The clonal selection theory

From the early 1940s, Burnet had attempted to develop a comprehensive, cohesive and concise view of the workings of the immune system—a theory of everything immunological. This effort resulted in the publication of a series of papers and books on the subject, including his two popular monographs titled The Production of Antibodies^{2, 3} and culminated in his drafting of the clonal selection theory, which was introduced in a brief monograph⁴ and subsequently explored in some detail in a published series of lectures presented to the Vanderbilt University in 1958.⁵ The development and implications of this model have been discussed in detail elsewhere, including by one of us.⁶

The clonal selection theory as published in 'A modification of Jerne's theory of antibody production using the concept of clonal selection' in the Australian Journal of Science⁴ can be presented in an abridged form (following that presented in reference⁶): Among [antibodies] are molecules that can correspond probably with varying degrees of precision to all, or virtually all, the antigenic determinants that occur in biological material other than that characteristic of the body itself. Each type of pattern is a specific product of a clone of [lymphocytes] and it is the essence of the hypothesis that each cell automatically has available on its surface representative reactive sites equivalent to those of the globulin they produce... It is assumed that when an antigen enters the blood or tissue fluids it will attach to the surface of any lymphocyte carrying reactive sites which correspond to one of its antigenic determinants... It is postulated that when antigen-[antibody] contact takes place on the surface of a lymphocyte the cell is activated to settle in an appropriate tissue... and there undergo proliferation to produce a variety of descendents. In this way, preferential proliferation will be initiated of all those clones whose reactive sites correspond to the antigenic determinants on the antigen used. The descendents will [be] capable of active liberation of soluble antibody and lymphocytes which can fulfil the same functions as the parental forms.⁴

In his original monograph, Burnet⁴ acknowledges Talmage's priority in the idea that cells were the replicating elements responsible for the exponential expansion of antibody production. Despite this, it is difficult to find in Talmage's original proposal⁷ so explicit an outline of a cohesive theory. Perhaps the closest passage is the following text: As a working hypothesis it is tempting to consider that one of the multiplying units in the antibody response is the cell itself. According to this hypothesis, only those cells are selected for multiplication whose synthesised product has affinity for the antigen injected. This would have the disadvantage of requiring a different species of cell for each species of protein produced, but would not increase the total amount of configurational information required of the hereditary process.⁷

Burnet agreed with this requirement, writing, '... any tenable form of [the] theory must involve the existence of multiple clones of globulin-producing cells, each responsible for one genetically determined type of antibody globulin'.⁵

The clonal selection theory raised two mechanistic issues and these were the principle sources of opposition over the first decade after publication: 'A major implication concerned the genetic processes which occurred in the development of the lymphocyte. There was no precedent not only for the randomization process involved in the generation of all possible antibody specificities but also in the proposal that an individual cell made antibodies of a single specificity'.⁸

A problem of Landsteinerian proportions

The view that antibody specificity arose, not from being cast to complement the inducing antigen, but by being selected from individual specificities within a pre-existing repertoire, raised the problem of how the genome could encode an apparently infinite number of specificities. Over a 15-year period, Landsteiner and Chase⁹ had raised antibodies against almost any chemical structure—even synthetic structures such as picryl chloride and 2,4-dinitroluorobenzene, providing they were first conjugated to a protein and then injected in the presence of killed tubercle bacilli. To put the problem another way, if an organism can generate antibodies against almost any protein derived from any other organism, how can its genome encode for more individual antibody molecules than the total of all its proteins?

Talmage's quantitative approach

Talmage's¹⁰ quantitative approach to immunology was imprinted during his training in Frank Dixon's laboratory in the early 1950s. There, he studied antibody responses by quantifying radiolabelled antigen in plasma following inoculation into naive, primed and irradiated rabbits.^{11, 12, 13}

