Biology 441 Spring 2012

Embryology   Biology 441   Spring 2012   Albert Harris


Lecture notes for Wednesday, February 22



What sorts of facts will we regard as explanations?

I will discuss this abstract question in terms of specific observations about somites.

Let us start with the abnormal formation of multiple rows of somites in the experiments by Ruth Bellairs and Claudio Stern that I mentioned last time.

A student asked me what they did to embryos that caused multiple rows to form, and I had forgotten the answer.
I then spent several hours searching on the web and in the book used as the textbook for this course last year.

Claudio D. Stern and Ruth Bellairs, "The roles of node regression and elongation of the area pellucida in the formation of somites in avian embryos" Journal of Embryology and Experimental Morphology vol. 81, pages 75-92 (1984)

I will quote Jonathan Bard’s book "Morphogenesis" page 169:

"The model can also explain one of the stranger observations in the study of somitogenesis, the formation of multiple rows of somites…"
Stern and Bellairs (1984) found that, when they cultured chick embryos on albumin/agar, a substratum to which the embryo did not adhere well, and so failed to extend properly, unexpectedly wide but short domains of presomitic mesenchyme resulted.
…within the majority of such embryos, somite organization had the appearance of 'bunches of grapes' up to five somites across.
[end of quote]

So it is analogous to the irregularity of feather rudiments in non-stretched skin!

Whole embryos can be cultured on agar gels (instead of on glass, or stretched flat), but they don’t adhere very well to the agar, don’t spread normally, and are not tense.

The result is less elongation, less mechanical tension and abnormal variation in somite formation, with more than one row of somites on both sides of the notochord, up to a maximum of five rows of somites.

Please note that this result was a big surprise to Bellairs and Stern. It wasn’t an example of someone inventing a hypothesis about what causes somites to form, and then figuring out that if a chick embryo were put on a non-adhesive surface, then the theory would predict that the abnormally un-stretched embryo would form five rows of somites on each side.

It wasn’t a total surprise, because Bellairs was the leading scientist who emphasized that lack of normal mechanical tension (for unknown reasons!) causes abnormal geometrical patterns in embryos. Stern and Bellairs made five specific conclusions in this paper, a,b,c,d,e
"d" was "The positions of the somites probably depends on mechanical tensions..."

They didn’t propose any specific reasons why this should be true, or what it predicts.
Jonathan Bard, in contrast, incorporates their discovery as part of a detailed theory about the mechanism of somite formation. He proposes that tension directly controls and causes somite formation, and synthesizes published observations of many researches, including me. For example, in footnote #25, Bard wrote,
"Harris et al., (1984) were chary of suggesting that traction could generate somites because such a mechanism could not account for the later spatial pattern of differentiation that took place. etc."

Bard’s later theory of somite formation was published in the journal that used to be named "Wilhelm Roux’s Archiv für Entwicklungsmechanik vol 197 pages 113-117.

Bard synthesized ideas somewhat like like the "Clock and Wavefront" (except with gradients of cell adhesion proteins, instead of gradients of growth factors), combined with observations of traction force exertion by embryonic cells, and an excellent insight (bottom of page 164) "…that somitogenesis is as much a process whereby a mass of cells becomes a tube, which then breaks up into small spheres." [rather than a process] "in which the mass breaks up into blocks."

Note that one of the you tube videos now linked to this web site shows Green Fluorescent Protein labeled paraxial mesoderm forming somites. Please watch this video carefully and tell me whether or not it looks as if randomly-arranged cells are first rearranging into a cylinder, and the cylinder is then progressively breaking up into a row of small spheres.. It does look that way to me. But it is difficult to make sense of the cell movements.

