HIV-1 vaccine design through minimizing envelope metastability

Abstract

Overcoming envelope metastability is crucial to trimer-based HIV-1 vaccine design. Here, we present a coherent vaccine strategy by minimizing metastability. For 10 strains across five clades, we demonstrate that the gp41 ectodomain (gp41_ECTO) is the main source of envelope metastability by replacing wild-type gp41_ECTO with BG505 gp41_ECTO of the uncleaved prefusion-optimized (UFO) design. These gp41_ECTO-swapped trimers can be produced in CHO cells with high yield and high purity. The crystal structure of a gp41_ECTO-swapped trimer elucidates how a neutralization-resistant tier 3 virus evades antibody recognition of the V2 apex. UFO trimers of transmitted/founder viruses and UFO trimers containing a consensus-based ancestral gp41_ECTO suggest an evolutionary root of metastability. The gp41_ECTO-stabilized trimers can be readily displayed on 24- and 60-meric nanoparticles, with incorporation of additional T cell help illustrated for a hyperstable 60-mer, I3-01. In mice and rabbits, these gp140 nanoparticles induced tier 2 neutralizing antibody responses more effectively than soluble trimers.

Fig. 1. Characterization of ExpiCHO-produced native-like Env trimers.

(A) SEC profiles of ExpiCHO-expressed, GNL-purified BG505 SOSIP.664 trimer, HR1-redesigned trimer (HR1 redesign 1), and UFO trimer from a Superdex 200 16/600 column. Transient expression in 100-ml ExpiCHO cells was used. (B) BN-PAGE of Env proteins for the three aforementioned trimers. The fractions used for antigenic profiling are circled by black dotted lines on the gel. (C) Antigenic profiles of the purified trimers measured against a panel of representative bNAbs and non-NAbs, with additional antibody binding profiles shown in fig. S1. Sensorgrams were obtained on an Octet RED96 using a trimer titration series of six concentrations (200 to 6.25 nM by twofold dilution). (D) Hydrophilic interaction liquid chromatography (HILIC)–UPLC profiles of the enzymatically released N-linked glycans of the HR1-redesigned trimer produced in ExpiCHO and 293 F cells followed by GNL and SEC purification. Oligomannose-type and hybrid glycans (green) were identified by their sensitivity to endoglycosidase H (Endo H) digestion. Peaks corresponding to complex-type glycans are shown in pink. The peaks are integrated, and the pie charts summarize the quantification of the peak areas. RF, retention factor. (E) Heavy-chain germline gene usage of the mouse antibody repertoire primed by ExpiCHO and 293 F–expressed BG505 gp140 trimers containing the HR1 redesign. The percentage of each germline gene family is plotted as a histogram, with a cartoon picture showing the immunization scheme. The results from the four mice in each group are colored in gray (M1, M5), cyan (M2, M6), light green (M3, M7), and orange (M4, M8). w0, week 0; w3, week 3; w6, week 6.

Fig. 2. Biochemical and biophysical characterization of the UFO-BG trimers for diverse HIV-1 strains.

(A) Design (left) and schematic representation (right) of the UFO-BG trimers. As shown on the left, BG505 gp41_ECTO of the UFO design is used to stabilize gp120s from other HIV-1 Envs in a hybrid form of gp140 trimer designated UFO-BG. The redesigned HR1 bend is highlighted in magenta. (B) SEC profiles of SOSIP, UFO, and UFO-BG trimers derived from 10 Envs across five subtypes (A, B, C, B/C, and A/E) following 100-ml ExpiCHO expression and GNL purification. The yield (in milligrams) of SEC-purified trimer protein (fractions corresponding to 53 to 57 ml) obtained from a 100-ml ExpiCHO expression is listed for each of the three trimer designs (SOSIP, UFO, and UFO-BG). (C) BN-PAGE of Env proteins after GNL purification but before SEC and of purified trimers following SEC and BN-PAGE for eight UFO-BG constructs. Two recombinant strains, B/C CN54 and A/E 95TNIH022, were not included due to low purity. (D) DSC analysis of eight UFO-BG trimers following GNL and SEC purification. Three thermal parameters (T_m, T_1/2, and T_onset) are listed for each trimer construct.

