Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Psychoneuroimmunology wikipedia , lookup
Major histocompatibility complex wikipedia , lookup
Immune system wikipedia , lookup
Lymphopoiesis wikipedia , lookup
Immunosuppressive drug wikipedia , lookup
Cancer immunotherapy wikipedia , lookup
Adaptive immune system wikipedia , lookup
Polyclonal B cell response wikipedia , lookup
Molecular mimicry wikipedia , lookup
)^ and Cell Biology {1996) 74, 278-285 Theoretical Article Signal minus 1: A key factor in immunological tolerance to tissue-specific self antigens? CHRISTOPHER R PARISH Division oj Immunology and Cell Biology. John Curtin School of Medical Research. Australian National University Canberra. ACT, Australia Summary Recent data suggest that many autoreactive T cells, particularly to tissue-specific self antigens, can escape thymic deletion. The current dogma is thai these autoreactive T cells are silenced by the failure of most tissues to provide co-stimulation (signal 2), antigen alone (signal 1) inducing T cell unresponsiveness. However. I propose that activation of autoreactive T cells frequently occurs but autodestruction by effector T cells is tightly regulated. This phenomenon is most evident with lymph node metastasizing tumour cells where the regional lymph node can mount a vigorous response to the invading tumour cells but tumour growth is unimpaired. I suggest that autodestruction is prevented by inhibitory receptors on T cells which recognize class I MHC structures on target cells. These receptors, which I propose deliver 'signal minus T to T cells, were recently described on NK cells and a subpopulation of peripheral T eells. They are also strikingly similar to a family of anti-self receptors that my laboratory described on murine T and B cells 15 years ago. In the 'signal minus 1" model, antigen-activated T cells acquire the inhibitory receptors when they become co-stimulation independent and gain the ability to exit lymphoid organs and enter non-lymphoid tissues. Thus, if autoreactive effector T cells encounter autoantigen in tissues they are functionally silenced by inhibitory receptor engagement and signal minus 1 delivery. In contrast. I propose that in response to intracellular infections, cells down-regulate expression of their ligands for inhibitory receptors. Such a model allows infected cells to be selectively eliminated by effector T cells. If correct, the model predicts that effector T cells, whether foreignantigen- or autoantigen-specific, can selectively respond to infected cells. This apparent 'usefulness' of autoreactive T cells may explain their observed persistence even after an encounter with autoantigen. It is also suggested that signal minus 1 may silence autoreactive B cells specific for tissue-specific cell surface antigens and lack of signal minus 1 may partially explain the vigorous T cell response to allogeneic MHC. Finally, it is hypothesized that, in evolutionar>' terms, inhibition of autodestruction by the recognition of a 'self marker' and delivery of signal minus 1 is an ancient process which probably emerged in early metazoans. Key words: alloreactivity, autoimmunity, inhibitory receptors, peripheral T cell tolerance, signal minus 1. Introduction One of the fundamental questions of immunology is the means by which peripheral T cell tolerance is maintained to tissue-specific self antigens. It cannot be assumed that all autoreactive T cells are eliminated during T cell selection in the thymus.'*-^ There must be considerable numbers of autoreactive T cell clones escaping from the thymus due to the failure of tissue-specific self antigens to be presented by thymic epithelium to maturing T cells. Furthermore, many developmentally regulated antigens are expressed after the appearance of a mature immune system. Thus it is reasonable to assume that autoreactive T cells to tissue-specific self antigens exist in secondary' lymphoid organs. Studies with transgenic mice support this Correspondence: Christopher R Parish, Division of Immunology and Cell Biology, John Curtin School of Medical Research, Australian National University. Canberra, ACT 2601, Australia. Received 31 January 1996; accepted 6 February 1996. view, autoreactive T cells often appearing in the periphery when the autoantigen is expressed extrathymically.'How then is reactivity against tissue-specific self antigens prevented while responsiveness to foreign antigens is maintained? The current dogma is that T cell immunity to tissuespecific self antigens is curtailed by the inability of most cells to constitutively express co-stimulator\' molecules (signal 2) required for T cell activation.'--^ A corollary of this model is that antigen alone (signal 1) induces nonresponsiveness in T cells.'"