Proc. Natl. Acad. Sci. USA Vol. 89, pp. 10365-10369, November 1992 Immunology Adaptive cellular interactions in the immune system: The tunable activation threshold and the significance of subthreshold responses (context dis nation/autoreavity/meanr/m oy units) Zvi GROSSMAN*t AND WILLIAM E. PAULt *Department of Physiology and Pharmacology, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv 69978, Israel; and tLaboratory of Immunology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD 20892 Contributed by William E. Paul, July 21, 1992

ABSTRACT A major challenge for Immunologists is to (iv) Memory: Static or dynamic? Memory cells show signs explain how the immune system adjusts its responses to the of activation but undergo limited division before challenge. microenvironmental context in which antigens are roi. Their inherent longevity has been questioned, but whether We propose that lymphocytes achieve this by tuning and they require continued stimulation by residual antigen re- updating their responsiveness to rcurrent signals. In partic- mains unresolved (6). ular, cellular anergy in vivo is a dynamic state in which the (v) The in vivo/in vitro dichotomy. Recent experiments in threshold for a stereotypic mode of activation has been ele- which animals are rendered deficient in specific genes have vated. Anergy is associated with other formsofcellular activity, provided some surprises. Interleukins 2- and 4-deficient mice not paralysis. Cells engaged in such subthreshold interactions appear surprisingly normal (7, 8). In transgenic mice over- mediate functions such as maintenance ofimmu l mem- expressing major histocompatibility complex (MHC) class II ory and control of infections. In such interactions, patterns of antigens, immune responses were largely unaffected and signals are recognized and ciasied and evoke selective re- there was no indication ofuncontrolled lymphoproliferation, sponses. The robust mechanis proposed for segregation of immunodeficiency, or autoimmunity (9). Finally, anergy as suprathreshold and subthreshold immune responses allows assayed in vitro correlated in some cases in vivo with memory lymphocytes to use recognition of self-antigens in executing (10) and activation (11) rather than tolerance. physiological functions. Autoreactivity is allowed where it is (vi) Complexity ofsignaling. Many peptide factors, includ- dissociated from uncontrolled on. ing lymphokines, are multifunctional, have a broad range of target cells, and the activity of one peptide in a single cell Despite the major contributions of molecular and cellular depends on the other signal molecules present (12). More- immunology, the physiological rules of immunity are poorly over, the existence of bidirectional communication between understood (1). Consequently, progress achieved in autoim- the immune system and the hemopoietic and neuroendocrine mune diseases, transplantation tolerance, allergy, and systems is now established. There is a growing schism (13) chronic infectious diseases has been limited (2). Some of the between the broadening scope of physiologic connections unresolved issues are listed below. and the traditional concept of antigen-oriented lymphocytes (i) The enigma of autoreactivity and anergy. There is that function as the specialized members of a stereotypical benign autoreactivity in healthy individuals, and the presence defense system. Why should lymphocytes, with their clonally of autoreactive lymphocytes is pronounced during infection distributed receptors for antigens, require so rich and so (3). Therefore, self/nonself discrimination may be much intricate communication among themselves and with other more complex than the simple failure of competent lympho- cells? cytes to recognize self-antigens. (vii) How does the immune system discriminate antigens by Cellular unresponsiveness (anergy) has been proposed as a context? There is a rich repertoire of cellular responses, major self-tolerance mechanism in addition to clonal deletion mediated by different molecules and their combinations, (4, 20, 27, 29). Natural autoreactivity could be ascribed to ranging from direct killing of the antigen-bearing cells to the cells that have escaped both mechanisms. However, the activation, or inactivation, of these cells and/or other lym- orderly nature of autoreactivity needs to be explained: the phocytes and accessory cells. The selection of a particular targets of both benign autoreactivity and autoimmunity in response depends not only on the nature and quantity of the different individuals are not random, but restricted and antigen, but also on the tissue and site of encounter and on predictable (5). The very concept ofanergy as a self-tolerance other cofactors. An illustration is provided by the striking mechanism is problematic: one might expect that the number differences in the response to the antigens ofMycobacterium of autoreactive cells, potentially the cause of autoimmunity, leprae in tuberculoid and lepromatous leprosy (14). The should be reduced to a minimum. immune system appears to recognize and classify