Altered Glycosylation Patterns Increase Immunogenicity of a Subunit Hepatitis C Virus Vaccine, Inducing Neutralizing Antibodies Which Confer Protection in Mice

ABSTRACT

Hepatitis C virus (HCV) infection is a global health problem for which no vaccine is available. HCV has a highly heterogeneous RNA genome and can be classified into seven genotypes. Due to the high genetic and resultant antigenic variation among the genotypes, inducing antibodies capable of neutralizing most of the HCV genotypes by experimental vaccination has been challenging. Previous efforts focused on priming humoral immune responses with recombinant HCV envelope E2 protein produced in mammalian cells. Here, we report that a soluble form of HCV E2 (sE2) produced in insect cells possesses different glycosylation patterns and is more immunogenic, as evidenced by the induction of higher titers of broadly neutralizing antibodies (bNAbs) against cell culture-derived HCV (HCVcc) harboring structural proteins from a diverse array of HCV genotypes. We affirm that continuous and discontinuous epitopes of well-characterized bNAbs are conserved, suggesting that sE2 produced in insect cells is properly folded. In a genetically humanized mouse model, active immunization with sE2 efficiently protected against challenge with a heterologous HCV genotype. These data not only demonstrate that sE2 is a promising HCV vaccine candidate, but also highlight the importance of glycosylation patterns in developing subunit viral vaccines.

IMPORTANCE A prophylactic vaccine with high efficacy and low cost is urgently needed for global control of HCV infection. Induction of broadly neutralizing antibodies against most HCV genotypes has been challenging due to the antigenic diversity of the HCV genome. Here, we refined a high-yield subunit HCV vaccine that elicited broadly neutralizing antibody responses in preclinical trials. We found that soluble HCV E2 protein (sE2) produced in insect cells is distinctly glycosylated and is more immunogenic than sE2 produced in mammalian cells, suggesting that glycosylation patterns should be taken into consideration in efforts to generate antibody-based recombinant vaccines against HCV. We further showed that sE2 vaccination confers protection against HCV infection in a genetically humanized mouse model. Thus, our work identified a promising broadly protective HCV vaccine candidate that should be considered for further preclinical and clinical development.

INTRODUCTION

It is estimated that over 2% of the world's population is chronically infected with hepatitis C virus (HCV) (1). Although recently approved direct-acting antiviral (DAA) drugs (2) have greatly improved upon the curing efficacy of the previous interferon (IFN)-based regimen, these new therapies are very expensive and thus unaffordable for the majority of HCV-infected individuals who live in developing countries, where most new infections occur. Since the approval of these highly effective DAAs, the number of chronic HCV carriers has not significantly declined. Furthermore, there is little evidence that patients cured of their chronic infections with DAAs retain antiviral immunity that is protective against future HCV exposures. Therefore, the development of a prophylactic HCV vaccine with high efficacy and low cost remains a high priority in the global control of HCV infection. Natural clearance of HCV correlates with the induction of vigorous T cell responses with broad specificity, which has prompted efforts to pursue T cell-based vaccines. Currently, the only vaccine candidate in clinical trials is based on the expression of HCV nonstructural proteins with adenoviruses and Modified vaccinia virus Ankara (MVA) to elicit T cell responses to infection (3, 4).

However, T cell-based vaccines cannot prevent the first steps of a viral infection, thus creating a rational for alternative/additional approaches geared toward induction of neutralizing antibodies (NAbs) that could putatively block HCV uptake. NAbs have been found to correlate with the protection offered by all the viral vaccines licensed thus far (5). However, the role of anti-HCV antibodies in humans is under debate. A strong correlation between viral clearance and the induction of an early and broad NAb response following HCV infection has been reported in a number of patient cohorts with HCV infection (6–8). In addition, in vivo challenge/protection studies have shown that the passive transfer of monoclonal and polyclonal NAbs was able to prevent HCV infection in chimpanzees (9) and in mice (10–14), highlighting the important role of NAbs in protecting against HCV infection. A single HCV vaccine that could induce NAbs against all seven known HCV genotypes would be ideal. However, due to the extreme genetic and antigenic diversity across and within HCV genotypes, this has been a particularly challenging goal (15).

The HCV envelope proteins E1 and E2 are responsible for mediating HCV entry into target cells by direct or indirect interaction with numerous host molecules (16, 17) and are thus the natural targets of NAbs (18). Consequently, all experimental HCV vaccines that aim to generate NAbs contain E2 and/or E1 components in a variety of modalities or prime-boost regimens (19–28). Although significant progress has been made toward the development of an efficacious HCV vaccine mediating protection by inducing humoral immune responses, several important issues remain: (i) the spectrum of NAbs elicited by existing vaccine candidates is still insufficiently broad to cover all seven HCV genotypes; (ii) the complexity of the heterologous prime-boost regimens with different antigen modalities renders vaccine production and vaccination difficult; (iii) the low yield of antigen manufacture hampers the application of some promising vaccine candidates, such as inactivated cell culture-derived HCV (HCVcc) (23); (iv) few of these vaccines have been evaluated in an immunocompetent-animal model by active immunization, as the utilization of chimpanzees is limited for ethical and financial reasons and murine models, such as human liver chimeric mice (29), are immunodeficient. To address some of these challenges, we aimed to develop an improved method for inducing broadly neutralizing antibodies (bNAbs). We demonstrate that expression of a transmembrane domain-truncated, soluble version of E2 (designated sE2) in insect cells results in an antigen with increased immunogenicity compared to equivalent constructs produced in mammalian cells. Mice mount strong immune responses to sE2, yielding high titers of anti-HCV E2 antibodies capable of neutralizing a diverse panel of HCV intergenotypic chimeras in vitro. Following sE2 vaccination, genetically humanized mice were protected against experimental challenge with a heterologous HCV strain.

MATERIALS AND METHODS

Establishment of HCVcc panels covering genotypes 1 to 7.

A panel of HCVccs covering genotypes 1 to 7 was used for the neutralization assay. All these HCVccs were produced following a previously described protocol (30), including in vitro transcription, HCV RNA electroporation, immunostaining, HCVcc titration, and amplification. Plasmids pUC-Con1/JFH1, pUC-H77/JFH1, and pUC-J6/JFH1 were constructed as previously described (31). Plasmids pJ8/JFH1, pS52/JFH1 (I793S, K1404Q), pED43/JFH1 (T827A, T977S), pSA13/JFH1 (A1022G, K1119R), pHK6a/JFH1 (F350S, N417T), and pQC69/JFH1 were generously provided by Jens Bukh at the University of Copenhagen (32). The chimeric genotype 1b HCVcc (PR52B6Mt and PR79L9) constructed from clinical isolates was reported previously (33).

Construction of expression plasmids.

