Design Trade-Offs between Target Binding and Self-Association Revealed by De Novo SARS-CoV-2 Binders

Abstract De novo protein binders have emerged as a powerful platform for molecular recognition, yet achieving high affinity while maintaining favorable biophysical properties remains a major challenge. Here, using the SARS-CoV-2 receptor-binding domain (RBD) as a model target, we generated two de novo protein binder design campaigns that deliberately imposed different trade-offs between binding and monomeric behavior, and benchmarked them against a set of experimentally validated binders. Through integrated experimental and computational analyses, we show that binder design is constrained by two distinct failure modes that require different optimization strategies. None of the 22 original designs showed measurable binding to the RBD, with one class remaining monomeric but failing to achieve sufficient target binding, while the other had more competent binding interfaces but exhibited excessive surface hydrophobicity that promoted self-association. By integrating AlphaFold-based predictions, molecular dynamics simulations, docking analyses, and surface aggregation propensity (SAP) calculations, we establish complementary computational signatures that distinguish these failure modes and reveal their underlying structural determinants. Guided by these insights, we prospectively redesigned a representative failed binder by reducing noninterface hydrophobicity while preserving folding and binding interface quality. Experimental characterization by SEC-MALS, surface plasmon resonance, and circular dichroism confirmed that the redesigned variants were stable monomers and acquired low μM binding affinities. Together, our results define the mechanistic trade-offs between binding and self-association in de novo protein binder design and provide a framework for diagnosing and rationally overcoming common design failure modes.