different (ii) Multiclonal activity. The extensive involvement in patterns ofsignals, even within the same tissue. The respond- specific immune responses of clones that lack receptors for ing cells, in turn, differ in the pattern of cytokines they the immunogen has not been satisfactorily explained. It is secrete and in the effector functions they execute, both often attributed to so-called bystander effects, to progression individually and collectively. The explanation, in terms ofthe of idiotypic signals along a network, or to other pathological properties of individual cells, of how the immune system effects. adjusts its response to the environmental context in which (iii) Suppression. Attempts to formulate the principles antigen is recognized is a major challenge. underlying the interplay of suppression and help have been The unresolved questions listed above are taken as indi- limited. The issue ofthe controversial suppressor cells is still cations that currently accepted hypotheses may be suscep- confusing. tible to modification. The following analysis is an attempt to

The publication costs ofthis article were defrayed in part by page charge Abbreviations: TAT, tunable activation threshold; TCR, T-cell re- payment. This article must therefore be hereby marked "advertisement" ceptor; APC, antigen presenting cell; MHC, major histocompatibility in accordance with 18 U.S.C. §1734 solely to indicate this fact. complex. 10365 Downloaded by guest on October 2, 2021 10366 Immunology: Grossman and Paul Proc. Natl. Acad Sci. USA 89 (1992) examine the logical consequences of varying the presuppo- The Tunable Activation Threshold (TAT) Model sitions about the nature of immunological functions by way of deductive reasoning. Let us assume that the direct effect on a cell of an external stimulus is changing the metabolic state of the cell in a way The Adaptable Lymphocyte Hypothesis: Presuppositions that can be expressed in quantitative terms. The metabolic change will be referred to as excitation and it is assumed, for Infection can, in theory, sneak through classical immunolog- simplicity, that there is a (positive) scalar measure, termed ical surveillance if its progression is gradual (15-17). Typical excitation level, of this excitation. of many real-life infections is a fast initial increase in antigen Two classes of models may be proposed: balance-of- concentration. It is this kinetic aspect that leads, partly due growth models (15-17) and cell-tuning models. Both regard to time delays along the feedback loops that regulate the the lymphocyte as a multistate system with alternative state immune response at the cell population level, to an overshoot transition pathways, e.g., (i) maintenance: rest -) excitation in the number of effectors and to effective elimination of the rest; (ii) self-renewal: rest -+ excitation proliferation antigen (15-17). (=self-renewal) -+ rest; (iii) activation: rest excitation On the other hand, a destructive immune response aimed proliferation (=amplification) -* differentiation. at the elimination of all antigen-bearing cells may be unde- Only the third process is defined as activation: induction of sirable in the case ofa slow but widespread infection. In such proliferation and (stepwise) differentiation of at least some cases of chronic infection, or residual infection (when the progeny. Balance-of-growth models are concerned with the initial growth burst of infectious agent has been controlled), balance between processes ii and iii, whereas cell-tuning different modes of resistance are required in which patterns models focus on i and ii and their dynamic segregation from of signals are recognized and classified and selective re- iii. sponses are evoked. We shall loosely refer to this class of While balance-of-growth models emphasize Darwinian- responses as environment regulated to underscore their type selection and the proliferation-differentiation interplay, adaptability. These responses are characterized by a variety more recent evidence has indicated that cell viability can be oflocally applied effector and regulatory functions and a fine promoted independently of growth (19) and that processes discrimination of antigenic stimuli based on the associated such as thymic selection of T lymphocytes, during a critical signals (the context). The functions could be implemented by phase, or memory cell maintenance do not involve significant directed secretion of lymphokines over extended periods of proliferation. These observations have refocused modeling time. efforts onto the so-called resting-state and the maintenance Presupposition 1. Conventional immune responses are pathway. It is largely through this pathway that environment- elicited by fast increases in antigen (or infected cells), require regulated responses are thought to occur. strong accessory stimuli (18), and are both transient and Proposition 1. For a stimulus to cause cell activation, the aggressive. Environment-regulated responses are associated excitation level must exceed an activation threshold. Con- with a moderate variation