Rod Langman - 8:43am May 13, 1997 (#24 of 32)  

I understand the urge to switch to real life (#23), but whose life is real, especially in California? The real-life host you refer to is a multicellular organism in which cells die either quietly of natural causes (apoptosis) or in a hail of bullets (necrosis), and the organism has figured out a way of knowing when there is a mess on hand - let's call it inflammation in a generic way. It then uses the environmental alarm of inflammation to trigger all the cells able to react with antigen in the vicinity. Now, these immune biodestructive reactions are craftily designed to induce apoptosis as the killing mechanism - if it is a cell that is being eliminated - and, in the case of extracellular parasites, endotoxins, exotoxins, and all manner of inflammatory irritants are simply eliminated, whereas self components are nontoxic and induce no immune response. Or at least I suspect this is close to what either you or Ephraim might argue. As an aside, I believe that complement lysis of cells is necrotic rather than apoptotic, and so the immune system is not all that crafty all the time - and how many mistakes can it afford?  

I am concerned about what happened to all the antigen-specific cells directed at self that seem to be missing when inflammation arrives. I suspect there is some kind of self-nonself discrimination lurking - namely, your Lederbergian negative-selection events. These we need to examine in more detail, as promised by the question for Day 3. 

You bring up a good point with the non-immune defense mechanisms of invertebrates. It is fun to do crude immunology sometimes and inject Listeria into mice and simply start counting bacteria in liver and spleen after the first 5 minutes. I spent a year doing this with Bob Blanden in Canberra and became convinced of the importance of infectious diseases in shaping the immune system and convinced that experimental immunology was damned hard work (hence my deep respect for the experiments of others). To my surprise, most of the inoculum ends up in liver, where 99.9% of it quickly disappears. Then, almost a day later, bacterial numbers begin to climb, and around the 4th and 5th days bacterial numbers either decline and the mouse survives or they continue to increase and the mouse dies. What happened to the original inoculum? It was destroyed by the Kupffer cells in liver sinuses in a quite non-immune manner. But we knew that T cells were important, because these cells transferred immunity to normal mice, yet immunity was not always specific - an immunosuppressive graft vs. host reaction gave solid protection to all kinds of intracellular bacterial infections. So, in the days long ago when neonatal thymectomy was still new and the AT x BM mouse was all the rage, we infected these mice with Listeria and found that they were protected. It turned out that the phagocytic systems of these immune-deficient mice were so greatly activated that they were almost as well protected as immune mice (better than normals). The moral of the story is that these non-immune defense mechanisms do deal very effectively with a large portion of the pathogenic load (resistance to viral infection is dependent on interferons, etc., but this story I know less well). The immune system deals with the remainder that have evaded the non-immune defenses. 

An interesting parallel can be drawn with the complement system. The so-called alternate pathway is activated by zymosan or yeast cell walls or many highly charged polymers at rather high concentrations. The antibody-dependent pathway is activated by very small aggregates of antibody on a cell surface - used to great effect to delete Thy 1, CD3, CD4, CD99, etc., T cells. The two pathways lead to exactly the same set of enzymes from C3 onward to C9 and finally cell lysis. The alternate pathway is a response to massive doses of inducer, whereas the immune pathway is thousands of times more sensitive and requires only traces of antigen to trigger lysis. Thus, I would argue that the immune system is a way of significantly increasing the threshold of detection for a primitive set of biodestructive effector reactions that long preceded the vertebrate immune system. However, the price for this specificity and sensitivity was a somatic self-nonself discrimination instead of the old germ-line-selected self-nonself discrimination.  

To bring this rambling collection of anecdotes to an end, you wrote (#10), ". . . unless immunology is redefined as the study of self-nonself discrimination in any biological organism, in which case it is something different from what I study." 

I'd argue that, indeed, the immune system is founded on an understanding of a somatically selected self-nonself discrimination and that all other germ-line-selected self-nonself discriminations are parts of the highly effective non-immune defense mechanisms. I would not want to give the impression that inflammation and these various non-immune defenses are trivial; they are the stuff that the specificity elements of the immune system absolutely depend on to provide protection. 

Antonio Bandeira - 6:59pm May 13, 1997 (#25 of 32)  

Is the self-nonself distinction (still) important in immunology? If grafted into an adult individual, a foreign (e.g., allogeneic) skin is destroyed, whereas an autologous (or syngeneic) graft always remains healthy. The immunologist sees this difference and says that the organism treats the two tissues differently. Nothing is more natural than to say that the system is able to "discriminate" self from nonself tissues. This is the phenomenon to be understood, and the field is then open to many questions, one of them being how to define self and nonself. To understand tolerance is to be able to explain self-nonself discrimination. To me, the term "self-nonself discrimination" is the best, as it is the least committed to interpretations; the term can simply mean that there is an observed difference, and by now everybody knows what we are talking about. Other terms like "integrity" or "danger" do not make it better, because we cannot escape our own language, structured as a system of references, and so danger immediately implies non-danger and integrity lack of integrity. Also, these terms are already committed to interpretations and inserted in the context of a particular theoretical model. 