We amplified optimized sE2 (amino acids [aa] 384 to 661) from the codon-optimized HCV E2 gene (strain Con1, genotype 1b; optimized and synthesized by GeneArt) with forward primer NcoI-sE2opti-F (5′-CTGCCATGGCCGGCACATACGTGACAG-3′) and reverse primer sE2opti-XbaI-R (5′-GCACTCTAGACTCGCTTCTGTCCCGAT-3′) and then inserted the PCR product into the vector pMT/BiP/V5-HisA (Invitrogen) to obtain pMT-sE2. To produce an sE2 without hypervariable region 1 (HVR1), we used the same approach (forward primer, NcoI-ΔHVR1-F, 5′-GCCCCATGGCAGCTGGTGAACACCAACGGCAGC-3′; reverse primer, sE2opti-XbaI-R) to obtain pMT-sE2ΔHVR1. To generate the mammalian sE2, the whole open reading frame (including the gene encoding the signal peptide and tags) was cloned from pMT-sE2 into the pcDNA3.1 vector (forward primer, 5′-CCCGCTAGCGCCACCATGAAGTTATGCATATTACTGGCCGTCG-3′; reverse primer, 5′-CCCGAATTCTCAATGGTGATGGTGATGATGACCGGTAC-3′) to obtain pcDNA3.1-sE2. To express the extracellular loop of human CD81 (hCD81LEL) (aa 112 to 202), we amplified the hCD81LEL gene from the construct pLEGFP-CD81 (34) by PCR (forward primer, hCD81LEL-F-NcoI, 5′-CAGCCATGGGCTTTGTCAACAAGGACCAG-3′; reverse primer, hCD81LEL-R-XhoI, 5′-GAACTCGAGCAGCTTCCCGGAGAAGAGGTC-3′) and inserted the fragment into pET26b to yield pET26b-hCD81LEL. All the plasmids constructed were verified by DNA sequencing.

Expression and purification of sE2 from stable Drosophila S2 cell clones.

To generate stable transfectants expressing sE2, we cotransfected S2 cells with pMT-sE2 and a selection vector, pCoBlast, followed by blasticidin screening. S2 cells cultured in complete Schneider's Drosophila medium (SDM) (Gibco) supplemented with 10% fetal bovine serum (FBS) (Gibco), 100 U/ml of penicillin-streptomycin (Gibco), and 100 mg/liter of l-glutamine (Gibco) were seeded at 3 × 10⁶ cells/well in a 6-well plate. After 12 h, we cotransfected 19 μg pMT-sE2 and 1 μg of the selection vector pCoBlast by calcium phosphate transfection and changed the medium after 16 to 24 h of incubation. Seventy-two hours posttransfection, we changed the medium to complete SDM containing 25 μg/ml blasticidin. After 1 to 2 weeks of selection, the medium was changed again for complete SDM containing 10 μg/ml blasticidin. To generate stable, high-yield cell clones, we seeded 1 or 2 of the stable transfected cells/well into 96-well plates. After 2 weeks of selection with SDM containing 10 μg/ml blasticidin, we detected sE2 expression from the supernatants of each well by enzyme-linked immunosorbent assay (ELISA) and Western blotting. Several monoclonal cells with high yields of sE2 were obtained. We chose one clone, named sE2B3, for large-scale culture and expression. For the large-scale expression, sE2B3 cells were cultured in complete Express Five SFM medium (SFM) (Gibco) supplemented with 100 U/ml of penicillin-streptomycin, 100 mg/liter of l-glutamine, and 10 μg/ml blasticidin in a 3-liter spinner flask (Bellco). When the cell concentration reached 2 × 10⁷ cells/ml, the cells were induced with chromic chloride at a final concentration of 5 μM. On day 5 postinduction, the supernatant was harvested and concentrated to a volume of 50 to 100 ml in binding buffer (0.5 M NaCl, 20 mM Tris, 10 mM imidazole, pH 7.9) by ultrafiltration in a Stirred Cell (Amicon 8400; Millipore) through a 5-kDa membrane. Then, sE2 proteins were purified from the concentrated supernatant using Ni-nitrilotriacetic acid (NTA) resins (Novagen) according to the manufacturer's instructions. The eluted sE2 proteins were analyzed by SDS-PAGE (12% acrylamide) and Western blotting. To obtain highly pure sE2 as a stimulus for enzyme-linked immunosorbent spot (ELISPOT) assays, the Ni-NTA-purified proteins were subjected to size exclusion chromatography (SEC) using a Superdex 200 10/300 GL column (GE Healthcare). The form with HVR1 deleted, sE2ΔHVR1, was expressed and purified using the same protocol.

Expression and purification of sE2 from HEK293T cells.

To generate sE2 from mammalian cells, HEK293T cells cultured in complete Dulbecco's modified Eagle medium (DMEM) (HyClone) supplemented with 10% FBS, 10 mM HEPES buffer (Gibco), 100 mg/liter of l-glutamine, 1% nonessential amino acids (NEAA) (Gibco), and 100 U/ml penicillin-streptomycin were transfected with pcDNA3.1-sE2 using branched polyethylenimine (PEI) (Sigma-Aldrich). The medium containing PEI was changed for complete DMEM with 10 μM sodium butyrate at 4 h posttransfection and then changed for FreeStyle 293 medium (FS293; Gibco) 12 h later. Mammalian sE2 was harvested, concentrated, and purified from the medium harvested 72 h posttransfection following the same protocol described above.

Deglycosylation of sE2.

Under reducing conditions, 10 μg of sE2 protein was denatured at 100°C for 10 min in 1× glycoprotein denaturing buffer and then digested with peptide-N-glycosidase F (PNGase F) (200 U) or endo-β-N-acetylglucosaminidase H (endo H) (200 U) at 37°C for 1 h according to the manufacturer's instructions (New England BioLabs). The deglycosylated protein was analyzed by Western blotting. Under nonreducing conditions, sE2 dissolved in phosphate-buffered saline (PBS) was digested with PNGase F at 4°C overnight. The resulting N-deglycosylated sE2 was purified with Ni-NTA resin and applied to ELISA, receptor-binding assay, and immunization experiments.

SDS-PAGE and Western blot assay.

Induced cell supernatants or purified sE2 proteins were separated on an SDS-12% PAGE gel and stained with Coomassie brilliant blue G-250 or transferred onto a polyvinylidene difluoride (PVDF) membrane (Pall). The membrane was probed with an anti-His tag monoclonal antibody (MAb) (1:1,000 dilution; M30111; Abmart) or anti-E2 MAb AP33 (35) (1:1,000 dilution; kindly provided by Arvind Patel, MRC-University of Glasgow), followed by a corresponding secondary antibody.

ELISA.

To determine the antigenicity of sE2, 96-well enzyme immunoassay (EIA)/radioimmunoassay (RIA) flat-bottom plates (Costar, Corning, NY, USA) were coated overnight with serially diluted sE2. The plates were blocked with PBS with Tween 20 (PBST) containing 5% nonfat dry milk for 1 h at 37°C. E2 MAb AR3A (kindly provided by Mansun Law and Denis Burton at the Scripps Research Institute) or AP33 (35) was then added, followed by incubation for 2 h at 37°C. Then, horseradish peroxidase (HRP)-conjugated anti-human IgG antibody (1:5,000 dilution; ab6858; Abcam) or HRP-conjugated anti-mouse IgG antibody (1:5,000 dilution; 31432; Invitrogen) was added and incubated for 1 h. After color development, colorimetric analysis was performed at 450 nm in a 96-well plate reader.

