Zlatko Dembic - 3:26pm May 11, 1997 (#13 of 32)  

If self-nonself discrimination would be perceived as an ubiquitous principle in life, then we find it neither in plants (they neither reject allografts nor xenografts; A.M. Silverstein and N. Rose, personal communication) nor in invertebrates, where an analog of transplant rejection involves species rather than individual incompatibility (Du Pasquier, 1993). In contrast, preservation of integrity trespasses living kingdoms as a principle, and it is the first principle observed when the first cell is formed during evolution. After the first cell's division, the daughter cells perhaps have a chance to compete and apply discrimination (Dembic, submitted).  

I dislike the self-nonself distinction (and the others: lethal-nonlethal, etc.) because it is based on a discriminatory principle. To explain life and its forms, integrative forces (and thoughts) are much more useful. Protecting one's integrity also allows distinction from others, but here we are left with the choice to tell the "parasites" from the "symbionts" (or friends from foes). To apply the self-nonself principle or any other discriminatory principle would mean that we a priori react to nonself (lethal, etc.), and so we are left with the problem of excusing ourselves to friends and potential useful collaborators that we had hit in the process. In fact, that is not what we have so far observed in immunology. The organisms respect symbionts. Thus, the protection of integrity can better explain autoimmunity and tolerance of symbionts than any discriminatory principle, already at the level of a general concept (Dembic, submitted).  

I would like to briefly comment on Bill's statement (#8) that the self-nonself concept should stay as a useful description. I agree. The words are not important, as long as we know what we mean by saying them - perhaps, similar to, for example, a statement that the Earth circles the Sun; we do not say, "The Earth ellipses the Sun."  

Doug's wish to leave the conceptual level and go into the mechanisms of the immunology is fine, but then, without the concept(s), we will remain confused. We will have innumerable (and expensive) possibilities and lines for further research. In contrast, with concept(s), much can be spared and concentrated on the effort to achieve specific goals or to follow unusual discrepancies (from the concept) to prove or disprove hypotheses. That is what I find more interesting than just screening information. By identifying key concepts in this debate, such a concentrated experimental future could be envisaged.  

Rod, now I understand your comment about starting the games! Okay, I will play. But first, let's put down some logical context in defining the stage with HT, PT, and BDT. As you claim, HT and BDT are the same except for that mutation that enabled BDT to kill PT. If HT is, in fact, the same as PT, then BDT (according to you) has to distinguish between the PT and BDT. But BDT cannot possibly exist in the first place with just this (BD) mutation because it will kill itself, leaving no progeny. The BD mutation (your words) does not discriminate between HT and PT. So, before a BD mutation can be transferred to progeny, HT has to protect itself (its integrity) by acquiring an integrity-protective (IP) mutation. Therefore, the integrity comes before the rest, and only an IPT-to-IPBDT pathway can exist, and BDT itself cannot. The rest of the story you can imagine yourself.  

Zlatko Dembic - 7:01pm May 11, 1997 (#14 of 32)  

I decided to tell the end of the story about thingies myself.  

Before I go any further, let us diverge for a moment and assume that lethal thingie (LT) developed a similar yet different biodestructive weapon (BD2) and thus formed IP2BD2LT. So, if we now have IPBDT that has to kill LT or IP2BD2LT in order to defend itself, it has to either (1) discriminate between self (IPBDT) and nonself (LT or IP2BD2LT) or (2) protect its integrity. The way to discriminate would be to check for differences, and they are located in the IP and BD regions. According to the self-nonself discrimination principle, HT is killed if infected with LT, and IPBDT kills both LT and IP2BD2LT (the lack or presence of different markers leads to identification of nonself), provided BD2 does not kill the IPBDT. 

Now, let's examine the protection-of-integrity theory: The way to protect IPBDT integrity would be to express the BD function together with the IP function, as they are inseparable (otherwise suicide ensues; however, there might be a lag in time). According to the integrity hypothesis, HT is killed unless it mutates to IPBDT. IPBDT neutralizes and kills both LT and IP2BD2LT. So both hypotheses work fine. But there is a major difference between the models, yet to come!  

Now let's consider a situation in which HT, LT, and also a potential symbiont (ST) are present. According to the self-nonself idea, IPBDT mercilessly kills ST (the lack of presence of self markers). But, according to integrity, IPBDT has a choice in expressing first IP and then, if necessary, BD function! This looks like inducing and effector arcs of the immune response. The choice lies in the assessment of the relative danger of the situation. In this primitive example, there is almost no distinction between the integrity and danger hypotheses; however, the danger hypothesis requires the integrity-protection principle for explanation.  

