Many people think that the immune system works by detecting which molecules are "non-self", and attacking them. This is a misunderstanding. It is not just an over-simplification, either. The truth of the matter is much more interesting, and is essential for understanding many medically important phenomena.

The shape of each antibody binding site is caused by the amino-acid sequence of that part of the antibody binding site. Each B lymphocyte makes billions of antibody molecules, all of which have the same shaped binding sites as every other antibody molecule made by that different lymphocyte. Other B lymphocytes make antibodies all of whose binding sites have other shapes. As a result of the randomness of the splicing and mutation, many B cells are produced whose antibody binding sites fit normal molecules of the body. These "anti-self" B cells are selectively weeded out, either by selective death or selective inactivation. In other words, when an embryonic lymphocyte makes contact with a molecule that exactly fits its binding site, that lymphocyte gets weeded out. Notice that this elimination does NOT result from the self-ness of whatever molecule fits the binding site. Contact with synthetic molecules can cause just as much weeding out, if these molecules fit the binding sites. So can contact with proteins from other animals, which was how the phenomenon got discovered.

The end result of this selective weeding-out is removal of all lymphocytes whose binding sites fit molecules present in the embryo. The remaining lymphocytes, that don't get weeded out, are those whose binding sites fit anything other than "self" molecules. But the reason those lymphocytes survive the weeding out process is not because they bind to non-self molecules; it's because they didn't bind to self molecules. This is kind of the opposite of that mistaken idea that the immune system detects which molecules are self, and leaves them alone. It's not the "selfness" or the "non-selfness" that controls whether lymphocytes will remain, it's how early during development lymphocytes encounter a molecule that fits its binding site. The reason we don't attack our own molecules is because we weeded out all the lymphocytes whose binding sites fit them.

As regards histocompatibility antigens, their function is not to serve as labels for "self" cells or molecules. Really, their function is to hold whatever peptides are around as one stage of deciding which molecules should be attacked. They got discovered by research on organ grafting. When a grafted organ has different histocompatibility antigens, as compared with the animal into which the graft is being put, the result is that the graft gets intensively attacked by the immune system of the latter ('host") animal. But being attacked is not the function of these proteins. It is not too much of an over-simplification to consider that "non-self" holding proteins get attacked as if they were holding "non-self" peptides.

Most importantly, as regards autoimmune diseases, we should learn to resist the idea that what goes wrong is the formation of lymphocyte binding sites that fit molecules of our own body. Their production is the inevitable price that has to be paid for using random splicing and mutation as our methods for producing the shapes of binding sites. Because randomness is used, we can't help generating just as many anti-self lymphocytes as we do lymphocytes specific for any particular molecules. We should be more surprised that auto-immune diseases are not much more common, and that there are not thousands of different autoimmune diseases, instead of just a few dozen.

Autoimmune disease results from failure of the weeding out, not from failure of the DNA splicing and mutation. There is nothing abnormal about that splicing producing binding sites that fit self molecules. Every vertebrate embryo produces billions of what would become anti-self lymphocytes, unless they get weeded out. Please don't think of the splicing as what goes wrong. It is a normal necessity for resisting germs, even germs never previously encountered by anyone.

Incidentally, you might wonder if any other physiological processes also use random splicing of DNA. The answer is: Probably not. The RAG genes that catalyze this splicing seem only to be transcribed in early B and T lymphocytes. Nevertheless, there are some reports of these or similar enzymes in the brain (which could have very exciting implications, if true).

Suppose that the wiring of parts of the brain were somewhat random, and that something analogous to "weeding" was the basis of learning and/or memory. Traditionally, brain functioning has been imagined as analogous to writing on a blank page (tabula rasa). Suppose the truth is some kind of reverse of that. Maybe learning will turn out to be selective erasure of what starts out as pages covered with random writing.

If ideas like that excite you, take a look at the book "Neural Darwinism" by Gerald Edelman or simply read the Wikipedia article about Gerald Edelman, himself. It's not about the nervous system evolving by natural selection.

--------

Researchers have discovered that cells of early embryos of mice and rats can be mixed together and will develop into surprisingly normal-looking animals. Such animals are fundamentally different from genetic hybrids, in that rat and mouse chromosomes are never together in the same cells. Animals made of mixtures of genetically different species are called "chimeras", and can be useful for some experiments.

For example, tests have shown that neither the rat or the mouse cells form any antibodies against each other. Both B and T lymphocytes develop, and some of these have only mouse chromosomes, and others have only rat chromosomes. If the immune system really worked by detecting difference between "self" and "non-self" cells, then would mouse T and B lymphocytes attack rat cells? And would you expect that rat lymphocytes would make antibodies against mouse cells? In fact, neither attacks the other. What would you conclude from these facts?