Receptor-binding assay.

The receptor-binding assay to analyze the capacity of sE2 binding to HCV entry factors (EF) has been previously described (34). Briefly, 100 μg of purified sE2 protein was incubated with parental CHO cells or with CHO cells expressing SR-BI or CD81 at 37°C for 1 h. After washing with PBS, the cells were incubated with anti-His mouse monoclonal antibody on ice for 1 h and then with Alexa Fluor 555-conjugated donkey anti-mouse IgG (1:200; A31570; Molecular Probes), followed by flow cytometry analysis on a BD LSRII flow cytometer (BD Biosciences, San Diego, CA). The results were analyzed with FlowJo software.

Human CD81LEL protein was expressed and purified to compete the binding of sE2. Briefly, pET26b-hCD81LEL was transformed into BL21 competent cells. Soluble hCD81LEL was obtained by low-temperature induction and purified using Ni-NTA resins according to the manufacturer's instructions. In the CD81LEL competitive-inhibition assay, sE2 was incubated with different doses (200 μg, 20 μg, and 2 μg) of CD81LEL at 37°C for 1 h, followed by the receptor-binding assay described above.

HCV infection-blocking assay.

To perform an HCV infection-blocking assay, Huh-7.5.1 cells cultured in complete DMEM were seeded at 1 × 10⁴ cells/well in a 96-well plate for 12 h. Then, approximately 4 × 10³ focus-forming units (FFU)/ml of Con1/JFH1 chimeric HCVcc was mixed with serially diluted sE2 protein, CD81LEL protein, or bovine serum albumin (BSA) (New England BioLabs) and added to the Huh-7.5.1 cells. After 4 h, the protein-virus mixture was changed to complete DMEM. After 72 h of culture, immunostaining was performed as previously described (36) with some modifications. Briefly, cells were fixed with fixation buffer (4% paraformaldehyde in PBS) and blocked with blocking buffer (3% BSA, 0.3% Triton X-100, and 10% FBS in PBS), followed by incubation with anti-HCV NS5A MAb (1:1,000 dilution; Abmart, Shanghai, China) with Alexa Fluor 488-conjugated donkey anti-mouse IgG (1:1,000 dilution; A21202; Molecular Probes) and Hoechst dye (1:10,000 dilution; Molecular Probes). The fluorescent foci were observed and enumerated on a standard fluorescence microscope (Leica).

Animal immunization.

BALB/c mice were purchased from Shanghai SLAC Laboratory Animal Inc. Prior to immunization, 50 μg sE2 antigens was formulated with 1 mg Inject Alum (Pierce), 1 mg Inject Alum plus 25 μg CpG 7909 (also known as CpG2006 or PF-3512676; 5′-TCGTCGTTTTGTCGTTTTGTCGTT-3′; synthesized by Sangon Biotech, Shanghai, China), or Freund's adjuvant (FA) (complete FA for prime and incomplete FA for boost; Sigma-Aldrich) at a volumetric ratio of 1:1 according to the manufacturer's instructions. To test the capacity of sE2-inducing NAbs, groups of 10 female BALB/c mice (6 weeks old) were injected intraperitoneally (i.p.) at weeks 0, 2, and 4. Blood samples were collected at weeks 0, 2, 4, 6, 13, 17, 22, and 25. To test sE2-induced immune memory, we boosted mice at week 25, when serum titers significantly decreased, and performed blood sampling at weeks 27 and 28, when the mice were terminated. The sera were kept at −80°C until use. To compare insect cell-derived sE2 and mammalian-cell-derived sE2 and to compare sE2 and sE2ΔHVR1, groups of six female BALB/c mice (6 weeks old) were i.p. injected at weeks 0, 2, 4, and 6. Blood samples were collected every 2 weeks. In a separate experiment, each of the three sE2 samples, representing different oligomeric states, was formulated with aluminum hydroxide (alum) and used for mouse immunization (6 mice/group) with the same dosage and immunization schedule described above, and mouse antisera were collected at week 12 for measurement of neutralization activities against HCVcc. All the animal immunization studies were approved by the Institutional Animal Care and Use Committee at the Institut Pasteur of Shanghai (protocol number A2013006). The animals were cared for in accordance with institutional guidelines.

Antibody measurement.

To measure E2-specific antibody responses in serum samples by ELISA, 96-well EIA/RIA flat-bottom plates were coated overnight with 100 ng/well of sE2. After blocking, serially diluted sera from mice were added as primary antibodies, followed by 1:5,000-diluted HRP-conjugated anti-mouse IgG antibody. After color development, colorimetric analysis was performed at 450 nm in a 96-well plate reader. For a given serum sample, the endpoint titer was defined as the reciprocal of the highest serum dilution that had an absorbance of > 0.1 optical density (OD) unit above that of the preimmune samples. For computation of geometric mean titers (GMTs), serum samples that did not yield positive readings at the lowest dilution tested (1:100) were assigned an endpoint titer of 50.

For measurement of sE2-specific IgG1 and IgG2a responses, sera from each mouse were diluted 1:10,000 and then subjected to ELISA analysis using the above-described protocol with some modifications: HRP-conjugated goat anti-mouse IgG1 and goat anti-mouse IgG2a (SouthernBiotech) were used as the secondary antibodies, respectively. The ratio of IgG1 and IgG2a isotypes was calculated by dividing the OD values for IgG1 by the OD values for IgG2a.

Competitive ELISA.

HRP-conjugated AP33 and AR3A antibodies were generated by using an EZ-Link Plus Activated Peroxidase kit (Thermo Fisher Scientific). For the competitive ELISA, 96-well EIA/RIA flat-bottom plates were coated overnight with 100 ng/well of sE2. After blocking, serially diluted mouse antisera were added to the wells and incubated at 37°C for 1 h, followed by three washes with PBS. Then, HRP-conjugated AP33 (2 μg/ml) or HRP-conjugated AR3A (2 μg/ml) was added to the wells and incubated at 37°C for 1 h. After color development, colorimetric analysis was performed at 450 nm in a 96-well plate reader.

Measurement of neutralization activities of antisera.

A panel of HCVccs covering genotypes 1 to 7 was established. All these HCVccs were used in neutralization assays to assess the neutralization breadth of the antisera. Briefly, serum samples were heat inactivated at 56°C for 1 h and diluted to their appropriate dilutions in complete DMEM. HCVccs were diluted in complete DMEM to approximately 4 × 10³ FFU/ml, mixed with an equal volume (50 μl) of diluted serum samples, and incubated at 37°C for 2 h. The virus-serum mixture was transferred to Huh-7.5.1 cells seeded 12 h previously in 96-well plates (1 × 10⁴ cells/well) and replaced with complete DMEM after 4 to 6 h of incubation at 37°C. The cells were incubated at 37°C for 72 h, followed by fixing and NS5A immunostaining. The percent neutralization was calculated by comparing the focus numbers of immune sera to that of a preimmune serum control at the same dilutions.

Measurement of cellular immune responses by ELISPOT assay.

