Efficacy of a T Cell-Biased Adenovirus Vector as a Zika Virus Vaccine

Zika virus (ZIKV) is a major public health concern due to the risk of congenital Zika syndrome in developing fetuses and Guillain-Barre syndrome in adults. Currently, there are no approved vaccines available to protect against infection. Adenoviruses are safe and highly immunogenic vaccine vectors capable of inducing lasting humoral and cellular immune responses. Here, we developed two Adenovirus (Ad) vectored Zika virus vaccines by inserting a ZIKV prM-E gene expression cassette into human Ad types 4 (Ad4-prM-E) and 5 (Ad5-prM-E). Immune correlates indicate that Ad5-prM-E vaccination induces both an anti-ZIKV antibody and T-cell responses whereas Ad4-prM-E vaccination only induces a T-cell response. In a highly lethal challenge in an interferon α/β receptor knockout mice, 80% of Ad5 vaccinated animals and 33% of Ad4 vaccinated animals survived a lethal ZIKV challenge, whereas no animals in the sham vaccinated group survived. In an infection model utilizing immunocompetent C57BL/6 mice that were immunized and then treated with a blocking anti-IFNAR-1 antibody immediately before ZIKV challenge, 100% of Ad4-prM-E and Ad5-prM-E vaccinated mice survived. This indicates that Ad4-prM-E vaccination is protective without the development of detectable anti-ZIKV antibodies. The protection seen in these highly lethal mouse models demonstrate the efficacy of Ad vectored vaccines for use against ZIKV.

. Ad4 has been shown to be effective as a viral vector in vaccination against influenza, either alone or in combination with Ad5 35,36 .
In this study, we developed both human Ad5 and Ad4 expressing the ZIKV full-length prM-E genes as a candidate ZIKV vaccine. Ad5-prM-E elicited both a strong antibody and T-cell response, whereas Ad4-prM-E did not induce any detectable anti-ZIKV antibodies but still induced a strong T-cell response. In a lethal ZIKV challenge model using an Ifnar −/− mice, Ad5-prM-E provided superior protection to Ad4-prM-E vaccination. However, both Ad vectors protected 100% of mice in a challenge model using anti-Ifnar1 blocking antibody. These data support the use of Ad vectors as a platform for ZIKV vaccine development.

Results
Construction of replication-defective Adenoviral-vectored vaccines. The full length prM-E genes of Zika virus (ZIKV) strain PRVABC59 (Puerto Rico, 2015) were cloned into the E1 region of Ad type 5 (Ad5-prM-E) or Ad type 4 (Ad4-prM-E) to create a replication-defective vector (Fig. 1A). The Ad5-prM-E vector is also E3 deleted to increase cloning capacity. In addition, the Zeocin resistance gene flanked by Frt regions (which was used for effective screening of recombinant clone) was left in the Ad4 vector for simplicity. However, this gene can be easily removed using FLP recombinase for future studies. These differences should not affect in vivo immunogenicity as both of these vectors are replication defective. ZIKV-E protein expression was confirmed in both Ad vaccines by western blotting (Fig. 1B). 293 cells were infected at on MOI of 1 for 48 hours and cell lysate was collected. No significant difference in ZIKV-E protein expression was detected. Data was collected from three separate biological replicates (Fig. 1C).
Antibody response to vaccination. To examine the humoral immunity induced by vaccination, groups of C57BL/6 mice were immunized with a single dose of 10 10 virus particles (vp) of Ad vaccine (Ad5-prM-E or Ad4-prM-E) by the intramuscular route (i.m.) on day 0 ( Fig. 2A,B). PBS was used as a vaccination negative control. All vaccinated mice were sacrificed on week 8 for immune response analysis. Mice that received single priming dose of Ad5-prM-E showed significantly higher levels of anti-ZIKV E protein antibodies as measured by ELISA (Fig. 2C) compared to the control PBS-treated mice. However, Ad4-prM-E vaccinated mice did not show detectable levels of specific anti-ZIKV-E protein antibodies or neutralizing antibodies (Fig. 2C,D).
To investigate if an Ad prime/boost vaccination strategy would increase immune responses, C57BL/6 mice were vaccinated at day 0 with 10 10 vp of Ad vaccine, boosted with 10 10 vp of the homologous or heterologous Ad vaccine on week 6, and sacrificed on week 8 ( Fig. 2A,B). Prime/boost strategies utilizing Ad5-prM-E as the prime showed significantly higher levels of neutralizing antibodies than control mice (Fig. 2C,D). In contrast, the Ad4-prM-E primed mice showed no boosting effect at all.