His quantitative approach to the problem of antibody specificity led to his beliefs that 'specificity is probably nothing more than a high degree of affinity'⁷ and that formation of a stable antigen/antibody complex 'does not require a perfect fit between two complementary configurations',¹⁴ as well as his acceptance of Haurowitz's¹⁵ view that that many antibodies are cross-reactive. These views are, of course, still acceptable. Similarly, few would have problem with his view that the specificity of an antiserum is likely to result from the combinatorial activities of antibodies of multiple specificities, each defining separate physical characteristics of the antigenic epitope, such that 'the dominant reactivity will be that common to the largest number of molecules present' and 'the specificity of an antiserum containing a mixture of different globulin molecules is likely to be very much greater than that of an antiserum in which all molecules are exactly alike'.¹⁴ Based on these premises, as well as estimates of the number of combinations required to account for the almost unlimited distinguishing ability of immune sera, the proportion of these configurations to which a single antibody could bind, and the number of different antibody molecules in an immune serum, Talmage developed a quantitative model of immune specificity. He proposed: If each globulin is so constructed that its reactions... are restricted to approximately one percent of all possible antigenic configurations, and there are 5000 molecular types, each antigen will react with approximately 50 different globulin molecules. In this case there would be ... approximately 3 10¹²⁰ different combinations and an equal number of antigens could be distinguished.¹⁴

His approach is well illustrated by Figure 2 in the same manuscript¹⁴ in which he plots the relationship between antibody concentration, the energy of combination of antibody and antigen, and the threshold required for detection. From a quantitative point of view, Talmage's model was elegant. It conformed to the principle of Ockham's Razor in that it required no hither-to-unforeseen structures or mechanisms, but merely emerged from a numerical approach to known facts. With the benefit of hindsight, his estimate of the cross reactivity of antibody molecules was probably unduly affected by studies of natural antibody, in contrast to immune sera raised against, for example, viral epitopes.

Burnet's qualitative approach

Burnet received an Australian medical education, which probably more than any other favours a qualitative approach. At the most superficial of levels, a patient is either sick or well. Clinical signs are mostly recorded as being present or absent, and even quantitative data, such as electrolyte levels tend to be regarded as either within, or outside, the normal range (never mind that 5% of the normal population lies outside this range). While Talmage also benefited from a medical education, his was a postgraduate degree following a thorough grounding in basic science, as is common in the USA.

Burnet's immunological writing was rarely quantitative. His explanations tended to rely on analogies of concrete experiences. In Enzyme Antigen and Virus¹⁶ he compared his instructive model of antibody formation to 'a mould made of fusible alloy which is used to make a Plaster of Paris cast'.¹⁶ Even the name of the clonal selection theory stems from his experience of pioneering studies culturing bacteria to characterize bacteriophage.¹⁷ In his original description of the clonal selection theory, he made no attempt to quantify the phenomena he was discussing; indeed, the only quantification he made was of the three theories of antibody production discussed in his first sentence. Instead, he glossed over details with the comment, 'Such a point of view is basically an attempt to apply the concept of population genetics to the clones of mesenchymal cells within the body. It is clear that the internal environment is an exceedingly complex one and in all probability many factors will impinge on clones of antibody-producing cells from that environment'.⁴

Without a quantitative basis, it was difficult to devise an experiment to falsify the theory. In personal correspondence (dated 14 May 2007), Nossal said, 'Burnet seemed to have no idea of how to set about testing his hypothesis.'

One cell one antibody

Even if Burnet did not know how to test the clonal selection theory, his PhD student Gustav Nossal did: 'When Burnet's clonal selection theory was presented to me as it went off to the Australian Journal of Science, I immediately realised that, in the best Popperian fashion, I could falsify the theory if a single antibody-forming cell from a doubly immunised animal managed to make both antibodies' (Nossal, personal correspondence dated 14 May 2007). Nossal's experiment depended critically on the contributions of Joshua Lederberg, who visited Burnet's laboratory for a 3-month period in 1957 and taught Nossal the techniques of single cell manipulation and culture.

The experiment by Nossal and Lederberg involved the immunization of rats with flagella of two serotypes of Salmonellae and single cell manipulation into microdrops into which live bacteria could be introduced in order to observe immobilization. Nossal's first publication on the subject,¹⁸ reported studies of so few cells that no conclusion could be drawn regarding the number of potential specificities of B cells (Table 1). A subsequent publication using the approach¹⁹ contained data that were highly significant (Table 1).

Table 1 - Frequencies (f) of antibody producing cells of two different specificities, S1 and S2, total numbers of cells examined and the observed and expected numbers of double producers.

