218426a0Nature2185140196805044264300028-0836196810.1038/218426a0ukNatureNatureNATUREnatureNature is a weekly international journal publishing the finest peer-reviewed research in all fields of science and technology on the basis of its originality, importance, interdisciplinary interest, timeliness, accessibility, elegance and surprising conclusions. Nature also provides rapid, authoritative, insightful and arresting news and interpretation of topical and coming trends affecting science, scientists and the wider public./nature/journal/v218/n5140issueJournal homeArchiveCurrent issueAdvance online publicationPrivacy policySubscribeNature Publishing GroupCurrent issue218426a0Evolution of the Immune Process in Vertebrates
AU  - BURNET, F. M.University of Melbourne, Victoria, AustraliaThis article is a slightly up-dated version of the presidential address to the Australian Society of Immunologists given by Sir Macfarlane Burnet in December 1967. He suggests that the mammalian immune system could have evolved from the haemocyte or primitive mobile cell of the invertebrate.THERE has been rapid progress in the understanding of the nature of antibody production in recent years, but there has been much less interest in the cellular side of immunity and very little attempt to consider the defence processes of invertebrates in relation to the standard immune system of mammals.
In discussing possible evolutionary processes from the cellular angle, I shall make two basic assumptions which nowadays are probably acceptable to most immunologists. The first is that adaptive immunity is characteristic of the vertebrates only, the second is to accept the selective approach to the origin of antibody pattern as outlined by Jerne1 in his summing up of the discussions at Cold Spring Harbor last June. Implicit in this approach is that the cells (the immunocytes) of an antibody-producing clone carry a receptor the reactivity of which has approximately the same specificity as the combining site of the corresponding antibody molecules. In other respects also, I shall follow the general concepts of the clonal selection approach2?4.
The first point to be made is that, like vertebrates, all invertebrates, large and small, live in a world in which potentially pathogenic bacteria and other micro-organisms are abundant and menacing. Invertebrates are, however, not normally subject to overt bacterial infection and we must assume that all possess appropriate defence mechanisms which are capable of recognizing, in at least a crude fashion, when foreign material enters the body and either eliminating it or rendering it harmless. If we are thinking only of defence against micro-organisms the invertebrate mechanism seems to be efficient and there is no obvious reason why the vertebrate system of adaptive immunity should have evolved.
One cannot, however, just dismiss the invertebrate system of defence by saying that there is no evidence of antibody production or any other adaptive character. Evolution demands some raw material from which to build a new functional system. I should like to explore the evidence in regard to recognition of foreignness?of not-self from self?in invertebrates in so far as it may be relevant to the evolution of the vertebrate immune system.
Some ability to recognize foreignness must go right back to the first utilization by protozoa of other organisms for food. Thirty years ago I wrote about the need to concentrate on "the most obvious aspect of all?that the engulfed micro-organism is not the amoeba itself. The fact that the one is digested, the other not, demands that in some way or other living substance of the amoeba can distinguish between the chemical structure characteristic of 'self and any sufficiently different chemical structure which is recognized as 'not-self? "5.
Since then, much work has been done on the defence mechanisms of invertebrates particularly in regard to their responses to metazoan parasites6. If one inserts into the body cavity of an insect a foreign object such as a larval parasite, a granule of celloidin or a little carbon black, there is a standard response. The nucleated cells, haemocytes, become attached to the foreign surface, flatten against it and build up a capsule of several layers of cells around the object. Subsequently the capsule becomes fibrous and the inner nuclei disappear. In the intermediate phase many haemocytes which had accumulated around the periphery disappear, presumably by release to the circulating fluid. From the indirect evidence of the ability of insect physiologists to graft organs almost at will and a very little specific evidence on the point7, one can assume that a piece of the insect's own living tissue would not provoke such a reaction. One of the most interesting features of the action for an immuno-logist is that there are many instances where a "usual" parasite provokes no reaction, while parasites of the same general quality but which normally parasitize a different type of host?"unusual" parasites?are actively encapsulated and killed. There is here a reasonable implication that it is an evolutionary possibility for a parasite to discard the chemical structures which allow its recognition by the host as a "foreign body". This has the further implication that the differences concerned are relatively minor ones because their removal leaves the parasite functionally intact.