Ad4-prM-E vaccination does not induce detectable antibodies.
To ensure that the lack of antibody response during Ad4-prM-E vaccination was not due to infectious differences in the Ad vaccine virus stocks, C57BL/6 mice were vaccinated based on infectious units (IFU). The IFU per mL was calculated for both Ad5-prM-E and Ad4-prM-E vaccine stocks. An analysis of virus particle (vp) to IFU showed a 2.3-fold difference between Ad5-prM-E and Ad4-prM-E (Fig. 2E). Mice were then vaccinated with 10 10 vp of Ad5-prM-E or 2.3 × 10 10 vp of Ad4-prM-E, resulting in a vaccination dose with equal IFU. In order to evaluate the response to vaccination near the predicated peak immune response, mice were euthanized 2 weeks later for immune correlate prM-E region with a VSV G signal peptide under the control of a CMV promoter was cloned into the Ad5 and Ad4 genomes in the E1 region. Ad5 was also deleted for the E3 region. (B) A representative western blot comparing the levels of ZIKV-E protein expression from Ad4-prM-E and Ad5-prM-E infected 293 cells infected at an MOI of 1. GAPDH is used as a cellular control. (C) The ratio of ZIKV-E protein to GAPDH of western blots from C. The western blot data represents three separate biological replicates (n.s. not significant, two-tailed t test). Data are expressed as the mean with standard error (SEM). analysis. Again, Ad4-prM-E vaccination did not result in any detectable anti-ZIKV antibodies as measured by ELISA (Fig. 2F) or plaque reduction neutralization test (Fig. 2G). This data indicates that the lack of antibody development in Ad4-prM-E vaccinated mice is not due to differences in vaccine stock infectivity.

T-cell responses after vaccination.
The T-cell response against the tested vaccine strategies and ZIKV infection was examined using an IFN-γ enzyme-linked immunospot assay (ELISPOT). Splenocytes were isolated and incubated with 15-mer peptides with 13 amino acid overlap spanning the entire E protein of ZIKV strain PRVABC59. Of the 164 tested peptides, 16 peptides were positive for a response ( Fig. 3; Table. S1). A response was considered positive if it was 4 times the response of the PBS sham vaccinated mice (Fig. 3A). Both Ad5-prM-E and Ad4-prM-E vaccination induced a significant T-cell response against ZIKV E protein peptides (Fig. 3B). Peptide 1 and 2 induced the strongest T-cell response in all Adenovirus vaccinated animals ( Fig. 3C) and encompassed the previously identified immunodominant CD8 + T-cell epitope E 4-12 (position number E 294-302 on the polypeptide) 37,38 . A strong T-cell response was generated against this epitope after a single shot of either Ad4-prM-E or Ad5-prM-E and was similar in intensity to the T-cell response seen after being boosted with either the homologous or heterologous Ad vaccine. Therefore, the prime/boost strategy and Adenovirus serotype-switching strategy did not significantly increase the T cell responses as judged by IFN-γ ELISPOT.