-^ This simple 'two signal" model has been used extensively to explain many aspects of T cell immunity. In this paper 1 propose, however, that the two signal model cannot totally explain T cell tolerance to tissue-specific self antigens, this deficiency of the model being particularly apparent with metastasizing tumour cells. An alternative model is presented which, I believe, can more adequately explain peripheral T cell tolerance. 279 Tolerance to tissue-specific self antigens The paradox of lymph node metastasizing tumour cells In many ways lymph node metastasizing tumour cells are an excellent model for studying the control of T cell immunity to tissue-specific self antigens. The tumour cells inevitably carry many tissue-specific self antigens, as well as tumour-specific antigens."*^ which can be recognized by substantial numbers of T cell clones that have survived thymic selection. In fact, it has been known for many years that the frequency of cytotoxic precursors for a syngeneic tumour can be high, resembling the precursor frequency for allogeneic tumour cells.^ Furthermore, by penetrating the lymph node the tumour cells are accessible to naive T cells which, it is believed, usually do not recirculate through tissues.^ Despite the existence of tumour-reactive T cells, many tumours metastasize via lymph nodes without apparently inducing a destructive immune response. In fact, the regional lymph node is a major route of metastasis for many epithelial-derived tumours.^ Nevertheless, it has been recognized for some time that substantial T cell proliferation can occur when tumour cells enter regional lymph nodes."' Despite this vigorous T cell response, the tumour spreads rapidly to other tissues and eventually kills the host. In contrast, when the same tumour ceils are injected into MHC-incompatible hosts they are usually rapidly rejected by a potent T-cell-mediated immune response. These observations provide several paradoxes which cannot be easily explained by the two signal model of T cell activation. First, despite the lack of co-stimulator\' molecule expression by the tumour cells^ substantial T cell activation is induced in the regional lymph nodes. Second, these activated T cells are very inefficient at eliminating the tumour. Third, if co-stimulation is the limiting factor why are the tumour cells rejected when they are transplanted into an allogeneic recipient? Surely increasing the frequency of tumour-specific T cells in the recipient, due to MHC disparity, will not overcome the lack of co-stimulation provided by the tumour cells. Clearly additional factor(s) must be operating which prevent effective T cell immunity against the tumour associated antigens. In this regard, a model can be developed based on the recently described inhibitory receptors for class I MHC molecules, which provides an explanation for immunoiogicai tolerance to tissue-specific self antigens, allows responsiveness to foreign antigens and also has some implications for alloreaetivity. Inhibitor) class I MHC receptors and autorosetting Recent studies with natural killer (NK) cells have revealed a remarkable family of inhibitory receptors which also may have profound implications for T cell reactivity.'"-'•* The receptors recognize MHC class I molecules on target cells and inhibit NK effector cell function. These inhibitory receptors are also functionally expressed on T lymphocytes. Families of both mouse and human inhibitorv' receptors have been defined and a comparison of their general properties is presented in Table 1. Although the mouse and human receptors share striking similarities in their Table 1 Inhibitory receptors for class I MHC on lymphocytes Receptor properly Mouse receptors Human receptors Nomenclature Ly49A-G p58. NKBl, NKAT(NK assoeiated transeripts) Chromosome location Chromosome 6 Chromosome 19 Cellular distribution NK cells, T eell subset NK cells, T eell subset Family type C-type lectin Ig superfamily Strueture 44kDa disulphide linked homodimer p58: 58 kDa. 2 Ig-like domains NKBl: 70 kDa. 3 Ig-like domains NKAT: 2-3 Ig-like domains Ligand(s) Class 1 MHC ( + speeific peptide?) Anionic carbohydrate Class I MHC + speeifie peptide Specifieity Ly49A: H-2D'' and D"^ Ly49C: H-:*' Ly49G2: H-2D'' and L*^ P58: HLA-A. -B and -C NKB1:HLA-B Expression Clonalty distributed Exhibit allelie exclusion Some adaptation to self MHC Clonally distributed Function Inhibit NK (and T?) eell killing of targets and lymphokine production Inhibit NK and T eell killing oi targets and tymphokinc production Based on data reviewed by Yokoyama;'° Raulet and Held;" Lanier and Phillips;'- Bottino <•/ a/," Liebson;''' and Gumperz and Parham.'* 280 CR Parish cellular distribution and function surprisingly they have no structural homology. The murine receptors (Ly49 members) are C-type lectins whereas the human receptors (p58, NKBl, NKAT) are Ig superfamily members. The reason for this structural discrepancy is unclear although it has been suggested that CD94 is the Ly49 equivalent in humans."