AI-based HIV-1 Env designs devoid of gp41 mimic native gp120 pre-fusion Env 9395

Abstract Description Viral fusion glycoproteins, such as HIV-1 Env, a trimer of gp120 and gp41 heterodimers, are key targets for neutralizing antibodies but are inherently metastable due to conformational changes required for function. Stabilizing these proteins without compromising their immunogenicity has been a long-standing challenge. Here, we leveraged deep learning-based protein design to create a gp41-free HIV-1 Env trimer platform, or GFT. These designs replace the metastable gp41 subunits with de novo structures, stabilizing the trimer while preserving the pre-fusion, native-like conformation of gp120. GFTs retain broadly neutralizing antibody (bnAb) and germline precursor binding, assemble efficiently into icosahedral nanoparticles with exceptional structural integrity, and activate specific B cells in vitro. Immunization with GFTs in gl-VRC01 KI mouse models induced robust germinal center B-cell responses specific to the CD4 binding site. Remarkably, cryo-EM analysis of GFT designs 36G and 38G on nanoparticles confirmed structural fidelity to native-like Env trimers at atomic resolution. We also demonstrate conservation of bnAb engagement of GFTs and validate computational predictions using cryo-EM to drive further design. Our results establish a transformative platform for presenting stable, gp41-free HIV-1 Env trimers in a closed conformation, providing a promising approach for HIV vaccine development and other complex antigen design applications. Funding Sources This work was supported by the Bill and Melinda Gates Foundation and the CAVD network (INV-002916), and in part by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (Award Number P01 AI110657). Topic Categories Vaccines and Immunotherapy (VAC)