Scientists commonly complain that jargon-loaded immunology is impossible to understand. While jargon abounds in immunology, it is no more dense than in other sciences. But immunology is somewhat peculiar in that the logic behind its jargon is at best circular, and often nonexistent.

There are many reasons for this. For example, antigens are defined by antibody and other immune responses, but antigens also define these responses. This is not a promising start for linear progression to ultimate destinations. Everything in immunology is defined by everything else, much of which is not indigenous to the field. Contextual importance demands that immunolological definitions be somewhat slippery.

In science, much of what is thought to be true is often explained by more fundamental phenomenas. But in immunology, many of the steps between the fundamentals and the truths are missing. In the antigen and antibody relationship, antigens beget antibodies, which bind antigens. However, antigens can also annihilate the ability to produce antibodies.

The surprise is that we do not know what or how many different processes cause these wildly different outcomes. Immunologists otherwise in agreement hold different views on this matter, as can be seen in the HMS Beagle Cutting Edge debate "Models of Immunologic Tolerance" [1], a monumental weeklong effort to reach a consensus. If one was reached it wasn't obvious to me, there being too many legitimate differences in viewpoints, facts, and approaches.

If like-minded immunologists cannot agree, immunology's periodic internecine struggles only worsen basic communication. These wars bring vigor to the field but, as wars do, they also bring confusion. Battles between those with chemical versus biological bents occupied the first half of this century. Immunobiologists are currently described as B cell people or T cell people, indicating their research's restricted focus. Signalists, who are currently invading immunology, rarely look at what signals do (inducing functions), but instead at some favored aspect of the signaling pathway. The list is endless. I, personally, am not immune to this problem, having happily studied immune responses to sheep erythrocytes as if they would be responsible for some dreaded disease of the future. During their heyday, sheep erythrocyte immunologists eclipsed the study of real infections.

Immunology was preceded by a vague understanding of resistance to epidemics due to prior infection by the epidemic-causing agent. It was put on a firm footing by the Society of Milkmaids, who made their wisdom known to Edward Jenner, the father of immunology. From Jenner's experience sprung forms of vaccination or immunization, captivating the attention of all. However, only certain infections were reduced by immunization. Along came antibiotics to help, so that during a lull in infectious problems, immunologists developed the concept of self-nonself discrimination.

There are two problems with the self-nonself paradigm. The majority of infections are not prevented by immunization, so recognition, attack, and repulsion of nonself is not a given. The second problem is that some infections are detected by their ability to stimulate antiself responses, and some of these responses cause damage, as in the case of syphilis. Blindness of the immune system to self is not absolute. Mild clinical immunity to self without apparent stimulus is not infrequent.

Most immunologists frame questions about self-nonself discrimination in terms of any antihost immune response, instead of restricting the discrimination to the damaging antiself immune responses that cause problems with host survival. They do not address the issue of controlling damaging antihost immune responses.

This not a new judgment. Paul Ehrlich stated that "the organism possesses certain contrivances by means of which the immunity reactions [are] prevented from acting against the organism's own elements. . . . One might be justified in speaking of a 'Horror autotoxicus' of the individual." To be sure that Ehrlich was talking about damaging antibodies, we can read his footnote, which states "a typical autospermotoxin is developed in the blood of guinea-pigs which have been treated with guinea-pig spermatozoa. . . . This is able in vitro to kill the spermatozoa of the animal itself. But such an injurious action on the spermatozoa does not take place, even in the slightest degree, in the living animal, because . . . only the immune-body combines with the spermatozoa, not the complement. [The complement is a system of proteins that kicks the living daylights out of anything to which an antibody, Ehrlich's "immune-body," attaches] an autotoxin within our meaning, one that destroys the cells of the organism which formed it, does not exist." [2]

The meaning of Ehrlich's statements is clear even today. Yet immunology labored for the next 50 or so years under the assumption that no immune responses to self could occur and that autoimmune diseases must have some explanation other than genuinely damaging autoimmunity. Even today, some immunologists ask if autoimmune disease is really just an immune response against tissue damaged by other mechanisms.

A converse of this situation is now being actively addressed: How do hosts pick a protective response, out of the candidates available, which will damage an invading infectious agent? If the host picks a response that does not damage the parasite, the host is in deep trouble. So, what the host must not do in mounting antihost responses, it must do in terms of antiparasitic responses. Damage is the action to avoid in harming the host and to encourage getting rid of the parasite. If theories and questions were framed to address the presence or absence of damaging immune responses, we might move closer to the issues in the real world.

Dominant views in immunology are sometimes needlessly circular. Consider clonal selection, a concept introduced by Talmage [3] and taken up by Burnet [4]. Lymphocytes proliferate in response to antigens to form clones of cells that make products of the immune system and, also, memory cells that synthesize more products faster on the next go-around with antigens. Many cite clonal selection to explain immune responses. However, clonal selection is merely a description of the immune response itself. The concept of clonal selection makes us little wiser about why the immune response occurs in the first place.

