Zlatko Dembic - 2:18am May 14, 1997 (#9 of 47) 

Integrity and danger hypotheses operate under the same dialectic principle and have nothing to do with the self-nonself discrimination principle. If, as originally proposed by Matzinger (1994), the immune system should discriminate between dangerous and not dangerous "intruders" or agents from the microenvironment, then the problem lies in what (who) decides on what is dangerous and which biological mechanism should be employed for this. Thus, the original danger model would suffer from the same problem as the self-nonself discrimination - which is what Rod was trying to point out. Following this thought, only a germ-line-encoded switch (decision to describe something as dangerous) could do so, and all that is dangerous would have to bear an uniform marker. If so, all that an immune system would have to do would be to associate the danger marker with the intruder, and consequently there would be no need to recognize anything but the danger. This line of thought is not able to explain, then, how the repertoire of specificities developed during the natural evolution (Langman and Cohn, 1996). But if we incorporate the danger hypothesis into the integrity dialectics, as discussed earlier in my messages from Day 2, it makes sense to have a danger-assessment switch (signal[0] according to Polly; a homologue of signal[3] in integrity) during the immune response's analytic part, because selective pressure is imposed on it in combination with a somatically diversified repertoire of immune activities. The rationale is that the breach of integrity always upregulates the switch as dangerous, and no upsetting of integrity does not (Ephraim defines this as necrotic cell death, or when the inside of the cell gets outside - i.e., after large-scale apoptotic death), and in combination with signal[1] (recognition) and/or signal[2] (costimulation) the selection-competition mechanism either selects or deselects the generation of effectors in and around the space (draining lymph node) affected by the intruder.  

So I would argue that the danger model does not discriminate but rather, like the integrity model, selects the fittest adaptation of the immune response to "parasite." The models differ in the definition of the initiation signal for the immune response - with integrity having a broader definition that may eventually include other than dendritic cells - and further in the way that the heightened alertness (including memory) is being generated and maintained. The mechanistic part of two models might in addition differ to some extent, but this could be discussed in detail within the topic(s) for Day 4. On the general level, for example, danger proposes that effector cells would inactivate themselves by default (after some period of time). This may be an easy way that some parasites might be able to trick the immune system into thinking they contain self antigens, as previously noted by Rod (#1), if the whole mechanism is not coupled to the formation of the memory cells. Integrity provides an explanation for this by including a cross-talk-like downregulation mechanism for the effectors with a formation of memory cells, a mechanism that can, perhaps, also explain a form of peripheral tolerance (and possibly anergy, if it exists). 

Antonio Bandeira - 5:30am May 14, 1997 (#10 of 47) 

I will help today, Kenneth. The French heuristic school sends an exam. Try to comment and explain in your own words (models) the following experimental situations.  

(1) Josselyne (Coutinho et al., 1993) took a day 16.5 embryo of mouse strain A (at which age there is no T-cell immune system) and skillfully grafted it in utero with skin from a day 16.5 embryo of full allogeneic skin B. Newborns had beautiful skin. Everything went perfectly for the next three weeks, until the graft was rapidly rejected by day 23/24 of postnatal life in all mice. She found out that Peter (Medawar and Woodruff, 1958) did a similar type of experiment 40 years ago, and so did Ruppert (Billingham and Silvers, 1960) for male skin grafted in day 0/1 newborn female. The results were the same as those 40 years ago - rejection.  

(2) In contrast to Peter and Ruppert (who used hemopoietic cells when they made one of the most beautiful discoveries in the history of tolerance), Josselyne found out that the problem of the embryonic skin was also solved - that is, it was kept for life - if instead she grafted pure thymic epithelium (TE) from a B embryo of day 10. In this case, she could graft the skin or heart at any time in life; the transplants would always be accepted. How was tolerance achieved to skin- or heart-specific antigens if she only grafted TE? With the help of the Antonios and Yves (Modigliani, 1995), it was found that the peripheral immune system of the TE chimera contained T cells that were immunocompetent to reject the skin. They found it okay because those cells could not possibly have been deleted by the TE, because the specific ligands should not be advertised in the thymus. But they also revealed that the periphery contained CD4, which specifically prevented the others from rejecting. These CD4 were therefore capable of suppression or regulation and had to be TE selected.  

