A single-dose live attenuated chimeric vaccine candidate against Zika virus

The mosquito-borne Zika virus is an emerging pathogen from the Flavivirus genus for which there are no approved antivirals or vaccines. Using the clinically validated PDK-53 dengue virus vaccine strain as a backbone, we created a chimeric dengue/Zika virus, VacDZ, as a live attenuated vaccine candidate against Zika virus. VacDZ demonstrates key markers of attenuation: small plaque phenotype, temperature sensitivity, attenuation of neurovirulence in suckling mice, and attenuation of pathogenicity in interferon deficient adult AG129 mice. VacDZ may be administered as a traditional live virus vaccine, or as a DNA-launched vaccine that produces live VacDZ in vivo after delivery. Both vaccine formulations induce a protective immune response against Zika virus in AG129 mice, which includes neutralising antibodies and a strong Th1 response. This study demonstrates that VacDZ is a safe and effective vaccine candidate against Zika virus.

Zika virus (ZIKV) is an emerging mosquito-borne pathogen from the genus Flavivirus1,2,3, which also contains other arthropod-borne pathogens such as yellow fever virus (YFV), dengue virus type 1 to 4 (DENV1 to DENV4), and Japanese encephalitis virus (JEV)4. ZIKV is typically spread by the Aedes aegypti and Aedes albopictus mosquitoes, although past outbreaks have also been linked to the Aedes hensilli mosquito5,6. ZIKV recently emerged to cause a larger outbreak in the Americas1,2,3. Between January 2015 and May 2017, ZIKV is estimated to have caused an estimated 8.5 million symptomatic infections in Brazil alone3.

ZIKV infection is reported to be asymptomatic in a majority of patients, with the reported asymptomatic rate varying from 29% to 82%, and with 80% being the commonly cited figure7. Symptomatic ZIKV infection typically presents as an acute and self-limiting febrile disease, with symptoms such as maculopapular rash, joint pain, and conjunctivitis1,2. In some patients, viable ZIKV can be recovered from patients weeks after the acute phase of the infection, and during this persistent infection phase ZIKV remains transmissible through sexual contact8. ZIKV infection is also associated with other severe complications. Infection in adults is associated with Guillain-Barré syndrome, and infection of pregnant women during the first two trimesters is associated with congenital microcephaly in newborns1,2,3. There are currently no approved vaccines or antiviral drugs for the prevention or treatment of ZIKV infection.

Flaviviruses share a common genome organisation, consisting of a positive sense single-stranded RNA genome that encodes for a single ORF that is flanked by 5′ and 3′ untranslated regions (5′UTR and 3′UTR)4. The ORF encodes for a polyprotein that is processed by host and viral proteases into the capsid, premembrane and envelope structural proteins (cap, prM and Env), as well as 7 non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5)4. The flavivirus structural proteins are functionally interchangeable to a certain extent, and it is possible to construct viable chimeric viruses which express the heterologous structural proteins of another flavivirus9,10,11.

There are successful live attenuated vaccines against YFV and JEV12. In particular, the live attenuated yellow fever vaccine that has been in use since 1938 is considered a gold standard as a safe and effective antiviral vaccine12. Live attenuated vaccines against DENV have also demonstrated good attenuation during clinical trials13,14,15. Live attenuated vaccines are highly immunogenic compared to inactivated vaccines12. For example the live JEV vaccine that is based on the SA14-14-2 strain only requires one dose, whereas inactivated JEV vaccines require multiple doses12. Therefore, a live attenuated vaccine against ZIKV would be a promising strategy for controlling ZIKV.

However, one major concern is that live vaccines that look promising during initial preclinical studies may show inadequate attenuation once they reach human clinical trials16. One way of addressing this risk is to repurpose a clinically validated vaccine strain to serve as a viral backbone for a chimeric virus9,10,11,13,14,15,17,18. Such a chimera would express the attenuated replication machinery of the vaccine strain, whilst also expressing the major structural antigens of the target virus. This strategy has been used in the development of several live attenuated tetravalent dengue vaccines, with the Sanofi-Pasteur, Takeda, and NIH dengue vaccines all containing at least one chimeric virus13,14,15,17,18. In particular, the Takeda TAK-003 (or DENVax) dengue vaccine that has recently completed phase II and III clinical trials contains four component viruses, one of which is the DENV2-PDK-53 vaccine strain that was developed by Mahidol University, Thailand9,10,13,14,18,19. The other three component viruses of TAK-003 are chimeric viruses that utilise DENV2-PDK-53 as their viral backbone.

