A next-generation cleaved, soluble HIV-1 Env trimer, BG505 SOSIP.664 gp140, expresses multiple epitopes for broadly neutralizing but not non-neutralizing antibodies

Abstract

A desirable but as yet unachieved property of a human immunodeficiency virus type 1 (HIV-1) vaccine candidate is the ability to induce broadly neutralizing antibodies (bNAbs). One approach to the problem is to create trimeric mimics of the native envelope glycoprotein (Env) spike that expose as many bNAb epitopes as possible, while occluding those for non-neutralizing antibodies (non-NAbs). Here, we describe the design and properties of soluble, cleaved SOSIP.664 gp140 trimers based on the subtype A transmitted/founder strain, BG505. These trimers are highly stable, more so even than the corresponding gp120 monomer, as judged by differential scanning calorimetry. They are also homogenous and closely resemble native virus spikes when visualized by negative stain electron microscopy (EM). We used several techniques, including ELISA and surface plasmon resonance (SPR), to determine the relationship between the ability of monoclonal antibodies (MAbs) to bind the soluble trimers and neutralize the corresponding virus. In general, the concordance was excellent, in that virtually all bNAbs against multiple neutralizing epitopes on HIV-1 Env were highly reactive with the BG505 SOSIP.664 gp140 trimers, including quaternary epitopes (CH01, PG9, PG16 and PGT145). Conversely, non-NAbs to the CD4-binding site, CD4-induced epitopes or gp41ECTO did not react with the trimers, even when their epitopes were present on simpler forms of Env (e.g. gp120 monomers or dissociated gp41 subunits). Three non-neutralizing MAbs to V3 epitopes did, however, react strongly with the trimers but only by ELISA, and not at all by SPR and to only a limited extent by EM. These new soluble trimers are useful for structural studies and are being assessed for their performance as immunogens.

Conflict of interest statement

I have read the journal's policy and have the following conflicts: A patent application on BG505 SOSIP.664 gp140. This does not alter our adherence to all PLoS Pathogens policies on sharing data and materials.

Figures

Figure 1. Design and biochemical characterization of BG505 SOSIP.664 gp140 trimers.

(A) Linear representation of the BG505 gp160, SOSIP.664 gp140, SOSIP.664-D7324 gp140 and gp120-D7324 Env proteins. Modifications compared to the original BG505 gp160 sequence are indicated in red and mentioned in the text. The following changes were made to the wild type BG505 amino acid sequence: 1) The tissue plasminogen activator (tPA) signal peptide replaced the natural one; 2) the gp41 transmembrane (TM) and cytoplasmic tail (CT) domains were deleted to create a soluble gp140; 3) the A501C and T605C substitutions were made to form a disulfide bond between gp120 and gp41ECTO ; 4) the I559P substitution was included to promote trimerization , ; 5) an optimal cleavage site (RRRRRR; R6) replaces the natural one, REKR ; 6) truncation of the MPER from residue-664 prevents aggregation , ; 7) the T332N substitution facilitates binding of bNAbs dependent on glycan-N332. The D7324- and His-tags are indicated in yellow. Env sub-domains are indicated: 5 conserved domains (C1–C5); 5 variable domains (V1–V5); heptad repeats 1 and 2 (HR1, HR2); the membrane proximal external region (MPER); the transmembrane domain (TM); and the cytoplasmic tail (CT). The glycan assignments in Env are based on previous studies using gp120 , , , but may be different for trimeric Env . The amino acid sequence of BG505 SOSIP.664 is given in Fig. S1. (B) SEC profile of 2G12-purified BG505 SOSIP.664 gp140 expressed in CHO-K1 cells. A Superdex 200 26/60 column was used. (C) Analytical SEC profile of 2G12/SEC-purified BG505 SOSIP.664 trimer re-run on a Superose 6 10/30 column. (D) BN-PAGE analysis of CHO-K1 expressed, 2G12-purified BG505 SOSIP.664 gp140, stained by Coomassie blue. The m.w. of marker (M) proteins (thyroglobulin and ferritin) are indicated. (E) BN-PAGE analysis of 2G12/SEC-purified BG505 SOSIP.664 gp140, stained by Coomassie blue. (F) SDS-PAGE analysis using a 4–12% Bis-Tris Nu-PAGE gel of 2G12/SEC-purified BG505 SOSIP.664 gp140, under non-reducing and reducing conditions, followed by Coomassie blue staining. (G) SDS-PAGE analysis using a 10% Tris-Glycine gel of 2G12/SEC-purified BG505 SOSIP.664 gp140, under non-reducing and reducing conditions, followed by silver staining. The conversion of the gp140 band to gp120 and the appearance of a gp41ECTO band under reducing conditions is indicative of cleavage.

Figure 2. Biophysical characterization of BG505 SOSIP.664 gp140 trimers.

DSC analysis of (A) purified BG505 SOSIP.664 gp140 trimers and (B) purified BG505 gp120 monomers. The melting profiles show that the trimer has a higher degree of stability than its monomeric counterpart, as its thermal transitions are initiated at a 14.4°C higher temperature. Raw data are shown in black, while the fitted curves from which Tm values were obtained are in red. (C) EM reconstruction of the BG505 SOSIP.664 gp140 trimer at 24-Å resolution. The 2D class averages and a Fourier Shell Correlation (FSC) curve are shown in Fig. S2.

