1. Genes and Antibodies (Lederberg 1959)

2. Two Signal Hypothesis (1968)

3. Positive Selection (1975)

4. Review of the above (2012)

Photo of Joshua Lederberg

Genes and Antibodies

Do antigens bear instructions for antibody specificity, or do they select cell lines that arise by mutation?

By Joshua LederbergScience (1959) 129, 1649-1653

(With copyright permission from the author and the publishers of Science, the AAAS)

[Colouring, bold font and comments in square brackets by DRF. Italics are as in original, unless otherwise stated.]

Joshua Lederberg (1925-2008)  

1. Antibody Globulin
2. Gene for Globulin Synthesis   
3. Genic Diversity of Precursor Cells
4. Hypermutability
5. Spontaneous Production of Antibody
6. Induction of Immune Tolerance
7. Excitation of Massive Antibody Formation
8. Proliferation of Mature Cells
9. Persistence of Clones

An antibody is a specific globulin which appears in the serum of an animal after the introduction of a foreign substance, an antigen (1). Each of the many globulins is specified by its reaction with a particular antigen (2). Our present concern is to formulate a plausible mechanism for the role of the antigen in evoking large amounts of a specific complementary globulin. An important element of any theory of antibody formation is its interpretation of self-recognition, the means by which an organism discriminates its own constituents from the foreign substances which are valid stimuli of the immune response.

    Recent speculation about antibody formation has been dominated by instructive theories which suppose that the antigen conveys the instructions for the specificity of the globulin synthesized under its governance [antigen directs]. Elective theories date from Ehrlich (9) and have been revived principally by Jerne (10), Talmage (2, 11) and Burnet (12). These postulate that the information required to synthesize a given antibody is already inherent in the organism before the antigenic stimulus is received, and the stimulus then functions to stimulate that mechanism electively [antigen selects]. Jerne had proposed an elective transport of antibody-forming templates to functioning sites; Talmage and Burnet have implicitly proposed an elective function based on cellular selection. The details which distinguish the various proposals are pointed out in the following discussion.

    Immunology does not suffer from a lack of experimental data, but still some of the most elementary questions are undecided, and it is not yet possible to choose between instructive and elective theories. However, the latter have had so little expression in the past decades that a detailed exposition may serve a useful function, if only as a target for experimental attack. This is an attempt to formulate an elective theory on the basis of genetic doctrines developed in studies of microbial populations.

Table 1. Nine propositions.

Al. The stereospecific segment of each antibody globulin is determined by a unique sequence of amino acids.

A2. The cell making a given antibody has a correspondingly unique sequence of nucleotides in a segment of its chromosomal DNA: its "gene for globulin synthesis."

A3. The genic diversity of the precursors of antibody-forming cells arises from a high rate of spontaneous mutation during their lifelong proliferation.

A4. This hypermutability consists of the random assembly of the DNA of globulin gene during certain stages of cellular proliferation.

A5. Each cell, as it begins to mature, spontaneously produces small amounts of the antibody corresponding to its own genotype.

A6. The immature antibody-forming cell is hypersensitive to an antigen-antibody combination: it will be suppressed if it encounters the homologous antigen at this time.

A7. The mature antibody-forming cell is reactive to an antigen-antibody combination: it will be stimulated if it first encounters the homologous antigen at this time. The stimulation comprises the acceleration of protein synthesis and the cytological maturation which mark a "plasma cell".

A8. Mature cells proliferate extensively under antigenic stimulation but are genetically stable and therefore generate large clones genotypically preadapted to produce the homologous antibody.

A9. These clones tend to persist after the disappearance of the antigen, retaining their capacity to react promptly to its later reintroduction.


Of the nine propositions here, only number 5 is central to elective theory. The first four are special postulates chosen as an extreme but self-consistent set; however, they might well be subject to denial or modification without impairing the validity of the elective approach. 

   The last four propositions are stated to account for the general features of antibody formation in cellular terms and may be equally applicable to instructive and elective theories. If this theory can be defended, and I know of no fatal refutation of it, then clearly elective theories of antibody formation, perhaps less doctrinaire in detail, should have a place in further experimental design, each proposition being evaluated on its own merits. 

    I am particularly indebted to Burnet (13) for this formulation, but Burnet should not be held responsible for some elaborations on his original proposal, especially in propositions 1 through 4. A connected statement of the nine propositions is given in Table 1, and each discussed in detail in the following sections.

Antibody Globulin

Al. The stereospecific segment of each antibody globulin is determined by a unique sequence of amino acids.

 This assertion contradicts the more popular notion, and the usual basis of instructive hypotheses, of a uniform sequence subject to differential folding. The chemical evidence is far from decisive. For example, Karush (14) rejects this proposition, not on analytical evidence, but on the cogent argument that miscellaneous antigenic compounds can scarcely convey instructions for sequence. But if instructive-sequence is implausible, this perhaps argues against instruction rather than [against] differential sequence [preexisting differences in primary sequence]. Karush has also demonstrated the remarkable stability of antibody through cycles of exposure to denaturing concentrations of urea. He attributes the structural continuity to stabilizing disulfide linkages, but determinant amino acid sequences may also be involved.

 Elective antibody formation is of course equally compatible with sequence or folding. In such a theory, the mechanism of assembly does not have to be specified, so long as the product (the prospective antibody) recognizes -- that is, reacts with -- the antigen. Differential sequence is proposed (i) to stress the ambiguity of present evidence and (ii) as being more closely analogous to current conceptions of genically controlled specificity of other proteins (15).

 The direct analysis of antibody structure by physicochemical methods has been equivocal. The fractionation of globulins by partition chromatography (16) might be interpreted by differential exposure of phenolic, amino, and carboxyl groups rather than differences in essential composition. Characterization of amino acid composition has given sharply different results with rabbit globulins, on the one hand, and equine and human globulins, on the other. Rabbit globulins, including various antibodies, apparently have a uniform N-terminal sequence, so far identified for five residues as (17):


  Various antibodies were, furthermore, indistinguishable in over-all composition (18). Any chemical differences would then have to attach to a central, differential segment. This possibility is made more tangible by Porter's recent finding (19) that rabbit antibody globulin could be split by crystalline papain into three fragments. One of these was crystallizable (and presumably homogeneous), devoid of antibodyactivity, but equivalent as an antigen to the intact globulin [i.e. react with antiglobulin antibody]. The remaining fractions were more heterogeneous and retained the antigen-combining specificity of the intact antibody. As these fractions may well correspond to the differential segments, their further immunological and chemical analysis will be of extraordinary interest.

  In contrast to the uniformity of rabbit globulins, normal and antibody globulins of horse serum proved to be grossly heterogeneous but equally so, a wide variety of N-terminal groups being found in all preparations (20). This merely confirms the concept of the plurality of antibodies evoked by a given antigen, which have in common only the general properties of normal gamma globulins and the capacity of reacting with the evoking antigen. The globulins of man, and in particular the characteristic globulins produced by different patients suffering from multiple myeloma, are likewise recognizably different, inter se, in amino acid composition (21).

Gene for Globulin Synthesis

A2. The cell making a given antibody has a correspondingly unique sequence of nucleotides in a segment of its chromosomal DNA: its gene for globulin synthesis

  This postulate follows plausibly from proposition Al, and would trace antibody-forming specificity to the same source as is imputed to other specific proteins. As the most deterministic of genetic hypotheses, it should be the most vulnerable to experimental test. For example, a single diploid cell should be capable of two potentialities for antibody formation, one for each chromosome.

 In tests of single antibody-forming cells from rats simultaneously immunized against two Salmonella serotypes, Nossal and I (22) could find only monospecific cells producing one or the other antiflagellin [antibody against the whip-like flagella born by the Salmonella bacterium]. Coons (23) and White (24) have reached a similar conclusion in applications of fluorescent labeling technique. However, Cohn and Lennox (25) have convincing evidence for some bispecific antibody-forming cells in rabbits immunized against two bacteriophages. Experiments pertinent to the possibility of a single cell's carrying more than two antibody-forming specificities remain to be done (26).

  The chromosomal localization of antibody-forming specificity is uncoupled from its elective origin in proposals (7, 8, 27) that an antigen induces a mutation in a gene for globulin synthesis, though not necessarily involving a new nucleotide sequence. Multiple specificity would stand against a simple chromosomal basis for antibody formation (28), leaving two alternative possibilities: (i) replicate chromosomal genes, or (ii) extrachromosomal particles such as microsomes. These might best be disentangled by some technique of genetic recombination.

  The differentiation of microsomes must be implicit in any current statement of a theory of antibody formation that recognizes their central role of protein synthesis. The main issue is whether or not their specificity is dependent on that of the chromosomal DNA. Autonomy of microsomes, in contradiction to proposition A2, is implicit in most instructive theories, the microsome carrying either the original, or a copy of, the antigenic message. On the other hand, a powerful elective theory is generated by substituting the term microsomal for the terms chromosomal DNA and gene in the various propositions. Since a single cell may have millions of microsomes, this theory would allow for any imaginable multiplicity of antibody forming information in a single cell. If the potential variety of this information approaches that of the total antibody response, further instructions in an antigenic input would become moot. In addition, the complexities of selection of cellular populations would be compounded by those of microsomal populations within each cell. These degrees of freedom which blur the distinction between microsomal instruction and election favor the utility of the chromosomal hypothesis as a more accessible target for experimental attack.

Genic Diversity of Precursor Cells

A3. The genic diversity of the precursors of antibody-forming cells arises from a high rate of spontaneous mutation during their lifelong proliferation.

Three elements of this statement should be emphasized:

  • (i) that antibody-forming cells are specialized,

  • (ii) that their diversity arises from some random process, and

  • (iii) that the diversification of these cells continues, in company with their proliferation, throughout the life of the animal.

Item (i) and its justification by various experiments have already been discussed as an aspect of proposition A2. Talmage (2) also stresses the specialization of antibody-forming cells by referring to their progressive differentiation. This is entirely consistent with propositions A3 and A4, which then postulate a specific mechanism of cellular differentiation, in this case, gene mutation. 

  If, on Talmage's model, fully differentiated cells are ultimately left with no more than one antibody-forming specificity per chromosome, the general consequences will be the same, whether this final state represents the unique activation of one among innumerable chromosomal loci (see 27), or the evolution of one among innumerable specific alleles at a given locus. Once again, the final resort for decision may have to be a recombinational technique.

  If the discrepancy between the experiments of Nossal and Lederberg (22) and those of Cohn and Lennox (25), as discussed under proposition A2, is real and depends on the timing of immunization, it may furnish strong support for (ii), the random origin of antibody-forming specificity. If antibody-forming cells can have two (or any small number of) specificities randomly derived, only a negligible proportion will have just the two being tested for. This would correspond to the case of simultaneous immunization with the two test antigens. If, however, a population of cells carrying one specificity is selected for, followed by selection for a second specificity among all available cells, this is the case of serial immunization and is precisely the method one would predict to obtain a clone "heterozygous" for two mutant alleles. Simultaneous versus serial immunization would be analogous to the suppression versus selection of bacterial mutants resistant to two antibiotics (29). Further experiments are needed to exclude more trivial reasons for the scarcity of bispecific antiflagellin-forming cells.

  Item (iii) diverges from Burnet's proposal that the "randomization" of antibody-forming cells is confined to perinatal life, thereby generating a set of then stable clones corresponding to the antibody-forming potentiality of the animal. These clones would then be irreplaceable if lost either by random drift or as a consequence of premature exposure to the corresponding antigen. 

