Immunogenicity of propagation-restricted vesicular stomatitis virus encoding Ebola virus glycoprotein in guinea pigs

Abstract

Vesicular stomatitis virus (VSV) expressing the Ebola virus (EBOV) glycoprotein (GP) in place of the VSV glycoprotein G (VSV/EBOV-GP) is a promising EBOV vaccine candidate which has already entered clinical phase 3 studies. Although this chimeric virus was tolerated overall by volunteers, it still caused viremia and adverse effects such as fever and arthritis, suggesting that it might not be sufficiently attenuated. In this study, the VSV/EBOV-GP vector was further modified in order to achieve attenuation while maintaining immunogenicity. All recombinant VSV constructs were propagated on VSV G protein expressing helper cells and used to immunize guinea pigs via the intramuscular route. The humoral immune response was analysed by EBOV-GP-specific fluorescence-linked immunosorbent assay, plaque reduction neutralization test and in vitro virus-spreading inhibition test that employed recombinant VSV/EBOV-GP expressing either green fluorescent protein or secreted Nano luciferase. Most modified vector constructs induced lower levels of protective antibodies than the parental VSV/EBOV-GP or a recombinant modified vaccinia virus Ankara vector encoding full-length EBOV-GP. However, the VSV/EBOV-GP(F88A) mutant was at least as immunogenic as the parental vaccine virus although it was highly propagation-restricted. This finding suggests that VSV-vectored vaccines need not be propagation-competent to induce a robust humoral immune response. However, VSV/EBOV-GP(F88A) rapidly reverted to a fully propagation-competent virus indicating that a single-point mutation is not sufficient to maintain the propagation-restricted phenotype.

Keywords: Ebola virus; biosafety; neutralizing antibody; reversion; vaccinia virus; vector vaccine; vesicular stomatitis virus; viral glycoprotein.

Fig. 1.

Genome maps of recombinant VSV vectors. (a) The original VSV contains five transcription units encoding the nucleoprotein N, the phosphoprotein P, the matrix protein M, the glycoprotein G and the large RNA-dependent RNA polymerase L. The VSV vector was modified by replacing the G gene with the EBOV-GP gene (either authentic or modified) and by inserting an additional transcription unit encoding either GFP, sNLuc or EBOV-VP40. Mq denotes a modified M gene encoding a mutant M protein which is characterized by the amino acid changes M33A, M51R, V221F and S226R. (b) Protein maps of authentic and modified EBOV-GP, indicating the location of the signal peptide (SP), receptor-binding domain (RBD), glycan cap, mucin-like domain (mucin), fusion loop (FL), heptad repeat (HR) region and transmembrane (TM) domain. The furin cleavage site (red line), the cleavage products GP1 and GP2, and the amino acid positions F88 and P537 are indicated as well.

Fig. 2.

Propagation competence of recombinant VSV vectors. (a) Multi-cycle replication of recombinant VSV. Vero cells grown in six-well plates were infected with the indicated recombinant viruses using an m.o.i. of 0.0001 ffu cell⁻¹. At the indicated times, aliquots of the cell-culture supernatant were collected and infectious virus titrated on Vero cells. Mean values and sd of three independent experiments are shown. Asterisks indicate significantly different infectious virus titres when compared to VSV*ΔG(EBOV-GP). (b) Virus yield on helper cells. BHK-G43 cells in 24-well plates were treated with mifepristone to induce VSV-G protein expression (dark grey bars) or were left untreated (light grey bars). Cells were infected with the indicated viruses using an m.o.i. of 0.1 ffu cell⁻¹ and maintained for 24 h in medium with or without mifepristone. Medium without mifepristone was supplemented with a neutralizing antibody directed to the VSV-G protein in order to inactivate any remaining input virus. Infectious virus released into the cell-culture medium was titrated on Vero cells. Results are shown as the mean plus sd of three independent experiments. Asterisks indicate significantly different infectious virus titres when compared to VSV*ΔG(EBOV-GP). (c) Propagation of GP mutant VSV on mammalian cell lines. The indicated cell lines were infected with VSV*ΔG(EBOV-GP_F88A) (blue bars), VSV*ΔG(EBOV-GP_P537R) (orange bars) and parental VSV*ΔG(EBOV-GP) (gey bars) using an m.o.i. of 0.1 ffu cell⁻¹ and incubated for 24 h in the presence of neutralizing antibody directed to the VSV-G protein. Infectious virus titres released into the cell-culture supernatant were determined. Mean titres and sd of three infection experiments are shown. Black asterisks indicate significantly different infectious virus titres when compared to VSV*ΔG(EBOV-GP). Red asterisks indicate significantly different virus titres when comparing VSV*ΔG(EBOV-GP_F88A) with VSV*ΔG(EBOV-GP_P537R).

Fig. 3.

Reversion of the growth-restricted phenotype of VSV*ΔG(EBOV-GP_F88A) and VSV*ΔG(EBOV-GP_P537R). (a) The indicated viruses were serially passaged on BHK-21 cells (replicates R1 to R6) and infectious virus titres (24 h p.i.) determined for passage 2 (white bars), passage 3 (light grey bars), passage 4 (dark grey bars) and passage 5 (black bars). Passage 5 viruses that were used to determine the cDNA sequence of GP are indicated by red arrows. (b) The complete primary sequence of GP from passaged viruses was determined but only regions containing amino acid changes are depicted. The same mutations were found in the second replicate of VSV*ΔG(EBOV-GP_F88A) and VSV*ΔG(EBOV-GP_P537R), respectively.