This question has been approached by a number of thoughtful scientists. The two-signal (associative recognition) model of Bretscher and Cohn [5] and its recent developments by Langman and Cohn [6], the costimulator model of Schwartz [7], the danger model of Matzinger and Fuchs [8], and others are ways to explain how immune responses are kick-started. They propose that lymphocytes see antigens via their antigen receptors but also see non-antigenic cues for activation via receptors other than antigen receptors. It is the interaction between antigen receptors and these other receptors that induces activation. We are beginning to understand what some of these interactions are. But guess what? Immunologists are battling furiously with each other for ascendancy of their favorite interacting mechanisms for activation. The understanding of these interactions is apparently tied up with intracellular signaling events.

The problem now is that we have little idea of what distinguishes a lymphocyte-response-generating pathway from those that are neutral or even inhibitory. Signaling pathways are not understood with the same precision as metabolic pathways. They are far from linear, many steps are missing from view, and signaling pathways bump into each other unpredictably. While one can construct detailed metabolic pathways that take glucose plus oxygen and end up with water, carbon dioxide, and energy, it is difficult to discern the steps between antigen binding and the stimulation of lymphocytes to synthesize end products.

All of this makes for an understandable degree of nebulosity regarding activation pathways. Adding to the confusion are the recent descriptions of inhibitory pathways. These off-signal pathways are distinct from the antigen-receptor signals but interact with them; therefore, they can be called coinhibitory mechanisms [9]. The first coinhibitory mechanism to be described was the inhibitory Fc-receptor pathway [10], which has however some degree of puniness associated with its activity. I say so reluctantly, since this mechanism has preoccupied me for a long time. More impressive in its effect is the CTLA-4 pathway which, if missing, leads to a rampaging form of autoimmunity [11]. This suggests that there are stop signals as well as go signals. Many such stop signals are mediated by transmembrane proteins with tyrosine-containing motifs inside cells that attract various phosphatases which appear to be important for the stop signal.

An even more fundamental question is, At what stage of their development do lymphocytes decide to react to foreign elements while avoiding damaging reactions to self? Probably at all developmental stages. This distinction was previously considered the business of antigen receptors and the establishment of the immunologic repertoire, i.e., lymphocytes are selected on the basis of the specificity of their antigen receptors to be included in the club, the repertoire. The rites of initiation to this club were hard to observe directly until the advent of transgenic mice. One could place the genes for specific antigen receptors into the cells of a developing mouse and watch what they did to the generation of lymphocytes in the absence or presence of the antigen recognized by these transgenic antigen receptors.

Early in their invasion of immunology, transgenics researchers got exactly what conventional wisdom predicted. A lot of deletion in the presence of antigens suggested that similar events were occurring under natural conditions. However, transgenics researchers are now finding other experimental models that indicate that alternate mechanisms also function.

There is a tiny problem with this approach. These transgenes, taken from differentiated lymphocytes, have flexed their muscles in response to antigens and, at least in B cells, have accumulated somatic mutations. During their development, lymphocytes undergo a multistep process of antigen-receptor maturation. Popping in mature transgenic antigen receptors may well be the equivalent of grafting a fully formed phallus on a fetus; the results are clear-cut and interesting, but probably prone to artifacts.

In CTLA-4 knockout mice, in which rabid autoimmunity develops, T cells go through the usual niceties of thymic education involving both positive and negative selection. However, T cells in these animals seem to loose it in the periphery, where they flip out of control. Other models of autoimmunity tell us the same tale, that is, positive and negative selection in the thymus can be okay, but animals still become autoimmune. This suggests that the control of late lymphocyte activities well beyond the establishment of the repertoire (what lymphocytes recognize when they are popped out of their formative thymus and bone marrow sites) also determines the effectiveness of the discrimination between self and nonself.

Discrimination between self and nonself depends on more than the antigens. High levels of costimulation and the absence of coinhibition can allow a normally inhibitory form of antigen to stimulate a good response. This definitely suggests that lymphocytes have both stop and go mechanisms during their responses to antigens. These mechanisms sense the context in which antigen is seen, independent of the specificity of the antigen, and if the receptors responsible are missing, lymphocytes cannot respond or not respond appropriately.

Even the signaling function of the antigen receptor, particularly on B cells, is in doubt. Despite its well-endowed signaling motifs and its numerous intracellular signaling connections to the nucleus, some immunologists suggest that the B cell antigen receptor does not impact on B cell function other than to endocytose the antigen and display antigenic fragments on B cell major histocompatibility molecules that T cells recognize. The T cells are given sole credit for activating B cells.

Immunologists frequently fall for either-or arguments rather than both-and approaches, usually by constructing experimental models that appear to justify their bias. These models tend to give only one side of a multifaceted story. It is more likely that the B cell antigen receptor has both an endocytic function and a signaling function. When T cells are necessary for a B cell response, T cells deliver to B cells some initial stimulating events along with those received through the B cell antigen receptor. T cells also provide both a costimulatory function and a means to decrease the effects of coinhibitory events.

The limiting factor is our active but ultimately feeble minds in grasping the full meaning of natural phenomenas, particularly those evidenced by the immune system. Given that immunology has many black holes, it seems less than wise to hang on for dear life to concepts that may have outlived their usefulness. Open-mindedness will help immunologists to communicate among themselves and with scientists in other disciplines.