Inactivated whole hepatitis C virus vaccine employing a licensed adjuvant elicits cross-genotype neutralizing antibodies in mice

Background & Aims: A prophylactic vaccine is required to eliminate HCV as a global public health threat. We developed whole virus inactivated HCV vaccine candidates employing a licensed adjuvant. Further, we investigated the effects of HCV envelope protein modi ﬁ cations (to increase neutralization epitope exposure) on immunogenicity. Methods: Whole virus vaccine antigen was produced in Huh7.5 hepatoma cells, processed using a multistep protocol and formulated with adjuvant (MF-59 analogue AddaVax or aluminium hydroxide). We investigated the capacity of IgG puri ﬁ ed from the serum of immunized BALB/c mice to neutralize genotype 1-6 HCV (by virus neutralization assays) and to bind homologous envelope proteins (by ELISA). Viruses used for immunizations were (i) HCV5aHi with strain SA13 envelope proteins and modi ﬁ cation of an O-linked glycosylation site in E2 (T385P), (ii) HCV5aHi(T385) with reversion of T385P to T385, featuring the original E2 sequence determined in vivo and (iii) HCV5aHi( D HVR1) with deletion of HVR1. For these viruses, epitope exposure was investigated using human monoclonal (AR3A and AR4A) and polyclonal (C211 and H06) antibodies in neutralization assays. Results: Processed HCV5aHi formulated with AddaVax induced antibodies that ef ﬁ ciently bound homologous envelope proteins and broadly neutralized cultured genotype 1-6 HCV, with half maximal inhibitory concentrations of between 14 and 192 l g/ml (mean of 36 l g/ml against the homologous virus). Vaccination with aluminium hydroxide was less immunogenic. Compared to HCV5aHi(T385) with the original E2 sequence, HCV5aHi with a modi ﬁ ed glycosylation and HCV5aHi( D HVR1) epitope immunogenicity.


Inactivated whole hepatitis C virus vaccine employing a licensed adjuvant elicits cross-genotype neutralizing antibodies in mice
Graphical abstract HCV processing and inactivation 1

HCV5aHi(∆HVR1)
with deletion of HVR1.  Inactivated whole hepatitis C virus vaccine employing a licensed adjuvant elicits cross-genotype neutralizing antibodies in mice Background & Aims: A prophylactic vaccine is required to eliminate HCV as a global public health threat. We developed whole virus inactivated HCV vaccine candidates employing a licensed adjuvant. Further, we investigated the effects of HCV envelope protein modifications (to increase neutralization epitope exposure) on immunogenicity. Methods: Whole virus vaccine antigen was produced in Huh7.5 hepatoma cells, processed using a multistep protocol and formulated with adjuvant (MF-59 analogue AddaVax or aluminium hydroxide). We investigated the capacity of IgG purified from the serum of immunized BALB/c mice to neutralize genotype 1-6 HCV (by virus neutralization assays) and to bind homologous envelope proteins (by ELISA). Viruses used for immunizations were (i) HCV5aHi with strain SA13 envelope proteins and modification of an O-linked glycosylation site in E2 (T385P), (ii) HCV5aHi(T385) with reversion of T385P to T385, featuring the original E2 sequence determined in vivo and (iii) HCV5aHi(DHVR1) with deletion of HVR1. For these viruses, epitope exposure was investigated using human monoclonal (AR3A and AR4A) and polyclonal (C211 and H06) antibodies in neutralization assays. Results: Processed HCV5aHi formulated with AddaVax induced antibodies that efficiently bound homologous envelope proteins and broadly neutralized cultured genotype 1-6 HCV, with half maximal inhibitory concentrations of between 14 and 192 lg/ml (mean of 36 lg/ml against the homologous virus). Vaccination with aluminium hydroxide was less immunogenic. Compared to HCV5aHi(T385) with the original E2 sequence, HCV5aHi with a modified glycosylation site and HCV5aHi(DHVR1) without HVR1 showed increased neutralization epitope exposure but similar immunogenicity. Conclusion: Using an adjuvant suitable for human use, we developed inactivated whole HCV vaccine candidates that induced broadly neutralizing antibodies, which warrant investigation in further pre-clinical studies. Lay summary: A vaccine against hepatitis C virus (HCV) is needed to prevent the estimated 2 million new infections and 400,000 deaths caused by this virus each year. We developed inactivated whole HCV vaccine candidates using adjuvants licensed for human use, which, following immunization of mice, induced antibodies that efficiently neutralized all HCV genotypes with recognized epidemiological importance. HCV variants with modified envelope proteins exhibited similar immunogenicity as the virus with the original envelope proteins.

