CRITICISMS OF THE DANGER MODEL

Not surprisingly the major critics of the danger model have generally been the foremost proponents of the self/nonself model. Langman & Cohn and Janeway attempt to reduce it to a detail of mechanism of the self/nonself theory. Silverstein, however, argues that it has all been said before and that the supposed threat to the ruling paradigm is “clearly hyperbole” (Silverstein & Rose 1997, p 199).

3.4.2.1 Langman & Cohn

Langman & Cohn scatter their criticisms of the danger model through several articles (primarily in Cohn 1998; Langman & Cohn 1996a; 1996b; 1997) with the most concentrated discussion in “Terra firma: a retreat from danger” (Langman & Cohn 1996b). In this article they claim that the theory has a key flaw,

If anti-S Ab [anti-self antibody] does exist in normal healthy individuals, then at least some anti-S cannot be lethal. Because it is also experimentally demonstrable that lethal anti-S can be produced, and does kill, S-Ags [self antigens] have to be divided into lethal (L) and nonlethal (NL) classes. The defect in the logical underpinning of the ADR [Associative Danger

Recognition - their term] model is that there is no basis for associating danger with NL-S-Ags, while not allowing danger to be associated with L-S- Ags. (1996b, p 4275)

They go on to apparently provide Matzinger’s defence, which in their words is that:

. . . for the number of anti-S iT [inducible T cell - a mature cell that has not yet met antigen] and iB [inducible B cell] cells to be at sufficiently low levels, danger is rather rare, local, and transient. Thus, most anti-S iT and iB cells find S-Ag when danger is absent. Any anti-S that is produced in a focus of danger cannot continue to be produced after the danger has been removed by an immune response. (1996b, p 4275)

And they then try to reduce the danger model to self/nonself on the basis that: . . . the ADR model, in fact, requires quite an accurate S-NS discrimination at the level of the iT and iB cell repertoires. Any claim that the ADR model noes not require a S-NS discrimination is simply disingenuous. (1996b, p 4275).

It is true that Matzinger does put forward this argument, which in effect is saying that if there are any self-reactive lymphocytes roaming around they will only be activated against self components if there is a local danger stimulus. Once this stimulus is removed, once the would is healed or the infection is eradicated, the self reactive lymphocytes would cease to be activated. However Matzinger provides a far more detailed discussion of the issue than this implies.

To start with she carefully makes the distinction between autoimmune reactions and autoimmune diseases (Matzinger 1994, p 1033), a distinction that Langman & Cohn fail to address in their critique. Autoimmune reactions are a common and transient aspect of any immune response. They are more frequent early in a primary response, and decline later as the danger signal diminishes and any autoreactive lymphocytes present are tolerized by receiving signal one in the absence of signal two (the default OFF setting).

These autoimmune reactions may come from several sources and their effects may be harmful, harmless or useful. In a perfect system these autoimmune reactions should remain at a low level and fade out, but in a complex system like the human body there are many ways that things can go wrong, and some of these ways may lead to the chronic states of autoimmune disease. Matzinger also

canvasses the hypothesis, extrapolating from the work of Bottazzo et al (1983):

. . . that autoimmunity may not be a defect in the immune response but in the expression of antigen, either in its concentration, location, or way in which it is presented. This is not the same as the popular view that an autoimmune disease may be initiated by a disruption of immune regulation during the response to a foreign antigen that cross reacts with self.

(Matzinger 1994, p 1034)

This discussion of the aetiology of both autoimmune reactions and disease is entirely consistent with the notion of danger as an immune stimulus and does not require regression to a self/nonself discrimination. Langman & Cohn’s arguments on these aspects are not convincing.

In their quest to reduce the danger model to a detail of mechanism of self/nonself discrimination they repeatedly make claims like the following:

There is a general inability to define ‘danger’ other than by a set of conveniences that are indistinguishable from a set of nonself markers. (Langman & Cohn 1997)

It is difficult to see how the release of heat shock proteins or other some other signal of cellular distress or unprogrammed cell death can be interpreted as a nonself marker. It is a benefit of the danger model that it accommodates the immune reactions that occur as part of normal wound healing regardless of the presence or otherwise of foreign pathogens. The attempt of Langman & Cohn to deal with the threat the danger model presents to their own theory by defining it as “self”, that is, as a detail of their own mechanism, is ultimately unsuccessful.

3.4.2.2 Silverstein

Silverstein takes a different line. His approach is basically to assert that it has all been said before and then throw all the theories out the window!