in amount of antigen and/or with ventional and environment-regulated responses are, essen- quasi-stationary sets of stimuli. tially, suprathreshold and subthreshold responses, respec- Each lymphocyte must develop and sustain its functional tively. characteristics in the face of a varying (noisy) stimulatory Proposition 2. A lymphocyte is inherently capable of environment. This means, in particular, that lymphocytes participating in both types of immune response. Rapidly engaged in environment-regulated activities may be resistant changing perturbances tend to induce suprathreshold re- to standard activation, either through differentiation or by sponses, whereas gradual ones impose subthreshold re- virtue of their ongoing activities. sponses. The state in which a lymphocyte is nonresponsive to a Proposition 3. When engaged in persistent subthreshold standard mode of activation conforms to the operational interactions, cells are protected against chance activation. definition of anergy. Thus, we suggest that anergic lympho- The resistance to the induction of activation is imposed by cytes from immunized or infected animals are often immu- their ongoing activity. nocompetent cells, differentiated or induced into unrecog- The TAT model accommodates these propositions by the nized functional states. following assumption: The activation threshold is dynam- The next logical step is linking these states to the contem- ically modulated by the cell's environment and it is geared to porary concept ofanergy as a barrier to autoimmunity. By the the recent excitation history of the cell. presupposed constraints on conventional immune responses, To be more precise, some time-dependent, weighted av- such responses are restricted, a priori, mainly to strongly erage of the cell's past excitation levels will be defined and invasive infectious agents and would rarely be elicited by named excitation index. Upon each excitation event, the cell autologous antigens. This, and the existence of other con- undergoes a metabolicperturbation, defined as the difference trols, may preempt the need for clonal elimination of all between the presently induced excitation level and the ex- self-reactive lymphocytes (with the exception, perhaps, of citation index. Thus, the perturbation is a measure of vari- is not ation (note that it may have positive as well as negative particularly high-avidity clones). If, indeed, anergy values). The assumption is that the perturbation must exceed defective responsiveness but a means for restricting cellular a fixed critical value to cause activation. The activation activity to within desirable bounds, then autoreactive cells threshold therefore equals the excitation index plus that could be available for environment-regulated activities. This critical value. As the excitation index is updated continuously would provide a rationale for the abundance of anergic cells in the course ofrepetitive stimuli, comparable changes occur in vivo. in the activation threshold (Fig. 1). Presupposition 2. Lymphocytes regularly employ recog- The excitation index, E(t), at time t could, for example, be nition of self-antigens (not only as restriction elements) in related to the excitation E(t) by the equation executing physiological functions. More explicit rules of organization, or models, need to be dE(t)/dt = aE(t)[E(t) -E(t)] [11 explored. Such rules should suggest, in particular, how the functional segregation of immunological responses may rea- where a is a constant. E0) then smoothly traces the excitation sonably come about. E(t). Downloaded by guest on October 2, 2021 Immunology: Grossman and Paul Proc. Natl. Acad. Sci. USA 89 (1992) 10367 facilitate intercellular communication and that allow an ex- cited cell to excite other cells. (iv) Induction of self-renewal division. Subcritical positive perturbations of a sufficiently large magnitude are associated with a probability of latent (16, 17), or sterile (20), proliferation-namely, proliferation that is not followed by a stepwise differentiation and that may involve kinetic characteristics and molecular mechanisms different from activation-related proliferation. It has been reasoned (16, 17) that lymphocytes are more readily induced into such proliferation than into a full activation. (v) Induc- tion ofadaptive, environment-instructed differentiation (as in the differentiation of thymocytes to express either CD4 or CD8 cell-surface markers; ref. 21). Such differentiation may be induced by appropriate stimuli exerting positive subcrit- ical perturbations ofsufficient magnitude and/or persistence. ts TIME The underlying difference between activation-associated dif- ferentiation and differentiation induced by subcritical pertur- FIG. 1. Schematic illustration of the variation of the cell's exci- tation, excitation index, and activation threshold with time (arbitrary bations is that ofapreprogrammed versus an adaptive change units). Numbers refer to excitation events. Sign indicates whether (22). Adaptive differentiation may need several maintenance- the associated perturbations are