^ Nevertheless, the receptors in both species recognize class I MHC molecules containing certain peptides and discriminate between different MHC alleles. For example, Ly49A, which is expressed on a subset of murine NK cells, only interacts with H-2D'' and D"^ (Table 1). Interestingly, some Ly49 molecules have also been demonstrated to bind anionic carbohydrates, an interaction which inhibits their binding to H-2 molecules.' ''^' ^ Recognition of carbohydrate structures by the human receptors has not been demonstrated; in fact the NKBi receptor still recognizes MHC molecules devoid of N-linked carbohydrate.'^ Currently, at least seven Ly49 genes have been described and ten cDNA of the p58/NKBI/NKAT family have been reported. "^"' ^ It is believed that considerable receptor diversity is generated in both species by alternative splicing of a small family of related genes. A particularly intriguing feature of both the mouse and human inhibitory receptor families is that they are expressed on subpopulations of T lymphocytes. Studies with NKBl suggest that it is predominantly expressed on T lymphocytes chronically stimulated by antigen and not by naive T cells.•^° Thus, the inhibitory receptors regulate the efFector function of T cells rather than activation of naive T cells. A fascinating aspect of the inhibitory receptors on NK and T cells is that they are strikingly similar to a family of anti-self receptors that my laboratory described 15 years ago (Table 2). We were examining a rather bizarre phenomenon, the ability of murine lymphocytes to bind autologous erythrocytes, a process termed autorosetting. Our research revealed that lymphocytes preferentially bind //-2-compatible erythrocytes,-'-- a phenomenon described independently by two other groups.--^-"* The autorosetting receptors were found to be H-2L/D restricted and resemble tbe Ly49 family in crossreacting Table 2 Properties of murine autorosetting receptors on lymphocytes Mediate binding of murine lymphocytes to autologous erythrocytes Receptors present on about 75% of thymocytes, 10-15% of peripheral T cells and 20% of B cells^'-^^'^' Binding class I MHC restricted,^'-^'' mapping to H-2L/D regions.-'-^ Recognize anionic carbohydrate structure on erythrocytes in a H-2L/D restricted manner'^—^ No evidence for 'thymic education", i.e. MHC restriction of receptors not determined by H-2 haplotype of thymic epithelium^^ Erythrocyte ligand blocked by the serum protein histidine-rich glycoprotein with some H-2 haplotypes.^'-^^ Furthermore, the autorosetting receptors, like Ly49, interact with anionic carbohydrates,^^"^^ the carbohydrate recognition also being under H-2L/D control.-^ Although MHC-restricted, analysis of chimaeric animals demonstrated that the thymic epithelium did not alter the MHC specificity of the autorosetting receptors.^^ The cellular distribution of the autorosetting receptors, however, differs somewhat from known Ly49 members being expressed on the majority of thymocytes (75%) and substantial subpopulations of peripheral T (10-15%) and B (20%) cells.^'-^^-^'' At the time, the functional relevance of these receptors was unclear, although it was concluded that they could not be directly involved in MHCrestricted recognition of foreign antigen by T cells.-^ It now seems highly likely that the autorosetting receptors belong to the Ly49 family of inhibitory receptors, presumably being family members which are yet to be characterized. Alternatively, the receptors may represent the murine equivalent of the p58/NKBl/NKAT receptors in humans. Further work is clearly required to resolve this issue and formally establish whether the autorosetting receptors can deliver an 'inhibitory signal' to lymphocytes. Finally, one other important feature of the autorosetting system should be noted. A unique serum protein was identified, termed histidine-rich glycoprotein (HRG), which blocks autorosetting by masking the erythrocyte ligand.-'^'^'^ Subsequent studies have demonstrated that HRG exhibits immunoregulatory properties,-^° but the effect of HRG on T and NK cell effector function has not been examined. Hypothesis: does signal minus 1 maintain peripheral T cell tolerance? Based on the available information on inhibitory class I MHC receptors it is possible to propose a plausible model for peripheral T cell tolerance. In this model the "inhibitory signal' delivered by these receptors will be called, for simplicity, 'signal minus V. Inhibition of autodestruction by signal minus 1 Figure I depicts the various stages in T cell activation and the proposed roles of signal 1, signal 2 and signal minus I in this process. It is proposed that initial activation of naive T cells follows the conventional 'two signal' model, antigen alone (signal I) resulting in T cell inactivation unless co-stitnulation (signal 2) is provided. The current dogma is that, due to the availability of appropriate APC, this signal 2 dependent clonal expansion of naive T cells is restricted to lymphoid organs.