Several studies have adopted the YFV-17D vaccine strain to serve as the backbone of a chimeric virus vaccine against ZIKV20,21,22, while another study adopted the JEV-SA14-14-2 vaccine strain for the same reason23. However, the mutations with the largest impact on attenuating the neurovirulence of JEV-SA14-14-2 lie within the Env protein, and these attenuating mutations are lost during chimerisation24. The situation is even more uncertain with the YFV-17D, because its attenuating mutations have yet to be identified25. Furthermore, live viral vaccines face several disadvantages in terms of genetic stability, storage, and scalability of production during sudden outbreaks26,27,28. These shortcomings resulted in a shortage of the live yellow fever vaccine during the 2016 YFV outbreak in Angola as the short shelf life of the vaccine had limited the size of stockpiles and manufacturers were unable to scale up production capacity in time28.

To address these problems, we developed a chimeric dengue/Zika virus called VacDZ as a vaccine against ZIKV (Fig. 1). DENV2-PDK-53 vaccine strain was chosen as a viral vaccine backbone because the attenuating mutations of DENV2-PDK-53 are located outside the prM and Env protein coding regions, meaning they are not affected by chimerisation19,29,30. The attenuating mechanism for one of these key attenuating mutations, NS1-G53D, was recently determined to be the enhanced induction of the type-I interferon response by the mutant NS1-G53D protein, a mechanism that should not be affected by chimerisation31. Furthermore, clinical trials have shown that chimeric vaccines that are derived from DENV2-PDK-53 have excellent safety profiles13,14. All this indicates that VacDZ is highly likely to retain the attenuation of the DENV2-PDK-53 backbone. We also developed VacDZ so that it can be delivered as either a live virus, or as a DNA-launched vaccine that produces live, functional VacDZ in vivo32,33,34. Here, we report the development of VacDZ as a safe and effective vaccine against ZIKV.

Construction of infectious clone of VacDZ chimeric dengue/Zika virus

We constructed the infectious clone of our chimeric dengue/Zika virus, VacDZ, (Fig. 1), by modifying our existing DENV2-16681 infectious clone35. The infectious clone of VacDZ, called pVacDZ, was constructed using conventional molecular cloning techniques (Supplementary Figure 1).

The 16681 strain of DENV2 is the parent strain of PDK-53. While PDK-53 differs from the parental wildtype DENV2-16681 by several mutations, only three mutations are necessary and sufficient for the attenuation of PDK-53: 5′UTR-c57t, NS1-G53D, and NS3-E250V9,10,11,29. Therefore, we only introduced these three attenuating mutations into our infectious clone to construct pVacDZ (Fig. 1, Supplementary Figure 1). We also replaced the region encoding for the prM signal sequence, prM protein, and Env protein with their counterpart from ZIKV strain PRVABC59 (Fig. 1b). The prM signal sequence at the capsid-prM junction was replaced because prior studies with dengue/West Nile virus chimeras showed that the prM protein requires a homologous signal sequence upstream for correct processing11. The signal sequence at the Env-NS1 junction is also shown (Fig. 1c).

Flaviviral infectious clone plasmids can be highly unstable, and this instability can prevent the construction of infectious clones for chimeric vaccines10,19. Therefore, we adopted two strategies in order to stabilise our pVacDZ infectious clone. First, we utilised an inducible CMV promoter (TRE-minCMV) that has been shown to stabilise flavivirus infectious clones35,36. Second, we cloned an intron sequence into the NS1 coding region37.

We also constructed a ΔGDD mutant of VacDZ as a control for DNA vaccination studies (Supplementary Figure 2). VacDZ-ΔGDD has a lethal deletion of the ‘glycine-aspartic acid-aspartic acid’ catalytic triad of the NS5 RNA-dependent RNA polymerase (RDRP)4. While VacDZ-ΔGDD can express a small quantity of viral proteins, it cannot enter into the exponential RNA-replication and protein expression stage of the virus replication cycle.