Full table

Lederberg's visit prompted him to think deeply on the clonal selection theory and to develop his own formulation of the theory. This was published as nine propositions in Science,²⁶ printed immediately following Talmage's quantitative analysis of the theory,¹⁴ and contains two insightful alterations, one critical to making the theory acceptable today, and the other correct, but distracting. The first insight was that he proposed 'the genetic diversity of the precursors of antibody-forming cells arises from a high rate of spontaneous mutation during their life long proliferation'.²⁶ (our italics) This change brings the theory into accord with our current understanding of the life-long production of lymphocytes that undergo a period of tolerance induction early in the cells' development, not necessarily early in the organism's development. '[This idea] diverges from Burnet's proposal that the 'randomisation' 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,' he wrote. Burnet did not appear to adopt this critical insight, which was capable of explaining why experimentally induced neonatal tolerance to a nonproliferating,^{27, 28, 29} or surgically removed³⁰ antigen only lasted a few weeks, while that to proliferating antigens was life long.³¹ One can only speculate as to the reasons for Burnet's failure in this regard, but it must have been obvious to him that a Nobel Prize was in the wind and that it would be for his work on neonatal tolerance. The second insight Lederberg contributed was the observation that as a somatic diploid cell, a lymphocyte 'should be capable of at most two potentialities for antibody formation, one for each chromosome'.²⁶ Burnet was impressed by this logic.

In the few years following Nossal and Lederberg's¹⁹ experiment, a number of other groups attempted to identify double antibody producing cells. Many of these studies were flawed. White,³² for example, immunized rabbits with diphtheria toxoid and ovalbumin and identified B-cell specificity using fluoresceinated antigens on spleen sections. No data are quantified and his figure illustrates perhaps a few dozen cells. Although the spleens of hyperimmunized animals were examined, which one would imagine would contain numerous Fc receptor bearing cells capable of binding both the antibody specificities circulating, White claimed no double staining cells. More quantitative approaches were later taken by Hiramoto and Hamlin³³ and Green et al.,³⁴ but they too relied on direct or indirect immunofluorescence of splenocytes, identifying not only cells producing specific Ig, but also those bearing Fc receptors.

Attardi et al.²⁴ immunized rabbits with Escherichia coli bacteriophage, and tested neutralization of activity after immersion in microdrops containing single cells. The number of B cells that produced antibody capable of neutralizing two phage was highly significantly greater than that expected by chance (Table 1)—exactly the opposite result from that of Nossal and Lederberg.¹⁹ The conflict between the two teams had not been resolved by the time Burnet delivered his acceptance speech for his Nobel Prize in physiology and medicine, which was awarded in December, 1960. In his speech, he reiterated the general value of the theory, but had started to retreat in matters of detail: At the present time I believe there is very little doubt among immunologists that some form of selective theory is needed... We can picture clones of cells arising which carry the capacity to synthesize, under appropriate stimulus, one, two or more specific patterns which, either as a cell receptor or as the active patch on an antibody molecule, could react each with a specific antigenic determinant... If a cell or clone is limited to one or two patterns, then it is practical to postulate that any clone carrying either one or two self-reactive patterns is eliminated... This is the form taken by the clonal selection theory, and provided two is adopted as the usual number of patterns for a diploid somatic cell, it provides a reasonable interpretation of the facts.³⁵ (our italics)

Over the subsequent 4 years, Nossal together with Mäkelä embarked on a heroic analysis of over 3000 single cells, rarely identifying a cell-secreting antibody reactive with two bacterial strains³⁶ (Table 1). Nossal attributes the conflicting results of Attardi, Cohn and Lennox and the other members of the collaboration,^{24, 25} to their failure to wash their cells adequately.³⁷ In contrast, Cohn believes that the low level of dual specific cells identified in their experiments reflect the real existence of a small proportion of T cells that have failed allelic exclusion: I believe today both on experimental and conceptual grounds that our findings were absolutely correct. Nevertheless, our failure to find perfect 'unispecific clonality' was the most costly result of our scientific careers.³⁸

The issue of one cell/one antibody separates the selection models proposed by Burnet and Talmage. While Burnet's clonal selection theory doesn't depend entirely on this outcome, it is certainly more elegant for it being true. In contrast, the issue is irrelevant to Talmage's model; highly cross-reactive antibodies would be expected to bind to multiple antigens.

