In the fly, segments fuse and display new features such as wings or legs. Illustration by Alison Schroeer
One of the fundamental features of the organization of multicellular animals is segmentation: We are initially built by subdividing a relatively undifferentiated embryonic tissue into smaller, repeated elements, like a stack of mostly identical building blocks. Look at an earthworm or a caterpillar or a maggot, and the organization is clear, with the wormlike animal showing the obvious seams and subdivisions that constitute its assembly from rings of similar chunks of tissue. Another property of this pattern of organization is that individual segments can then acquire specializations. In a caterpillar, the front end is modified with mouthparts and sense organs to form a head, while other segments will bear stubby limbs or be festooned with bristles or colored spots and patterns. Specialization is carried further when a maggot becomes a fly. Segments become much more obscure, retaining their visible identity in the abdomen, but are otherwise fused, elaborated upon, and display new features, such as wings or legs or mouthparts, that make the segments, ultimately, look very different from one another.
We vertebrates were also overtly segmented animals early in our embryonic development. As with the fly, the nature of our construction from similar blocks of tissue has been obscured by later additions in development, with limbs patched on and some segments (like human tails) reduced to near invisibility. Others (like significant portions of our brains) have been telescoped, contorted, and fused so that the boundaries between the original segments are detectable only to sensitive molecular probes. As with the fly’s abdomen, we also retain some still apparent vestiges: the chain of vertebrae in our backs and the muscles of our torsos.
Before those specializations intruded, however, there was an early period in development when all was simple, and the only job the embryo had to do was to set up partitions, segregate small sections of tissue that were all nearly identical, and let local developmental programs proceed within them. The most obvious expression of the process of segmentation is seen in the mesoderm—the embryonic tissue that will form bone and muscle—which begins as a long strip of cells in a continuous mass stretching the length of the embryo, and ends with the mass clumping into a chain of small segments of mesoderm, called somites. One somite (or a few) forms early, and then another coalesces just behind it, and a little later another one behind that, until the entire chain is constructed sequentially, from the front of the animal to the back. We can even watch this happen. I’ve put a short time-lapse recording of the process in a zebrafish embryo at http://scienceblogs.com/pharyngula/zebrafish. The movie covers about two hours of the embryo’s life, but has been sped up 1200 times so that you can easily see the events. At the beginning, four lozenge-shaped somites are visible on the left, and to the right is a ribbon of tissue. As the movie proceeds, portions of the ribbon are pinched off and added to the stack of new formed somites. The zebrafish continues to pinch off new somites until it reaches a total of 30 to 34. Some animals, such as snakes, may continue to form as many as 400.
One prediction from this observation is that there must be a molecular signal that sweeps in a progressive wave down the length of an animal to regulate the addition of these segments. Another marvelous property of this process is that it is clocklike. The new somites form with a predictable, regular rhythm. It’s so regular that developmental biologists can use the number of somites present in the segmentation period to figure out exactly how old an embryo is. In a zebrafish, for instance, since we know when segmentation begins and the rate at which new segments form (one every 30 minutes at 28° C), we can confidently state that if an embryo has 20 somites, it must be 19 hours old, plus or minus 15 minutes. The segmentation clock is species-specific—the chick takes 90 minutes to do the same thing that takes mice two hours, and humans eight. The additions are so regular and uniform, that one other thing we can predict about the mechanism is that it must involve an oscillator or clock with a period equal to the time it takes to add a new somite.
Thanks to the work of Olivier Pourquié and many other researchers, we’re beginning to puzzle out the underlying molecular mechanisms of the clock and wave front that generate vertebrate somites. The wave is provided by two molecules, the fibroblast growth factor fgf8 and the signaling molecule Wnt3a. Both of these molecules are expressed in a gradient, highest in the tail and lowest at front end of the animal, and this gradient slowly recedes backwards at the same rate that new somites are added. This wave represents the determination front; as cells lose the fgf8/Wnt3a signal, they organize themselves into somites.
The clock is more complicated, and it turns out that what vertebrates may have is a collection of little clocks that work together to generate a rhythm. All of them belong to a family of what are called Notch-related cyclic genes. Notch is a receptor protein that turns up again and again in development and evolution and is, to describe it in the most general terms, often involved in boundary formation. This makes it a natural to turn up in a process where we need to set up a series of segmental boundaries.
The simple explanation (there are more complicated interactions that I’ll set aside for now) for how these clock molecules work is that they are all transcription factors that exhibit auto-repression. That is, these genes are expressed to make a protein that binds to DNA and binds to its own gene, turning it off. What happens then is that the gene product accumulates, turns off the gene that makes it, and then the protein is slowly degraded by natural cellular processes. When the protein is sufficiently depleted, the gene is released from repression and is expressed again. The cycle continues over and over again.
The fgf8/Wnt3a wave front and the Notch-related cyclic clock work together to elegantly generate a precise spatial pattern over time. One way to picture this is to imagine standing at the shore, with waves lapping at the beach; as each wave rolls in, it leaves a little line of sea foam at its farthest reach. The waves are regular and clocklike, and with no other factors involved, would produce that line of sea foam at the same place every time.
But now we add another idea: The tide is going out. Each succeeding wave travels a little less far up the beach, and each one leaves its line of sea foam a little farther back toward the sea. Over time, what you’d see is a series of periodic lines of sea foam receding with the ebbing tide. In this example, the ebbing tide is like the receding gradient of fgf8/Wnt3a, while the individual waves are cycles of the Notch-related cyclic genes. Working together, two time-related processes can generate a regular spatial pattern of gene expression.
To read the Pharyngula blog, visit www.scienceblogs.com/pharyngula.
Originally published October 17, 2007