Rescue of VacDZ by DNA-launch

Virus rescue for our infectious clones takes place by DNA-launch, whereby the infectious clone plasmid is co-transfected with a pTet-Off Advanced accessory plasmid35,36. After transfection, the accessory plasmid activates the TRE-minCMV promoter of the infectious clone, which then recruits host nuclear machinery to transcribe the viral RNA. The viral RNA is eventually processed into a functional viral genome that can establish a normal virus replication cycle. VacDZ was rescued by DNA-launch in BHK-21 cells to produce a seed stock. VacDZ seed stock was then amplified by passaging the virus in BHK-21 cells to produce a working stock, which was then titrated by plaque assay.

VacDZ expresses ZIKV envelope protein

We first verified if VacDZ was expressing the ZIKV Env protein. BHK-21 cells were infected with VacDZ or with parental DENV2-16681 or ZIKV. Mock infected cells were used as a control. The cells were then fixed and viral protein expression was analysed using immunofluorescence assay. Cells that were infected with VacDZ, DENV2-16681, or ZIKV were found to express flaviviral NS1 protein, whereas mock infected cells did not (Supplementary Figure 3). Cells that were infected with VacDZ or ZIKV were also found to express ZIKV Env protein, while mock infected cells and DENV2-16681 infected cells did not (Supplementary Figure 3). This demonstrates that VacDZ expresses ZIKV Env protein as intended.

VacDZ retains the small plaque phenotype of PDK-53

In order to validate the potential safety of VacDZ, we investigated if VacDZ retains the key markers of attenuation of the DENV2-PDK-53 vaccine strain, namely small plaque phenotype, temperature sensitivity, attenuation of neurovirulence in suckling mice, and attenuation of pathogenicity in AG129 mice9,10,11,19,29.

We started by investigating if VacDZ retains the small plaque phenotype. We compared the plaque formation phenotype of VacDZ and parental wildtype DENV2-16681 by performing a plaque assay in BHK-21 cells. VacDZ formed significantly smaller plaques than DENV2-16681 (Figs. 2a, b), confirming that VacDZ retains the small plaque phenotype.

VacDZ replication kinetics and temperature sensitivity

Next, we investigated the replication kinetics of VacDZ in comparison to the parental wildtype DENV2-16681 and ZIKV. We also investigated if VacDZ retains the temperature sensitivity phenotype, which is an attenuation marker that is defined as a reduction of the viral titre of 1-log or more at 39 °C relative to 37 °C11.

We compared the growth kinetics of these viruses in four cell lines which are commonly used for flavivirus studies: BHK-21 baby hamster kidney cells, Vero African green monkey kidney cells, Huh-7 human hepatoma cells, and C6/36 Aedes aegypti mosquito cells. The cells were infected with VacDZ, DENV2-16681, or ZIKV. After infection, the supernatant was harvested daily, and the extracellular virus titre was determined by using plaque assay. Temperature sensitivity was investigated in BHK-21 cells.

In BHK-21 cells, ZIKV had the highest peak titre, followed by VacDZ and then DENV2-16681 (3.33 × 107, 1.30 × 106, and 3.53 ×105 PFU/ml respectively) (Figs. 2c-e). Both VacDZ and ZIKV were temperature sensitive, whereas DENV2-16681 was not (Fig. 2c-e).

In Vero cells, DENV2-16681 had the highest peak titre, followed by ZIKV and then VacDZ (2.00 × 106, 1.93 × 106, and 1.90 × 106 PFU/ml respectively) (Supplementary Figure 4a). In Huh-7 cells, ZIKV had the highest peak titre, followed by VacDZ and then DENV2-16681 (4.33 × 106, 1.13 × 106, and 4.63 × 105 PFU/ml respectively) (Supplementary Figure 4b). For viral replication in mammalian cells, VacDZ generally seems to replicate to titres which are intermediate to the parental DENV2-16681 and ZIKV. In C6/36 mosquito cells, ZIKV had the highest peak titre, followed by DENV2-16681 and then VacDZ (7.33 × 107, 2.30 × 107, and 2.53 × 105 PFU/ml respectively) (Supplementary Figure 4c). This demonstrates that VacDZ has attenuated replication kinetics in mosquito cells, which is a predictive marker for reduced transmissibility in mosquitoes11. This is ideal since vaccine strains should not be transmissible by mosquitoes11.