positive or negative. Only the 4th like processes to be completed. Indeed, differentiation often event could lead to activation. At time t. and later, activation is requires continuous, active regulation and is controlled by prohibited. mechanisms that are dynamic and reversible (23). This view of subthreshold excitations (or, equivalently, The excitation index could be measured by the cell through subcritical perturbations) emphasizes the role of cellular the concentration of some molecule(s) undergoing storage- interactions and self-organized cellular environments in the degradation cycles during excitation events in a certain facilitation and regulation of cell function. Some corollaries domain of the cell. The opposing action of tyrosine kinases are (i) to maintain cellular viability over extended periods of and phosphatases may represent a process of this type. time, a compound measure of the frequency of encounters Supposing that the production and degradation rates are both and of the cells' excitabilities must exceed a cell-coupling increasing functions of the excitation E(t), we may have, for threshold: below this threshold, both the excitability and the example, viability decay, between excitation events, at a faster rate than the rate at which these quantities are promoted during dC(t)/dt = E(t)[bE(t) - aC(t)], [2] such events. (ii) A feedback relationship between cell density and the probability of self-renewal division could act as a where C(t) is the molecule's concentration and a and b are local homeostatic mechanism, so that a loss of cells by constants. Denoting (a/b)C(t) = 4ft), Eq. 2 reduces to Eq. 1, migration or death would trigger a compensatory prolifera- so the excitation index is proportional to C(t). tion: as the number of cells is transiently reduced, the Propositions 2 and 3 become interpretable in mechanistic encounters become less frequent but also less regular, and terms as corollaries of the assumption described above. (i) this could, at first, enhance the probability that larger-than- Activation resistance. Cells maintained at a high level of average perturbations occur, leading to self-renewal events. excitation have correspondingly high activation thresholds (iii) A high average level of excitation may prohibit cells not and therefore are relatively insensitive to activation or are only from being activated but also from displaying the full partly anergic. Moreover, since there is some upper limit to range of subthreshold responses. Under some circum- the excitation level that can be reached, and since perturba- stances, such restriction may be detrimental to the cell. tions are bounded by the difference between this saturation The concept of tunable responses to sets of stimuli, de- level and the excitation index (see Fig. 1), a sufficiently high pending on the magnitude and time contingency of these average level of excitation may prohibit cells from being stimuli, provides a powerful approach for the consideration activated-i.e., they are completely anergic. (ii) Rate-of- of immune activation, tolerance, and memory. change sensitivity. Upon gradual increase in the levels of excitation, reached through a sequence ofincreasingly strong Biological Insights excitation events, a cell is not likely to be activated. Rather, it will become progressively anergic. This is because the Conventional immune response. According to the TAT perturbation associated with each excitation event is small. model, lymphocyte activation requires strong, positive per- turbations. In other words, excitation levels must signifi- Subthreshold Interactions cantly exceed the excitation index. For primed lymphocytes to regain responsiveness for a subsequent challenge, their Although major phenotypic alterations occur as a result of excitation index must reassume a sufficiently low value. activation, subthreshold excitation also results in a response Under the selective influence of the antigen, clusters of and in changes in the response characteristics ofthe cell. The cells are assembled (24) in which both antigenic and other effector functions that are expressed by activated versus signals may be delivered to lymphocytes with the time and subthreshold-excited cells differ in terms of intensity, local- space contingencies required for rapid induction of a high ization, reversibility, and tunability, although the same cy- level of excitation. These other signals may be particularly tokines may be secreted. Excitatory stimuli that do not potent when induced by infectious agents (18); the coordi- induce activation may have the following consequences: (i) nated delivery of such signals together with the antigenic Tuning ofthe activation threshold. (ii) Promoting the viability stimulus makes these agents inherently strong immunogens. of the cell (19). The effect of a stimulus is transient, so that In the course of such an immune response, a portion ofthe repeated stimuli are required to maintain the viability. (iii) activated lymphocytes will undergo irreversible differentia- Promoting the excitability of the cell. This is