At the risk of moving too far and too fast into a void of speculation, I should like to attempt to fit this picture of recognition of foreignness in invertebrates into relationship with the modern genetic approach to antibody production in vertebrates. It has, I believe, the virtue that it suggests definite experimental approaches.
It is axiomatic that a haemocyte must not adhere to or react damagingly with any body cell with which it comes into normal contact. It seems a likely corollary to this that any mutant haemocyte so doing would be automatically destroyed, and that a damaged body cell could equally be recognized as foreign and sequestered. There is more than one way in which this necessary quality could be associated with recognition of foreignness, but, having due regard to what is the accepted situation in mammals, the following may be suggested.
There are globulin-like proteins associated with the haemocyte surface and probably leaking into the fluid of the body cavity. These can act as receptors and can be adsorbed to a foreign configuration, the result of stimulation being broadly damaging to the haemocyte and leading to capsulation. It would presumably require a range of patterns of this globulin to allow ability to respond to a wide range of foreign material.
If diversity of pattern in cell globulins arose by a stochastic process it would be necessary to have a means of eliminating self-reactive patterns. This could be done most effectively if diversification of globulin pattern was provided at a cellular level, that is, if each haemocyte produced globulin of one pattern only, presumably with phenotypic restriction extending to any descendant clone. The alternative that each cell contains a variety of globulins all available at the surface is less attractive for the same sort of reasons as hold for the corresponding mammalian hypothesis.
The absence of adaptive immunity in insects and other invertebrates would then depend on the failure of specific stimulation of a reactive haemocyte to provoke its multiplication. By hypothesis the response of the wandering cell to specific contact is limited to damage and conversion to an encapsulating fibroblast-like cell. The evolutionary step towards an adaptive response was the acquisition of the potentiality of a specifically patterned mobile cell to respond to at least a proportion of specific contacts by proliferation to produce a descendant clone of the same specificity.
There are some relevant experimental findings and a number of new experimental approaches which could be considered in relation to this hypothesis. Cushing et al.8 reviewed work on the haemagglutinins of crustacean body fluids, which show considerable specificity in relation to human ABO antigens. TJhlenbruck et al.9 found that a globulin-like protein can be obtained from a snail. Tripp7 found that implantation of living homologous tissue into the snail (Australorbis) produced no response, but formalized homologous tissue and living tissue from another genus of snails were quickly encapsulated.
It might be suggested that insect physiologists should carry out experiments to see whether haemocytes can be divided into specific sub-populations, for example, by their power to adsorb specific particles such as bacteria or red cells. If this is the case it should not be impossible to devise ways of inducing specific tolerance to a given type of foreign material. It would also be within the bounds of possibility that by using subcellular antigen in a very small dose, an abnormal reaction leading to proliferation of a specific clone might occur. A study of immunization of caterpillars with bacterial flagellin and its monomer and test of haemocytes with the corresponding motile bacteria might provide interesting results.
At the biochemical level, I feel that an active search in haemocytes of large invertebrates for small proteins of 100?110 amino-acids (12,000 [plusmn] molecular weight) with adsorptive and polymerizing properties might result in the isolation of the primitive building block for the immunoglobulins of vertebrates.
My second point is that, as Hildeman and Owen10 showed, homograft immunity as exemplified in scale transplantation in goldfish is exquisitely specific in coldblooded vertebrates. Antibody production, on the other hand, is characteristically poor in fish and amphibians, although more evident when their temperature is raised. There is a strong suggestion here that specific cellular responses of the type characteristic of homograft immunity and delayed hypersensitivity came earlier than antibody production during evolution.
There is wide acceptance of the view that delayed hypersensitivity and homograft reactions are intimately related and that in neither does antibody play an essential part. A helpful hypothesis in regard to this relationship derives from Thomas's suggestion11 that the adaptive immune system arose out of the need for dealing with any foreign antigens which arose within the body by somatic mutation of some equivalent chromosomal process.
