So all three growth factors are involved in the newly discovered tail-organizer activity.
The wider significance of these findings concerns how they relate to existing models for region-specific organizers, in which Wnts, BMPs and Nodals are no strangers. The head- and trunk-organizing activities of the Spemann organizer are known to require combinations of these three types of growth factor4. Head-organizer activity requires inhibition of all three signals; trunk formation requires Nodal signalling to be active but BMP signalling to be inhibited; and we now know that the tail organizer requires signalling by all three factors. The drawback of previous organizer models was that they failed to explain the difference between the trunk and tail organizers in molecular terms, except for tail development at later stages5,6. Thanks to Agathon et al., this riddle seems well on the way to being solved.
A few issues remain, however. First, Agathon and colleagues' model1 emphasizes qualitative differences in combinatorial signalling, with key signals being either 'on' or 'off' — and this is enough to explain the phenomenon of discrete organizers as they have been observed in transplantation experiments for the past 75 years. For instance, the authors propose that Wnt signalling needs to be 'on' to generate the tail, but 'off' to produce the trunk. Yet we know that all three growth factors act in a concentration-dependent fashion during early axis formation. A gradient of Nodal protein instructs cells at different points along the gradient to take on different mesodermal fates; this then translates into an increasing dorsal-to-ventral gradient of BMP7,8 and an increasing head-to-tail gradient of Wnt9, generating a continuum of positional information. Completely blocking Wnt inhibits trunk formation, so trunk-organizer activity must require some — probably low levels of — Wnt signalling, rather than the complete inhibition suggested by the authors.
Second, unlike BMPs and Wnts, Nodal proteins affect the head-to-tail patterning of neural tissue only indirectly, inducing mesendodermal tissue to produce, for instance, Wnts, Wnt inhibitors and BMP inhibitors at different threshold activities of Nodal. A model complementary to that of Agathon et al. focuses on the more direct players — BMP and Wnt — in regulating axis formation (Fig. 1b), and takes into account their activity gradients.
Finally, how does the tail organizer uncovered by Agathon et al. interact with the Spemann organizer to induce complete tails? And how can we integrate the process of trunk 'segmentation', which is controlled from the tail bud? Whatever the answers, the new work1 should generate greater interest in questions relating to the tail.
Agathon, A., Thisse, C. & Thisse, B. Nature 424, 448–452 (2003).
Harland, R. M. & Gerhart, J. Annu. Rev. Dev. Biol. 13, 611–667 (1997).
Mangold, O. Naturwissenschaften 21, 761–766 (1933).
Piccolo, S. et al. Nature 397, 707–710 (1999).
Beck, C. W. & Slack, J. M. Development 126, 1611–1620 (1999).
Beck, C. W., Whitman, M. & Slack, J. M. Dev. Biol. 238, 303–314 (2001).
De Robertis, E. M., Larrain, J., Oelgeschlager, M. & Wessely, O. Nature Rev. Genet. 1, 171–181 (2000).
Schier, A. F. Curr. Opin. Genet. Dev. 11, 393–404 (2001).
Kiecker, C. & Niehrs, C. Development 128, 4189–4201 (2001).
Gilbert, S. F. Developmental Biology 7th edn (Sinauer, Sunderland, Massachusetts, 2000).
Gilbert, S. F. & Saxen, L. Mech. Dev. 41, 73–89 (1993).
About this article
Seminars in Cell & Developmental Biology (2004)