Functional and structural characterization of a combination of pan-sarbecovirus antibodies with potent antiviral activity

Sarbecoviruses, a subgenus of coronaviruses, include strains with zoonotic spillover risk as exemplified by recent outbreaks (SARS-CoV-1, SARS-CoV-2). Monoclonal antibodies targeting conserved spike protein regions (RBD class 4, S2 fusion machinery) exhibit broad sarbecovirus neutralization, but their utility has been impacted by immune selection leading to escape and suboptimal neutralization. Using structure-conditioned machine-learning protein design we optimized two broadly neutralizing sarbecovirus antibodies to rescue the activity of a class 4 anti-RBD antibody against Omicron variants and improve the neutralization profile of an anti-S2 stem antibody, resulting in the identification of the lead candidates PRO-37587 and GB-0669. Both antibodies displayed remarkable resilience to SARS-CoV-2 evolution. Structural analyses elucidated the mechanistic basis of their broad neutralization profiles. In combination, these antibodies exhibited improved SARS-CoV-2 neutralization in vitro and in vivo and a higher barrier to resistance. These findings support further evaluation of such antibody combination as countermeasure against current and emerging sarbecoviruses.