Introduction
Worldwide 2 million acute HCV infections occur yearly, leading to chronic infection in 75% of cases. 1 More than 70 million individuals are estimated to be chronically infected, resulting in 400,000 annual deaths, mostly due to cirrhosis and hepatocellular carcinoma. 1 Given the asymptomatic nature of HCV infection prior to development of severe (and often irreversible) liver damage and the lack of screening programs, it is estimated that less than 20% of infections are diagnosed. Further, not all diagnosed individuals receive treatment with direct-acting antivirals, partly due to their high cost. 1 Moreover, treatment does not protect against reinfection, can be associated with severe side effects in hepatitis B virus co-infected individuals, and does not always eliminate the risk of hepatocellular carcinoma following HCV clearance. Finally, observed emergence of antiviral resistance could compromise future treatment efficacy. Therefore, a prophylactic vaccine is essential to achieve the World Health Organization's objectives for HCV elimination as a major public health threat. 2 Ideally, a vaccine should protect against different HCV variants. 3 Among the 8 reported major HCV genotypes, genotype 1-6 are epidemiologically significant. There are various subtypes. Genotypes and subtypes differ in 30% and 20% of their sequence, respectively, and show differential sensitivity to neutralizing antibodies (nAbs). [4][5][6][7][8][9] The HCV envelope (E) Keywords: hepatitis C virus; hepatitis; neutralizing antibodies; HCV vaccine; whole viral particle vaccine; HCV inactivation; HCV downstream processing; adjuvant; envelope glycoprotein; hypervariable region; glycosylation. glycoproteins (gp) E1 and E2 are the main targets of nAbs. 10 The highly variable 27 amino acid motif hypervariable region 1 (HVR1) at the E2 N-terminus mediates HCV evasion from nAbs; it acts as immunological decoy that induces nAbs, which are rendered inefficient due to mutational escape, and facilitates closed envelope protein conformational states, which restrict nAb access to conserved neutralization epitopes. 4,11,12 Further, HCV envelope protein glycosylation induced closed envelope protein states that protect conserved epitopes. 12,13 Thus, in vitro deletion of HVR1 and mutation of glycosylation sites increased HCV sensitivity to nAbs. 4,[11][12][13][14] Protective immunity against HCV is achievable as 25% of acute infections are cleared, likely by nAbs and T cells. 15,16 In a vaccine setting, nAbs might be sufficient for protection, as most licensed viral vaccines protect by nAbs. [16][17][18] Further, early development of nAbs was predictive of HCV clearance in humans and passive immunization with nAbs prevented HCV infection in chimpanzees. [19][20][21][22][23] Moreover, nAbs induced by a recombinant gpE1/gpE2 vaccine had protective effects in chimpanzees. 24,25 In contrast, a viral vector-based vaccine inducing HCV-specific T cells did not protect against chronic HCV infection in chimpanzees and humans. 26,27 In humans, the gpE1/gpE2 vaccine yielded robust nAbs in <50% of immunized individuals. 28,29 Compared to such subunit vaccines, whole virus vaccines show superior immunogenicity owing to a more native envelope protein conformation, a broader epitope array and a denser epitope presentation. 30,31 The development of a whole virus HCV vaccine only became feasible following the development of infectious cell culture systems for HCV production. 32 However, proof of the immunogenicity of cellculture-produced HCV was only obtained with non-licensed adjuvants. 31,33 Additionally, it remains an unresolved question whether envelope protein modifications designed to expose conserved neutralization epitopes improve HCV immunogenicity in the context of a whole virus vaccine.
