Protective efficacy of Zika vaccine in AG129 mouse model

Zika virus (ZIKV) is a mosquito-borne flavivirus that causes asymptomatic infection or presents only mild symptoms in majority of those infected. However, vaccination for ZIKV is a public health priority due to serious congenital and neuropathological abnormalities observed as a sequelae of the virus infection in the recent epidemics. We have developed an inactivated virus vaccine with the African MR 766 strain. Here we show that two doses of the vaccine provided 100% efficacy against mortality and disease following challenge with homotypic MR 766 and the heterotypic FSS 13025 ZIKV strains in the Type I and Type II interferon deficient AG129 mice. Two doses of the vaccine elicited high titer of neutralizing antibodies in Balb/c mice, and the vaccine antisera conferred protection against virus challenge in passively immunized mice. The studies were useful to rationalize vaccine doses for protective efficacy. Furthermore, the vaccine antisera neutralized the homotypic and heterotypic ZIKV strains in vitro with equivalent efficiency. Our study suggests a single ZIKV serotype, and that the development of an effective vaccine may not be limited by the choice of virus strain.


Results
Vaccine efficacy in AG129 mice. Two groups of 4-6 week old female AG129 mice (n = 8/group) were vaccinated with 10 μ g per dose of the alum adsorbed purified inactivated vaccine on days 0 and 21 by intramuscular route, and challenged subcutaneously with 10 4 PFU (plaque forming unit) of FSS 13025 or MR 766 on day 28 as schematically outlined in Supplementary Fig. S1a,b. Two groups of control mice (n = 8/group) received equivalent concentration of alum (placebo) and were challenged with 10 4 PFU of FSS 13025 or MR 766. All the four groups were monitored daily for 20 days post infection for clinical health, mortality, body weight and temperature. Vaccinated mice were fully protected (100% survival) against infection with either of the virus strains. Death in the control groups challenged with MR 766 occurred earlier with mean time to death (MTD) of 8 days as compared to the group challenged with FSS 13025 with MTD of 12 days (Fig. 1a). Mean health scores for clinical disease based on appearance, mobility and alertness were calculated based on observations described in Table 1. The vaccinated group had a perfect health score of 1 throughout the observation period and no disease symptoms were observed, whereas the control animals exhibited progressive morbidity before succumbing to infection (Fig. 1b). Clinical score was supported by the lack of weight loss in the vaccinated animals ( Supplementary Fig. S2a,b). Serum viremia was estimated on days 4 and 6 post infection in 3 animals from each group. Vaccinated mice exhibited undetectable viral load, whereas viremia peaked on day 4 in the control animals regardless of the challenge virus (Fig. 1c). Disease progression was more rapid in MR 766 group when compared to the FSS 13025 group. Body temperatures across all the groups did not vary greatly with the exception of the control MR 766 group which had a large temperature drop just prior to death ( Supplementary Fig. S3a,b). The vaccine protected against mortality and disease caused by ZIKV strains of the Asian and African genotypes in the AG129 mice.

Immunogenicity.
To study the immune responses induced by the alum adsorbed purified inactivated ZIKV vaccine, we immunized Balb/c mice (n = 8/group) with two doses of 5 μ g or 10 μ g of the vaccine, and the control group with alum (placebo) by intramuscular route on day 0 and on day 21 as schematically outlined in Supplementary Fig. S1a,b. Vaccinated mice developed ZIKV neutralizing antibodies on day 14 after prime dose, and on day 28, one week after boost immunization. Mean log neutralizing antibody titers by PRNT 50 were 2.14 after prime, and 3.38 after boost dose in the 5 μ g group (P < 0.0001), and 2.41 and 3.74 in the 10 μ g group (P < 0.0001) (Fig. 2a). The control animals did not develop detectable antibody responses and hence the results are not shown. The corresponding mean log antibody titers by ELISA using purified inactivated ZIKV as the coating antigen were 3.35 and 4.26 (P < 0.0001) in the 5 μ g group, and 3.50 and 4.41 (P < 0.0001) in the 10 μ g group (Fig. 2b). Protective efficacy of the vaccine was assessed in the vaccinated and in control groups by intravenous injection of 10 5 PFU of MR 766 and the virus titers in plasma were estimated every 24 hours up to 144 hours post-infection. Viremia was undetectable in the 5 μ g and 10 μ g vaccine groups up to 144 hours (Fig. 2c,d), but peaked at 72-96 hours post-infection in the control animals (Fig. 2e). The vaccine protected against virus replication as infectious virus particles could not be detected in plasma by plaque assay even after three serial amplifications in vitro in Vero cells. Vaccinated animals showed good anamnestic response to virus challenge with saturating mean log PRNT 50 titers of 4.30 and 4.26 in the 5 μ g and 10 μ g dose groups respectively (Fig. 2f).