  The arguments against Burnet's proposal are by no means decisive; however, the correspondence between cells and antibodies is made more difficult by having to maintain each clone at a sufficient population size to compensate for loss by random drift. Further, the recurrence of antibody-forming specificity is supported by experiments showing the decay of immune tolerance in the absence of the corresponding antigen (30; see comment on proposition A6). Since immune reactivity in these experiments may return during adult life, susceptibility to the induction and maintenance of tolerance by the timely introduction of the antigen may have only a coincidental relationship to the immunological incompetence of the new-born animal.


A4. This hypermutability consists of the random assembly of the DNA of the "globulin gene" during certain stages of cellular proliferation.

  This ad hoc proposal is doubtless the least defensible of the propositions, and certainly the furthest removed from experimental observation. It is stated to illustrate that accurate replication rather than mutability is the more remarkable phenomenon, whatever the detailed mechanism for the variation. If, as has been suggested, many nucleotide triplets are nonsensical (31), the triplets rather than single nucleotides would have to be posed as the unit of assembly in this case.

  To carry this speculation one step further, heterochromatin has been proposed to be, on the one hand, a random sequence, and, on the other hand, a dissynchronously assembled segment of the genome (32). If both views are correct, proposition A4 might be restated: "the globulin gene is heterochromatic during certain stages of cellular proliferation" (becoming by implication, euchromatic in the mature stages of propositions A8 and A9).

   For the theory of microsomal election it might be postulated that globulinogenic microsomes are initially fabricated as faulty replicas of the globulin gene, but are then capable of exact, autonomous replication.

   Pending more exact knowledge and agreement of opinion on the morphogenetic relationships of antibody-forming cells, the term certain stages cannot be improved upon. On the other hand, as is shown under proposition A8, a model might be constructed even on the basis of a constant but high mutation rate of all antibody-forming cells.

  Further insight into the mechanism of cellular diversity in antibody fomation may be won by studies on the genetic control of reactivity to various antigens in inbred animals (33); two cautions, however, must be stated: (i) for effects on the transport of particles of different size, and (ii) for effects from cross-reactions with gene-controlled constituents evoking autotolerance.

Spontaneous Production of Antibody

A5. Each cell, as it begins to mature, spontaneously produces small amounts of the antibody corresponding to its own genotype.

   Note the implication that antibody is formed prior to the introduction of the antigen into the antibody-forming cell.

 The function of spontaneous antibody is to mark those cells preadapted to react with a given antigen, either to suppress these cells for the induction of immune tolerance (proposition A6) or to excite them to massive antibody formation (proposition A7). Therefore, the antigen need participate in no type of specific reaction with cell constituents other than antibody itself, the one type of reaction available to chemically diverse antigens that requires no further special pleading. 

  There is no agreement whether the reactive globulins found in the serum of untreated animals ["natural antibody"] are produced spontaneously or by casual exposure to cross-reacting antigens (see 2). Accordingly, the spontaneous antibody postulated in proposition A5 may, or may not, be produced in the quantity and form needed for it to be liberated and detected in the serum. The nonspecific fragment of antibody-globulin described by Porter raises the possibility that the same determinant segment may be coupled either to a diffusible or to a cell-bound residue, the latter corresponding to various aspects of cellular immunity, including the suppression or excitation of antibody-forming cells by reactions with the corresponding antigen.

Induction of Immune Tolerance

A6. The immature antibody-forming cell is hypersensitive to an antigen-antibody combination: it will be suppressed if it encounters the homologous antigen at this time.

  This is the first of four propositions which bear less on the source of antibody-forming specificity than on its subsequent expression in terms of cellular behavior. These propositions are therefore equally applicable to instructive theories.

  The duality of reactions of antigens with antibody-forming cells is simply a restatement of the experimental observations of tolerance versus immunity (34). It seems plain that every cell of the antibody-forming system must be marked to inhibit its reactivity both to the autologous antigens of the same animal and extraneous antigens introduced and maintained from a suitably early time of development. 

  In the light of current evidence for the persistence of antigenic molecules (5, 6) and for the loss of tolerance when a given antigen has dissipated (30), there are no more plausible candidates for the self-markers then the antigens themselves. The distinction between the function of an antigen as inhibitor (self-marker) or as inducer of antibody formation is then the time when the antigen is introduced into the potential antibody forming cell, We may profitably define maturity in terms of the progression of the cell from sensitivity towards reactivity.

  The suppression of this process of maturation is a sufficient attribute to account for tolerance, and this need not involve so drastic an event as the destruction of the cell. However, the elective hypothesis proposes that only a limited number of cells will spontaneously react with a given antigen, so that their destruction by premature reaction can safely be invoked as the means of their suppression. It may be hoped that presently documented phenomena of cellular hypersensitivity may furnish a precedent for cellular destruction by such reactions. The cytotoxicity of the antigen to hypersensitive cells is still controversial even in the historical case of tuberculin sensitivity (35). However, the destruction of invading lymphocytes of the host in the course of rejection of a sensitizing homograft (36) supports the speculation of some role of cellular destruction of immature antibody-forming cells in the induction of tolerance.

  The nature of immaturity remains open to question. It might reflect the morphogenetic status of the antibody-forming cell -- for example, sensitive lymphocyte --> reactive plasma cell (37), some particular composition of immature sensitizing antibody, or merely a very low level of antibody so that complexes are formed in which antigen is in excess.

  Finally, one additional hint of an implication of hypersensitivity in the early stages of the antibody response: the transient skin sensitivity of delayed type (and transferable by cells) appearing in the course of immunization, as observed by several workers (38). If these skin reactions reflect the destruction of some antibody-forming cells, it would speak for some overlapping or reversibility of the two stages of maturation.

  The implications of proposition A6 in the elective theory may be summarized as follows: If an antigen is introduced prior to the maturation of any antibody-forming cell, the hypersensitivity of such cells, while still immature, to an antigen-antibody reaction will eliminate specific cell types as they arise by mutation, thereby inducing apparent tolerance to that antigen. After the dissipation of the antigen, reactivity should return as soon as one new mutant cell has arisen and matured. As a further hopeful prediction, it should be possible to induce tolerance in clones of antibody-forming cells from adult animals by exposing a sufficiently small number of initials [?] to a given antigen.

Excitation of Massive Antibody Formation

A7. The mature antibody-forming cell is reactive to an antigen-antibody combination: it will be stimulated if it first encounters the homologous antigen at this time. The stimulation comprises an acceleration of protein synthesis and the cytological maturation which mark a "plasma cell."

  These principles of the cellular response to secondary antigenic stimulation are widely accepted and are readily transposed to the primary response on the elective hypothesis whereby some cells have spontaneously initiated antibody formation according to proposition A5.

Proliferation of Mature Cells

A8. Mature cells proliferate extensively under antigenic stimulation but are genetically stable and therefore generate large clones genotypically preadapted to produce the homologous antibody.

  This proposition takes explicit account of the secondary response, the magnitude of which is a measure of the increase in number of reactive cells (26). However, the antigen need play no direct part in the stabilization of antibody-forming genotype which might accompany the determinate maturation of the cell, whether or not it is stimulated. 

  In fact, it may be possible to dispense with the postulate that mature cells are less mutable by adopting a mutation rate which is an effective compromise: to furnish a variety of genotypes for the primary response while selected genotypes may still expand for the secondary response. For example, by mutation of one daughter chromosome per ten cell divisions, on the average, after ten generations about 600 chromosomes of the same type would have been produced, together with 100 new genotypes distributed among the other 400 or so cells. Selection must then compensate for the mutational drift if a given clone is to be maintained.

Persistence of Clones

A9. These clones tend to persist after the disappearance of the antigen, retaining their capacity to react promptly to its later reintroduction.

  This is a restatement of the possibly controversial phenomenon of lifelong immunity to viruses (4, 5). A substantial reservoir of immunological memory should be inherent from one cycle of expansion of a given clone. Its ultimate decay might be mitigated either by continued selection (that is, persistence of the antigen), stabilization of genotypes, or dormancy (to cell division or remutation, or both) on the part of a fraction of the clone.


Each element of the theory just presented has some precedent in biological fact, but this is testimony of plausibility, not reality. As has already been pointed out, the most questionable proposition is A4, and it may be needlessly fanciful to forward a too explicit hypothesis of mutability for antibody formation when so little is known of its material basis anywhere.

  Theories of antibody formation have, in the past, been deeply influenced by the physiology of inducible enzyme synthesis in bacteria. In particular, instructive theories for the role of the substrate in enzyme induction have encouraged the same speculation about antibody formation. This interpretation of enzyme induction, however, is weakened by the preadaptive occurrence of the enzymes. at a lower level, in uninduced bacteria (39).

  One of the most attractive features of the elective theory is that it proposes no novel reactions: the only ones invoked here are:

  • (i) mutability of DNA;

  • (ii) the role of DNA, presumably through RNA, as a code for amino acid sequence and

  • (iii) the reaction between antibody and antigen, already known to have weighty consequences for cells in its proximity.

  The conceptual picture of enzyme induction would be equally simplified if the enzyme itself were the substrate-receptor. Clearly, susceptibility to enzymic action is not a necessary condition for a compound to be an inducer -- for example, neolactose and thiomethylgalactoside for the beta-D-galactosidase of Escherichia coli (39, 40), but formation of complexes with the enzyme may be. The picture is somewhat complicated by the intervention of specific transport systems for bringing the substrate into the cell (40).

  Antibody formation is the one form of cellular differentiation which inherently requires the utmost plasticity, a problem for which the hypermutability of a patch of DNA may be a specially evolved solution. Other aspects of differentiation may be more explicitly canalized under genotypic control. Nucleotide substitution might still play a role here by modifying the level of activity rather than the specificity of neighboring loci, and elective recognition of transient states spontaneously derived then remains as a formal, if farfetched, possibility for other morphogenetic inductions.

References and Notes

1. This definition excludes antibody-like substances such as the hemagglutinins found in normal human sera. These reagents do not, however, pose the problem of the mechanism of specific response which is the burden of this discussion.

2. Talmage, in this issue of Science, discusses various aspects of antibody specificity, including the number of antibodies, which may be exaggerated in current immunological thought. For the present discussion, however, this number is left open for experimental determination, for it would embarrass a theory of cellular selection only if it is large compared with the number of potential antibody-forming cells in the organism. To anticipate proposition A1, as few as five determinant amino acids would allow for 206 = 3,200,000 types of antibody.

3. L. Pauling, J. Am. Chem. Soc. 62, 2640 (1940).

4. F. M. Burnet and F. Fenner, Heredity 2, 289 (1948).

5. F. Haurowitz, Biol. Revs. Cambridge Phil. Soc. 27, 247 (1952).

6. D. H. Campbell, Blood 12, 589 (1957).

7. A. H. Coons, J. Cellular Comp. Physiol. 52, Suppi. 1, 55 (1958).

8. R. S. Schweet and R. D. Owen, ibid. 50, Suppl. 1, 199 (1957).

9. P. Ehrlich, Studies in Immunity (Wiley, New York, 1910).

10. N. K. Jerne, Proc. Natl. Acad. Sci. U.S.A 41, 849 (1955).

11. D. W. Talmage, Ann. Rev. Med. 8, 239 (1957).

12. F. M. Burnet, Australian J. Sci. 20, 67 (1957).

13. I am also indebted to the Fulbright Educational Exchange Program for furnishing the opportunity of visiting Burnet's laboratory in Melbourne.

