mRNA Vaccine Against Japanese Encephalitis Virus Genotype IV Protects Against Lethal Infection

TL;DR

Researchers developed an mRNA vaccine against Japanese encephalitis virus genotype IV that protected mice from lethal infection and reduced viremia.

Contribution

The study demonstrates rapid in-country development of an mRNA vaccine for a local JEV GIV outbreak with promising protective efficacy.

Findings

Two mRNA vaccines encoding prME provided ≥80% protection against disease and weight loss in mice.

Both vaccines reduced viremia by five to six logs and showed significant neutralizing antibody responses.

The Shorter vaccine induced lower but overlapping antibody responses compared to the approved Imojev vaccine.

Abstract

In 2022, Australia saw an unprecedented outbreak of Japanese encephalitis virus genotype IV (JEV GIV). The outbreak involved 42 human cases with 7 fatalities, as well as affecting >80 pig farms in New South Wales and Queensland. Herein, we designed, constructed, and tested two JEV GIV mRNA vaccines encoding prME, which provided protection against a lethal JEV GIV challenge in an Ifnar-/- mouse model. The vaccines were not codon optimized and included either the Native (full-length) or a Shorter signal peptide, with the latter missing the N-terminal n-region. Two vaccinations with 5 µg of the Shorter vaccine provided neutralizing antibody responses that were significantly lower but overlapped with those seen after vaccination with Imojev, a live attenuated vaccine approved for use in humans. Both mRNA vaccines provided approximately a five to six log reduction in viremia, ≥80% protection against overt disease and weight loss, and mortality. The paper illustrates in-country mRNA vaccine generation in response to a local outbreak, with JEV mRNA vaccines potentially emerging to be easier to manufacture, cheaper, and more suitable for immunocompromised individuals.

Figures

Full text

1. Introduction

In 2022, the World Health Organization (WHO) launched the Global Arbovirus Initiative, with pathogenic orthoflaviviruses representing a major burden to human and animal health globally [1]. In response, a number of mRNA vaccines for orthoflaviviruses are being developed, and one of the most advanced is a Zika virus (ZIKV) mRNA vaccine [2] that has entered phase II trials (mRNA-1893; ClinicalTrials.gov ID NCT04917861). Preclinical studies have also demonstrated the efficacy of mRNA vaccines for Yellow fever virus [3], Duck Tembusu virus [4], Powassan virus [5], genotype III Japanese encephalitis virus (JEV) [6,7], and single serotype and NS1-based vaccines for dengue [8]. These vaccines usually encode the pre-membrane and envelope (prME) polyprotein of the respective orthoflaviviruses. The prME polyproteins are likely proteolytically processed in the vaccine recipients [4,6], and self-assemble into subviral particles (SVPs) [9,10]. Although often smaller than orthoflavivirus virions [9,10], SVPs nevertheless present authentic tertiary and quaternary structures to the immune system [11], a key feature for generating neutralizing (and other protective) antibody responses [1].

In 2022, Australia saw an outbreak of a genotype IV (GIV) JEV, with 42 human cases and 7 fatalities. Over 80 pig farms in New South Wales, Victoria, and Queensland were also affected, with the JEV GIV outbreak associated with fetal deaths, piglets born with severe abnormalities, and prolonged infertility in boars [12,13]. Phylogeographic analysis suggests JEV GIV was previously limited to Indonesia and Vietnam from 1980 and spread to the Tiwi Islands between 2005 and 2020 [14]. The unprecedented outbreak in Australia resulted in the Acting Chief Medical Officer of Australia declaring the event a “Communicable Disease Incident of National Significance” on 4 March 2022 (ended 16 June 2023) [15]. Imojev is a live attenuated, chimeric single-dose human JEV vaccine used in Australia during the outbreak; it is relatively expensive (≈$AU 300) and contraindicated for immunocompromised individuals and pregnant women [16]. Imojev is based on a genotype III (GIII) JEV isolate, but provides neutralizing antibodies against GIV viruses [12,17], with a GIV isolate from the Australian outbreak JEV_NSW2022_ showing 93.4% amino acid identity with Imojev in the E protein sequence [18]. Nevertheless, sera from cohorts of Imojev vaccinated individuals showed significantly lower neutralization titers for JEV_NSW2022_ when compared with GIII viruses [18].

mRNA vaccine technology is now well-established for use in humans [19], with substantial evidence showing a favorable safety profile for this technology [20]. A number of veterinary applications are also being pursued [21]. For instance, a Sequivity platform mRNA vaccine (MSD Animal Health/Merck Animal Health) was approved in the USA in 2022 for use in swine (influenza vaccine, IAV-S NA) [22,23].

