Mapping the antibody response to Lassa virus vaccination of non-human primates

Abstract

Background: Lassa fever, caused by Lassa virus, is a severe disease, endemic in Western Africa, for which no vaccines or therapeutics are yet approved. Understanding the immune responses elicited by candidate vaccines is key for approval, including characterisation of antibody epitopes recognised and capacity for neutralisation.

Methods: Here we used negative-stain electron microscopy polyclonal antibody epitope mapping (EMPEM), in-vitro pseudovirus neutralisation assays, and biophysical antibody competition assays to uncover components of polyclonal antibody responses elicited in nonhuman primates 26 days after receipt of a single immunisation with a fully protective, recombinant, replication-competent vesicular stomatitis virus-based vaccine bearing the Lassa virus glycoprotein GPC.

Findings: Although the vaccinee sera are overall poorly-neutralising, we do directly visualise, within the polyclonal pool, antibodies targeting epitopes on GPC that are consistent with neutralisation, as well as competition with known neutralising mAbs. Nearly every animal, for example, produced antibodies that compete with mAbs against GP1-A and GPC-A neutralising epitopes. The most abundant classes of antibodies, however, are directed against interior interfaces of GPC, while other antibodies recognise post-fusion GPC epitopes not consistent with neutralisation.

Interpretation: It may be that some individual antibodies in the pool are neutralising, but that the abundance of non-neutralising epitopes reduces potency as measured at the polyclonal level. The finding, however, neutralisation-consistent sites and competition with known neutralising antibodies are important steps in vaccine design toward eliciting more potent neutralisation.

Funding: A complete list of funding bodies that supported this study is presented in the Funding section.

Keywords: Electron microscopy polyclonal antibody epitope mapping; Lassa virus; Neutralising antibody; Vaccine design.

Fig. 1

Antibodies in non-human primates (NHP) appearing around four weeks after a single dose of rVSV$\mathbf{\Delta }$G-LASV-GPC (Josiah) recognise protomeric, and pre-fusion and post-fusion conformations of LASV GPC. (a) NHP Immunisation schedule. Two groups of NHPs received either a high- or low-dose single immunisation of rVSV $\Delta$ G-LASV-GPC on Day 0; the control group received only saline. NHP serum samples were collected before vaccination and 26 days later. All animals were homologously challenged with LASV Josiah on Day 28. (b) Sandwich ELISA binding titres (midpoint EC50) for serum samples collected on Day 0 and Day 26 shown as a line indicating the mean of all animals in the indicated group with the 95% confidence interval (CI) (n = 5 for vaccinated groups; n = 3 for control groups) and time point. Each data point is the average of two technical replicates (n = 2) for each serum sample at every time point. ∗, p < 0.05 by Kruskal–Wallis test with Dunn's correction.

Fig. 2

Neutralisation by vaccine-elicited antibodies in some animals is partially impaired by glycans at position N390 and N395 that occlude the GPC-B epitope. (a) Neutralisation of wild-type rVSV-LASV pseudovirus (ppLASV) by purified pAbs from NHPs in Group-1 (high dose, left) and Group-2 (low-dose, right). (b) Neutralisation of ppLASV bearing N390D and N395D mutations by purified pAbs from NHPs in Group-1 (left) and Group-2 (right). Each data point is the average of two biological duplicates, where each replicate is performed in technical duplicate; error bars indicate $\pm$ the 95% CI from the mean. Data are relative neutralisation compared to infection with wild-type or glycan-deleted pp-LASV in the absence of antibody.

Fig. 3

NHP vaccinee pAbs target multiple epitopes on pre- and post-fusion GPC. (a) Schematic of biolayer interferometry antibody competition assay. Biotinylated pre-fusion GPCys^R4-TD or post-fusion GPC-linked is first captured on streptavidin-loaded biosensors to establish a baseline, and the biosensor is then dipped into either buffer or a mAb solution such that GPC is fully saturated with bound mAb. The resulting GPC-mAb complexes are then dipped into pAb solution from the indicated animal to measure pAb binding interference when mAb is pre-bound to GPC as compared to pAb binding to GPC in the absence of mAb. The total binding of pAbs to GPC in the absence of mAb was set as 100% relative binding (dotted line). (b) Binding analysis in an antibody competition assay measuring binding of pAb when GPC is pre-bound with the indicated mAb from a representative panel of mAbs against eight GPC epitopes, including three known to be recognised by neutralising mAbs (GP1-A, GPC-A, and GPC-B) and five known to be recognised by non-neutralising mAbs (GP1-B, GP2-L1, GP2-L2, GP2-L3, and GP2-B). Error bars indicate the mean ± the SD of two independent experiments, each performed in technical duplicate. The data are normalised to the binding of pAb to GPC in the absence of mAb.

Fig. 4

pAbs from NHP vaccinees recognise external epitopes on trimeric pre-fusion GPC that are associated with neutralization. Negative stain electron microscopy 3-D reconstructions of pFabs from vaccinated NHPs (animals 7, 8, and 12) bound to GPC-TD. (a) Panel: LASV GPC atomic model (PDB: 5VK2; GP1, green; GP2, purple) docked into a 3-D negative stain reconstruction of unliganded GPC. Density corresponding to the trimerisation domain (PDB: 1NOG) is labelled with “TD”. The “side view” is orthogonal to the LASV GPC trimeric axis while the “top-view” is head-on with the trimeric axis. (b–d) From left to right: composite 3-D reconstructions of polyclonal Fabs from vaccinated animals NHP-7 and NHP-8 of the high-titre group, and NHP-12 of the low-titre group in complex with pre-fusion GPC-TD. Fab density is coloured according to the putative epitope with blue, magenta and pink, and yellow corresponding to the GP1-A, GPC-A, GPC-B, and GP2-L3 epitopes, respectively.

Fig. 5

Vaccine-elicited antibodies recognise internal epitopes on pre-fusion GPC that promote GPC disassembly. (a) pAbs from NHP vaccinee four weeks after prime dose recognise internal epitopes on pre-fusion GPC. Left panel: atomic model of pre-fusion LASV GPC trimer depicting one protomer as ribbons with the GP1 and GP2 subunits coloured light and dark turquoise, respectively. Right panel: representative 2-D class averages of unliganded GPC-TD, side and top view, and a class of unbound polyclonal Fab (orange). (b) Left panel, 3-D negative stain reconstruction of polyclonal Fab bound to an internal epitope on pre-fusion GPC that is normally hidden in the trimeric conformation of GPC. Right panel, a zoomed-in view of an orthogonal slice (side view clip) along the axis of the pFab shows that the inferred pFab footprint appears to span internally across two protomers with potential contributions from GP1 to GP2 subunits of both protomers. A representative 2-D class average of a pFab bound to an internal epitope on GPC is shown on the lower left. (c) 3-D reconstructions and 2-D class averages depicting pFabs bound to different components of disassembled GPC. The 2-D and 3-D reconstructions indicate that pFabs may mechanically destabilise GPC by wedging in between protomers in the trimer that promotes disassembly of trimeric GPC into individual protomers and likely even individual subunits. (d) Cartoon schematic depicting a potential mechanism of action for pAb-mediated disassembly of GPC. pFab binding to external regions on GPC first destabilises the pre-fusion conformation that in turn exposes internal sites on the trimer allowing binding of a second wave of pFabs that mechanically deconstruct GPC from a trimer into protomeric components.