14. F. Karush, in Strategical and Biochemical Comparisons of Proteins, W. H. Caie, Ed. (Rutgers Univ. Press, New Brunswick, N.J., 1958), chapter 3.

15. V. M. Ingram, Scientific American 238, No. 1, 68 (1958).

16. R. R. Porter, Biochem. J. 59, 405 (1957).

17. ----- , ibid. 46, 473 (l950); M. L. McFadden and E. L. Smith, J. Biol. Chem. 214, 185 (1955).

18. E. L. Smith, M. L. McFadden, A. Snockell, V. Buettner-Janusch, J. Biol. Chem. 214, 197 (1955).

19. R. R. Porter, Nature 182, 670 (1958).

20. M. L. McFadden and E. L. Smith J. Biol. Chem. 216, 621 (1955).

21. E. L. Smith, D. M. Brown, M. L. McFadden, V. Buettner-Janusch, B. V. Jager, ibid. 216, 601 (1955); F. W. Putnam, Science 122, 275 (1955).

22. G. J. V. Nossal and J. Lederberg, Nature 181, 1419 (1958); G. J. V. Nossal, Brit. J. Exptl. Pathol. 39, 544 (1958).

23. A. H. Coons, J. Cellular Comp. Physiol. 50, Suppl. 1, 242 (1957).

24. R. G. White, Nature 182, 1383 (1958).

25. M. Cohn and E. S. Lennox, private communication.

26. An indirect measure of polyspecificity would be the total number of antibodies multiplied by the proportion of competent cells initially recruited to yield a particular species. Coons (7) has not attempted to count the antibody-forming cells in the primary response, but his statements are compatible with an incidence of 10-5 to 10-3 of cells forming antialbumin in lymph nodes 4 days after inoculation. Nossal ( Brit. J. Exptl. Pathol., in press) found about 2 percent of yielding cells in a primary response after 7 days. These figures are subject to an unknown correction for the extent of proliferation in the interval after innoculation. They perhaps also raise the question whether all the yielding cells are indigenous to the lymph node, or whether circulating cells of appropriate type can be filtered by a node in which locally administered antigen has accumulated.

27. J. Schultz, Science 129, 937 (1959). Schultz makes an analogy between antibody formation and serotype determination in Paramecium, stressing the role of cytoplasmic feedback mechanisms in the maintenance of specificity.

28. A diploid cell should be heterozygous for at most two alleles at one locus, but strictly speaking, this is a restriction of genotype, not phenotype. A cell whose proximate ancestors had mutated through a series of different states might carry a phenotypic residue of information no longer represented in its chromosomes [see linear inheritance in transduction clones: B. A. D. Stocker, J. Gen. Microbiol. 15, 375 (1956); J. Lederberg, Genetics 41, 845 (1956)]. Pending tests on clones from single cells, bi- or polyspecificity of antibody-forming phenotype remains subject to this qualification.

29. V. Bryson and M. Demerec, Am J. Med. 18, 723 (1953).

30. C. H. Tempelis, H. R. Wolfe, A. Mueller, Brit. ]. Exptl. Pathol. 39, 323 (1958); R. T. Smith and R. A. Bridges, J. Exptl. Med. 108, 227 (1958); P. B. Medawar and M. F. A. Woodruff, Immunology 1, 27 (1958); G. J. Nossal, Nature 180, 1427 (1957).

31. F. H. C. Crick, J. S. Griffeth, L, E. Orgel, Proc. Natl. Acad. Sci. U.S. 43, 416 (1957).

32. C. D. Darlington and K. Mather, Nature 149, 66 (1942); J. Schultz, Cold Spring Harbor Symposia. Quant. Biol. 12, 179 (1947); A. Ficq and C. Pavan, Nature 180, 983 (1957).

33. J. H. Sang and W. R. Sobey, J. Immunol. 72, 52 (1954); M. A. Fink and V. A. Quinn, ibid. 70, 61 (1953).

34. M. Cohn, Ann. N.Y. Acad. Sci. 64, 859 (1957).

35. C. B. Favour, Intern. Arch. Allergy 10, 193 (1957); B. H. Waksman and M. Matoltsy, J. Immunol. 81, 220 (1958).

36. J. M. Weaver, G. H. Algire, R. T. Prehn, J. Natl. Cancer Inst. 15, 1737 (1955).

37. J. W. Rebuck, R. W. Monto, E. A. Monaghan, J. M. Riddle, Ann. N.Y. Acad. Sci. 73, 8 (1958).

38. L. Dienes and T. B. Mallory, Am. J. Pathol. 8, 689 (1932); M. Tremaine, J. Immunol. 79, 467 (1957); J. W. Uhr, S. B. Salvin, A. M. Pappenheimer, Jr., J. Exptl. Med. 105,11 (1957); S. Raffel and J. M. Newel, ibid. 108, 823 (1958).

39. J. Lederberg, in Enzymes: Units of Biological Structure and Function, , O. H. Gaebler, Ed. (Academic Press, New York, 1956), p. 161. A feeble attempt in this paper to homologize antibody formation with elective enryme induction was hindered by an uncritical rejection of proposition Al and by the want of a tangible cellular model such as Burnet and Talmage have since furnished.

40. J. Monod, ibid., p. 7.


Two contrasting responses to antigen are immunity and tolerance. The cellular basis for this, deriving from the clonal selection theory in the late 1950s, was considered to be that immunologically competent cells responded differentially to antigen depending on their developmental stage. 

  While not excluding the latter, in the mid 1960s it was considered that a cell might also respond differentially to different antigen concentrations. At low antigen concentrations a cell might be stimulated (i.e. immunity), and at high antigen concentrations a cell might be inhibited (i.e. tolerance; Forsdyke, 1966; 1967; 1968; 1969; Doherty & Robertson 2004). This "two signal" hypothesis led  to the view that the education of immunologically competent cells (ICC; later known as B and T cells) would involve, not only negative selection (tolerance) as had been suggested by Burnet, but also positive selection. 

  Negative selection meant that the ICC reacting with "self" would be eliminated, leaving holes in the ICC repertoire. The remaining ICC would have to cope with a large, and apparently unpredictable, spectrum of antigens associated with "non-self" foreign agents. Could the specificities of these remaining ICC be broad enough to cope with all invaders? Would the specificities arise randomly? Alternatively, could there be some sort of forearming? For example, foreign agents that could, step-by-step, mutate to appear more like the "self" of their host organism, would have an edge over those that could not. They would then be able to exploit the holes in the ICC repertoire. Could the host, in some way, bias its ICC repertoire to counteract this?

  This  idea was developed in the early 1970s (see below), while  the "two signal hypothesis" was ingeniously explored by Cohn and his colleagues (Bretscher & Cohn, 1968,1970; Cohn 1989,1994; Langman & Cohn 1996). However, Cohn postulated inhibition at low antigen concentrations, and stimulation at high antigen concentrations. While not excluding multiple zones of sensitivity to different concentrations (and modes of presentation) of antigenic determinants, most studies now tend to favour inhibition by high antigen concentrations (e. g. Alexander-Miller et al., 1996; Gaudin et al. 2004), as argued in the following paper. 

   There is also more evidence that, as suggested by Azar and coworkers (1968, 1971), some forms of tolerance may require complement (e.g. Prodeus, A. P. et al. 1998; Manderson et al. 2004; Ferry et al. 2007), and that natural autoantibodies are secreted by positively selected B cells (Gaudin et al. 2004), and can exert a "buffering" function (Coutinho & Avrameas 1992).

The Lancet (1968) 1, 281-283 (10th February)

With copyright permission from The Lancet Ltd.

The Liquid Scintillation Counter as an Analogy for the Distinction between "Self" and "Not-self" in Immunological Systems


Liquid Scintillation Counting
Two Antibody Sites Buffered By Natural Antibody
Elimination Of Self-Reacting Cells
End Note

Summary. The natural selection theory of immunity of Jerne, with its emphasis on natural antibody, is combined with the clonal selection theory of Burnet, with its emphasis on cells, to produce a simple "two site" hypothesis of the mechanism of immune self-recognition in vivo. An analogy is made with a similar process occurring in liquid scintillation counters containing two photocells and a coincidence circuit.



    I explore here the extent to which the principles of a known process in a non-biological system can be applied to the biological problem of immune self-recognition in vivo. The result is a scheme of interrelationships between various cellular and humoral factors which, although it rests on tenuous experimental evidence, seems to be not too inconsistent with the currently accepted facts of immunology.

   Selective theories of immunity (1-4) have four main elements:

  • A randomisation process by which a wide spectrum of units, each able to direct the formation of a specific antibody, are established.

  • A mechanism forbidding immune reaction with "self" and permitting recognition of foreign ("not-self") antigenic determinants.

  • Activation of the synthesis of specific antibody by foreign antigenic determinants.

  • Amplification of this response.

A number of more detailed, and hence more controversial, postulates will be made here. These are that:

  • (a)  Randomisation occurs in the thymus (5, 6).

  • (b)  The units involved are single cells (7, 8).

  • (c)  Such cells are released from the thymus to seed other tissues (9, 10)
  • (d)  At some stage during this process self-recognition occurs.

A possible mechanism by which self-recognition could occur in vivo will be derived from a consideration of an analogous process occurring in modern liquid scintillation counters. Insofar as activation is closely linked with the postulated recognition process it will also be discussed. Neither the mechanism of randomisation nor amplification, will be considered here.


  Radioactivity in a scintillation solution emits pulses of light energy which may be detected by a photocell (11) which converts light energy into electrical energy. In theory, simply connecting the photocell to a counting meter should provide a record of the disintegrations occurring in the solution, and thus of the amount of radioactivity present. In practice, however, the background "noise" within the photocell seriously reduces the sensitivity of the method. The counting meter needs to distinguish between pulses of energy originating within the scintillation solution ("not-self "), and those originating within the photocell ("self").

  The difficulty is overcome by using a coincidence circuit in which the scintillation solution is looked at, not by one, but by two photocells. Spontaneous discharge in each photocell individually is virtually eliminated by permitting only electrical pulses arriving simultaneously from both photocells to activate the counting meter (12; Fig. 1).

lancet03.gif (423262 bytes)

Fig. 1 - Mechanism of self-recognition in a liquid scintillation counter.

The radioactivity to be measured is mixed with a scintillation solution in a glass bottle and placed between two photocells. These convert light energy into electrical energy. The system is in complete darkness.

  • In A a radioactive disintegration causes the emission of a pulse of light energy (X), which is seen by both photocells simultaneously. These feed electrical pulses through a coincidence circuit to a counting meter. Since both pulses arrive simultaneously at the coincidence circuit they are permitted to activate the meter.

  • In B spontaneous discharge (X) within one photocell does not activate the counting meter since the electrical pulse cannot pass the coincidence circuit. Thus the machine successfully distinguishes between spontaneous internal discharge which it does not wish to measure ("self"), and discharge caused by radioactivity in the scintillation solution ("not-self").


    A postulate of most clonal theories of immunity is that somewhere in or on a cell is a site capable of reacting with an antigenic determinant against which the cell is uniquely able to respond immunologically. A cell is considered to bear an antibody site which it has itself synthesized in accordance with the unique information coded in its nucleic acid. A coincidence circuit model would require at least two such "cell-borne" sites. Activation of one site might mean "self" and activation of both sites simultaneously might mean "not-self" (or vice versa).