Fig. 3. Structural characterization of the UFO-BG trimers derived from diverse HIV-1 Envs.

(A) Crystal structure of a clade B tier 3 H078.14 Env spike determined at a resolution of 4.6 Å. Molecular surface of the H078.14 UFO-BG trimer in complex with bNAb Fabs PGT124 and 35O22 is shown on the left (top view and side view), with a ribbon model of the gp140 protomer and two Fabs shown in the middle, and a zoomed-in view of the redesigned HR1 bend (alone and superimposed onto two available structures, 5JS9 and 5CEZ) on the right. (B) Reference-free 2D class averages derived from negative-stain EM of eight UFO-BG trimers produced in ExpiCHO cells followed by GNL and SEC purification, with the full sets of images shown in fig. S3B. The percentage of native-like trimers is indicated for each trimer construct. aa, amino acid.

Fig. 4. Antigenic map of diverse HIV-1 strains and structure-informed optimization of the H078.14 UFO-BG trimer.

(A) Antigenic profiles of 10 UFO trimers (left) and 10 UFO-BG trimers (right) against 11 bNAbs and 8 non-NAbs. Sensorgrams were obtained from an Octet RED96 using a trimer titration series of six concentrations (200 to 6.25 nM by twofold dilution) and are shown in fig. S4. The peak values at the highest concentration are summarized in the matrix, in which cells are colored in red and green for bNAbs and non-NAbs, respectively. Higher color intensity indicates greater binding signal measured by Octet. To facilitate antigenic comparison between UFO and UFO-BG trimers, the average peak value (AVE) and SD are listed for each antibody in the two matrices. P values calculated from paired t test are listed in the last column of the UFO-BG matrix, with statistically significant P values (<0.05) highlighted in gray. (B) Top-down view of the H078.14 UFO-BG trimer apex and zoomed-in view of the H078.14 V1V2 apex superposed with that of the BG505 SOSIP.664 trimer (PDB: 5CEZ). Glycans at N130, N160, and N171 are labeled for H078.14. The turn between strands B and C of H078.14 and the V2 loop of BG505 are shown as dotted lines in blue and orange, respectively. (C) Sequence alignment of V1V2 regions from BG505, 6240.08.TA5.4622 (clade B), WT H078.14 (clade B), and a modified H078.14 (termed H078.14Mut) with mutations at positions 156, 170, and 172 colored in red and “KDGS” deletion at the turn of strands B and C highlighted in yellow. (D) Characterization of an H078.14Mut construct that also contains a disulfide bond (I201C-A433C) to prevent CD4-induced conformational changes. Trimers produced in 100-ml ExpiCHO cells are characterized by SEC (left), BN-PAGE (middle), and antigenic evaluation against the V2 apex–directed bNAbs PGDM1400 and PG16 and a CD4i-specific non-NAb 17b (right). The direction and magnitude of the change of peak binding signal (in nanometers) are labeled on the sensorgrams of the H078.14Mut UFO-BG trimer, with an arrow colored in red and green for bNAbs and non-NAbs, respectively.

Fig. 5. Ferritin nanoparticles presenting diverse UFO-BG trimers and I3-01–based gp140 nanoparticles with embedded T cell help signal.

