Embryology   Biology 441   Vertebrate Embryology, Spring 2014   Albert Harris


Monday March 19 Lecture: Geometric Patterns and Gradients in Embryonic Development :

The key breakthrough regarding induction was realizing that cell differentiation is usually controlled by close-range chemical signals between adjacent cells.

Discovering specific proteins that produce induction has been an anti-climax.

The next question is how to produce consistent geometric patterns like stripes and ribs. Spemann's use of the word "organizer" suggests that he expected induction somehow to create patterns. Pattern formation will be my main topic today.

BUT FIRST, notice the importance of embryonic induction to "stem cells". Methods are needed to stimulate undifferentiated cells to differentiate into whichever of our 250 cell types that are needed, and to form correct geometric arrangements.

Exactly what do people mean by stem cells?

Some parts of our bone marrow are full of undifferentiated, rapidly-growing cells that continually replace nearly all kinds of blood cells (red blood cells, platelets, and all the kinds of white blood cells, except lymphocytes [which are also replaced, but from different precursors).

Two other tissues that are constantly replaced by undifferentiated cells are the skin and the lining of the digestive tract.

Until about the last decade, the phrase "Stem Cell" used to refer specifically to these undifferentiated skin, intestine and bone marrow cells that are not yet differentiated, although irreversibly committed to which cell types they can replace. For example, bone narrow stem cells can't replace intestine or skin, and stem cells of skin and intestine can't replace each other, or skin.

Commitment to differentiate into just a certain few cell types was as much implied by the phrase "stem cell" as their lack of visible differentiation, and their rapid growth.

The meaning of stem cell has now changed to mean any undifferentiated cell that might be able to be stimulated to differentiate and replace any desired cell type of the body, for medical reasons. For example, nerve cells of damaged spinal cords might eventually be replaceable; And noggin, etc. proteins might turn out to be adequate signals to stimulate stem cells to differentiate into nerve cells, instead of skin or blood.

The great success of bone marrow transplants and skin transplants implied that lots of other tissues can possibly also be repaired with "stem cells", regardless of whether those other tissues are normally replaced by undifferentiated, fast growing cells. In other words, stem cells are not a recent discovery, but a change in vocabulary to fit hypotheses. A first step in getting undifferentiated cells to repair damaged tissues is to use the phrase "stem cells" to refer to what you hope will be possible to develop.

Embryologists have been hoping for such goals since before I was a graduate student. But you need to get undifferentiated cells, and then learn how to stimulate them to differentiate into whatever you need to replace.

[My own first effort was to dissect out undifferentiated cells from asexual buds of sea squirts, grow these in tissue culture, and study how/whether these undifferentiated tissue culture cells formed organs. The two reasons for using sea squirts were #1) They are members of the same phylum as we are (Chordata), and #2) They continually replace their whole bodies from stem cells, about once a week. I only got as far as making tissue cultures of stem cells of the tunicate Botryllus , but I didn't discover methods to induce differentiation. After a year of trying, I switched to studies of cell adhesion and forces.]

How to make geometric patterns? Most hypotheses use diffusion gradients.

Also, embryos have real gradients, not all of which are diffusion gradients.

Examples of gradients:

* Gradually increasing diameters and packing density of yolk granules near the vegetal pole of frog oocytes. (evidently not a diffusion gradient)

* Gradients of attractant substances in chemotaxis. (Is caused by diffusion)

* Gradual spatial and time changes in transcription of hox genes.
(which can't be due to diffusion unless the RNA and/or proteins can diffuse from cell to cell, through the plasma membranes; and also if they don't get carried around by blood flow)

How to Simulate Diffusion? (mathematically, or by computer program)

What mechanisms can make the same patterns as diffusion does?

Repeated changes in local properties that make them equilibrate, or otherwise become closer to the average properties of their nearest neighbors

"The Parable of the Nearsighted Conformists"


The take-home lessons:

Most theories of embryological pattern formation depend on diffusion gradients. (Ignoring stirring effects of blood circulation; ignoring how substances can diffuse from cell to cell; ignoring gradients in properties that can't diffuse)

Scientists should study behaviors of actual gradients, not just imaginary ones.

Please invent theories to explain how co-linear transcription of hox genes looks so much like diffusion gradients (How else do they form gradients).

Explaining this could be as important as Spemann's induction discovery.

As a specific example, suppose that cells inside developing embryos somehow detect which hox genes are being transcribed by next-door cells, and transcribe Whichever of their own hox genes are closest (along their chromosomes) to the hox genes being transcribed by the other cells closest to them.

The resulting patterns of hox gene expression will look as if hox gene m-RNA were diffusing from cell to cell.

This would be equivalent to the parable of the near-sighted conformists.

Any time changes are made that cause variable properties to change so as to become more like the average properties nearby, then the result will closely resemble a diffusion gradient. (Whether or not anything is diffusing)

We shouldn't assume every gradient is caused by diffusion.

Every theory (Turing's, Wolpert's) that assumes the existence of diffusion gradients should be re-examined to determine whether some different kind of spatial interaction is "simulating" diffusion.

This insight led to what is now called "Biomechanical Theories", in which pulling and pushing forces serve the function of diffusion: But the principle is even more general that that.

Even non-quantitative variables can produce geometry, if some are between others. Nearsighted conformists could create gradients of stylishness. Between-ness can substitute for amount-ness.

Researchers need to invent or borrow criteria for distinguishing causes of spatial variations, instead of always assuming diffusion gradients

Point #1) Embryological theories about pattern formation (almost?) always depend on diffusion gradients.

Wolpert's "Positional Information".
Turing's "Reaction Diffusion systems".
The wave-front of the "Clock and Wavefront Theory" is a gradient.

Researchers need to invent or borrow criteria for distinguishing causes of spatial variations, instead of always assuming diffusion gradients.

Point #2) Many real gradients occur in embryos.

Gradients of concentrations of ephrin and eph proteins (adhesion proteins in neural projections)
Gradients of diameters of yolk platelets in frog oocytes.
Gradients in amounts of transcription of hox genes.

Notice that none of these three is a DIFFUSION gradient. Their causes are not known.

Point #3) To make smooth diffusion gradients isn't as easy as theoreticians imagine.

Flow of blood will carry small molecules away. (Acts as a "sink")
Many variables (Like yolk granule size, Ephrin concentrations, m-RNA amounts per cell) can't diffuse.
Large proteins can't diffuse from cell to cell. (syncytial arthropod embryos are a special case)
Even diffusion through paper doesn't produce smooth gradients.



Diffusion isn't the only way to produce gradients.

These other methods should be included in our theories. (They may be more important than diffusion)

HOW ELSE CAN GRADIENTS BE CREATED? (In addition to, or instead of, diffusion)


back to syllabus