We aimed to develop inactivated HCV vaccine candidates employing adjuvants licensed for human use and to investigate their immunogenicity in mice. We evaluated the capacity of vaccine-induced antibodies to broadly neutralize cell-cultureinfectious HCV genotype 1-6 recombinants. Finally, we evaluated the impact of envelope protein modifications on vaccine immunogenicity by deleting HVR1 and mutating a glycosylation site.
HCV cell culture Huh7.5 cells were maintained as described. 38,39 Generation of HCV virus stocks and further cell-culture-adaptation of HCV5a-Hi(DHVR1) by serial passage are described in the supplementary materials and methods.
Production of HCV for vaccine generation HCV-infected Huh7.5 cells were seeded in 10-layer cell factories (Thermo Fisher Scientific) and maintained in serum-free Adenovirus Expression Medium (ThermoFisher Scientific) 40 during the HCV production phase. For details see the supplementary materials and methods.

Evaluation of HCV-infected cell cultures
Details regarding the evaluation of the percentage of HCV antigen-positive cells in infected cell cultures by immunostaining, 37,39 as well as HCV infectivity, 38,41,42 RNA 39 and core titers 43 in culture supernatants, are provided in the supplementary materials and methods.

HCV sequencing
Sanger sequencing of the HCV sequence of DNA maxipreparations or of amplicons of the HCV RNA genome of cellculture-derived HCV generated by reverse-transcription PCR was carried out at Macrogen Europe as described 38,44 and as detailed in the supplementary materials and methods.
HCV concentration HCV in serum-free culture supernatant was concentrated using a multistep process consisting of tangential flow filtration, ultracentrifugation and chromatography steps, as specified in the supplementary materials and methods.

Mouse serum IgG purification and quantification
IgG was purified (Amicon ® Pro Affinity Concentration Kit Protein G, Millipore), concentrated (Vivaspin ® 500, 30,000 molecular weight cut-off (MWCO), GE Lifesciences) and quantified (IgG (total) Mouse Uncoated ELISA Kit, ThermoFisher or Cedex Bio Analyzer, Roche) according to the manufacturers instructions and as specified in the supplementary materials and methods.
In vitro neutralization assays Neutralization assays for characterization of HCV with envelope protein modifications were performed as described, 35 using human monoclonal antibodies (mAbs: AR3A, 45 AR4A 46 ) or human polyclonal IgG preparations (C211 12 or H06 47 ). Neutralization assays with purified mouse IgG were carried out in a smaller volume than assays for characterization of HCV. The E1/E2 sequence of viruses used in neutralization assays was sequence confirmed to be identical to the plasmid sequence. For details see the supplementary materials and methods.
E1/E2 complexes and soluble E2 (sE2) ELISA Binding of mouse serum IgG to recombinantly expressed native HCV5a (strain SA13) E1/E2 complexes derived from cell lysates of transfected HEK293T cells or his-tag purified HCV5a SA13 sE2 derived from supernatant of transfected HEK293T cells was evaluated by ELISA, as described in the supplementary materials and methods.

Results
Production of HCV vaccine antigen HCV for immunizations was produced in Huh7.5 cells in 10-layer cell factories under serum-free conditions and processed using a multistep protocol. A representative production with the cellculture-infectious HCV recombinant HCV5aHi 34 (Fig. S1) is shown in Fig. 1A-E. Following UV inactivation, the HCV antigen was formulated with adjuvant and used for 4 subcutaneous immunizations of BALB/c mice at 3-week intervals (Fig. 1F).
Immunization of mice with an inactivated whole virus HCV vaccine formulated with Alum+MPLA elicited nAbs with limited efficacy First, mice were immunized with inactivated HCV5aHi formulated with Alum+MPLA, licensed for human use. Following sacrifice, serum IgG was purified and tested for nAbs against homologous HCV5aHi, showing dose-dependent neutralization, however, with limited efficacy. The half maximal inhibitory concentration (IC 50 ) ranged from 64 to 589 lg/ml (mean 313 lg/ ml) for 4 animals, while for 2 animals 50% neutralization was not observed at 1,000 lg/ml, the highest IgG concentration used ( Fig. 2A). Near complete neutralization was only observed for 1 animal, while 18-72% was achieved for the other animals at the highest IgG concentration. No HCV-specific nAbs were detected in control mice immunized with ovalbumin and Alum+MPLA (Fig. S2). Vaccine-induced IgG specifically bound to homologous E1/E2 complexes and less strongly to sE2 (Fig. S3). Immunization with HCV5aHi formulated with the widely used experimental Freund's adjuvant induced slightly better nAb responses, with mean IC 50 at 286 lg/ml IgG and mean maximum neutralization of 83% against HCV5aHi (Fig. S4).