VacDZ is genetically stable

The potential safety of VacDZ is dependent on its three attenuating mutations: 5′UTR-c57t, NS1-G53D and NS3-E250V. These attenuating mutations are known to be stable for the component viruses of the TAK-003 live dengue vaccine, which also utilise the DENV2-PDK-53 backbone19,30. Therefore, we investigated if these attenuating mutations were also stable in VacDZ. We serially passaged VacDZ in BHK-21 cells. At passage 10, we extracted the viral RNA and analysed the reversion rates of the attenuating mutations using next-generation sequencing (Omics Drive Pte Ltd). All three attenuating mutations had reversion rates of less than 1% (Supplementary Table 1). This degree of genetic stability is similar to DENV2-PDK-53 and other DENV2-PDK-53 derived chimeras19. This demonstrates that VacDZ is genetically stable, with a low risk of reversion.

VacDZ retains attenuation of neurovirulence in suckling mice

We proceeded to evaluate the in vivo safety and efficacy of VacDZ using two different mouse models. We used a suckling mouse model to evaluate the attenuation of neurovirulence, while the interferon deficient AG129 mouse model was used to evaluate the overall attenuation and immunogenicity of VacDZ9,10,11,19.

Newborn outbred white ICR mice were challenged by intracranial inoculation within 24 h of birth with different doses of VacDZ, DENV2-16681, ZIKV, or a PBS vehicle control. The mice were then observed daily for clinical signs for a period of four weeks (Fig. 3a, Supplementary Table 2).

The clinical signs that were commonly observed in mice that were challenged with ZIKV or DENV2-16681 were weight loss, wobbling gait, limb weakness or dragging, paralysis, hunching, or inactivity. Mice that were challenged with 1, 10, or 100 PFU of ZIKV per mouse had a mortality rate of 80%, 100%, or 100% respectively. Mice that were challenged with 10 or 100 PFU of DENV2-16681 per mouse had a mortality rate of 66.7% or 75% respectively. In contrast, mice that were challenged with 1, 10, or 100 PFU of VacDZ per mouse or with the PBS vehicle control had a mortality rate of 0%. This demonstrates that VacDZ has highly attenuated neurovirulence in suckling mice compared to the parental wildtype DENV2-16681 and ZIKV.

VacDZ has attenuated pathogenicity in AG129 mice

We proceeded to evaluate VacDZ for safety and efficacy in the interferon deficient AG129 mouse model. AG129 mice are commonly used for flavivirus vaccine studies because they are highly susceptible to lethal flavivirus challenge, but retain the ability to develop an adaptive immune response10,38,39.

ZIKV strain PRVABC59 (ZIKV-PRVABC59) is known to be lethal for AG129 mice, whereas DENV2-16681 is not38,40. We started by investigating the pathogenicity of ZIKV-PRVABC59 in AG129 in order to determine a suitable dose for subsequent lethal challenge studies. Because VacDZ encodes the prM and Env proteins of ZIKV-PRVABC59, we also investigated if expression of these ZIKV proteins resulted in any gain of pathogenicity for VacDZ.

AG129 mice were challenged by intraperitoneal inoculation with different doses of VacDZ, ZIKV, or PBS vehicle control. The mice were then observed daily for clinical symptoms for a period of four weeks (Fig. 3b). Mice that were challenged with 103 or 104 PFU of ZIKV per mouse had a mortality rate of 100%. The most commonly observed clinical signs in mice that were challenged with ZIKV was weight loss, with most of the mice reaching a humane endpoint of cumulative 20% weight loss relative to the initial weight. In contrast, mice that were challenged with 106 PFU of VacDZ per mouse or with the PBS vehicle control had a mortality rate of 0%. The changes in mouse weights are shown in Supplementary Figure 5. This demonstrates that VacDZ has attenuated pathogenicity in AG129 mice as well.

VacDZ is immunogenic in AG129 mice

Next, we investigated if vaccination with VacDZ could induce an immune response against ZIKV in the form of neutralising antibodies or a T helper cell response.

