Journal of General Virology (1995), 76, 889-897. Printed in Great Britain 889 All rabbits immunized with type A influenza virions have haemagglutination-inhibition antibody response biased to in antigenic site B a serum a single epitope R. Lambkin and N. J. Dimmock* Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, UK Nine rabbits were immunized with type A influenza virions and the epitope specificities of the secondary serum haemagglutination-inhibition (HI) antibody re- sponse were analysed with a panel of neutralizing monoclonal (MAb) antibody double escape mutants. Each of the latter was made by sequential selection using a MAb directed to an epitope of a discrete antigenic site, site A, site B or site D, of the haemagglutinin (HA). Thus the epitope reactivity of the escape mutants was represented as A+B-D -, A-B+D and A-B-D +. The HI antibody response of all antisera was biased to the site B epitope. In 9/12 antisera, obtained from seven rabbits immunized with whole virions, the site B epitope was predominant, representing 65-82% of the total HI antibody. The restriction of HI antibody was unaffected by strain of rabbit, route of inoculation (intravenous or subcutaneous), use of Freund's adjuvant, and up to four immunizing injections. In 3/7 rabbits immunized with whole virus, there was a HI antibody response to the HC2 (site A) or HC10 (site D) epitope, but not both, of equal magnitude to the site B epitope. The HI antibody response in one of the rabbits (#40) became more biased to the site B epitope between the third and fourth immunizing doses. Two further rabbits were immunized with virions which had been partially digested with bromelain and then purified from free HA. Both of these made equal HI antibody responses to the site B epitope and the site D epitope, possibly because their remaining HA spikes were better exposed. Overall, these data demonstrate an unexpected degree of restriction in the production of biologically relevant antibody, such that some rabbits (e.g. #45) mount an HI antibody response which is essentially epitope-specific. Implications for epitope specificity of HI antibody stimulated by human influenza vaccines, and also for the generation of antigenic drift variants are discussed. The reason for the non-responsiveness of the immune system to the many other HI epitopes of the HA is not known. Introduction There are 14 subtypes of type A influenza virus (H1-14) which are defined by the antigenicity of their haemag- glutinin (HA) protein. Of these, three have caused pandemic and epidemic disease in man. The latter is believed to result from antigenic drift of the viral HA, a phenomenon in which accumulation of amino acid substitutions in key epitopes allows infection of indi- viduals who had previously acquired immunity as the result of infection with the parent virus (Wilson & Cox, 1990; Webster et al., 1992). It is widely believed that antigenic drift is driven by the immune response, and largely by neutralizing antibody specific to the HA. However, in the experimental version of antigenic drift, * Author for correspondence. Fax +44203523568. e-mail nd@dna.bio.warwick.ac.uk the production of an antibody escape mutant by a monoclonal antibody (MAb), this is documented as only taking place in the presence of a single MAb (Laver et al., 1979; Yewdell et al., 1979; Webster & Laver, 1980). This is because two independent amino acid substitutions are required if two MAbs to two non-overlapping epitopes are used, and this occurs at the vanishingly low frequency of the product of the individual rates for amino acid substitution (i.e. about 10 -5 × 10-5 = 10 -l°) (Laver et al., 1979; Yewdell et al., 1979; Portner et al., 1980; Lubeck et al., 1980). Thus there is a paradox as to the mechanism of antigenic drift in nature since the antibody response is thought to be too diverse to permit the emergence of drift variants. This belief is based mainly on the knowledge that the HA has up to five antigenic sites each composed of several non-overlapping epitopes, and the assumption that the convalescent neutralizing antibody response would be directed in approximately equal amounts to at least several of these. For these reasons antigenic drift 0001-2940 © 1995 SGM