Dark times for clonal selection

A major point in favour of the clonal selection theory was its ability to explain self/nonself discrimination,³⁹ something that Burnet felt was not covered by the instructive theories of antibody production.⁴⁰ By the early 1960s, work on immunological tolerance and autoimmunity was largely interpreted from the perspective of the clonal selection theory. Certainly, Burnet's own experimental interests became focussed on the autoimmune diseases, in particular those of the NZB and NZB/NZW F1 hybrid mice. Because it offered a simple paradigm, the clonal selection theory also made experimental immunology more accessible to newcomers who would otherwise have been dissuaded by its complexity and obtuse jargon. As a result, Burnet was exposed to a wide range of experimental results, not always produced under optimum conditions or with requisite skill. Burnet was not a man to think well on his feet, and did not always respond incisively to issues raised at conferences.⁴¹

For example, at the Sixteenth Annual Symposium on Fundamental Cancer Research at the University of Texas in 1962, Trentin and Fahlberg⁴² presented a model purported to test the clonal selection theory, in which lethally irradiated mice were rescued by transfusion of limiting numbers of bone marrow cells. The nodules of haemopoiesis resulting were picked as 'single clones' and used to repopulate additional lethally irradiated mice. The second and third generation of recipients were found to be able to make antibodies of multiple specificities. Even though it was already known, and the authors also reported, that transfer of lymphoid colonies did not rescue the lethally irradiated mice from death, and that the transferred cells had to be pluripotent for the mice to survive, Burnet interpreted the results as disproving his theory—at least in its original form: 'I agree entirely. This blows out the original clonal selection theory'.⁴⁰ His reasons for so easily accepting this experimental result appear to relate to his general approach to science as well as his newly formed research interest in autoimmunity. Nossal⁴³ has described Burnet's 'ready willingness to accept as 'probably true' new findings that are not yet proven or universally accepted... He embraces new data as just so much grist to the mill.' Furthermore, Burnet's studies of autoimmunity had led him to collect histopathological photomicrographs of germinal centre-like collections in affected organs and thymi of individuals afflicted by myasthenia gravis, rheumatoid arthritis, thyroiditis, lupus and hepatitis. He interpreted the epithelioid collections at face value and attributed them to an epithelial origin of lymphocytes, stating, '... there can be a reversion of lymphocytes to epithelial cells and redifferentiation of epithelial cells to lymphocytes with immunological competence'.⁴⁴ This view then attributed lymphocytes to the differentiation of ectoderm (or perhaps endoderm) instead of mesoderm, from which the rest of the haemopoietic system developed. Trentin and Fahlberg's model⁴⁵ was subsequently published, by which time, Feldman and Mekori⁴⁶ had published a similar study of in vivo analysis of splenic colony formation in irradiated recipients.

The view that the clonal selection theory was an appropriate framework for considering tolerance was far from universal. Haurowitz was still eloquently defending his template theory on the basis of his original argument, that 'It seems... difficult to believe that the body should contain preformed antibodies against azophenylarsonate, azophenyltrimethylammonium ions and other artefacts of the chemical laboratory'.⁴⁷ Burnet acknowledged, 'The most widely held view among working immunologists [is] that the problem of tolerance has been overstated and that the findings can be simply interpreted on a basis of (a) the immunologic nonreactivity of the embryo, and (b) the existence of immunologic paralysis by any continuing excess of antigen'.⁴⁰

Perhaps the greatest concern for Burnet was obtained using the Simonsen phenomenon, in which the intravenous inoculation of adult chicken leucocytes into a chick embryo induced proliferative cellular lesions in the chick spleen and thymus.⁴⁸ His problem was that too high a proportion of lymphocytes were able to initiate foci in allogeneic transfers. Szenberg and Warner⁴⁹ reported that as many as 1:40 large lymphocytes could produce foci, although the median value was usually in the range of 1:1000 to 1:2500. 'Clearly', he wrote, 'We cannot assume that the possible antigenic patterns to be found in another chick embryo represent two percent of all the possible patterns of foreign antigen'.⁴⁰ This troubled him greatly because it was the Simonsen phenomenon that had impressed the cellular aspect of immunity on him inspiring the clonal selection theory, and because Szenberg and Warner worked within his own institute.

To explain these results, Burnet proposed an extreme modification to the theory. He suggested 'the elaborate pattern of antibodies which we can work with is based on development from a few inborn patterns on which the recognition and tolerance phenomena are based... It seems that what is needed to bring a clonal selection theory back into touch with experimental realities is an acceptance of wide (genetically based) mutability of a few basic [receptor] patterns... Modification of antibody pattern to fit the circumstances, the central feature of immunology, can now be based on high genetic liability of some or all regions of the genome which are concerned in determining antibody patterns'.⁴⁰

In discussion with Lesley Brent at the cancer symposium in 1962, Burnet admitted: 'Frankly, I would not now support the original clonal selection theory that I put forth 3 years ago, but I still believe that it was a very useful start to the development of this concept of population dynamics'.⁴⁰

