Antonio Bandeira - 6:52pm May 15, 1997 (#18 of 47) In response to Rod (#15): One major reason to put the questions was to hear alternative explanations to the one we offer with our model, in order to be able to explain observations that we think are important. Although we think that they are not satisfactorily explained by current alternative concepts, it is my concern that I may not be correctly counterarguing in my mind on the behalf of the alternative models. I don't recall seeing those observations debated in a convincing way, so this would be a good opportunity to put them forward. But you are probably quite right that some more theoretical clarifications should be available before, in order to make things progress. So let's go. In Day 2's message 29, you asked whether it would be fair to summarize our two views as follows: ours based on dominant suppression to achieve self nondestruction and yours based on dominant immunity to achieve nonself destruction, with self being simply forgotten (recessively deleted). I do think it is a fair summary, and certainly an important start. I also do agree entirely with you that the classical difficulty of dominant-suppression models is how to explain the shift from tolerance to immunity - that is, briefly, how does the system overcome this physiological suppression and generate destructive immune responses? This is a fundamental problem to which we think our model offers an explanation. Conversely, I would say of your view that it would be more difficult to explain why, for example, after an efficient immunization with a self antigen like MBP (we know it is efficient because the mouse gets sick), the system develops an immune state of memory to resistance and not to disease. How do we explain then the shift from tolerance to immunity? You agree also (#11) that definition of self antigens is the result of historical learning and, in most cases at least, persistence. Persistence, however, is not enough; nor is it a guarantee for tolerance in the presence of a given antigen in the critical period of early ontogenetic development, as Medawar himself demonstrated (1958). While risking some too-abrupt shortcuts, we think the shift from tolerance to immunity is at the start established during the ontogenetic development. For a given tissue-specific self ligand that was always present in the organism, we say that the pool of peripheral T cells that can recognize it contains both naive (potentially aggressive) cells and memory, effector-regulatory cells. In contrast, the pool of cells that can recognize a foreign ligand (still to show up in the organism) is only made of naive cells, without the counterpart of regulatory cells. One of the key events that follows a viral infection is the drastic change in the repertoire of MHC-bound peptides advertised on the surface of the infected cell, and now many self peptides are going to disappear and be substituted by a viral peptide. We argue that this could now hide the ligand for regulatory cells that recognize a given tissue-specific antigen and so negatively disbalance the local ratio of aggressive to regulatory cells. A second main difference between self ligands and this
viral peptide is related to the density of the ligand. I was always impressed
that up to 1-5% of all class I molecules can now be occupied by the same
ligand. This is not at all the case for self ligands, I believe. (It seems
to me that the way a lymphocyte is going to "sense" ligands is of fundamental
importance, and I believe that the system is "tuned" right from the start
- that is, thymic positive selection.) By the same token, the reason that
the immune system is so bad with tumors is exactly because most of the
so-called tumor-specific antigens are in fact self ligands (so they bring
in locally regulatory cells), and those that are truly neoantigens are
probably at low copy numbers, in contrast to the example of the viral peptide
density achieved during infection. Although for our model we don't need
to postulate a qualitative difference of APCs into professional and nonprofessional,
the game of ligand density and coreceptors could contribute to favor the
aggressive compartment. The system of regulation is, however, robust and
dynamic, and this easily accommodates the observation that after immunization
of a normal mouse with MBP, the mouse is primed for resistance. However,
if the number of naive cells is too high, the same protocol of immunization
kills the mouse in experimental allergic encephalomyelitis, as shown by
Lafaille and Tonegawa using anti-MBP TCR transgenic mice. One more point
on densities and function: Ana O'Garra in Dinax has shown that TCR-transgenic
T cells interacting with an homogeneous APC population presenting the high-affinity
ligand for this TCR make different interleukins according to the density
of the ligand. A potentially aggressive cell (no matter if against self
or nonself)can thus be selected into different functions.
Rod Langman - 7:07pm May 15, 1997 (#19 of 47) Excellent, Antonio. I'd say that suppression is real but
part of the class-discrimination process rather than self-nonself. The
phenotypes at our crude levels of observation are roughly the same. The
big question is always, How does the balance get tipped to tolerance or
immunity? What is being measured, and can we also measure this experimentally?
I'd like to digest your comments before making a more detailed response.
Antonio Bandeira - 7:36pm May 15, 1997 (#20 of 47) Rod, I also now have to read your "alarm" message. Just
a quick comment that I forgot in the previous message: I think that our
two views being "dominant," that is, immediately implying quantitative
balance, I am prone to think that maybe the distance between them is shorter
than what we may think. I thought, reading AAR, that the first wave of
regulatory T cells (TE selected) could solve your problem of the induction
of the first helper cell.
Zlatko Dembic - 1:11pm May 16, 1997 (#21 of 47) In response to Rod (#17): In discussion about cross-reacting antigens as tolerance breakers, when using a simple example of an antigen made up of S and F epitopes, a statement was made that the appearance of first common and then unique activities to S tolerogen is observed after breaking tolerance by immunization with decreasing levels of SF cross-reacting antigen. It also has been noted that alarmist theories would predict the simultaneous occurrence of common and unique responses. I think it would not. Here is the explanation: The B cells are not the problem.
