Improving the Immunogenicity of Native-like HIV-1 Envelope Trimers by Hyperstabilization

Abstract

The production of native-like recombinant versions of the HIV-1 envelope glycoprotein (Env) trimer requires overcoming the natural flexibility and instability of the complex. The engineered BG505 SOSIP.664 trimer mimics the structure and antigenicity of native Env. Here, we describe how the introduction of new disulfide bonds between the glycoprotein (gp)120 and gp41 subunits of SOSIP trimers of the BG505 and other genotypes improves their stability and antigenicity, reduces their conformational flexibility, and helps maintain them in the unliganded conformation. The resulting next-generation SOSIP.v5 trimers induce strong autologous tier-2 neutralizing antibody (NAb) responses in rabbits. In addition, the BG505 SOSIP.v6 trimers induced weak heterologous NAb responses against a subset of tier-2 viruses that were not elicited by the prototype BG505 SOSIP.664. These stabilization methods can be applied to trimers from multiple genotypes as components of multivalent vaccines aimed at inducing broadly NAbs (bNAbs).

Graphical abstract

Figure 1

Design, Antigenicity, Structure, and Conformational Dynamics of BG505 SOSIP.v5 Trimers Containing Novel Disulfide Bonds between gp120 and gp41 (A) Linear schematic of the BG505 SOSIP.664, SOSIP.v4, SOSIP.v5, SOSIP.v5.2 I201C-A433C, and SOSIP.v6 constructs. Modifications that create the SOSIP.664 construct are indicated in red (Sanders et al., 2013). The E64K and A316W substitutions added to make the SOSIP.v4 construct are colored blue (de Taeye et al., 2015). The engineered disulfide bonds in SOSIP.v5 are shown in purple. There are two variants of the SOSIP.v5 construct: the SOSIP.v5.1 has a disulfide bond between H72C and H564C and the SOSIP.v5.2 has a disulfide bond between A73C and A561C. The I201C-A433C disulfide bond, previously described by Kwon et al. (2015), was introduced into the SOSIP.v5.2 background. The E49C-L555C disulfide bond was introduced into the SOSIP.v5.2 construct. The resulting constructs are designated SOSIP.v5.2 I201C-A433C and SOSIP.v6, respectively. (B) SPR analysis of the binding of bNAbs PG16, PGT145, PGT151, and 35O22 to quaternary epitopes (upper panel), and of CD4-IgG2 and non-NAb 17b CD4i (±prior addition of CD4-IgG2) (lower panel), to the indicated BG505 trimer variants. (C) Crystal structure of the quaternary complex of BG505 SOSIP.664 H72C-H564C trimer (cyan) in complex with PGT122 Fab (pink), 35O22 Fab (orange), and NIH45-46 scFv (deep blue). The three different antibody Fab or scFv fragments are represented as colored surfaces, and the trimer N-linked glycans as cyan spheres. One of the three protomers is highlighted for clarity, and the gp41 central HR1 helix location proximal to the engineered H72C-H564C disulfide is marked with a rectangle. (D) Detail of the interaction between gp41 HR1 and gp120. A 2Fo-Fc composite omit map contoured at 1.0σ around gp41 HR1 (blue mesh) reveals ordered electron density (green mesh) for gp41 HR1_N residues 547–569 (dotted lines), implying that the H72C-H564C disulfide bond helps stabilize this region of the trimer in a ground-state prefusion conformation. The figure was rendered using PyMOL. (E) Differences in hydrogen/deuterium (H/D) exchange rates between the unliganded and CD4-bound forms of the indicated SOSIP variants. The difference between the two states is the difference obtained by subtracting the deuterons exchanged in the unliganded condition from the deuterons exchanged in the CD4-bound forms. While positive values indicate that the residues were destabilized in the presence of 2D-sCD4, negative values indicate stabilization. The net difference in H/D exchange as a sum of all time points is plotted for each observable peptide. Only differences that were outside the error range were included in the summation process. The same set of peptides was used for each trimer construct. The individual exchange plots are shown in Data S1A.

Figure 2

Immunogenicity of BG505 and ZM197M SOSIP.v5 Trimers in Rabbits (A) Schematic representation of immunization schedule. (B) Color coding for the immunogens tested. (C–I) Shows data for BG505-based immunogens. (J –P) Shows data for ZM197M. (C and J) Midpoint antibody binding titers (ED₅₀) as measured by D7324-capture ELISA for the trimer variants indicated on the x axis at the bottom of the figure, and by the color-coding scheme outlined in (B). (D–H and J–O) Neutralization of HIV-1 viruses in the TZM-bl assay by sera from animals immunized with BG505 or ZM197M trimer variants. The plots show ID₅₀ values, the serum dilution at which infectivity is inhibited by 50%. (D–H and K–O) Autologous viruses (D and K); SF162 heterologous tier-1A (E and L); 92RW020 heterologous tier-2 virus (F and M); SHIV162P3 heterologous tier-2 virus (G and N); and TV1.21 heterologous tier-2 virus (H and O). (I and P) The cumulative numbers of heterologous viruses neutralized with an ID₅₀ >40 by the five sera from each group are shown. In the particular case of the BG505 SOSIP.v5 group, where 15 animals were analyzed (SOSIP.v5.1: five animals and SOSIP.v5.2: ten animals), the number of heterologous hits was divided by three. Statistical tests were performed using Mann-Whitney t test for (E)–(H) and Kruskal-Wallis with Dunn’s post-test for (I) and (P). Note that Kruskal-Wallis tests were performed on the ID₅₀ values. The horizontal bars represent medians. Neutralization titers with SOSIP.v6 sera against 92RW020, SHIVP3, and TV1.21 (plotted in F–H and M–O) were obtained from two or three independent experiments performed in duplicate. The average values are plotted, and the data from the replicate experiments can be found in Table S4C.

Figure 3

Properties of the Autologous NAb Specificities (A) Neutralization depletion assays with sera from one BG505 SOSIP.v5.2 recipient and all five BG505 SOSIP.v6-immunized animals using the autologous BG505.T332N virus and a BG505 V3 peptide, as well as a set of different BG505 proteins. Neutralization of BG505.T332N from each individual sera was set at 100%, and the percent depletion was calculated when the V3 peptide or the different BG505 proteins were added in the assay. The percentage shown is the mean of two different experiments. (B) Neutralization assays with BG505.T332N PNGS mutants. The ID₅₀ of the individual sera with the parental BG505.T332N virus was set at 100%, and the percent reduction in ID₅₀ when the respective PNGSs are introduced is shown. See also Table S4B. (C) Neutralization of maternal MG505 viruses and heterologous TV1.21 mutant viruses. The ID₅₀s are shown. Strong neutralization (ID₅₀ > 1,000) is shown in red; 1,000 > ID₅₀ > 100 is shown in orange; 100 > ID₅₀ > 50 is shown in yellow, and no neutralization (ID₅₀ < 50) is shown in white.