Passive immunization. We tested the protective efficacy of the vaccine antisera to virus challenge by passive immunization in Balb/c mice. Rabbit vaccine antisera (log PRNT 50 titer of 4.28) at either 1:1, 1:2, 1:4 or 1:8 dilutions (groups I-IV) was administered in a volume of 0.3 ml by i.p. route in Balb/c mice (n = 5/group). The animals were challenged 6 hours later with10 5.5 PFU of MR 766 by i.v. The control animals received pre-immune serum by i.p. and equivalent virus dose by i.v. as the passively immunized animals. The serum mean log PRNT 50 titers 24 hours after antisera administration in the recipient mice groups I to IV were 3.07, 3.02, 2.52, and 2.11 respectively (Fig. 3a). No infectious particles could be detected in animals from 24 to 144 hours after passive immunization (Fig. 3b), whereas viremia peaked at 72-96 hours in control animals that received pre-immune serum (Fig. 3c). The lack of detectable virus particles in the passively immunized mice was further confirmed by plaque assay after three serial amplifications in vitro in Vero cells.
Vaccine cross-neutralization. While MR 766 vaccine protected against the homotypic and heterotypic virus challenge in AG129 mice, we tested if the vaccine sera cross neutralized the African and Asian strains with equivalent efficiency. The vaccine sera cross neutralized MR 766 and FSS 13025 with mean log PRNT 50 titers of 4.25 and 4.26 respectively ( Supplementary Fig. S4a,b), that suggests a single serotype of ZIKV.

Discussion
The envelope E protein is the target of ZIKV neutralizing antibodies 25,26 . The Asian lineage of the virus has maintained an amino acid identity of 99.4 to 100% in the E protein ( Supplementary Fig. S5) suggesting that immune pressure has not been a dominant factor in virus evolution and fitness, and that any Asian ZIKV strain can confer protective immunity against other strains of this lineage 27 . The ZIKV Asian strain FSS 13025 shares 99.2 to 99.8% amino acid identity in the E protein with the contemporary strains of the same lineage, while MR 766 shares 97.03 to 97.62% amino acid identity with the Asian strains in the same region ( Supplementary Fig. S5). ZIKV convalescent sera from the recent epidemics neutralized H/PF/2013 isolated from French Polynesia in 2013, Paraiba/2015 strain from Brazil, and the MR 766 with similar efficiency 28 , and Rhesus macaques infected with MR 766 were protected when challenged with the heterologous H/PF/2013 strain 29 . While the studies with convalescent sera identified a single serotype, we show that the ZIKV vaccine derived from the African strain induced cross-lineage protective efficacy against a strain of the Asian genotype in an appropriate animal model, and the vaccine antisera neutralized the homotypic and heterotypic strains in vitro with equivalent efficiency. Our studies also showed rapid progression in disease pathogenesis with MR 766 as compared to equivalent challenge with FSS 13025 in AG129 mice. Vaccines derived from single antigenic serotype have also been effective against other flaviviruses such as Yellow fever virus 30 , Tick-borne encephalitis 31,32 , and Japanese encephalitis 33,34 viruses that exist as multiple genotypes.
Any ZIKV MR 766 sequence data needs to be interpreted with caution as multiple MR 766 sequences in GenBank show deletion of the signature 'VNDT' glycosylation site in the E protein. Sequence heterogeneity  Table 1. Error bars represent s.d. (c) Viremia was estimated on days 4 and 6 after virus challenge in three animals from each group, and expressed as PFU/ml. Red bars reflect the mean of three values. The results are shown from a single experiment. could perhaps be due to passage history in different labs subsequent to the extensive passage of the original MR 766 strain in mouse brain 35 . Here we show that the MR 766 strain from ATCC which is the original repository of the virus has the conserved glycosylation motif and identical number of amino acids as the contemporary Asian strains. The twice plaque purified virus strain sequenced in this study, matches the sequence available in GenBank under the accession numbers KX377335, HQ234498 and KU720415, and henceforth should represent the correct sequence of MR 766.