achieved by tion into an effector state, while other excited cells will be inducing the transient expression of membrane-associated rendered temporarily anergic to activation, constituting the receptors, adhesion molecules, and secretory factors that memory cell population. The transient nature ofmany natural Downloaded by guest on October 2, 2021 10368 Immunology: Grossman and Paul Proc. Natl. Acad. Sci. USA 89 (1992) challenges, with an increase and then a decrease in antigen between cell density and the probability of self-renewal concentration, would allow for resetting the excitation index would act as a local homeostatic mechanism. of memory cells to a lower value and for recovery of These highly selective interactions may occur within responsiveness between challenges. On the other hand, when unique microenvironments in lymphoid organs. In these the infectious agent persists, the continued updating of the environments, cells assembled under the selective influence excitation index may result in no stimulus causing a pertur- of the antigen at the time of challenge would continue to bation that exceeds the critical value. cooperate, even in the absence of antigen, to provide stimuli T-cell development. The generation ofself-tolerance within to each other. The concept of an autocatalytic stimulation of the thymus (4, 27, 30) involves positive and negative selec- memory lymphocytes should pertain equally to the accessory tion, as well as the induction of anergy. These events may be cells. The accessory functions of nonlymphoid cells are correlated with the affinity of the T-cell receptor (TCR) for inducible (26), as are the helper functions of T-helper cells. the ligand expressed on the thymic antigen presenting cell We suggest that they are sustained by the very same inter- (APC) (25). Cells with the highest affinity are clonally de- actions discussed above. This would imply the existence of leted. Furthermore, it has been suggested that cells with a memory accessory cells. According to this view, residual TCR of intermediate affinity often survive but are rendered antigen is not required to sustain dynamic memory, although anergic. its effect is not excluded. If it does play a role, its role is to This scenario can be explained by the TAT model. Thy- excite, not to activate. The fact that antigen is required to retain memory in cell transfer experiments can be explained mocytes that do not recognize self-MHC molecules fail to by the necessity of reassembling operative memory units in receive excitation signals and consequently are short-lived. a new environment and initiating the self-driven memory We propose that the excitation of thymocytes that recognize maintenance process. self-MHC molecules on the thymic APC is gradually en- Subthreshold interactions and infection. Chronic or resid- hanced due to increased expression of the TCR and acces- ual infection elicits, preferentially, subthreshold responses. sory molecules, entailing an increase in the cell's activation The first step in such a response depends on receptor- threshold. This would result in desensitization to activation mediated binding to antigen-expressing target cells. The rather than in activation. The degree of desensitization, or resulting excitation, however, depends not only on binding anergy, is directly related to the affinity of the thymocyte for per se but also on the availability ofother signals and on other the self-antigen. Since the lower-affinity cells are less po- response characteristics of both the responding and the tently desensitized, having reached a lower excitation index, stimulating cells. If the excitation exceeds the cell coupling they are usually capable of being activated by some combi- threshold, a positive cycle ofenhancement at the levels ofcell nation offoreign peptides and the selecting MHC molecules. adhesion and excitation is started. Therefore, the formation Thus, these cells are anergic to the positively selecting of local anti-infection networks of lymphocytes, target cells, element but not to conventional foreign antigens. However, and accessory cells depends both on antigen recognition and there is no reason to assume that the anergy expressed by on the stability of the dynamic relationship. These networks these cells and by truly anergic cells (cells rendered anergic may control, but not necessarily eliminate, infection through to a self-peptide plus MHC) are essentially different in nature. the production of factors such as interferons, tumor necrosis The elimination of high-affinity, and therefore highly ex- factor, or lymphotoxin. cited, cells can be attributed to the putative detrimental effect Anergic cells should be thought of as functionally active, of protracted high excitation. A cell's early developmental not inert. In fact, in several situations of a chronic infection stage may, or may not, be relevant. It is possible that strong with a virus or with other parasites, lymphocyte anergy as desensitization of high-affinity cells by hemopoietic APCs assayed in vitro is paradoxically (11) associated with