Molecular insights into antibody-mediated protection against the prototypic simian immunodeficiency virus

Abstract

SIVmac239 infection of macaques is a favored model of human HIV infection. However, the SIVmac239 envelope (Env) trimer structure, glycan occupancy, and the targets and ability of neutralizing antibodies (nAbs) to protect against SIVmac239 remain unknown. Here, we report the isolation of SIVmac239 nAbs that recognize a glycan hole and the V1/V4 loop. A high-resolution structure of a SIVmac239 Env trimer-nAb complex shows many similarities to HIV and SIVcpz Envs, but with distinct V4 features and an extended V1 loop. Moreover, SIVmac239 Env has a higher glycan shield density than HIV Env that may contribute to poor or delayed nAb responses in SIVmac239-infected macaques. Passive transfer of a nAb protects macaques from repeated intravenous SIVmac239 challenge at serum titers comparable to those described for protection of humans against HIV infection. Our results provide structural insights for vaccine design and shed light on antibody-mediated protection in the SIV model.

Fig. 1. Isolation of SIVmac239 neutralizing rhesus mAbs.

a FACS sorting layout from r10039 rhesus PBMCs, in which population is gated on lymphocytes/singlets. CD3⁻CD4⁻CD8⁻CD14⁻CD20⁺IgM⁻IgG⁺ SIVmac239 SOSIP.664²⁺ memory B cells were single-cell sorted. Red dots represent sorted cells. b Immunogenetics of isolated rhesus mAbs. Rhesus mAbs were annotated with a rhesus germline database. c, d SIVmac239 pseudovirus neutralization curves. Data are presented as mean values ± SD and error bars are from two technical replicates. Data are representative for at least three independent experiments. c and summary table d of neutralization IC₅₀ and maximum percentage of neutralization (MPN) for denoted mAbs. IC₅₀ potency was colored according to the key.

Fig. 2. SIVmac239 nAbs recognize two specificities: a glycan hole and the V4 loop.

a Negative-stain electron microscopy (nsEM) of K11 (red) bound to SIVmac239 SOSIP.664 (gray). b Representative neutralization curves of K11 (red), ITS90.03 (blue), and FZ019.2 (purple) against wild-type SIVmac239 virus (solid line) and K254N mutant with N-linked glycan filled in (dashed lines). Data are presented as mean values ± SD and error bars are from two technical replicates. Data are representative for at least two independent experiments. c nsEM reconstruction of FZ019.2 Fab (purple) bound to SIVmac239 SOSIP.664. d Model of the complex using BG505 SOSIP.664 (PDB:6X9R) with the SIVmac239 delta V1/V2 gp120 monomer crystal structure (PDB:6TYB) aligned to one protomer showing that FZ019.2 targets the V4 loop (pink). Residues within 5 Å of the Fab are colored yellow. e Neutralization breath and potency (μg/mL) of isolated mAbs against an SIV pseudovirus panel. CD4-IgG2 served as positive control. IC₅₀ potency was colored according to the key.

Fig. 3. Mapping of polyclonal antibody epitope specificities.

a ELISA competition between a panel of SIVmac239 Env-binding mAbs and SIVmac239-infected macaque sera for binding to SIVmac239 SOSIP.664. Maximum percentage of competition is colored according to the key. b ELISA screening of infected rhesus sera to SIVmac239 Env 15-mer peptides. Peptide ID numbers were listed on the left and sequences were shown in Supplementary Table 2. SIVmac239 neutralizing mAbs (K11, FZ019.2), non-neutralizing mAbs (FZ012.16, 5L7), and HIV bnAbs (PGDM1400, VRC01) served as controls. Absorbance intensities at OD405 nm were colored according to the key. c, d EMPEM mapping of SIVmac239-infected c non-neutralizing macaque sera Rh33519, Rh31186, Rh34620 and d neutralizing macaque sera r11008, r11002, r11004 for binding to SIVmac239 SOSIP.664. Data shown are representative 2D class averages of trimeric and monomeric particles bound to polyclonal Fabs along with segmentations of 3D reconstructions showing the location of bound Fabs relative to a molecular surface model of the SIVmac239 structure. Epitope specificities are colored according to the key.

Fig. 4. Cryo-EM structure of SIVmac239.K180S SOSIP trimer in complex with the nAb K11.

