Here are three animals. If you had to classify them on the basis of this superficial glimpse, which two would you guess were most closely related to each other, and which one would be most distant from the others?

On the left is a urochordate, an ascidian, a sessile, filter-feeding blob that is anchored to rocks or pilings and sucks in sea water to extract microorganismal meals. In the middle is a cephalochordate, Amphioxus, also a filter feeder, but capable of free swimming. On the right are some fish larvae. All are members of the chordata, the deuterostomes with notochords. If you’d asked me some years ago, I would have said it’s obvious: vertebrates must be more closely related to the cephalochordates—they have such similar post-cranial anatomies—while the urochordates are the weirdos, the most distant cousins of the group. Recent developments in molecular phylogenies, though, strongly suggest that appearances are deceiving and we vertebrates are more closely related to the urochordates than to the cephalochordates, implying that some interesting evolutionary phenomena must have been going on in the urochordates. We’d expect to see some conservation of developmental mechanisms because of their common ancestry, but the radical reorganization of their morphology suggests that there ought also be some significant divergence at a deep level. That makes the urochordates a particularly interesting group to examine.

What has long made the urochordates a recognized member of the phylum Chordata in spite of their relatively unusual adult morphology is their embryonic/larval affinities to the chordates. Before they settle down into their adult sessile life style, they go through a larval dispersal stage—they have tadpole-like larvae, like the one on the right, that have a characteristic chordate tail, with blocks of muscle arranged around an elastic rod, the notochord (which has been lit up with green fluorescent protein in that picture).

What I find fascinating about the ascidians is that they have an extreme mixture of attributes that are conserved with other attributes that are deeply divergent from the vertebrate pattern. Two papers have come across my desk lately that illustrate both of these aspects—what we find in the ascidian are another attribute once thought unique to vertebrates, the neural crest, a variety of different timing shifts in development that might be a clue to their diversity, and more evidence that the control of axial patterning is profoundly different from what we see in vertebrates.

Neural crest-like cells in ascidians

The neural crest are a population of set-aside cells found in early development. They are a marginal set of cells that arise during neurulation when, as the neural tube closes, some of the cells between the prospective neural tissue and the overlying ectoderm delaminate and transform from an epithelial sheet into multipotent, freely migrating mesenchymal cells. These cells are very plastic, and contribute to diverse tissues—much of your face below the eyes is dependent on contributions from neural crest cells, and they also provide the myelin sheaths for peripheral nerves, and are the source of the melanocytes that provide for pigment in your skin. Without the neural crest, you’d be very, very pale, slow-moving and relatively insensate, and you’d have a face that would make a hagfish look lovely in comparison. Be thankful.

The tree diagrams below illustrate the old phylogenies, which made vertebrates and cephalochordates sister groups (A), with more recent phylogenies, which place vertebrates and urochordates together (B). This distinction was made on the basis of molecular evidence, but this diagram also shows another piece of evidence that unites vertebrates and urochordates: both have neural crest or neural crest-like cells (NCLCs).

Neural crest evolution. (A) Traditional view. (B) View incorporating new insights from ascidians. Alternative deuterostome trees are based on (A) rDNA sequences, by which cephalochordates are the closest living relatives of vertebrates, and (B) meta gene phylogenies suggesting that urochordates are the sister group of vertebrates. NCLC: neural crest-like cells.

(By the way, hemichordates are very interesting in this context, too.)

Jeffery makes the case for the existence of NCLCs in ascidians by showing that they have cells that a) migrate like neural crest cells, b) express molecular markers like neural crest cells, and c) adopt fates that are like those of neural crest cells. Here’s some evidence of migratory populations of cells in an ascidian:

DiI tracing of migratory NCLC in Ecteinascidia turbinata. Dorsal views of brightfield (left) and fluorescence (right) images of the same mid-tailbud
embryo injected in the anterior neural tube near the otolith/ocellus (blackdot) at the early tail bud stage. Images shown at 10 min (A and B), 4h (C and D), and
14h (E and F) after DiI injection.

Diagram showing the three pathways of cell migration: anterior (red), posterior (blue), and ventral (yellow).

That looks familiar—I’ve done a lot of work like that myself. What they’ve done in the first panels is place a small dot of DiI in the embryo. DiI is a wonderfully useful lipophilic dye. You don’t have to inject it directly into cells, but instead, just place it in contact. The dye is bright and lipid soluble, and it gets taken up into the membrane of any cells that touch it. It’s so bright that the site of injection is often a glowing, unresolvable smear, so what you look for are cells, if any, that move away from the injection. That’s what we see here in this time series: a bright injection site, which isn’t so interesting, and small dots of cells moving farther away, which is. Using this tool in multiple experiments, you can eventually work out patterns of cell movements and map out pathways.