Somites form in Amphioxus embryos (a chordate that is not a vertebrate) by budding of epithelial spheres from the roof of the archenteron. (As if a row of lenses budding)

Incidentally, a puzzling fact is that somite formation in frog embryos seems to be geometrically different from somite formation in any other kind of vertebrate. Instead of budding or breaking up into small spheres, frog paraxial mesoderm cells first line up in the direction perpendicular to the notochord and neural tube, and then (as if ranks of soldiers were being ordered "Right Face", one group at a time) clusters of these medio-laterally elongated cells rotate their axis of elongation by 90 degrees, so as to become lined up in the anterior-posterior axis. This puzzles the heck out of me! Based on a few studies I have done, my only suggestion is that frog myotomes are a much bigger fraction of somites than in any other group of vertebrates (like 90%), instead of 10 or 20% myotome in the somites of birds, mammals and fish. I don’t know if salamander somites segment "by rotation", as frog somites do, But salamander myotomes are also a huge fraction of their somites.

In all vertebrates, the myotome part of somites becomes stringently oriented along the anterior-posterior direction. This is easy to see in histological sections, such as we look at in the laboratory. Also, if you dissect somites out of living early chicken embryos, and observe with a polarizing microscope, then it’s amazing how birefringent they are, and how strong the anterior-posterior alignment of myotome cells. They are already differentiating as muscle cells in the somite.

Some other interesting facts mentioned in Bard’s book is that chicken somites are made of about a thousand cells per somite, at the time of their formation, in contrast to feather papillae, which contain about 1,200 cells, each. So those regularly-arranged condensations that control feather location are about a fifth bigger than somites. I hadn’t realized that, and would have guessed somites would be bigger than feather papillae.

What are we going to accept as an explanation for somite formation?
What force rearranges cells? What mechanism controls numbers and locations of somites? Why form somites at all (considering they are transient structures, whose component cells will redistribute to form vertebrae, muscles and dermis).

And how do somite location control which stripe-shaped areas of your skin will later get innervated by which spinal nerves?
Because of a "slipped" disk between my 7th and 8th vertebrae, intense and constant pain is spread out down my left arm, and extending onto the ring finger and the little finger on that hand. The area of pain has sharp edges, that don’t change.

The word "dermatome" is used by neurologists to mean these areas of people’s bodies that are innervated by the nerve fibers of each particular vertebral ganglion.. Neurologists often have framed diagrams on their office walls that map out human dermatomes, in this sense of the word.

In contrast, the embryological meaning of the word dermatome is one of the three subdivisions of each somite. These become arranged as an epithelium, then become mesenchymal, and crawl outward to the skin, where they secrete collagen fibers and become the leathery, inner 9/10s of the skin, by mass.

What about the sensory nerve ganglia, made of neural crest cells, that aggregate just above each somite. Do the nerve fibers from each of these ganglia innervate only that stripe-shaped area of skin whose dermis is made of cells of the dermatome of the specific somite, above which that sensory nerve ganglion formed?

If so, what guides the axons to that area of skin? What prevents them from trespassing into adjacent dermatomes? Textbooks say nothing. Neurologists don’t know what embryologists mean by a dermatome; and embryologists (usually!) don’t know what the word "dermatome" means to neurologist. (Except if an embryologist gets a slipped disk! Or is the son of a neurologist.)
Neither has a good, testable theory about what causes this geometric pattern of innervation.

Wolpert and Tickle believe that boundaries between somites are controlled by a variation of Cooke and Zeeman’s "Clock and Wave-Front" model..
"Model" means a complicated theory, usually involving mathematics.

They mention papers that describe a moving gradient of concentration of one of the "Fibroblast Growth Factor" families of cytokine proteins, combined with time fluctuations on synthesis of proteins coded for by some genes named "Hairy.".

How do they deal with the fact that up to five rows of somites get formed, by embryos cultured on non-adhesive surfaces?

What is their explanation for how smaller embryos segment proportionally smaller somites, so as to produce the same eventual total number? (Driesch)

Why separate somites from each other? They are only temporary.
Why don’t the chemical gradients directly cause vertebrae to form?
Why don’t these gradients cause dermatomes directly?
Likewise, why don’t the gradients cause aggregations of sensory nerves to form ganglia. Ditto for causing locations of side branches from the aorta?