Vaccine protection to lethal ZIKV infection in mice. Initially, protection was examined in
Ifnar1 −/− mice on a C57BL/6 background, which are vulnerable to ZIKV infection 39 . Since no significant differences were seen in immune correlates between vaccination based on virus particles or infectious units (Fig. 2), challenge studies were carried out with 10 10 vp per mouse. Ifnar1 −/− mice were primed via i.m. injection at day 0 with 10 10 vp Ad4-prM-E or Ad5-prM-E, or PBS as a sham control. Animals were boosted with the same dose of vaccine at week 3 and challenged at week 7 with 10 6 FFU of mouse-adapted ZIKV strain 41525 40 (Fig. 4A). A comparison of the genetic differences between the PRVABC59 vaccination strain and the Dakar 41525 challenge strain can be seen in Supplementary Fig. 1. Studies have identified a single ZIKV serotype indicating that one . ZIKV E protein specific antibodies were measured using an ELISA in (C) and neutralization titer (PRNT50) was determined by plaque reduction assay in (D) ( * p < 0.05; ** p < 0.01, *** p < 0.001, **** p < 0.0001; one-way ANOVA with Bonferroni multiple comparisons). To ensure that the undetectable antibody response to Ad4-prM-E vaccination was not sue to differences in infectivity of Ad vaccine virus stocks, the virus particle (vp) to infectious unit (IFU) ratio was determined (E). C57BL/6 mice (n = 5) were then vaccinated with an equal IFU dose and the antibody response measured via ELISA (F) and plaque reduction assay (G) ** p < 0.01; one-way ANOVA with Bonferroni multiple comparisons). Data are expressed as the mean with standard error (SEM). ZIKV vaccine antigen should protect against all ZIKV strains 13,41 . After challenge, mice were monitored and euthanized when a weight loss of 25% or more was detected. Ad5-prM-E primed and boosted mice showed an 80% survival rate and Ad4-prM-E primed/boost mice had a 33% survival rate whereas all of the sham vaccination animals succumbed by day 7 (Fig. 4B,C; Fig. S2A). ZIKV viremia in mice peaks between day 3-4 post 42,43 , so animals were bled 4 days after infection to determine viremia using RT-qPCR (Fig. 4D). Ad5-prM-E animals showed a highly significant decrease in viremia (p < 0.001). Ad4-prM-E vaccination also decreased viremia compared to those in sham vaccinated mice. In addition, a plaque assay was performed to determine infectious ZIKV levels in sera. Both Ad4-prM-E and Ad5-prM-E vaccination significantly reduced infectious plaque forming  4G) confirmed a vaccine-induced immune response that was similar to wild type C57BL/6 mice. Due to the immunocompromised nature of the Ifnar1 −/− mice, we tested a second challenge model that utilized wild type C57BL/6 mice and a blocking anti-Ifnar1 antibody (MAR1-5A3) administered after vaccination but before challenge, which makes naïve mice susceptible to lethal ZIKV infection 43 . Groups of C57BL/6 mice were immunized via the i.m. route with 10 10 vp of Ad4-prM-E or Ad5-prM-E, or PBS (Fig. 5A). Mice were boosted at week 3 with 10 10 vp of Adenoviral vector, either heterologous or homologous to the prime Adenovirus type. One day prior to challenge, mice were administered 2 mg of anti-Ifnar1 mAb via i.p. injection. An additional 0.5 mg of mAb was administered on day 4 post infection to ensure lethal challenge. At week 7, all mice were challenged with 10 6 FFU of mouse adapted ZIKV Dakar strain 41525 and monitored for weight loss (Fig. 5B). All Adenoviral vaccinated mice survived challenge, regardless of prime/boost strategy (Fig. 5C). In comparison, all sham vaccinated mice showed significant weight loss by day 5 (Fig. 5B). All Adenovirus vaccinated animals showed decreased viremia compared to sham vaccinated mice by RT-qPCR (Fig. 5D) and plaque assay (Fig. S2C).