Both AAR and integrity work fine there under the similar premises. The
naive T helpers - let us call them, with your annotation, "iTh" - would
be tolerized (killed) by meeting signal[1] only in the presence of tolerogen
S. After SF immunization, signal[3] provides anti-F iTh to help B cells
that can now respond to both anti-F and anti-S. This is possibly the reason
why simultaneous occurrence was predicted by Rod. But the net anti-S potential
of all B cells present in the body would be negligible or small at the
beginning of the tolerance break. This is so because B cells reactive to
anti-S were also deleted by meeting tolerogen (signal[1] alone), and the
newly generated virgin B cells (which were not yet tolerized) would be
in smaller numbers. With anti-S B cells being either not existent or not
as abundant as the anti-F or anti-SF common B-cell population, with decreasing
levels of tolerogen, they would show up later and thus conform to the observed
finding. The iTh with anti-S would be the same. Initially, they are in
lower numbers than anti-F. With the immunization, only the newly generated
anti-S cells could be activated by the SF cross-reactive antigen, and so
it would take time to reach the number of their counterparts. In conclusion,
both models can explain these observations equally well.
Zlatko Dembic - 1:12pm May 16, 1997 (#22 of 47) Another point: "Alarmist" is a word describing one part
of the process and not a state that should be discriminated, and it might
be inappropriate to use for "integrity." Thus I urge that we preserve the
distinction between infectious self-nonself, danger, and integrity models,
because from my viewpoint the last of these offers something different
(it can integrate a part of the Coutinho and Bandeira viewpoint also).
I would like to clarify this: Integrity allows modulation of the two basic
signals [1] and [2] by the microenvironment. Signal[1] can be modulated
by presenting different sets of peptides by different integrity-disruption
mechanisms (including cellular polarity of the antigen presentation). Signal[2]
can be different for memory cells and might involve active mechanisms for
eTh to die: For example, an eTh that needs only signal[1] to express its
function would be killed by meeting a cell that also provides signal[2]
(perhaps this could be envisioned as a premature switch of CD28 to CTLA4).
The net result of any response to a dendritic cell's activation attempt
would then critically depend on the relative ratios in populations of naive,
effector, and memory cells. Their competition for the epitopes presented
by APCs would determine the winner. That is how, now, our alarmed dendritic
cell would tolerize (kill) cells of the same specificity, instead of providing
the activation to naive or memory cells. The effectors would still be effectors,
because on target cells they would find only signal[1] and, therefore,
exert their function. But because dendritic cells are the most efficient
presenters (in quantitative terms), they might seem to be the most important
mediators of activation and, as explained, tolerance. It would be the same
for the B cells. Furthermore, without the appropriate signal[2] for the
memory cells provided by the dendritic or any other APC, the memory cells,
if abundant, might outcompete naive cells in accessing the same epitopes.
Thus, the net result might be unr esponsiveness with prolonged survival
of these memory cells, something similar to the C and B regulatory cell
hypothesis. In the end, it seems important for an organism to generate
the appropriate signal[3], which in turn would send the appropriate signals[1]
and [2] for the right cellular population. This could be enhanced by allowing
migration of APCs to lymph nodes or spleen, where the majority of T cells
might be of naive or memory type if activation or reactivation is desired,
respectively. That may be why alarmed skin dendritic cells migrate to lymph
nodes, where the newly thymic emigrants - Mel-14+ T cells - await the current
"weather report" from the boundary.
Rod Langman - 4:47pm
May 16, 1997 (#23 of 47) Rod Langman - 5:05pm May 16, 1997 (#24 of 47) In response to Zlatko (#22): I'm getting the strong impression that the decision we all agree has to be made, the one that allows the immune system to rid pathogens and not rid itself, is ending up under alarmists' models in the hands of the APC. The APC does not discriminate between processing self components and nonself components, and yet it is being asked to turn on responses to cells in the nonself-reactive compartment and not the self-reactive compartment. I should add that I understand (I think) the positions laid out clearly by Ephraim, that in the presence of an alarming event both antiself and antinonself are locally induced to immunity, and that the little bit of antiself we know to exist is no more harmful than the alarming event itself. Moreover, Ephraim argues that immune elimination mechanisms are not in and of themselves alarming. However, if the steady-state load of infectious encounters is continuous, even though failures of immunity allow expression of disease only occasionally, then a constant level of alarm is sounded. From a purely practical observation, cell-mediated immunity
in the form of a delayed type hypersensitivity reaction is a large local
inflammatory reaction, and antibody-mediated elimination of antigen is
no less inflammatory in immediate hypersensitivity reactions. The apoptosis/necrosis
distinction is a very narrow one, because the immune response includes
the recruitment of all manner of non-immune effectors, which in many cases
are in the end responsible for eliminating the pathogen.