And why does the process of somite formation look like a cylinder breaking up into spheres, with a narrower cylinder forming shorter somites, thereby producing the same total number of somites in smaller embryos. (Or bigger ones.)

Wolpert always explains patterns as secondary responses to chemical gradients.
What causes the gradient? Never mind.
How the chemical causes the structure? Never mind that either.
What matters are the names of the genes needed for the gradient or the response?

C. H. Waddington suggested a mechanical explanation, back in the 1940s & 50s.
(He got famous for discovering that grafting Hensen’s node can induce a second embryo: the equivalent to Spemann’s phenomenon, for birds and mammals. In WWII, he was chief of British anti-U-Boat Tactics for the Bay of Biscay, wrote a book about it that was top secret until surprisingly recently, and is usually considered to have originated the field of operations research. No kidding. He was a close friend of HG Wells, who called him "Waddy". I read that in a biography of Wells. He also knew Tolkien.)

The one time I ever met Waddington, I was a graduate student giving a visiting lecture at MIT, and he sat in the first row, smoking a pipe, and blowing enormous smoke rings right at me, one after another. Each time, I stepped to my left and let the ring go by.

I wish I knew how well Waddington understood the math of the process of streams of water breaking up into spheres. It took me years to learn the following:
If a liquid, or a mass of cells adhering to each other, has a surface tension that is equally strong in every direction, then its stable shape will be a sphere. But if the tension at a fluid surface is stronger in one direction than in the axis perpendicular to that, then the stable shape can be a cylinder.

Therefore, consider the following variation on Waddington’s water drop theory.
If paraxial mesoderm cells stick to each other rather strongly, and if their collective surface tension is stronger in one axis than another, then they will spontaneously rearrange into a cylinder; Then, if their surface tension becomes equally strong in all directions, this cylinder will break up into spheres. To regularize both processes, you could use a gradient of one protein to cause the change in directionality of contraction, And use fluctuations of concentration another protein to control the periodic constrictions. These patterns of differing protein concentration would be easily misinterpreted as a clock and wave-front. But cell surface contractions would really be what pinches the mesoderm into separated somites.

Now, please help me figure out why reduced tension would cause somites to form multiple rows, as many as five! Computer simulation will be needed.


Near the top of page 432 of the book by Wolpert and Tickle:
Reaction-Diffusion Systems:

…An activator molecule, that stimulates its own synthesis and that of an inhibitor molecule, which in turn inhibits the synthesis of the activator, a type of lateral inhibition will occur, such that synthesis of activator is confined to one area. Under appropriate conditions, which are determined by the reaction rates and diffusion constants of the components… etc.

This set of rules is rather close to those that I used for synthesis and destruction of A and B. One difference is that they don’t specify that their "inhibitor" diffuse faster than their "activator", but they do say "under appropriate conditions, which are determined by the reaction rates and diffusion constants".
(Notice, they don’t really mean "determined by". They mean "which are reaction rates and differences in diffusion rates…")
A more important difference is that their inhibitor "inhibits the synthesis of the activator", but my substance B causes the destruction of A, as well as of B.

"Inhibiting the synthesis" is not quite the same thing as "causing the destruction".
First, Wolpert hijacks Turing’s word "Morphogen"; now, this!

We need a computer simulation in order to prove, true or false, whether wave patterns can also be generated by a combination of rules in which B merely reduces increases in A, as contrasted with B causing reductions in A.

Causing reduction is not the same thing as reducing increase.
In particular, the former can lower concentration of A, not just reduce increase.

Can you predict whether Wolpert’s set of rules really produce waves?.

My goal here is to prove to you why computer simulations are often the only way to find out the answers to fairly simple logical questions. .



contact the webmaster

back to syllabus