^ However, a key feature of the new model is that antigen-activated T eells acquire inhibitory receptors (and, therefore, can receive signal minus 1) when they gain the ability to exit lymphoid organs and enter non-lymphoid tissues. Such a proposition ensures that effector T cells specific for tissue-specific self antigens are functionally silenced if they encounter antigen in tissues. Certainly, the limited evidence available is consistent with this proposition, only antigen- Tolerance to tissue-specific self antigens Naive T cell I Signal I LYMPHOID Signal I ORGAN Inaclivalion Activation RESTRICTEDi 1 Signal 2 Clonal Expansion Expression or Inhibitory Receptors ABLE TO Inactivation Min Signal I ENTER TISSUES Signal Signal EfTector T cell Inaclivalion I Signal I Cylotoxicily Lymphokine Release Figure 1 Stages in T cell activation demonstrating the role of signal 1 (antigen), signal 2 (co-stimulation) and signal minus 1 (inhibitory receptor engagement) in T cell activation. In this model it is proposed that initial activation of naive T cells occurs in lymphoid organs. Acquisition of inhibitory receptors occurs when activated T cells become signal 2 (co-stimulation) independent and gain the ability to exit lymphoid organs and migrate into tissues. 'Inaetivation' refers to functional silencing of T cells, not their deletion. experienced T cells appearing to express the inhibitory receptors.'--^ Another important aspect of the T cell activation pathway depicted in Fig. 1 is that acquisition of inhibitory receptors oeeurs when T cell clones become costimulation (signal 2) independent. Obviously T cells do not need to recognize co-stimulatory molecules to mediate their effector function. If co-stimulation was required, target cell recognition would be extremely limited. However, co-stimulation independence poses serious problems of autoreactivity, a problem ameliorated by the expression of inhibitory receptors by effector T cells. Thus, based on the above model, the immune system can tolerate considerable autoreactivity. What is tightly controlled is autodestruction by effector T cells. In fact, it could be argued that thymic deletion is primarily aimed at eliminating T cell clones which recognize self antigens expressed by 'professional' APC, a point already highlighted by Matzinger.^ In contrast, the bulk of autoreactive T cells are regulated extrathymically by signal minus 1. How is selective recognition of foreign antigen achieved? So far this discussion has concentrated on the functional silencing of autoreactive T cells by signal minus 1. If one 28 accepts this concept how does the immune system discriminate between self and foreign antigens? For this discrimination to occur I propose that, in response to intracellular infections, cells down-regulate expression of their ]igand(s) for inhibitor^' receptors. Such a response allows infected cells to be selectively eliminated by effector T cells. At this level the model somewhat resembles the 'danger hypothesis" proposed by Polly Matzinger.-^ Loss of the ligand for inhibitory receptors on infected cells is a fundamental feature of the 'signal minus V hypothesis and warrants further discussion. In the case of NK cells it is clear that target cells become susceptible to lysis if class I MHC is down-regulated and, therefore, inhibitory receptors are unable to engage."^"'^ However, displacement of self peptides from class I MHC molecules can also render targets susceptible to lysis by human NK cells.-^' Thus it appears that subtle changes in the inhibitory ligand on target cells following intracellular infection could prevent signal minus I delivery. In Figs 2 and 3 the inhibitory ligand is depicted as class I MHC + self peptide associated with an oligosaccharide structure, thus incorporating features of both the human and murine inhibitory receptor systems (Table 1). Following viral infection, it is proposed that the oligosaccharide structure is lost and an inappropriate (self or viral) peptide is bound to class I MHC. Whether the inhibitors' ligand is modified as depicted, or in some other way, remains to be determined. Certainly, such molecular details are not required for the development of the hypothesis. Nevertheless, loss of the oligosaccharide component of the inhibitory ligand is an attractive possibility which requires investigation. By selectively down-regulating oligosaccharide expression inhibitory receptor engagement is prevented but class I MHC expression is retained for viral peptide presentation to T cells. Farlier studies from my laboratory suggesting that oligosaccharide structures are associated with MHC molecules may have relevance here.