Cancer, malignant disease, is something which appears to be confined to vertebrates. Given the characteristics of malignant cells as we know them, their existence would be a special threat to the young if there were no means of preventing contagious transfer of cancer cells from one individual to another. Cancer as a disease of the post-reproductive period has no intrinsic significance for evolution, but if it were transferable from old to young it would be disastrous to any species in which this occurred. In vertebrates we all know that it does not occur and it is worth analysing why it does not. It depends, of course, on the diversity of histocompatibility antigens and on the capacity of immunocytes in the body to recognize and dispose of cells carrying alien histocompatibility antigens.
I should like to explore the possibility that these two new characteristic ways of generating diversity are closely related evolutionary; that the adaptive system of immunity arose from the need of countering the onset of malignancy in vertebrates.
In the last analysis I expect that we shall find all three characteristics of vertebrates?proneness to malignant change, diversity of histocompatibility antigens and the adaptive immune system?to be related to the necessity for more flexibility of the genome both at germ cell and somatic cell level. What concerns us is how, granted a potential mutability of the somatic genome, evolution could have moulded an adaptive immune system out of a non-adaptive system of defence in invertebrates or the earliest vertebrates.
As I have discussed already, there is definite evidence that there is at least a limited capacity to recognize foreignness in the haemocytes of invertebrates. It is also known that there are proteins in invertebrate body fluids which have pseudo-immunological capacities, for example, they can agglutinate mammalian red cells. There is clearly raw material from which an antibody-based system could be moulded by strong enough evolutionary forces. What recognition of foreignness there is in invertebrates is a function of wandering phagocytic cells and we can think of them as in some sense ancestral to the immuno-cyte, polymorphonuclear and macrophage of the vertebrates (Fig. 1).
Among the globulins present in the surface membrane of wandering cells there may well have been proto-antibodies of the type I have already spoken of. Two developments are required to convert this to an adaptive immune system. The first is an increased flexibility of the part of the somatic genome concerned with coding for the pattern of these protoglobulins. This conceivably arose as a phase of changing patterns or mechanisms of differentiation. Its function was (a) to provide a means of greater diversification of pattern?of the sort discussed by Smithies and others at the Cold Spring Harbor Symposium in June 1967?and (b) to associate this diversification with an increasingly absolute phenotypic restriction. In this way there would be an increasing number of foreign patterns that could be recognized and a concentration on a few cells of high ability to deal with any one specific pattern of foreignness. The second requirement is that contact of foreign pattern (antigenic determinant) with recognition globulin (combining site) should in appropriate conditions allow proliferation of the haemocyte concerned; with retention of its specific character throughout the descendant clone. This at the cellular level is the essential feature which makes an immune system adaptive.
If Thomas's hypothesis is correct that the primary function of adaptive immunity was to recognize and remove body cells which by somatic mutation or otherwise had become recognizably foreign, then the earliest manifestations of this function would be expected to be wholly at the cellular level. On the effector side it would take the form of recognition, by specific union, of foreignness in the surface of another cell with destruction of the "foreign" cell, and probably of the effector cell, by mutual liberation of damaging agents.
Fig. 1. Diagram suggesting the evolution of the mammalian immune system from the haemocyte or primitive mobile cell of the invertebrate. MPH, Macrophage; MC, monocyte; DPC, dendritic phagocytic cell of lymphoid follicles; PMN, polymorphonuclear leucocyte; DH, immunocytes mediating delayed hypersensitivity and homograft rejection (thymus-dependent); A, M, G, immunocytes differentiated to produce the immunoglobulins shown (GALT-dependent).
The only original suggestion I want to make?if it is original?is that there are many persisting indications that mammalian immune responses arose from cell to cell associations. This is particularly the case in regard to delayed hypersensitivity. It is in accord with the recent reviews of both Uhr12 and Humphrey13 to assume that the challenge side of delayed hypersensitivity represents the liberation, by sensitized cells under the stimulus of antigen contact, of agents which can stimulate and damage associated unsensitized cells, even when the latter form a very large percentage of the cells in the system.