Discussion
Adenoviruses have been widely used as vaccine vectors due to their ability to induce humoral and cellular immune responses [44][45][46] . Replication-defective Adenovirus vectors have been shown to be safe and well tolerated in clinical trials 32,47,48 . Here we developed replication-defective human Ad5 and Ad4 vectored ZIKV vaccines and characterized their immune responses and protection against ZIKV infection. Ad5-prM-E vaccination . Two experimental replicates were performed. Mice were boosted with the same vaccine and dose at week 3 and then challenged with 10 6 FFU of mouse-adapted ZIKV strain 41519 at week 7. Weight loss was monitored (B) and mice were sacrificed when a 25% weight loss was reached (C). Asterisks indicate significance in weight compared to PBS sham vaccinated mice as determined by two-way ANOVA (p < 0.05). Survival data was analyzed using a log rank test ( *** p < 0.001, **** p < 0.0001). Blood was sampled at 4 days post infection (d.p.i) to determine the viral load in the sera using RT-qPCR (D) ( ** p < 0.01; **** p < 0.0001; one-way ANOVA). To determine immune correlates in Ifnar1 −/− mice, groups (n = 5) were immunized with 10 10 vp of the indicated Ad vaccine and sacrificed 2 weeks later. Sera was used to determine anti-ZIKV IgG (E) and neutralizing antibodies (F). In addition, an ELISPOT was performed to determine spot-forming cells per million splenocytes against the immunodominant E 4-12 epitope (G). ELISA, PRNT50, and ELISPOT are analyzed with one-way ANOVA ( ** p < 0.01, **** p < 0.0001). Data are expressed as the mean with standard error (SEM). induced a strong anti-ZIKV humoral immune response. However, Ad4-prM-E vaccination induced no detectable anti-ZIKV antibodies, even with vaccination with an equal IFU dose to the Ad5-prM-E vaccination or in a prime-boost strategy. Vaccination by either Ad5-prM-E and Ad4-prM-E resulted in the development of a T-cell immune response. This is important because CD8 + T-cells have been shown to protect against ZIKV infection 37,38,49 . Notably, both Ad vectored vaccines elicited T-cell responses against the immunodominant epitope E 4-12 . The ability of Ad4-prME vaccination to induce significant T-cell response but no detectable antibody response could demonstrate preferential development of cellular over humoral immune response in Ad4 vectored vaccination. Ad has a self adjuvanting property, and the Ad4 and Ad5 vectors may initiate very different cytokine profiles that are highly Th 1 biased in Ad4 immunized mice but more Th 1 /Th 2 balanced in the Ad5 immunized mice 50 . A phase I clinical trial utilizing Ad4 as an oral, replication-competent vector for H5 influenza vaccination found similar results. Participants vaccinated with an Ad4-H5 vector developed robust T-cell responses but failed to seroconvert even after 3 vaccinations (11% in vaccinated vs 7% in placebo groups) 51 . This could indicate that the immune profile of a strong T cell response with limited antibody development that we see with our Ad4-prM-E vaccination is not limited to mouse models or the ZIKV antigen, but further studies are needed to fully evaluate this.
In the anti-Ifnar1 blockade challenge model, both Ad5-prM-E and Ad4-prM-E vaccinated mice protected against weight loss and death. On day 4 post infection, vaccination significantly reduced levels of ZIKV RNA and infectious virus in serum. Even in the highly lethal Ifnar1 −/− mouse challenge model, the Ad vectored vaccines were partially protective. The immunocompromised state of the mice along with the challenge dose made this challenge highly stringent. However, as ZIKV infection is not lethal in adult human infection, further studies need to be performed to determine if the partial protection in this highly lethal challenge study translates to protective efficacy in human vaccination.
In conclusion, we constructed two replication-defective Adenovirus (Ad) vectored vaccine candidates by inserting the ZIKV structural prM-E genes into the E1 region of human Ad type 4 and Ad type 5. Both vectors have shown promise by protecting against ZIKV infection in a lethal antibody blockade challenge model. However, immune correlates indicate that Ad4-prM-E vaccination established a different immune profile than Ad5-prM-E vaccination. The ability of Ad4-prM-E vaccination to induce a strong T-cell response with undetectable levels of antibody, while Ad5-prM-E vaccination induces both strong antibody and T-cell responses, warrants further investigation. A comprehensive cytokine/chemokine and cellular profile analysis may elucidate these mechanisms. Due to the potential for antibody dependent enhancement (ADE) of Dengue virus from ZIKV antibodies 22,52,53 , the lack of antibody development in Ad4-prM-E vaccination could be explored as a potentially safer vaccine platform. Our study highlights the significant differences in immune profiles between these two Ad vectors and supports further investigation of the use of Ad vectors as a platform for ZIKV vaccine development. Weight loss was monitored (B) and mice were sacrificed at 25% weight loss (C). Asterisks indicate significant in weight compared to PBS sham vaccinated mice as determined by two-way ANOVA (p < 0.05). Blood was sampled at 4 d.p.i to determine the viral load in the sera using qPCR (D) ( **** p < 0.0001; one-way ANOVA). Data are expressed as the mean with standard error (SEM).

