Do vertebrate embryos exhibit significant variation in their early development? Yes, they do—in particular, the earliest stages show distinct differences that mainly reflect differences in maternal investment and that cause significant distortions of early morphology during gastrulation. However, these earliest patterns represent workarounds, strategies to accommodate one variable (the amount of yolk in the egg), and the animals subsequently reorganize to put tissues into a canonical arrangement. Observations of gene expression during gastrulation are revealing deeper similarities that are common in all deuterostomes—not just vertebrates, but also the invertebrate chordates (tunicates and cephalochordates) and echinoderms.

What does all that mean? If you think of development as a formal dance, the earliest stages are like the prelude; everyone is getting out of their chairs around the ballroom, looking for partners and working their way towards the floor. The dispositions of the dancers are variable and somewhat chaotic, and vary from dance to dance. Once they get to their positions, however, we're finding that not only is there a general similarity in their arrangements, but they're all dancing to the very same tune. In this case, one of the repeated motifs in that tune is a gene, Nodal, which is active in gastrulation and shows a similar pattern in animal after animal.

Holoblastic cleavage of a sea urchin embryo

Let's begin with the players beginning to take their places. There is much variation in how different deuterostomes carry out their earliest cell divisions, variation that we can reduce to two different categories: holoblastic vs. meroblastic cleavage.

Holoblastic cleavage is the easiest to visualize. The fertilized one-cell zygote divides completely into two separate cells, then each cell divides completely again to produce four cells, and again and again. The end result is a clustered mass of cells that resembles a bunch of grapes, or the druplets of a raspberry. This is the pattern of cleavage that we humans (and all mammals) use.

Meroblastic cleavage of a zebrafish embryo (the two partially separated cells are on top)

Meroblastic cleavages are incomplete divisions. The nuclei of the cell replicate and divide completely, but when cytokinesis (the actual division of the cytoplasm and membrane) occurs, only a partial membrane forms. This kind of division occurs in very yolky eggs, such as a chicken egg, and is a reasonable compromise to allow replication to occur rapidly by not bothering with the assembly of massive amounts of membrane. The yolk of a chicken egg is a single cell; imagine that having to be subdivided with new membranes at every division. Instead, the embryo typically forms as one small disc on the surface of the massive ball of yolk.

Holoblastic cleavage is almost certainly the ancestral condition, and is common in the invertebrate chordate phyla. Any animal that packs its eggs with lots of yolk, however, evolves the meroblastic shortcut at some point, and so a cladogram of the patterns of early cleavage looks rather chaotic—no nested hierarchies are easily visible here!

Cleavage patterns within the Chordates. Ascidians, lancelets, and lamprey display holoblastic cleavage, suggesting that it is the ancestral cleavage pattern in chordates. Ascidians have bilateral cleavage, while lancelets display radial cleavage. Lamprey and amphibians have meso-lecithal eggs, with larger, yolky cells confined to the vegetal half of the embryo. Present day teleost fish show meroblastic cleavage, which evolved independently in fish and amniotes. Mammals show a reversion to holoblastic cleavage, but exhibit rotational cleavage.

We can see the evolutionary relationships in other ways, though. Animals with meroblastic cleavage produce many cells, and eventually those cells migrate to engulf the uncleaved mass of the yolk. The cells that migrate eventually produce a thin membrane around it called the yolk sac. We mammals are a revealing example: from the cladogram, you'd predict that we had a reptilian ancestor that divided meroblastically, and that we secondarily lost that pattern of cleavage and readopted the ancestral holoblastic pattern as a consequence of not bothering to pack our eggs with yolk (instead, our eggs stay resident in women's bodies and draw nourishment directly). Yet our zygotes still divide into many cells, and then a sub-population migrates to surround a fluid-filled, yolkless space—they make a yolk sac.

Out of the variability of the early divisions, however, comes order. The dance begins at gastrulation, when cells begin to make consistent patterns of migration to set up the layered tissues of the embryo. The milling masses of cells begin to line up and order themselves into a bilayered structure. In the diagrams below, you can see that most of the animals form a hollow ball, with one set of tissues above at the animal pole (a), and another set below at the vegetal pole (v). The animals with a more meroblastic cleavage tend to form discs instead of hollow balls, but it's the same structure; the ball has just been flattened. The cell movements that follow consist of directed migrations towards the dorsal side, and invagination into the center of the hollow ball or disc that will produce a mesodermal layer.

Gastrulation in the deuterostomes. The animal pole (a) is marked by polar bodies; (v) is vegetal pole. (A) is future anterior; (P) is future posterior. In echinoderms and hemichordates, invagination of the endoderm begins at the vegetal pole and the early gastrula is radially symmetrical. The blastopore becomes the future anus of the larvae. In most echinoderms, the animal- vegetal axis becomes the larval A-P axis, except in echinoid and ophiuroid pluteus larvae. In chordates, invagination during gastrulation also begins at the vegetal pole. However, in the Chordata, the future anterior and posterior of the larva are determined by the point of sperm entry, which breaks the symmetry of the radial egg. The embryo posterior develops opposite the point of sperm entry.