(3) Yves took this regulatory T-cell population and injected them into syngeneic mice that never saw B skin (or any B antigen) and were able to reject this skin immediately. However, the more regulatory cells he injected, the more days the skin survived, until, at a given ratio, skin was no longer rejected. A quantitative balance of nontolerant/regulatory T cells was required. Yet, as soon as he took the regulatory cells away, the accepted skin was soon destroyed. He concluded that the skin-reactive cells were still there and were not anergized. But then he did another experiment. Once again, he took syngeneic mice, but this time he grafted them with B skin at a point at which no peripheral T cells existed. As soon as the thymus put out the first T cells, they could, if they wanted, see the graft. And they did, and the graft was rejected. However, in another group of mice he did the same, but this time his famous regulatory T cells (B specific) were going around in the periphery. The skin this time survived for life. But the interesting thing was that the number of regulatory cells needed was much smaller. The process was much more efficient. Why? What happened now to the host T cells that were under suppression? To make a long story short, Yves showed that removing the regulatory T cells did not end the state of tolerance to B skin. But he could show more: The T cells that came out from the thymus were now able to specifically suppress rejection of B skin by other nontolerant T cells. That is, they themselves became the regulatory phenotype. They were educated. But the funniest part came when he saw that these educated T cells, in contrast to the original regulatory T cells, could not prevent rejection of B heart. Because antigen recognition was required, it looked as though it was the potentially aggressive skin-specific (but not heart) cells that became of a regulatory phenotype. As in all learning processes, it is when we are small, and during a given "window," exactly as Peter (Billingham et al., 1953) showed 40 years ago, that we can be educated to use guns or play the cello (editorial comment). 

(4) Irun (Cohen, 1986) told me a similar story about the coexistence of aggressive and regulatory cells, but this time in normal animals. He told me, for example, that normal adult mice (as well as humans) contain readily detectable anti-myelin basic protein (MBP) T cells. Tough immunizations with MBP lead to the induction of experimental allergic encephalomyelitis. Interestingly, disease is transient and animals recover. But most striking is that if you boost again, nothing happen in terms of disease. The mouse is now resistant, and this state is T-cell mediated and is a dominant process.  

(5) Fiona (Powrie et al., 1994) found the following: If you isolate subpopulations of CD4 T cells from a normal healthy adult mouse according to certain activation/memory markers and transfer them into immunodeficient recipients, one of the populations (resting T-cell type) generates a mess of an inflammation in the bowel of the mouse, with tremendous production of aggressive interleukins, and the mouse dies. She protected the recipient by cotransferring the other T-cell population (activated/memory) phenotype. Protection appears to be TGF-beta dependent, and it seems that the thymus is important for the generation of protective cells.  

(6) Shimon (Sakaguchi et al., 1982) also tells an interesting story about regulation. Neonatal thymectomy causes a big deletion in the number of peripheral T cells of the mouse, which will have to make it with a few dozen of them. Interestingly, in order to try to get more T cells, the mouse expends a lot those he has (a feature that does not happen in normal ontogeny - by the way, what would make them grow, what do they see?). The outcome is that they get multiple organ-specific autoimmune diseases. The T cells can transfer disease into T-cell-deprived mice, but a subtype of CD4 T cells taken from a normal adult mouse again protects the host.  

(7) Joseph K. is a middle-aged fellow, a smoker who likes good salted food, drinks, and women. It is likely that he has no thymus anymore. He all of a sudden has a terrible pain in the chest and is in danger: He just had a massive heart attack. Terrible necroses, inflammation, etc., and he may even develop autoantibodies. And yet, like many millions in the world, he did not acquire a heart-specific autoimmune disease.  

(8) What mechanisms can you think of to understand why some immunologists are trying to tolerize against MBP through the gut, exactly where the immune system is supposed to develop immune responses against bacterial antigens of the flora to protect the organism?  