Herein, we illustrate the local production and preclinical evaluation of JEV GIV mRNA vaccines, illustrating how this technology can be used to respond rapidly to emerging infectious diseases, both locally [24] and internationally, in the face of increasing threats from arboviral diseases [25]. To evaluate these vaccines, we used a previously established mouse model for JEV_NSW2022_, which involves use of a type I interferon receptor 1 deficient (Ifnar-/-) mouse strain [13,18]. Ifnar-/- mice have significantly impaired type I interferon (IFN) responses, thus infection with a range of arboviruses, including JEV_NSW2022_, often results in rapidly rising and high viremias leading to acute mortality [13,18,26]. The model thus provides a robust challenge model, wherein vaccine-induced adaptive immunity must rapidly control the infection given that the early innate anti-viral responses are significantly diminished. However, the Ifnar-/- model does not recapitulate the infection of brain cells or encephalitis but is used herein, because JEV_NSW2022_ replicates poorly in C57BL/6J mice [18]. Induction of cellular immunity might be suboptimal in Ifnar-/- mice as IFNβ promotes cellular immunity after vaccination with standard pseudouridine-modified mRNA vaccines [27,28] (also used herein). However, such vaccines can stimulate adjuvant activity via a number of mechanisms [29], with the induction of antibody responses by mRNA vaccines in Ifnar-/- mice not overly confounded by the loss of the type I IFN receptor [30,31].

2. Materials and Methods

2.1. Ethics Statement and Regulatory Compliance

Mouse work was conducted in accordance with the “Australian code for the care and use of animals for scientific purposes” as defined by the National Health and Medical Research Council of Australia. Mouse work was approved by the QIMR Berghofer Animal Ethics Committee (P3746, A2108-612. Approved 9 January 2025). The breeding and use of GMO mice were approved under a Notifiable Low Risk Dealing (NLRD) Identifier: GTSC_075_2025: NLRD1.1(a).

The use of Imojev, the live attenuated chimeric JEV vaccine strain, was approved under NLRD_Suhrbier2_Mar2024: NLRD2.1(d), NLRD2.2. All infectious JEV work was conducted in a dedicated suite in a biosafety level 3 (PC3) facility at the QIMR Berghofer (Australian Department of Agriculture, Fisheries and Forestry certification Q2326 and Office of the Gene Technology Regulator certification 3445). All work was approved by the QIMR Berghofer Safety Committee (P3746).

Research with JEV at QIMR Berghofer was approved under the Queensland Biosecurity Act, Scientific Research Permit (restricted matter)—Permit number PRID000916.

2.2. Genotype IV JEV Isolate

JEV_NSW2022_ (GenBank: OP904182) was kindly provided by Dr. Peter Kirkland (Elizabeth Macarthur Agriculture Institute, New South Wales, Australia) [13], and represents a genotype IV isolate from the recent outbreak in Australia [32]. Virus stocks were generated from supernatant of the infected C6/36 cells and titers determined by CCID_50_ assays as described [18].