Gastrulation is a complex process with many specific cell rearrangements and requiring the activation of many molecules. To keep things manageable, this paper focuses on one, nodal.

Nodal is a signaling molecule, a member of the TGF-β family, which plays a crucial role in gastrulation. In the middle diagram below, it is expressed as a graded signal which specifies where the Spemann organizer is to form—the organizer defines the site where the invagination of cells mentioned above is to occur, and is also a site where other molecules are expressed that signal migrating cells what position and fate they should take. Nodal is a central player in the formation and maintenance of mesoderm in early development.

Nodal is also a deuterostome universal: this molecule is used in similar ways cephalochordates, tunicates, echinoderms, and vertebrates. It has not been found in arthropods, although there are suggestions that polychaetes and molluscs may have a form of it. (All do have TFG-β molecules, just not the specific Nodal gene).

There is another critical function for nodal. It is later expressed in an asymmetric pattern, with a domain of expression on just one side. Slap your left side; it doesn't matter whether you are left- or right-handed (although it does matter if you have situs inversus), you've found the side of your body where nodal was active when you were a little embryo. It's also the same side where nodal was turned on in mice and frogs and fish and cephalochordates and tunicates, as you can see in the Late Gastrulation diagram below. It is also turned on asymmetrically in sea urchins, but it's got a rather more complicated relationship to morphology there—it's turned on in the region of the sea urchin's prospective mouth.

Comparison of vertebrate nodal and nodal-related gene expression to lancelets, ascidians, and sea urchins. Animals are shown at early gastrulation (top, A–K) and after completion of gastrulation (bottom, B–L). Areas of nodal expression are indicated by heavy areas of shading. A: In mouse gastrulae, nodal is expressed bilaterally around the node. B: By early somite stages, nodal is expressed asymmetrically in the left lateral plate mesoderm and in the node where ingression is occurring. C: In early Xenopus gastrulae, nodal is expressed at the dorsal lip and later (D) in the left lateral plate mesoderm at late neurulation. E: Expression of cyc is localized to the hypoblast layer of the embryonic shield in early zebrafish embryos while sqt is expressed in the dorsal forerunner cells. F: During somitogenesis in zebrafish, cyc and spaw are both expressed in the left lateral plate mesoderm.
G: In lancelets, nodal expression is observed in the hypoblast layer of the dorsal lip. H: At late neurulation, nodal is asymmetrically expressed on the left side of the paraxial mesoderm, gut endoderm, and presumptive oral ectoderm. I: In ascidians, nodal is expressed bilaterally within the presumptive endoderm, epidermis, and trunk lateral cells from the 32-cell stage to early gastrulation. J: By the initial tailbud stage, ascidian nodal is expressed in the left side of the epidermis. K: In sea urchins, nodal expression begins in the presumptive ectoderm at the 60-cell stage, but decreases once gastrulation begins. L: Nodal transcripts are localized in the presumptive oral ectoderm during prism and pluteus stages.

The association with the echinoderm mouth is intriguing. In cephalochordates (the lancelet), the oral opening is initially asymmetric, too…and it forms on the left side. The authors suggest that one very interesting thing to examine would be the pattern of expression of nodal at the time the vertebrate mouth forms; it may have a role in that, too.

I've said a few things before about generating left-right asymmetries in development, and the key observation in us mammals at least is that the initial source of the asymmetry is the clockwise rotation of cilia on the surface of the node, the site of gastrulation. This rotation sets up a net leftward flow of fluids over the gastrula, which is illustrated below. In B, we're looking at a cross-section of an embryo, with a couple of ciliary hairs sprouting up and spinning. At the same time, cells lining the surface are secreting nodal vesicular parcels (NVP), small membrane-bound blobs that contain a payload of signaling molecules, sonic hedgehog (shh) and retinoic acid (ra).

Breaking left-right symmetry at the node. A: A leftward fluid flow (indicated by arrows) is the earliest known asymmetric event in mammalian left – right determination, resulting in activation of the nodal signaling cascade in the left lateral plate. In contrast in the chick, a cascade of asymmetric gene expression at the node is upstream of the nodal signaling cascade. B: FGF signaling in the node permits nodal vesicular parcel (NVP) production. Nodal flow carries NVPs to the left side of the node where they fragment, releasing lipid, SHH and RA, activating left-sided Ca²⁺ signalling.

When the NVPs are released, they enter the bulk leftward flow, and are catapulted toward the left side of the embryo—when they contact the surface, they splatter apart, releasing their contents and triggering a signal cascade that eventual stimulates nodal expression. Now that is a Rube Goldberg machine, a weird way generate left-right differential expression of a gene.

As odd and unexpected as the mechanisms might be, the important message is that the deuterostome gastrulation dance seems to be choreographed in similar ways and with similar players. There are fascinating variations in individual species, though, and this molecule is a promising target for studies of the evolving regulation of development.

Chea HK, Wright CV, Swalla BJ (2005) Nodal signaling and the evolution of deuterostome gastrulation. Developmental Dynamics 234:269-278.

Norris D (2005) Breaking the left-right axis: do nodal parcels pass a signal to the left? Bioessays 27:991-994.