Guiding Competitive Binding Assays Using Protein–Protein\nInteraction Prediction: The HER2–Affitin Use Case

Affitins are a class\nof small artificial proteins, designed as\nalternatives to antibodies for therapeutic, diagnostic, and biotechnological\napplications. Recent patents by Bracco Imaging S.p.A have demonstrated\nthe potential of two engineered affitins for designing imaging probes\nto detect and monitor human epidermal growth-factor receptor 2 (HER2)\nlevels <i>in vivo</i>. Targeting HER2 is critical, as its\noverexpression is linked to poor prognosis of several cancer diseases,\nmaking it a key marker for treatment strategies and diagnostic tools.\nInterestingly, these affitins do not compete with the commonly used\nmonoclonal antibodies trastuzumab and pertuzumab for HER2 binding\nsites, allowing their concurrent use <i>in vivo</i> and\nmaking them suitable for imaging or diagnostic purposes. Since these\ntwo affitins compete for the same yet unidentified binding site on\nHER2, structural insights into these interactions are essential for\nfacilitating the design and development of more effective diagnostic\ntools and treatments. In this study, we used protein–protein\ndocking and molecular dynamics simulations to model the binding of\nthese affitins to HER2. The stability of the predicted complexes was\nquantified by using the DockQ parameter, a widely used metric for\nevaluating protein–protein docking predictions. The docking\nposes were then compared with HER2 sites likely to interact with a\nprotein partner, as predicted by the matrix of local coupling energies\nmethod. The combination of these two computational methods allowed\nfor the identification of the most likely docking poses. Comparative\nanalysis with HER2-protein complexes from the Protein Data Bank suggests\nthat both affitins may bind HER2 at the same epitopes as an antibody\nfragment and an affibody. These findings indicate that targeted competitive\nbinding assays could efficiently reduce the experimental efforts to\nmap the HER2–affitin interactions. The computational approach\nproposed in this study not only provides insights into this specific\ncase but also establishes a robust framework applicable for facilitating\nthe structural modeling and interaction prediction of other affitin–protein\nsystems.