Rod Langman - 6:19pm May 16, 1997 (#25 of 47) In response to Antonio (#18): With regard to the persistence of antigen being a necessary, but not sufficient, condition for self tolerance, the antigen has to be present before the first eTh and preferably the first iTh. Exactly when this occurs in embryogenesis I'm not sure, but as I recall, the mouse thymus is formed around day 12 and is exporting cells by days 14-15; for other species, I have no information at my fingertips, but clearly birth is not a particularly useful marker for immune development. The MBP example you refer to seems to be to be a case of switching the class of the response, just like an immunization that starts with IgM and switches to IgG, and then the IgG inhibits the response in the IgM class. I think that transient disease is CMI (the effective class) and resistance is antibody (an ineffective class). But what is the evidence of tolerance to MBP? As far as I can tell, MBP is normally invisible, but activated cells can cross the blood-brain barrier and cause local inflammation and partial temporary paralysis. As an aside, I think there is a rule in the immunology databank, filled with obscured oddities of dubious general applicability, that there exists for any given notion, model, or theory two lists: one of facts that support the idea and the other of antifacts that contradict the idea. However, when all the chaff is blown away, there are precious few grains to form the substantial food for thought. It would be nice one day to have an edited list of observations with all their experimental limitations that form the foundation of stuff we really know (not, as Bill pointed out with low zone tolerance, stuff we merely believe with religious fervor). Here I will pool my comments on immune regulation, including a response to these issues as raised in Day 4, message 7. First, let me see if I have the picture right. In embryonic life, the first T cells that arise are in a state that allows antigen to amplify them into a suppressive state; these are the antiself T-cell set. I have to assume that this is a result of the state of the newborn cell, not a set of pacifist hormones. There are some T cells arising at this time that fail to interact with antigen, and so they just sit patiently waiting, unamplified and unsuppressive; these are the antinonself T-cell set. Much later in life, a pathogen arrives and interacts with the waiting antinonself T-cell set and starts to drive them to suppression - oops, no, it must be immunity; so, while the antinonself T cells have been waiting, they have undergone an antigen-independent maturation and become unable to generate suppressors, only immunity. Or maybe the pacifist hormone option applies, and as the animal gets older, the pacifism gives way to aggression and immunity. In either case, any antigen that is a mixture of self and nonself epitopes will dominantly suppress the antinonself response. The example you cite of a viral infection displacing self peptides and replacing them with viral peptides would not help me explain the kind of result that Bill finds with serum globulins and that I described in some detail. Reduced to a long sound bite: The immune system has to
make a choice between responding or not responding to any given antigen.
The system has two options in this dominant-suppression model, the cell
has two options in the AAR model, and the APC has two options in the alarmists'
models.
Zlatko Dembic - 8:40pm May 16, 1997 (#26 of 47) In response to Rod (#23):
I still see no problem. The threshold for tolerance for iB is higher than
for iT. So when the first available iB of the anti-U and anti-C kind is
available for receiving help, there is unfortunately (lower threshold)
neither iTh anti-C nor iTh anti-U around yet to help them. It follows that
anti-F Th effectors can help only anti-C iB cells and not anti-U iB cells
(because these do not present the F epitope). Thus, antibodies to C emerge
before antibodies to U.
William O. Weigle - 8:56pm May 16, 1997 (#27 of 47) In response to Rod (#14): The mechanism that prevents our own serum proteins from being processed and activating APCs is most likely the same as that preventing heterologous serum protein from being processed and activating APCs. Whether processing occurs is governed by the physical nature of the protein. Thus, some discrimination is at play in the processing of pathogens and the failure to process our own serum proteins. I postulate that there is only one signal[2] and that it is generated by a series of events including processing of antigen, activation of the cytokine cascade, and upregulation of costimulatory elements. I object to Ephraim's necrosis/apoptosis argument in that cytokines are released and T-cell activation occurs in the absence of necrosis or apoptosis. However, in a situation where necrosis/apoptosis results in activation of APCs and cytokines are released in the microenvironment of the T cell-APC interaction, then a second signal would obviously be delivered. However, much more subtle processing is sufficient to trigger the events needed for signal[2]. There are many ways that dampening of the immune system
in vivo occurs. A finely tuned in vivo regulatory system is an essential
part of the immune response. As a result of cross-regulating among T-cell
subsets, elements of T-cell reactivity are dampened to allow expansion
of others. Such dampening is responsible for selective subset usage in
CD4+ cells and obviously dictates what arm of the immune response will
prevail. Intervention of such dampening may have therapeutic applications.
Rod Langman - 9:26pm May 16, 1997 (#28 of 47) I'm still not quite sure why there is a kind of gradient in the sort of self markers that make self molecules, even if adjuvant, etc., is nonimmunogenic, near self tolerogenic unless mixed with adjuvant, and far self (e.g., chicken gamma globulin) very hard to tolerize and not too hard to immunize without adjuvant. I'm also not sure why you object to using eTh to ignite the cascade of events we can lump for the moment as signal[2] in the style that you use to describe these factors. The eTh is a symbol for some antigen-specific regulator that permits an immune response, and in the absence of this permission, no response can occur. |