(A) Surface model of UFO-BG gp140-ferritin nanoparticle, with gp120, BG505 gp41_ECTO of the UFO design, and ferritin colored in cyan, magenta, and gray, respectively. (B) BN-PAGE of eight UFO-BG-FR nanoparticles after a single-step 2G12 affinity purification. (C) Reference-free 2D class averages derived from negative-stain EM of five representative UFO-BG-FR nanoparticles. (D) Antigenic profiles of five representative UFO-BG-FR nanoparticles against six bNAbs and four non-NAbs. Sensorgrams were obtained on an Octet RED96 using a trimer titration series of six concentrations (starting at 35 nM by twofold dilution) and are shown in fig. S6. The peak values at the highest concentration are summarized in the matrix, in which cells are colored in red and green for bNAbs and non-NAbs, respectively. Higher color intensity indicates greater binding signal measured by Octet. (E) Left: Surface model of the I3-01 nanoparticle (colored in gray), with the subunits surrounding a front-facing fivefold axis highlighted in dark gray and three subunits forming a threefold axis colored in sky blue, magenta, and green, respectively. Middle: Spacing between N termini of three I3-01 subunits surrounding a threefold axis (top view) and the anchoring of a gp140 trimer onto three I3-01 subunits by flexible peptide linkers (indicated by black dotted lines). Right: Schematic representation of I3-01 nanoparticle constructs containing both gp140 and a T helper epitope, with sequences listed for three such T helper epitopes: PADRE, D, and TpD. (F) SEC profiles of I3-01 nanoparticles presenting an HR1-redesigned BG505 trimer (termed gp140.664.R1) with a 10–amino acid GS linker (left) and three T helper epitope linkers (right). The yields (in milligrams) of nanoparticle protein obtained from a 100-ml ExpiCHO expression and after 2G12 and SEC purification are labeled on the SEC profiles. (G) BN-PAGE of two I3-01 nanoparticles, containing a GS linker and a T helper epitope linker (PADRE), after a single-step 2G12 affinity purification. (H) Micrograph derived from negative-stain EM of 2G12-purified I3-01 nanoparticle presenting an HR1-redesigned BG505 trimer with a PADRE linker (termed gp140.664.R1-PADRE-I3-01). (I) Antigenic profiles of BG505 gp140.664.R1-PADRE-I3-01 nanoparticle against six bNAbs and four non-NAbs. Sensorgrams were obtained on an Octet RED96 using a trimer titration series of six concentrations (starting at 14 nM by twofold dilution).

Fig. 6. Evaluation of trimers and nanoparticles in B cell activation assays and in two small-animal models.

(A) Ca²⁺ mobilization by various gp140 nanoparticles in B cell transfectants carrying PGT145, PGT121, and VRC01 bNAb receptors. WEHI231 cells expressing a doxycycline-inducible form of bNAb BCR were stimulated with anti-BCR antibodies or the indicated antigens at a concentration of 10 μg ml⁻¹: anti-human Ig κ-chain F(ab′)₂; anti-mouse IgM; an UFO-BG-FR nanoparticle derived from a clade A, B, C, B/C, or A/E strain; or a BG505 gp140-PADRE-I3-01 nanoparticle containing a redesigned HR1 bend within gp41_ECTO. (B) Top: Assessment of immunogenicity in WT mice. Schematic representation of the mouse immunization protocol. Bottom: Neutralization curves for groups 3, 6, and 10, which correspond to a scaffolded full-length gp140 trimer (gp140.681.R1-1NOG), a gp140-ferritin nanoparticle (gp140.664.R1-FR), and a gp140-I3-01 nanoparticle with T cell help (gp140.664.R1-PADRE-I3-01), respectively. Structural models of these three immunogens are placed next to their group-combined neutralization curves. The neutralization curves are also included for individual mice whose serum IgGs neutralized BG505.T332N. (C) Assessment of immunogenicity in rabbits. Schematic representation of the rabbit immunization protocol (top), longitudinal analysis of midpoint titers of antibodies reactive with the HR1-redesigned BG505 trimer (gp140.664.R1) and an N332 nanoparticle probe (bottom, left), and longitudinal analysis of neutralization against autologous tier 2 BG505.T332N and clade B tier 1 SF162 (bottom, right). Percent neutralization (%) at the 50-fold plasma dilution and ID₅₀ (50% inhibitory dose) are plotted for BG505 and SF162, respectively. An unpaired t test was performed to determine whether trimer and ferritin groups were significantly different (P < 0.05) in plasma binding and neutralization. P values are shown for time points w6, w14, w22, and w30, with asterisks indicating the level of statistical significance. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Detailed plasma ELISA and neutralization curves are shown in fig. S7, D and E. −d10, day −10.