A Domain organization, site-specific N-linked glycan distribution, and cryo-EM structure of SIVmac239.K180S SOSIP.664 in complex with K11. B Close-up of the V1-V5 loops with the structures of SIVcpz MT145K (PDB:6OHY) and HIV-1 BG505 SOSIP (PDB:6X9R) Env trimers overlayed for comparison—glycans not shown. C Close-up views of the novel SIVmac239 V1 and V4 loop conformations showing the positions of N-linked glycans (Asn-NAG only) for all three structures to emphasize the clashes that would occur with HIV and SIVcpz glycans. D Cryo-EM map density for the novel V1 and V4 loop conformations with the sequence of the disordered V1 fragment indicated above. E Epitopes of both K11 and ITS90.03 (PDB:6TYB) mapped to the SIVmac239 trimer structure (white indicates residues within 5 Å of Fab residues). F Close-up view of the K11 epitope-paratope showing hydrogen bonds with gp120 residues and N-linked glycans. Also indicated is the intra-CDRH3 disulfide bond.

Fig. 5. Site-specific MS and computational modeling of the SIVmac239 Env trimer glycan shield.

a Predominant glycoforms at each site as determined by MS used for computational modeling along with cartoon representations of each type of complex glycan and a visualization of 10 (1000 total) different fully glycosylated models with glycans in green and gp120 and gp41 in white and black respectively. b Glycoform distribution for the two highly under-processed glycan sites at N37 and N278 showing the detection of glucosylated Man9 at both sites along with views of each glycan structure showing stabilizing hydrogen bonds with neighboring protein and glycan residues. c Glycan encounter factor calculated for every surface-exposed residue using 1000 fully glycosylated models of both SIVmac239 and BG505 SOSIP.664 and projected as a color map onto the molecular surface representation. Blue indicates the most shielded surfaces and red the most exposed. Demarcated with dashed circles are some prominent regions including common neutralizing epitopes such as the CD4 binding site (yellow), glycan hole (green), V2 apex (magenta), and the V1/V4 epitope identified in this study (black). Other regions of interest such as the gp120 inner and outer domains (OD), V1 and V4 loops, and high-mannose patch (HMP) are labeled. d Per-residue glycan encounter factor and per-glycan root mean squared fluctuation (RMSF) calculated from 1000 fully glycosylated trimer models.

Fig. 6. Repeated SIVmac239 intravenous (IV) challenge of 5L7-LS-infused, K11-infused, and control animals.

a SIVmac239 passive Ab protection experimental design. Rhesus macaques were allocated to three groups; Group 1 and Group 2 were infused with 60 mg/kg of 5L7-LS and K11 antibody, respectively, 1 day prior to challenge. Group 3 animals were untreated controls. Animals were challenged weekly by the IV route with a repeat 1 AID₅₀ dose of SIVmac239. Blood was drawn and plasma viral load was measured at day 7 and day 10 following each challenge. Animals with detectable viral loads at either of those time points were no longer challenged. After three repeated challenges, a second antibody infusion was given to all Group 1 and Group 2 animals. Uninfected animals were subsequently repeatedly challenged intravenously with 1 AID₅₀ dose of SIVmac239. b Kaplan–Meier curves comparing three test groups colored according to the key. Protection was measured by log-rank (Mantel-Cox) test. ***P < 0.001. ns not significant. c Plasma viral loads after productive infection with SIVmac239 in the 3 animal groups: K11 (red), 5L7-LS (blue), and control (black). The threshold of the assay employed was 15 SIV RNA copies/mL. The plasma viral concentrations for each animal were indicated by different symbols. d Comparison of peak plasma viral loads among the 3 groups of infected animals based on two-tailed Mann–Whitney U test. ns not significant (p > 0.05). e Comparison of set point viral load among three groups of infected animals based on two-tailed Mann–Whitney U test. Set point viral load was the geometric mean of all values from each animal between weeks 10 and 20. **P = 0.0043. f Plasma nAb ID₅₀ titers and geometric mean of ID₅₀ titer 7 days prior to infection in the K11 group using the SIVmac239 pseudovirus assay. N/A not applicable. g ADCC activity was measured for rhesus mAbs using SIVmac239 virus on CEM.CCR5 target cells and macaque NK effector cells. CD4-IgG2 served as positive control and an anti-dengue mAb Den3 as a negative control. Data are presented as mean values ± SD, and error bars are from two technical replicates.