Assuming that none of my readers are perfectly spherical, you all possess notable asymmetries—your top half is different from your bottom half, and your front or ventral half is different from you back or dorsal half. You left and right halves are probably superficially somewhat similar, but internally your organs are arranged in lopsided ways. Even so, the asymmetries are relatively specific: you aren’t quite like that Volvox to the right, a ball of cells with specializations scattered randomly within. People predictably have heads on top, eyes in front, arms and legs in useful locations. This is a key feature of development, one so familiar that we take it for granted.

I’d go so far as to suggest that one of the most important events in our evolutionary history was the basic one of taking a symmetrical ball of cells and imposing on it a coordinate system, creating positional information that allowed cells to have specific identities in particular places in the embryo. When the first multicellular colony of identical cells set aside a particular patch of cells to carry out a particular function, say putting one small subset in charge of reproduction, that asymmetry became an anchor point for establishing polarity. If cells could then determine how far away they were from that primitive gonad, evolution could start shaping function by position—maybe cells far away from the gonad could be dedicated to feeding, cells in between to transport, etc., and a specialized multicellular organism could emerge. Those patterns are determined by interactions between genes, and we can try to unravel the evolutionary history of asymmetry with comparative studies of regulatory molecules in early development.

In developmental biology, those who study early events are often looking at the formation of these asymmetries. One way to think of development at these stages is as a process that establishes fields of interacting cells, each field with a general expected role to play in the embryo, and each with its own different patterns of gene expression. The distribution of the fields, these fundamental asymmetries, are set up early in development, and we can map them out. Here, for instance, is a fate map of an early frog embryo, at a time when it looks like little more than a ball of cells.


The embryo itself looks nothing like this, but is just a ball of cells with some orienting features: there’s more pigment at the top (the “Animal” side), and at this early gastrula stage, there’s a little pucker called the dorsal lip, indicated by the short green line on the right side of the image. What scientists do is mark individual cells—in the old days, they might use particles of chalk or carbon, pressed up against the surface of the cells, but nowadays we more commonly inject the cells with a fluorescent dye—and then let the animal develop and ask where the marker ends up. Mark cells just above the dorsal lip, for instance, and then look a few days later at the tadpole, and you discover that only the notochord is fluorescing, and nothing else. Mark cells a little further towards the animal pole, and only cells in the nervous system are lit up. Do thousands of experiments like that, and you end up with a map like the one above, where you’ve got a good idea of what each cell will do in the embryo. You’ve also exposed otherwise invisible asymmetries in the animal.

Way, way back in my grad school days, this kind of work was central to my Ph.D. thesis project. Others in the Kimmel lab at the University of Oregon were working to build a fate map of the zebrafish, and I was taking advantage of their work to look at the developing nervous system. They told me where the cells in the early embryo were fated to become the nervous system, and Judith Eisen and I would inject single cells in that region, let the embryo grow up to a stage we were interested in, and then we’d have animals where subsets of neural cells were glowing green, making them easy to watch.


Fate maps give an idea of prospective changes in the formation of tissues and organs, but what we ultimately want to examine is the pattern of gene expression in these cells. For that, we use another tool: the in situ stain. In this procedure, if you know the sequence of a gene, you can make a probe that complements the sequence of the mRNA it produces, and tag it with a colored marker. Soak the embryo in your probe, it sticks to RNA in any cells that are expressing the gene of interest, and can be washed out of cells that don’t have any complementary RNA. The image to the right is of some other old work I did with Scott Stachel and David Grunwald on zebrafish. Those spherical balls are early embryos around the time of gastrulation, and we labeled them with a probe to the gene goosecoid (gsc). That gene is turned on in cells that will become the notochord, and we could use it as a marker for an early asymmetry—at a time when the animal was a ball of cells, we could detect that blue lozenge of cells, which were the population about to become a specific tissue.

Again, after many experiments by many people, we can determine what genes are turned on where and when in setting up these asymmetries, and here is a partial map of important genes superimposed on a frog fate map.

Diagram of a fate map for an amphibian embryo indicating the positions of ectodermal (including the CNS), mesodermal, and endodermal germ layers and some of the signaling molecules located on the ventral and dorsal (Spemann Organizer) sides involved in D-V polarity formation.

This is a tiny snippet of what we know—these are just some of the genes involved in setting up one plane of asymmetry, the dorsal/ventral distinction. BMP4, for instance, is a ventralizing gene that specifies cells to become part of your front half, while Chordin/Noggin/Gsc are genes that suppress BMP4 or specify dorsal fates, contributing to the formation of your notochord and nervous system.

These are old, old genes. We share them with other animals, like insects, so they arose and acquired their functions in generating asymmetries before our lineages separated, sometime way back in the pre-Cambrian. They are part of our makeup as members of the Bilaterian superphylum, the animals with bilateral symmetry, and, we thought, distinguish us from the Radiata (diploblasts with radial symmetry) and Parazoa (multicellular animals with no discrete tissues or organs.

They seem to be older than we thought, though. Recent work by Matus et al. has delved into the developmental molecular biology of the starlet anemone, Nematostella vectensis, specifically plumbing the sequenced anemone genome for the same genes used in frogs and flies and people to define our dorso-ventral axis, and they’re there. What’s more, they are expressed asymmetrically, and the anemone contains a huge amount of hidden complexity in its organization.

Here’s a picture of the planula larva of Nematostella—it’s really not much of an animal, little more than a hollow ball with an opening at one end and a tuft of cilia at the other. Asymmetry along one axis from blastopore to apical tuft is obvious, but anything equivalent to the dorso-ventral axis isn’t; it seems to be radially symmetrical about that axis.