    How might such a differential activation of sites arise? Clearly, differences in determinant concentration would permit a distinction. If an animal were injected with a dose of antigen sufficient only to activate one site, then twice the dose of antigen might activate both sites. However, the minimum concentration of an antigenic determinant required to react with a cell would be very low (low threshold), and the range of discrimination (one or two sites occupied) would be severely restricted. The known ability of animals to respond to a very wide range of antigen concentrations conflicts with this simple idea.

    A resolution of the conflict may be obtained by combining the early "
natural selection" theory of Jerne (l), which emphasizes the role of natural antibody, with the clonal theories of others (2,3), which emphasize the role of cells. It may be postulated that immunologically competent cells constantly secrete very small quantities of natural antibody so that at any one moment there are a large number of free antibody molecules for each cell capable of reacting with the same determinant. 

  The natural antibody molecules would "buffer" the cell against changes in antigenic determinant concentration; the threshold for reaction with a cell would be raised, and the range of discrimination would be increased. The critical factor governing combination of a determinant with a cell would be, not the determinant/cell concentration ratio, but the determinant/natural-antibody concentration ratio. (Factors such as the number and distribution of the cells of a particular clone, antigen diffusion gradients, and dissociation constants will not be considered here.)


    In what ways do self determinants differ from foreign determinants? Self determinants can vary over a wide range of concentrations (compare, for example, serum-albumin with the serum levels of some protein hormones; 13). In theory, foreign determinants can also vary over an equally wide range of concentrations. The only definitive statement which can be made about foreign determinants is that they are unlikely to be constantly present; for much of the time their concentration is likely to be zero. Self-determinants are likely to be constantly present at some concentration higher than zero.

    If the factor critical to the differential activation of cell-borne antibody sites is the determinant / natural -antibody concentration ratio, then such activation will be influenced either by a change in determinant concentration, or a change in natural antibody concentration. A consideration of the temporal sequence of events following completion of the thymic randomisation process may now clarify the problem.

    I suggest that a cell endowed with a unique ability to form a particular antibody emerges from the thymus with its two cell-borne antibody sites already synthesized and placed in a position accessible to extracellular determinants; the cell has not yet initiated the secretion of the corresponding natural antibody.  At this unique moment in time, circulating self determinants, even at very low concentration, if possessing sufficient avidity, would combine with both cell-borne sites. It is postulated that such a union destroys a cell, possibly by a mechanism involving complement. Cells not so destroyed would initiate the secretion of natural antibody, so that at later times the cell would be buffered against changes in antigen concentration. Occupation of one cell-borne site would activate the cell to respond immunologically (Fig. 2).

lancet04.gif (23790 bytes)

Fig. 2 - Hypothetical mechanism by which immunologically competent cells capable of reacting with "self" are eliminated in vivo.

  Following a "randomisation" process, lymphocytes are released from the thymus each bearing two antibody sites of identical specificity in close proximity at its surface. "Self " antigenic determinants, present at the unique moment of release, destroy self-reacting cells by combining with both cell-borne antibody sites, and fixing complement. Surviving non-self-reacting cells then initiate the secretion of natural antibody. Free natural antibody molecules buffer cells against changes in antigenic determinant concentration so that generally only one of the cell-borne sites reacts with determinant, and an immune response is initiated.


  That immune responses in vivo can be critically dependent on antigen concentration is well documented (14,15). There are a number of in-vitro studies which may be interpreted as showing that lymphoid cells respond to low concentrations of antigen and are inhibited at higher concentrations:

  • It has been shown that whereas antigen added directly to a suspension of lymphoid cells containing macrophages will not induce a primary immune response, the antigen extracted in lower concentrations from antigen-treated peritoneal-exudate cells will induce such a response. When the concentration of antigen added to the peritoneal cells was increased, the ability of the extract to induce a response disappeared (16-20).

  • Phenol extracts of the spleens of antigen-treated animals also appeared inhibitory in an in-vitro lymphoid-cell system. On dilution, the extracts could induce a primary immune response (21).

  • Others have found that a primary response of lymphoid-tissue fragments to erythrocytes is only demonstrable by lowering the ratio of antigen concentration to tissue concentration (22).
  • Experiments designed to examine more precisely the antigen dose-response requirements of in-vitro immune responses have provided less controversial evidence. The secondary response in vitro of circulating blood leucocytes showed an inhibition at high antigen concentrations (23, 24); but attempts to demonstrate the specificity of this inhibition were unsuccessful (25). Recently, however, the specificity of inhibition by excess antigen has been shown in an in-vitro primary-response system (26).

  Although a case can be made out for the existence of natural antibody (27, 28), there is no evidence that it buffers cells. However, an interesting analogy with the postulated cell buffering has arisen from studies of the in-vitro activation of lymphocytes by phytohaemagglutinin. The transforming activity of solutions of phytohaemagglutinin could be adsorbed out with leucocytes (29, 30); this indicated that the activating principle might produce its effect by reacting with cells directly. However, the actual quantity of phytohaemagglutinin required to produce a given activation response of cells was not proportional to the cell concentration, but to the concentration of serum in the culture medium. The response was proportional to the phytohaemagglutinin/serum concentration ratio (31-33).

   The requirement of the combination of two antigenic determinants with two cell-borne antibody sites for cell destruction, is compatible with current knowledge of the action of complement on cells. It is likely that not one, but two, separate reactions in close proximity of antigenic determinants with antibody sites must occur for complement to be fixed and destroy a cell (34, 35). In the presence of complement, anti-lymphocyte antibody stimulates lymphocytes to transform in vitro; however, at high antibody concentrations cells are destroyed (36-38). Anti-immunoglobulin antibody also activates lymphocytes to transform in vitro, and evidence has been obtained suggesting a reaction of the anti-immunoglobulin antibody with a cell-borne immunoglobulin (39).

    Fig. 2 summarizes the main outlines of the hypothesis. A preliminary account of this work has been published (32). A more extensive hypothesis is presented elsewhere (40).


1. Jerne, N. K. Proc. natn. Acad. Sci., Wash. 1955, 41, 849.
2. Burnet, F. M, The Clonal Selection Theory of Acquired Immunity. 1959.
3. Lederberg, J. Science, N. Y. 1959, 129, 1649.
4. Eisen, H. N., Karush, F. Nature, Lond. 1964, 202, 677.
5. Claman, H. N., Talmage, D. W. Science, N.Y. 1963, 141, 1193.
6. Cross, A. M., Leuchars, E., Miller, J. F. A. P. J. exp. Med. 1964, 119, 837.
7. Nossal, J. G. V., Szenberg, A., Ada, G. L. G., Austin, C. M. ibid. 1963, 119, 485.
8. Friedman, H. P. Experientia, Basel, 1964, 20, 564.
9. Nossal, J. G. V. Ann. N.Y. Acad. Sci. 1964, 120, 171.
10. Murray, R. G., Woods, P. A. Anat. Rec. 1964, 150, 113.
11. Bell, C. G., Hayes, F. N. (editors). Liquid Scintillation Counting. London, 1958.
12. Morton, G. A., Robinson, K. W. Nucleonics, 1949, 4, 25.
13. Hales, C. N., Randle, P. J. Lancet, 1963, i, 790.
14. Dixon, F. J., Maurer, P. H. J. exp. Med. 1955, 101, 245.
15. Mitchison, N. A. Proc. R. Soc. B. 1964, 161, 275.
16. Fishman, M. J. exp. Med. 1961, 114, 837.
17. Fishman, M., Adler, F. L. ibid. 1963, 117, 595.
18. Fishman, M., Van Rood, J. J., Adler, F. L. in Molecular and Cellular Basis of Antibody Formation (edited by J. Sterzl); p. 491. New York, 1965.
19. Askonas, B. A., Rhodes, J. M. Nature, Lond. 1965, 205, 470.
20. Friedman, H. P., Stavitsky, A. B., Solomon, J. M. Science, N. Y. 1965, 149, 1106.
21. Friedman, H. P. ibid. 1964, 146, 934.
22. Globerson, A., Auerbach, R. J. exp. Med. 1966, 124, 1001.
23. Cowling, D., Quaglino, D. J. Path. Bact. 1965, 89, 63.
24. Caron, G. A. Int. Archs Allerg. 1966, 30, 331.
25. Caron, G. A. ibid. 1967, 31, 441.
26. Diener, E., Armstrong W. D. Lancet, 1967, ii, 1281.
27. Boyden, S. V. Adv. Immunol. 1966, 5, 1.
28. Kerman, R., Segre, D., Myers, W. L. Science, N.Y. 1967, 156, 1514.
29. Kolodny, R. S., Hirschhorn, K. Nature, Lond. 1964, 201, 715.
30. Nordman, C. T., De La Chapelle, A., Grasbeck, R. Acta med. scand. 1964, supplement no. 412, p. 49.
31. Forsdyke, D. R. Lancet, 1966, i, 715.
32. Forsdyke, D. R. in The Biological Effects of Phytohaemagglutinin (edited by M. W. Elves);  pp. 115, 195. R. Jones and A. Hunt Orthopaedic Hospital, Oswestry, 1966.
33. Forsdyke, D. R. Biochem. J. 1967, 105, 679.
34. Humphrey, J. H., Dourmashkin, R. R. in Ciba Fdn Symp. Complement (edited by G. E. W. Wolstenholme and J. Knight); p. 175. London, 1965.
35. Borsos, T., Rapp, H. J. Science, N. Y. 1966, 150, 505.
36. Grasbeck, R., Nordman, C., De La Chapelle, A. Acta med. Scand. 1964, supplement no. 412, p. 39.
37. Holt, L. J., Ling, N. R., Stanworth, D. R. Immunochemistry, 1966, 3, 359.
38. Forsdyke, D. R. Unpublished, 1967.
39. Sell, S., Gell, P. G. H. J. exp. Med. 1965, 122, 923.
40. Forsdyke, D. R. Ph. D. thesis, University of Cambridge, 1967.[This provided the basis for a paper in the J.Theoretical Biology in 1969.]

END NOTE  Commentary on the above paper may be found in:

Cohn, M. (1989) History of Associative Recognition Theory. In Foreword to The Immune System, Academic Press, San Diego.

Cohn. M. (1994) The Wisdom of Hindsight. Annu. Rev. Immunol. 12, 1-62.

Doherty, M. & Robertson, M. J. (2004) Some early Trends in Immunology. Trends Immunol. 25, 623-631.

Langman, R. E. & Cohn, M. (1996) A short history of time and space in immune discrimination. Scand. J. Immunol. 44, 544-548.

Podolsky, S. H. & Tauber, A. (1997) The Generation of Diversity: Clonal Selection Theory and the Rise of Molecular Immunology. pp. 151-152.

Further References

Alexander-Miller, M. A., Leggatt, G. R., Sarin, A. & Berzofsky, J. A. (1996) J. Exp. Med. 184, 485-492. Role of antigen, CD8, and cytotoxic T lymphocyte (CTL) avidity in high dose antigen induction of apoptosis of effector CTL.

Bretcher, P. & Cohn, M. (1968) Nature 220, 444-448. Minimal model for the mechanism of antibody induction and paralysis by antigen.

Bretcher, M. A. & Cohn, M. (1970) Science 169, 1042-1049. A theory of self-nonself discrimination.

Coutinho, A. & Avrameas, S. (1992) Speculations on immunosomatics: potential diagnostic and therapeutic value of immune homeostasis concepts. Scand. J. Immun. 36, 527-532.

Ferry H. et al. (2007) Increased positive selection of B1 cells and reduced B cell tolerance to intracellular antigens in C1q-deficient mice. J. Immunol. 178, 2916-2922.