^The behaviour of autoantigen-specific and viral antigen-specific effector T cells will now be considered. In the case of autoantigen-specitic T cells (Fig. 2), presentation of the autoantigen by uninfected cells results in engagement of the inhibitory' receptor and no response. In contrast, presentation of the autoantigen by virus infected cells allows a response to occur, the presence of a modified inhibitory ligand resulting in signal minus I not being received. This is an intriguing prediction as it implies that autoreactive T eells could selectively eliminate infected target cells. Certainly the frequently reported appearance of autoreactive T cells at early stages during a viral infection^-'' may be a manifestation of this phenomenon. Also, it should be noted that inactivation of NK and T cell cytotoxicity by inhibitory receptors is reversible and transient.'-' The apparent 'usefulness" of autoreactive T cells also may explain the observed persistence of these cells in the periphery.'--^-* Based on the current dogma deletion of autoreactive T cells would seem the safest option rather than their observed persistence in an anergic state requiring the continual presence ofautoantigen.'--^" In order to explain this paradox it has been proposed that persistent autoreactive T cells maintain self tolerance by suppressor meehanisms.^^-^* However, if autoreactive T 282 (a) CR Parish TCR Autoantigen Specific TCell Uninfected Cell [nhibilory Receptor Uninrected Cell Viral Antigen Specific TCell Peptide NO RESPONSE (b) NO RESPONSE (b) TCR TCR Virus Infected Cell Autoanligen Specific TCell No Signal Virus Viral Antigen Specific TCell Infected Cell I Signal Inhibitory Receptor RESPONSE Figure 2 The role of inhibitory receptors in the interaction of autoantigen-specific T cells with (a) uninfected cells or with (b) virus-infected cells carrying the autoantigen. The MHC inhibitory ligand recognized by the inhibitory receptor is depicted, for convenience, as class 1 MHC-^ self peptide associated with an oligosaccharide structure. On virus infected cells the MHC inhibitory ligand is modified by loss of the oligosaccharide and an inappropriate peptide being presented by class I MHC. Engagement of the autoreactive TCR results in signal 1 ( + ), whereas engagement of the inhibitory receptors results in signal minus 1 ( - ) . 'Response' refers to cytotoxicity., lymphokine release and T cell proliferation. Note that autoantigen-specific T cells can respond to virus infected cells. In the figure the TCR is depicted as interacting with MHC + self peptide. The class of the MHC is not specified but. in theory, could be either class 1 or class II MHC molecules. cells can be harnessed to eliminate infected cells in tissues their persistence can be justified on these grounds as well. In the case of viral antigen-specific T cells (Fig. 3), uninfected cells are not recognized due to the lack of MHC + viral peptide on these cells. When the viralantigen-specific T cells encounter virus-infected cells, however, a prompt response occurs due to TCR recognition in the absence of inhibitory receptor engagement. Incidentally, it is presumed throughout this discussion that inhibitory receptor engagement and signal minus I delivery does not occur unless the TCR interacts with antigen (see Fig. 3a). In Figs 2 and 3 the TCR is depicted as interacting with RESPONSE Figure 3 The role of inhibitory receptors in the interaction of viral antigen-specific T cells with (a) uninfected cells or with (b) virus-infected cells. For more of the interaction details see Fig. 2. Note that inhibitory receptors do not deliver a negative signal (signal minus 1) unless TCR is engaged (compare uninfected cells in Figs 2 and 3). MHC + self or viral peptide. The class of the MHC is not specified but, in theory, could be either class I or class II MHC molecules. However, it would be anticipated that the signal minus 1 hypothesis, as proposed, would predominantly control recognition of uninfected or infected tissue cells by class I MHC restricted T cells. Since class II MHC has a restricted cellular distribution and is involved in the presentation of extracellular processed antigen to T cells by 'professional' APC. inhibitory receptors would only control the response of class II MHC restricted T cells under special circumstances. For example, one can envisage situations where tissue cells are induced to express class II MHC molecules following stimulation by various pro-inflammator\' cytokines. In such cases, the inhibitory receptors could control autoreactive T cells which are class II MHC restricted. In fact, it has been shown that the inhibitory receptors can prevent T cell cytolysis triggered by class II MHC recognition.-'^ Implications for alloreactivity and autoreactive B cells The signal minus I hypothesis presented above also may have relevance to two other types of immune responses; Tolerance to tissue-specific self antigens the activation of T cells by allogeneic MHC and the response of B cells to tissue-specific self antigens. It is generally believed that the unusually violent response of T cells to allogeneic MHC molecules is due to the high frequency of alloreactive T cells.