Macromolecular interactions are central to cellular regulation and biological function, and antibody-antigen complexes are often used as a paradigms for molecular recognition. Protein-protein interactions are studied utilizing monoclonal antibodies (mAbs) specific for hen egg white lysozyme (HEL), a protein which has long served as a prototype for investigating the specificity of immune recognition. Four structurally and functionally related mAbs, H8, H10, H26, and H63, recognize highly coincident epitopes and share over 90% sequence homology, but they differ significantly in their fine specificity properties. Using BIAcore surface plasmon resonance technology (SPR), we have defined new methodology and experimental criteria for the analysis of receptor-ligand binding kinetics using SPR. Our results established that the real-time association kinetics of all 4 Fabs specific for HEL are best described by a 2-step kinetic model, the most simplified scheme that quantitatively describes an encounter followed by a docking/conformational rearrangement. Although bimolecular association has for over 20 years generally been considered to occur in 2 steps, in practice, experimental systems which allow monitoring of the 2 steps are rare. Our results are the first to experimentally confirm a 2-step binding mechanism for antibody-protein complexes. The significance is that we have been able to experimentally resolve 2 steps in the binding kinetics, allowing experimental separation and analysis of each, providing insight about molecular mechanisms not obvious from a simpler Langmuir model. The results will be applicable to a broad range of receptor-ligand complexes. Our results have informed experimental design of other biophysical measurements by our collaborators (calorimetry, R.C. Willson; atomic force microscopy, P. Hinterdorfer). We have extended transition state theory to describe the free energy changes of 2-step binding of bimolecular complexes, an advance in the theoretical foundation of protein dynamics. Our application of the model to the 2-step (3-state) binding of antibody complexes reveals experimentally determined free energy change profiles signicantly different from those proposed in the literature, and provides new insight into the structural mechanism determining affinity and antibody affinity maturation. Using novel protocols significantly different from those prevailing in the literature, we have applied van't Hoff analysis to the kinetics of these complexes, in order to calculate binding and activation thermodynamics of the separate steps as well as of net association of 2-step binding. We find that the thermodynamics of the encounter and docking steps are significantly different from each other. The thermodynamics of encounter are consistent with a hydrophobically driven process. The energetics of docking are consistent with formation of noncovalent bonds and conformational rearrangements. There is entropy-enthalpy compensation within each step, and also between the steps. These are the properties that have long been predicted for the 2 steps of the association process, but protein-protein complexes that display these properties are rarely reported, probably because the 2 steps usually cannot be separated. Preliminary results suggest that antigenic mutations alter the thermodynamic balance between the steps. These results support the hypotheses that (i) docking includes a conformational change, (ii) hydrophobic interactions predominate energetics of one of the antibodies, while those of the other 3 reflect a range of more polar and electrostatic interactions The net changes agree with calorimetry measurements on the same complexes by our collaborator R.C. Willson, U Houston, and provide new insight into the mechanisms underlying frequently observed inconsistencies between theromdonamics derived from calorimetry and kinetics. Calorimetry is apparently biased towards the docking step while van't Hoff thermodynamics calculated from BIAcore kinetics using widely prevailing methods are biased towards the encounter step. It is becoming apparent that there may be fundamental thermodynamic differences in receptor binding by agonist and antagonists. Understanding thermodynamics of binding can provide information on the nature of transition states and underlying reaction mechanisms. A detailed understanding of the thermodynamics of these structurally and functionally characterized complexes is therefore of theorectical and practical interest. In collaboration with R. Mariuzza, CARB, we have determined a new antibody-antigen x-ray structure which confirms our epitope mapping, shows induced fit of the combining site, and validates our modeling and computational studies. We have developed this model system because the structurally similar antibodies show a range of functional and structural variations which we believe will provide major insights into the interrelationships of structure, flexibility, specificity, thermodynamics, and kinetics. Among the 4 antibodies, the X-ray structure of the H10-antigen complex was determined in 1989 in collaboration with the E. Padlan. Since then we have built models of 2 of the antibodies, and last year we published the structure of a fourth complex (H63 with and without complexed antigen). The H63 structure is very similar to that of H10, and shows conformational changes on binding. It also validates our homology-based modeling of the other 2 antibodies. We are testing the hypothesis that cross-reactivity requires conformational flexibility of the Ab combining site, largely modulated by intramolecular salt links and networks; while rigidly preorganized binding sites produce more specific binding. A major obstacle to the de novo structure-based design of ligands and of receptors with predefined specificity, including Abs, is an incomplete understanding of the roles of receptor mobility and conformational changes that accompany ligand binding. Our results are directly applicable to rational design of antibodies with predesigned specificity and dynamics for diagnostic and therapeutic applications They also should be applicable to other protein-protein receptor-ligand interactions.
