A rugged binding landscape unifies static and dynamic paradigms in protein-protein interactions

Predicting protein-protein interaction affinities from structural data remains a challenge. Although binding funnel theory describes the formation of native complexes, the topography of the funnel bottom and its influence on affinity are often overlooked. Using nanobody-antigen complexes as model systems, we identified two series of nanobodies that adopt nearly identical binding poses toward their respective antigens yet exhibit diverse affinities. These series correspond to two distinct binding paradigms: a static paradigm in which affinity is accurately ranked by Rosetta scoring of co-crystal structures alone, and a dynamic paradigm in which affinity can only be ranked using molecular dynamics-sampled ensembles. The two paradigms differ in interfacial dynamics. In the dynamic series, the relative motion between binding partners (ΔF) is temperature-sensitive, with optimal affinity correlation at 298 K where experimental affinities were measured. In the static series, ΔF is minimal and insensitive to temperature. Local frustration analysis reveals the energetic basis for this dichotomy: dynamic interfaces exhibit increased frustration upon thermal sampling, enabling escape from local minima and sampling of functionally relevant microstates, whereas static interfaces show minimal frustration changes across temperatures. The temperature-dependent frustration difference mirrors ΔF sensitivity, establishing local frustration as the determinant of interfacial dynamics. Furthermore, static interfaces are characterized by a higher density of canonical hotspot residues, while dynamic interfaces utilize interfacial ruggedness to modulate affinity. Together, these results demonstrate that conserved binding modes can encode distinct energy landscapes and provide a framework for determining when ensemble-based sampling is required for accurate affinity prediction.