Imperial Cancer Research Fund Tumour Immunology Unit, Department of Zoology, University College London, Gower Street, London, WC IE 6BT, England

Interest in helper and suppressor epitopes has been growing rapidly. It is now generally accepted that most antigens present structures (epitopes) to the immune system which are recognized preferentially by dif ferent sets of lymphocytes, as shown in Fig. I. These sets belong to the effector compartment containing B cells, Tc cells (cytotoxic cells), and delayed-type hypersensitivity-mediating T cells (not shown), or to the regulatory compartment containing Th and Ts cells (helper and suppressors ). Structures preferen tially recognized by Th cells are termed helper epitopes, and so on, and the balance of the two types of regulatory epitope is known to be an important factor in determining the outcome of at least some immune responses. The presence of even a single suppressor epitope may be enough to prevent a response from occurring. The fullest analysis of an antigen along these lines has been carried out on lysozyme [I] and ß-galactosidase [ 18, 9], and other antigens which have been examined in this way include ferredoxin [19] and tumour antigens [5, 7]. Note however that cleavage of serum albumin does not yield fragments with distinct helper and suppressor epitopes [2, 4]. Understanding the nature of helper and suppressor epitopes is a matter of importance to leukaemia research because it could link tumour idiotype to membership

Fig. I. How a vaccine looks to T cells. Th- Ts repertoire differences arc potentially valuable

of a lymphocyte set, and also because of its relevance to any future tumour immunotherapy. It is important to immunological diseases at large, because it is likely to help explain why particular types of antigen tend to generate a harmful response. And at the present time it is most obviously important to development of the new generation of vaccines. This is an area of great excitement because these vaccines, based on bioengineering, promise to control and eventually eradicate the major tropical diseases. Of course one needs to exercise caution in evaluating this promise and as yet engineered vaccines are only just entering veterinary trials, but there is no doubt that this hope has given new heart to vaccine research. What is now being done, world-wide, is to take a parasite such as the plasmodium of malaria, clone cDNA into an expression vector, and screen for antibody-defined antigens [6, 7]. A complementary approach is to screen for antibody-reactive synthetic peptides [ II ]. Both of these strategies concentrate initially on the ability of vaccine molecules to interact with the effector compartment, simply because antibodies and to a lesser extent cytotoxic T cell clones are the only practical screening agents. But in the long run this is too limited an approach, particularly for the major tropical diseases all of which are long-term and characteristically display an ineffective host response. Surely the main hope is for a vaccine which can perturb this balance between parasite and host, through manipulation of the regulatory compartment. The nature of suppressor and helper epi topes is still poorly understood. I wish here to offer a contribution towards a general theory of what makes them different from one another. Any discussion of the problem should start with a distinction between differences based on antigen processing and differences based on the receptor repertoire of lymphocytes. As regards the former, interpretations have tended to diverge, with some authors emphasizing the importance of relatively crude factors such as the gross anatomical localization of antigen, while others emphasize the importance of the interaction of fragments of antigen with particular cell receptors. Thus chemical modification can greatly effect anatomicallocalization [3], and the route of immunization or form in which a determinent is administered can greatly influence which sets of lymphocyte respond [14]. On the other hand it has been suggested that peptide fragments of antigens associate selectively with particular major histocompatibility complex components, and thus determine which regulatory cells respond to "aggretope selection" [ 10]. Or suppressor cells may resemble B cells rather than helper cells in their interaction with antigen, and consequently tend to respond to conformational rather than sequence determinants [ 12]. These are fascinating questions, but we have little hard information with which to answer them. In contrast there is something more definite to say about repertoire differences between helper and suppressor cells, even if at present this is of a rather general character. Thanks to recent advances in our understanding of the interactions of four

Fig.2. Only medium concentration self-macromolecules generate Th- Ts repertoire differences (see text for details)

proteins with the immune system as they occur naturally, we can begin to define what is probably the main factor responsible for moulding this repertoire difference. The proteins in question are C5, a complement component, F, a liver and serum protein of unknown function, AFP , a-fetoprotein, and Tg, thyroglobulin. The argument is summarized in Fig. 2, which requires some explanation. Each stippled area represents the repertoire of antigen receptors for a set of Iym phocytes ( each dot representing, as it were, a single clone of cells). The circles represent holes punched out of the repertoire by tolerance of a single self-protein; it is assumed that self-tolerance results from purging the repertoire of clones reactive with self-proteins. The top row describes what happens to B cells: their repertoire is purged only by proteins which occur in the body at high concentrations, over 10-5 M, such as serum albumin or the constant part of immunoglobulin. The next row deals with helper T cells: their repertoire is purged by proteins occurring down to lower levels of concentration, 10-10 M, but no doubt there are still other molecules which occur at concentrations too low to be noticed at all by the immune system, and for which no purging occurs. Thyroglobulin is an example at the

