How to make a tadpole

i-0971a73433e3e9f1ec34d16d40e543e1-tadpole.jpg

I've been tinkering with a lovely software tool, the 3D Virtual Embryo, which you can down download from ANISEED (Ascidian Network of In Situ Expression and Embryological Data). Yes, you: it's free, it runs under Java, and you can get the source and versions compiled for Windows, Linux, and Mac OS X. It contains a set of data on ascidian development—cell shapes, gene expression, proteins, etc., all rendered in 3 dimensions and color, and with the user able to interact with the data, spinning it around and highlighting and annotating. It's beautiful!

Unfortunately, as I was experimenting with it, it locked up on me several times, so be prepared for some rough edges. I'm putting it on my list of optional labs for developmental biology—3-D visualization of morphological and molecular data is one of those tools that are going to be part of the future of embryology, after all—but it isn't quite reliable enough for general student work. At least not in my hands, anyway. If one of my students were to work through the glitches and figure out how to avoid them, though, it could be a useful adjunct to instruction in chordate development.

If you want to play with it, I'll give you a quick overview of what's going on in the dataset. A paper by Munro et al. has used these kinds of data to summarize key events in the transformation of a spherical ball of cells into an elongate, swimming tadpole larva.

Ciona is a tunicate, a marine invertebrate chordate, that in its adult form is a kind of sessile barrel-shaped blob called a sea squirt that attaches to a substrate and spends its life filter-feeding. As an embryo, though, it rapidly forms a tadpole shaped larva that displays its chordate affinities: it has a notochord, and gill slits, and a long post-anal tail, and a dorsal nerve cord. It uses these features to swim, as a dispersal apparatus, before settling down and metamorphosing into the adult form.

i-d64d6b181f325bdf4902a6cc352030fb-ciona.gif
The simple Ciona tadpole. Schematic overview of the major tissue types. The colour
scheme used here and in Figures 2-4 is as follows: brown, mesenchyme; dark blue, nerve cord; green, palps; gray, epidermis; light blue, brain;
orange, muscle; red, notochord; yellow, endoderm.

Their development is different from ours in some significant ways. They exhibit an invariant, stereotyped pattern of cleavages, with a mosaic pattern of development in which each newly divided cell is assigned to a specific and relatively inflexible fate. Each cell has a specific and predictable lineage, which has been a real boon to embryologists since Conklin described them in 1905—you can recognize each cell in the embryo, give it a name, follow what it does, experiment on it, and ask what tissues it will generate. The cells also have different distributions of pigment, so even at the turn of the last century it was relatively easy to follow their development; it was as if they came with a colored label.

The diagram below shows some of the stages. As you can see, the divisions can be asymmetric, and shape and size are also indicators of cell identity.

i-f9aa8b848bdf0b7178fb0588d4060788-ciona_cleavage.jpg
Cleavage stages in ascidian embryos. (a) Patterns of induction and
cleavage that accompany early fate specification in ascidian embryos.
For each hemisphere, the left column illustrates the progressive fate-
restriction of each blastomere, and the right column illustrates
corresponding patterns of induction and asymmetric cell division. In the
left columns, fate-restricted blastomeres appear in the colour of the
corresponding tadpole tissue of Figure 1b. Blastomeres giving rise to
two or more tissue types are not coloured. In the right columns,
blastomeres that are induced are coloured in pink; blue lines link sister
blastomeres with equal volume; red lines link sisters with unequal
volumes; black lines link sisters whose volume has not been determined.
In the vegetal view of the 110-cell stage, cells marked by an asterisk
have not undergone cleavage since the 64-cell stage. (b) Visualisation in
green of the surface of contact between the a6.5 and A6.2 blastomeres;
made using the 3D Virtual Embryo module.

One other quirk of ascidians is that there is almost no cell migration, and a cell generally hangs on to its nearest neighbors throughout embryonic development. In vertebrates, we're used to seeing cells uncouple from each other extensively and stream to new locations in the embryo. One classic example is gastrulation: in the two-layered early embryo, a subset of cells move into the interior of the animal to establish a third germ layer, the mesoderm. Ciona also gastrulates, but cells don't dissociate from each other. Instead, the embryo forms a flattened disc, and then curls in on itself to move cells into an interior space.

i-8a63e446e77d97c969dc52bc969e08c2-ciona_gastrulation.gif
Phases of gastrulation in Ciona. Top panels shows coloured tracings of the same SEMs. Bottom panels shows tracings of
single parasagital confocal sections taken at corresponding stages. Anterior is up in all panels, and
animal is left in the bottom panels. (a,b,c) First phase: formation of cup-shaped embryo. (a) 76-cell stage: the vegetal plate first flattens.
(b) 110-cell stage: apices of the vegetal cells (indicated by asterisks) constrict as the vegetal half of the embryo bends inwards to form a shallow
cup; the animal-half spreads as the animal cells divide once and then flatten. (c) Mid-gastrula: the cup deepens asymmetrically as anterior and
lateral mesoderm cells fold inwards, while endoderm, muscle and epidermal precursors execute a single round of cell divisions. (d) Second
phase: extension of anterior tissues, and asymmetrical blastopore closure produces a completely asymmetrical embryo. Red arrows indicate
AP-oriented cell divisions in the neural plate and notochord lineages that drive extension of the anterior plate. Along the anterior lip of the
blastopore, apices of neural plate cells [white arrow in middle panel of (d) and notochord cells (not shown)] constrict as the blastopore closes.