Until biological activities can be represented by mathematical functions, we will not be able to escape our own language. As I guess everybody agrees, the problem is not the words per se but the context, and it is in this context that we should look for a meaning. And in which context should we discuss it? Within the most fundamental one in the discipline: the paradigm. For me there are two alternative ones: the Burnetian/Lederberg and developments thereof, of which the brightest are those of Bretscher and Cohn and of Langman and Cohn (sorry, Rod, if I still have difficulties in seeing a major conceptual difference between Lederberg's and the two-signal theory, but I am ready to change), and the most obscure, the danger model. The alternative paradigm or, as Coutinho may say, an embryo of a paradigm is derived from the network ideas of Niels Jerne. 

The meaning of self-nonself is to be understood within the two major theoretical frames (Coutinho and Stewart, 1991): the network "self" is not the same entity as the Burnetian, classical "self." In the latter case, it is defined as the total molecular composition of the organism to which the immune system belongs, defined by the external observer. The immune system will ignore self because tolerance will be achieved by elimination of all self-reactive cells. Tolerance is then a state of default, established by ignorance of what self is. The system will only react to nonself. It is based on recessive mechanisms of tolerance (deletion, anergy, etc.). In the immunocompetent state, the system will only react by destruction of all that it recognizes. All autoreactivity is pathological. The end practical result for autoimmune diseases and transplantation is immunosuppression as the logical therapy. 

For the network paradigm, macromolecular self is primarily defined by the activity of the immune system itself. Self can only be defined by what the system can recognize (self components will not be similarly perceived by the system, and many will not be perceived at all).The immune system positively defines self components by activation into an internal physiological activity of self-reactive cells that interact with each other in a nondestructive manner. The pool size of internally activated cells is tightly regulated. This internal activity is built up during development and represents a dynamic state of memory (in the immunologic sense) of the antigenic composition of self along development. The distinction between a self and a nonself antigen is, at the start, related to its presence or absence during early ontogenetic development. Tolerance is systemic and maintained by dominant mechanisms (suppression, "education") in the adult by continuous recruitment into the same (non-aggressive) effector pool of newly produced, self-reactive cells. By this perspective, autoimmune diseases are rather the result of failure of regulatory mechanisms. Hence, they should be considered as immunodeficiencies and therefore treated as such, and immunostimulation should be considered as a potential therapy.  

I would like at this point to ask the participants if they would consider that not all immunologic functions of T cells are of the inflammatory/aggressive type and therefore that self-reactive T cells can be activated and differentiated in nondestructive effector functions, without the need to destroy every autoreactive cell. For example, T cells may have the ability to secrete immunosuppressive type of factors and suppress the effector activity of other T cells or deviate the program of functional differentiation of another T cell that happens to be stimulated in the same microenvironment - in a similar manner as people use to deal with the Th1/Th2 dichotomy.  

Zlatko Dembic - 9:07pm May 13, 1997 (#26 of 32)  

Thanks to discussion with Rod, I finally see the difference between the self-nonself discrimination and the analytic-synthetic principle that I was, in fact, trying to reinvent. The self-nonself is static (markers are being selected), and integrity protection/restoration is an active (signals select) mechanism. I don't agree that all integrity-disrupting signals should be packed into the nonself group, because they are not static markers (they are not being selected) and because this is oversimplification of the "integrity" as a hypothesis. To distinguish static markers, somebody, someone, or something has to distinguish between them. In contrast, in the "analysis/synthesis" mechanism, a selection-of-the-fittest kind of response (including no response at all) operates. As in natural evolution, there is no outside force that selects the fittest specific response. This is my main objection to the self-nonself discrimination principle - and, in addition, it is an oversimplification of the immune response. As Rod pointed out, natural selection is acting on the action and not on the static features of life. So, as the organism selects its kind of specific immune response, it also uses its other active mechanism, the generation of immune diversity in the repertoire, that has been selected for the benefit of the analytic process in the protection/restoration principle.  

