B Mechanisms of Antibody Binding to a Protein ELIZABETH D. GETZOFF,* H. MARIo GEYSEN, STUART J. RODDA, HANNAH ALEXANDER, JOHN A. TAINER, RicHARD A. LERNER The mehanisms of antibody binding to a protein were studied by an analysis of specific amino acid residues critical to nine antigenic sites on myohemerythrin. Rabbit antisera to the whole protein were assayed for binding to more than 1500 distinct peptide analogs differing from the protein sequence by single amino acid replacements. The results, combined with information from the three-dimensional crystallographic structure, were used to evaluate probable mhanisms of antibody binding at individ- ual sites. The data from all sites exmined indicate that initial binding to solvent- exposed amino acid residues may promote local side-chain displacements and thereby allow the participation of other, previously buried, residues. A NTIGENIC SITES ON THE PROTEIN myohemerythrin (MHr) were cho- en from antigenicity profiles of sev- en rabbit antisera produced against the whole protein (anti-MHr) and mapped with the use of 113 overlapping hexapeptide homologs of the MHr sequence (1). These sites (referred to by inclusive amino acid residue numbers) represent peaks of fre- quent antigenicity, as probed with peptides, as well as one peptide (residues 16 to 21) recognized with high titer by a single antise- rum. The roles of individual side chains in each site were characterized by measuring the reactivities of appropriate antisera with sets of 120 peptides that included all possi- ble analogs (replacement nets) differing in sequence from the parent peptide by re- placement of a single amino acid residue (Fig. 1). Competition data measured in a solution phase assay show that native MHr inhibits the anti-MHr binding of peptides corresponding to the antigenic sites exam- ined (1). This competition suggests that the antibody binding role for an individual side chain in the protein can be inferred from its role in the peptide. Such replacement net data made it possi- ble to assign each residue of a peptide to one of the following four categories (Table 1): essential, selected, partially replaceable, or generally replaceable. In site 4 to 9 (Fig. IA), antigenic reactivity was lost when re- placing Glu6 or Tyr8 with any of the other 19 commonly occurring amino acids; thus these residues are classified as essential to antibody binding. This ability of antibodies to distinguish single amino acid changes in antigens has also been shown in fine speci- ficity studies of other systems (2, 3). In contrast to the required specificity of Glu6 and Tyr8, Ile4 can be replaced by all common 6 MARCH I987 amino acids without affecting peptide bind- ing by the antiserum (generally replaceable). Likewise, Pro5 is replaced by most other amino acids without influencing antibody binding; Pro7, however, is partially replace- able and Val9 is selected (selectively replace- able only by Ala, Ile, Ser, or Thr). In some cases, replacement net data are only partially interpretable and may reflect the existence of multiple overlapping antigenic sites recog- nized by polyclonal antisera. For example, replacement nets in which a given level of reactivity is partially restored by many sub- stitutions at a given position and completely restored by a few, seem to indicate that two overlapping sites are present, one in which that residue is essential or selected, and another in which it is generally replaceable. Overlapping antigenic determinants have also been indicated by experiments examin- ing the repertoire of monoclonal antibody specificities to protein antigens (4). Residues identified as critical (that is, essential or selected) to interaction with anti-MHr from one rabbit are usually critical to interaction with polyclonal antibodies from other responding rabbits (Table 1). This applies both within the same peptide (for example, Phe80 and Lys83 in site 80 to 85) and between overlapping peptides (Tyr67, Glu69, and Val70 in sites 63 to 68, 65 to 70, and 68 to 72). Using monoclonal antibodies, another investigator (5) has identified within the C helix of MHr the same sites and patterns of critical residues. Thus, intrinsic stereochemical properties of specific residues within the three-dimen- sional environment of a protein bias the selection of the critical residues recognized by the immune system, at least for antigenic sites probed by peptides. Although MHr antisera from different individual animals may recognize different sites, the chemical basis for interaction at a given recognized site appears to be similar among the applica- ble rabbit antisera. As is shown in Table 2, eight of the nine sites each include one or more highly ex- posed and accessible critical side chains (>55 percent exposed side-chain area, acces- sible to an antibody binding domain as modeled by a 15 A radius sphere), and one or more largely buried and inaccessible criti- cal side chains ('5 percent exposed side- chain area, nearly or totally inaccessible to a solvent-sized probe of 1.4 A radius) within the native protein structure. Moreover, criti- cal, buried side chains seem to be directly involved in binding to the antibody, rather than simply stabilizing an appropriate con- formation of the peptide, although confor- mational stabilization cannot be definitively ruled out. First, short peptides are very unlikely to have fixed or strongly favored conformations in solution and would be expected to adapt to the more structured antibody binding site (6). Second, often only one amino acid type at a given position permits binding, rather than a set of accept- able amino acid replacements with similar properties (Fig. 1), as would be expected for side chains performing a conformational role. Third, buried critical residues are not randomly located, but show a specific posi- tional relation in the protein structure to the solvent-exposed critical residues, which is most frequently seen as a gating phenome- non (see below). This proposed antibody binding role of buried hydrophobic side chains agrees with the analysis of known protein complexes in which binding affinity depends in part on maximizing the hydro- phobic surface area in the buried interface (7), with the previously noted importance of hydrophobic and aromatic residues in anti- genicity (8), and with the prevalence of complementary hydrophobic and aromatic residues in the binding sites of antibodies (3, 9). The critical residues within the character- ized sites form microassemblies: three-di- mensional clusters of closely interacting side chains (Fig. 2). A network of microassemb- lies interconnects to form the backbone of the most frequently reactive sites (1). From this work, three levels of structural hierarchy can be proposed within a protein antigen: (i) microassemblies of sequence-local, criti- E. D. Getzoff, H. Alxander, J. A. Tainer, R. A. Lerner, Department of Molecular Biology, Research Institute of Scripps Clinic, La Jolla, CA 92037. H. M. Geysen and S. J. Rodda, Department ofMolecular Immunology, Commonwealth Serum Laboratories, Parkville, Victoria 3052, Australia. *To whom correspondence should be addressed. REPORTS 1191 on December 19, 2014 www.sciencemag.org Downloaded from on December 19, 2014 www.sciencemag.org Downloaded from on December 19, 2014 www.sciencemag.org Downloaded from on December 19, 2014 www.sciencemag.org Downloaded from on December 19, 2014 www.sciencemag.org Downloaded from on December 19, 2014 www.sciencemag.org Downloaded from