De novo protein design has been increasingly successful in expanding beyond nature's sequence and structural space. However, most de novo designed proteins lack biological function, in part due to the structural complexity required for functional purposes. An important domain where protein design has raised expectations was on the induction of precise antibody responses that may lead to improved vaccines. Here, we showcase two computational design approaches to stabilize irregular and discontinuous binding motifs in de novo designed immunogens and tested them for the induction of respiratory syncytial virus neutralizing antibodies in vivo. The designs mimic the native conformations of the neutralization epitopes with sub-angstrom accuracy. In vivo, cocktail formulations of the immunogens induce robust neutralizing serum responses targeting three epitopes, and re-focus pre-existing antibody responses towards bona fide neutralization epitopes. Our work provides a blueprint for epitope-centric vaccine design for pathogens that have frustrated traditional vaccine development efforts, and a general methodological pipeline to create novel proteins with functional sites within tailored protein topologies.Recently, Correia and colleagues (12) have shown that computationally designed immunogens could elicit epitope-specific responses. The RSVF antigenic site II, a linear helixturn-helix motif, was transplanted onto a heterologous protein scaffold, which elicited nAbs in non-human-primates (NHPs) after repeated boosting immunizations. Despite being a proofof-principle for the induction of functional antibodies using a computationally designed immunogen, several major caveats emerged; namely, the lack of applicability of the computational approach to structurally complex epitopes, and the inconsistent neutralization titers observed in the immunogenicity studies.