Zlatko Dembic - 9:54pm May 11, 1997 (#15 of 32)  

I have been inspired by Rod's little game, so if he doesn't stop me, I will end up with thingies running around like Smurfs!  

We can even go further and assign a danger-assessment (DA) mutation as a modulator of the biodestructive function. So, to repeat, IPDABDT thingie is friendly toward ST. If we consider tolerance as a state where danger will be assessed at a low probability, then ST can be comprehended as DLT, as opposed to a situation where the probability for danger would be high (DH), like, for example, in an LT (or PT). So the LTs and PTs are DHTs. If in IPDABDT, during the symbiosis with DLT (formerly ST), a mutation in DLT occurs that would lead to DHT, IPDABDT will notice this with its DA function and will engage BD effector function.  

Thus, the integrity-protection function regulates the initiation of the response and the danger-assessment function provides tolerance and also regulates the effector function. All is based on the difference among the alleles of IP, DA, DL/DH, and BD domains.  

Rod Langman - 2:07am May 12, 1997 (#16 of 32)  

A simple factual question: Is there any known demonstration of new antigens appearing after birth, or after puberty? For example, in inbred mice, can adult tissue or proteins, etc., be rejected by juveniles? The only situation I can think of offhand that comes even close are the tumors that develop in one individual that can be rejected when transferred to another syngeneic individual. However, this is not a general case, and when observed, there is usually a sharp dosage effect: Small tumor inocula are rejected, and large inocula are not. I ask this question because there have been several hints in comments that the universe of self antigens changes throughout life. While self components change as measured by all manner of non-immune instruments, can any changes measured by immune instruments be observed in the normal case? (Those who may want to argue a case of immunoglobulin idiotypes inducing anti-idiotype antibodies I'd put to one side and argue separately along lines similar to tumors.) Anyone reading this who is not a debater can fax or e-mail me. 

Zlatko Dembic - 12:04pm May 12, 1997 (#17 of 32)  

Just to conclude the "thingie exercise" before I go into factual specifics. There are two major conclusions that I want to make:  

(1) In order to successfully defend itself, the simplest thingie (IPBDT) does not necessarily need a discrimination principle.  

(2) There is a discrimination principle involved in the IPDABDT that is limited to IP, DA, DL/DH, and BD differences. We can call them self-nonself markers or anything else, but this is not relevant at this point. However, if we fail to name them properly, in higher-complexity thingies we are going to get confused! (That is why I argue against the use of the self-nonself.) 

So now we can go a step further and describe a bit more complex thingie (evolution works). Let's suppose that the IP domain includes protection against BD, self killing (SK), and eventually species-specific IP elements based on chemical (microenvironment) and physical (habitat) differences (perhaps later this can develop into individual specifics). Thus, IP gets larger than BD. I would argue that the immune response starts by activating the integrity-protection-linked function, which, in the course of evolution of IPDABDT, has reached an elaborate network of signals. A loss of the integrity signal(s), sensed by, for example, a specialized factor (dendritic-cell equivalent), leads to activation of a cross-talk function to reestablish integrity by assessing danger and, consequently, destroying the parasite or tolerating the symbiont (these functions can be interpolated to communication with the specific part of the immune system, and perhaps nonspecific mediators). Therefore, we can group DA-DL/DH pairing into a novel category, which I would like to call integrity reestablishment (IR) function. Because there are many ways to perform IR function, these will evolutionarily differ at the same rate as IP. So, we end up with an IPIRBDT. This is the discrimination principle I call the integrity protection/reestablishment principle.  

In conclusion, as I said before (Dembic, 1996), the immune system evolved to do integrity protection and integrity reestablishment, and, as a consequence, it uses a discriminatory principle based mostly on IP and IR alleles. 

Thus, we only need to define integrity (as an integrated being; Day 1, message 6) and not the self, and we might have it all! I feel that, if you agree on this point, we can continue together, having a unified concept - you name it. 

Zlatko Dembic - 4:36pm May 12, 1997 (#18 of 32)  

Just one small addition about the most complicated thingie (IPIRBDT). A conceptual leap might take us to envisage major histocompatibility complex (MHC) class I as possible descendants of the IP (integrity protection) and MHC class II (immune response genes) as descendants of IR (integrity reestablishment) regions, with corresponding receptors also being descendants of the putative (and not yet mentioned) IP and IR receptors of the IPIRBDT. So, at the end, we ended up with a possible link to factual data. 