immune interferes with their continued interactions with thymic epi- hyperactivity in vivo. TAT interprets this association as thelial cells, which may be essential for their further differ- resulting from persistent interactions that raise the excitation entiation and/or survival. index ofthe participating cells. Such interactions may reflect, Subthreshold interactions and memory. The concept of in part, an effort by foreign antigen-specific and autoreactive excitability implies, by definition, the existence of a short- cells to control the infection. term cellular memory and suggests an interesting viewpoint Thus, under the cover of a relatively high activation of the dynamic relationship between memory cells and their threshold, autoreactive lymphocytes could participate in environment. Short-term memory at the cellular level can providing a variety of effector functions. This suggests a facilitate long-term memory at the levels ofthe cell group and rationale for autoreactivity and for anergy as a selective the cell population. Transitory changes in excitability can barrier to autoimmunity. stabilize mutual interaction of the memory cell with other While the expression by positively selected lymphocytes of cells in its environment that display appropriate counter various anti-self receptors may be largely random, the actual receptors (e.g., TCR, CD4, and/or CD8 interact with class II recruitment of autoreactive cells into relatively stable dy- or class I molecules; CD28 interacts with B7; CD5 interacts namic relationships with potential target cells in the periphery with CD72). This can be accomplished even if unstimulated is restricted by the requirement for an active cooperativity cells have a relatively short life-span and the induced en- between the cells. This could explain the orderly nature of hancement in the excitability lasts for only a short interval. autoreactivity and autoimmunity: only a subset of the self- Upon each encounter with the appropriate stimulator cell(s), recognizing lymphocytes, directed at tissue cells with appro- the lymphocyte's viability and responsiveness may be suffi- priate physiological activity profiles, will manifest autoreac- ciently augmented to allow for a viable, responsive state to be tivity under certain conditions. maintained indefinitely. Like many other positive feedback The definition of T cells as anergic and autoreactive processes, this one would also be threshold dependent in formally applies to all T cells, including those specific for terms of the initial increase in the responsiveness and the MHC-foreign peptide combinations; the differences are frequency ofencounters (the cell-coupling threshold). Tuning quantitative. Memory cells could use their nominal reactivity the activation threshold to the average excitation level acts to for self-MHC molecules in the proposed interactions with minimize the perturbations, reducing the probability for cell accessory cells. In an infected tissue, this same nominal activation, while at the same time maintaining a degree of autoreactivity of some (primarily) foreign antigen-specific sensitivity suitable for reactivation in response to the original lymphocytes might be upgraded via the synergistic effect of antigen in an immunogenic context. A feedback relationship other infection-induced signals. This might explain, beyond a Downloaded by guest on October 2, 2021 Immunology: Grossman and Paul Proc. Natl. Acad. Sci. USA 89 (1992) 10369 bystander effect, the extensive participation of nonspecific (vi and vii) The last issues in the list are interrelated: the lymphocytes in specific immune responses. complexity of signaling and regulatory pathways probably Suppression. The rationale for the dynamic constraints has to do with context discrimination, since context can be imposed on cell activation is that such constraints are, represented by associations of signals (12, 13, 28). The physiologically, a barrier to autoimmunity. The constraints scenario that emerges from this notion of context discrimi- on the induction of latent proliferation are less stringent. nation is of a range of effector cells, distinguished by the Various situations can be envisioned in which partly anergic patterns of cytokines they secrete, interacting simultane- or low-affinity lymphocytes would undergo extensive self- ously and selectively with a range of respective target cells renewal (16, 17). Tolerant cells could inhibit the activation of and accessory cells that provide distinct patterns of signals. nontolerant cells through competition for stimulatory sig- This scenario makes sense only under the constraints of nals-e.g., if the number of APCs is limiting-or by exerting subthreshold interactions, where both effector cell turnover direct suppressive effects. Such inhibition may be a regular and target cell destruction are restricted. Therefore, the correlate of cell excitation in general and not associated with adaptable lymphocyte hypothesis, by rationalizing the feasi- a particular cellular phenotype. Thus, tolerant-but excit- bility of such interactions, sets the ground for further theo- able-cells could be the real, elusive suppressor cells (17), as retical and experimental explorations of generalized immu- has been recently reiterated (27). Since, in general, these nological functions. anergic suppressor cells are assumed to mediate positive The work was supported in part by the Harry Palley Endowment immunological functions, the suppression is not an immuno- Fund of the United Jewish Federation Foundation. logical function for its own sake; rather, suppression of one immune response is a facet ofsome different, more quiescent, 1. Paul, W. E. (1987) J. Immunol. 