De Novo Design of High-Affinity Minibinders Targeting the Occluded Head-Interface Epitope of Influenza A Hemagglutinin
Abstract
Influenza A viruses pose a persistent global health threat due to rapid antigenic drift in the hemagglutinin (HA) globular head. The monomer-monomer head-interface epitope is highly conserved across epidemic H3N2 strains but sterically occluded within the intact HA trimer, rendering it inaccessible to full-size monoclonal antibodies (IgG, ~150 kDa). We report a scaled computational pipeline for de novo design of small protein binders (minibinders) targeting this occluded site. An initial 34-candidate library produced a lead with high local fold confidence (pLDDT = 0.898) but failed interface docking (i_PTM = 0.167, i_PAE = 26.11 Å, RMSD = 11.60 Å). To address this, we generated 300 RFdiffusion backbones across multiple topology classes, applied geometric pre-filtering to isolate 64 scaffolds, and designed 512 candidate sequences using ProteinMPNN across multiple sampling temperatures. High-throughput screening via AlphaFold2 initial-guess prediction identified 13 candidates clearing all quality thresholds (i_PTM > 0.50, i_PAE < 10.0 Å, complex RMSD < 2.0 Å). The lead candidate, Design 142_n4, achieved pLDDT = 0.912, i_PTM = 0.724, i_PAE = 4.82 Å, and 1.42 Å complex RMSD. PyMOL contact modeling indicates a pre-organized mixed α/β scaffold inserting into the inter-monomer groove with hydrophobic packing at target hotspots PHE 58, TRP 186, and LEU 190. All results are computational; no wet-lab validation has been performed. Code and pipeline scripts are available at https://github.com/NolanYu1111/ha-minibinder-design