To determine the IFN-γ- and interleukin-4 (IL-4)-secreting cells in splenocytes, 96-well PVDF plates (Millipore) were precoated with anti-mouse IFN-γ capture antibody (an-18; eBiosciences) or anti-mouse IL-4 capture antibody (11B11; eBiosciences) at 4°C overnight. The plates were blocked with complete RPMI 1640 medium (Gibco) for 1 h at 37°C, and freshly isolated splenocytes were added to the plates. sE2, medium (negative control), or concanavalin A (positive control) was diluted in complete RPMI 1640 medium, added to each well at a final concentration of 10 μg/ml, and incubated for 48 h at 37°C and 5% CO₂. Subsequently, the plates were incubated with biotinylated anti-mouse IFN-γ detection antibody (R4-6A2; eBiosciences) or biotinylated anti-mouse IL-4 detection antibody (BVD6-24G2; eBiosciences) diluted in PBST for 2 h and then with alkaline phosphatase (AP)-conjugated streptavidin (Mabtech) diluted in PBS for 1 h. After washing, nitroblue tetrazolium (NBT)/BCIP (5-bromo-4-chloro-3-indolylphosphate) substrate (Promega) was added for color development. The cytokine-secreting cell spots were imaged and counted on a CTL Immunospot reader (Cellular Technology Ltd.).

Determination of the oligomeric state.

Ni-NTA-purified sE2 was subjected to SEC using the AKTA fast protein liquid chromatography (FPLC) system. Briefly, sE2 was loaded on a Superdex 200 10/300 GL column, and SEC was performed in PBS buffer at a flow rate of 0.5 ml/min. The eluent was analyzed by SDS-PAGE under reducing or nonreducing conditions. For the nonreducing SDS-PAGE, protein samples were mixed with sample buffer without β-mercaptoethanol and loaded into the gel without boiling.

Generation of recombinant HCV.

Construction of Jc1-derived bicistronic HCV encoding Cre recombinase (BiCre-Jc1) was described previously (12). Briefly, the plasmid carrying BiCre-Jc1 was linearized with XbaI and transcribed using Megascript T7 (Ambion). RNA was electroporated into Huh-7.5 cells using an ECM 830 electroporator (BTX Genetronics). The transfected Huh-7.5 cells were cultured in complete DMEM until 72 h after electroporation, when the medium was replaced with serum-free DMEM, and the supernatants were harvested every 6 h for 3 days. The supernatants were pooled, filtered through a 0.45-μm bottle top filter (Millipore), and concentrated using a stirred cell (Millipore). Viral titers (50% tissue culture infective dose [TCID₅₀]) were determined using Huh-7.5 cells as previously described (37).

Generation and production of recombinant adenovirus.

Adenovirus encoding human homologs of the two species-specific HCV entry factors CD81 and OCLN (AdV-hCD81-2A-hOCLN) was constructed via overlapping PCR of hCD81 and hOCLN templates and cloned into pShuttle. The pSh-hCD81-2A-hOCLN was then electroporated into BJ5183 AD-1 cells (Agilent), which were pretransformed with pAdEasy-1 to facilitate recombination with the pShuttle vector. Colonies were selected, prepped, and screened for a 3-kb band following PacI digestion.

The adenovirus constructs were then transfected into HEK293 cells (American Type Culture Collection) cultured in complete DMEM using the calcium phosphate method. The transfected cultures were maintained until the cells exhibited complete cytopathic effect (CPE) and then harvested and freeze-thawed. The supernatants were serially passaged two more times, with harvest at complete CPE and freeze-thaw. For virus purification, the cell pellets were resuspended in 0.01 M sodium phosphate buffer (pH 7.2) and lysed in 5% sodium deoxycholate, followed by DNase I digestion. The lysates were centrifuged, and the supernatant was layered onto a CsCl gradient (1.2 to 1.46 g/ml) and spun at 23,000 rpm in a Beckman Optima 100K Ultracentrifuge using an SW28 spinning bucket rotor (Beckman Coulter). The adenovirus bands were isolated and further purified on a second CsCl gradient using an SW41 spinning bucket rotor. The resulting purified adenoviral bands were isolated using an 18.5-gauge needle and twice dialyzed against 4% sucrose. Adenovirus concentrations were measured at 10¹² times the dilution factor times the OD at 260 nm (OD₂₆₀) reading on a Nanodrop 2000 (Thermo Fisher Scientific). The adenovirus stocks were aliquoted and stored at −80°C.

Active immunization and bioluminescence imaging.

Gt(ROSA)26Sor^{tm1(Luc)Kaelin} mice (Rosa26-Fluc) were obtained from the Jackson Laboratory. The mice were bred and maintained at the Laboratory Animal Resources of Princeton University according to guidelines established by the Institutional Animal Committee (protocol number 1930). After sE2 immunization, the Rosa26-Fluc mice were injected with 10¹¹ adenovirus particles 24 h before intravenous injection with 2 × 10⁷ TCID₅₀ HCV-BiCre-Jc1. At 72 h postinfection, the mice were anesthetized using an isoflurane inhalation anesthesia and injected intraperitoneally with 1.5 mg of d-luciferin (Caliper Life Sciences). Bioluminescence was measured using an Ivis Lumina II platform (Caliper Life Sciences).

Statistics.

Significance comparisons were calculated with a two-tailed Student t test or Kruskal-Wallis one-way analysis of variance (ANOVA). For correlations, a nonparametric Spearman test was used. All statistical analyses were performed with GraphPad Prism 5.0c (GraphPad Software, CA).

RESULTS

High-yield production of soluble E2 glycoprotein in Drosophila S2 cells.

Previous efforts to develop a recombinant HCV vaccine relied on expression of E2 or E1/E2 heterodimer in mammalian cells (27, 28, 38–40). It was reported that N-glycans associated with HCV E2 could modulate receptor-binding affinity, as well as neutralizing-epitope recognition (41). Thus, we hypothesized that glycosylation might influence the antigenicity and immunogenicity of sE2. To directly determine the contribution of glycosylation to the induction of bNAbs by sE2, we produced sE2 in mammalian and insect cell lines known to yield altered glycan structures on proteins (42). To produce the E2 protein of the HCV strain Con1 (genotype 1b) in insect cells, we used an established Drosophila S2 cell expression system (43). Transgenic cell lines expressing sE2 comprised of residues 384 to 661 were recovered following transfection of Drosophila S2 cells with pMT-sE2 and subsequent antibiotic selection (Fig. 1A). The sE2 protein was secreted into the cell culture supernatant and remained stable for at least 9 days (Fig. 1B). sE2 was readily purified from the cell culture supernatant to near homogeneity (Fig. 1C) at high levels of up to 100 mg/liter of supernatant. The purified sE2 migrated at ∼45 kDa (Fig. 1C), which is much higher than the predicted mass (∼34 kDa) based on its amino acid sequence, suggesting possible glycosylation. To examine the extent and pattern of glycosylation, sE2 was digested with the endoglycosidases endo H and PNGase F. As shown in Fig. 1D, PNGase F treatment of sE2 generated a band at the calculated molecular mass of ∼34 kDa, suggesting that sE2 is fully deglycosylated by PNGase F. Endo H digestion resulted in a band only slightly below that of untreated sE2, suggesting that sE2 contains endo H-resistant glycan types, such as paucimannose N-glycans, as previously reported for other S2 cell-produced glycoproteins (42).