God

Prior to writing 'A modification of Jerne's theory of antibody production using the concept of clonal selection' for the Australian Journal of Science,⁴ Burnet sketched out his views in a draft. This draft is now held in the University of Melbourne Archives as Item 06/079 from Box Number 30, Series 6 of the Records of Frank Macfarlane Burnet, and appears in its entirety as Figure 1. It is dated 6 September 1957—about 6.5 weeks before publication of the theory in the Australian Journal of Science and 'sets out to produce a Jerne type theory from first principles...' This document is of significant interest because it appears to be the most comprehensive description of Burnet's view of the origin of diversity of specificity of lymphocytes—what Mel Cohn later termed the generator of diversity or 'GOD'.⁵⁰ In his draft, Burnet wrote: The real problem is how in the early stage of development a randomization mechanism involves the appearance of a range of globulin patterns sufficiently wide to cover all determinants.
Suppose for instance there is some aspect of protein synthesis by which a segment or set of segments can be filled in at random—or the same thing for RNA or even the DNA blueprint.

Figure 1.

Item 06/079 from Box Number 30, Series 6 of the Records of Frank Macfarlane Burnet from University of Melbourne Archives, headed '1st draft of clonal selection theory of antibody formation.'

Full figure and legend (767K)

Despite the tentative reference to the DNA blueprint, he then sketches out a possible mechanism by which the DNA encoding the antibody molecule might be randomized: In the ancestors of mesenchymal cells for a certain period in embryonic life DNA is of the nature
_______ . . . . . . . . . ._______ . . . . . . . . . . . _______ . . . . . . . . . .
where _______ are defined codes and . . . . . . . is randomised for the production of globulins.
At a later stage [the pattern undergoes] stabilization of random sequence. The same process may occur as a result of somatic mutation later in life giving AICF (auto-immune complement fixation).

In stark contrast to the draft, the published version of the 1957 Australian Journal of Science manuscript lacks any molecular mechanism addressing 'the real problem'. Instead, Burnet wrote: The theory requires at some stage in early embryonic development a genetic process for which there is no available precedent. In some way we have to picture a 'randomisation' of the coding responsible for part of the specification of gamma globulin molecules, so that after several cell generations in early mesenchymal cells there are specifications in the genomes for virtually every variant that can exist as a gamma globulin molecule. This must then be followed by a phase in which the randomly developed specification is stabilized and transferred as such to descendant cells.⁴

Even in his book The Clonal Selection Theory of Acquired Immunity,⁵¹ where he took the opportunity to greatly expand on other aspects of the theory, he barely discusses GOD and certainly did not enlarge on the model he had previously described: If at a certain stage of embryonic development certain synthetic elements in mesenchymal cells were 'randomized', the possibility might well emerge of producing all of the 10 000 or more patterns required... Suppose that at the appropriate stage of development a limited genetic determinant carrying the coding responsible for globulin patterns releases control in such a fashion that purely random arrangements are allowed which will be different at each replication.⁵¹

Nossal remarked, 'It is curious how little Burnet seemed to care about the real nature of the 'generator of diversity' creating the B-cell repertoire. He used to refer to 'somatic mutation' but clearly indicated that he meant thereby any genetic mechanism that might make a daughter cell different from its parent, be this through point mutation, deletions, sister chromatid exchange or other recombinations, translocations, etc. The molecular mechanism simply did not engage his interest'.⁵² Ada has suggested that Burnet's reticence to explore potential molecular mechanisms of clonal selection probably stems from the criticism he attracted in his attempt to provide a molecular basis for his earlier self marker theory,¹⁶ saying, 'Because one of his earlier 'bright ideas' books had been severely criticised, Burnet published [only] a brief account of his concept'.⁵³ In The Clonal Selection Theory of Acquired Immunity, Burnet⁵ wrote, 'Several immunologists, including Jerne himself, have suggested that the self marker theory is semi-mystical in character and generally unattractive.' Much of the criticism stemmed from his proposal of 'the existence and modification of self-replicating units within the antibody-producing cells', in which he claimed that if the concept were established, it would bring 'the antibody-producing mechanism into line with the growing number of cytoplasmic entities which have demonstrably, or probably, power of self-replication in response to environmental, rather than nuclear stimuli'.⁵⁴ Indeed, he specifically argued against the suggestion supported by the geneticists Haldane and Sturtevant, that the molecular patterns responsible for antigenic specificity could be traced back to the gene, saying, 'We feel that this may be an unduly sweeping generalisation'.⁵⁴