2.3. Construction of mRNA

Plasmid construction and colony screening. DNA plasmids encoding JEV_NSW2022_ (Native) and JEV_NSW2022_ (Shorter) vaccine sequences were supplied by Twist Bioscience (Decode Science, Mt Waverly, VIC, Australia). mRNA vaccine sequences were amplified with Phusion Plus Master Mix (Thermo Fisher, Woolloongabba, Australia) (primers 5ʹ to 3ʹ: Native F CAACCTCAAACAGGATCCACCATGGGAGGGAATGGAGGAACAGTCTTGTGGCTCATGAGC, R TTGGACAGCAAGAAAGCGAGCTTATCAAGCGTGCACATTGGTTGCTAAAAA—CAAGAGTGTTCC; Shorter F CAACCTCAAACAGGATCCACCATGTTGTGGCTCATGAGCTTGACAATTGCGCAGTCAGTG, R TTGGACAGCAAGAAAGCGAGCTTATCAAGCGTGCACATTGGTTGCTAAAAACAAGAGTGTTCC). PCR products were purified on GeneJET columns (Thermo Fisher, Woolloongabba, Australia) and size-verified by automated gel electrophoresis on an Agilent TapeStation (Agilent, Santa Clara, CA, USA). Purified PCR products were cloned by DNA assembly into a plasmid vector, designed for mRNA IVT template production. The vector includes a kanamycin resistance gene, high-copy number origin of replication, T7 RNA polymerase promoter, NcoI site flanked by 5′/3′ human beta-globin UTRs, a poly(A) tail, and a 3′ BspQI type IIS restriction site for linearization. The vector backbone was linearized with NcoI and PCR product assembled into this site. DNA assembly reactions were performed using the NEBuilder^®^ HiFi DNA Assembly Master Mix (New England Biolabs (NEB), Ipswich, MA, USA) and transformed into NEB Stable Competent E. coli (NEB, Ipswich, MA, USA), following the manufacturer’s instructions. Transformation plates were incubated for 24 h at 30 °C. Single colonies were picked and inoculated into Luria–Bertani (LB) medium supplemented with 50 µg/mL kanamycin and cultured overnight at 30 °C with shaking at 300 rpm. Plasmid DNA was extracted from overnight culture using the GeneJET Plasmid Miniprep Kit (Thermo Fisher, Woolloongabba, Australia). Successful clones were identified by restriction digestion followed by automated gel electrophoresis using the Agilent TapeStation system. The sequence identity of successful clones was determined using Oxford Nanopore Sequencing with the Rapid Barcoding Kit 24 V14. Nanopore sequencing was performed using the MinION Mk1b device and Flongle flow cell (FLO-FLG114) from Oxford Nanopore Technologies, following the manufacturer’s instructions. Base calling was carried out using Guppy Dorado basecaller (v7.3.9) in the MiniKNOW software (v24.02.6). The assembly and annotation of the plasmids were achieved using the Fastq Clone Validation workflow in EPI2ME Lab (v5.0.2). The resulting FASTA output reads were aligned to a reference construct sequence using SnapGene (v7.0.1).

Cell line generation. Purified plasmid DNA of correct identity was transformed into NEB Stable Competent E. coli (NEB, Ipswich, MA, USA), and transformation plates were incubated for 24 h at 30 °C. Single colonies were picked from the transformation plates for inoculation of overnight cultures in LB medium containing kanamycin, as detailed for clone screening. Glycerol stocks were prepared from the overnight cultures and promptly stored at −80 °C.

Plasmid DNA template production. A single clone for JEV_NSW2022_ (Native) and JEV_NSW2022_ (Shorter) confirmed by nanopore sequencing was used to inoculate a primary culture of medium supplemented with 50 µg/mL kanamycin. The culture was seeded and grown for 20 h at 30 °C with shaking at 300 rpm in an orbital incubator. Secondary cultures were prepared under the same conditions as the primary culture. Plasmid DNA was purified using the PureLink™ HiPure Plasmid Filter Midiprep Kit (Thermo Fisher, Woolloongabba, Australia) following the manufacturer’s instructions. The purified supercoiled plasmid was linearized with BspQI (Hongene, Shanghai, China) and incubated at 50 °C for 5 h. Linearized plasmid DNA was subsequently purified using QIAGEN (Clayton, VIC, Australia) silica columns. Both supercoiled and linearized plasmid preparations were analyzed by agarose gel electrophoresis to assess size, integrity, and completion of the restriction digest. Plasmid concentrations were quantified by UV absorbance using a NanoPhotometer^®^ NP80 (Implen, Munich, Germany).

mRNA IVT production. Linearized and purified plasmid was used as template for mRNA IVT using T7 RNA polymerase (Roche, Indianapolis, IN, USA), ribonucleotide triphosphates (CTP, GTP, ATP and N1-methylpseudoUTP; Roche and Thermo Fisher), Co Cap A (SRNA), and mM Magnesium Acetate (Merck, Rahway, NJ, USA). IVT reactions were buffered with Tris-HCl, pH 8.0 (Merck) and supplemented with Inorganic Pyrophosphatase (NEB, Ipswich, MA, USA), DL-1,4-Dithiothreitol (Thermo Fisher, Woolloongabba, Australia), Spermidine (Sigma Aldrich, North Ryde, Australia), and Triton X-100 (Sigma Aldrich, North Ryde, Australia). To mitigate the risk from RNase contamination, Murine RNase inhibitor (NEB, Ipswich, MA, USA) was incorporated, and the reaction was prepared in an RNase-free environment following aseptic techniques. The IVT reaction was incubated at 37 °C for 240 min and then terminated by the addition of EDTA. The integrity of unpurified mRNA post-IVT samples was assessed by automated gel electrophoresis on an Agilent TapeStation system.