The candidate MR 766 vaccine conferred protective efficacy comparable to vaccines developed using the ZIKV strains of the Asian genotype. The log median neutralizing antibody titers of 3.66 induced by two 5 μ g doses of the purified inactivated PRVABC59 ZIKV strain in rhesus macaques 15 is comparable to log mean neutralizing antibody titers of 3.38 induced by two equivalent doses of the inactivated MR 766 vaccine in Balb/c mice in this study. The values are also comparable to log antibody titers of ~3.0 elicited by two 50 μ g doses of prME DNA vaccine in Balb/c mice 17 , and to two doses of H/PF/2013 strain prME DNA vaccine in non-human primates 16 . A single dose of 10 11 virus particles of rhesus adenovirus 52 vectored vaccine encoding prME also induced comparable neutralizing antibody titers in rhesus macaques 15 . In all the studies, the vaccine-induced antibodies conferred sterilizing immunity against virus challenge either in mice and/or non-human primates. Passive antibody transfer studies suggests a threshold of ~2.0 (log neutralizing antibody titers) for conferring protection against virus infection in these animal models 15,16 . The values are comparable to log mean PRNT 50 titer of ~2.11 that conferred complete protection against virus challenge reported in this study.
The ZIKV vaccine induced neutralizing antibody titers are much higher than the threshold of 1:10 PRNT 50 titer shown to be protective for other flavivirus vaccines 36 . A single boost dose significantly enhanced the neutralizing and binding antibody titers. To further rationalize the protective doses, we found that there was no significant difference between the 5 μ g and 10 μ g of vaccine in prime and prime-boost immunization. Also, the neutralizing antibody titers induced by a single 5 μ g dose was comparable to the log PRNT 50 titer of ~2.11 that conferred passive protection against virus replication. Protection at lower antibody levels and duration of protective immunity conferred by passive antibody transfer were not studied. Vaccination conferred complete protection against viremia and disease up to 14 and 20 days after virus challenge which is 35 and 48 days after prime dose vaccination in Balb/c and AG129 mice respectively. A strong anamnestic response to virus challenge with high increase in mean neutralizing antibody titers indicates that vaccine could potentially confer longer duration of protective immunity when evaluated at similar dose levels and by the same route in clinical trials.
Safety evaluation of the ZIKV vaccine in clinical trials have to factor in the potential for enhancement of ZIKV infection by Dengue antibodies that has been demonstrated in vitro 37,38 . Also definitive diagnosis of ZIKV infections will be difficult in the background of pre-existing Dengue and other flavivirus antibodies that are known to be broadly cross-reactive 39,40 . Such cross reactive antibodies could potentially impair the assessment of vaccine efficacy particularly in the absence of knowledge whether heterologous flavivirus antibodies can quench or potentiate immune response to the vaccine. These are major challenges to be addressed during further pre-clinical and clinical development of the vaccine.
Inactivated vaccines are easily administered, have good safety profile and might be preferable for vaccinating women of child bearing age in case of an emergency during an epidemic outbreak. Our study has demonstrated that the choice of ZIKV strain may not be a limiting factor in vaccine development. This study is a pioneering effort in ZIKV vaccine space that was initiated in late 2014, and was limited by the choice of virus strain at that period. In the absence of strain specific differences in neutralizing potential, the purified MR 766 virus strain with its ability to grow to high titers in Vero cells might pave the way for development of an effective vaccine. Availability of vaccine is important as an emergency response in regions with impending threat of ZIKV epidemics, and in the long term to restrict local transmission in regions endemic for the virus.     International requirements. Animals were randomly allocated to groups. The study tested the efficacy of two doses of alum adsorbed 10 μ g vaccine in 4-6 week old female AG129 mice (n = 8/group) against challenge with ZIKV. The groups 1 and 3 were vaccinated on day 0 and day 21, and groups 2 and 4 received equivalent volume of placebo (alum only) by intramuscular route (100 μ l total, 50 μ l in each hind leg). Prior to virus challenge, all the mice were microchipped for daily temperature monitoring. The mice in groups 1 and 2 were challenged with 10 4 PFU of FSS 13025 via subcutaneous route on day 28. The mice in groups 3 and 4 were challenged with 10 4 PFU of MR766 subcutaneously on day 28. Three mice from each group were test bled for serum by retro-orbital route on days 4 and 6 post infection, and were assessed for viremia by plaque assay. All the groups were monitored for weight loss, clinical score, mortality, and body temperature daily for 20 days post-challenge or until time of sacrifice. Mice displaying severe illness as determined by > 20% weight loss, a health score of 5 or above (see Table 1 to 144 hours post-infection, and expressed as PFU/ml. Further, plasma samples were serially amplified three times for 5-7 days in vitro in Vero cells for detection of infectious particles if any, by plaque assay. Virus titer. Plaque assays were used for virus titrations and expressed as plaque forming units (PFU/ml).