Figure 3. BG505.T332N virus neutralization and ELISA binding to BG505 SOSIP.664 gp140 trimers by bNAbs or non-NAbs.

Midpoint neutralization concentrations (IC50, in ng/ml) were derived from single cycle experiments involving Env-pseudovirus infection of TZM-bl cells. The values represent the averages of 2–5 independent titration experiments, each performed in duplicate, with the standard error recorded. Half-maximal binding concentrations (EC50, in ng/ml) were derived from D7324-capture ELISAs. The values represent the averages of 2–6 independent single titration experiments, with the standard error recorded.

Figure 4. BG505 SOSIP.664 gp140 antigenicity by ELISA with bNAbs.

(A) Schematic representation of D7324-capture ELISAs using BG505 gp120-D7324 monomers and/or SOSIP.664-D7324 gp140 trimers. (B) Representative binding curves of bNAbs VRC01, VRC03, VRC06, VRC06b, PGV04, 3BNC117, 12A12, 45–46, 45–46W, 1NC9, 8ANC195, CH31, CH103, CH106, PGT121, PGT123, PGT125, PGT126, PGT128, PGT130, and also CD4-IgG2, to purified BG505 SOSIP.664-D7324 gp140 trimers. (C) Representative binding curves of bNAb 3BC315 with purified BG505 SOSIP.664-D7324 gp140 trimers and gp120-D7324 monomers. (D) Representative binding curves of quaternary structure dependent bNAbs PG9, PG16, CH01 and PGT145 to purified BG505 SOSIP.664-D7324 gp140 trimers and gp120-D7324 monomers. The legend is the same as for panel C. (E) BG505 SOSIP.664-D7324 gp140 trimers were 2G12-affinity purified and fractionated using a Superose 6 10/30 SEC column. The SEC fractions were analyzed for PG9, PG16, PGT145 and b6 binding by D7324-capture ELISA. Note that the scales on the y-axes and x-axes vary from MAb to MAb.

Figure 5. BG505 SOSIP.664 gp140 antigenicity by ELISA with non-bNAbs.

Representative binding curves of: (A) gp120-directed non-NAbs 15e, F91, F105 and b6 to purified BG505 SOSIP.664-D7324 gp140 trimers and gp120-D7324 monomers, with bNAbs 2G12 and PGT128 included as controls. (B) CD4i MAbs 17b, A32 and 412d to purified BG505 SOSIP.664-D7324 gp140 trimers and gp120-D7324 monomers in the absence (open symbols) and presence (closed symbols) of sCD4. (C) gp41-directed non-NAbs 7B2 and F240 to BG505 SOSIP.664-His and WT.664-His ( = gp41ECTO-His) proteins. Both proteins were expressed in the presence of furin, yielding cleaved gp140. The lack of the SOS disulfide bond results in gp120 shedding from gp41ECTO-His, as illustrated by the poor reactivity with VRC01 (left panel). (D) V3 MAbs 39F, 19b and 14e to purified BG505 SOSIP.664-D7324 gp140 trimers and gp120-D7324 monomers. Note that the scales on the y-axes and x-axes vary from MAb to MAb.

Figure 6. Correlation between MAb binding to BG505 SOSIP.664 gp140 trimers and BG505.T332N neutralization.

The midpoint binding concentrations (EC50) for MAb binding to BG505 SOSIP.664-D7324 gp140 trimers (y-axis) were plotted against the IC50 values for neutralization of the BG505.T332N Env-pseudotype virus (x-axis). The Pearson's correlation coefficient, r, was calculated using Prism software version 5.0. When accurate midpoint concentrations could not be calculated because of lack of binding or neutralization, the highest concentration tested was included in the correlation analysis (i.e., when the IC50 of neutralization was >30,000 ng/ml, a value of 30,000 ng/ml was used). The data points for NAbs are indicated in black, those for non-NAbs in gray. Also note that 11 data points are overlapping in the right upper corner; they were derived using MAbs that neither neutralized the virus nor bound the trimer (IC50 of neutralization >30,000 ng/ml, EC50 in ELISA >10,000 ng/ml). Only MAbs whose epitope could be shown to be present on at least one form of BG505 Env protein were included in this analysis; MAbs that were non-reactive, presumably because of sequence variation, were excluded. The Pearson's correlation was also calculated without the data points for 2G12, 39F, 14e and 19b, as discussed in the text. The fitted line is based on all data, i.e. including 2G12, 39F, 14e and 19b.

Figure 7. BG505 SOSIP.664 gp140 antigenicity by SPR.

