A bivalent Epstein-Barr virus vaccine induces neutralizing antibodies that block infection and confer immunity in humanized mice

Abstract

Epstein-Barr virus (EBV) is the major cause of infectious mononucleosis and is associated with several human cancers and, more recently, multiple sclerosis. Despite its prevalence and health impact, there are currently no vaccines or treatments. Four viral glycoproteins (gp), gp350 and gH/gL/gp42, mediate entry into the major sites of viral replication, B cells, and epithelial cells. Here, we designed a nanoparticle vaccine displaying these proteins and showed that it elicits potent neutralizing antibodies that protect against infection in vivo. We designed single-chain gH/gL and gH/gL/gp42 proteins that were each fused to bacterial ferritin to form a self-assembling nanoparticle. Structural analysis revealed that single-chain gH/gL and gH/gL/gp42 adopted a similar conformation to the wild-type proteins, and the protein spikes were observed by electron microscopy. Single-chain gH/gL or gH/gL/gp42 nanoparticle vaccines were constructed to ensure product homogeneity needed for clinical development. These vaccines elicited neutralizing antibodies in mice, ferrets, and nonhuman primates that inhibited EBV entry into both B cells and epithelial cells. When mixed with a previously reported gp350 nanoparticle vaccine, gp350D₁₂₃, no immune competition was observed. To confirm its efficacy in vivo, humanized mice were challenged with EBV after passive transfer of IgG from mice vaccinated with control, gH/gL/gp42+gp350D₁₂₃, or gH/gL+gp350D₁₂₃ nanoparticles. Although all control animals were infected, only one mouse in each vaccine group that received immune IgG had detectable transient viremia. Furthermore, no EBV lymphomas were detected in immune animals. This bivalent EBV nanoparticle vaccine represents a promising candidate to prevent EBV infection and EBV-related malignancies in humans.

Figure 1.. Structure-based design of single chain gH/gL and single chain gH/gL/gp42 nanoparticles.

(A) A schematic representation is shown for the single chain gH/gL, single chain gH/gL/gp42, single chain gH/gL-NP, and single chain gH/gL/gp42-NP constructs. EBV gL (green) is fused to the N terminus of gH (cyan) through a flexible amino acid linker (indicated by the black line between gL and gH). EBV gp42 (gray) is fused to the C-terminus of gH. Single chain gH/gL-NP or single chain gH/gL/gp42-NP constructs show the gH/gL or gH/gL/gp42 fused to H. pylori ferritin (represented by the letter “F” in beige) by a flexible amino acid linker (line), respectively. (B) Left: The crystal structure of the single chain gH/gL was resolved at 5.5Å (gL in green and gH in cyan) with superposition of the previously solved crystal structure of gH/gL complex (white, PDB: 3PHF) (Root-Mean-Square (RMS_ = 0.33). Right: The crystal structure of the single chain gH/gL/gp42 was resolved at 2.9Å and superposition with the previously solved gH/gL/gp42 heterotrimer complex crystal structure (white, PDB: 5T1D) (RMS value = 0.96). (C) Left: Size exclusion chromatography (SEC) elution profiles of single chain gH/gL-NP and single chain gH/gL/gp42-NP are shown. mAU indicates milli-absorbance unit at 280 nm. Right: Size of single chain gH/gL-NP and single chain gH/gL/gp42-NP were determined by dynamic light scattering. (D) Negative stain electron microscopy (EM) images of single chain gH/gL-NP (left) and single chain gH/gL/gp42-NP (right) are shown. Close-up images of the nanoparticles are displayed in the upper right corners. A structural model of the single chain gH/gL-NP or single chain gH/gL/gp42-NP is shown on the right of their respective EM images (gH: cyan; gL: green; gp42: gray; ferritin: orange). The surface density is a model built from crystal structures solved in (B) and ferritin core from PDB 3BVE (DOI:10.2210/pdb3bve/pdb) using Chimera (47) and is not reconstructed from EM.

Figure 2.. Neutralization responses are induced in mice by vaccination with single chain gH/gL-NP or single chain gH/gL/gp42-NP alone or in combination with gp350D₁₂₃-NP.