139, 1-6. ongoing response. 2. Coutinho, A. (1989) Immunol. Rev. 110, 63-87. 3. Shoenfeld, Y. & Isenberg, D. (1989) The Mosaic of Autoim- Conclusion munity (Elsevier, Amsterdam). 4. Blackman, M., Kappler, J. & Marrack, P. (1990) Science 248, Our hypothesis consists of three elements, which are self- 1335-1341. 5. Cohen, I. R. (1992) Immunol. Today, in press. consistent but not entirely interdependent: (i) functionally 6. Beverley, P. C. L. (1990) Immunol. Today 11, 203-205. distinct immune responses to infection, and the redefinition 7. Schorle, H., Holtschke, T., Hunig, T., Schimpl, A. & Horak, of anergic cells as activation-resistant but excitable cells, (ii) I. (1991) Nature (London) 352, 621-624. tunable cellular response characteristics, and (iii) biologically 8. Kuhn, R., Rajewsky, K. & Mueller, W. (1991) Science 254, important subthreshold interactions. 707-710. The positive definition of anergy contradicts most cur- 9. Fehling, H. J., van Ewijk, W., Pasquali, J. L., Waltzinger, C., rently held views, with some exceptions (2). TAT is consis- LeMeur, M., Gerlinger, P., Benoist, C. & Mathis, D. (1990) J. Immunol. 144, 2865-2874. tent with an inherent potency ofinfectious agents (18), but the 10. Bandeira, A., Mengel, J., Burlen-Defranoux, 0. & Coutinho, constraints imposed here on lymphocyte activation are dy- A. (1991) Int. Immunol. 3, 923-931. namic in nature and have a different range of connotations. 11. Ascher, M. S. & Sheppard, H. W. (1990) J. AIDS 3, 177-191. To study these dynamic aspects, molecular measures of 12. Sporn, M. B. & Roberts, A. B. (1988) Nature (London) 332, viability, excitation, excitability, and excitation index, or at 217-219. least some surrogate markers of those, would have to be 13. Grossman, Z. (1989) in Theories of Immune Networks, eds. identified. It should be noted that sensitivity tuning is con- Atlan, H. & Cohen, I. R. (Springer, Berlin), pp. 38-52. 14. Yamamura, M., Uyemura, K., Deans, R. J., Weinberg, K., ceived of as the adaptive behavior of cells which are com- Rea, T. H., Bloom, B. R. & Moldin, R. L. (1991) Science 254, ponents of organized structures and are subject primarily to 277-279. subthreshold interactions. Our ability to reconstruct or mon- 15. Grossman, Z. & Berke, G. (1980) J. Theor. Biol. 83, 267-296. itor such conditions is at present extremely limited. 16. Grossman, Z. (1982) Eur. J. Immunol. 12, 747-756. Looking back at the unresolved questions listed at the 17. Grossman, Z. (1984) Immunol. Rev. 79, 119-138. beginning, it may be seen that the adaptable lymphocyte 18. Janeway, C. A. (1989) Cold Spring Harbor Symp. Quant. Biol. hypothesis opens some avenues to approach these questions. 54, 1-13. (i) Autoreactivity is dissociated from uncontrolled aggres- 19. Williams, G. T., Smith, C. A., Spooncer, E., Dexter, T. M. & sion. The anergy Taylor, D. R. (1990) Nature (London) 343, 76-79. paradoxes of have been accommodated. 20. Sprent, J., Gao, E.-K. & Webb, S. R. (1990) Science 248, The orderly nature of autoreactivity and autoimmunity is 1357-1363. consistent with the concept of tuning. (ii) A positive expla- 21. Borgulya, P., Kishi, H., Muller, U., Kirberg, J. & von Boeh- nation ofapparently nonspecific responses has been offered. mer, H. (1991) EMBO J. 10, 913-918. (iii) A concept of suppression arises as a facet of a compe- 22. Kirschner, M. & Mitchison, T. (1986) Cell 45, 329-342. tition between different modes of immunological response. 23. Blau, H. M. & Baltimore, D. (1992) J. Cell Biol. 112, 781-783. (iv) The hypothesis reconciles dynamic and static aspects of 24. Mitchison, N. A. (1989) Chem. Immunol. 46, 157-168. memory cells. (v) The dichotomy of in vitro and in vivo 25. Sprent, J. & Webb, S. R. (1987) Adv. Immunol. 41, 39-133. observations is inevitable ifin vitro we observe only the most 26. Weaver, C. T. & Unanue, E. R. (1990) Immunol. Today 11, stereotypic modes of cellular behavior. Redundancy of cells 49-55. 27. Ramsdell, F. & Fowlkes, B. J. (1991) Science 248, 1342-1348. and molecules alone may not be sufficient to explain the 28. Bloom, F. E. (1984) Am. J. Physiol. 246, C184-C194. resilience of in vivo processes. For example, the property of 29. Burkly, L. C., Lo, D. & Flavell, R. A. (1990) Science 248, autoscaling cellular responsiveness predicts and explains the 1364-1368. observations in transgenic mice overexpressing MHC anti- 30. von Boehmer, H. & Kisielow, P. (1990) Science 248, 1369- gens. 1373. Downloaded by guest on October 2, 2021