Immunization of mice with an inactivated whole virus HCV vaccine formulated with AddaVax elicited potent crossgenotype nAb responses Next, we immunized mice with inactivated HCV5aHi formulated with AddaVax, an analogue of MF-59; MF-59 is licensed for human use. IgG purified from 9 vaccinated animals neutralized homologous HCV5aHi at IC 50 of 15-66 lg/ml (mean 36 lg/ml) and near complete neutralization at the highest IgG concentration (Fig. 2B). Thus, AddaVax showed superior efficacy, with mean IC 50 more than 15-fold lower than that of the Alum+MPLA vaccination group (Fig. 2C). Further, mean maximum neutralization at the highest IgG concentration was 97% vs. 61% for IgG induced by AddaVax vs. Alum+MPLA (Fig. 2D). Importantly, IgG pooled from these 9 animals efficiently cross-neutralized HCV genotypes 1-6 at IC 50 of 14-192 lg/ml, with near complete neutralization at the highest IgG concentration (Fig. 2E). No HCVspecific nAbs were detected in control mice (Fig. S2). Further, IgG purified from pooled serum of these 9 animals after the 2 nd and 3 rd immunization neutralized HCV5aHi at IC 50 of 50 and 76 lg/ ml, respectively, and with near complete neutralization at the highest IgG concentration (Fig. 2F). Moreover, vaccine-induced IgG specifically bound to homologous E1/E2 complexes and sE2 individually (Fig. 3), showing stronger binding than IgG induced by Alum+MPLA (Fig. S3). IgG induced following 3 and 4 immunizations showed similar, concentration-dependent binding.
Slightly lower binding was observed for IgG derived following 2 immunizations (Fig. 3).
HCV with modified envelope proteins showed differential sensitivity to nAbs Deletion of HVR1 or mutation of glycosylation sites has been reported to influence the neutralization sensitivity of cellculture-derived HCV, presumably by influencing exposure of neutralization epitopes. 4,[11][12][13][14] However, how such modifications influence immunogenicity in the context of a whole virus vaccine has not been studied. HCV5aHi harbored the E2 substitution T385P, which has been shown to increase neutralization sensitivity. 34 To investigate the influence of T385P on immunogenicity, we produced HCV5aHi(T385) with the substitution reverted to T385, 34 yielding 5.6 log 10 focus-forming units (FFUs)/ ml in pooled supernatant from cell factories. Further, to study the influence of HVR1 on immunogenicity we developed HCV5a-Hi(DHVR1) by deleting HVR1 from HCV5aHi and by further culture adaptation, carrying out 18 passages in Huh7.5 cells to compensate for fitness impairment. The resulting polyclonal virus had acquired 3 additional dominant substitutions based on Sanger sequencing: N532D and L735I in E2 as well as K1609R in non-structural protein 3 (amino acid positions are related to the polyprotein of the 1a H77 reference sequence (GenBank accession no. AF009606)) (Fig. S1). Pooled passage 20 supernatant from cell factories inoculated with a passage 19 seed stock yielded 5.8 log 10 FFUs/ml.

