From the structure of antibodies to the diversification of the immune response.

When an animal is infected, either naturally or by experimental injection, with a bacterium, virus, or other foreign body, the animal recognises this as an invader and acts in such a way as to remove or destroy it. There are millions of different chemical structures that the animal has never seen and yet which it is able to recognise in a specific manner. How is this achieved? Scientists have been fascinated by this question for most of this century, and we continue to be fascinated by the intricacies and complexities that still need to be clarified. Even so, looking back over the years since I myself became involved in this problem, progress in the understanding of the process has been phenomenal. Suffice it to remind our younger colleagues that 20 years ago we were still trying to demonstrate that each antibody differed in its primary amino acid sequence. What attracted me to immunology was that the whole thing seemed to revolve around a very simple experiment: take two different antibody molecules and compare their primary sequences. The secret of antibody diversity would emerge from that. Fortunately at the time I was sufficiently ignorant of the subject not to realise how naive I was being. Back in 1962, when I had by accident become the supervisor of Roberto Celis in Argentina, it occurred to me that antibody diversity might arise from the joining by disulphide bridges of a variety of small polypeptides in combinatorial patterns. I don't know whether anybody else had the same idea at that time, but of all the prevailing theories about antibody diversity that I am aware of, this is one that was widest of the mark. I hold it to my credit that I never put it into print. But it was of great value to me as it provided an intellectual justification to work on disulphide bonds of antibodies. By the time I joined the Laboratory of Molecular Biology in 1963, the model of two heavy and two light chains joined by disulphide bonds (Figure 1) had been established (1), and I was eager to accept Dr Sanger's proposal that I should engage in studies of antibody combining sites. The nature of antibody diversity At first I looked for differences in fingerprints of digests of iodinated antibodies directed against different antigens. The pattern that emerged from those studies implied that purified antibodies were too complex and differed only in a subtle quantitative way from the totally unfractionated immunoglobulin. I never published those results, which only led me to the conviction that the protein chemistry of antibodies at that level was too difficult to tackle, and that a different approach was needed. The study of the amino acid sequence around the disulphide bonds of the immunoglobulins was my own short-cut to the understanding of antibody diversity. I soon recognised the existence of what appeared to be a variable disulphide bridge and a common disulphide bridge (2,3), but the full meaning of that observation only became obvious when Hilschmann and Craig described the variable and constant halves of antibody light chains (4). The variable half contained one disulphide bond, and the constant half the other. This was followed, in later studies with Pink, Frangione, Svasti and others, by the observation of the repeating pattern of similar S-S loops as a distinctive common architectural feature of the different classes and subclasses of immunoglobulin chains. What distinguished them from each other was the diversity of interchain S-S bonds (5). The period between 1965 and 1970 was full of excitement, both at the experimental and theoretical level. How were these variable and constant regions going to be explained? It was now not only a problem of millions of antibody structures, but that in addition those millions of structures were part of a polypeptide which otherwise had an invariant primary sequence encoded by only one or very few genes. How to solve the puzzle? Dreyer and Bennett (6) suggested that there were thousands of genes in the germline and that the paradox was easy to solve if we postulated a completely unprecedented scheme. This became known as the 'two genes-one polypeptide' hypothesis. At the time we did not like that, and proposed a mechanism of hyper-mutation