Table I. Documentation ofreactivity of the main lymphocyte sets to four medium concentration self-proteins

borderline between the medium and low ranges of concentration: it purges the helper cell repertoire to a significant extent, but incompletely. At the bottom come suppressor T cells: their repertoire is no more than partially purgcd by proteins occuring at medium concentrations, as exemplified by the proteins F and C5. The important point is that purging of helper and suppressor cells occurs down to different levels, and that this difference defines a medium concentration range at which their repertoires must differ. No other factor can be identified which is known to generate a difference between their repertoires. This does not mean that other factors do not operate, but simply that at present we do not know what they are. For instance at the time of writing there is a suspicion, but at present no more than that, that helper and suppressor T cells draw from different sets of V genes. The evidence for the medium concentration range that is crucial to this argument is cited in Table 1 for the four proteins C5, F, AFP and Tg, and is discussed in greater detail elsewhere [13]. What are the practical consequences? It is ironic that this, the only definite piece of information which we have about the helper-suppressor epitope difference should produce so little in the way of practical advice about how to design a particular epitope. Even if we knew the full threedimensional structure of the proteins listed in Table 1 we would only be a little nearer this goal. From this point of view research on differential antigen processing perhaps has more to offer, even if so far its achievements have been small. For the time being immunologists will be kept busy cloning genes and synthesizing peptides of potential value in vaccines. More and more of these new products will enter immunization trials without much rhyme or reason, and as they do so we shall no doubt acquire empirical information about which kinds of structure are good immunogens as distinct from ones which merely react well with antibodies. It will be important to have some kind of theoretical framework into which this information will fit. I believe that medium concentration self-proteins as defined here will be an important part of that framework.

References

I. Adorini L, Harvey MA, Miller A, Sercarz EE ( 1979) Fine specificty of regulatory T cells. II. Suppressor and helper T cells are induced by different regions of hen egg-white lysozyme in a genetically non-responder mouse strain. J Exp Med 150:293-306
2. Benjamin DC, Daigle CA, Riley RL (1983) The antigenic structure of bovine serum albumin. T-cell. B-cell, and la determinants. In. Celada F. Schumacher YE, Sercarz EE (eds.) Protein confrontation as an immunological signal. Plenum, New York, pp 261 -280
3. Dailey MO, Hunter RL (1974) The role of lipid in the induction of hapten-specific delayed hypersensitivity and contact sensitivity. J ImmunoII12.1526-1534
4. Ferguson TA, Peters TJr, Reed R, Pesce AJ, Michael JG ( 1983) Immunoregulatory properties of antigenic fragments from bovine serum albumin. Cell ImmunoI78.1-12
5. Harris DE, Cairns L, Rosen FS, Bore! Y (1982) A natural model of immunologic tolerance. Tolerance to murine C5 is me diated by T cells, and antigen is required to maintain unresponsiveness. l Exp Med 156:567-584
6. Kemp Dl, Koppel RL, Cowman AF, Saim RB, Brown Gv, Anders RF (1983) Expression of Plasmodium falciparum blood stage antigens in Escherichia coli. detection with antibodies from immune humans. Proc Natl Acad Sci USA 80.3787 -3791
7. Klein BY, Sharon R, Tarcic N, Naor 0 (1982) Induction orantitumour reactive cells or suppressor cells by different molecular species isolated from the same nonimmunogenic tumour. Immunobiology 163.7-21
8. Kong YM, Okayasu I, Giraldo AA, Beisel KW, Sundiale RS, Rose NR, David D, Audibert F, Chedid L ( 1982) Tolerance to thyroglobulin by activating suppressor mechanisms. Ann NY Acad Sci 392.191-209
9. Krzych U, Fowler A V, Miller A, Sercarz EE ( 1982) Repertoires of T cells directed against a large protein antigen, ß-galactosidase. I Helper cells have a more restricted repertoire than proliferative cells. l Immunol 128: 1529 -1534
10. Krzych U, fowler A, Sercarz EE (1983) Antigen structures used by regulatory T cells in the interaction among T suppressor, T helper, and B cells. In. Celada F, Schumacher YE, Sercarz FEE (eds) Protein confrontation an immunological signal. Plenum, New York, p 395-408
II. Lerner RA ( 1983) Synthetic vaccines. Sci Am 248.66--74
12. Lukic ML, Mitchison NA, Self and allo specific suppressor T cells evoked by intravenous injection of F protein. Eur J ImmunoI 14.8:766-768 13. Mitchison NA, Lukic ML, Nardi N, Proteins at the margin between being tolerated and being noticed. In. Proc Ninth Int Covocation on Immunology. Karger, Basel 6
14. Plak W, Rozycka D, Askenase PW, Gershon RK ( 1980) Role of antigen presenting cells in the development and persistence of contact hypersensitivity. l Exp Med 151 :362-375
15. Ruoslahti E, Wigzell H ( 1975) Breakage of tolerance to alpha-foetoprotein in monkeys. Nature 255.716-117
16. Ruoslahti f, Pihko H, Becker M, Makela M ( 1975) Rabbit alpha-f)etoprotein. normal levels and breakage of tolerancc with haptenated homologous alpha-fetoprotcin. Eur l ImmunoI5.7-10
17. Romball CG, Weigle WO (1984) Tccll competence in heterologous and homologous thyroglobulins during thc induction of experimental autoimmune thyroiditis fur l ImmunoI14.10:887-893
18. Turkin 0. Sercarz EE ( 1977) Key antigenic determinants in regulation of the immune response. Proc Nat Acad Sci USA 74.3934 -3987
19. Weaver M, Singhai R, Sikora L Levy lC (1983) Identification of an idiotypic marker of a major regulatory T cell of the immune response in BlO.BR mice to ferredoxin. The relationship of idiotypic regulation too conventional hapten-carrier effects. l Exp Medi 157:285-300
20. Wedderburn L, Lukic ML, Edwards S, Kahan MC, Nardi N, Mitchison NA (1984) Single-step immunosorbent preparation of F protein from mouse liver with conservation of the alloantigenic site, and determination of concentration in liver and serum Mol Immunol21:10:979-984