mRNA chromatography purification. An ÄKTA™ avant 150 chromatography system (Cytiva. Marlborough, MA, USA) was used for chromatography, with UNICORN 8 (Cytiva) software. A CIMmultus Oligo dT18 monolithic column (Sartorius, Göttingen, Germany) was used to capture and purify polyA mRNA at room temperature. The column was pre-equilibrated in binding buffer (50 mM Sodium Phosphate, 2 mM EDTA, 250 mM NaCl, pH 7.0). Contents of an mRNA IVT reaction were diluted with Oligo dT binding buffer and loaded onto a pre-equilibrated column, followed by a wash step of binding buffer and a second low-salt wash of wash buffer (50 mM Sodium Phosphate, 2 mM EDTA), followed by elution with water for injection and peak fractions being collected. Fractions with purified mRNA were combined and concentrated down (ultrafiltration) to approximately 1 mg/mL using an Amicon™ Ultra-15 Centrifugal Filter Units. The total volume was measured by serological pipette and Sodium Citrate Buffer, and pH 6.5 was added to a final concentration of 1 mM. Purified and concentrated mRNA was filter sterilized using Sartorius minisart 0.22 μm PES filters. The size and integrity of mRNA in-process (IPC) samples during purification were assessed by automated gel electrophoresis on an Agilent TapeStation system. The final purified mRNA drug substance (concentrated and filter sterilized) was analyzed by Capillary Gel Electrophoresis on a Fragment Analyser 5300 (Agilent, Santa Clara, CA, USA) using an HS RNA Kit (15NT) 500 (Agilent, Santa Clara, CA, USA) and following manufacturer’s instructions. mRNA was quantified by UV absorbance measurements conducted with a NanoPhotometer NP80.

2.4. Western Blotting

HEK293 Lenti-X cells (10^6^ cells per 6 well plate) were transfected a day after seeding with the vaccine RNA species using Opti-MEM and Lipofectamine™ MessengerMax (Invitrogen, Woolloongabba, Australia) a transfection reagent according to manufacturer’s protocol. After 24 h, cells’ supernatants were harvested. Cells were washed twice in cold PBS and were scrapped into RIPA lysis buffer (Thermo Fisher, Woolloongabba, Australia). Cellular debris was removed from supernatants by 20 min centrifugation at 13,000× g at 4 °C. Proteins (and any viral particles [33]) were precipitated from supernatants by mixing 1 + 3 v/v of 40% PEG 6000, followed by slow rotation overnight at 4 °C. Protein precipitates were pelleted by centrifugation at 12,000 rpm (Beckman Coulter, Avanti J-26 XPI, Beckman Coulter, Inc., Brea, CA, USA) 3000 g at 4 °C for 1 h. The pellets were suspended in RIPA buffer lysis. Protein determination was undertaken using a Pierce BCA Protein Assay Kit (Thermo Fisher, Woolloongabba, Australia). Cells and precipitates from supernatants were run on SDS PAGE with 35 µg loading per well. Western blotting for the envelope was then undertaken using 4G2, a pan-orthoflavivirus anti-E monoclonal antibody [11,34]. An HRP labeled goat anti-mouse secondary (Dako, Glostrup, Denmark) was used, with the SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Fisher, Woolloongabba, Australia). Blots were scanned using the ChemiDoc Touch Imaging System (v6.1) (Biorad, Hercules, CA, USA). Western blotting for GAPDH was undertaken using an anti-GAPDH monoclonal antibody (MA5-15738) (Thermo Fisher, Woolloongabba, Australia).

2.5. Encapsulation of RNA into Lipid Nanoparticles

The following lipids were used for the generation of lipid nanoparticles (LNP) as described [35]. Heptadecan-9-yl 8-((2-hydroxyethyl)(6-oxo-6-(undecyloxy)hexyl)amino)octanoate (SM102) was purchased from Cayman chemicals (Ann Arbor, MI, USA; Cat. No 33474). 1,2-distearoyl-snglycero-3-phosphocholine (DSPC), cholesterol (Chol) and 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMGPEG2000) were purchased from Avanti Polar Lipids (Alabaster, AL, USA, Cat. No. 850365P-1G; 700100P-50MG; 880151P-1G).

