Deep mutational scanning of Env from the HIV-1 strain DU422 for binding to CD4 and PG16

The HIV-1 genome gains access to the inside of a cell via the mechanism of the viral spike protein Env, which undergoes a series of major conformational rearrangements after binding target receptors that ultimately drive virus-cell membrane fusion. Env is expressed as a heterogenous ensemble of conformations, which can inappropriately misdirect the host immune response towards the production of non-protective, strain-specific antibodies. Potent, broadly neutralizing antibodies (bnAbs) frequently recognize a 'closed' Env conformation, and therefore Env has undergone significant engineering to stabilize the closed state for vaccine incorporation. Previously, we used deep mutational scanning of Env from a prototypical tier 1 clade B strain (BaL) to characterize the sequence-activity landscape for binding to PG16, a bnAb that preferentially binds the closed state. Mutations were identified that increased expression of closed Env and reduced conformational heterogeneity, but these mutations were only partially transferable to Env sequences from other strains. To generate an expanded set of mutations that may be broadly applicable to diverse HIV-1 strains, we present here the deep mutational scanning of Env from the tier 2 clade C strain DU422 for interactions with CD4 and PG16. Residues across the trimerization domain and trimer interface have low mutational tolerance for maintaining PG16 recognition. New mutations are identified that enhance presentation of the closed Env conformation, and these are applied to Env sequences spanning multiple clades and tiers. Overall design: The extracellular domains of Env (referred to here as gp140) from the clade C HIV-1 DU422 strain were encoded by a synthetic codon-optimized gene with a CD5 leader sequence, and fused at the C-terminus via a glycine/serine-rich linker to a hexahistidine tag and canonical transmembrane helix from class I MHC for surface display. The Env sequence was diversified by overlap-extension PCR to generate three single-site saturation mutagenesis libraries spanning the full length of gp140. The libraries were expressed in human Expi293F cells (a suspension derivative of HEK 293) and sorted by FACS for binding to soluble CD4 and bnAb PG16. Enrichment ratios were calculated for all single amino acid substitutions by comparing the frequencies of sequence variants in the sorted cell populations (from RNA transcripts) with the naive libraries (plasmid DNA). Experiments were independently duplicated.