Forsdyke, D. R. (1969) J. Theor. Biol. 25, 173-185. A theory of immunity.

Gaudin E. et al. (2004) J. Exp. Med. 199, 843-853. Positive selection of B cells expressing low densities of self-reactive BCRs.

Kappler, J. W., Skidmore, B., White, J. & Marrack, P. (1981) Antigen-inducible, H2-restricted, interleukin-2-producing T cell hybridomas. Lack of independent antigen and H-2 recognition. J. Exp. Med 153, 1198-1213.

Manderson, A. P. et al. (2004) Annu. Rev. Immunol. 22, 431-456. The role of complement in the development of systemic lupus erythematosus.

Prodeus, A. P., et al. (1998) Immunity 9, 721-731. A critical role for complement in the maintenance of self tolerance.



Jerne and Positive Selection

Some Early Quotations

Further Implications of a Theory of Immunity (1975)

End Note on the Differential Avidity/Affinity Model

"Alternative Self" or "Near Self" (2005)

Review of the above (2012)


It followed from the two signal hypothesis that a concentration of antigen which could inactivate a cell with high specificity receptors might, at the same time, activate a cell with low specificity receptors. This is illustrated in the following "Figure 6" from Immunology (1973b) 25, 597-612, which deals with the incorporation of tritium-labeled thymidine by cultures of lymph-node cells in the absence or presence of added antigen. 

Figure 6.immunol2.gif (13787 bytes)

Legend to Figure 6. Theoretical antigen dose-response curves for three subpopulations (clones) of cells (A, B, C) of decreasing specificity for the antigen, in the presence (curve 1) or absence (curve 2) of a complement inhibitor [held to prevent high-dose inhibition]. At low antigen concentrations clone A is stimulated. As the concentration of antigen increases clone A begins to be inhibited, but clone B begins to be stimulated. At still higher concentrations, clone B begins to be inhibited, but clone C begins to be stimulated. The shape of the resultant curve depends on the size of each population. At early stages of the immune response there would be more low specificity cells (C) than high specificity cells (A; Forsdyke, 1969), and a curve similar to curve 2 would be obtained. If the inhibitory component of these curves could be eliminated with a complement inhibitor, then curve 1 would be obtained.

   This led  to the view that the education of immunologically competent cells (B and T cells) would involve not only negative selection, but also positive selection. Jerne (1971) made the interesting suggestion that some germ-line genes for immunoglobulins had been selected over evolutionary time for reactivity with self MHC. The following letter shows that he did not think positive selection would result.

Jerne and Positive Selection   LETTER


Immunology Today (1995) 16, 105.
(With copyright permission from Elsevier Publications.)

In a delightful review (1) entitled "Why Positive Selection?" Polly Matzinger adopts a position of "unabashed advocacy" and declares that Niels Jerne "invented positive selection", citing his paper published in 1971 (2). This puts those of us who have not been citing Jerne in this respect (3) in rather an embarrassing position. Have we been unfair to Jerne?

  I believe that Matzinger may, from a modern perspective, have read more into Jerne’s paper than he originally intended. In 1971, immunologists were particularly concerned with explanations for high rates of cell proliferation and death in the thymus and the relatively high proportion of lymphocytes involved in allorecognition phenomena. Jerne made the novel postulate that an organism’s germline "V-region genes" encode the combining sites of antibodies directed against the histocompatibility antigens of its own species. He distinguished two lymphocyte subsets:

  • One with receptors directed against allogeneic histocompatibility antigens (subset A).

  • One with receptors directed against self histocompatibility antigens (subset S).

Cells of the former subset:

"proliferate to some extent after which the descendent cells leave the thymus and become antigen-sensitive cells specifically capable of reacting against allogeneic histocompatibility antigens that the individual does not possess" (2).

    Jerne stated that a cell of subset S:

"proliferates, perhaps because a hormone stimulates the proliferation of all stem cells entering the thymus."

He then comes very near to the idea of positive selection by adding that:

"Possibly, the histocompatibility antigens in the thymus which fit to cell receptors provide additional stimulation."

However, he went on to say that:

"I assume, however, that for this same reason none of the cells of this forbidden clone will be permitted to leave the thymus as antigen-sensitive cells, but that they will eventually all die out".

Only cells that can mutate survive, so that:

"finally, cells will arise that have entirely lost their fit to the histocompatibility antigens of the individual itself" (my italics).

  Thus, there is ongoing proliferation in the thymus and it is the absence of negative selection, not positive selection, that in Jerne’s view allows mutant cells to survive. The role of intrathymic self-histocompatibility antigens in driving the proliferation of the S subset was added as an afterthought with no implication of a positive selection for the ability to react with histocompatibility antigens (the role of the A subset). Indeed, Jerne concluded (2) that:

"The restriction of ontogenic selection to random mutants of cells expressing V-genes of subset S thus determines both the responsiveness to certain types of antigen and the range of antibody specificities that an individual animal can produce, so that, indirectly, these properties are under dominant control of histocompatibility genes" (my italics).

(1) Matzinger, P. (1993) Immunol. Rev. 135, 81-117.
(2) Jerne, N. K. (1971) Eur. J. Immunol. 1 1 1, 1-9.
(3) Forsdyke, D. R. (1975) J. Theor. Biol. 150, 451-456.


Another unabashed advocist of Jerne's ideas was Philippa Marrack who, from the Presidential podium of the American Association of Immunologists was, as late as 2001, describing Jerne's 1971 paper as "visionary" (J. Immunology 167, 617-621) in proposing that "lymphocyte receptors were selected evolutionarily to react with MHC proteins." 

  However, she noted that while "immunologists [have] clung to the idea," from structural evidence they "have had to reluctantly conclude that TCR and MHC may not have some conserved fit for each other." Thus, "the germline repertoire of ... TCRs [T cell receptors] may actually be completely random and the high frequency with which TCRs react with MHC may ... simply be because of positive selection in the thymus for low reaction with self-MHC (plus peptide) and heteroclitic cross reaction between self and foreign MHC."

   It should be noted that, although the idea was in circulation, 1971 was prior to Tonegawa's demonstration of multiple V regions which translocate to a limited number of constant (C) regions, from which free (secreted) and cell-bound immunoglobulins (receptors) would be derived. In 1971 the problem of how antibody variability was generated was a black box, and "mutation" was as good a way to allude to what might be going on in that box as any other. However, as for "visionary," perhaps we should pause. Visions can be either correct or incorrect.

  Jerne's work was presented in September 1969 at a WHO meeting, and again at a Brook Lodge conference on Immunological Surveillance in 1970. Copies were widely circulated, and I received a copy second hand. Jerne's 1971 paper carried ad hoc assumptions such as: "I assume that antibodies directed against self-components on cell surfaces have some important function in ontology." He should have known that mules are sterile because gametogenesis is impaired, so that there are no offspring. Yet he wrote: "Mules may be viable because their cells contain complete haploid sets of v-genes for the relevant histocompatibility antigens of both parents, but the cells of their offspring would most often be deficient in this respect."

   The high point of Jerne's paper came early with the statement, of which the Brook Lodge version (1970; full reference below) is clearer: 

"The clonal selection theory deals with the responsiveness of the immune system to foreign antigens. In order to respond, however, the immune system must already be functional before foreign antigens arrive. My present theory deals, in the first instance, with the period in ontogeny which precedes the invasion of the body by foreign antigens. If the ontogenic development of the immune system involves antigens -- and such is my a priori argument -- then these must be the antigens of the individual himself."

This might well be the introduction to a paper on positive selection. However, Jerne's is not that paper. In 2004 Marrack and her colleagues seemed to agree, noting that, at that time, Jerne "could not have predicted the idea of positive selection". They also stated that "the notion of death-by-neglect had certainly not come up" (Huseby et al. 2004. Eur. J. Immun. 34, 1243-50).

    In 1998 Thomas Soderqvist's biography of Jerne was published in Danish, and in 2003 an English translation by David Paul became available. However, I did not get round to reading it until September 2010. Several issues were clarified, as is indicated elsewhere in these pages where I press for reform of the peer-review system using, as example, the "ultimate" test of that system, the awarding of a Nobel prize in the context of the clonal selection theory (Click Here).

     In 1984 Jerne shared the prize with Kohler and Milstein who had discovered how to make large quantities of monoclonal antibodies - a finding of much clinical value. Sodersvist's book went some way in examining the rationale for Jerne's Nobel prize. First we should note that, despite numerous prior nominations, it was not until 1922 that Albert Einstein received the Nobel prize. It was for his discovery of the photoelectric effect, and did not take "into account the value which will be accorded your relativity and gravitational theories after these are confirmed in the future."

     Formally, Jerne's Nobel award was "for theories concerning the specificity in development and control of the immune system." But these, in themselves, had little to do with the Kohler-Milstein work, and have not yet been "confirmed in the future." Jerne's natural selection theory (1955), hinted at the truth in very important ways, but was mechanistically wrong. Since he was aware of Paul Ehrlich's work, many were surprised that he did not take the clonal step. And, as mentioned above, his ideas on the selection of the immunological repertoire (1971), while again hinting at the truth, are wrong. Jerne's 1974 network theory (an elaboration of Ehrlich's idea that an antigen-binding site on an antibody molecule can itself be antigenic) has to date gained little support. A historian (A. I. Tauber) recently commented (Isis, 2010) on the account of Jerne's network theory in the second edition (2009) of Arthur Silverstein's monumental A History of Immunology:

This theory enjoyed a great vogue from its presentation in 1974 through the 1980s, and although Silverstein presents the conceptual antecedents, he neglects to appraise the reader of the theory's demise and the implications of its downfall. That major textbooks completely omitted any mention of Jerne's network by the late 1990's reflects how other theoretical concerns vying for dominance signify a key conceptual struggle within the discipline that dates to its origins.

    Soderqvist notes that a professor of immunology at the Karolinska Institute publicly "explained that Jerne received the prize for his 'visionary theories' that had enabled modern immunology to make major leaps in progress, and in a private letter to Jerne pointed out that "it was good that we are also able to reward theories; actually, this is the first time [in medicine], as far as I know."' Soderqvist points out that formally the Nobel Foundation is required to give prizes to those that have made "the most important discoveries within the domain of" the subject (in this case physiology or medicine). Thus the scope of the word "discovery" had been extended beyond facts to theories. One can "discover" a theory. Further information on the matter will not be available until 2034 when the Nobel Foundation opens its 1984 files.


End Note (Oct 2012) The Jerne (1971) hypothesis of germ-line bias for MHC reactivity was further thrown into doubt by Holland et al. (2012), which reviewed evidence from the Singer lab. (2007, 2012).

   Holland SJ et al. (2012) The T-cell receptor is not hardwired to engage MHC ligands. Proceedings of the National Academy of Sciences, USA doi/10.1073/pnas.1210882109. (See also Forsdyke 2012; Click Here)


Some early quotations on Positive Selection

On incorporation of [3 H] thymidine by control cultures of lymph-node cells with no added antigen. [Forsdyke, D. R.  Immunology (1973a) 25, 583-595. Serum factors affecting the incorporation of [3 H]thymidine by lymphocytes stimulated by antigen. I. Serum concentration.]
"..Labelling in control cultures might represent a response to endogenous antigens; these might be bound to, or released from, phagocytic cells present in culture (Garvey and Cambell, 1966). Such antigens would include
  • (i) the antigen under study in experiments with preimmunized animals,
  • (ii) other antigens acquired either before or after immunization, and
  • (iii) 'self' antigens capable of stimulating low specificity anti-self cells. A high concentration of such low specificity cells would be predicted from theoretical considerations previously advanced (Forsdyke, 1969)."
On incorporation of  [3 H]thymidine by cultures of lymph-node cells stimulated by antigen. [Forsdyke, D. R.  Immunology (1973b) 25, 597-612. Serum factors affecting the incorporation of [3 H]thymidine by lymphocytes stimulated by antigen. II. Evidence for a role of complement from studies with heated serum. (See Fig. 6 above).]
"The results are discussed in relationship to models which require that the size of a specific lymphocyte clone be positively or negatively regulated by the concentration of antigen specific for that clone."