Finding that some cells migrate isn’t a definitive signature of neural crest, since cell movements are a fairly common hallmark of developing animals. Another step is to look for molecular markers that are also specialized for neural crest. And Jeffery has done that, too:

Expression of NC markers and body pigmentation in Ecteinascidia turbinata. Comparison of orange pigment cells (C), HNK-1 stained cells (D), and Etzic expressing cells (E) in the larval body wall.

That’s also familiar territory for me—I’ve used some of the same markers to look at neural crest and related cells in zebrafish. The ascidians have pigment cells on their skin — and remember, in vertebrates the pigment cells are all derived from neural crest — and cells with a similar distribution express the neural crest markers HNK-1 and Etzic.

At this point it’s not clear whether there is a one-to-one correspondence in ascidians between what are considered neural crest derivatives and the molecular markers. It’s also not clear whether the ascidian neural crest-like cells contribute to anything other than pigment cells. There are many unresolved ambiguities at this point, which is why these are called neural crest-like rather than outright neural crest cells. There is at least partial similarity, though, which justifies suggesting that a population of plastic, migratory cells might be just as much a chordate character as having a notochord. Now, of course, another interesting question would be the identification of homologous set-aside cells in hemichordates and echinoderms…

Timing changes in development

I was not previously aware of the extent of the developmental variation in ascidians, but here it is—they vary immensely. A good part of that overt variation can also be credited to changes in the timing of developmental events.

The simple view has been that ascidians have a tadpole larva with one set of structures, like the tail and notochord, which undergo a metamorphosis in which they throw away the larval structures and adopt the adult form. There is a distinct larva, a sharp metamorphic transition, and a distinct adult. It’s not that simple: in different species, there are different times for the onset of development of adult structures.

The diagram below illustrates the process of adultation. In the stereotypical ascidian (A), there is a tadpole larva (and sometimes even larvae that don’t form the tadpole tail!), and that sharp transition to the adult form. In others, though, there is a pattern of pre-assembling adult features in the head of the tadpole. We’ve got everything from a tadpole that is simple, but forms a branchial sac with a gill slit or two (Ciona, B), to forms that make more gill slits and a well developed siphon primordium (Botryllus, C), to what is nearly a complete ascidian adult in the tadpole head, with a tadpole tail that just needs to be discarded at metamorphosis, leaving the adult ready to get to work.

Adultation and complexity of ascidian larvae. Category A: Simplified larvae with no adultation as in Molgula occulta. Category B: Minimal adultation as in
Ciona intestinalis. Category C: Moderate adultation as in Botryllus schlosseri. Category D: Extreme adultation as in Ecteinascidia turbinata. (A-D) Left: structure
of larval heads. Right: line diagrams illustrating the extent of overlap in larval (blue) and juvenile adult (red) development. Vertical line: metamorphosis. Asterisks:
Approximate timing of the beginning of HNK-1 expression.

It’s the difference between shipping lumber and nails and so forth to a building site, versus having a pre-fab house shipped out. We see a whole range of timing shifts in this group, different solutions for different species.

One important thing is not known. What is the evolutionary sequence here? If you read the diagram from top to bottom, you could imagine a progression from a distinct larval and adult form that, with increasing parental investment, led to greater and greater amounts of pre-development of adult structures in the larva, making Molgula the primitive form, an Ecteinascidia the derived form. Alternatively, read it bottom to top: the primitive form could have been a tailed filter feeder with a complete set of adult organs, and that we’re seeing over evolutionary time is an increasing segregation of larval and adult functions—more temporal specialization. I’m inclined to favor the latter interpretation, but right now, there’s no evidence to support one over the other. I only lean one way because of other evidence that implies a trend towards increasingly specialized patterns of gene expression in this lineage as a whole.

Radical renovations in pattern formation

Here’s another different strategy within the urochordates: the larvaceans are also urochordates, but they don’t bother with that metamorphosis business at all—they retain the larval morphology throughout their life and become sexually mature while still looking like a tadpole (this may be an example of progenesis). These animals also develop very rapidly—they put a premium on quickly assembling the information for pattern formation, and this has led to some dramatic modifications of the pattern forming genes.

Best known of these pattern forming genes are the Hox genes. In most familiar organisms, the Hox genes are ordered in sequential banks, a cluster of genes that are turned on in sequential order along the length of the body. In us, for instance, there are global regulatory elements that control the ordered expression of these genes; there are also gradients in the embryo of important regulators like retinoic acid (RA) that control the boundaries of gene expression. It’s a pattern that’s tightly locked in, and the colinear organization of these genes are a necessary part of their function.

In the diagram below, each Hox gene is indicated by a colored arrow and numbered from 1-13. In the mammal at the top right, for instance, you’ll see the arrows arrayed in tight sequential strings on single strands of DNA. Look below the mammal, at the urochordate; the arrows are all dissociated from one another, scrambled around in the genome. The organizing principle, the global gene regulator, has been lost, and they have instead evolved different mechanisms for controlling where the Hox genes are expressed. The particular urochordate used in this diagram is the larvacean, Oikopleura dioica, which carries this fragmentation of the Hox cluster to an extreme.