Briefly, confluent Vero cells in 6-well plates were inoculated with log dilutions of the virus for 90 min in serum free MEM, followed by addition of 0.85% methyl cellulose overlay. After incubation for 4 days, cells were fixed with 10% buffered formalin, and stained with 0.1% crystal violet. Each sample was assayed in triplicates, and virus titers were calculated by multiplying the enumerated plaque count by virus dilution and volume used for infection, and expressed as PFU/ml. PRNT 50 . Neutralizing antibody titers were estimated by 50% Plaque Reduction Neutralization Test (PRNT 50 ).
Prior to the assay, 6-well tissue culture plates were seeded with 2.5 × 10 4 Vero cells per well, and incubated at 37 °C until confluent. Four fold sera dilutions in MEM with equal volume of standardized titer of MR 766 were incubated for 90 min and added to Vero cells in the 6-well plates. After incubation for 90 min, 0.85% methyl cellulose overlay was added and incubated for 4 days. The cells were fixed with 10% formalin and stained with 0.1% crystal violet. Each sample was assayed in triplicates. Plaques were enumerated, and the estimated serum dilution causing 50% reduction in the plaques formed by the control virus sample without antibody was estimated as the PRNT 50 titer and expressed in log values. The mean of three replicate values for each animal per time point was used in the calculation of neutralizing antibody titers by PRNT 50 . Samples with titer of ≥10 were considered as seropositive. Cross neutralization of MR 766 and heterotypic FSS 13025 strains by vaccine antisera was outsourced to IBT Bioservices, Gaithersburg, MD, USA. PRNT 50 titers were calculated using a 4PL curve fit.

ELISA.
Binding antibody titers were estimated by ELISA. Briefly, purified inactivated ZIKV virus was coated at the standardized concentration in 96-well Nunc ™ MaxiSorp ™ plates overnight in 50 mM carbonate-bicarbonate buffer, pH 9.6 and blocked with 0.5% skim milk powder in coating buffer. Two fold serial dilutions of the vaccine antisera was added in triplicate wells per dilution, incubated at 37 °C for 90 min and washed with PBST (phosphate buffered saline with 0.05% tween-20) and PBS three times each. Anti-mouse-IgG HRPO secondary antibody (Sigma-Aldrich, St. Louis, USA) at 1:2500 dilution was added and incubated for 60 min, and washed three times each with PBST and PBS. Freshly prepared O-phenylenediamine dihydrochloride (Sigma Aldrich, St. Louis, USA) and hydrogen peroxide were added and incubated for 10 min. The reaction was stopped with 2 M sulfuric acid and absorbance was read at 490 nm. The mean of three replicate values was used to calculate the titer. The cut-off (endpoint titer) for seroconversion was the pre-exposure titer + 3x standard deviation. The reciprocal of penultimate serum dilution above the cut-off was taken as antibody titer and expressed in log values.
Statistical Analysis. The antibody titers were log transformed for statistical analysis using GraphPad Prism v5 (GraphPad Software, CA, USA). The values are expressed as mean with 95% c.i. Parametric paired t-tests with two-tailed limits were used to calculate differences between time points within a dose group (n = 8). Parametric unpaired t-test using confidence level of 95% and two-tailed limits was used to calculate variance between the groups. The values were considered statistically significant if the P < 0.05. The normality and homogeneity of variance assumption was checked using tests of assumptions in NCSS software version 9.0.14 prior to performing all the parametric tests.