"..relationship to models in which the size of a specific lymphocyte population (clone) may be regulated by positive or negative signals dependent on the concentration of antigenic determinants with specificity for receptor sites borne by cells of the clone (e.g. Forsdyke, 1966, 1968, 1969)."

For more on antigen-dose response curves see Forsdyke (1977; Click Here)

Further Implications of a Theory of Immunity


J. Theoretical  Biology (1975) 52, 187-198  (Received 28th May 1974)

(With copyright permission from Academic Press)



Restatement of Main Features of the Theory


Further Implications




An Overview

End Note on the Differential Avidity/Affinity Model

Abstract . A theory of immunity presented previously showed that many immune phenomena could be explained in terms of a simple model involving interactions between:

  • (i) Receptors on immunologically competent cells.

  • (ii) Antigenic determinants.

  • (iii) Natural antibody.

  • (iv) Complement.

Several features of the theory subsequently gained experimental support. The present paper extends the analysis and predicts that two mechanisms operate in the induction of immunological tolerance to self antigenic determinants. High specificity anti-self cells are destroyed by a mechanism involving complement. In contrast, low specificity anti-self cells are stimulated by self determinants and increase in numbers. This increases the concentration of the natural antibody secreted by these cells which acts as a "blocking" antibody preventing their continued stimulation by self determinants.

  The expanded clones of low specificity anti-self cells, which may be of high specificity for "near-self" determinants:

  1. Are responsible for the greater immunological responsiveness between non-identical members of the same species than between members of different species ("alloaggression").

  2. Provide a barrier opposing the progressive evolution of the surface determinants of a pathogen into forms identical with the surface determinants of its host.

Following exposure to a given self or not-self antigenic determinant, the distribution curve for cells of varying specificities for the determinant shows a sharp cut-off point between high and low specificity cells. The position of this cut-off point is critical in determining the subsequent response of the organism to the determinant. Variables affecting the cut-off point include, antibody present prior to the exposure of cells to the determinant, complement, complement inhibitors, the cell membrane and certain drugs. Autoimmune diseases are improved by drugs (e.g. chloroquine) which move the position of the cut-off point towards cells of low specificities for self determinants.


1. Introduction

  A theory of immunity was presented previously (Forsdyke, 1966, 1967, 1968, 1969). The theory, based on the clonal selection theory (Burnet, 1959), placed special emphasis on the role of two groups of humoral factors, natural antibody and complement, in modulating the response of immunologically competent cells to varying concentrations of antigenic determinant. A mechanism by which higher organisms distinguish immunologically between "self" and " not-self " was derived from a consideration of the possible molecular events occurring at the lymphocyte surface in the process of the induction of high-dose immunological tolerance.

  Several features of the theory have since gained experimental support and this has been reviewed (Forsdyke, 1973a,b,c). The purpose of the present paper is to explore more fully certain aspects of the theory as they reflect on recent developments in immunology. In particular, attention is drawn to the role of expanded clones of anti-"near-self " cells in:

(i) The phenomenon of "alloaggression" (Jerne, 1971).

(ii) The stabilization of the surface determinants of a pathogen relative to those of its host.

2. Restatement of Main Features of the Theory

  In this paper an antibody combining site is described as a "high specificity receptor" if, at a given concentration of an antigenic determinant, it combines with the determinant in a fashion such that there is a higher probability of bringing about an effect dependent on such an interaction than the combination of a "low specificity receptor" with the determinant. 

  The production of a particular effect, such as cell proliferation in the case of a cell-borne antibody receptor, probably requires close receptor- determinant interaction for a discrete period of time. This is likely to relate to the chemical affinity of the receptor for the determinant (Eisen, 1966).


  The theory postulated that a lymphocyte with the potential to make a particular antibody against an antigenic determinant bears receptor antibody molecules on its surface in a form capable of combining with the antigenic determinant. In addition the cell secretes the antibody at a low rate. This "natural" antibody normally buffers the cell against reaction with the antigenic determinant. The critical factor affecting the number of receptors which combine with the determinant at a given determinant concentration, is not the determinant concentration itself, but the determinant/natural antibody concentration ratio. Normal variations in the number, distribution or mobility of surface receptors are also not critical; for the purposes of the theory all the immunologically competent cells within an individual organism are considered equal in these respects.

  A stimulation of the cell to initiate an immune response follows reaction of antigenic determinants with a limited number of cell-borne receptors. This condition is achieved at "low" or "moderate" determinant concentrations due to the buffering of the cell against reaction with excess determinants by natural antibody. 

  However, at "high" determinant concentrations the buffering is overcome and there is an increased probability of the simultaneous occurrence of two reactions in close proximity at the lymphocyte surface between cell-borne receptors and antigens determinants. By analogy with the proposed molecular mechanism of hemolysis by IgG and complement (Rapp & Borsos, 1970), this condition permits complement to bind to the cell which is then destroyed. Thus, following exposure to a high dose of antigen, an organism is rendered immunologically tolerant to the subsequent exposure to a normally immunogenic dose of antigen.


  The main problem was how self antigenic determinants, even at very low concentrations, could react with newly formed immunologically competent cells in a manner such that those cells with the potential to respond immunologically against self determinants were destroyed. For economy it was desirable that the proposed mechanism should:

  • Resemble the mechanism of induction of high dose tolerance described above.

  • Invoke no properties of a newly-formed cell different from those of a mature cell.

A possible solution was afforded by the postulate that the critical factor affecting cell exposure to an antigens determinant is the determinant/natural antibody concentration ratio.

  Shortly after its generation in a "mutant breeding organ" such as the thymus (Jeme, 1971), an anti-self immunologically competent cell has receptor antibody molecules on its surface, but the rate of secretion of the corresponding natural antibody is insufficient to cause a rapid rise in the concentration of specific humoral buffering capacity. Thus, even for very low concentrations of self determinants, the determinant/natural antibody ratio is very high. This ensures that the exposure of the cell to self determinants is such that there is a high probability of two simultaneous receptor-determinant interactions in close proximity at the cell surface. Complement is then bound to the cell which is destroyed. If the cell does not possess receptors for self antigenic determinants it is not destroyed. Eventually the concentration of the natural antibody molecules which it secretes rises to an equilibrium value so that the cell is buffered against reaction with specific foreign antigens determinants which may enter the organism at subsequent time-points.


  Several features of the theory have gained experimental support. For example:

  • There is now in vitro evidence that high antigen concentrations can specifically inhibit immunologically competent cells (Diener & Armstrong, 1969; Mo1ler & Kashiwagi, 1972); lower antigen concentrations stimulate the cells.

  • There is now evidence that inhibition by high antigen concentrations requires complement both in vivo (Azar, Yunis, Pickering & Good, 1968; Azar & Good, 1971a,b) and in vitro (Forsdyke, 1973b,c).

  • There is also increasing evidence both for the existence of natural antibody (Wilson, Nossal & Lewis, 1972; Vitetta, Grundke-Iqbal, Holmes & Uhr, 1974), and for its ability to buffer cell-borne receptors against antigenic determinants (Forsdyke, 1973a).

  • The theory also suggested a cellular mechanism to explain the change an antibody affinity with time after immunization and this also has gained experimental support (Andersson, 1970; Smith, Hammarstrom & Moller, 1974).

3. Further Implications of the Theory


   Figure 1 shows theoretical curves for the number of immunologically competent cells in an organism against their specificity for a given antigenic determinant. A "high specificity cell" bears antibody receptor sites of high specificity for the determinant, whereas a "low specificity cell" bears antibody receptor sites of low specificity for the determinant. The area under the curves is finite and represents the total number of immunologically competent cells in the organism.

  Curve 1 shows the expected distribution of specificities towards a determinant to which the organism has not yet been exposed. The majority of cells are of very low specificity for a given determinant and a minority are of very high specificity. There is a smooth transition, as shown, between the number of very high specificity cells and the number of very low specificity cells.

  Curve 2 shows the expected distribution of specificities of cells capable of binding with a given self antigenic determinant at an early stage in the immunological maturation of the organism. "High" and "moderate" specificity cells have been eliminated by the complement-dependent mechanism described above. 

  In contrast, the self determinant has not been at a sufficient concentration, relative to natural antibody, to destroy low specificity cells by the same mechanism. However, the determinant has been at a sufficient concentration to stimulate the low specificity cells to proliferate. Thus there is an increased number of low specificity cells against the self determinant and there is not a smooth transition between zero numbers of high and moderate specificity cells and high numbers of very low and zero specificity cells. 

  The expanded clones of low specificity cells are buffered collectivity by the natural antibody they secrete against continued stimulation by self determinants. Thus tolerance against self determinants involves different mechanisms for high and low specificity cells. High specificity cells are eliminated by the complement mechanism. Low specificity cells are buffered against continued reaction with self determinants by natural "blocking" antibody (Wright, Hargreaves, Bansal, Bernstein & Hellstrom, 1973). The natural antibody does not bind with self determinants with sufficient affinity to interfere with the normal function of the determinants.

possel03.gif (47341 bytes)

FIG. 1. The distribution of immunologically competent cells of varying specificities for a given antigenic determinant.
  • Curve 1: distribution prior to exposure of cells to the determinant.
  • Curve 2: distribution of cells reacting with a given self antigenic determinant, at an early stage in the maturation of the immune system.
  • Curve 3: distribution of cells reacting with a given self antigenic determinant at a later stage in the maturation of the immune system.

  At later stages in the immunological maturation of the organism there is a high probability that freshly generated cells bearing receptors of a particular specificity will be partially buffered against reaction with antigenic determinant by natural antibody molecules of varying specificities secreted by cells formed at earlier stages of immunological maturation. This means that high specificity anti-self cells are destroyed less easily by the complement mechanism, whereas low specificity anti-self cells are less easily stimulated to proliferate.

  Curve 3 (Fig. 1) shows the expected distribution of cell specificities against a self determinant under these conditions. With increasing immunological maturation the curve moves progressively to the right until a new cut-off point is reached between cell destruction and cell stimulation. Thus the lack of smoothness in the cell specificity distribution curve at early stages of immunological maturation (curve 2) becomes even more apparent at later stages of immunological maturation (curve 3).


  Figure 2 shows the expected distribution of cell specificities at various stages of the primary immune response to a foreign antigenic determinant. Curve 1 shows the distribution before immunization as in curve 1 of Fig. 1. Curve 2 shows the distribution of specificities after immunization with "weakly immunogenic" (low) concentrations of the antigenic determinant. Since initially there are more low specificity cells than high specificity cells (curve 1) there is a greater probability of cells of low specificity encountering the determinant than cells of higher specificity. Thus low specificity cells are stimulated to proliferate and high specificity cells are neither stimulated to proliferate nor destroyed.