FIG 1.

Expression and characterization of sE2 derived from stably transfected Drosophila S2 cells. (A) Schematic diagrams of sE2 expression constructs. A truncated E2 gene (aa 384 to 661) from the Con1 strain was inserted between the BiP signal peptide and the His tag. (B) Western blot analysis of sE2 accumulation in the supernatant of the stably transfected S2 cell culture after different periods of induction. (C) SDS-PAGE analysis of purified sE2. (D) Analysis of sE2 glycosylation by PNGase F or endo H digestion. Glycosidase-treated and untreated samples were then subjected to Western blotting with anti-E2 (AP33) or anti-His MAb as the detection antibody. (E) Receptor-binding assay. sE2 protein was incubated with wild-type CHO (CHO-WT), CHO-CD81, or CHO-SRB1 cells stained with anti-His MAb, followed by Alexa Fluor-555-conjugated anti-mouse IgG, and detected by flow cytometry. PE-A indicates the signal from the PE laser. (F) Dose-dependently competitive inhibition of sE2 binding to CHO-CD81 cells by CD81LEL. sE2 was incubated with different doses of CD81LEL before performing the receptor-binding assay. (G) Blockade of HCVcc infection by sE2. Serially diluted sE2 was mixed with HCVcc, and the mixtures were added to Huh-7.5.1 cells to allow infection for 4 h. CD81LEL and BSA were set as controls. NS5A immunostaining was performed at 72 h postinfection. Means ± standard errors of the mean (SEM) of triplicates are shown. (H) Recognition of sE2 by neutralizing MAb AR3A (1 μg/ml) or AP33 (1 μg/ml). Means ± SEM of the OD₄₅₀ readings from triplicate wells are shown.

Functional and conformational characterization of sE2.

Proper folding of recombinant sE2 is critical for inducing antibodies capable of binding to the envelopes of HCV particles. Thus, we performed a variety of tests to ensure that sE2 retained its native conformation. CD81 and SRB1 are two critical HCV entry factors and directly interact with E2 (44, 45), but only when the E2 protein is properly folded. To determine whether S2 cell-produced sE2 can bind CD81 and SRB1, CHO cells stably expressing human CD81 (designated CHO-CD81) or SRB1 (CHO-SRB1) were incubated with sE2, and the binding of sE2 to the CHO-CD81 or CHO-SRB1 cells was assessed by flow cytometry (34). As shown in Fig. 1E, sE2 bound to both CHO-CD81 and CHO-SRB1, and binding to CHO-CD81 was inhibited by recombinant soluble large extracellular loop of CD81 (CD81LEL) in a dose-dependent manner (Fig. 1F), in agreement with the previous finding that CD81LEL is the binding domain of E2 (45). Next, we tested sE2 for its ability to block HCVcc infection. As shown in Fig. 1G, HCVcc infection of Huh-7.5.1 cells was inhibited by sE2 in a dose-dependent manner, demonstrating that sE2 competes with HCVcc for receptor-binding sites.

The conformation of sE2 was further assessed by using two well-characterized, E2-specific, broadly neutralizing MAbs, AR3A (10) and AP33 (35), which target conformational and linear epitopes, respectively. ELISA analysis revealed that sE2 efficiently reacted with both AR3A and AP33 in a dose-dependent manner (Fig. 1H). Because AR3A recognizes a conformational epitope that overlaps the CD81 binding site (10, 46), our results demonstrate that sE2 acquires a conformation critical for binding both key receptors and bNAbs and therefore has a high potential to elicit bNAbs. Thus, collectively, our data strongly suggest that our sE2 antigen is properly folded.

Immunization of mice with sE2 induces HCV E2-specific antibodies that can neutralize a diverse panel of HCV genotypes.

The immunogenicity of our recombinant sE2 vaccine was first evaluated in BALB/c mice in combination with aluminum hydroxide (alum), alum plus CpG7909, or FA. As shown in Fig. 2A, the sE2-specific antibody titers of all the mice from the three vaccine groups peaked at week 6 and gradually decreased, but remained above 1,000 until week 22. Following a homologous boost at week 25, a drastic increase in sE2-specific antibody titers was observed at week 27 for the three vaccine groups, but not the control group (Fig. 2A), indicating the presence of sE2-specific B cell memory. Among the three experimental vaccines, sE2 adjuvanted with FA, which is not approved for human use, elicited the highest antibody titers at week 6 (Fig. 2B). The sE2-specific antibody titers of the sE2/alum/CpG7909 group were significantly higher than those of the sE2/alum group (Fig. 2B). Moreover, the sE2/alum group predominantly produced the IgG1 subclass of antibodies (Fig. 2C), confirming that alum is a T helper 2 (Th2)-biased adjuvant. The addition of CpG7909 resulted in a low IgG1/IgG2a ratio (Fig. 2C), indicating that CpG redirects the Th cell bias from Th2 to Th1.

FIG 2.

Induction of antibody- and cellular-mediated immune responses in mice. (A) Kinetics of sE2-specific antibody titers. BALB/c mice (n = 10 per group) were immunized intraperitoneally at weeks 0, 2, 4, and 25 (arrows), and the serum titers were measured by ELISA. The data are expressed as the means ± SEM of the endpoint titers for each group. (B) Anti-sE2 IgG titers at week 6. (C) IgG1/IgG2a ratios of sE2-specific antibodies at week 6. The horizontal lines indicate the geometric means for each group. The asterisks represent significant differences between groups. Statistical significance was calculated by Kruskal-Wallis one-way ANOVA. (D) Cellular immune responses as measured by IFN-γ and IL-4 ELISPOT assays. Splenocytes were isolated from mouse spleens at week 27, pooled, and stimulated with sE2 protein. The results are expressed as spot-forming cells (SFCs) per 10⁶ splenocytes. Means and SEM of triplicate wells are shown. The asterisks represent significant differences (two-tailed Student t test) between medium (white bars) and sE2 stimulation (blue bars) in each group. ns, no significance (P ≥ 0.05); *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Antigen-specific T cell responses induced by the immunizations were assessed by ELISPOT assay in splenocytes. The three vaccine groups, but not the control, showed significantly higher numbers of IFN-γ- and IL-4-secreting cells (Fig. 2D). Therefore, immunization of mice with sE2 can elicit antigen-specific IFN-γ and IL-4 T cell memory, which can be rapidly recalled to respond to subsequent viral antigen exposure.

The ability of the mouse antisera to inhibit HCV infection in vitro was assessed by a microneutralization assay using a panel of chimeric HCVccs containing envelope proteins of 12 different strains, which cover nine subgenotypes in all seven genotypes (30–33). As shown in Fig. 3A, anti-sE2 sera at a 1:40 dilution were able to efficiently neutralize (neutralization ≥ 50%) almost all of the HCVccs in the panel. Antisera from a minority of the mice had relatively poor neutralization (neutralization < 50%) against only J6 and QC69. Notably, PR52B6mt and PR79L9, containing envelope proteins constructed directly from two clinical isolates (33), were effectively neutralized. As shown in Table 1, for the three sE2 vaccine groups, the pooled serum titers that yielded ≥50% neutralization (NT₅₀) were no less than 40 against all the HCVccs tested. Among them, the sE2/alum/CpG antisera appeared to be the most effective in terms of overall neutralization potency and breadth, with NT₅₀s of ≥160 for all the HCVcc strains.