We tested VacDZ as both a live virus vaccine (live VacDZ) and also as a DNA-launched vaccine (DNA-launched VacDZ). The DNA-launched vaccine is essentially an alternative delivery method, whereby the pVacDZ infectious clone is transfected in vivo, after which the production of live VacDZ occurs in vivo32,33. This live, DNA-launched VacDZ can then proceed to establish a virus replication cycle as per normal.

Most DNA-launched flavivirus vaccines reported to date have used electroporators for in vivo transfection32,33. In order to simplify our DNA vaccination protocol, we used the in vivo-jetPEI transfection reagent (Polyplus transfection) since it does not require an electroporator for delivery.

For live VacDZ, the dose for vaccination or boosting was 104 PFU of VacDZ per mouse, and PBS was used as a vehicle control. For DNA-launched VacDZ, the dose for vaccination or boosting was 80 μg of pVacDZ and 20 μg of pTet-Off Advanced per mouse, while a dose of 80 μg of pVacDZ-ΔGDD mutant and 20 μg of pTet-Off Advanced per mouse was used for the RDRP-defective control, and a 5% glucose solution was used as a vehicle control.

AG129 mice were first vaccinated with live VacDZ, DNA-launched VacDZ, or their respective controls via intraperitoneal inoculation. Four weeks after primary vaccination, the mice were boosted with the same dose of their respective vaccine or control. Four weeks after boosting, the mice were killed and their blood and organs were harvested.

To investigate if there was induction of neutralising antibodies in the vaccinated mice, we analysed their serum using plaque reduction neutralisation test (PRNT). AG129 mice that were vaccinated with live VacDZ or with DNA-launched VacDZ were all found to have neutralising antibodies against ZIKV (100% seroconversion rate), with live VacDZ inducing a stronger response, though this difference was not significant (Fig. 4, Supplementary Table 3). In contrast, AG129 mice vaccinated with the PBS vehicle control, 5% glucose solution vehicle control, or pVacDZ-ΔGDD mutant control all had a 0% seroconversion rate (Fig. 4, Supplementary Table 3).

Live VacDZ vaccinated mice had anti-ZIKV PRNT titres ranging from 1:640 to 1:20480 (geometric mean titre of 3044.37) (Fig. 4a). DNA-launched VacDZ vaccinated mice had anti-ZIKV PRNT titres ranging from 1:640 to 1:1280 (geometric mean titre of 1114.3) (Fig. 4b).

To investigate if there was induction of a type 1 or type 2 T helper cell (Th1 or Th2) response in the vaccinated mice, we analysed their splenocytes by using IFNγ or IL-4 ELIspot assays respectively. The mouse splenocytes were activated using ZIKV at a MOI of 0.5, or with control RPMI medium supplemented with 2% FCS. AG129 mice that were vaccinated with live VacDZ developed a significant Th1 and Th2 response (Fig. 5a, b). However, the Th1 response was stronger compared to the Th2 response. For the mice vaccinated with live VacDZ, the Th1 response was 10.2-fold higher compared to the mice vaccinated with the PBS control, whereas the Th2 response was only 2-fold higher.

Mice that were vaccinated with DNA-launched VacDZ developed a significant Th1 response, but no significant Th2 response (Fig. 5c and d). As expected, mice that were vaccinated with the ΔGDD mutant control did not develop any significant Th1 or Th2 response compared to the 5% glucose solution vehicle control (Fig. 5c and d). For the mice vaccinated with DNA-launched VacDZ, the Th1 response was 12.2-fold higher compared to the mice vaccinated with the 5% glucose solution control.

This demonstrates that both live VacDZ and DNA-launched VacDZ induce an immune response against ZIKV in the form of neutralising antibodies and a Th1 response. This Th1 biased response and the associated IFNγ response are favourable because they are associated with protective immunity against flaviviruses, as well as milder disease outcomes during flavivirus infection41,42,43.

VacDZ confers protective immunity against lethal ZIKV challenge

Finally, we investigated if vaccination with VacDZ confers protective immunity against lethal ZIKV challenge. For the protective immunity studies, we adopted a single dose regime that did not involve any boosting: mice were vaccinated with a single dose of vaccine or control, and then challenged with ZIKV.