  Although there may be a temporary increase in the total number of immunologically competent cells in an organism in the course of an immune response, homeostatic mechanisms operating randomly on the whole lymphocyte population (Forsdyke, 1973b) tend to keep the total population size constant. Hence when the low specificity cells increase in number, the number of cells of higher and lower specificities tends to decrease; i.e. the number of cells of zero, very low, moderate, high and very high specificities decreases. Since the majority of cells in the total population are of zero and very low specificities, the absolute decrease mainly affects these cell species.

  Curve 3 shows the distribution of specificities at an early stage after immunization with "optimally immunogenic" concentrations of antigenic determinant. Very high specificity cells are destroyed by the complement mechanism. High, moderate and low specificity cells are stimulated to proliferate to varying extents and hence their numbers increase. This increase is only partially compensated for by the loss of the few very high specificity cells, so that homeostatic mechanisms tend to slightly reduce the concentrations of very low and zero specificity cells.

possel04.gif (43878 bytes)

FIG. 2. The distribution of immunologically competent cells of varying specificities for a given not-self antigenic determinant.
  • Curve 1: distribution prior to exposure of cells to the determinant.
  • Curve 2: distribution following primary exposure to a "weakly immunogenic" (low) concentration of determinant.
  • Curve 3: distribution shortly after primary exposure to an "optimally immunogenic" concentration of determinant.
  • Curve 4: distribution after maturation of the primary immune response to an "optimally immunogenic" concentration of determinant.

  Curve 4 illustrates the maturation of the immune response with time after immunization. Compared with Curve 3, the curve shifts to the right along the abscissa indicating the appearance of new very high specificity cells. These arise de novo from the "mutant breeding organ", but are not destroyed by remaining antigenic determinants for two reasons.

  • (i) The concentration of free antigenic determinants is declining.

  • (ii) The antibody secreted in the course of the immune response increases the buffering of cells against reaction with antigenic determinants.

Thus the very high specificity cells destroyed during the initial encounter with the antigenic determinant are replaced by fresh cells of very high specificity able to survive in an environment of decreasing antigen concentration and increasing concentration of buffering antibody (i.e. a decreasing determinant/antibody concentration ratio). These same conditions favour the continued stimulation of moderate and high specificity cells and discourage the proliferation of low specificity cells. The de novo generation of high specificity cells might be of particular importance for the maturation of the immune response of a very small organism with a low total lymphocyte population.

 Following the administration of a "tolerogenic" (high) concentration of antigenic determinant the moderate and high specificity cells are destroyed while low specificity cells are stimulated, so that curves similar to curves 2 and 3 in Fig. 1 are obtained at various stages of the response.

4. Discussion


  It follows from the above analysis that higher organisms should possess expanded clones of immunologically competent cells with receptors of low specificity for self determinants (Fig. 1). These self determinants are more likely to resemble the self determinants of other members of the same species than the self determinants of other species. Indeed, a cell of low specificity for self determinants may be of high specificity for the determinants of other members of the same species (i.e. a cell may be of high specificity for "nearself" antigenic determinants). This would explain why there is greater immunological responsiveness between non-identical members of the same species than between members of different species ("alloaggression"; Jerne, 1971).

  The expanded clones of low specificity anti-self cells may play a primary role in the immune response against foreign pathogenic organisms which attempt to confuse the immune system of the host by modifying their surface determinants so that they come to resemble those of the host (Snell, 1968). This type of adaptation will only tend to be successful if the pathogen adapts so that its determinants are identical with the host's determinants. If the pathogen adapts only partially then the expanded clones of low specificity anti-self cells (high specificity anti-"near-self"), will eliminate the pathogen by normal immunological mechanisms.

  The adaptation of the surface determinants of a pathogenic organism may reflect a change in genotype or a change in phenotype alone. A genetic change in the pathogen will be acted upon by selective forces in the environment which will include the host's clones of anti-"near-self" cells. These will act as a barrier preventing the evolution of the surface determinants of the pathogen into forms progressively more and more like those of the host. Only a quantal, one step, adaptation of the pathogen from complete foreignness to complete resemblance with the host, would be successful. Thus the expanded clones of anti-"near-self" cells would tend to stabilize the surface determinants of a pathogen relative to those of its host.

  An example of a phenotypic adaptation of the surface determinants of a pathogenic organism would be the inclusion of cell surface components from a previous host into the surface of the pathogen. In the case of a species specific pathogen, cell surface components from a previous host are likely to possess "near-self" determinants for the next host, particularly if the surface components show a high degree of polymorphism (Bodmer, 1972). In this situation, the rejection of the pathogen by the host's immune system would be a further example of the "alloaggression" phenomenon.

  In that a diseased, effete or tumor cell might also possess modified surface determinants with a "near-self" conformation, then the same mechanism could play a role in the elimination of such cells (Rosenberg & Rogentine, 1972).


  Following exposure of immunologically competent cells to a given self or not-self antigenic determinant, a sharp cut-off point is established in the distribution curve for cells of varying specificities for the determinant (Figs 1 and 2). To the right of the cut-off point there are no cells. Immediately to the left of the cut-off point there is an increased number of cells. These are the cells of highest specificity against the determinant which the organism possesses.

  The positions of the cut-off points for particular antigenic determinants may vary between individual organisms of a species, between individual strains of a species and between species. Several factors, which may be individual, strain or species specific, may be identified as likely to influence the positions of cut-off points and hence the qualitative and quantitative characteristics of the total population of immunologically competent cells in an organism. These characteristics, in turn, will influence the "responder" versus " non-responder" status of the individual, strain or species, towards a particular antigen. The factors include the following:

  • (i) Antibody. If initial exposure of a cell to self or non-self antigenic determinants is dependent on the determinant/antibody concentration ratio, then variables affecting the qualitative and quantitative characteristics of the total antibody population prior to such exposure, will be of importance. These variables will include: (a) the previous immunological history of the individual organism, (b) the characteristics of the organism's spectrum of self determinants, and (c) the homeostatic mechanisms affecting the total immunoglobulin concentration.

  • (ii) Complement and complement inhibitors. Variations in the concentrations and activities of rate-limiting complement components and of complement inhibitors will affect the cut-off point by modulating the complement dependent destruction of cells by high antigen concentrations. In this respect it is of interest to note recent evidence that a genetic locus influencing the concentration of a complement component maps near an immune response locus in mice (Demant, Capkova, Ninzova & Voracova, 1973). The view is growing (Fisher, 1958; Bodmer, 1972), that one reason for the close linkage of such genes might be a close functional relationship between the gene products.

  • (iii) Cell surface. The characteristics of the surface of an immunologically competent cell will influence such factors as receptor mobility, which in turn will influence the ease with which two receptor-determinant interactions of close proximity can occur (for complement to bind). Thus immune response loci would be associated with loci for cell surface determinants (Benecerraf & McDevitt, 1972).

  • (iv) Drugs. The position of the cut-off point would also be influenced by drugs which affect either of the above three factors. For example, recent studies in a system involving chloroquine and the antigen-analogue concanavalin-A (Forsdyke, 1974), indicate that this membrane-active drug causes the cut-off point to move to the left, thus favouring the destruction of anti-self immunologically competent cells. This could explain the therapeutic benefit of chloroquine in certain autoimmune diseases (Popert, Meijers, Sharp & Bier, 1961).


    If immunologically competent cells of particular specificities are generated randomly in the "mutant breeding organ" (Jerne, 1971), then there is a greater probability that during the immunological maturation of an organism a cell of low specificity towards a particular determinant will be generated before a cell of high specificity towards the determinant. 

  Thus an immune response which required high specificity cells would be more likely to be impaired by the early removal of the "mutant breeding organ" than immune responses requiring cells of lower specificities. For example, if cell-mediated immune responses require higher specificity cells than humoral immune responses (Levine, 1965), then cell-mediated immune responses would be more susceptible to inhibition by removing the "mutant breeding organ" at an early stage than humoral immune responses (Miller, 1963). 

  By the same token, an antigen presented to an immunologically immature organism in a manner designed to preferentially stimulate a cell-mediated response, would be likely to have fewer effective immunogenic determinant groups than the same antigen presented in a manner designed to stimulate the humoral immune response.

            5. An Overview

The importance to progress in immunology of a continuing and critical interplay between experiment and theory has recently been stressed by Nossal (1972) in a historical overview of the possible scope of future developments in immunology. The theory, the further implications of which have been discussed here, attempted to describe certain key features of the immune process in terms of interactions between a limited number of elements within a highly simplified model organism. The theory did not directly address itself to many of the phenomena currently interesting immunologists, such as:

  • "T" and "B" cell cooperation (Smith et al, 1974).

  • The role of the macrophage (Diener, Shortman & Russell, 1970).

However, the limited success of the theory in predicting the results of subsequent experiments is encouraging. Certain features of the theory are shared with several other elaborations of the clonal selection theory which have appeared (Eisen, 1966; Moller, 1968; Siskind & Benecerraf, 1969; Cohen, 1970, 1971; Bell, 1970a,b), and this is also a source of encouragement.

  It is our belief that when a final description is given of the immune process, some of the complexities which confront us today in our experimental systems will be seen as the result of two main causes:

  1. The great sensitivity of some of the assays now in use means that a variety of "technical" factors are confusing the interpretation of experiments; some examples of this are given elsewhere (Forsdyke, 1973d; Milthorp & Forsdyke, 1973a,b).

  2. Our systems, derived of necessity from higher organisms, are not solely the sites of immunological processes, but the sites of interactions between immunological processes, metabolic processes under hormonal influence and the processes concerned with overall cell population size homeostasis.

The author thanks the Medical Research Council of Canada for support of the experimental work referred to in this paper.