(9) If you inject newborn mice with allogeneic (mixed bone marrow and spleen) cells and later graft skin, the skin is accepted, exactly as Medawar found many years ago. What will happen if you do the same but this time use allogeneic cells from an allogeneic donor that lacks T and B cells (a RAG/KO mouse) but certainly has dendritic and other professional APCs? Yves did the experiment and found that skin grafts are accepted. You can even use full allogeneic combinations because there is no risk of graft vs. host disease.  

I think that's enough for now.  

This exam was also done in Paris and the answers published (Modigliani et al., 1996) 

Rod Langman - 7:56am May 14, 1997 (#11 of 47) 

First, I need to explain that Bill's office is quite a bit closer to the beach than either mine or Doug's, so it is true, he is much better situated to see real life in action (#7). I know that Bill has been itching to deal with how adults become tolerant, and I agree it is a very important practical problem in clinical medicine.  

For the past few days, we have been focused on practical problems in evolution, and hopefully, if we have understood anything, we should be able to explain all kinds of immune behaviors. But, as Bill points out, there are few antigens that can be given to adults and, as he bluntly put it, "induce solid unresponsive states" similar to those we enjoy to our own body constituents. And, as careful experimentalists would also agree, split tolerance, or what I'd call class switching, is certainly not usefully termed "tolerance"; nor are suppression and other generalized downregulation processes real tolerance. Some of the best antigens we know to be capable of inducing solid adult tolerance are deaggregated serum immunoglobulins of closely related species. The same proteins when aggregated, or emulsified in oil and dead bacteria, induce solid immune responses. Thus, it is not unreasonable to suspect that the tolerogenicity of deaggregated antigen is based on how closely it can mimic normal self components.  

From a practical point of view, I think it would be fair to summarize Bill's interpretation of his experiments along the lines that antigens that avoid being processed and inducing inflammatory mediators of various kinds all tend to push the immune response toward tolerance; conversely, the presence of these agents tends to push the system toward immunity. The next step is to determine how the availability of these inductive stimulants is regulated by antigen. This is where we break up into groups again; although Bill contends that neither Bretscher-Cohn-Langman nor danger provides a convincing explanation, I'd like to try by suggesting that, normally, eTh cells produce many of the stimulants as well as the key signal that allows the iT or iB cell to proceed past the activation stage to become an effector. I want to emphasize normally, because I doubt that in extreme circumstances other pathways of cellular induction can be used to bypass what is normal. Insisting on a single unique pathway or on a rabble of indiscriminate alternate pathways is equally unrealistic. Thus, I'd argue that it is important to distinguish when an override system is in play vs. a normal system. The example of immunity to Listeria in T-cell-deficient mice is not an argument that in normal mice resistance is T-cell independent, though it would have been easy to argue with only a limited amount of information.  

I would like to come back to the problems of inducing and breaking adult tolerance after we have drawn some lines in the sand regarding tolerance in embryonic life. 

Rod Langman - 7:58am May 14, 1997 (#12 of 47) 

In response to Ephraim (#8): Now I'm clear on a few points, but I'm still rather lost. First, it seems that as long as we keep firmly in mind that the categories "self" and "nonself" refer to how the immune system treats an antigen, then the terms remain useful, and despite the constant grumbling, no better or more useful word pair has stepped forward. However, I'd make a suggestion that we could refer to immune self and immune nonself. How about a vote? Meanwhile, out of sheer exhaustion, I'll keep using self and nonself with all previous caveats implied. Also, I did not spell out that the alarms set off by the "thingies to be eliminated" are started by the pathogen and, as Ephraim notes, the final step may be a host cell response ("danger," for example). Sorry for any confusion; my shorthand stopped at the first step.  

