



















used in immune responses of inbred mice to many hap- ten, polysaccharide, and even protein antigens, includ- ing p-azophenylarsonate (Pawlak et al., 1973; Makela et al., 1976; Manser, 1990), phenyloxazolone (Makela et al., 1978; Kaartinen et al, 1983; Berek and Milstein, 1987), (4-hydroxy-3 - nitrophenyl)acetyl (NP) (Imanishi and Makela, 1974, 1975; Blier and Bothwell, 1988), DNP (Dzierzak et al., 1980, 1985), phosphorylchol ine (Potter and Lieberman, 1970; Sher and Cohn, 1972; Barstad et with the findings that murine B cells that have a re- arranged heavy chain immunoglobulin gene in the germ line always change the CDR3 sequence and often the V as well when they deviate from the original antigen reactivity (Cascalho et al., 1996; López-Maclas et al., 1999). Our data may explain the long-standing puzzle of "canonical" CDR structures in antibodies (Chothia and Lesk, 1987; Chothia et al., 1989). We would suggest that uniform CDR shapes are important for the stability of a binding site, with the exception of V CDR3, which has to "fit" the antigen surface much more precisely. al., 1974; Crews et al., 1981), phosphatidylcholine (Seidl et al., 1997), dextran (Blomberg et al., 1972; Riblet et The nese results a tuctural anal- ysis shows that about 30 side chains from each protein centor whe Potter, 1978), Staphvlococcal nuclease (Fathman et al. 1977), and influenza hemagglutinin (McKean et al., 1984; Clarke et al., 1985). T cell recognition of peptide/MHC complexes is often (but not always) fairly restricted with respect to However, it is not clear whether the preferred V gene segment(s) are absolutely required for such epitopes or are merely the result of extensive clonal expansion where a slight advantage in binding would allow the most optimal VV combination to predominate. In the current study, we find that mice having only one func- tional VHand (effectively) two Vi gene segments are able Lit are in close contact but only a fraction of those residues have any contribution to the binding affinity and only two account for more than three-quarters of the free energy (Wells, 1996). Another interesting issue raised by the data presented here is the role of somatic mutation in producing higher- affinity antibodies upon rechallenge with antigen. Our data shows that, at least for protein antigens, having an arbitrarily chosen human V and one of two mouse A chains as a germl ine repertoire is no barrier to generat gene usage (e.g., Hedrick et al., 1988). sistont k on the chicken i is finding is con- where only a sing le functional V and V are utilized, but goes beyond this in two important respects. One is that most m likly that the restricted V es. to mount usage observed in most immune responses represents the consequence of antigen-driven clonal expansion of kinetically favorable In the HC1+/ lgH Igkmice, we find a remarkable restriction in V CDR3 diversity with respect to both length and amino acid composition in the primary im- mune response to both h