FIG 3.

Induction of bNAbs against HCVccs of genotypes 1 to 7 in mice. (A) Neutralization assay. Mouse antisera collected at week 27 were diluted 1:40 and then tested for neutralization of a panel of HCVccs consisting of 12 strains from 7 genotypes. The names of the HCVccs are shown as “strain(genotype)” in each graph. Each symbol represents one animal, and the horizontal lines indicate the geometric means for each group. The data are representative of the results of three independent experiments. (B) Competitive ELISA. Mouse antisera were serially diluted and tested for inhibition of AR3A and AP33 binding of sE2. Means ± SEM of the OD₄₅₀ readings for all animals in each group are shown.

TABLE 1.

NT₅₀s of mouse antisera against HCVccs of genotypes 1 to 7^a

Antiserum	NT50
	1a (H77)	1b			2a		2b (J8)	3a (S52)	4a (ED43)	5a (SA13)	6a (HK6a)	7a (QC69)
		Con1	PR52 B6mt	PR79 L9	JFH1	J6
PBS	<5	<5	<5	<5	<5	<5	<5	<5	<5	<5	<5	<5
sE2 + alum	160	320	160	640	1,280	160	80	80	80	160	80	160
sE2 + alum + CpG	320	320	160	640	1,280	160	160	160	320	160	160	160
sE2 + FA	80	640	160	1,280	640	80	80	40	160	80	80	40

^aThe mouse antisera collected at week 27 were pooled for each group and then used for neutralization tests. The NT₅₀ was defined as the highest dilution of serum able to neutralize 50% of HCVcc infectivity. The data are representative results from three independent experiments.

The above-mentioned data demonstrate that sE2 can elicit the production of bNAbs. To determine whether the induced bNAbs target different neutralizing epitopes, we developed a competitive ELISA in which HRP-conjugated AP33 and AR3A were used as the detection antibodies. The results from competitive ELISAs showed that sE2 binding to both AP33 and AP3A was inhibited by preincubation with the antisera from the three sE2 vaccine groups in a dose-dependent manner, whereas the control sera from the PBS group did not exhibit significant inhibitory effects (Fig. 3B). These data indicate the presence of both AP33-like and AR3A-like broadly neutralizing antibodies in the anti-sE2 sera, which may contribute to the observed broad neutralization by the anti-sE2 sera.

The oligomeric state of sE2 does not affect bNAb induction.

A previous study showed that a similar version of sE2 produced in mammalian cells existed primarily in monomeric and dimeric forms (40). To define the oligomeric state of our insect cell-produced sE2, we subjected the Ni-NTA-purified sE2 to SEC analysis. Three major peaks were identified in the SEC chromatogram (Fig. 4A). Nonreducing SDS-PAGE showed that monomers (∼45 kDa) and dimers (∼90 kDa) were predominantly detected in peak 3 and peak 2, respectively, whereas peak 1 contained large-molecular-mass proteins probably representing sE2 megamers (Fig. 4B). To investigate the impact of the sE2 oligomeric state on bNAb induction, we immunized three groups of BALB/c mice with equal amounts (50 μg) of sE2 antigens concentrated from the three peaks, respectively, and compared the neutralizations of the resulting mouse antisera against different HCVcc genotypes. As shown in Fig. 4C, despite the fact that the antisera from mice immunized with the peak 3 antigen exhibited decreased neutralization activity against the homologous strain Con1 compared to the antisera from the other two groups (peak 1 and peak 2), no significant difference in neutralization against all of the heterologous strains tested was observed for the three sE2 vaccine groups. These results indicate that the oligomeric state of sE2 does not affect its ability to induce bNAbs.

FIG 4.

Analysis of oligomeric states of sE2 and their bNAb-inducing abilities. (A) Size exclusion chromatography of insect sE2 revealed three major peaks representing different oligomeric states. (B) Analysis of the samples from the three peaks by SDS-PAGE under nonreducing and reducing conditions. Megamers, tetramers (∼170 kDa), trimers (∼130 kDa), dimers (∼90 kDa), and monomers (∼45 kDa) are indicated by the arrows. (C) Neutralization activities of antisera. Four groups of mice were immunized with PBS or one of the antigens from the three peaks, and the resulting antisera were tested at 1:40 dilution for neutralization of a panel of HCVccs, as indicated. The asterisks represent significant differences (Kruskal-Wallis one-way ANOVA) between groups: ns, no significance; *, P < 0.05; **, P < 0.01.

The unique glycosylation of insect cell-derived sE2 is critical for its ability to induce broadly neutralizing antibodies.

Next, we analyzed the contribution of glycosylation to the induction of bNAbs by directly comparing the immunogenicity of sE2 produced in human HEK293T cells (Fig. 5A and B) with that of sE2 produced in Drosophila S2. Upon PNGase F treatment, both proteins yielded the same ∼34-kDa band (Fig. 5A and C); however, endo H-treated mammalian sE2 showed a smeared (from 70 kDa to 35 kDa) banding pattern (Fig. 5A), which differed from that of endo H-treated insect sE2 (Fig. 1D). These results demonstrate that the glycosylation patterns of insect and mammalian HCV sE2s are different, which is in line with previous studies comparing the envelope proteins of other viruses, e.g., influenza A virus, expressed in cells of heterologous species (42). Then, insect sE2 and mammalian sE2 were compared for MAb recognition. Both sE2 versions reacted with the AP33 MAb in identical fashions, whereas mammalian sE2 was poorly recognized by the MAb AR3A compared to insect sE2 (Fig. 5D); however, after PNGase F digestion to remove N-linked glycans (Fig. 5C), both deglycosylated sE2 proteins reacted equally with AR3A and AP33 antibodies (Fig. 5E). These antibody binding data indicate that glycosylation can affect epitope exposure on sE2. Moreover, mammalian sE2 bound CD81 and SRB1 less efficiently than insect sE2 did, whereas their deglycosylated forms performed equally (Fig. 5F), although both sE2 versions compete with CD81LEL (Fig. 5G and H) and HCVcc (Fig. 5I). These results suggest that the better receptor recognition by insect sE2 is due to its unique glycosylation pattern.

FIG 5.