According to this speculation, danger assessment (being part of IR; DA recognition of either DL or DH) would play a regulatory switch; however, where would it lead conceptually? To kill the host or the parasite? Thus, I argue that integrity reestablishment (to a level that could at least secure reproductive functions) is a purpose of the immune system. 

I think it is a good idea from Rod to engage the audience of this debate. Because we are not databases, I would also appreciate an e-mail or fax for the same purpose - to discuss differences among the adult, embryonic, and juvenile states of the antigenic world. 

Rod, I would be glad if you could explain to me how you view the origin of self-nonself (lethal-nonlethal, or the like) discrimination?  

Rod Langman - 9:07pm May 12, 1997 (#19 of 32)  

Despite appearances, the dreaded "thingies" may have done some good. Let me address some of the key points raised by Zlatko. 

In the last paragraph of message 13, the point was made that no biodestructive process can arise if it is immediately self destructive (i.e., no BDT can arise unless it avoids destroying its HT). Calling this an integrity-protective mutation is fine, but I would like to know how this was achieved. 

In message 14, a whole new problem is introduced with the notion that the pathogens themselves need to be protected and may have their own BDTs. This is like discussion of the resistance of bacteria to viral infections when the bacteria are overrunning the immune system. I'm not aware of any bacterial antiviral defense mechanisms that end up killing the host with an immune system, so I think this is getting off the track of understanding the immune system as a biodestructive defense mechanism that somehow avoids self destruction - our focus is on how this is achieved. 

In message 17, the two major conclusions can be reworded a little to say that: 

(1) A biodestructive protective mechanism that does not destroy self does not need a self-nonself, or any other, discrimination principle.  

(2) Integrity protection is limited to those elements that are capable of upsetting integrity and restoring integrity. These can be called self-nonself markers.  

True, the integrity protection that has been built into the BDT, making it an IPBDT, has in effect distinguished the host from the pathogen. But how do IPs work? The second conclusion makes it clear that this is a system of discrimination based on the germ-line selection of self markers. 

We all seem to agree that host and pathogens are made up of the same building blocks, so recognizing the blocks themselves is of no help. If not the blocks, then it must be some way in which the blocks are organized, and this in turn implies some kind of recognitive element that can distinguish the organizational features that allow the best discrimination between host and pathogen. It is essential to repeat that evolution cannot do anything to perfection; it must make mistakes, and the pathogens must have ways of escape, but overall there are sufficient survivors among hosts and pathogens to ensure the perpetuation of each species. There is obviously no way to select for an immune system that provides, in the absurd, 99% protection against 100% of pathogens if it means that there are so many lymphocytes in blood that 90% of individuals die of asphyxiation because there is no room for the erythrocytes.  

Rod Langman - 11:00pm May 12, 1997 (#20 of 32)  

In response to Doug: First, sorry about the abuse of Greenian thingies - I'm not as creative with cute acronyms, so I'll abandon the beasts. 

Re: message 11. First, in search of a clarification: You wrote, "As a first approximation, for example, the process of central tolerance (negative selection) creates an illusion of self-nonself discrimination because any protein that is always available for presentation to developing lymphocytes will eliminate those cells from the population." This is reminiscent of your earlier comment about a simple self-nonself discrimination based on the Lederberg principles of strictly sequential tolerance only and then immunity only; I wonder what mechanistic principles apply to achieve negative selection, and, although you make no special reference, is this strictly intrathymic, or can it occur in the periphery also? 

Second, a challenge: With regard to peripheral events, you assume there are cells present that are able to respond to any antigen not eliminated by negative selection (I suspect that you have restricted this to the thymus only) and that these cells have open to them two pathways, tolerance or immunity. The choice of tolerance or immunity is postulated to be governed by environmental concerns such as inflammatory mediators and others too numerous to count. You want to rule out any criterion that could be described as self vs. nonself because these criteria are not described in terms of what actually happens to ligands, receptors, and other molecular events. My challenge is to ask whether you would consider the possibility of a set of mechanistic details and boundary conditions that offer a way of establishing a set of criteria that amount to a workable (i.e., imperfect but adequate) self-nonself discrimination that does not require nonself markers such as "danger," "disintegration," "inflammation," "toxicity," etc. 

