PerspectiveDevelopmental Biology

How Many Ways to Make a Chordate?

See allHide authors and affiliations

Science  26 May 2006:
Vol. 312, Issue 5777, pp. 1145-1146
DOI: 10.1126/science.1128784

Ascidians are invertebrate chordates that belong to the tunicates, the closest living sister group of vertebrates (1), from which they diverged more than 500 million years ago. On page 1183 of this issue, Imai and co-workers (2) report how they analyzed the ascidian Ciona intestinalis (sea squirt) to generate the first metazoan whole-embryo gene regulatory network. Surprisingly, it appears quite different from vertebrate networks despite the conservation of a common, tadpole-like larval body plan (see the figure).

Gene regulatory networks consist of functional linkages between transcription factors, cell signaling components, and the cis-regulatory modules that control their expression at the transcriptional level (3). The action of such networks is a major force driving animal development, from a simple egg to a complex larva. These networks are proposed to be free standing—that is, the regulation of each network component can be accounted for by the presence of other components. Ultimately, the networks control the precise expression of differentiation genes that confer specific attributes to each embryonic cell. The identification of networks in echinoderms (sea urchins) (4) and vertebrates (5, 6) has illuminated the developmental logic underlying animal embryogenesis. In addition, comparing gene-regulatory networks that give rise to homologous anatomical structures across taxa or phyla have helped to elucidate the evolutionary origin of these structures (3).

Like lower vertebrates, developing Ciona make tadpole larvae. But these tadpoles have only 2600 cells, and the size of the Ciona genome is only 1/20 that of the mouse. In addition, ascidian genomes have not undergone the vertebrate-specific duplication events (7). The extreme genetic and cellular simplicity of Ciona is a boon to biologists attempting to reconstruct gene regulatory networks. It is, however, unlikely to reflect the condition of the ancestral chordate. Rather, most of it probably results from secondary simplification that has occurred since the tunicate and vertebrate lineages separated.

Thus, despite a shared larval body plan, the extent of conservation between ascidian and vertebrate gene regulatory networks has been uncertain. For instance, unlike in vertebrates, most ascidian tail muscle cells form cell autonomously, following the inheritance of the localized maternal Zic-family transcription factor Macho-1 (8). Notochord formation in ascidians and vertebrates involves the induced expression of the T-box transcription factor Brachyury by signals elicited by fibroblast growth factor (9). On the other hand, signals from bone morphogenetic protein play opposite roles in the formation of this tissue in ascidians and vertebrates (10). Neural tissue development is induced by fibroblast growth factor signals in ascidians and vertebrates, but it is unclear whether the transcription factors that act in the fibroblast growth factor signaling cascade are shared between ascidians and vertebrates (11). Finally, involvement of the transcription factors Mesp, Nkx, and HAND in heart formation appears to be conserved (12). These examples suggest a mixture of conservation and divergence between ascidian and vertebrate developmental strategies.

Look-alikes, early on.

Although ascidians such as Ciona intestinalis have a very peculiar adult body plan that is adapted to their marine filter-feeding life-styles, their larvae are very similar to frog tadpoles (such as Rana sylvatica), demonstrating a close common evolutionary history. Nevertheless, their genetic circuits are different.


The extent to which different components of a large gene regulatory network are evolutionarily conserved may, however, differ. It was proposed that the only parts of the network that are well conserved, the “kernels,” are highly connective subnetworks that specify the organization in space of different tissues (the body plan) rather than specifying precise cell types (3). The major function of kernels, and the reason for their high level of evolutionary conservation, would be to “lock” a given body plan in place (3). If this is true, chordate-specific kernels should be detected in ascidian gene regulatory networks despite the simplification of ascidian development.

Imai et al. provide the first whole-embryo gene regulatory network for a chordate, covering development from the 16-cell stage to the gastrula stage in Ciona. Previously, Imai and colleagues (13) generated a spatiotemporal atlas of gene expression of nearly 500 genes coding for transcription factors, signaling ligands, and receptors. They identified 65 genes encoding transcription factors and 26 genes encoding signaling molecules that are zygotically expressed up to the onset of gastrulation. In the present work, Imai et al. focus on those of these genes with no maternal expression and show that their expression patterns define an unambiguous transcriptional code for each blastomere up to the gastrula stage. This suggests that the combination of these genes is sufficient to give rise to the cellular diversity in the early embryo, and that they should therefore form a reasonably complete network. Imai et al. microinjected morpholino antisense oligonucleotides into Ciona eggs to inhibit the expression of most of these genes individually. Using quantitative polymerase chain reaction and in situ hybridization, they assessed the expression of all candidate genes (nodes of the network) in response to each perturbation. Only 27 out of 70 (39%) morpholinos tested produced a specific phenotype. This relatively low percentage reflects either imperfect morpholino design or the existence of genetic redundancy. The network constructed from these data is far from complete because the regulation of many genes cannot be accounted for by the presence of other genes from the collection. In particular, the network does not address how the expression of early zygotic genes is turned on by maternal factors. Finally, it does not include cis-regulatory analysis, so that the links established may be direct or indirect. Yet, it provides the first bird's-eye view of the regulatory circuitry that creates an early metazoan gastrula, at which stage most cell fates are restricted.

A first surprise is the low level of connectivity of the network in each embryonic territory. In contrast to the situation for the echinoderms and vertebrate networks, only a small minority of signaling molecules (fibroblast growth factor, Nodal) and transcription factors (ZicL, FoxD, FoxA-a, Otx) affect the expression of a large fraction of the regulatory genes assayed. It will be interesting to test whether the majority of genes studied, which have few or no targets in the network, directly control differentiation genes. This would suggest that the simple and rapidly developing Ciona embryo may not need the cross-regulatory interactions used to stabilize gene expression patterns found in more slowly developing and also more complex embryos. Because extensive cross-regulatory interactions are a feature of kernels, the low level of connectivity of the Ciona network suggests their absence at the stages analyzed. Another surprise is the prevalent—though probably indirect—use of negative autoregulatory loops in the network. In contrast, very few positive autoregulatory loops were identified. This conflicts with the proposal that cell fate determination, which occurs very early in Ciona, is associated with the establishment of positive regulatory loops that lock in a given fate (4).

The structure of the current early Ciona network differs substantially from those of other deuterostomes. The maintenance of a chordate body plan in ascidians, in the absence of detectable kernels, may cast doubt upon the proposal that this type of subnetwork is important to stabilize a body plan across large evolutionary distances. It should, however, be considered that within a phylum, developmental strategies can be diverse in early embryos, converge at the phylotypic stage, and diverge again when terminally differentiated structures form. The network analyzed in the present work mainly covers pregastrula stages, and forms a necessary first step toward reconstructing networks for later stages in which chordate-specific kernels may be present. As it stands, the Ciona network has already allowed researchers to identify novel key regulators of specific fates and illustrates that whole-embryo reconstruction of gene regulatory networks is feasible provided that a suitable model organism is chosen. As was noted a few years ago, “Ascidians are back in the limelight, with a good chance of staying there” (14).


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
View Abstract

Navigate This Article