Comparative analyses of mammalian-cell- and insect cell-derived sE2 proteins. (A) Western blot analysis of mammalian-cell-derived sE2 without or with PNGase F or endo H digestion. (B) SDS-PAGE analysis of insect sE2 and mammalian sE2. (C) SDS-PAGE analysis of PNGase F-treated insect sE2 and mammalian sE2. N-deg., N-deglycosylated. (D and E) Reactivities of different sE2 forms with neutralizing MAb AR3A (1 μg/ml) or AP33 (5 μg/ml) in ELISAs. The error bars indicate means ± SEM. Statistical significance was calculated by two-tailed Student t test: ns, no significance (P ≥ 0.05); ***, P < 0.001. Representative results of three independent experiments are shown. (F) Binding of glycosylated or N-deglycosylated mammalian and insect sE2 to CHO-WT, CHO-CD81, or CHO-SRB1 cells measured by flow cytometry. Representative results of three independent experiments are shown. PE, phycoerythrin. (G and H) Competitive inhibition of insect sE2 (G) or mammalian sE2 (H) binding to CHO-CD81 cells by CD81LEL. (I) Comparison of HCVcc infection blocking by insect sE2 and mammalian sE2. Means ± SEM of triplicates are shown. (J) sE2-specific antibody endpoint titers. Groups of mice (n = 6 per group) were immunized with 40 μg of mammalian sE2, insect sE2, N-deglycosylated mammalian sE2, or N-deglycosylated insect sE2 in the presence of 500 μg alum adjuvant. Another group was injected with PBS plus alum, serving as the control. sE2-specific antibody endpoint titers were determined by ELISA. Each symbol represents one animal, and the line indicates the geometric mean value of the group. Statistical significance was calculated by Kruskal-Wallis one-way ANOVA; *, P < 0.05; ***, P < 0.001. (K) NT₅₀s of the pooled antisera against a panel of HCVccs.

To assess the impact of glycosylation on sE2's immunogenicity and NAb-inducing ability, we immunized four groups of BALB/c mice with insect sE2, mammalian sE2, or their N-linked deglycosylated counterparts. Among the four groups, the one immunized with insect sE2 produced the highest binding antibody titers (Fig. 5J), as well as the highest neutralization antibody titers, against all HCVccs tested (Fig. 5K). Specifically, antisera induced by insect sE2 showed NT₅₀s of ≥160 against all the tested HCVcc strains, whereas antisera of the mammalian sE2 group exhibited NT₅₀s of ≥160 for only two strains (Con1 and JFH1) (Fig. 5K). In addition, removal of N-glycans from insect sE2 resulted in reduced sE2 binding and neutralization antibody titers (Fig. 5K). These data suggest that the unique glycosylation in insect cell-derived sE2 is critical for its ability to induce pangenotypic neutralizing antibodies.

The mutant with HVR1 deleted is equivalent to the intact sE2 in eliciting bNAbs.

HVR1 is a highly variant region in the N terminus of E2 and is a predominant neutralizing epitope in the context of natural infection. However, anti-HVR1 NAbs are mostly isolate specific (18, 46), and HVR1 has been shown to play a detrimental role in virus neutralization (47–49). We therefore evaluated whether the deletion of HVR1 from sE2 (sE2ΔHVR1) (Fig. 6A) would affect its conformation and ability to induce cross-reactive NAbs. sE2ΔHVR1 retained the ability to bind CD81 but bound SRB1 poorly (Fig. 6B), indicating that HVR1 is involved in interacting with SRB1, in agreement with previous findings (49). The ability of sE2ΔHVR1 to inhibit HCVcc infection was significantly impaired compared to that of full-length sE2 (Fig. 6C), likely due to the failure of sE2ΔHVR1 to block the interaction between HCVcc and SRB1. Both sE2ΔHVR1 and full-length sE2 were examined for the ability to react with MAbs AR3A and AP33. As shown in Fig. 6D, the reactivity of sE2ΔHVR1 to AR3A was nearly identical to that of full-length sE2, whereas binding of sE2ΔHVR1 to AP33 was much greater than that of full-length sE2, suggesting that HVR1 (aa 384 to 410), present on full-length sE2, somewhat masks the adjacent AP33 epitope (aa 412 to 423).

FIG 6.

Effect of HVR1 in induction of bNAbs. (A) SDS-PAGE analysis of purified sE2ΔHVR1 and full-length sE2. (B) Receptor binding functions of sE2ΔHVR1 and sE2 analyzed by flow cytometry. The deletion of HVR1 diminished sE2 binding to SRB1, while it did not affect sE2 binding to CD81. The results shown are representative of three independent experiments. (C) Significantly weaker blockage of HCVcc infection by sE2ΔHVR1 than by sE2. Mean values ± SEM from three independent experiments performed in duplicate are shown. The asterisks represent significant differences (two-tailed Student t test) between the sE2 and sE2ΔHVR1 groups; *, P < 0.05; **, P < 0.01; ***, P < 0.001. (D) Reactivity of sE2 and sE2ΔHVR1 to AR3A or AP33 MAb. ELISA plates were coated with serially diluted sE2, sE2ΔHVR1, or BSA samples, which were detected by AR3A or AP33 MAb. The data are mean values ± SEM of triplicate wells. (E and F) Comparison of neutralization activities of antisera induced by sE2 and by sE2ΔHVR1. (E) BALB/c mice (n = 6 per group) were injected intraperitoneally using sE2, sE2ΔHVR1, or PBS (all adjuvanted by alum) four times. sE2-specific or sE2ΔHVR1-specific antibody titers of mouse sera in each group were measured by ELISA. The mean values ± SEM of all animals in each group are shown. (F) Neutralization levels of sE2- and sE2ΔHVR1-immunized mouse sera (1:40 diluted) against HCVccs of genotypes 1 to 7 were compared. The horizontal lines indicate geometric means of each group. The results are representative of three independent experiments.

Next, we immunized BALB/c mice with sE2 and sE2ΔHVR1. The sE2- and sE2ΔHVR1-binding activities of antisera from the two groups were similar (Fig. 6E). Although the neutralizing ability of the anti-sE2ΔHVR1 sera against the homologous Con1 strain was significantly impaired compared to that of full-length sE2, no significant difference in neutralization of the 11 heterologous strains was observed between the two antisera (Fig. 6F). These results demonstrate that sE2ΔHVR1 is equivalent to the full-length sE2 in inducing broadly neutralizing antibodies.

Prophylactic efficacy of sE2 vaccine in a genetically humanized mouse model of HCV infection.

The in vivo protective efficacy of the sE2 vaccine was evaluated using a genetically humanized mouse model (12). Rosa26-Fluc mice immunized with sE2/alum generated high-titer sE2-specific antibodies, whereas those injected with PBS/alum did not (Fig. 7A and B) (n = 6 per group). Following injection of adenoviruses expressing HCV EF CD81 and OCLN, the immunized Rosa26-Fluc mice were subsequently challenged with BiCre-Jc1 (genotype 2a) (12). As shown in Fig. 7C and D, the bioluminescence signals were drastically reduced in the sE2/alum-immunized mice compared to the PBS/alum and the “no EF” groups. These results indicate that active immunization with sE2 can efficiently protect against HCV infection in vivo.

FIG 7.

