Molecular studies of tunicate development show that genetic programmes for early embryonic patterning can change radically during evolution, without completely disrupting the basic chordate body plan.
The tunicates are our closest invertebrate relatives, being members, along with us and all other vertebrates, of the phylum Chordata. There are three groups of tunicates, and investigations of a member of one of them, the appendicularian Oikopleura dioica, has revealed an instructive anomaly among chordates. As they report in Developmental Biology, Cañestro and Postlethwait1 find that although Oikopleura (Fig. 1) has retained the fundamental chordate body plan, it has lost the mechanism of retinoic-acid signalling that otherwise operates during chordate development. That loss raises the question of the evolutionary constraints that have prevented similar changes in the other chordates — the vertebrates and amphioxus, another invertebrate.
The currently understood phylogenetic context of this work is shown in Figure 2, with the three groups of chordates — the tunicates (ascidians, thaliaceans and appendicularians), the vertebrates and the cephalochordates (amphioxus, or lancelets) — being at the centre of the story. Although tunicates were long considered as the earliest offshoot of the chordate lineage, and amphioxus as the closest group to vertebrates2,3, recent analyses have reversed their positions. Amphioxus is now viewed as the most 'basal' chordate4, and tunicates as the sister group, or closest relatives, of the vertebrates, with the appendicularians (sometimes called larvaceans) as 'basal' among them5. The simplified body plans of ascidian larvae and appendicularians, and their small genomes (about 180 megabases and 72 megabases, respectively), are now recognized as evolutionary reductions from a more complex ancestor similar to amphioxus with its 500-megabase genome.
Retinoic acid is a derivative of vitamin A, and its function in specifying position along the embryonic anterior/posterior (A/P) axis is a chordate innovation. It acts most notably through the Hox set of genes, which control A/P patterning in early development. In both amphioxus and vertebrates, retinoic-acid signalling regulates the anterior limits of Hox expression in the embryonic central nervous system (CNS) and certain other tissues. Retinoic acid acts by binding to heterodimers of the retinoic-acid receptor (RAR) and the retinoid X receptor (RXR), which in turn bind to retinoic-acid response elements (RAREs) in the regulatory regions of direct targets (including Hox genes), thereby activating gene transcription. Control of the levels of retinoic acid is exercised by a suite of proteins including retinaldehyde dehydrogenase (Aldh1a, which catalyses the conversion of retinaldehyde to retinoic acid), and another enzyme, Cyp26 (retinoic acid hydroxylase, which inactivates retinoic acid).
In vertebrates and amphioxus, excess retinoic acid severely perturbs embryonic development. In contrast, Cañestro and Postlethwait found that the same treatment has no apparent effect on the A/P patterning of O. dioica. The head/trunk is foreshortened. But the A/P extent of the activity of a β-galactosidase-like gut enzyme, and of expression of Hox-1 and other developmental genes (three Otx genes and two Pax-2/5/8 genes), is unaffected. This lack of effect is not altogether surprising, because O. dioica has lost Aldh1a, Cyp26 and RAR, as well as several Hox genes6,7. Moreover, unlike Hox genes in amphioxus and vertebrates, the remaining O. dioica Hox genes are not clustered together in the genome, and their expression along the embryonic A/P axis is only approximately co-linear.
Cañestro and Postlethwait make the point that although ascidians have retained genes for Aldh1a, Cyp26 and RAR, the ancestral-chordate mechanism of retinoic-acid signalling has undergone alteration in these tunicates as well. Although excess retinoic acid causes a foreshortened head/trunk in ascidians, and induces misplaced expression of Hox-1, RAR is not autoregulated as it is in amphioxus and vertebrates, and no RAREs have been found in ascidian Hox-1. As in O. dioica, Hox clustering is also disrupted in ascidians. Consequently, the authors argue that a breakdown of Hox clustering may be causally related to evolutionary changes in retinoic-acid signalling in all tunicates.
More broadly, however, the significance of their findings is twofold. First, the results put into relief a paradox — although tunicates are now generally agreed to be the sister group of vertebrates, their exceptionally rapid evolution has effaced much information that might have been useful for suggesting how the vertebrates evolved from chordate ancestors. Instead, they are excellent for understanding what evolution can do.
Second, the results raise a question. Why, if genes can be lost and developmental programmes greatly changed without loss of the fundamental chordate body plan, have amphioxus and vertebrates retained their early developmental programmes over half-a-billion years of evolution? The answer may lie in constraints imposed by the mode of early development.
Tunicates have evolved 'determinate cleavage': the fate of cells is set early in embryonic development, with reduced cell numbers (for example, the CNS of O. dioica has only about 100 cells and that of an ascidian larva about 330), and their genomes are evolving rapidly. In contrast, in amphioxus and vertebrates, in which the retinoic-acid-sensitive period for A/P patterning occurs relatively late in development, cleavage is indeterminate: cell fates are decided late, there are many more cells (an estimated 20,000 neurons alone in the amphioxus CNS), and genome evolution is relatively slow.
To date, there have been few studies of the possible relations between the timing of cell-fate decisions in development and rates of genome evolution. In the nematode worm Caenorhabditis (early decision of cell fate) there is more selection against duplicates of genes expressed very early in development than against those expressed late, suggesting that constraints on genome evolution are greater early in development8. On the other hand, in the fruitfly Drosophila, there is relatively little difference in selection against duplicates of early and late developmental genes9.
Caenorhabditis and Drosophila have very different body plans, however, making comparisons difficult. In contrast, tunicates in general — and Oikopleura in particular, as the work by Cañestro and Postlethwait1 shows — lend themselves to comparative experiments with amphioxus and vertebrates. Such experiments can address the relationship of developmental constraints and rates of genome evolution against a background of conservation of the fundamental chordate body plan.
Cañestro, C. & Postlethwait, J. H. Dev. Biol. 305, 522–538 (2007).
Garstang, W. Q. J. Microsc. Sci. 72, 51–187 (1928).
Whittaker, J. R. Am. Zool. 37, 237–249 (1997).
Bourlat, S. J. et al. Nature 444, 85–88 (2006).
Swalla, B. J., Cameron, C. B., Corley, L. S. & Garey, J. R. Syst. Biol. 49, 52–64 (2000).
Cañestro, C., Postlethwait, J. H., Gonzà lez-Durate, R. & Albalat, R. Evol. Dev. 8, 394–406 (2006).
Seo, H. C. et al. Nature 431, 67–71 (2004).
Castillo-Davis, C. I. & Hartl, D. L. Mol. Biol. Evol. 19, 728–735 (2002).
Yang, J. & Li, W.-H. Gene 340, 237–240 (2004).
About this article
LRH-1 senses signaling from phosphatidylcholine to regulate the expansion growth of digestive organs via synergy with Wnt/β-catenin signaling in zebrafish
Journal of Genetics and Genomics (2017)
Repression of GPRC5A is associated with activated STAT3, which contributes to tumor progression of head and neck squamous cell carcinoma
Cancer Cell International (2017)
Intra-laboratory validated human cell-based in vitro vasculogenesis/angiogenesis test with serum-free medium
Reproductive Toxicology (2017)
Retinoid acid-induced microRNA-27b-3p impairs C2C12 myoblast proliferation and differentiation by suppressing α-dystrobrevin
Experimental Cell Research (2017)