Lipids were prepared at a molar ratio of 50:10:38.5:1.5 (SM102, DSPC, Chol, DMGPEG2000). Lipids in ethanol were mixed with RNA in an aqueous phase at a N/P ratio (positively charged amine groups/negatively charged phosphate groups) of 3:1 using the NanoAssemblr Benchtop machine (Precision NanoSystems; Vancouver, BC, Canada), at a flow rate of 12 mL/min. The LNP mixture was immediately diluted to 1:4 in 20 mM Tris-hydrochloride (pH = 7) and 8.7% sucrose solution (Tris/sucrose) before further dilution to 15 mL in Tris/sucrose and concentration using 10 kDa Amicon Ultra-15 filter columns (Millipore; Burlington, MA, USA) at 3000× g for 60 min at room temperature. The dilution and concentration steps were repeated with centrifugation for 120 min. Concentrated LNPs were stored at 4 °C.

The LNP size and polydispersity index were measured using the Zetasizer Nano ZS (Malvern Instruments, Malvern, UK). Encapsulation efficiency was measured using the Ribogreen assay (Thermo Fisher, Cat No. R11490) and was >85%.

2.6. Mice and Vaccination

Ifnar-/- mice were originally kindly provided by Dr P. Hertzog (Monash University, Melbourne, Australia) and were bred in-house at QIMR Berghofer. The mice are on a C57BL/6J background and do not express Ifnar1 [36,37]. Standard housing conditions (22 ± 1 °C) and environmental enrichments were used as detailed previously [38].

Female Ifnar-/- mice were allocated into groups of 5 or 6 mice per group, such that the mean age per group was similar (range of mean ages was 19.2 to 19.7 weeks per group), and the age distribution of mice within each group was also similar (range of mouse ages was 13–15 to 22–24 weeks).

Mice were vaccinated intramuscularly (i.m.) with a total of 1 or 5 µg of mRNA vaccine divided equally into both quadriceps muscles in a volume of 50 µL per muscle. Control mice received empty LNP at the same dose, site, and volume (LNP control).

Imojev is a live-attenuated chimeric vaccine strain wherein the prME genes of the Yellow Fever 17D virus were replaced with those from JEV SA14-14-2 (a genotype III virus). Imojev was propagated using Vero cells with supernatant used to vaccinate mice. Mice were given a dose of 200 CCID_50_/mouse in 50 µL, i.m. into quadriceps muscles.

2.7. Antibody Response Assessments

ELISA assays. Total IgG responses were determined by standard ELISA developed in-house using whole JEV_NSW2022_ as an antigen. In brief, JEV_NSW2022_ viral particles were concentrated from infected C6/36 cell supernatants by polyethylene glycol precipitation (PEG6000; Sigma-Aldrich; North Ryde, Australia). The supernatant was mixed 1 + 3 v/v of 40% PEG 6000, incubated overnight at 4 °C with slow rotation, and the precipitate pelleted by centrifugation at 12,000 rpm (Beckman Coulter, Avanti J-26 XPI) for 1 h at 4 °C. The pellet was dissolved and diluted ≈1:100 in ELISA Carbonate Coating Buffer (Thermo Fisher, Woolloongabba, Australia), which was then used to coat MaxiSorp Immuno 96-well plates (Thermo Fisher, Woolloongabba, Australia). Mouse serum samples, starting at a 1:30 dilution, were serially diluted 2-fold in duplicate. Bound serum IgG was detected using biotin-labeled rat anti-mouse-IgG (Thermo Fisher, Woolloongabba, Australia), streptavidin HRP (Biosource, Camarillo, CA, USA), and ABTS substrate (Sigma-Aldrich). Endpoint titers were interpolated when OD_405_ values reached the mean OD_405_ + 3 standard deviations for naïve serum.