Please consider another experiment: first make a rat-mouse chimera, and later make skin grafts from mouse skin and rat skin to the chimera. Would you expect that either of the skin grafts would be "rejected", in the sense of being attacked by the chimera's immune system. How would you explain the lack of such immune rejection?

Next, imagine if you grafted skin from the chimera into rats and into mice. Would you expect any of these grafts to be attacked by the immune system? What if some grafts were rejected, but other grafts were not rejected? What is the most likely interpretation for such results?

Another kind of grafting experiment would be to graft mouse tissue culture cells onto rat embryos at various stages of development. Then, when these rat embryos have grown up into rats, would you expect their immune system to attack grafts of mouse tissues?

Probably the result would differ, depending on the stage of embryonic development that had been reached by the rat embryos at the time the mouse cells had been grafted into them. What could you learn from such experiments about the stage at which immune tolerance develops?

Some experiments of great historical importance consisted of making skin grafts between pairs of twin cattle, for the purpose of finding out which were identical twins ("monozygotic" twins is a synonym for "identical"). The experimenters assumed that grafts between non-monozygotic twins would be rejected (= attacked by the twins' immune systems). The were very surprised to observe that none of the pairs of twins rejected grafts from the other member of that pair. In other words, the results seemed to be that all the twins were monozygotic.

Instead of ignoring these unexpected results, they invented a hypothesis to explain them in terms of the development of immune tolerance, and eventually shared a Nobel Prize for this achievement. Can you figure out what their hypothetical explanation probably was?

Incidentally, they would have gotten different results if they had studied twins of other species, including humans. Please suggest one or more hypothetical explanations. In other words, non-identical twin people are usually not tolerant to grafts from their twin.

Can you figure out what would need to be true in order for non-identical twins' immune systems not to attack tissues grafted from their twin?

Hint: there are at least three different kinds of possible reason: Having to do with genetic similarities between their parents: Having to do with the permeability of their placentas: having to do with cells getting transferred from each twin to the other during early development.

Explain how each of the following could result in an autoimmune disease:

A) If no molecules of a certain protein get synthesized until after birth. B) If no lymphocytes penetrate to the locations of certain proteins or other chemicals until later in life (for example because those proteins are tightly wrapped in myelin, and/or enclosed within the blood-brain boundary) C) If some lymphocytes delay their DNA recombination until abnormally late in development or after birth. D) If lymphocytes whose binding sites have certain shapes are for some reason unable to undergo apoptosis. (Like maybe they bind to some abnormal location in the body, where apoptosis is inhibited?)

For which of these preceding causes would you expect the harmful self-attacking lymphocytes to be 1) Only B lymphocytes? 2) Only T lymphocytes? 3) Both B and T lymphocytes? 4) Sometimes both B and T, but in other people only one or the other?

If drug-induced Lupus is really Lupus (not just a different disease that has the same symptoms) then what possible changes can these drugs be causing in certain lymphocytes? 1) Reactivating DNA recombination? (creating new genes for lymphocyte binding sites) 2) Stimulating lymphocytes to penetrate into locations (like inside myelin sheaths) that usually are inaccessible. 3) De-repressing lymphocytes that had been de-activated as a result of embryonic contact with "self" antigens. 4) Try to invent one or more possibilities. *5) Bringing dead lymphocytes back to life, even though they had undergone apoptosis, induced by contact with molecules of antigens that fit their particular binding site. The point is that apoptosis is irreversible, whereas drug-induced lupus gets interpreted as a temporary reversal of tolerance, which should be impossible unless tolerance consists of some kind of inactivation that is reversible.

Lupus is caused by mixtures of lymphocytes that specifically bind many different molecules. This is in contrast to other autoimmune diseases, in each of which different antigens get attacked.

When you do an experiment, and the results seem impossible, what should you do? a* Ignore the results? Just don't tell anybody about them. b* Make a careful list of all the assumptions you have been making, and figure out which assumption would need to be changed in order that the results would not be impossible? c* Figure out what mistakes you might have made (using an impure chemical, grafting the wrong cells) d* Submit a manuscript to a less-competitive journal, including the statement that you don't understand what caused the results. e* Repeat the experiment two or three more times, and then submit a paper reporting what you can't explain. f* Submit a manuscript to Nature or PNAS, claiming that your results disprove the dominant paradigm? g* Something else? Propose that it's caused by thermodynamics.

A great mathematician suggested that some areas of experimental research get systematically ignored because big differences in results can be produced be very small differences in initial conditions. This might be especially true of self-organizing systems that use small triggers (or even random Brownian movement) to initiate big changes.

Some scientists conclude that such phenomena are too "complicated". Others tend to believe that researchers must be testing the wrong variables. Another opinion is that measurements need to be more quantitative in order to fit the correct equation. A few scientists realize that some equations in