The integrity hypothesis differs from discriminatory hypotheses at the philosophical level, because it uses a dialectic analysis/synthesis principle with thesis/antithesis/synthesis as integrity/disintegration/restoration of integrity. The self-nonself discrimination principle involves distinction in "space," whereas the dialectic principle simultaneously incorporates the "time" dimension. Because of this they cannot be compared, and if they are, integrity would lose its major explanatory power - its analytic-selective part. This is a completely different view of the immunologic world. Thus, defining not-changing self becomes imperative. 

Response about the antigenic changes in the self: The static features of life - the self markers - change in parallel with the antigenic content during life. First, every immune response gives a load of idiotypic determinants that may constantly change during the life of an individual. Are those molecules cleaned by the nondestructive immune response? Second, bacterial flora change during life, too. Sonicates of bacterial flora of one individual can provoke an immune response in another individual (Powrie et al., 1994). Because symbionts are tolerated, they provide the host with a load of different antigens as they change during a lifetime. Then, sexual maturation requires expression of sex-related antigens. 

Response to the question about achieving the IP function: It might be achieved by, perhaps, selecting an adaptation (mutation) of the worker-on-integrity molecule - a chaperone, capable of destroying the BD weapon of the host, which could evolve either simultaneously or later (for example, from a pool of mutated IPs, a BD antagonist is being selected as soon as BD arises). This chaperone-like molecule would be capable of destroying the conformation of BD or chopping BD into pieces. From IPs and IRs, molecules might have evolved that now represent a machinery for degradation of proteins with peptide-presentation characteristics. The IP/IR functional significance is in providing signal[1] to specific immunocytes. The initiation of the immune response depends on disintegrative processes grouped as signal[3] and as such might be perceived by, for example, dendritic cells. The danger-assessment function of the IR region now has two choices: to switch the effectors either on or off by providing signal[2] or not, respectively. At the same time signal[3], as a nonspecific signal, exerts a very important additional function. Like the electronic mesh in a triode, it modulates signal[1], by changing the set of peptides that are being loaded and expressed; this is understandable, because dendritic cells, in an active state, would tend to pick up the antigens in their microenvironment (giving rise to signal[1M]). This is the driving force of the diversity of the system. Next, the cell receiving signals[1M] and [2] has:  

(1) No choice in accepting signal[1M] - that is why it is called a specific immunocyte, after all; and if it is the only signal it gets, it is tolerized - killed; or 

(2) A choice in either accepting or not accepting signal[2].  

If the cell accepts signal[2], the effector mechanism ensues. If it does not accept signal[2], it gets tolerized after it received the activation signal (combined signal[1M] and signal[2]) by, for example, downregulating the receptor for signal[2]. Acceptance might depend on the balance of soluble or membrane-bound molecular mediators present during the cross-talk. The time difference between the initial activation signal (signals[1M+2]) and the rejection of signal[2] is, perhaps, a way to (1) downregulate the normal immune response to "parasites," (2) create memory, or (3) achieve tolerance (anergy?). Soluble mediators, in addition, affect the class of the response (for example, Th1 vs. Th2). 

The effector mechanism would aim to allow reinstatement of the best possible level of integrity as before the response started. If it cannot, the organ or tissue loses function, and the parasite wins by secluding itself in the disintegrated area, to the frustration of the response. For the same reason, if a vital organ is affected similarly, the organism succumbs. The initiation of the response starts when the "parasite" intruder breaks certain integrity signals that are perceived as signal[3] and result, for example, in the activation of dendritic cells. This in turn would upregulate signal[2] and modulate signal[1] to become signal[1M]. Tolerance to self antigens is primarily the function of signal[1], but it also can be achieved by not accepting signal[2], or if for some reason signal[3] does not upregulate signal[2] and after some time vanishes. Tolerance to symbionts is perhaps the result of not accepting signal[2].  

Doug Green - 11:24pm May 13, 1997 (#27 of 32)  

Another small comment on the question of whether "self" changes. Zlatko gives the response that "self" changes because of three things: the collection of idiotypes, the collection of bacterial flora, and the proteins that are expressed during sexual maturation. These are the usual answers that one hears. The question, though, is whether these new proteins are truly "tolerated." We are usually led to believe that new idiotypes are themselves antigenic (at least, in most network schemes). And, as I stated, I don't know of any evidence that new proteins produced during maturation would be tolerated by the maturing individual while being immunogenic in a syngeneic individual of another age. Finally, are you saying, Zlatko, that sonicates of gut flora from one individual produce a response in a different individual but not in the individual from which it was derived? (I haven't read your paper, but I'd be willing to bet that the sonicates would be immunogenic in both, because gut flora are extremely immunogenic when they happen to leave the gut). 