Neutralization assays. Serum neutralization titers were determined as described [13,18]. In brief, mouse serum samples were heat-inactivated (56 °C for 30 min) and incubated in duplicate with 500 CCID_50_ of JEV_NSW2022_ or 100 CCID_50_ of JEV_Nakayama_ at 37 °C for 1 h before Vero E6 cells were added (10^5^ cells/well of a flat bottomed 96 well plate). The initial serum dilution was 1:10 with 2-fold serial dilutions in duplicate. After 7 days, cells were fixed and stained with formaldehyde and crystal violet [39] and the 50% neutralizing titers interpolated from optical density (OD_590_) versus serum dilution plots as described [13,18]. The 500 CCID_50_ dose for of JEV_NSW2022_ is relatively high compared with other orthoflaviviruses [40] but was required to generate clearly measurable cytopathic effects similar to those seen when using 100 CCID_50_ of JEV_Nakayama_.

2.8. JEVNSW2022 Challenge and Monitoring

Mice were challenged as described [13]. In brief, mice received 5 × 10^3^ CCID_50_ in 100 μL of JEV_NSW2022_ subcutaneously (s.c.) in the base of the tail. Mice were monitored using a score card system for signs and symptoms of disease [38] and were weighed daily; mice reaching clinical defined endpoints were euthanized using carbon dioxide as described [13,38]. Serum and tissue titers were determined by CCID_50_ assays as described [13,18]. In detail, C6/36 cells were plated in 96-well flat-bottom plates at 2 × 10^4^ cells per well in 100 μL of medium. For tissue titrations, tissues were homogenized in tubes, with each containing 4 ceramic beads twice at 6000× g for 15 s, followed by centrifugation twice at 21,000× g for 5 min. Samples underwent 10-fold serial dilutions in 100 μL RPMI 1640 supplemented with 2% FBS, performed in duplicate. The serially diluted samples (100 µL) were then added to each well of 96-well plates in duplicate containing C6/36 cells, and the plates were cultured for 5 days at 37 °C and 5% CO_2_. Thereafter, 25 μL of supernatant from the C6/36-containing wells were transferred (well by well) to parallel Vero E6 cells plated the day before at 2 × 10^4^ cells per well in 96-well flat-bottom plates. Vero E6 cells were cultured for 5 days and cytopathic effects were scored, and the virus titer was calculated using the method of Spearman and Karber.

2.9. Statistics

The t-test was used when the differences in variances between groups was <4 fold. Otherwise, the non-parametric Mann–Whitney U exact tests or the Kolmogorov–Smirnov exact test was applied (GraphPad Prism 10; Boston, MA, USA).

3. Results

3.1. Design of JEVNSW2022 GIV mRNA Vaccines

mRNA vaccines for orthoflaviviruses generally comprise the prME structural gene cassette and include a signal peptide sequence that directs this polyprotein into the endoplasmic reticulum (ER), thereby trafficking it through the secretory pathway and the associated post-translational processing. The signal peptide is cleaved from prME by host-cell signal peptidases; however, this process can be inefficient [41]. A range of different signal peptides has previously been evaluated seeking to identify sequences that optimize secretion and immunogenicity of specific orthoflavivirus vaccines [4,5,7,9,42].

Herein, we evaluated two signal peptide sequences, the full-length 23 amino acid (a.a.) Native signal peptide sequence from JEV_NSW2022_, and a Shorter 16 a.a. sequence from JEV_NSW2022,_ with the latter based on a JEV signal peptide sequence used for a ZIKV mRNA vaccine (Moderna, mRNA-1893) [42,43] (Figure 1a and Figure S1a,b). Signal peptide sequences can be viewed as having an N-terminal n-region, a central hydrophobic membrane spanning h-region, and a C-terminal sequence that helps direct cleavage by the cellular signal peptidase (Figure 1a and Figure S1b) [41,44]. The Shorter signal peptide could thus be viewed as missing the n-region (GGNGGT). Capsid is not present in these prME mRNA constructs. However, in the viral polyprotein capsid would be located upstream of the signal peptide (Figure S1c). During virus infection, cleavage at the N-terminal end of the signal peptide (to release capsid) would be mediated by the NS2B/NS3 viral protease [45,46] (Figure S1c). A codon-optimization approach was not employed for JEV GIV prME, with the native RNA sequence of JEV_NSW2022_ prME retained in the mRNA vaccines (Figure 1a and Figure S1a). A segmented poly(A) tail was used (Figure 1a), as this has been shown to significantly reduce plasmid recombination in E. coli, without compromising mRNA stability and translation [47]. Human beta globin (HBB) gene 5′ and 3′ UTRs were employed (Figure 1a and Figure S1a) and are well-described for mRNA therapeutics and vaccines [48].