Doug Green - 7:18am May 13, 1997 (#21 of 32)  

Rod, it's a very good challenge, and I think it's our assignment for Day 3, so if it's okay I'll come back to it then. I have no doubt, though, that as rigorous as we try to be, there will always be some play in the system (for example, as you point out, "inflammation" is not rigidly defined - I've been using it as shorthand, and I have to be much more careful. Thanks for that.) 

Doug Green - 07:19am May 13, 1997 (#22 of 32)  

There is a small point that I'd like to bring up, as it seems to be an assumption in some of the models that are being discussed here - that is, that the antigenic nature of self is always changing. Are we sure that this is true? Yes, we can state that there are developmental changes, and we should expect new molecules to appear that weren't present before. But must these be antigenic? I know of no evidence (but of course I may just be uninformed) that aging produces new antigens - in the sense that cells from an older animal put into a syngeneic younger animal will elicit an immune response. Is it possible that this just isn't a problem? 

Can the function of the immune system be pretty well explained by a combination of two fairly concrete principles: (1) negative selection, wherein immature lymphocytes (immature in the development of the lymphocyte, which can occur any time in the life of the individual) die if their antigen receptors are ligated, and (2) costimulation, where ligation of an antigen receptor induces a state of anergy (or deletion) unless another signal generated by ligation of a different receptor occurs? The presence of the costimulatory molecule (e.g., on a dendritic cell) depends upon exposure to bacterial products (endotoxin) or inflammatory mediators (such as IL-1 or tumor necrosis factor). A cell whose antigen receptor contacts its ligand in the absence of costimulation (which will only be present in sites of infection or damage) does not pose a threat to the system, because it will be functionally eliminated. Do we need much more than this? I think the answer may be yes, but perhaps not much. (I've tried to keep this discussion as straightforward as possible, without invoking any process that cannot be rigorously defined in concrete terms. Even terms like "antigen" could be replaced with more concrete terms like "protein," in a first pass at explaining the function of the immune system. No doubt, though, I've still left huge holes in the description.) 

Doug Green - 7:20am May 13, 1997 (#23 of 32)  

I like Rod's thingie game as well, but I don't think that the answer is necessarily that tough if we switch to real life. Without specificity, it would seem that a problem quickly emerges if the host thingie makes the biodestructive thingie every time it perceives that its own tissues are being damaged (which I'll suggest is a straightforward way even very simple creatures can make responses - damaged cells release something that invokes a response). The BDT is released and destroys the PT, but because there's now more tissue damage, there will be more BDT until the host self-destructs. Right? 

Well, not necessarily. As Ephraim noted (#5), there are basically two ways our cells can respond to irreparable damage: they can die by necrosis or they can die by apoptosis. When stressed by an intracellular pathogen, our cells attempt to die by apoptosis, which triggers a rapid clearance and destruction of the cell (and pathogen) in phagocytes. Extracellular parasites either produce tissue damage or release substances that we are hardwired to make inflammatory responses to. We (the HT) respond by a rapid series of events (none of which involve somatic diversity) that results in the accumulation of some very good BDTs - the neutrophils. These eat everything in sight and then die. However, they die via apoptosis, which results in them (and everything they contain) being digested quietly in the macrophages with no further proinflammatory effects. 

Many intracellular pathogens have mechanisms to block apoptosis, and eventually the cells they infect die by necrosis as the pathogen replicates. So, in response to tissue damage, the host releases another BDT. Cytotoxic cells release cytotoxic granules that bypass viral anti-apoptotic defenses to induce apoptosis in infected cells. If the stressed cell can trigger the cytotoxic cell to release its granules, there may be no need for a "specific" response. But responses cause damage, and, as I've suggested, the extent of host damage has to be limited if the host is to benefit from the response. Specificity of the sort we've been discussing is important to focus and amplify nonspecific effector mechanisms, giving the host an advantage over the parasite, which might be rapidly proliferating.  

Here's a fun puzzle, then. Defense mechanisms exist in virtually all metazoans, but only the vertebrates have immune systems that generate somatically diversified, specific responses (some small amount of specificity has been seen in some other organisms, but it is limited, questionable, or both). These organisms have been around a long time, and appear to do pretty well - much better than we do when our specific immune system is even marginally impaired. Why? Why, when we lose our specific immune responses due to congenital or acquired deficiency diseases, don't we simply do as well as, say, starfish, insects, and sponges in defending ourselves against disease (and these creatures obviously do survive to reproduce)? There are several good answers, I think, and they may be enlightening.