^^ Since a high frequency of cytotoxic precursors has been reported against both syngeneic and allogeneic tumour cells'^ this explanation may be overly simplistic. On the other hand, since inhibitory receptors on T cells are MHC-restricted (Table 1), the failure of these receptors to recognize allogeneic MHC may partially explain the violent T cell response observed. Certainly, it appears likely that the rejection of allogeneic tumours may be due to the inability of the tumour cells to inactivate effector T cells by signal minus 1. So far this essay has concentrated on the control of T cell autoreactivity. However, as with T cells, one would expect the continual appearance in the periphery of autoreactive B cells.•'••'^ Of particular importance would be B cells specific for tissue-specific self antigens expressed on cell surfaces. These B cells would not be deleted in the bone marrow during development,^* and yet could produce undesirable blocking antibodies against important cell surface molecules. Do inhibitory class I MHC receptors control these autoreactive B cells? At present there is no evidence for inactivation of B cells by inhibitory receptors. It is clear, however, that the autorosetting receptors detected on a subpopulation of peripheral B cells are strikingly similar to the murine inhibitory receptors, being class I MHC-restricted and anionic carbohydrate specific (Table 2). Thus, one could envisage a situation where autoreactive B cells could interact with tissue-specific cell surface antigens via their surface Ig and subsequently be inactivated by inhibitory receptor engagement. Autoimmunity Here I have proposed that autoreactivity is common but autodestruction is rare due to signal minus I intervention. Since inactivation of effector T cells is mediated by the overriding of signal 1 by signal minus 1 (Figs 1 and 2). it is relatively easy to envisage situations where autoimmunity could occur. For example, T cell clones with high affinity TCR for tissue-specific self antigens could receive sufficient signal 1 to overcome inhibitory receptor inactivation.'^ Another, possibly more likely, scenario is that effector T cells, although co-stimulation independent are still co-stimulation responsive. Thus, the abnormal expression of co-stimulator molecules by target cells may provide sufficient positive signals to override signal minus 1. A clear example of this possibility is the ability of B7 transfected tumour cells to be killed by NK cells, despite the engagement of inhibitory receptors.^** At the genetic level, predisposition to autoimmune disease could occur in individuals where signal minus 1 delivery is impaired. This genetic effect could occur either syslemically or at the organ-specific level. An interesting possibility, for example, is that a predisposition to organspecific autoimmunity could be due to low level expres- 283 sion of MHC class I inhibitory ligands in the target organ prone to autoimmune attack. Evolutionary significance of signal minus I It is highly likely that the active recognition of self (i.e. signal minus 1) in order to prevent autoreactivity is an evolutionarily ancient process. In fact, it has been proposed on several occasions that specific recognition of self is the basis of self/non-self discrimination in all multicellular animals.**^-*^ In these models it has been suggested that a 'self marker', present on all cells of the individual, is recognized by an anti-self receptor on phagocytes. Engagement of the anti-self receptor prevents phagocytosis of self, whereas the lack of the self marker on foreign cells allows them to be ingested. In other words, phagocytes will ingest any cell or particle (presumably recognized by promiscuous receptors) unless self is specifically recognized. A common suggestion has been that histocompatibility antigens represent the self marker.•''^"•*^ Antigenspecific recognition of the MHC by T cells is, therefore, a sophisticated variation which has evolved in higher vertebrates. Interestingly, a number of authors have also suggested that a carbohydrate structure may be the self marker."*^""' When one considers the general properties of the inhibitory' receptors for class I MHC on mammalian lymphocytes (Table 1) these ideas are remarkably prophetic. There is some experimental evidence in invertebrates which supports the concept of active recognition of self. Colonial tunicates only fuse if they share one allele of the fusion gene. Active rejection occurs if all alleles are incompatible.""^ Similarly, gorgonians accept autografts but reject all xenografts and most allografts by a mechanism which appears to involve positive recognition of self.'