This statement naturally attracted criticism from geneticists, and Francis Crick, who was at that time promoting his Central Dogma of genetics, named Burnet as someone who's views conflicted with the dogma. Indeed, in his landmark publication on the dogma, Crick, after outlining the belief that 'once information has passed into protein it cannot get out again',⁵⁵ wrote, 'This [view] is by no means universally held—Sir Macfarlane Burnet, for example, does not subscribe to it—but many workers now think along these lines'.⁵⁵ It seems unlikely that Burnet recovered from this slight, as even after retirement, he compared molecular biologists to the physicists who developed the atomic bomb and implied that Crick held the belief 'that there is nothing substantial in biological research apart from working out the implications of the sequence of nucleotides in DNA'.¹⁷

Given this antipathy, it is ironic that the strongest support for the clonal selection theory came from molecular biology. By 1967, the sequencing of large numbers of myeloma proteins had provided strong evidence that the variety in sequence of the light and heavy chains of antibody was likely to be reflected in DNA sequence, because one cell produced many identical molecules of immunoglobulin and passed on this capacity to its progeny. Furthermore, variation in amino acid sequence was often consistent with single base changes in DNA.⁵⁰ Dreyer and Bennet⁵⁶ had proposed a model in which a gene encoding the constant region of an antibody isotype combined with one or more of a large number of oligonucleotide 'rings', each encoding a different variable portion of the chain, and this model was further developed by Brenner and Milstein⁵⁷ to incorporate sequence variation at junction sites. 'Perhaps the greatest force for acceptance of the theory,' said Nossal,⁵² 'was the progressive unravelling of the molecular side of the antibody story, both the protein chemistry and the molecular genetics.'

Again, while Burnet's clonal selection theory supported—even demanded—a genetic mechanism for generating an enormous variety of antibody sequences, Talmage's model had settled for the much more modest requirement of only 5000 specificities, and did not provide the dramatic raison d'être of Burnet's theory. It seems likely that the original importance of Talmage's contribution was lost in the first decade of experimentation under the new paradigm of clonal selection because it was not as relevant as Burnet's model to the experiments that were possible at the time. It is ironic that as the theoretical basis for generating the enormous breadth of the immune repertoire was developed, and its existence in reality established, Talmage began to misquote his original proposal. For example, in the recently published 'Reflections on the clonal selection theory' in Nature Reviews Immunology, Talmage wrote, 'I published a paper in Science in 1959 that demonstrated how an almost unlimited number of different combinations of approximately 50 000 different globulins might explain immunological specificity'.⁵⁸ Whether an honest mistake, or an attempt to rewrite history, this was unnecessary. His original contribution was outstanding.

Note Added in Proof

Following the presentation of this paper by Alan Baxter, who received a qualitative medical education, but now performs quantitative genetic studies, Peter Doherty, who received a qualitative veterinary education, but now performs quantitative virological studies, pointed out that although Burnet's immunological work was qualitative, his earlier virological studies were quantitative. We accept this criticism