The problem of the induction of hypersensitivity is less widely studied and there is no consensus of opinion in the matter. I have recently completed a discussion (unpublished) of the thesis that to provoke delayed hypersensitivity, homograft immunity and related reactions, the progenitor immunocyte (antigen reactive cell) must react with the appropriate antigenic determinant carried in the cell surface of a mobile cell. Once this point of view is adopted we are almost compelled to adopt the paracortical region of Turk and Heather14, the thymus-dependent areas of Parrott et al.15, as the sites in lymph nodes and spleen where induction occurs. Here both progenitor immunocytes from the thymus and antigencarrying cells from the periphery are in a position to make mutually stimulating contact. The concentration of pyroninophil cells in this region at the stage of development of skin sensitivity to a chemical like oxosalone has been stressed by Oort and Turk16. It is reasonable to assume that the lipoprotein surface of a cell is a highly dynamic system and that there may be frequent transfer of potentially antigenic material from one cell to another.
This concept derives very largely from considerations of Lawrence's transfer factor experiments17'18. The only interpretation I can see?and it covers only about 90 per cent of the facts?is that in man an antigen of the type capable of producing delayed hypersensitivity characteristically concentrates on the surface of mobile cells, in such a form that it is both highly sensitizing and readily transferable to other cells of the same general quality. All the quality of the phenomenon is of an active rather than a passive character. The transferred sensitivity, for example, may last a year.
There are many other aspects of delayed hypersensitivity and related phenomena which seem to become clearer on this hypothesis of cellular carriage of the effective antigen. These include the low molecular weight of some typical sensitizing agents, tuberculin, the effective peptide in experimental allergic encephalitis and the various skin sensitizing chemicals; the difficulty of inducing delayed hypersensitivity by the intravenous route and the many resemblances of delayed hypersensitivity and homograft rejection reactions.
On such a view, delayed hypersensitivity would be regarded as a phenomenon essentially identical with the primitive surveillance function of which homograft responses are the standard laboratory model. It is probably also implicit in the concept that the immunocytes mediating these responses can be regarded as a distinct immunoglobulin type. The possibility that the reactive immunocytes are IgA producers in which no antibody is actually liberated into the environment is suggested by Rothman and Liden's19 work, but this has not yet been confirmed or generalized.
Active antibody production has all the marks of a later addition to the more primitive cell-based system concerned with delayed hypersensitivity and homograft rejection. Despite the assumption implicit in most of the teaching of immunology that antibody and antibody production were the central features of defence against infection, the evolutionarily significant features of antibody are rather marginal ones. Its most important functions may be to act as opsonin especially when transferred to the foetus or newborn, to provide effective protection on mucous surfaces and perhaps to act as a feedback preventing stimulation of new progenitor cells.
Such functions can be regarded as extending the primitive capacity of cell to cell recognition to the molecular level with correspondingly increased flexibility and adaptability. The details of this evolutionary change may never be elucidated and they will certainly remain obscure until the relationships of the cell lines responsible for delayed hypersensitivity and those responsible for the production of IgM, IgG and IgA antibody are clarified.
What may be a potentially productive approach is to follow the implications of the rapidly crystallizing view that only certain immune functions are "thymus-dependent". Homograft immunity is the clearest example. There is no unanimity of the exact role of the thymus, but most immunologists would probably accept as concordant with the facts the view that the thymus cortex is a site at which stem cells from the bone marrow lodge and are differentiated under local hormonal influences to immunocytes. A proportion of these pass as small lymphocytes to the peripheral lymphoid tissues. For some years I20 have favoured the ancillary hypothesis that a newly differentiated immunocyte meeting in the thymic environment antigen with which it can react is destroyed. This provides a simple and direct way of interpreting natural and acquired immunological tolerance.
The relation of the thymus to antibody production has been much less clear. Mice neonatally thymectomized can still produce a variety of antibodies and may show little reduction in immunoglobulins. What seems to be needed to bring antibody production into line is the postulation of a new site of primary differentiation. From the behaviour of the bursa of Fabricius in birds it seems likely that mammals possess an analogous differentiating system in the gut-associated lymphoid tissue which has a similar relationship to antibody-producing cell lines that the thymus has to those concerned with delayed hypersensitivity and homograft immunity.