*'' The capacity to discriminate between self and nonself may also exist in protozoa. Studies over 70 years ago showed that amoebae do not ingest their own or another amoeba's pseudopodia,'''^' despite the considerable phagocytic potential of these organisms. Observations such as this suggest that the emergence of active recognition of self may have even been essential for speciation to occur in early metazoans in some ecosystems. Concluding remarks At this stage it is worth re-examining the immunological paradoxes raised earlier by lymph node metastasizing tumour cells. Can these paradoxes be explained by the signal minus 1 hypothesis? Certainly the failure of the host to reject a tumour despite a vigorous T cell response in lymph nodes can be explained by the action of inhibitory receptors. As depicted in Fig. 1, autoreactive T cell clones only express inhibitory receptors following substantial clonal expansion and the acquisition of the ability to exit the node and enter tissues. Since metastasizing tumour cells are already present in the node and express inhibitory receptor ligands they could rapidly inactivate these T cells immediately they expressed inhibitory receptors (Fig. 2). A very different situation applies when the tumour cells are implanted in an allogeneic recipient. In 284 CR Parish this case tumour rejection ensues as the inhibitory receptors on T ceils are class I MHC-restricted and, therefore, unable to recognize inhibitory ligands on the allogeneic tumour cells. One interesting paradox is the ability of lymph node metastasizing tumour cells to induce a vigorous T cell response despite the tumour cells lacking co-stimulator molecules. This observation implies that bystander costimulation can be provided by professional APC in lymph nodes. I am unaware of in vitro studies demonstrating that naive T cells can respond to tumour cells when bystander co-stimulation is provided. Nevertheless, the juxtaposition of T cells, tumour cells and professional APC in the microenvironment of the lymph node appears to allow T cell activation to proceed. In conclusion, in this essay I have attempted to address one of the most tantalizing mysteries of immunology, the manner by which the immune system is tolerant to tissuespecific self antigens but retains responsiveness to foreign antigens. A fundamental feature of the model is that autoreactivity is common but autodestruction is rare. The recent explosion of information on inhibitory receptors on NK cells and T cells suggests that regulation of autoimmunity at the effector cell level may be a common feature of peripheral T cell tolerance. If this concept is correct it has profound implications for the treatment of autoimmune disease and for the immunotherapy of cancer. Acknowledgements I thank Hilary Warren, Phil Hodgkin and Meredith Bradbury for reading the manuscript and providing honest and constructive comments. References 1 Fowlkes BJ, Ramsdell F. T cell tolerance. Curr. Opin. Immunol. 1993; 5: 873-79. 2 Arnold B, Schbnrich G, Hammerling GJ. Multiple levels of peripheral tolerance. Immunol. Today. 1993; 14: 12-14. 3 Matzinger P. Tolerance, danger, and the extended family. Annu. Rev. Immunol. 1994; 12: 991-1045. 4 Boon T, Cerottini, J-C, Van den Eynde B, van der Bruggen P, Van Pel A. Tumor antigens recognized by T lymphocytes. Annu. Rev. Immunol. 1994; 12: 337-65. 5 Hellstrom, KE, Hellstrom I, Chen L. Can co-stimulated tumor immunity be therapeutically efficacious? Immunol. Re^'. 1995; 145: 123-45. 6 Warren HS, Lafferty KJ. A high frequency of cytotoxic precursor cells for a syngeneic tumour. Seand. J. Immunol. 1979; 10: 349-52. 7 Mackay CR. Homing of naive, memory and effector lymphocytes. Curr. Opin. Immunol. 1993; 5: 423-7. 8 Hart IR. The spread of tumours. In: Franks LM, Teich N (eds) Introduction to the Cellular and Molecular Biology of Cancer. Oxford: Oxford University Press, 1987; 27-39. 9 Tachibana T, Yoshida K. Role of the regional lymph node in cancer metastasis. Cancer Metast. Rev. 1986; 5: 55-66. 10 Yokoyama WM. Right-side-up and up-side-down NK-cell receptors. Curr. Biol. 1995; 5: 982-5. I1 Raulet DH, Held W. Natural killer cell receptors: The offs and ons of NK cell recognition. Cell 1995; 82: 697-700. 12 Lanier LL, Phillips JH, Molecular and cell biology of inhibitory MHC class I receptors on NK cells and T cells. Immunol. Today 1996; 17:86-91. 13 Bottino C. Vitale M, Pende D, Biassoni R, Moretta A. Receptors for HLA class I molecules in human NK cells. Semin. Immunol. 1995; 7: 67-74. 14 Liebson PJ. MHC recognizing receptors: They're not just for T cells anymore. Immunity. 1995; 3: 5-8. 15 Gumperz JE, Pamham P. The enigma of the natural killer cell. Nature 1995; 378: 245-8. 16 Moretta A, Vitale M, Sivori S el al. Human natural killer cell receptors for HLA-class I molecules. Evidence that the Kp43 {CD94) molecule functions as a receptor for HLA-B alleles. / Exp. Med 1994; 180: 545-55. 17 Daniels BF. Nakamura MC, Rosen SD, Yokoyama WM, Seaman WE. Ly-49A, a receptor for H-2D*', has a functional carbohydrate recognition domain. Immunity 1994; 1: 785-92 18 Brennan J, Takei F. Wong S, Mager DL. Carbohydrate recognition by a natural killer cell receptor, Ly 49C. / Biol. Chem. 1995; 270: 9691-94. 19 Gumperz JE, Litwin V, Phillips JH, Lanier LL. Pamham P. The Bw4 public epitope of HLA-B molecules confers reactivity with natural killer cell clones that express NKB1, a putative HLA receptor. / E.xp. Med. 1995; 181: 1133-44. 