Prophylactic efficacy of active vaccination with sE2. Rosa26-Fluc mice were injected intraperitoneally at weeks 0, 1, 3, and 6 (green arrows [A]) with 50 μg sE2 plus 1 μg alum per mouse (n = 6) or with PBS plus alum (n = 6). (A) Anti-sE2 endpoint titers were measured by ELISA and plotted for the indicated time points. (B) Anti-sE2 IgG titers for the day of challenge with HCVcc (red arrow [A]; day 72 after the first injection). Mean values ± SD of all animals in each group are shown. (C) In vivo imaging of HCVcc-challenged Rosa26-Fluc mice. At week 11, human HCV EF were delivered via adenovirus to the liver, and the mice were challenged with HCVcc expressing Cre recombinase (HCV-BiCre-Jc1, genotype 2a). An untreated cohort (n = 6) was challenged with HCVcc only without adenovirus delivery of human EF. (D) Bioluminescence was quantified 72 h following HCV–BiCre-Jc1 infection. The error bars represent standard deviations (SD). Statistical significance was calculated by Kruskal-Wallis one-way ANOVA; ****, P < 0.0001.

DISCUSSION

Glycosylated envelope proteins of viruses are often the targets for vaccine development. Thus far, the role of glycosylation in vaccine immunogenicity remains controversial (50–53). Structural analysis suggested that N-linked glycans in HCV E2 mask the exposed face on the E2 surface from NAbs (54, 55). In the present study, we aimed to analyze the impact of glycosylation on the immunogenicity of an HCV E2 protein-based recombinant vaccine candidate. We demonstrate that the glycosylation pattern associated with S2 cell-derived sE2 may be more favorable for eliciting bNAbs than mammalian-cell-produced sE2 (40). In our study, we show that the removal of N-glycans in sE2 markedly reduced the antibody titers (Fig. 5J). In addition, S2 cell-produced sE2, which is decorated with less complex glycans, not only was more immunogenic (Fig. 5J), but also had more exposed neutralizing epitopes (Fig. 5B to E) and receptor-binding domains (Fig. 5F) than its mammalian-cell-derived counterpart. This suggests that the unique glycosylation of S2 cell-produced sE2 is important for its bNAb-inducing ability.

In addition to the altered glycosylation of insect cell-derived sE2, we reason that increased conformational flexibility contributes to the greater immunogenicity we observed. sE2 expressed in S2 cells appears to have acquired a critical conformation suitable for inducing bNAbs, and such a conformation may not readily occur when E2 is presented in a more complex structure, such as a virion or E1/E2 heterodimer. For example, HVR1 present on HCV particles has been shown to obstruct the CD81 binding site, thereby masking conserved neutralizing epitopes (49), and an initial contact with SRB1 may be needed to unmask the binding site for E2 to interact with CD81. In addition, it has also been reported that deletion of HVR1 enhances the CD81 binding activity of E1/E2 heterodimers in HCV pseudoparticles (56). In contrast, we found that the deletion of HVR1 did not significantly affect the CD81 binding activity of sE2 (Fig. 6B), indicating that the HVR1 domain in the context of insect cell-produced sE2 does not functionally mask the CD81 binding site and therefore may facilitate the elicitation of CD81 binding-site-specific antibodies. In support of this notion, structural studies suggest that in sE2, the region from HVR1 (residues 384 to 410) to a conserved epitope (residues 412 to 423) is conformationally flexible (57, 58), which may aid in exposure of the conserved epitope. Conceivably, this flexibility enables sE2 to represent a critical E2 intermediate form(s) in a transition phase of HCV entry and may therefore be key to the ability of sE2 to induce pangenotypic NAbs.

Another important consideration for prioritizing HCV vaccine candidates is scalability for potential mass production. Although the yields of insect and mammalian sE2s may differ for other HCV strains, in this study, correctly folded sE2 of strain Con1 can be produced at high levels in the S2 expression system (100 mg/liter), considerably higher than the yields (1 to 2 mg/liter) of mammalian sE2 or a similar version termed eE2 (40). The yield could potentially be further improved by using a perfusion culture technology, as described previously (59). The identification of such a production system will be important for practical scaling up.

Thus far, only a few NAb-based vaccines have been evaluated in vivo due to the scarcity of small-animal models for HCV infection (12, 19, 20, 23). For example, an inactivated HCV vaccine was found to induce NAbs capable of protecting chimeric uPA/SCID mice from lower-dose (10³ RNA copies of HCVcc) HCV challenge, but not higher doses (10⁴ and 10⁵ RNA copies of HCVcc) (23). In this study, we used immunocompetent mice to assess the immunogenicity, as well as the prophylactic efficacy, of the vaccine candidate. Proof of concept for the utility of this genetically humanized mouse model to test antibody-based vaccine candidates was previously established with a recombinant vaccinia virus vector expressing the structural proteins of a genotype 1a HCV vaccine strain. Here, we employed this model to assess the efficacy of our recombinant vaccine candidate. We found that genetically humanized mice immunized with Con1 (genotype 1b)-derived sE2 were protected from heterologous challenge with 2 × 10⁷ TCID₅₀ of BiCre-Jc1 (genotype 2a) HCVcc. As BiCre-Jc1 HCVcc is a strain that is relatively difficult to neutralize (Fig. 3A) (10–13), the observed cross-protection highlights the potential of sE2 as a broadly protective HCV vaccine. While evaluation of the efficacy of this vaccine in chimpanzees, the only other known species readily susceptible to HCV infection, would be desirable, experimentation in great apes is banned in most countries and/or is no longer supported by federal funding due to ethical concerns. Thus, to date, the genetically humanized mouse model remains the only immunocompetent-animal model available to test preclinically the efficacy of vaccine candidates. We should point out that this mouse model is not without caveats. For example, since not all of the identified human receptors/cofactors for HCV (17) are present in the model, it may be easier to prevent infection in the model than with natural infection. Nevertheless, we have previously shown that levels of protection correlate well with the immunogenicity of vectored vaccines in the model (12), and thus, such humanized mice seem adequate for evaluating the efficacy of our recombinant HCV vaccine.

In conclusion, an sE2 vaccine based on expression of viral envelope proteins in insect cells has multiple benefits. First, the simple glycosylation and high flexibility of insect cell-derived sE2 may aid its ability to induce bNAbs, as demonstrated by the very broad NAbs observed in mice. Second, industrial-production-related advantages, such as simple composition, high yield, and ease of purification, will greatly reduce production costs and also make it possible to develop a multivalent vaccine formulation for broader coverage, if needed. Third, the in vivo protective efficacy of sE2 was confirmed by an active immunization, instead of a passive immunization, in a humanized mouse model, which is the only immunocompetent-animal model (except chimpanzees) of HCV infection currently available. These results should encourage further preclinical and clinical development of an sE2-based, broadly protective HCV vaccine.

Funding Statement

This work, including the efforts of Jin Zhong, was funded by the National Natural Science Foundation of China (81330039), the Chinese National 973 Program (2015CB554300), and the CAS-SAFEA International Partnership Program for Creative Research Teams. This work, including the efforts of Zhong Huang, was funded by the Ministry of Science and Technology of the People's Republic of China (MOST) (2012ZX10002007-003). This work, including the efforts of Alexander Ploss, was funded by HHS | National Institutes of Health (NIH) (AI079031-05), the Burroughs Wellcome Fund (BWF), and the American Cancer Society (ACS) (RSG-15-048-01 MPC). This work, including the efforts of Markus von Schaewen, was funded by the German Research Foundation.