(click for larger image)
Cladogram depicting Hox gene chromosomal organization for representative animals. At the base is shown a cnidarian (Nematostella vectensis), which has a dispersed genomic organization of Hox genes and lacks posterior Hox paralogs. The left branch displays fragmented Hox clusters for the lophotrochozoan flatworm Schistosoma mansoni and the ecdysozoan fruit fly (Drosophila melanogaster) and nematode (Caenorhabditis elegans). The right (deuterostome) branch portrays the rearranged but coherent Hox cluster of the sea urchin Strongylocentrotus purpuratus, the “prototypical” Hox cluster of Branchiostoma floridae (a cephalochordate), the dispersed genomic organization of the Hox genes of a urochordate (Oikopleura dioica), and the quadruplicated Hox clusters of a mammal (Mus musculus), which remain coherent but have experienced losses of multiple paralogs. Similar to the mammals but not shown diagrammatically, the ray-finned fish have multiple duplicate Hox clusters that are mostly coherent and have experienced gene loss, as exemplified by the zebrafish (Danio rerio), pufferfish (Takifugu rubripes), and medaka (Oryzias latipes). At the base of the cladogram is the likely Hox cluster organization of the last common ancestor of bilaterians. Genes are typically assigned to the Hox class if they encode homeodomain sequences that group with the founder HOX protein sequences from Drosophila and vertebrate clusters, and then into Hox homology groups arbitrarily designated 1 through 14. Even having a Hox-like homeobox sequence and mapping in a cluster of Hox genes is not an invariably useful standard for Hox axial patterning function in some animals, because one of the Drosophila Hox clusters contains the ftz (marked 6* in fly Hox cluster) gene, derived from Hox ancestors, but with novel developmental functions.

This pattern of disintegrating Hox clusters isn’t unique to urochordates. Other organisms show it, as well, and it’s associated with increasing rapidity and efficiency of development by adding specialized regulatory elements to the individual Hox genes, which would presumably make their expression more robust and more precise in developmental regimes where tight schedules make flexibility undesirable. Adding new, specialized regulatory elements gradually makes the old, whole-cluster regulators superfluous, and they can then be lost.

A paper by Cañestro and Postlethwait (another connection to my previous work—John Postlethwait was on my thesis committee) shows that Oikopleura has carried the process of discarding the primitive regulators of the Hox complex to a great degree. I mentioned that retinoic acid gradients are one of the regulators of boundaries of Hox gene expression…well, larvaceans have gotten rid of large chunks of the retinoic acid regulatory machinery as well, which is a bit of a surprise. RA is such a ubiquitous signaling molecule that it’s a shock to see it so thoroughly dismissed.

Schematic representation of the newly proposed deuterostome phylogeny highlighting the presence of RA genetic machinery and important events for the evolution of urochordates. Analysis of the RA genetic machinery (Aldh1a1, Cyp26, Rar and Rxr) in extant deuterostomes reveals that these genes were already present before the origin of chordates. Gene presence is indicated with “+” and absence with “-“. Urochordates (= tunicates, grey background) include larvaceans and ascidians. Four important events that probably occurred in the stem urochordates might have been crucial for the evolution of this subphylum: AP axial patterning became independent of RA-signaling, the Hox cluster was disrupted, Gbx was lost, and a determinative
mode of development was adopted. Larvaceans appear to have lost most of the components of RA genetic machinery.

This is another reason that ascidians are so darned interesting: they’ve taken great strides towards shucking major elements of the foundation of embryonic pattern formation, which is very cool.

It’s not at all contrary to evolutionary principles, however; there’s nothing contrary about taking a general set of molecular regulators, adding additional regulators that reinforce the original ones, and that eventually become so useful and dominant that the original regulatory mechanisms are reduced to an unnecessary redundancy, to the point that they can eventually be lost. To continue the house analogy—it’s like moving a house. House movers will go in and hoist a house on an additional foundation of jacks; during this process, the house will eventually be supported on a dual foundation of the new jacks and the old pillars or cinder blocks or other structural elements. Once the jacks are fully in place, the old foundation can be knocked away, and you’ve got the same old house, but now erected on a completely different foundation.

That’s what we’re seeing in these ascidians: a similar structure to ours, but built on novel regulatory elements that are not shared with us. We can still see the scars of our shared primitive evolutionary foundation in the Hox genes and the retinoic acid receptors (they haven’t lost every vestige of the RA system!). It’s simply a beautiful example of the diversity of developmental mechanisms that can be produced by evolutionary processes.

Cañestro C, Postlethwait JH (2007) Development of a chordate anterior-posterior axis without classic retinoic acid signalling. Dev. Biol. 395:522-538.

Jeffery WR (2007) Chordate ancestry of the neural crest: New insights from ascidians. Semin Cell Dev Biol [Epub ahead of print].