Recombinant adenovirus type 4 plasmid construction. The Adenovirus type 4 genome was
cloned and a shuttle plasmid to replace the E1 region was created as previously described 35 . Briefly, the complete Adenovirus type 4 (pAd4) genome was cloned into a single low-copy plasmid, and a shuttle plasmid (pAd4-recomb-cloner) to replace the E1 region of Ad4 was created using overlapping PCR products. The CMV-prM-E-PolyA region from the pShuttle-CMV-prM-E plasmid was amplified using the Platinum PCR SuperMix High Fidelity (Invitrogen) and fused to an Frt-Zeo-Frt fragment using overlapping PCR. This fragment was then cloned into pCR8/Gw/TOPO for sequence confirmation (Invitrogen). After sequence confirmation, the shuttle was linearized and transformed with the pAd4 plasmid into BJ5183 electrocompetent cells for homologous recombination. Recombinants were screened on kanamycin and zeocin and confirmed with restriction digest. A successful recombinant was then transformed into XL1 cells, and plasmid DNA was purified using the Qiagen Plasmid Midi Kit (Qiagen).

Recombinant adenovirus virus rescue and purification. The recombinant Adenovirus 5 and 4
genomes with the ZIKV prM-E insert were linearized and buffer exchanged using a Strataprep PCR purification kit (Agilent Technologies). The linearized recombinant gDNA was transfected into 293 cells using the PolyFect Transfection Reagent (Qiagen). After virus rescue was observed via plaque formation, cells were harvested and virus was released by 3 freeze-thaw cycles. Virus was amplified by sequential passages in 293 cells until a final amplification using a Corning 10-cell stack (~6300 cm2). The virus was purified by 2 sequential CsCl ultracentrifuge gradients, desalted using Econo-Pac 10DG Desalting Columns (Bio-Rad), and stored at −80 °C in Ad-tris buffer (20 mM Tris-HCl, 100 mM NaCl 2 , 1 mM MgCl 2 •6H 2 O, 10% glycerol).

Recombinant adenovirus quantification.
After CsCl purification and desalting, the virus particle quantity was determined on a NanoDrop Lite spectrophotometer (Thermo Fisher) with an OD260. The infectious unit titer was determined using the QuickTiter Adenovirus Titer Immunoassay Kit (Cell Biolabs, Inc. ZIKV foci forming unit assay. Virus stocks were titrated by focus forming assay (FFA) on Vero cells as previously described 39 and stored in aliquots at −80 °C.
Mouse challenge studies. Female Ifnar1 −/− mice (8-9 weeks old) were immunized intramuscularly at day 0 with 10 10 vp of Ad4-prM-E (n = 9), Ad5-prM-E (n = 10), or control PBS (n = 10). Two experimental replicates were performed however, after day 4, three mice in the Ad4-prM-E group were lost to unrelated causes due to anesthesia complications. Female C57BL/6 mice were immunized intramuscularly at day 0 with 10 10 vp of the indicated Ad vaccine or control PBS. All mice were boosted at week 3 with 10 10 vp of indicated Adenovirus or sham vaccine and then challenged at week 7 with 10 6 FFU of mouse-adapted ZIKV-Dakar 41525 via the intraperitoneal (i.p.) route. C57BL/6 mice were administered 2 mg of anti-mouse Ifnar1 mAb (MAR1-5A3, BioXcell) via the i.p. route one day before ZIKV infection and given 0.5 mg more at day 4 post infection. All mice were monitored daily for weight loss and mice were euthanized at 25% weight loss.