HCV vaccine candidates with modified envelope proteins induced similar nAb responses in mice
To evaluate if differences in neutralization sensitivity affected immunogenicity, mice were immunized with HCV5aHi, HCV5a-Hi(DHVR1) and HCV5aHi(T385) equivalent to 7.5 log 10 FFUs using AddaVax. First, we proved that serum IgG purified from each animal neutralized HCV5aHi with similar efficacy (Fig. 2B and  S6). Then, for each group, IgG pools were generated and used to neutralize HCV5aHi, HCV5aHi(DHVR1), HCV5aHi(T385) and the HCV5a reference virus (Fig. 5A-D). In these assays, IC 50 ranged from 17 to 654 lg/ml. The HCV5aHi IgG pool showed somewhat higher neutralization efficacy than the other pools with IC 50 of 17, 267, 126 and 69 lg/ml against HCV5aHi, HCV5aHi(T385), HCV5aHi(DHVR1) and HCV5a, respectively. Further, the IgG pools neutralized HCV1a at IC 50 of 14-143 lg/ml (Fig. 5E) and HCV3a at IC 50 of 192-295 lg/ml (Fig. 5F), with HCV5aHi-induced IgG showing the highest efficacy. Finally, IgG pools showed similar concentration-dependent binding to E1/E2 complexes and sE2 (Fig. 6). Overall, while modifications of the envelope proteins did not have major effects on immunogenicity, we observed a trend

Discussion
We provide proof-of-concept for immunogenicity of a whole virus inactivated HCV vaccine employing an adjuvant analogue of MF-59, which is licensed for human use. Immunizations resulted in induction of potent antibodies broadly neutralizing all major HCV genotypes with recognized epidemiological importance. HCV envelope protein variants with deletion of HVR1 and mutation of a putative O-linked glycosylation site showed differential neutralization sensitivity but overall similar immunogenicity.
Broadly nAbs are associated with protection from chronic HCV infection. 16,20,23,48 Our results suggest that induction of broadly nAbs by a single virus antigen is possible. This is in line with proof-of-concept findings in chronically HCV-infected patients and in humans immunized with the gpE1/gpE2 vaccine, where cross-nAb targeting conformational epitopes were elicited by single isolates. 7,28,49 HCV5aHi vaccine-induced nAbs exhibited slight differences in efficacy against different HCV isolates, overall reflecting differences previously observed using other nAbs, such as the relatively low efficacy against the HCV3a virus. 4,8,9,24,49,50 These observations are in line with the hypothesis that conformational epitopes targeted by potent crossgenotype neutralizing mAbs, such as AR3A and AR4A, are conserved among HCV isolates, and that different neutralization sensitivity is mainly caused by isolate-specific epitope protection by HVR1-and glycan-dependent closed envelope protein conformational states. 4,7,12,51 The highest IgG concentrations used in in vitro neutralization assays, yielding close to complete neutralization, were 10-fold lower than mean IgG concentrations in human serum. IC 50 were comparable to those achieved in previous immunogenicity studies using whole inactivated genotype 2a HCV, however, formulated with non-licensed adjuvants; in these studies cross-neutralization of genotype 1a/b and 3a was observed, while genotypes 4-6 were not tested. 31,33 Moreover, IC 50 were in the range of IC 50 of IgG in chimpanzees protected from HCV challenge following vaccination with the gpE1/gpE2 vaccine 24,25 and at least comparable to that of IgG induced by vaccines based on sE2 in small animal models. 14,[52][53][54] Future development of an inactivated HCV vaccine candidate will be facilitated by development of optimized serumfree bioreactor-based upstream and downstream processes. 38,40,55 Future immunogenicity studies should aim at defining an optimal immunization schedule and HCV antigen dose. For production of whole virus vaccines, amounts of viral particles required to achieve a given immune response are of interest. In this study, antigen doses were defined by infectious unit equivalents, as infectivity titrations were used to monitor viral processing, providing evidence for the presence of intact particles, and as reported by others. 56,57 This might not allow for an optimal comparison of amounts of particles for viruses showing different specific infectivities. Therefore, we retrospectively determined amounts of HCV core and genome copies in vaccine preparations. This analysis suggested that 3-to 6fold higher amounts of viral particles might have been contained in the HCV5aHi(DHVR1) vaccine compared to the HCV5aHi and HCV5aHi(T385) vaccines (Table S1). This could explain the slightly higher E1/E2 binding by HCV5aHi(DHVR1)induced IgG compared to IgG induced by the other 2 viruses. Based on core determinations, vaccine doses used in this study were comparable to doses used in previous inactivated HCV vaccine studies. 31,33 A future research focus should be establishment of assays to quantify the amount of HCV envelope proteins, being the main antigenic proteins, in vaccine preparations.