The features necessary for the progressive incorporation of antibody as a functioning component of the immune system would be (1) the evolution of immunocytes capable of actively synthesizing and liberating IgM and IgG and if the earlier suggestion of the relationship of IgA to delayed hypersensitivity is found to be valid of IgA liberators, (2) the development of lymphoid tissue associated with the alimentary tract as a tissue capable of differentiating stem cells to immunocytes, and (3) the specialization of the dendritic phagocytic cells (DPC) of the lymph follicles as the standard presenters of antigen to immunocytes which are potential producers of antibody21.
There is more than one way of looking at the evolution of IgM and IgG antibody-producing cells. If antibody is needed, there must not only be cells differentiated so as to be capable of producing it but also a control mechanism possibly quite elaborate to see that it is produced in appropriate amount and on the appropriate occasions.
In seeking a more detailed interpretation I have felt that very great weight must be placed on two now well established dichotomies in the immune system. The first is the differentiation, gradually becoming clearer over the years, of the thymus-dependent and the bursa-dependent functions in the domestic fowl22?25. This has led largely from the study of immuno-pathological conditions such as congenital agammaglobulinaemia and some even rarer anomalies to the conclusion formulated largely by Good and his associates26'27, that in the mammal there are, as in the bird, both thymus-dependent immunological functions and another set dependent on gut-associated lymphoid tissue. In general the thymus-dependent functions comprise delayed hypersensitivity and homograft rejection, while general antibody production is dependent on gut associated lymphoid tissue (GALT).
The second set of experiments and interpretations is due primarily to the recent work of Miller's group28'29 but owes much to many previous investigators concerned with the function of thymus and bone marrow cells in murine immunity. Miller and Mitchell have shown that in experiments designed, by the use of irradiated animals and biologically recognizable cells from appropriate donors, to trace the origin of functionally active cells: A, that antigen-reactive cells which in the presence of antigen give rise to antibody-producing (plaque forming) cells are not of thymic origin; in part or wholly they are of recent bone marrow origin; and JE?, that cells of thymic origin which have reacted with antigen can multiply specifically30, are not antibody producers31, but are necessary if cells of bone marrow origin are to become specific antibody producers.
Fig. 2. Diagram of the suggested interrelationship of the thymus-dependent and GALT-dependent systems. DH, Delayed hypersensitivity and cells mediating this; AB, antibody or immunocytes of antibody-producing line; AG, antigen; AG[ast], antigen not requiring the co-operation of thymus-dependent cells.
There is, I believe, a simple way in which these two groups of findings can be brought into the picture of the evolution of the mammalian immune system.
Simple interpretations are always less subtle than biological realities but they can sometimes be helpful. The functions of a DH-type immunocyte making contact with antigen of type appropriate to its receptors is to liberate pharmacologically active material which can stimulate, damage or destroy adjacent cells. This is basic to the primary surveillance function that we have ascribed to such cells. There is evidence from various sources, for example, Dutton and Harris32, that stimulation by antigen can induce immunological activation of adjacent cells. The hypothesis I wish to suggest is simply that in most or all immunological situations where antigenic stimulation is taking place we have in the relevant lymphoid tissue: A, progenitor immunocytes (lymphocytes) of thymic origin (PI-T), and progenitor immunocytes of GALT origin (PI-G); and B, antigen (i) as free molecules or particles, (ii) on, the surface of monocytic cells, and (iii) held perhaps in modified form on the surface of dendritic phagocytic cells.
In such circumstances one can envisage the first step as stimulation by antigen of the DH-type cells (PI-T) to develop to pyroninophil blasts and to liberate pharmacological agents which can influence and activate adjacent cells. One of these effects is assumed to be on any PI-G cells of potential reactivity with the antigen concerned. Only when they are so stimulated do they become susceptible to react with antigen (especially with antigen on DPC) by taking on plasmablast morphology and initiating clones of antibody-producing cells.