BALB/c mice (n=5 per group) were immunized intramuscularly in the presence of AF03 adjuvant at weeks 0 and 3 with 1μg of (A) monovalent single chain gH/gL-NP, gp350D₁₂₃ or bivalent gH/gL-NP+gp350D₁₂₃-NP or (B) monovalent single chain gH/gL/gp42-NP or bivalent gH/gL/gp42-NP+gp350D₁₂₃-NP. Control indicates pre-immune serum samples. Neutralization antibody titers from immune serum collected 2 weeks after the second injection were determined in Raji B cells and SVK CR2 epithelial cells. The IC₅₀ indicates the log titer that resulted in 50% inhibition of EBV entry into target cells. The data are shown as box-and-whiskers plots, where the box indicates lower and upper quartiles with line at median, and whiskers span minimum and maximum data points; *p<0.05 compared to control as measured by Mann-Whitney test.

Figure 3.. Immunogenicity of single chain gH/gL-NP+gp350D₁₂₃-NP or single chain gH/gL/gp42-NP alone or in combination with gp350D₁₂₃-NP was measured in ferrets and NHPs.

(A) Ferrets (n=6 per group) were immunized intramuscularly at weeks 0 and 4 with either 15 μg single chain gH/gL-NP + 15 μg gp350D₁₂₃-NP or 15 μg single chain gH/gL/gp42-NP + 15 μg gp350D₁₂₃-NP bivalent vaccines. Serum samples were collected 2 weeks after immunization and were assayed for neutralizing activity in both Raji B cells (left) and 293 epithelial cells (right). Means and standard errors of mean are shown. *p< 0.05 compared to pre-immune serum as measured by Mann-Whitney test. No neutralizing activity was detected from the pre-immune serum samples and concentrations were at the limit of detection in the graphs (dashed lines). (B) Rhesus macaques (n=4 per group) were vaccinated with the bivalent vaccine composed of 25μg gH/gL-NP + 25μg gp350D₁₂₃-NP or 25μg gH/gL/gp42-NP + 25μg gp350D₁₂₃-NP at weeks 0, 4, and 10. AF03 was used as adjuvant. Immune serum samples were collected 2 weeks after the third injection and neutralizing antibody titers were determined in both Raji B cells (left) and SVK CR2 epithelial cells (right). Means and standard errors of meanare shown. *p< 0.05 compared to pre-immune serum as measured by Mann-Whitney test.

Figure 4.. Passive transfer of bivalent vaccine serum confers protection against EBV infection in humanized NSG mice.

Humanized NSG mice (n=6 per group) were injected with IgG (20μg/g of mouse body weight) purified from naïve (control), single chain gH/gL-NP+ gp350D₁₂₃-NP, or single chain gH/gL/gp42+gp350D₁₂₃-NP immunized BALB/c mice. Passive transfer of IgG was delivered intraperitoneally on day −1, 0, and 1 and EBV challenge was performed intravenously on day 0. (A) Viremia from each group was measured at weeks 5, 7, and 9 post challenge. Medians with 25% and 75% percentiles are shown by the boxes. Whiskers indicate minimum and maximum values. *p< 0.05 compared to control at the same week as measured bystudent’s t-test. (B) The heatmap shows EBV encoded RNA 1 (EBER1) positivity of tissues (graded 0 to 3) from mice receiving IgG from naïve, single chain gH/gL-NP+ gp350D₁₂₃-NP, or single chain gH/gL/gp42+gp350D₁₂₃-NP immunized BALB/c mice after challenge with EBV. A score of 0 indicates no EBER1 staining whereas a score of 3 indicates marked infiltration of tissues by EBER1-positive cells.

Figure 5.. Passive transfer of bivalent vaccine serum confers protection against EBV lymphoma in vivo.

Pathologic and immunohistochemical analyses of the liver are shown for representative mice receiving IgG from non-immune (Control, left) or vaccinated mice (Immune, middle and right) after challenge with EBV. Tissues were collected 27 weeks after challenge and stained with hematoxylin and eosin (H&E, yellow arrows indicate representative region with lymphoma, top) or anti-CD20 antibody (brown staining, red arrows, middle). In situ hybridization was also performed with a probe to EBER1 (purple staining, cyan arrows, bottom). CD20 and EBER staining were apparent in control samples, but not in any of the tissues receiving IgG from vaccinated mice. No EBV-positive B cell lymphomas were observed in recipients of IgG from vaccinated mice. Scale bars, 100 μm.