AG129 mice were vaccinated with live VacDZ, DNA-launched VacDZ, or their respective controls, using the same dose described above for immunogenicity studies. Four weeks after vaccination the mice were challenged with a lethal dose of ZIKV (105 PFU per mouse). The mice were then observed daily for clinical signs for a period of four weeks. Four weeks after lethal challenge, the mice were killed and their blood and organs were harvested.

Mice that were inoculated with the PBS or 5% glucose solution vehicle controls had a survival rate of 0% when challenged with ZIKV (Fig. 6, Supplementary Table 4). Mice that were vaccinated with the replication-defective pVacDZ-ΔGDD mutant control had a survival rate of 20% when challenged with ZIKV (Fig. 6b, Supplementary Table 4). In contrast, mice that were vaccinated with live VacDZ or DNA-launched VacDZ had a survival rate of 100% or 80% respectively when challenged with ZIKV (Fig. 6, Supplementary Table 4). Mouse weights are shown in Supplementary Figure 6. This demonstrates that both live and DNA-launched VacDZ confer protective immunity against ZIKV, with live VacDZ having better protective efficacy.

Next, to investigate if there was induction of neutralising antibodies in the surviving mice, we analysed their serum by using PRNT. All surviving mice were found to have neutralising antibodies against ZIKV (seroconversion rate of 100% in surviving mice) (Fig. 7, Supplementary Table 4). Live VacDZ vaccinated mice had anti-ZIKV PRNT titres ranging from 1:640 to 1:20480 (geometric mean titre of 2726.51). Surviving DNA-launched VacDZ vaccinated mice had anti-ZIKV PRNT titres ranging from 1:1280 to 1:10240 (geometric mean titre of 3620.39).

Curiously, the pVacDZ-ΔGDD mutant that was used as the RDRP-defective negative control for DNA vaccination seemed to offer weak protection against ZIKV. For the immunogenicity studies in AG129 mice, the strongest Th1 response from the pVacDZ-ΔGDD vaccinated group was higher than the weakest Th1 response from the group vaccinated with DNA-launched VacDZ (mean spot forming counts of 215.0 and 201.7 respectively for IFNγ-ELIspot) (Fig. 5c). However, pVacDZ-ΔGDD vaccination did not induce any neutralising antibodies against ZIKV, even with a vaccination and boost regime (Fig. 4b). Therefore, we hypothesise that pVacDZ-ΔGDD is able to elicit a protective T-cell response in a minority of mice. This may have allowed one of the mice in the pVacDZ-ΔGDD vaccinated group to survive lethal challenge with ZIKV (Fig. 6b), and in the process the mouse was able to seroconvert as well (Fig. 7).

In this study, we developed a chimeric dengue/Zika virus called VacDZ as a vaccine candidate against ZIKV. VacDZ utilises the clinical validated DENV2-PDK-53 vaccine strain as the backbone of the chimeric virus13,14. This was to ensure that our vaccine would be sufficiently attenuated. Unlike other flavivirus vaccine strains, the attenuating mutations of DENV2-PDK-53 vaccine are known, and they are not affected by chimerisation of the structural proteins29. This is a similar approach to the chimeric rZIKV/D4Δ30-713 vaccine for ZIKV, which is derived from the live attenuated dengue 4 backbone, rDEN4Δ3044. At the time of writing, the rZIKV/D4Δ30-713 is set to undergo phase I clinical trials, but the preclinical data of the vaccine has not been published44. The attenuating mutations of both the DENV2-PDK-53 and rDEN4Δ30vaccine strains are known, and in each vaccine strain they are located outside the structural protein coding region29,45. Both DENV2-PDK-53 and rDEN4Δ30 have been successfully repurposed as backbones for chimeric vaccines45,46,47. The Takeda TAK-003 (or DENVax) tetravalent dengue vaccine contains DENV2-PDK-53 itself, as well as three chimeric viruses which utilise DENV2-PDK-53 as their backbone13,14,18. The NIH TV003 tetravalent dengue vaccine contains DEN4Δ30 itself, as well as a chimeric DENV2/4 which utilises rDEN4Δ30as its backbone45,46,47. However, the DENV2-PDK-53 vaccine strain is the more established vaccine backbone, as the Takeda TAK-003 has already completed phase II & III clinical trials, while the NIH TV003 vaccine has only completed phase II clinical trials13,14,45,46,47.