. . . appeals to some type of teleological concept of the immune response such as Matzinger’s “danger signal” or Janeway’s “stranger signal” as the stimulus for activation of the immune response are unnecessary.. . . there is little evidence on which to conclude that the central issue in the evolution of the immune system is the distinction between self and nonself. (Silverstein & Rose 1997, p 204)

. . . all antigenic challenges obey the same rules. . . . all immunogenic substances activate the full measure of responses available to the immune system, both specific and non-specific. . . no recognitive component of the immune system, is able to discriminate between dangerous and innocuous stimuli. (Silverstein & Rose 1997, p 204)

But different immunogenic substances activate quite different responses. The immune response to an extracellular bacterium is quite different to the response mounted to an intracellular virus. The immune system does make very fine discriminations. Also Silverstein’s assertion that “no recognitive component of the immune system is able to discriminate between dangerous and innocuous stimuli” is probably simply wrong.

Some nonpathogenic substances can elicit immune responses, but almost never do so in the absence of adjuvants that contain bacterial products (Fuchs, Ridge & Matzinger 1996). Janeway discusses this in detail, referring to the “immunologist’s dirty little secret” (Janeway in Tauber 1998, p 460) being the need to use adjuvant material (mostly bits of killed bacteria) to obtain an immune response to various proteins and their conjugates. Janeway states that the belief that:

. . . all foreign macromolecules are equally able to give rise to an immune response . . . is wrong. . . immunogenicity . . . requires both the presence of a suitable antigenic determinant to signal the lymphocyte through its antigen receptor and distinct signals derived from host cells. (Janeway 1989, p 5)

It is difficult, in any case, to envisage how Silverstein’s version of the immune system would function, with all components of the immune system mobilised equally against every foreign invader, dangerous or otherwise. At a very fundamental level this does not account for the tolerance of commensal

microorganisms, let alone do justice to the complexities of a highly specific and finely tuned network of cellular interactions.

Janeway is a supporter of the self/nonself discrimination model. He accepts the findings of Matzinger and co-workers (Forsthuber, Yip & Lehmann 1996; Ridge, Fuchs & Matzinger 1996; Sarzotti, Robbins & Hoffman 1996) with regards to neonatal tolerance. He agrees that the findings of the three articles indicate that there is no “window” of neonatal tolerance, and that neonates are immunologically competent; that the differences in their responses is quantitative rather than qualitative. However he does not follow with support for her danger model. He asserts that although

. . . no longer can we blithely speak of a window of tolerizablility in the neonate [they] have not, however, upset current paradigms of immunology. (Janeway, Goodnow & Medzhitov 1996, p 522).

Janeway stands out from other commentators because he critiques her

experimental approach rather than just engaging in debate at a conceptual level. He claims that by using H-Y male antigen in female mice in her experiments on neonatal tolerance she is in fact demonstrating that a belief in self/nonself

discrimination is integral to the design of “virtually all experiments in immunology” (Janeway, Goodnow & Medzhitov 1996, p 520). However, Matzinger claims that their experiment is entirely consistent with the danger model. They used the H-Y male antigen in female mice so that they could be sure that it was a protein the females’ immune systems had never encountered before, it is also, apparently, a protein to which it has traditionally been hard to provoke a response. Their results were different to those previously obtained for two reasons: firstly they

administered the antigen with a dose of antigen presenting cells (or dendritic cells) and secondly they claim that the surgery required to obtain the dendritic cells would have provided the activating danger signal (Pennisi 1996, pp 1666-67; Ridge, Fuchs & Matzinger 1996). Although the H-Y antigen was foreign to the female mice, the foreigness on its own was not a sufficient condition to provoke a response, there is far more required to provoke an immune response, therefore,

than just the presence of a foreign antigen.

3.4.3 THE DANGER MODEL SURVIVES ITS CRITICS

None of the criticisms of the theory so far have been convincing. It has the benefits of being able to explain many observations about the immune system for which the self/nonself discrimination model has been unable to account. As another sign of its merit, it is proving useful in areas such as oncology and transplantation surgery (Blumberg & Heal 1996; Lafferty and Gazda 1997), and is also now being referred to in articles on various aspects of immunology as a replacement for self/nonself discrimination (Colaco 1998; Gaston 1998; Nickoloff & Turka 1996; Uetrecht 1999). It also remains convincing when compared with the following range of recently offered alternative theories of immune function.