Zlatko Dembic - 12:27am May 14, 1997 (#28 of 32)  

In the reference I cited before, bacterial sonicates from autologous intestine stimulated proliferation of neither peripheral blood mononuclear cells (PBMC) nor lamina propria mononuclear cells (LPMC) in vitro but did so when tested with other individuals' PBMCs and LPMCs. In other words, heterologous sonicates induced strong proliferation of PBMCs and LPMCs associated with the expansion of CD8 and CD4 markers, production of IL-12, interferon gamma, and IL-10.  

In addition, autologous sonicates also activated LPMCs isolated from sites corresponding to active inflammatory bowel disease but neither PBMCs nor LPMCs from non-inflamed intestine of these patients. Proliferation was inhibitable by MHC class II specific antibodies. 

Rod Langman - 12:47am May 14, 1997 (#29 of 32)  

In response to Antonio (#25), I'm not so sure that translating biological activities into mathematical functions is a real escape from language, just a substitution of one for another. After all, math functions are just rules for manipulating symbols of some kind of stuff, but the stuff biology is made of seems rather resistant to symbolic manipulation - even the thingies had to be put out of their misery. But you are right - it is the problem and not the words that concern us. Which brings me to your restatement of the Burnetian view of self. I'm not sure to whom this view should be properly ascribed, but the definition of self as the total molecular composition of the organism to which the immune system belongs is wide of the mark. I an not aware of anyone wanting to use the word "self" in an immune context to describe such an artificial entity. I, at least, would be quite content with your conclusion: Self can only be defined by what the system can recognize.  

Is the following a fair summary of our two views? The view you are supporting is one of dominant suppression to achieve self nondestruction, and the one I would support is one of dominant immunity to achieve nonself destruction, with self being simply forgotten (recessively deleted).  

I doubt that there can be much disagreement that helper T cells perform a strictly regulatory function and that, when they associatively recognize antigen on B cells, thereby promoting their differentiation to secreting plasma cells, these helper T cells do not kill the poor B cell. There may be a little grumbling, but I doubt we can dismiss suppressive T-cell activities either. It is not the activities that are of concern, but what they are directed at. I am on record as claiming that the same cells that are biodestructive cannot also be regulatory. And, so far as I'm aware, regulatory T cells do not perform destructive effector functions. Although helper T cells secrete all kinds of stimulants that juice up the surrounding cells, it is these cells that have their performances enhanced, operating often at lowered threshold levels. These stimulants made by regulatory T cells are not toxic, and, as best as I can determine, suppressive activities of T cells are not toxic either, but rather act as depressants.  

Rod Langman - 01:26am May 14, 1997 (#30 of 32)  

Glad to have been of help, Zlatko (#26). But I'm still not sure what integrity protection is, so let me try to say in my own words what I understand to date. The organism is a complicated thing with lots of different activities going on inside. We assume that a normal organism is in a state of homeostasis, under the control of many regulators. These regulators have to know when the system is becoming disordered, and these regulators then attempt to restore the old order. The initiation of wound healing might be a good example of a breach in integrity, and a response to restore integrity is wound healing. The part of the immune system that was around when the wound occurred would recognize the release of epidermal growth factor and all the other "alarm" or "disintegration" substances, and these would permit any cell in the area that happened to contact antigen to respond to that antigen and eliminate it. When the antigen is all gone, the immune response stops, and the immune system comes to rest too. This would seem awfully close to the danger principles envisaged by Ephraim and Doug. The major difference would be that disruption of integrity covers many more situations than are covered by "danger" and inflammation. Am I getting close? Doug and Ephraim can help me here too.  

Zlatko Dembic - 1:32am May 14, 1997 (#31 of 32)  

Yes, you are.  

Doug Green - 5:28am May 15, 1997 (#32 of 32)  

In response to Rod (#30): Remarkably, I think we are somehow reaching a consensus (I wasn't sure that could be possible). Ephraim and I are in agreement about the nature of some of the triggers, and it seems that we are describing the things that Zlatko envisions. I have been reaching a similar conclusion regarding the meaning of integrity (and its loss), but a few more concrete examples would be useful. I hope people are not getting too frustrated by my harping on concrete mechanisms, but my problem is that so many concepts in this field are defined in such circular ways that they are only useful as a kind of shorthand for something we think we mean (e.g., "self is only defined by what the system can recognize " - we know what we mean, but we can easily get into difficulties, as we've seen). The trouble is, we then begin to think that the shorthand (e.g., self, danger, etc.) is rigorously defined, and we think we've then explained something without realizing that the foundation is sand (the circular definitions we started with). Here, I think we are successfully pushing beyond those problems and making some progress.