Protein-protein interactions (PPIs) are central to cellular function, yet the precise recognition sites that drive binding remain difficult to predict. To pinpoint key residues that mediate recognition, we use a perturbation-based method within the elastic network model (ENM) framework called perturbation response scanning (PRS). This method works by applying a small external force to individual residues and quantifying the resulting structural responses via linear response theory. Residues that exhibit a high propensity to detect these perturbations and respond via altered structural dynamics are designated as sensor residues. We hypothesize that these dynamically sensitive residues are structurally predisposed to receive and transduce mechanical signals, thus facilitating molecular recognition and complex formation. To overcome the limitations of a single static structure, we extend PRS to an ensemble of monomeric conformations for each protein. This ensemble-based approach effectively captures functional dynamics and identifies sensor residues that are consistently responsive across different conformational states. Applying this method to a dataset of monomers that form non-obligate complexes, we demonstrate that the sensor residues are significantly enriched at known protein-protein interfaces. These residues map to flexible surface-exposed regions known to serve as recognition hotspots. Our findings suggest that this perturbation-based ENM approach provides a robust method for detecting PPI recognition sites, offering new insights into the dynamic nature of molecular recognition.
Parwana et al. (Sun,) studied this question.