ELISPOT.
The T-cell epitopes were mapped using a peptide array of the ZIKV strain PRVABC59 E protein from BEI resources (Catalog No. NR-50553). This 164-peptide array spans the entire E region and consists of 15-mers with 12 amino acid overlap. Potential immunogenic peptides were identified using a matrix of peptides pools, and the epitopes were confirmed using individual peptides. Splenocytes were isolated from mice using a 40 μm Nylon cell strainer (BD Labware). Red blood cells were lysed using ACK lysis buffer and the splenocytes were resuspended in cRPMI at a concentration of 10 6 splenocytes/mL. Ninety-six well polyvinylidene difluoride-backed plates (MultiScreen-IP, Millipore) coated with 50 μl of anti-mouse IFN-γ mAb AN18 (5 µg/ ml; Mabtech) overnight at 4 °C. Plates were washed and blocked with RPMI at 37 °C for 1 hour. Equal volumes (50 µL) of the single-cell suspension splenocytes and peptide (5ug/mL) were added to the wells in duplicate. Plates were incubated overnight at 37 °C with 5% CO 2 . The plates were washed 6X with PBS and incubated with 100 μl of biotinylated anti-mouse IFN-γ mAb (1:1000 dilution; Mabtech) diluted in PBS with 1.0% FBS for 1 hour at RT. Plates were washed 6X with PBS and incubated with 100 µl of streptavidin-alkaline phosphatase conjugate (1:1000 dilution; Mabtech) diluted in PBS 1.0% FBS. After 1 hour at RT, the plates were washed 6x with PBS. To develop, 100 µl of BCIP/NBT (Plus) alkaline phosphatase substrate (Thermo Fisher) was added to each well and development was stopped by washing several times in dH 2 O. The plates were air dried and spots were counted using an automated ELISpot plate reader (AID iSpot Reader Spectrum). Results are expressed as spot-forming cells (SFC) per 10 6 splenocytes.
qPCR for viral load quantification. Challenged mice were bled from the submandibular vein at day 4 post infection and sera was isolated. RNA from sera was extracted using the PureLink Viral RNA/DNA Mini Kit according to manufacturer's instructions (Invitrogen). Real time RT-qPCR was performed using the Luna Universal Probe One-Step RT-qPCR Kit (NEB). It was run on a QuantStudio 3 Real-Time PCR System (Applied Biosystems) using the Luna kit cycling conditions (55 °C for 10 min, 95 °C for 1 min, and 40 cycles of 95 °C for 10 s and 60 °C for 1 min). Results were compared to a standard curve created using ZIKV RNA extracted from a known quantity of infectious virus. Unamplified samples were set at the limit of detection (100 FFU equivalent/mL). The following primer probe set was used: and 5% CO 2. This was then added to a 24-well plate of confluent Vero cells. The plaque assay was continued as described above.
Statistical analysis. GraphPad Prism software was used to analyze all data. Data are expressed as the mean with standard error (SEM). Survival curves were analyzed using the log rank test and weight loss was analyzed using two-way ANOVA multiple comparison test. ELISA, PRNT50, T-cell data, and viral burden were analyzed using one-way ANOVA with Bonferroni multiple comparisons. A p value < 0.05 was considered statistically significant ( * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001).