Unsolved Problems Nov 13, 2017

I) More about bioassays: Sometimes their method of measurement isn't specific enough. For example, a researcher who was trying to purify cell-cell adhesion molecules (like cadherins turned out to be) used larger size of cell aggregation as his measurement of concentrations of the hypothetical substance that he was looking for. Unfortunately, many cells were killed in the process of separating cells, and the DNA they released is sticky. Therefore his bioassay led him to purify DNA.

Another example was that optic nerve axons were thought to be guided by concentrations gradients of adhesive proteins. Much evidence seemed to support this, but didn't identify or purify the adhesion molecule. Now it is turning out that there are gradients of proteins that stimulate de-adhesion of nerve axon tips (Ephrin proteins and their receptors, which got discovered by accident because a line of cancer cells secreted them).

II) Genetic Screens have produced more important biological discoveries than any other strategy in the past 50 years. Suppose that you wanted to identify all the proteins used by an alga to move toward brighter light. You could start (#1) by causing lots of mutations in a population of your chosen species of alga. Next you could put a few million individual algae in the water at one end of a chamber, and (#2) shine a bright light on the other end, sufficient to attract ~ 99% of the algal cells to the bright end .

Paradoxically, (#3) you then collect from among the rare one % of algae that failed to move to the bright end . Your goal is to collect as many different mutant algae as you can (algae with as many different mutations as you can find that either prevent the algae from detecting the light, or prevent them from swimming, or cause their swimming not to be directional, or any other abnormality that prevents or reduces movement toward the light.) This method can reveal phenomena that you hadn't even imagined.

You might find a hundred different mutations that block movement toward the light. If a hundred different proteins participate in any way to the algae's detection, orientation, swimming, or anything else, then you should find mutant strains corresponding to each of those hundred proteins. In fact, you should keep looking until you isolate a dozen or so different strains for each one of these proteins. That way you can be more sure that you have found all the proteins needed for light responses or any other phenomenon you study by this approach.

Notice how the algae were used to separate mutants automatically. All the non-mutants swam away from where you first put them. You didn't need to look at one alga at a time, and try to notice abnormalities. Therefore, even if only one in a billion were mutated in a way that caused them not to move toward the light, you would find that one in a billion. Mutant strains of "model organisms" are valued more than gold or diamonds by geneticists. Even more valued are ingenious methods ("genetic screens") by which mutant organisms can be caused to separate themselves. Very much brain-power has been focused on inventing new screens, which have been the foundation of entire careers.

III) Please notice that lethal mutations are a blind spot of the genetic screen strategy. If mutating a certain gene causes the organism to die, then screens can't find such mutants. Some of the most important genes and proteins can be missed.

A great advantage of the alga Chlamydomonas is that it can live either by photosynthesis or by absorbing acetate as food. Other plants would die if mutated in any way that prevented photosynthesis. That would prevent genetic analysis of photosynthesis. Another advantage is that they use flagella to swim, but mutations that paralyze flagella are not lethal because immobilized algae can live by photosynthesis.

The world's greatest and craziest chess players don't focus their minds as sharply as geneticists.

IV) Some specific problems that puzzle me: Two uses of the word dermatome.

What guides sensory axons to consistent stripes (called "dermatomes") in the skin.


Embryologists use this word "dermatome" to mean the part of each somite that underlies the developing skin, the cells of which become the inner, leathery layer of skin (called dermis; leather is cow dermis.) Incidentally, the painful medical condition called "Shingles" gets its name from the geometric pattern of striped bands of inflammation and blisters that develop across the chests or older people who had chicken-pox as children. Each inflamed stripe marks the location of a dermatome.

What is the causal linkage, if any, between the little sheets of cells on the tops of somites and the broad and elongated areas of skin surface innervated by particular sensory nerves? Both are called dermatomes. Did neurologists and embryologists separately invent the same word? Does each dermatome of the skin develop from cells that had been part of particular somite? Or maybe what matters is the guidance of sensory nerve fiber outgrowth, with each segmental spinal nerve innervating only tissues derived from one particular somite.

By means of what sorts of experiments (Or mutations? Or birth defects? Or virus infections?) will the causation of dermatomes eventually be discovered?

V) What mechanism causes the gradient-like spatial patterns of hox genes, sonic hedgehog, wnt bicoid and other transcription factors in developing embryos?

These were discovered by genetic screens followed by in situ hybridization as a means of selective marking of locations where particular mRNAs are concentrated, not to forget antibody staining of locations of particular proteins.

Because flies and most arthropods continue to be syncytial up through about the stage of six thousand nuclei, it wouldn't be surprising for proteins to diffuse freely through the early embryo.

The earliest stages of vertebrate embryos, in contrast, separate their cytoplasm into separate cells, with two plasma membranes to cross in order to diffuse from one cell to the next. Large proteins shouldn't be able to leak from one cell to the next, let alone distances of dozens or hundreds of cell widths?

Another problem for diffusion gradients in embryos occurs at later stages when the heart starts pumping blood, because all diffusible chemicals will be carried along with the blood and lymph, quickly smoothing out any diffusion gradient.

Nevertheless, labeling with antibodies and complementary-sequence nucleic acids often shows what looks like gradients. How are they formed, and preserved over time? Let me suggest a possibility that I thought of but haven't published. The idea comes from the algorithm used in computer simulations of diffusion.

You make a grid-like spatial distribution of numbers, which can be in two or three dimensions. Each number represents the concentration of a chemical at a point, and the adjacent numbers represent concentrations at adjacent locations. Things like this are called "finite element" simulations, because they apply equations to finite numbers of point locations. They can be very accurate.

Each number was recalculated by averaging the neighboring numbers around each point, and then comparing each number with the average of its neighbors. Sometimes the average of neighboring numbers turns out to be lower than the number they surround, other times the average of neighbors is the same, and some times the average of neighbors turns out to be more than the number they surround.

See http://www.albertkharris.com/number_diffusion_1.mov for an example.

How the program simulates diffusion is by changing each number to make it closer to the value of its neighboring numbers. It works quite well, even if you only use the 4 closest neighboring numbers. And now, the concept! Imagine that cells in a sheet could measure the concentrations of some chemical in the cells immediately adjacent value, and then either synthesized or destroyed enough of that chemical in their own cytoplasm. The result would exactly mimic a diffusion gradient!

This is probably not how concentrations of signal molecules are controlled, but maybe it is. It could rapidly produce gradients of many different geometries, over distances of as much of centimeters or more. Repeated recalculations based on comparisons of neighbors produces patterns that look just like diffusion gradients, and their patterns automatically scale (adjust proportions, make scale models and all that.) It can even do that using variables that are not quantities. It is sufficient that properties can be compared. A test of this idea would be to use radioactive labels to determine whether the gradient molecules are synthesized only at the high end, or are synthesized by all cells along the length of the gradient.