Native-like Env trimers as a platform for HIV-1 vaccine design

Abstract

We describe the development and potential use of various designs of recombinant HIV-1 envelope glycoprotein trimers that mimic the structure of the virion-associated spike, which is the target for neutralizing antibodies. The goal of trimer development programs is to induce broadly neutralizing antibodies with the potential to intervene against multiple circulating HIV-1 strains. Among the topics we address are the designs of various constructs; how native-like trimers can be produced and purified; the properties of such trimers in vitro and their immunogenicity in various animals; and the immunization strategies that may lead to the eventual elicitation of broadly neutralizing antibodies. In summary, native-like trimers are a now a platform for structure- and immunology-based design improvements that could eventually yield immunogens of practical value for solving the long-standing HIV-1 vaccine problem.

Keywords: Env trimers; HIV-1 vaccines; neutralizing antibodies.

Figure 1

Schematic presentations of various native‐like trimer constructs. Linear schematic of the SOSIP.664, SOSIP.v4, DS‐SOSIP, SOSIP.664 E49C‐L555C, SC, and UFO gp140 constructs that are mentioned in the text. Each schematic is based on the BG505 genotype, with the conserved (C) and variable (V) domains of gp120 and the two heptad repeats (HR) of gp41_ECTO depicted in gray. The locations of N‐linked glycans are indicated above the construct bars. The N332 glycan that was knocked into the BG505 gene used to make every construct is highlighted in black (T332N). The assignment of the various glycans to the oligomannose (green), complex (purple), or mixed or hybrid (orange) type is based on data derived using BG505 SOSIP.664 trimers.124 The glycan site‐specific composition of the other trimers has not been determined and may differ to various extents from what is depicted. The SOS bond (501C‐605C) linking gp120 to gp41_ECTO and the I559P substitution in gp41_ECTO (not included in the UFO construct) are shown in black, as is the RRRRRR (R6) change to the Furin cleavage site (not present in the SC and UFO designs). The additional disulfide bonds engineered into the DS‐SOSIP and SOSIP.664 E49C‐L555C constructs are represented by solid red and dotted blue lines, respectively. The point substitutions, compared to SOSIP.664, that are present in the SOSIP.v4 constructs are shown in green (SOSIP.v4.1, E64K; SOSIP.v4.2, H66R).123 The flexible linker between gp120 and gp41_ECTO in the SC construct is indicated by an undulating purple line, and the same linker as well as an additional one in HR1 (HR redesign) that is present in the UFO construct are both in orange.128, 132 The NFL construct is essentially identical to the SC design except that it does not include the SOS bond and the flexible linker length varies slightly.132 All constructs terminate at residue 664102

Figure 2

Structure and glycosylation of the BG505 SOSIP.664 trimer and comparison with the membrane‐associated native JR‐FL trimer. (A) Structure of the BG505 SOSIP.664 trimer (4TVP)107 with the gp120 subunits in green and gp41_ECTO in blue, except for HR2 (orange). The position of the viral membrane is shown for orientation. The glycans were attached to the protein using GlyProt (http://www.glycosciences.de/glyprot/) according to the composition determinations described in Behrens et al.124 For glycans where the composition could not be determined in that study, we used information from other reports.126, 212, 213, 214 Overall, Man₉GlcNAc₂ glycans are shown at positions N156, N160, N234, N262, N276, N295, N332, N339, N363, N386, N392, and N448; Man₅GlcNAc₂ glycans at positions N197, N301, N355, and N411; and fucosylated biantennary complex glycans at position N88, N133, N137, N462, N611, N618, and N637. We note however, that the composition of many glycans is heterogeneous.124 We did not add a glycan to residue N625 because that position may be only partially occupied.126 We were unable to add glycans to positions 190, 190c, 398, and 406 because these asparagine residues were not resolved in the structure and hence are not included in the model. (B) Structure of one BG505 SOSIP.664 protomer (5CEZ)114 with the same color code as in panel‐A. Residues P559, C501, and C605 are shown as red spheres. (C) A cartoon rendering of panel‐B. (D) The same structure as in panel‐A is shown, but with the protein component in white as a space‐filling representation and with residues 241 and 289 (i.e. the location of the ‘glycan hole’) shown in red and indicated by arrows. (E) The protein structure of one BG505 SOSIP.664 protomer (as in panel‐B) is overlaid with the structure of the native, membrane‐associated JR‐FL trimer in white.89 (F) Detail of the overlay from panel‐E showing positions A501 and T605 in the JR‐FL trimer together with the cysteine substitutions (C501 and C605) and the linking disulfide bond that are present in the BG505 SOSIP.664 trimer. (G) Detail of the overlay from panel‐E showing part of gp41 and the effect the I559P substitution has on the local structure

Figure 3

Autologous NAbs induced in rabbits and guinea pigs by native‐like trimers of different designs and genotypes and purified by different methods. The figure depicts autologous NAb titers derived from published studies that have evaluated the immunogenicity of various designs of native‐like trimers in (A) rabbits or (B) guinea pigs. (A) The rabbit data are reported in the following papers: BG505 SOSIP.664 (red)84, 123, 150; BG505 SOSIP.v4.1 (red)123; B41 SOSIP.664 (blue)84, 123, 150; B41 SOSIP.v4.1 (blue)123; AMC008 SOSIP.664 (magenta)123; AMC008 SOSIP.v4.1 (magenta)123; JR‐FL SOSIP.664 (green)155; DU422 SOSIP.664 (purple)150; CZA97 SOSIP.664 (black).150 The JR‐FL SOSIP.664 trimers were used as single trimers (triangles) or presented on liposomes (inverted triangles).155 The DU422 and CZA97 data were derived when these trimers were used as boosting reagents in rabbits previously immunized with BG505 or B41 trimers.150 Their immunogenicity as de novo immunogens has not yet been reported. (B) The guinea pig data are reported in the following papers: BG505 SOSIP.664 (red)131, 154; BG505 NFL (red)131; BG505 NFL Xlink (red)131; JR‐FL SOSIP.664 (green)131; JR‐FL NFL (green)131; JR‐FL NFL Xlink (green)131; 16055 SOSIP.664 (orange).131 ‘XLink’ refers to glutaraldehyde cross‐linking of trimers prior to immunization.131 Four guinea pigs received BG505 SOSIP.664 trimers as complexes with the PGT145 bNAb (red diamonds).154 On both panels, the symbol shapes indicate the affinity column‐based method used to purify the trimers: 2G12 followed by SEC (circles); PGT145 (squares); lectin followed by non‐NAb depletion and then SEC (triangles, upright for soluble trimers, inverted for liposome presentation). All the NAb titers were derived using the TZM‐bl cell assay, but not all in the same laboratory. Our experience is that titers can vary by up to threefold when nominally the same TZM‐bl cell assay is carried out with the same sera and test viruses in different laboratories. Immunization conditions, including adjuvants, also vary between different research groups and sometimes between experiments, introducing another variable of unknown magnitude. For details, the primary papers must be consulted. In addition, the genotype affects both the immunogenicity of the trimer and the overall neutralization sensitivity of the autologous virus. All the virus genotypes referred to here have Tier‐2 neutralization resistance phenotypes except for AMC008 (Tier‐1B), but there is a spectrum of sensitivities within the Tier‐2 classification. Overall, the multiple variables across the totality of experiments mean that this figure should be used only as a guide to the relative immunogenicity of different trimers of various designs and genotypes. Only directly comparative experiments, whether completed or for the future, allow more precise conclusions to be drawn