20 Phillips JH, Gumperz JE, Pamham P, Lanier LL. MHC class I receptors on T lymphocytes inhibit superantigendependent cell-mediated cytotoxicity. Science 1995; 268, 403-5. 21 Sia DY. Parish CR. Anti-self receptors. I. Direct detection of H-2L region-restricted receptors on murine thymocytes. /. E.xp. Med. 1980; 151: 553-65. 22 Sia DY. Parish. CR. Anti-self receptors. II. Demonstration of H-2L region-restricted receptors on subpopulations of peripheral Tand B lymphocytes. / Immunol. 1980; 124: 236671. 23 Primi D, Lewis GK, Triglia R. Goodman JW. Rosette formation between murine lymphocytes and erythrocytes. A new locus in the H'2 region. / Exp. Med. 1979; 149: 1349-59. 24 Charreire J, Camaud C, Bach JF. Studies on mouse autoreactive cells. I. Role of H-2 antigens in mouse autologous rosette formation. Celt Immunol. 1980; 49: 372-8. 25 Sia DY. Parish CR. Anti-self receptors. IV //-^restricted receptors on thymocytes recognize carbohydrate structures on target cells. Immunogenetics 1981; 12: 587-99. 26 Kolb H. A ganglioside receptor on lymphocytes mediates recognition of self. Biochem. Biophys. Res. Comm. 1982; 85; 678-83. 27 Parish CR, Rylatt DB. Snowden JM. Demonstration of lymphocyte surface lectins that recognize sulphated polysaccharides. / Celt Sci. 1984; 67: 145-58. 28 Sia DY, Parish CR. Anti-self receptors. III. Lack of allelic exclusion and thymic epithelium dependence of H-2L region-restricted receptors on lymphocytes. Scand. J. Immunol. 1981; 13; 535-40. 29 Rylatt DB, Sia DY, Mundy JP, Parish CR. Autorosette inhibition factor; isolation and properties of the human plasma protein. Eur. J. Biochem. 1981; 119: 641-6. 30 Shatsky M, Saigo K, Burdach S, Leung LLK, Levitt U . Histidine-rich glycoprotein blocks T cell rosette formation and modulates both T cell activation and immunoregulation. J. Biol. Chem. 1989; 264: 8254-9. 31 Malnati MS, Peruzzi M, Parker KC el al. Peptide specificity Tolerance to tissue-specific self antigens 32 33 34 35 36 37 38 39 40 in the recognition of MHC class I by natural killer cell clones. Science 1995; 267; 1016-18. Parish CR. The paradox of carbohydrate histocompatibility antigens. In; Hoffmann GW, Levy JG, Nepom GT (eds) Paradoxes in Immunology, Boca Raton. CRC Press, 1986; 187-98. Kesson AM, Blanden RV, Mullbacher, A. The secondary in vitro murine cytotoxic T cell response to the flavivirus. West Nile. Immunol. Cell Biol. 1988; 66: 23-32. Alferink J, Schittek B, Schonrich G. Hammerlmg G, Arnold B. Long life span of tolerant T cells and the role of antigen in maintenance of peripheral tolerance. Int. Immunol. 1995; 7: 331-6. Coutinho A, Bandcira A. Tolerize one, tolerize them all: tolerance is self assertion. Immunol. Today 1989; 10: 264-6. McCullagh P. The significance of immune suppression in normal self tolerance. Immunol. Rev. 1996 (in press). MacDonald HR, Cerottini J-C. Ryser J-E et al. Quantitation and cloning of cytolytic T lymphocytes and their precursors. Immunol. Rev. 1980: 51: 93-120. Goodnow CC. B-cell tolerance. Curr. Opin. Immunol. 1992; 4: 703-10. Gcldhof AB. Racs G. Bakkus M. Devos S, Thielcmans K, Debaetsclier P. Expression of B7-1 by highly metastatic mouse T lymphomas induces optimal natural killer cellmediated cytotoxicity. Cancer Res. 1995; 55; 2730-3. Hildcmann WH. Specific immunorecognition by histocompatibility markers: The original polymorphic system of immunoreactivity characteristic of all multicellular animals. Immunogenetics \911\ 5: 193-202. 285 41 Kolb H. On the phylogenetic origm of the immune system-a hypothesis. Dev. Comp. Immunol. 1977; 1; 193206. 42 Langman RE. Cell-mediated immunity and the major histocompatibility complex. Rev. Physiol. Biochem. Pharmacol. 1978; 81: 1-37. 43 Cohn M. The T-cell receptor mediating restrictive recognition of antigen. Cell 1983; 33: 657-69. 44 Coombe DR, Ey PL, Jenkin CR. Self/non-self recognition in invertebrates. Q. Rev. Biol. 1984; 59: 231-55. 45 Coombe DR, Parish CR. Sulfated polysaccharide-mediated sponge cell aggregation: The clue to invertebrate self/nonselfrecognition? In: Grosberg RK, Hedgccock D. Nelson K. (eds) Invertebrate Historecognition. New York. Plenum Publishing Corp., 1988; 31-54. 46 Parish CR. A simple model for self non-self discrimination in invertebrates. Nature 1911; 267; 711-12. 47 Rothenberg BE. The self recognition concept: an active function for the molecules of the major histocompatibility complex based on the complementary interaction of protein and carbohydrate. Dev. Comp. Immunol. 1978; 2: 23-38. 48 Scofield VL, Schlumpberger JM. Weissman IL. Colony specificity in the colonial tunicate Botryltus and the origins of vertebrate immunity. .4m. Zooi 1982; 22; 783-94. 49 Theodor JL. Histo-incompatibility in a natural population of gorgonians. Zooi J. Linn. Soc. 1976; 58: 173-6. 50 Reynolds BD. Interactions of protoplasmic masses in relation to the study of heredity and environment in .Arcella polypora. Biol. Bull. 1924; 46; 106-40.