278 THE ANTIGENIC STRUCTURE OF BACTERIA 



of change in antigenic structure. We have very good reasons for believing that the 

 change in antigenic structure determines the loss of virulence, and it certainly 

 determines the change in antigenic value of the variant strain regarded as an" 

 immunizing agent. We may, then, for our immediate purpose and to avoid con- 

 fusion, explicitly limit ourselves to this antigenic variation ^nd its consequences, 

 noting that if we define the S— >R variation in this way we are acting in defiance 

 of general usage. 



In terms of antigenic structure we may say that this variation consists in the 

 loss of the heat- stable somatic antigen that characterizes the surface of the normal virulent 

 bacterial cell. This loss may be associated with an uncovering of some other somatic 

 antigen, which then dominates the antigenic behaviour of the strain. 



Variation of this type has now been described in a wide range of different 

 bacterial species — in the Pasteurella group (de Kruif 1921, Websterl925),in pneumo- 

 cocci (Griffith 1923, Reimann 1925, 1927), in staphylococci (Bigger, Boland and 

 O'Meara 1927), in streptococci (Todd 1928rt) and in a host of other organisms. 

 We are probably justified in regarding it as a type of variation to which pathogenic 

 bacteria are inherently liable. 



Sometimes (see White 1932, 1933, Henderson 1939) this variation may proceed 

 still further, and another antigenic constituent may be lost. The antigenic be- 

 haviour of the organism may then be dominated by a component that, in the 

 normal state, was altogether latent. It may be noted (a) that the S — > R variation 

 is quite independent of the H — >■ variation (rough variants are often flagellated), 

 and (b) that loss variations of the S — > R or more deeply seated types are often 

 irreversible. 



There is another kind of antigenic variation to which reference must be made 

 before we pass to the next section of our discussion. 



Andrewes (1922, 1925) described a curious phasic variation in the H antigens 

 of certain flagellated species. Taking two bacteria, x and y, that showed the 

 flagellar type of agglutination when tested either against an anti-x or against an 

 anti-^ serum, and therefore possessed at least one H antigen in common, he absorbed 

 the anti-x serum with bacillus y and thus obtained a serum that agglutinated x but 

 not y. He then took a broth culture of x and plated it on a solid medium, thus 

 obtaining separate colonies. Subculturing from several of these, he obtained 

 different cultures, each representing a single bacillus in the original broth culture. 

 When he tested these different cultures against the anti-a; serum, rendered specific 

 by absorption, and against the anti-y serum, containing the common, or " group " 

 antibody, he found that his subcultures fell sharply into two classes. Those of one 

 class were agglutinated to titre by the specific anti-x serum, but not at all by the 

 anti-t/ serum. Those of the other were agglutinated to titre by the anti-?/ serum, but 

 not at all by the specific anti-a; serum. The only possible conclusion would seem 

 to be that the original culture of x showed cross-agglutination with the anti-y 

 serum not because each bacillus in the culture possessed two antigenic components, 

 say a and b, in their flagella, a being specific for x and b being shared by y, but 

 because some of the bacilli possessed a alone, and others b alone. We need no t, 

 for the moment, worry as to whether the a bacilli do or do not possess a trace of b 

 and vice versa. Similarly Andrewes was able to show that the y culture, which 

 agglutinated with anti-a; and anti-t/ sera, contained some bacilli possessing the 

 common flagellar antigen b and others containing an antigen c specific for y. The 

 bacilli that possessed the group antigen only were referred to as being in the group 