ANDERSSON, B. (I 970). J. exp. Med. 132, 77.
AZAR, M. M. & GOOD, R. A. (1971a). J. Immun. 106, 241.
AZAR, M. M. & GOOD, R. A. (1 971 b). Proc. Soc. exp. Biol., N.Y. 137, 429.
AZAR, M. M., YUNIS, E. J., PICKERING, R. J. & GOOD, R. A. (1968). Lancet i, 1279.
BELL, G. I. (1970a). Nature, Lond. 228, 739.
BELL, G. I. (1970b). J. theor. Biol. 29,191.
BENECERRAF, B. & MCDEVITT, H. 0. (1972). Science, N.Y. 175, 273.
BODMER, W. F. (1972). Nature, Lond. 237, 139.
BURNET, F, M. (1959). The Clonal Selection Theory of Acquired Immunity. Cambridge:Cambridge University Press.
COHEN, S. (1970). J. theor. Biol. 27, 19.
COHEN, S. (1971). Immunochem. 8, 1099.
DEMANT, P., CAPKOVA, J., NINZOVA, E. & VORACOVA, B. (1973). Proc. natn. Acad. Sci. U.S.A. 70, 863.
DIENER, E. & ARMSTRONG, W. D. (1969). J. exp. Med. 129, 591.
DIENER, E., SHORTMAN, K. & RUSSELL, P. (1970). Nature, Lond. 225, 731.
EISEN, H. N. (1966). Cancer Res. 26, 2005.
FISHER, R. A. (1958). The Genetical Theory of Natural Selection, 2nd edn. New York:Dover.
FORSDYKE, D. R. (1966). In The Biological Effect of Phytohaemagglutinin (M. W. Elves,ed.), p. 115. Oswestry, England: R. Jones and A. Hunt, Orthopaedic Hospital Management Committee.
FORSDYKE, D. R. (1967). Ph.D. thesis: University of Cambridge.
FORSDYKE, D. R. (1968). Lancet i, 281.
FORSDYKE, D. R. (1969). J. theor. Biol. 25, 173.
FORSDYKE, D. R. (1973a). Immunology 25, 583.
FORSDYKE, D. R. (1973b). Immunology 25, 597.
FORSDYKE, D. R. (1973c). Immunology 25, 613.
FORSDYKE, D. R. (1973d). Expl Cell Res. 77, 216.
FORSDYKE, D. R. (1974). Proc. Can. Fedn Biol. Socs 17, 92.
JERNE, N. K. (1971). Eur. J. Immun. 1, 1.
LEVINE, B. B. (1965). J. exp. Med. 121, 873.
MILLER, J. F. A. P. (1963). Br. med. J. ii, 459.
MILTHORP, P. & FORSDYKE, D. R. (1973a). J. Immun. Methods 2, 269.
MILTHORP, P. & FORSDYKE, D. R. (1973b). Biochem. J. 132, 803.
MOLLER, G. (1968). J. exp. Med. 127, 291.
MOLLER, G. & KASHIWAGI, N. (I 972). Immunology 22, 441.
NOSSAL, G. J. V. (1972). Cell. Immun. 4,109.
POPERT, A. J., MEIJERS, K. A. E., SHARP, J. & BIER, F. (1961). Ann. Rheum. Dis. 20,18.
RAPP, H. J. & BORSOS, T. (1970). Molecular Basis of Complement Action. New York: Appleton-Century-Crofts.
ROSENBERG, S. A. & ROGENTINE, G. N. (1972). Nature New Biol. 239, 203.
SISKIND, G. W. & BENECERRAF, B. (1969). Adv. Immun. 10, 1.
SMITH, G. D., HAMMARSTROM, L. & MOLLER, E. (1974). Scand. J. Immun. 3, 61.
SNELL, G. D. (1968). Folia Biol. 14, 335.
VITETTA, E. S., GRUNDKE-IQBAL, I., HOLMES, K. V. & UHR, J. W. (1974). J. exp. Med. 139, 862.
WILSON, J. D., NOSSAL, G. J. V. & LEWIS, J. (I 972). Eur. J. Immun. 2, 225.
WRIGHT, P. W., HARGREAVES, R. E., BANSAL, S. C., BERNSTEIN, I. D. & HELLSTROM, K. E. (1973). Proc. natn. Acad. Sci. U.S.A. 70, 2539.


End Note on the Differential Avidity/Affinity Model

The above paper treated lymphoid cells generically without a distinction between B and T lymphocytes, or a consideration of the role of the macrophage (which was poorly understood at that time). The model has come to be known as the differential affinity/avidity model of the development of the immunological repertoire, and applies to both B and T lymphocytes, which may be located either "centrally" (in a mutant-breeding and educating organ), or peripherally. 

  Cell surface antigens were also treated generically, although MHC involvement was suggested ("particularly if the surface components show a high degree of polymorphism (Bodmer, 1972)". -- "Thus immune response loci would be associated with loci for cell surface determinants (Benecerraf & McDevitt, 1972)". Association of antigen with MHC was invoked in the discussion (p. 195), but not in a way that can now be viewed as prescient.

   In December 1973 Zinkernagel & Doherty submitted a paper to Nature which, from a quite different perspective, reached some similar conclusions. Instead of the term "near self", which emphasized how close an antigen might be to self, the authors used the term "altered self", which emphasized the fact of a difference from self. Extending earlier work on MHC-restriction of B and T cell cooperation, they showed that T cell-mediated cytotoxicity was also MHC-restricted (Zinkernagel & Doherty 1974, 1997).

   Under my model (derived from the first principles set out in the above Lancet paper) MHC-restriction would be an automatic consequence of the selection of anti-near-self cells during their initial maturation (education). Thus, the ideas of positive selection and of its mechanism came as one conceptual package. The need to take into account total population-size homeostasis was emphasized, so that the increase in a population of one affinity would require proportionate decreases in some other populations.

   Although my 1975 paper invoked much circumstantial evidence, the first definitive experimental evidence generally acknowledged as supporting positive selection came in 1977 (see Jameson et al. 1995).The differential affinity/avidity model was independently suggested by Sprent & Webb (1987), and although for some time opposed by the "peptide" model (Marrack & Kappler, 1987), it eventually became widely accepted as applying generically to both T and B lymphocytes (Von Boehmer, 1994; Hayajawa et al. 1999; Detours & Perelson 1999; Gaudin et al. 2004; Cancro & Kearney, 2004). Janeway (2001) concluded:

"Thus, both the mature, naive T cell repertoire and the mature, naive B cell repertoire are generated by interaction with self-ligands rather than non-self ligands. These self ligands can signal B and T lymphocytes to mature and to survive, ... ."

  My 1975 paper also defined what later unfortunately became known as "death-by-neglect". The paper invoked homeostatic mechanisms operating randomly on the whole lymphocyte population that would tend to keep the total population size constant. Hence, when low specificity cells increased in number, the number of cells of higher and lower specificities would decrease. Since the majority of cells in the total population were of zero and very low specificities, the absolute decrease mainly affected these cell species. Thus death-by-neglect (implying passivity) was not a good description since, although poorly defined, active homeostatic processes were involved. 

  The principle of cell population homeostasis was further invoked in 1991 (Forsdyke, 1991; Click Here).

"The 'education' of T-lymphocytes involves both positive and negative selection (7, 8). Positive selection generates sets of T-lymphocytes with the potential to respond to various 'self' determinants (e.g. MHC, CDI, TI and Qa-1 antigens; 9-11). Negative selection eliminates cells responding to self with high specificity. The final immunological repertoire consists of numerous small clones of cells. Members of a particular clone are each capable of recognizing a particular set of 'nearself' antigenic determinants with varying degrees of specificity.

    The range of specific responsiveness exhibited by an individual reflects the outcome of these selection processes (and further positive selections by foreign antigens), over many years. To renew the educated T-lymphocyte population after depletion (perhaps due to haemorrhage), could be a protracted process if renewal required reeducation. Individual T-cells (end cells), rather than stem cells, should be responsive to homeostatic control mechanisms affecting the size of the total T-lymphocyte population (7).

   Thus peripheral immunologically-competent clones of T-cells should be responsive not only to the cues provided by foreign antigenic determinants (through the determinant- specific T-cell antigen receptor) but also to cues provided by the growth factors concerned with T-lymphocyte population size homeostasis (through appropriate receptors)."

Later work recognized that, in addition to the latter growth factors, self antigens have a role "not only to select a receptor repertoire in the thymus, but also to keep naive T cells alive and 'ready for action' in the periphery" (Goldrath & Bevan 1999; Ernst et al. 1999). However, employment of the term "naive" for cells which have been well educated in the mutant-breeding and educating organ (thymus in the case of T cells and some B cells; Akashi et al. 2000), is unfortunate.

  The need to know "the underlying biological rationale" of lymphocyte positive selection was stressed by Cancro & Kierney (2004), whose ideas seemed to favour the model advanced in my 1975 paper that anti-'near-self' cells "Provide a barrier opposing the progressive evolution of the surface determinants of a pathogen into forms identical with the surface determinants of its host." For more on the relative merits of the terms "alternative self" and "near-self" see Forsdyke (2005) (Click Here).

     Many immunologists in the 1970s (e.g. in Melbourne, Basel, Denver) approached the idea of positive selection solely from the perspective of T-lymphocytes, rather than generically (B and T lymphocytes) as here. Consequently, a somewhat different history of the discovery of positive selection is usually given (see refs. below). However, there remains general agreement (Bevan et al. 1994) that:

"Imprinting on self produces a circulating T cell population  that is poised to respond to pathogen-modified self."

For more see Lewin's "Great Experiments in Biology" where Harald von Boehmer gives his account.

Donald Forsdyke


Akashi, K., Richie, L. I., Miyamoto, T., Carr, W. H. & Weissman, I. L. (2000) B lymphopoiesis in the thymus. J. Immun. 164, 5221-26.

Benecerraf, B. & McDevitt, H. 0. (1972) Histocompatibility-linked immune response genes Science 175, 273-279.

Bevan, M. J., Hogquist, K. A. & Jameson, S. C. (1994) Selecting the T cell receptor repertoire. Science 264, 796-797.

Bodmer, W. F. (1972) Nature 237, 139-145. Evolutionary significance of the HLA system.

Cancro, M. P. & Kearney, J. F. (2004) J. Immunol.173, 15-19. B cell positive selection: road map to the primary repertoire?

Detours, V. & Perelson, A. S. (1999) Explaining alloreactivity as a quantitative consequence of affinity-driven thymocyte selection. Proc. Natl. Acad. Sci. USA 96, 5153-5158.

Ernst, B., Lee, D-S., Chang, J. M., Sprent, J. & Surh, C. D. (1999) The peptide ligands mediating positive selection in the thymus control T cell survival and homeostatic proliferation in the periphery. Immunity 11, 173-181.

Goldrath, A. W. & Bevan, M. J. (1999) Selecting and maintaining a diverse T-cell repertoire. Nature 402, 255-262. Low-affinity ligands for the TCR drive proliferation of mature CD8+ cells in lymphopenic hosts. Immunity 11, 183-190.

Gaudin, E., Hao, Y., Rosado, M. M., Chaby, R., Girard, R. & Freitas, A. A. (2004) Positive selection of B cells expressing low densities of self-reactive BCRs. J. Exp. Med. 199, 843-853.

Hayakawa, K., Asano, M., Shinton, S. A., Gui, M., Allman, D., Stewart, C. L., Silver, J. & Hardy, R. R. (1999) Positive selection of natural autoreactive B cells. Science 285, 113-116.

Huseby, E., Kappler, J. & Marrack, P. (2004) TCR-MHC/peptide interactions: kissing cousins or a shotgun wedding? Eur. J. Immunol. 34, 1243-50.

Jameson, S. C., Hogquist, K. A. & Bevan, M. J. (1995) Annu. Rev. Immunol. 13, 93-126.

Janeway, C. A. (2001) How the immune system works to protect the host from infection: a personal view. Proc. Natl. Acad. Sci. USA 98, 7461-7468.

Jerne, N. K. (1970) Generation of antibody diversity and self tolerance - a new theory. Immune Surveillance. Proceedings of a International Conference Held at Brook Lodge, Augusta, Michigan. Edited by R. T. Smith & M. Landy. Academic Press, New York. pp. 343-363.

Marrack, P. & Kappler, J. (1987) Science 238, 1073-1079. The T cell receptor.

Sprent, J. & Webb, S. R. (1987) Adv. Immunol. 41, 39-133. Function and specificity of T cell subsets in the mouse.

Von Boehmer, H. (1994) Cell 76, 219-228. Positive selection of lymphocytes.

Zinkernagel, R. M. & Doherty (1974) Nature 248, 701-702. Restriction of in vitro T cell-mediated cytotoxicity in lymphocytic choriomeningitis with a syngeneic or semiallogeneic system.

Zinkernagel, R. M. & Doherty (1997) Immun. Today 18, 14-17. The discovery of MHC restriction.

Next for: Intracellular Self/Not-self Discrimination (Click Here)

Go to: "Alternative Self" or "Near Self" (Click Here)

Go to: Historical Overview 2012 (Click Here)

Return to: Theoretical Immunology index (Click Here)

Return to: Home-Page (Click Here)

This page was established circa 1999, and last edited on 23 October 2014, by Donald Forsdyke. For previous versions of this page see http://www.archive.org