If I have understood Ephraim's posting, he is firmly in the historian camp. He has, however, some question about how to measure time. I'd settle for some kind of biological clock to avoid controversy over the endorsement of particular brand of wristwatch. The more important issue is to determine what interval the immune system has to measure. You say it is the history of the individual that matters, but you go on to define the time factor you have in mind in terms of the state of differentiation of a lymphocyte. This made it difficult for me to untangle this crucial sentence: "If the antigen was present early enough during the life of the animal such that there is no immature thymocyte that is not exposed to the antigen, then a state of deletional tolerance will result, and the antigen will be treated as self." Here is my best guess:  

Under the AAR model, the very first iTh cells that arise in the embryo do so in the complete absence of eTh, which arise some time later in embryonic life. How long is later? The time it takes for an iTh to differentiate into an eTh in the absence of antigen (no measure of physical time is useful). I can see no a priori reason to make this time the same physical measure in all animals. It is the effect we are after, namely, to have the iTh exposed to antigen in an environment that is devoid of eTh and thereby ensuring that the antigens that are present will drive the iTh to terminal differentiation and death. The iTh has at all times the potential to be driven to immunity by a mixture of antigen and "help" derived from eTh, making the absence of eTh a sure way of allowing antigen to drive iTh specific for these antigens out of the system. As an aside, all T and B subsets are born in the i-state, and only the iTh can undergo the key antigen-independent differentiation to eTh. This describes the establishment of tolerance to all antigens present in the embryo at the time and with access to the immune system.  

I'm going to pause and post this response so that any disagreements can be aired. You may not like where this is going, but is there a logical failure or biological absurdity revealed at this point? Future steps may show that this proposal cannot lead to an answer to immune self-nonself discrimination, but let's see whether this first step is okay.  

You may not have a disproof of the steps I proposed above but prefer a different set of steps to establish the historical foundation - if so, try to give us a short paragraph summary. 

I'd particularly like Antonio to spell out what he thinks is going on during the early stage of establishing suppression in his historical model, without prejudice to any subsequent steps. 

I realize that an obligatory role for eTh in controlling the induction of immunity in all iT and iB cells is controversial, but I'd settle for some kind of generic controller of your choice, provided it appeared in embryogenesis some time after the iTh has been thoroughly bathed in self antigens.  

I also wonder if there are any hard-core alarmists who reject any kind of historical process as essential to their model. 

William O. Weigle - 10:07am May 14, 1997 (#13 of 47) 

Before dealing with some of the more clinically oriented questions raised by the proposed topic for Day 4, I would like to briefly continue my discussion of different antigens in models of tolerance. As pointed out yesterday (#7), one can readily induce tolerance with heterologous serum proteins, especially if they are presented in the monomeric form free of APC-activating agents such as endotoxin. However, it is extremely difficult to induce tolerance in either the neonate or the adult with viruses, bacteria, intracellular parasites, or their components. In fact, having reviewed the literature, I know of no case of solid tolerance induced to such antigens, even in such cases of chronic viral infections in humans (HIV) or in experimental animals. A solid state of tolerance is not induced even in such experimental infections by viruses such as lymphocytic choriomeningitis virus, despite the ability of the virus to cause destruction of APCs (McChesney and Oldstone, 1989). As I mentioned yesterday, all reported cases of tolerance to such agents are the result of preferential T-cell subset selection. In cases where there is suppression of the immune response by such infections, such suppression results from regulatory mechanisms rather than a true tolerant state. It appears that such antigens are obligate immunogens in both the neonate and adult. Why, then, are there antigens such as serum proteins (in neonates and adults) and allogeneic cells (in neonates) recognized as self despite different epitope usage? It appears to me that it is highly possible that at least some self antigens are recognized as self by a nonspecific mechanism. It may be that this recognition occurs at the APC level based on the physicochemical nature of the protein rather than by its serological difference. These antigens would not be processed by APC, and no APC activation or release of cytokines would occur and thus no second signal.  

I believe, as does Ephraim, that such events as immune regulation, preferential subset usage, and ignorance should not be considered as models of immunologic tolerance. I disagree with him on the nature of the required second signal, which he terms "danger." I cannot imagine how the contents of any damaged cell that are spilled out into its environment can cause any signal to the immune system. If Ephraim would accept that it is activation of APCs and subsequent release of cytokines that cause the second signal (possibly by upregulating costimulatory factors on T cells), he and I would be in closer agreement.