I n 1946, when I was starting work as a research student under the supervision ofDr F. Sanger, tbe second edition of Karl Lansteincr's book "The Specificity of Serological Reactions' [I] reached England. In it WHS summarized Lhe considerable body of information available on the range of antibody specificity and much of it was Landsteiner's own work or by others using his basic technique of preparing antibodies against haptcncs and testing their ability to inhibit the precipitation of the antisera and the conjugated protein. Also described in this book was the work in Uppsala ofTiselius and Pederson in collaboration with Heideiberger and Kabat in which they showed that all rabbil antibodies were in the y globulin fraction of serum proteins and that they had a molecular weight of 150.000. This combination of an apparently infinite range ofantibody-combining specificity associated with what appeared to be a nearly homogeneous group of proteins astonished me and indeed still does.
ACTIVE FRAGMENTS OF ANTIBODIESThe preparation of antibodies by dissociation of specific precipitates with strong salt solutions or in acidic conditions had been described, as had the preparation in fair yield of 7 globulin fractions by salting-out techniques from whole serum, so an experimental approach to the structural basis of antibody-combining specificity was possible. A start had indeed been made by showing that the whole molecule was not required for the combining specificity. Parventjev [2] had introduced pepsin treatment of serum as a commercial tncthod of purification ofhorse antitoxins and Pctcrmann & Pappcnhcitiicr [3] studied the reaction but using purified horse anlidiphthcria to.xin rather than whole serum for the peptic digestion. They showed that a product able to flocculate wiih the toxoid or neutralize toxin could be obtained and thai it had a moleeular weight of 113,000, i.e. substantially less than that of the original molecule. Petermann [4] showed later that human 7 globulin could be split by papain to give what she estimated, by using the ultraccntrirugc. to be quarter molecules. No antibody activities were, however. investigated in this study. At about the same time Landsteiner [5] was extending his investigation of the antigcnie specificity of protein antigens and had found that crude but apparently low-molecular-weight peptides from an acid digest of silk fibroin could inhibit the precipitation of soluble fibroin with its rabbit antiserutii. This finding, in conjunction with the haptcnc and other studies, suggested that antigenic sites and presumably, therefore, antibody combining sites were small, certainly very much smaller than the antibody molecules and further attempts to obtain fragments of an antibody molecule, which retained the power to combine with the antigen seemed worthwhile. Testing for such active fragments was by their ability to inhibit the combination of the antigen and whole antibody. However, although a variety of conditions of hydrolysis by acids or enzymes were investigated [6] only papain g...