An important achievement of our study is the development of a whole inactivated HCV vaccine candidate employing an analogue of a licensed adjuvant, facilitating its use in humans. In line with previous findings, MF-59 analogue AddaVax, also used in the gpE1/gpE2 vaccine, 29 was superior in inducing binding Abs and nAbs compared to Alum used in most human vaccines and the experimental golden standard Freunds adjuvant. 56,58 Further, whole inactivated genotype 2a HCV formulated with Alum yielded nAbs showing <20% neutralization. 33 Nevertheless, comparatively high neutralization sensitivity of our vaccine antigens might have contributed to comparatively high immunogenicity. Thus, compared to the genotype 2a (isolate J6) HCV, 31,33 HCV5aHi showed at least 200-fold higher sensitivity to the human antibodies used for characterization of HCV neutralization sensitivity (unpublished results). 12 For other viruses, increased neutralization epitope exposure resulted in enhanced induction of nAbs following vaccination. [59][60][61] According to assays applied in this study, increased HCV neutralization epitope exposure mediated by deletion of HVR1 did not result in increased immunogenicity, while modification of an O-linked glycosylation site in HCV5aHi resulted in somewhat increased induction of nAbs. While this question had not been addressed using whole HCV vaccines that would be expected to show close to native envelope conformations, similar findings were reported using envelope protein subunit vaccines. Thus, gpE1/gpE2 heterodimers 50 or sE2 14 without HVR1, with or without an additional modification of a glycosylation site, enzymatically deglycosylated sE2, 14 or insect cell produced sE2 52,54 showed no or slightly increased immunogenicity. Somewhat increased immunogenicity was observed for sE2 lacking all 3 variable regions. 53 Future vaccine studies with HCV showing greater differences in neutralization sensitivity than the viruses used in this study might further clarify the potential of envelope protein engineering to increase immunogenicity. Compared to the reference virus HCV5a, both HCV5aHi and HCV5aHi(DHVR1) showed up to 20-fold increased neutralization sensitivity, IgG (μg/ml) Fig. 3. Immunization of BALB/c mice with HCV and AddaVax induced IgG binding to HCV5a E1/E2 complexes and sE2. Binding to HCV5a (A) E1/E2 complexes and (B) sE2 in ELISA was tested for pooled purified serum IgG from 9 mice immunized 4 times with HCV5aHi (Fig. 2B) and 3 mice immunized with OVA (Fig. S2) using AddaVax. Furthermore, IgG purified from pooled sera from 9 mice following 2 and 3 IMM was tested (Fig. 2F)    In conclusion, using an analogue of the licensed adjuvant MF-59 and whole inactivated cell-culture-derived HCV, we have developed an attractive HCV vaccine approach for further pre-clinical development. While a recently tested immunogenic T cell-based vaccine did not protect against chronic HCV infection in a phase I/II study, 27  HCV5aHi(ΔHVR1) IgG (log 10 μg/ml) IgG (log 10 μg/ml) IgG (log 10 μg/ml) IgG (log 10 μg/ml) IgG (log 10 μg/ml) IgG (log 10 μg/ml) IC   inducing potent nAbs. In the future, following optimization of the vaccine preparation processes, clinical studies will be needed to elucidate if this vaccine approach will confer protection against chronic HCV infection.   Fig. 6. Immunization with HCV5aHi variants resulted in IgG with similar binding to HCV5a E1/E2 complexes and sE2. Binding to HCV5a (A) E1/E2 complexes and (B) sE2 in ELISA was tested for pooled purified serum IgG from mice immunized with HCV5aHi (Fig. 2B), HCV5aHi(T385) (Fig. 5), HCV5aHi(DHVR1) (Fig. 5), or OVA (Fig. S2) using AddaVax. Data from HCV5aHi and OVA immunizations are reproduced from Fig. 3 for comparison. Datapoints are means of duplicates with SEM. E1/E2, HCV envelope glycoprotein heterodimer; OD, optical density; OVA, ovalbumin; sE2, soluble HCV envelope glycoprotein E2.