BG505 SOSIP.664-His gp140 trimers were immobilized on NTA chips (A–D). The sensorgrams show the response (RU) over time (s) using IgGs at 1,000 nM (150,000 ng/ml) (A–C) or Fabs at 500 nM (25,000 ng/ml) (D). The association phase was 300 s and dissociation was followed over 600 s. (A) 2G12 (high), PGT135 (intermediate), PGV04 (high), b6 (marginal), b12 (undetectable) and F240 (undetectable); (B) PG9, PG16, PGT145 (all intermediate); (C) PGT121 (intermediate), PGT123 (high), PGT128 (high), 14e (low); (D) Fabs of PGV04 (intermediate), b6, b12, and F240 (undetectable). (E) In an alternative approach, Env-reactive MAb was captured by anti-Fc Ab on the chip and the responses to BG505 SOSIP.664 gp140 trimers (200 nM; 78,000 ng/ml) were followed: 2G12 (high), PGT128 (high), b6 (low), b12, F240 (undetectable). Each curve represents one of 2–3 similar replicates.

Figure 8. BG505 SOSIP.664 gp140 trimer antigenicity by ITC.

The top panels show the raw data and the bottom panel the binding isotherms for representative ITC binding experiments measuring the binding of BG505 SOSIP.664 gp140 trimers to: (A) PGT121 Fab, (B) PGT128 Fab, (C) 2G12 IgG (domain-exchanged). The thermodynamic parameters of binding are listed in Table 1.

Figure 9. BG505 SOSIP.664 gp140 trimer antigenicity by negative stain EM.

(A) EM reconstruction of the BG505 SOSIP.664 gp140 trimer in complex with Fab PGV04 at 23 Å resolution. The 2D class averages and the Fourier Shell Correlation (FSC) are shown in Fig. S3. The crystal structure of gp120 in complex with PGV04 (PDB:3SE9; [85]) was fitted into the EM density. (B) EM reconstruction of the BG505 SOSIP.664 gp140 trimer in complex with sCD4 and Fab 17b at 22 Å resolution. The 2D class averages and the Fourier Shell Correlation (FSC) are shown in Fig. S4. The crystal structure of gp120 in complex with sCD4 and 17b (PDB:1RZK; [86]) was fitted into the EM density. (C) Comparison of unliganded BG505 SOSIP.664 gp140 trimers with complexes of the same trimers with PGV04, PGT122, PG9, PGT135, or sCD4 with 17b. The unliganded trimer is shown in mesh. The EM reconstructions of PGT122, PG9, PGT135 with BG505 SOSIP.664 trimers, expressed in HEK293S cells, have been published elsewhere , , .

Figure 10. Negative stain EM data of the BG505 SOSIP.664 gp140 trimer in complex with Fabs b6, F240, 14e or 19b.

2D class averages of trimers and trimer∶Fab complexes, with 2D class averages of BG505 SOSIP.664 gp140 with PGV04 shown on the lower left panel for comparison. Examples of complexes are circled. Blue circle: no Fab bound to trimer; yellow circle: one Fab bound; red circle: two Fabs bound; green circle: three Fabs bound. For generating b6 complexes, BG505 SOSIP.664 (44 nM; 17,200 ng/ml) was incubated with Fab b6 (480 nM; 24,000 ng/ml) prior to imaging. For generating F240 complexes, BG505 SOSIP.664 (48 nM; 18,720 ng/ml) was incubated with Fab F240 (860 nM; 43,000 ng/ml) prior to imaging. For generating 14e complexes, BG505 SOSIP.664 (14 nM; 5460 ng/ml) was incubated with Fab 14e (240 nM; 12,000 ng/ml). For generating 19b images, BG505 SOSIP.664 (40 nM; 15,600 ng/ml) was incubated with Fab 19b (600 nM; 30,000 ng/ml). The results are summarized in Table 2.

Figure 11. BG505 SOSIP.664 antigenicity summary for bNAbs and non-NAbs.

The following scoring was used for neutralization: +: IC50<30,000 ng/ml; −: IC50>30,000 ng/ml. See Table 1 for details. The following scoring was used for SPR analyses: +: >70 RU; +/−: >30 RU, <70 RU; −: <30 RU, based on plateau estimates of 1,000 nM for IgG and 500 nM for Fab. See Fig. 7 for details. In SPR CD4 binding results were obtained with sCD4 not CD4-IgG2. b6 binding was absent with b6 Fab (−) and low with b6 IgG (+/−). The following scoring was used for ITC experiments: +: N>1.2 (except for PG9, PG16 and PGT145 where N>0.4) and Kd<300 nM; −: N<1.2 and Kd>300 nM, where N is the stoichiometry of binding. See Table 1 and Fig. 8 for details. . The following scoring was used for EM analyses: +: N = 3 (except for PG9, PG16 and PGT145 where N = 1) for >50% of the trimers; +/−: N = 3 for <50% of the trimers; −: N = 1,2 or 3 for <10% of the trimers, where N is the number of Fabs bound per trimer. See Table 2, Fig. 9 and Fig. 10 for details. Note that ITC data with PG16 and PGT127 were obtained using BG505 SOSIP.664 trimers produced in GnTI-defective HEK293S cells. No ITC experiments were performed with non-NAbs. The following scoring was used for ELISA experiments: +: EC50<10,000 ng/ml; EC50>10,000 ng/ml. See Figs. 3–5 for details.