References

1. Watson JD. The Double Helix. Touchstone: New York, 1968.
2. Burnet FM, Freeman M, Jackson AV, Lush D. The Production of Antibodies. Macmillan: Melbourne, 1941.
3. Burnet FM, Fenner F. The Production of Antibodies, 2nd edn. Macmillan: London, 1949.
4. Burnet FM. A modification of Jerne's theory of antibody production using the concept of clonal selection. Aust J Sci 1957; 20: 67–69.
5. Burnet FM. The Clonal Selection Theory of Acquired Immunity. Cambridge University Press: Cambridge, 1959.
6. Hodgkin PD, Heath WR, Baxter AG. The clonal selection theory: 50 years since the revolution. Nat Immunol 2007; 8: 1019–1026. | Article | PubMed | ChemPort |
7. Talmage DW. Allergy and immunology. Annu Rev Med 1957; 8: 239–256. | Article | PubMed | ISI | ChemPort |
8. Ada GL. The conception and birth of Burnet's clonal selection theory. In: Mazumdar PMH (ed). Immunology 1930–1980. Wall and Thompson: Toronto, 1989, pp 33–40.
9. Landsteiner K, Chase MW. Studies on the sensitization of animals with simple chemical compounds: ix. Skin sensitization induced by injection of conjugates. J Exp Med 1941; 73: 431–438. | Article | ISI | ChemPort |
10. Talmage DW. The acceptance and rejection of immunological concepts. Annu Rev Immunol 1986; 4: 1–11. | Article | PubMed | ChemPort |
11. Dixon FJ, Talmage DW. Catabolism of I¹³¹ labelled bovine gamma globulin in immune and non-immune rabbits. Proc Soc Exp Biol Med 1951; 78: 123–125. | PubMed | ChemPort |
12. Dixon FJ, Talmage DW, Maurer PH. Radiosensitive and radioresistant phases in the antibody response. J Immunol 1952; 68: 693–700. | PubMed | ISI | ChemPort |
13. Maurer PH, Dixon FJ, Talmage DW. A quantitative comparison of antibody produced by x-radiated and normal rabbits. Proc Soc Exp Biol Med 1953; 83: 163–166. | PubMed | ChemPort |
14. Talmage DW. Immunological specificity. Science 1959; 129: 1643–1648. | Article | PubMed | ISI | ChemPort |
15. Haurowitz F. The nature of the protein molecule: problems of protein structure. J Cell Comp Physiol 1956; 47 (Supp 1): 1–33. | Article | ISI | ChemPort |
16. Burnet FM. Enzyme Antigen and Virus. Cambridge University Press: Cambridge, 1956.
17. Burnet FM. Men or molecules? A tilt at molecular biology. Lancet 1966; 1: 37–39. | Article | PubMed | ChemPort |
18. Nossal GJV. Antibody production by single cells. Br J Exp Pathol 1958; 39: 544–551. | PubMed | ChemPort |
19. Nossal GJV, Lederberg J. Antibody production by single cells. Nature 1958; 181: 1419–1420. | Article | PubMed | ISI | ChemPort |
20. Mäkelä O, Nossal GJ. Study of antibody-producing capacity of single cells by bacterial adherence and immobilization. J Immunol 1961; 87: 457–463. | PubMed |
21. Nossal GJV, Mäkelä O. Elaboration of antibodies by single cells. Annu Rev Microbiol 1962a; 16: 53–74. | Article | PubMed | ISI | ChemPort |
22. Nossal GJV, Mäkelä O. Kinetic studies on the incidence of cells appearing to form two antibodies. J Immunol 1962b; 88: 604–612. | PubMed | ISI | ChemPort |
23. Mäkelä O. The specificity of antibodies produced by single cells. Cold Spring Harbor Symp Quant Biol 1967; 32: 423–430.
24. Attardi G, Cohn M, Horibata K, Lennox ES. Symposium on the biology of cells modified by viruses or antigens. II. On the analysis of antibody synthesis at the cellular level. Bacteriol Rev 1959; 23: 213–223. | PubMed | ISI | ChemPort |
25. Attardi G, Cohn M, Horibata K, Lennox ES. Antibody formation by rabbit lymph node cells. I. Single cell responses to several antigens. J Immunol 1964; 92: 335–345. | PubMed | ISI | ChemPort |
26. Lederberg J. Genes and antibodies. Science 1959; 129: 1649–1653. | Article | PubMed | ISI | ChemPort |
27. Nossal GJV. Induction of immunological tolerance in rats to foreign erythrocytes. Nature 1957; 180: 1427–1428. | Article | PubMed | ISI | ChemPort |
28. Smith RT, Bridges RA. Immunological unresponsiveness in rabbits produced by neonatal injection of defined antigens. J Exp Med 1958; 108: 227–250. | Article | PubMed | ISI | ChemPort |
29. Terres G, Hughes WL. Acquired immune tolerance in mice to crystalline bovine serum albumin. J Immunol 1959; 83: 459–467. | PubMed | ISI | ChemPort |
30. Medawar PB, Woodruff MF. The induction of tolerance by skin homografts on newborn rats. Immunology 1958; 1: 27–35. | PubMed | ChemPort |