Fig. 4.

Induction of type I IFN synthesis in VSV vector-infected cells. (a) Release of type I IFN by infected cells. NHDF were infected with the indicated viruses (m.o.i. of 10). At 24 h p.i., cell-culture supernatants were sampled and heated for 30 min at 55 °C to inactivate infectious virus. The antiviral activity released into the cell-culture medium was titrated on HeLa cells using a previously described VSV replicon-based bioassay [24]. The antiviral activity was expressed as inhibitory concentration 50 % (IC₅₀). Mean values and sd of three infection experiments are shown. The broken line indicates the lower limit of detection. Asterisks indicate secretion of type I IFN at levels that are significantly different (P<0.05) from the mock control. (b) Spreading of chimeric VSV in cell culture. NHDF and Vero cells were infected with either VSV*∆G(EBOV-GP) or VSV*Mq∆G(EBOV-GP) using an m.o.i. of 0.0001 ffu cell⁻¹. Spreading of virus in the cell monolayer was monitored by detection of GFP fluorescence with an inverted fluorescence microscope. The bar represents 200 µm. (c) Multi-cycle replication of sNLuc reporter viruses in IFN-competent cells. NHDF were infected with either VSV∆G(EBOV-GP,sNLuc) (red line) or VSVMq∆G(EBOV-GP,sNLuc) (black lines) using an m.o.i. of either 0.001 ffu cell⁻¹ (circles), 0.01 ffu cell⁻¹ (rhombs) or 0.1 ffu cell⁻¹ (squares). sNLuc activity was determined in the cell-culture medium at the indicated times. Mean luciferase values and sd of three infection experiments are shown. Asterisks indicate significant different luciferase values compared to VSV∆G(EBOV-GP,sNLuc).

Fig. 5.

Recombinant VSV-driven expression of EBOV-GP. Vero cells in 12-well plates were infected with the indicated viruses using an m.o.i. of 10 ffu cell⁻¹. (a) At 14 h p.i., cell surface proteins were labelled with biotin, precipitated from cell lysates by immobilized streptavidin, and analysed by Western blot using either guinea pig polyclonal anti-EBOV-GP serum or rabbit anti-VP40 serum. The positions of proteins with defined molecular weight are indicated on the left-hand side. (b) Flow cytometric analysis of EBOV-GP expression. Infected Vero cells were incubated with a LIVE/DEAD fixable violet dead cell marker. Subsequently, cells were stained either directly for EBOV-GP cell surface expression (left panel) or fixed and permeabilized to allow detection of intracellular EBOV-GP (right panel). EBOV-GP was detected using a mouse polyclonal anti-EBOV-GP anti-serum and anti-mouse IgG-allophycocyanin. Expression levels of EBOV-GP are represented in histogram plots of live, GFP-positive, i.e. infected cells.

Fig. 6.

Analysis of the antibody responses of guinea pigs vaccinated with recombinant VSV. (a) FLISA. Vero cells were grown in 96-well plates and infected with MVA-BN-EBOV-GP (m.o.i. of 0.05 ffu cell⁻¹). At 24 h p.i., the cells were fixed with paraformaldehyde and incubated with serially diluted serum pools from vaccinated guinea pigs (four to five animals per vaccine group) and subsequently with Alexa 488-conjugated anti-guinea pig IgG serum. FLISA antibody titres were calculated by determining the reciprocal value of the highest immune serum dilution allowing discrimination of infected from non-infected Vero cells by indirect immunofluorescence. Mean values and sd of three independently performed titrations are shown. (b) Analysis of guinea pig sera by PRNT. Serially diluted serum from vaccinated guinea pigs (n=4 to 5 animals per vaccine group) were incubated for 60 min with 100 ffu of VSV*∆G(EBOV-GP). Vero cell monolayers grown in 96-well cell-culture plates were inoculated with the virus/antibody mixture for 1 h and then replaced by 200 µl of medium containing 0.8 % methyl cellulose. Following an incubation period of 24 h, the GFP-positive cell foci were counted under an inverted fluorescence microscope. The reciprocal serum dilution causing a reduction of plaque numbers by 80 % (PRNT₈₀) was calculated. Mean titres and sd were calculated for the immune sera collected from four to five individual guinea pigs per group. (c) Inhibition of virus spreading in vitro. Vero cells were infected with VSV*∆G(EBOV-GP,sNLuc) using an m.o.i. of 0.005 ffu cell⁻¹ and maintained in medium containing serial dilutions of immune sera which were collected 4 weeks after the first and 4 weeks after the second immunization of guinea pigs with recombinant VSV expressing the indicated antigens. At 24 h p.i., sNLuc activity in the cell-culture supernatant was determined. The reciprocal serum dilution leading to 90 % inhibition of reporter activity (IC₉₀) was determined (relative to virus-spreading experiments in the presence of naïve guinea pig serum). Mean IC₉₀ titres and sd were calculated for immune sera that were collected from four to five individual guinea pigs per vaccine group. (a–c) Black asterisks indicate significantly different titres (P<0.05) with respect to the reference vaccine VSV*∆G(EBOV-GP). Red asterisks indicate significantly different antibody titres when comparing first and second immunization.