Finally, a round ball of cells has to reshape itself into that elongate tadpole form you can see at the top of this page. In this process, we get into more familar ground for vertebrate embryologists: elongation of a cluster of cells is accomplished by intercalation. Cells push medially, into the interior of the tissue. When a cell moves inward, it pushes neighboring cells to either side, at right angles to the direction of movement. All of the cells jostling to squeeze in between each other, to intercalate, generates a force for elongation. Imagine a pile of coins loose on your desk. If you put your hands on the edges of the pile and push all the coins inward, they'll stack on one another and pile higher while reducing their overall footprint on the desk. The same principle operates here; cells are more obliging than rigid metal coins, though, and will eventually get into a perfect stack that maximizes the length of the column. (You'll often see the notochord described with the term "stack of coins" in the literature—it actually looks a bit like that.)

i-9ef42cb39dd534295fc31bfd1a7dc5c5-ciona_elongation.gif
Cellular processes that accompany tail elongation during neurula and
tailbud stages. (a,d,g) Early neurula; (b,e,h) early tailbud; and (c,f,i)
late tailbud stages. The view is of the dorsal side, and posterior is up
for all panels. Black arrows indicate whole-tissue deformations; white
arrows and outlined cells indicate local cell shape change and
rearrangements. (a-c) Notochord: during neurulation, (a) the monolayer
notochord plate invaginates to form a cylindrical rod, while mediolateral
intercalation within the plate drives AP extension. During tailbud stages,
(b) intercalation about the circumference of the rod drives AP elongation
of the notochord. (d-f) Muscle: during neurulation, (d) cell
rearrangements transform a roughly 6
x 6 array of cells into an array
3 cells high and 12 cells long. During tailbud stages, (e) the initially
isodiametric cells elongate individually along the AP axis as the tail
further extends. (g-i) Neural plate and tube: during neurulation,
(g) invagination turns the neural plate into a tube, beginning at the
posterior and progressing anterior. During tailbud stages, (h)
mediolateral intercalation (at the top and bottom of the tube), and
oblique divisions (not shown), followed by mediolateral shearing
(along the sides of the tube), accompany elongation of the tube.

That should get you started in figuring out what's going on. Now go play with the free software and watch all these events happen! The paper by Tassy et al. listed below also describes how the software is used to examine cell shape changes and interactions in ascidian development.


Munro E, Robin F, Lemaire P. (2006) Cellular morphogenesis in ascidians: how to shape a simple tadpole. Curr Opin Genet Dev. 16(4):399-405.

Tassy O, Daian F, Hudson C, Bertrand V, Lemaire P (2006) A quantitative approach to the study of cell shapes and interactions during early chordateembryogenesis. Curr Biol 16:345-358.

More like this

Ciona is a tunicate....

Sounds like a guy I knew in high school.

By Tukla in Iowa (not verified) on 10 Sep 2006 #permalink

They exhibit an invariant, stereotyped pattern of cleavages, with a mosaic pattern of development in which each newly divided cell is assigned to a specific and relatively inflexible fate.

It was pointed out that insects develop the same way (in the link about gynandromorphs from the post about the butterfly with differently-coloured wings).

Is highly deterministic cellular division more common than not among the various animals? Can it be broken down by phylum (that is, all arthropods develop deterministically; no vertebrates do so)?

By Owlmirror (not verified) on 10 Sep 2006 #permalink

Can it be broken down by phylum (that is, all arthropods develop deterministically; no vertebrates do so)?

Er, since Cionia is a vertebrate, obviously some do. But anyway, the essense of my question still stands.

Or to put it another way, since humans don't develop deterministically, is it just amniotes (or some other subdivision of vertebrates) that don't?

By Owlmirror (not verified) on 10 Sep 2006 #permalink

What happens to the other 18 cells between the 64 and the 110 cell stages? Do they suffer "programmed cell death"? Or do 9 of them not divide?

The divisions are not synchronous, so yes, some don't divide during that period.

Ciona is a chordate. It isn't a vertebrate.

The mosaic vs. regulative development is a continuum, not an absolute, so different members of different groups show differing degrees of mosaicism. Animals that develop very rapidly tend to skew towards the mosaic end of the spectrum; it reflects a greater amount of preparation in the embryo by maternal determinants. Invertebrates tend to be more likely to exhibit mosaicism, but it's not black and white. Echinoderms are regulative, and so are cephalopods.

Very, very cool. Outstanding! And by the way, where did the free software come from? Oh ... you mean someone designed the software? Okay ... mums the word, I won't tell anyone.

Why, yes, and that someone was French. By the law of inappropriate and excessive analogy, we can therefore conclude that the designer of all the universe was French.

Boy, that's going to piss off the wingnuts no end when they arrive at the Pearly Gates and a guy in a beret hands them a phrase book and tells them zey will not be seated until zey have mastered zee proper pronunciation.

I'll concede that the universe was designed as soon as the IDiots produce the source code.

By Tukla in Iowa (not verified) on 10 Sep 2006 #permalink

hmm... an active youth followed by a sessile, barrel-shaped adulthood of feeding. sounds like many humans have ciona-like atavisms.

interesting that mosaic development is tightly correlated with cleavage patterns in most cases, but recent work in zebrafish shows that the dorsal axis is specified earlier than was thought (4-cell stage), but is independent of cleavage planes (gore et al., nature 2005).

I couldnt get it to work. Do I need other plug ins?

A new version of the software is now available that should avoid the crashes due to the frequent change of embryos.
You can download it in:
http://crfb.univ-mrs.fr/aniseed/virtual_embryo.php

To install it, you'll need a recent version of Java JRE *AND* Java3D. Instructions are available in the webpage.

By Daniel Sobral (not verified) on 13 Sep 2006 #permalink