Abstract
CD4-based decoy approaches against HIV-1 are attractive options for long-term viral control, but initial designs, including soluble CD4 (sCD4) and CD4-Ig, were ineffective. To evaluate a therapeutic that more accurately mimics HIV-1 target cells compared with monomeric sCD4 and dimeric CD4-Ig, we generated virus-like nanoparticles that present clusters of membrane-associated CD4 (CD4-VLPs) to permit high-avidity binding of trimeric HIV-1 envelope spikes. In neutralization assays, CD4-VLPs were >12,000-fold more potent than sCD4 and CD4-Ig and >100-fold more potent than the broadly neutralizing antibody (bNAb) 3BNC117, with >12,000-fold improvements against strains poorly neutralized by 3BNC117. CD4-VLPs also neutralized patient-derived viral isolates that were resistant to 3BNC117 and other bNAbs. Intraperitoneal injections of CD4-CCR5-VLP produced only subneutralizing plasma concentrations in HIV-1-infected humanized mice but elicited CD4-binding site mutations that reduced viral fitness. All mutant viruses showed reduced sensitivity to sCD4 and CD4-Ig but remained sensitive to neutralization by CD4-VLPs in vitro. In vitro evolution studies demonstrated that CD4-VLPs effectively controlled HIV-1 replication at neutralizing concentrations, and viral escape was not observed. Moreover, CD4-VLPs potently neutralized viral swarms that were completely resistant to CD4-Ig, suggesting that escape pathways that confer resistance against conventional CD4-based inhibitors are ineffective against CD4-VLPs. These findings suggest that therapeutics that mimic HIV-1 target cells could prevent viral escape by exposing a universal vulnerability of HIV-1: the requirement to bind CD4 on a target cell. We propose that therapeutic and delivery strategies that ensure durable bioavailability need to be developed to translate this concept into a clinically feasible functional cure therapy.