This concept avoids the very unattractive necessity of postulating any transfer of genetic information between somatic cells, is in accord with the experimental facts as far as I am aware of them and should in principle be capable of experimental test.Jerne, , N. K., Cold Spring Harbor Symp. Quant. Biol., 32 (in the press).Burnet, , F. M., Austral. J. Sci., 20, 67 (1957).Burnet, , F. M., The Clonal Selection Theory of Acquired Immunity (Cambridge and Vanderbilt University Presses, 1959).Burnet, , F. M., Nature, 203, 451 (1964).ArticlePubMedISIBurnet, , F. M., Biological Aspects of Infectious Diseases, first ed. (Cambridge University Press, 1940).Salt, , G., Parasitology, 53, 527 (1963).PubMedISIChemPortTripp, , M. R., J. Parasitol., 47, 745 (1961).PubMedISIChemPortCushing, , J. E., Calaprice, , N. L., and Trump, , G., Biol. Bull., 125, 69 (1963).ISIUhlenbruck, , G., Kim, , Z., and Prokop, , O., Nature, 213, 76 (1967).ArticleISIChemPortHildeman, , W. H., and Owen, , R. D., Transplant. Bull., 3, 132 (1956).Thomas, , L., in Cellular and Humoral Aspects of the Hypersensitive State (edit. by Lawrence, H. S.), 529 (Hoeber, New York, 1959).Uhr, , J. W., Physiol. Rev., 46, 359 (1966).PubMedISIChemPortHumphrey, , J. H., Brit. Med. Bull., 23, 93 (1967).PubMedISIChemPortTurk, , J. L., and Heather, , C. J., Intern. Arch. Allergy, 27, 199 (1965).ISIChemPortParrott, , D. M. V., de Sousa, , M. A. B., and East, , J., J. Exp. Med., 123, 191 (1966).ArticlePubMedISIChemPortOort, , J., and Turk, , J. L., Brit. J. Exp. Pathol., 46, 147 (1965).ISIChemPortLawrence, , H. S., J. Clin. Invest., 34, 219 (1955).PubMedISIChemPortLawrence, , H. S., in Cellular and Humoral Aspects of the Hypersensitive State (edit. by Lawrence, H. S.), 279 (Hoeber, New York, 1959).Rothman, , U., and Liden, , S., Nature, 208, 389 (1965).ArticlePubMedISIChemPortBurnet, , M., Brit. Med. J., ii, 807 (1962).Ada, , G. L., Nossal, , G. J. V., and Pyre, , J., Austral. J. Exp. Biol., 42, 295 (1964).PubMedISIChemPortGlick, , B., Chang, , T. S., and Jaap, , R. G., Poultry Sci., 35, 224 (1956).ISIGood, , R. A., and Papermaster, , B. W., Adv. Immunol., 4, 1 (1964).ISIWarner, , N. L., and Szenberg, , A., Ann. Rev. Microbiol., 18, 253 (1964).ISIChemPortMoore, , M. A. S., and Owen, , J. J. T., Develop. Biol., 14, 40 (1966).ArticlePubMedISIChemPortCooper, , M. D., Peterson, , R. D. A., South, , M. A., and Good, , R. A., J. Exp. Med., 123, 75 (1966).ArticlePubMedISIChemPortGood, , R. A., Cooper, , M. D., Peterson, , R. D. A., Hoyer, , J. R., and Gabrielsen, , A. E., in Germinal Centres in Immune Responses (edit. by Cottier, H.), 386 (Springer, Berlin, 1967).Miller, , J. F. A. P., and Mitchell, , G. F., Nature, 216, 659 (1967).ArticlePubMedISIChemPortMitchell, , G. F., and Miller, , J. F. A. P., Proc. US Nat. Acad. Sci. (in the press).Cross, , A. M., Leuchars, , E., and Miller, , J. F. A. P., J. Exp. Med., 119, 837 (1964).ArticlePubMedISIChemPortDavies, , A. J. S., Leuchars, , E., Wallis, , V., Marchant, , R., and Elliott, , E. V., Transplantation, 5, 222 (1967).Dutton, , R. W., and Harris, , G., Fed. Proc., 22, 266 (1963).ISI