Body plan of the anemone N. vectensis. (A and B) Micrographs of a lateral view of a planula-stage N. vectensis embryo. (A) Differential interference contrast micrograph. (B) Confocal micrograph with F-actin (green) and nuclei (red) stained with BODIPY-phalloidin and propidium iodide. The asterisk marks the blastopore and site of the future mouth. at, apical tuft; b.end, body wall endoderm; b.ect, body wall ectoderm; pha, pharynx; p.end, pharyngeal endoderm; p.ect, pharyngeal ectoderm.

Not so fast, though—a cross section reveals some subtleties. Look at the pharyngeal structure in the center, which is flattened along one plane. The authors don’t want to call it a dorso-ventral axis, reasonably enough since it isn’t quite the same, and call it the “directive axis” instead. It is a plane of symmetry that gives a bilateral aspect to this member of the Radiata.

Cross-sectional diagram of the N. vectensis body plan showing the different germ layers.
Gsc is asymmetrically expressed along the directive axis during development. NvGoosecoid (gsc) is expressed initially in body wall endoderm (b.end) (A) and pharyngeal endoderm (p.end) and ectoderm (p.ect).

Now this paper by Matus et al. is a fairly typical representative of a modern developmental biology paper. It contains in situs. Lots of in situs. They threw an amazing amount of work at this creature, searching genome databases for homologs to standard bilaterian genes, making probes, staining Nematostella embryos and larvae for these genes, and noting patterns of asymmetry. I’ll make it short and show you just one: to the right is animal stained for goosecoid gene expression, and as you can see it is not radially symmetric, but instead is more heavily expressed on one side of the directive axis. If you look back up at that photo of goosecoid expression in the zebrafish embryo, it’s much more discrete in the vertebrate, with expression confined to just the dorsal side of the dorso-ventral axis. So there are significant differences, but the key thing is that Nematostella is exhibiting strongly asymmetric patterns of gene expression.

Here is a diagrammatic summary of all that work. There are an awful lot of genes with asymmetrically patterned expression, and with homology to bilaterian patterning genes.

Summary of asymmetric gene expression patterns during development in N. vectensis. (A) Lateral view with the oral (anterior) pole toward the left. (B) Cross section through the pharynx. Asterisk marks the site of gastrulation and mouth. The mesoglea (yellow line) separates the ectodermal epidermis from the endodermal gastrodermis. Colored bars represent different gene expression patterns, with hash-marked color bars representing transient early gene expression. Tentacle and apical tuft expression is not show for clarity. A suite of bilaterian patterning genes (transcription factors, secreted ligands, and their antagonists) is dynamically expressed along both the oral/aboral axis and the directive axis in both germ layers. The asymmetric expression of NvChordin, NvDpp, NvNoggin, and NvAnthox8 in the pharynx on the side opposite the expression of NvNetrin, three Hox genes, the larger domain of NvGsc, and NvGDF5-like, suggests homology of the directive axis to the D-V axis of bilaterians. The bilateral expression of NvGbx and NvChordin marks, the left–right axis.

In addition to the surprising sophistication of asymmetric gene expression in Nematostella, there were a few other fascinating tidbits.

The function of some of these genes is highly conserved. In frogs, for instance, you can inject extra frog noggin, a dorsalizing gene, into frog embryos and induce extra body axes—two-headed tadpoles and that sort of thing. Injecting the Nematostella version of noggin into frogs does exactly the same thing, dorsalizing that region of the embryo and inducing extra body axes.

A classic teratological agent (chemical that induces birth defects) is lithium chloride. The salt interferes with an important signaling pathway which regulates the expression of dorsal genes, so that animals express genes like goosecoid everywhere, rather than in one asymmetric spot (in the work I mentioned above with zebrafish goosecoid, we made quite horrible looking embryos that were radialized by exposing them to LiCl). Lithium does the same thing to Nematostella, radializing the expression of goosecoid. This suggests that the signaling mechanisms that regulate these genes are also conserved.

One of the genes in the diagram above is Netrin. Netrins are a whole family of molecules found in the nervous systems of bilaterian animals: they are secreted agents that diffuse from sources at the ventral midline (in arthropods) or the ventral floor of the nervous system (in chordates) and play a role in guiding growing nerve fibers to their appropriate destinations. Unfortunately, we know very little about the Nematostella nervous system—like other coelenterates, it doesn’t have much of one, and what it does have is probably a rather diffuse nerve net, scattered throughout the body wall. It would be very interesting to see if there is any hints of a greater centralization, associated with the distribution of Netrin.

The most important message, though, is that the morphological complexity of modern bilaterians has molecular antecedents in seemingly simple organisms. It suggests that the cnidarian ancestor might well have been bilaterally symmetrical, and that the apparent simplicity of modern jellyfish and anemones may be the product of a loss of bilaterality, meaning that the basic regulatory interactions that define asymmetry in animals may have a greater antiquity than we usually think.

Matus DQ, Pang K, Marlow H, Dunn CW, Thomsen GH, Martindale MQ (2006) Molecular evidence for deep evolutionary roots of bilaterality in animal development. Proc Natl Acad Sci U S A. 103(30):11195-200.

Stachel SE, Grunwald DJ, Myers PZ (1993) Lithium perturbation and goosecoid expression identify a dorsal specification pathway in the pregastrula zebrafish. Development 117(4):1261-74.