We found that VacDZ retained the key attenuation markers of the DENV2-PDK-53 vaccine strain: small plaque phenotype, temperature sensitivity, attenuation of neurovirulence in suckling mice, and attenuation of pathogenicity in AG129 mice. We found that both live VacDZ and DNA-launched VacDZ could induce a protective immune response against ZIKV. This immune response included neutralising antibodies against ZIKV, a strong Th1 (and IFNγ) response against ZIKV, and a weaker or negligible Th2 T-cell response. This Th1 biased response and the associated IFNγ response are favourable because they are associated with milder disease outcomes during flavivirus infections and with a protective CD8 + T-cell response41,42,43.

Because VacDZ is a chimeric vaccine that is derived from the DENV2-PDK-53 vaccine strain, we based our vaccine design and development process on prior studies which describe the development on other chimeric vaccines which are derived from DENV2-PDK-5310,11,18. Therefore, we adopted the interferon deficient AG129 mouse strain that was described in some of these prior studies because it serves as a highly reproducible lethal mouse model10,18,39. In particular, we found that live VacDZ and DNA-launched VacDZ were able to induce comparable neutralising antibody titres against their respective target viruses in AG129 mice (1:1114.3 to 1:3044.37) when compared to chimeric DENV2/1, DENV2/3, and DENV2/4 vaccines (1:320 to 1:2560)10. However, while the AG129 strain is a highly reproducible mouse model that is used in many flavivirus vaccine studies, the defective interferon response may affect its relevance as a model for pathogenic flavivirus infections in humans10,38,39,48. Therefore, in future studies we hope to evaluate our VacDZ vaccine in a non-human primate model or in a immunocompetent ZIKV mouse model, such as a recently developed human STAT2 knock-in mouse model49,50.

The World Health Organisation has published a target product profile (TPP) for ZIKV vaccines51. Two potential vaccination strategies are proposed. The first is a routine vaccination programme where there will be predictable demand, production, and delivery of the vaccine. In such circumstances, live VacDZ would be an ideal candidate. Live VacDZ had better efficacy compared to DNA-launched VacDZ. Furthermore, live attenuated virus vaccines, such as the YFV-17D vaccine, are a proven technology with a long, safe, and successful track record. The longer production time for live vaccines should not pose a problem given the predictable turnover of vaccine stocks. However, the WHO TPP is oriented towards the deployment of ZIKV vaccines in response to an outbreak. In these circumstances, DNA-launched VacDZ is not without its merits. DNA-launched VacDZ combines the advantages of DNA vaccines and live vaccines. First, DNA plasmids are more stable and easier to store, with simplified cold chain requirements26,27,28,34. While live VacDZ needs to be stored at -80 °C, DNA-launched VacDZ can be stored at -20 °C for many years, or even at 4 °C for a few months. This simplified storage requirement is one of the desired characteristics described in the WHO TPP for ZIKV vaccines51. Second, compared to live viral vaccines, it is easier to scale up the production of DNA vaccines during an outbreak26,27,28. Third, DNA-launched VacDZ is far more immunogenic compared to traditional DNA vaccines that express non-replicating antigens26,27,34. Fourth, while prior DNA-launched flavivirus vaccines require electroporation for in vivo delivery26,32,33, we show for the first time that a PEI based reagent, in vivo-jetPEI, can be used as an alternative. According to the manufacturer’s protocol (Polyplus transfection), in vivo-jetPEI can be stored at the same temperature as DNA (-20 °C). Furthermore, once the DNA-reagent transfection mixture is prepared, it remains stable for up to 7 days when it is stored at 4 °C. By circumventing the need for an electroporator and by adopting a transfection reagent with simplified storage requirements, our approach maintains the convenience that DNA vaccines are meant to offer. Future work may focus on improving this reagent-based approach for vaccine delivery, as this would eventually make it easier to deploy DNA-launched VacDZ for mass vaccinations during outbreaks, especially in rural areas.

In this study, we demonstrate that VacDZ is a safe and effective vaccine candidate against ZIKV. VacDZ can be delivered as a live virus formulation or as a DNA-launched formulation. This convenience and flexibility would greatly benefit the future development and deployment of the vaccine, especially in vulnerable populations. All this makes VacDZ a promising vaccine candidate against ZIKV.