Single-Chain Soluble BG505.SOSIP gp140 Trimers as Structural and Antigenic Mimics of Mature Closed HIV-1 Env

Abstract

Similar to other type I fusion machines, the HIV-1 envelope glycoprotein (Env) requires proteolytic activation; specifically, cleavage of a gp160 precursor into gp120 and gp41 subunits creates an N-terminal gp41 fusion peptide and permits folding from an immature uncleaved state to a mature closed state. While the atomic-level consequences of cleavage for HIV-1 Env are still being determined, the uncleaved state is antigenically distinct from the mature closed state, and cleavage has been reported to be essential for mimicry of the mature viral spike by soluble versions of Env. Here we report the redesign of a current state-of-the-art soluble Env mimic, BG505.SOSIP, to make it cleavage independent. Specifically, we replaced the furin cleavage site between gp120 and gp41 with Gly-Ser linkers of various lengths. The resultant linked gp120-gp41 constructs, termed single-chain gp140 (sc-gp140), exhibited different levels of structural and antigenic mimicry of the parent cleaved BG505.SOSIP. When constructs were subjected to negative selection to remove subspecies recognized by poorly neutralizing antibodies, trimers of high antigenic mimicry of BG505.SOSIP could be obtained; negative-stain electron microscopy indicated these to resemble the mature closed state. Higher proportions of BG505.SOSIP-trimer mimicry were observed in sc-gp140s with linkers of 6 or more residues, with a linker length of 15 residues exhibiting especially promising traits. Overall, flexible linkages between gp120 and gp41 in BG505.SOSIP can thus substitute for cleavage, and sc-gp140s that closely mimicked the vaccine-preferred mature closed state of Env could be obtained.

Importance: The trimeric HIV-1 envelope glycoprotein (Env) is the sole target of virus-directed neutralizing antibody responses and a primary focus of vaccine design. Soluble mimics of Env have proven challenging to obtain and have been thought to require proteolytic cleavage into two-component subunits, gp120 and gp41, to achieve structural and antigenic mimicry of mature Env spikes on virions. Here we show that replacement of the cleavage site between gp120 and gp41 in a lead soluble gp140 construct, BG505.SOSIP, with flexible linkers can result in molecules that do not require cleavage to fold efficiently into the mature closed state. Our results provide insights into the impact of cleavage on HIV-1 Env folding. In some contexts such as genetic immunization, optimized cleavage-independent soluble gp140 constructs may have utility over the parental BG505.SOSIP, as they would not require furin cleavage to achieve mimicry of mature Env spikes on virions.

FIG 1

Design of single-chain gp140 constructs. (A) Single-chain gp140 constructs that retain a linkage between gp120 and gp41 in BG505.SOSIP (top) were designed by replacing either both the furin cleavage site (residues 508 to 511) and adjacent residue segments (middle, orange) or only the cleavage site (sc-gp140) (bottom, red) by a flexible peptide linker with variable length. Highlighted are the SOSIP mutations (SOS, a disulfide between residues 501 and 605; IP, an Ile-to-Pro mutation at residue 559) as well as a designed glycosylation site at residue 332 and the truncation of gp140 at residue 664. (B) Mapping of the sc-gp140 design concept onto a BG505.SOSIP structure (PDB ID 4TVP). The three protomers are shown in different shades of gray, with two of the protomers shown as transparent surface and one protomer shown as cartoon ribbon. The residues between 505 and 518 are not observed in the structure and are marked as dotted lines, while the designed flexible linker replacing residues 508 to 511 is shown in red.

FIG 2

Characterization of transiently expressed and purified BG505.SOSIP sc-gp140. (A) Gel filtration profiles for selected BG505.SOSIP sc-gp140 constructs and wild-type BG505.SOSIP. Two rounds of size exclusion chromatography (SEC) were performed for each of the constructs, as described in Materials and Methods. (B) SDS-PAGE analysis of purified sc-gp140 under nonreducing (left) and reducing (right) conditions. Lanes M, molecular weight markers. Bands corresponding to gp41 can be seen under reducing conditions in the BG505.SOSIP lane and are highlighted with a bracket and an arrow. (C) DSC measurements for selected constructs (colored curves) after F105 negative selection are shown along with T_m temperatures and ΔH denaturation enthalpies.

FIG 3

Negative-stain EM of BG505 sc-gp140 constructs. (A) Example of counting trimer structures in EM images. Representative images (e.g., left) were classified (right) according to the presence of trimer versus aberrant structures. The criteria for identifying an aberrant structure were that (i) the negative stain has to be strong enough to clearly identify it as a particle, (ii) it should not have 3-fold symmetry, and (iii) it has to be larger than the native trimer in at least one dimension. We note that whereas the top views of trimer structures are structurally well defined, a number of factors such as staining blotches, side views of native trimers, etc., could cause a structure to be inadvertently categorized as aberrant. Trimer structures mimicking closed mature Env are shown in small blue circles, whereas aberrant structures are shown in larger white circles. Large black circles show examples of particles that were not classified as native trimers but were not counted as aberrant structures because they did not fulfill all of the criteria. (B) EM analysis was performed on proteins before and after F105 negative selection (see Materials and Methods). Shown are images for three of the sc-gp140 constructs (with 1-, 6-, and 15-residue linkers), BG505.SOSIP, and gp140-foldon after F105 negative selection. Representative structures are shown as magnified insets. The scale bar represents 50 nm in the main panels. (C) Classification of negative-stain EM for selected sc-gp140 constructs. Structures were analyzed before and after F105 negative selection (F105-selection). Constructs were categorized according to the fraction of “good” trimer structures observed: −, <10%; +, 10 to 50%; ++, 50 to 70%; +++, 70 to 85%; n/a, experiment not performed.

FIG 4

Antigenic characterization of purified BG505.SOSIP sc-gp140 constructs by biolayer interferometry. (A) The panel of antibodies included V1V2-targeting broadly neutralizing antibodies VRC26.09 and PGT145, CD4-binding-site poorly neutralizing antibody F105, CD4i poorly neutralizing antibody 17b, and gp120-gp41 antibodies PGT151, 35O22, and 8ANC195, as well as CD4-binding-site antibody VRC01 as a control. The heat map (white to red) is colored according to the maximum biolayer interferometry response observed in an Octet binding assay; values for the gp120(full-length) control are not directly comparable to the values for the gp140 molecules due to its substantially lower mass. (B) Binding kinetics for BG505.SOSIP and 15-sc gp140 with VRC26.09 and PGT145.

FIG 5

Antigenicity characterization by ECLIA. An extended panel of both broadly and poorly neutralizing antibodies was assessed using ECLIA for cleaved (black open circles) and 15-sc gp140 (red filled circles) BG505.SOSIP constructs.

FIG 6

Antigenicity characterization by ELISA. An extended panel of both broadly and poorly neutralizing antibodies was assessed using lectin capture ELISA for cleaved (black open circles) and 15-sc gp140 (red filled circles) BG505.SOSIP constructs.