31. Billingham RE, Brent L, Medawar PB. Actively acquired tolerance of foreign cells. Nature 1953; 172: 603–606. | Article | PubMed | ISI | ChemPort |
32. White RG. Antibody production by single cells. Nature 1958; 182: 1383–1384. | Article | PubMed | ISI | ChemPort |
33. Hiramoto RN, Hamlin M. Detection of two antibodies in single plasma cells by the paired fluorescence technique. J Immunol 1965; 95: 214–224. | PubMed | ISI | ChemPort |
34. Green I, Vassalli P, Nussenzweig V, Benacerraf B. Specificity of the antibodies produced by single cells following immunization with antigens bearing two types of antigenic determinants. J Exp Med 1967; 125: 511–526. | Article | PubMed | ChemPort |
35. Burnet FM. Immunological recognition of self. Science 1961; 133: 307–311. | Article | PubMed | ISI | ChemPort |
36. Nossal GJV. Turning points—mentors and manipulation. Nature 2004; 429: 811. | Article | PubMed | ChemPort |
37. Nossal GJV. One cell one antibody; prelude and aftermath. Immunol Rev 2002; 185: 15–23. | Article | PubMed | ISI | ChemPort |
38. Cohn M. The wisdom of hindsight. Annu Rev Immunol 1994; 12: 1–62. | Article | PubMed | ChemPort |
39. Baxter AG. Self/nonself recognition. In: Pollard KM (ed). Autoantibodies and Autoimmunity: Molecular Mechanisms in Health and Disease. Wiley-VCH: Weinheim, Germany, 2006, pp 37–61.
40. Burnet FM. Theories of immunity. In: Conceptual Advances in Immunology and Oncology. Harper and Row: New York, 1963, pp 7–21.
41. Baxter AG. Germ Warfare—Breakthroughs in Immunology. Allen and Unwin: Sydney, 2000.
42. Trentin JJ, Fahlberg WJ. An experimental model for studies of immunologic competence in irradiated mice repopulated with 'clones' of spleen cells. In: Conceptual Advances in Immunology and Oncology. Harper and Row: New York, 1963, pp 66–74.
43. Nossal GJV. Burnet and science—an appreciation. Aust Ann Med 1969; 4: 311–315.
44. Burnet FM, Mackay IR. Lymphoepithelial structures and autoimmune disease. Lancet 1962; 280: 1030–1033. | Article |
45. Trentin J, Wolf N, Cheng V, Fahlberg W, Weiss D, Bonhag R. Antibody production by mice repopulated with limited numbers of clones of lymphoid cell precursors. J Immunol 1967; 98: 1326–1337. | PubMed | ChemPort |
46. Feldman M, Mekori T. Antibody production by 'cloned' cell populations. Immunology 1966; 10: 149–160. | PubMed | ChemPort |
47. Haurowitz F. The template theory of antibody formation. In: Conceptual Advances in Immunology and Oncology. Harper and Row: New York, 1963, pp 22–36.
48. Simonsen M. The impact on the developing embryo and newborn animal of adult homologous cells. Acta Pathol Microbiol Scand 1958; 40: 480–500.
49. Szenberg A, Warner NL. Qualitative aspects of the Simonsen phenomenon. I. The role of the large lymphocyte. Brit J Exp Pathol 1962; 43: 123–128.
50. Lennox ES, Cohn M. Immunoglubulins. Annu Rev Biochem 1967; 36: 365–406. | Article | ChemPort |
51. Burnet FM. The Clonal Selection Theory of Acquired Immunity. Cambridge University Press: Cambridge, 1959.
52. Nossal GJV. The coming of age of the clonal selection theory. In: Mazumdar PMH (ed). Immunology 1930–1980. Wall and Thompson: Toronto, 1989, pp 41–48.
53. Fenner F, Ada G. Frank Macfarlane Burnet: two personal views. Nat Immunol 2007; 8: 111–113. | Article | PubMed | ChemPort |
54. Burnet FM, Fenner F. Genetics and immunology. Heredity 1948; 2: 289–324. | ISI |
55. Crick FH. On protein synthesis. Symp Soc Exp Biol 1958; 12: 138–163. | PubMed | ChemPort |
56. Dreyer WJ, Bennett JC. The molecular basis of antibody formation: a paradox. Proc Natl Acad Sci USA 1965; 54: 864–869. | Article | PubMed | ChemPort |
57. Brenner S, Milstein C. Origin of antibody variation. Nature 1966; 211: 242–243. | Article | PubMed | ISI | ChemPort |
58. Cohn M, Mitchison A, Paul WE, Silverstein AM, Talmage DW, Weigert M. Reflections on the clonal-selection theory. Nat Rev Immunol 2007; 7: 823–830. | Article | PubMed | ChemPort |

Acknowledgements

We would like to acknowledge the help, correspondence and suggestions of GJV Nossal and G Ada; J McCluskey for provision of a transcript of Burnet's draft of the clonal selection theory and the financial support of the Australian National Health and Medical Research Council. Permission to print Burnet's draft of the clonal selection theory was granted by Ms Elizabeth Dexter.