3.2. Generation of the JEVNSW2022 GIV mRNA Vaccines

The generation of purified vaccine mRNA species was undertaken at Southern RNA (see Materials and Methods). The final vaccine mRNA products were analyzed by capillary gel electrophoresis, illustrating a sharp peak at the expected size (Figure 1b), with an estimated purity of >90%.

The vaccine mRNA species were used to transfect HEK293 cells, and after 24 h cells supernatants were harvested. Proteins precipitated from the supernatants and cells were analyzed by Western blotting using 4G2, a pan-orthoflavivirus anti-envelope (E) monoclonal antibody. The molecular weight of the JEV_NSW2022_ E protein is ≈50 kDa [6,11], with a ≈50 kDa band present in supernatant and cells transfected with the Shorter RNA construct (Figure 1c, left). Supernatants from cells transfected with the Native RNA construct also provided a ≈50 kDa band, although this band was not detected in the cells (Figure 1c, left). A higher molecular weight band at ≈60 kDa was present in all samples, and this was the only band detected in cells transfected with the Native RNA construct (Figure 1c, left). This band likely represents incomplete proteolytic processing of the prME protein.

The Shorter RNA construct would appear to generate higher levels of E in the cells, as assessed by Western blotting (Figure 1c, left, red arrow). In support of this contention, SDS PAGE and Coomassie blue protein staining (Figure 1c, middle), and Western blotting for GAPDH (housekeeping protein) (Figure 1c, right), indicated that loading was actually slightly lower for the Shorter RNA samples. Levels of envelope in the supernatants showed no overt differences for the two constructs (Figure 1c, Supernatant). The results (Figure 1c) illustrated that after transfection with the mRNA constructs, vaccine antigens were translated, processed, and secreted.

The mRNA species were subsequently encapsulated in lipid nanoparticles (LNP), formulated as for Moderna’s Spikevax^®^ (COVID-19 Vaccine). The LNPs had a mean diameter of ≈150 nm (Figure 1d), a size within the optimal range for LNP vaccines [49]. The polydispersity indices (PdI) were 0.167–0.187 (Figure 1d). The PdI is a measure of the uniformity of particle-size distribution, with values below 0.3 considered acceptable.

3.3. Antibody Responses in Ifnar-/- Mice After Delivery of JEV GIV mRNA Vaccines

Mice were vaccinated with two doses of the Shorter or the Native LNP vaccines at 1 µg or 5 µg RNA doses, with serum taken at 3 and 8 weeks and antibody titers determined by ELISA. ELISA titers were detected in all mice that received the JEV mRNA vaccines. A significant dose effect was seen after two doses for the Shorter vaccine, with 5 µg providing titers ≈3-fold higher than 1 µg (Figure 2b, p = 0.016). The Shorter vaccine also gave ≈3-fold higher titers than the Native after two 5 µg doses (Figure 2b, p = 0.016).

Neutralization titers for JEV_NSW2022_ were detected after two doses of 5 µg of either the Shorter or Native vaccines in four out of five mice (Figure 2c, left). Although the Shorter vaccine induced mean neutralization titers that were 4.4-fold higher than those induced by the Native vaccine, this did not reach significance (Figure 2c, n.s.). Two doses of 5 µg of the Shorter vaccine induced mean titers (reciprocal 50% neutralization titer of 187 ± SD 115) that were 3.5-fold lower than the mean titer seen after the Imojev vaccination (655 ± SD 551) (Figure 2c, p = 0.048); however, a small level of overlap was seen for individual titers (Figure 2c).

Neutralization titers for a GIII isolate, JEV_Nakayama_, were also evaluated for serum samples from mice that had received two doses of 5 µg doses of the mRNA vaccine (Figure 2c, right). Clear cross-neutralization was evident (Figure 2c, right), consistent with the 93.4% amino acid sequence identity for envelope proteins from JEV_NSW2022_ versus JEV_Nakayama_ [18].

3.4. Weight Loss and Survival for Vaccinated Mice Post-Challenge

All Ifnar-/- mice lost weight post-challenge, with the mean weight loss reaching ≈3–8% in most groups of mice (Figure 2d) that survived (Figure 2e), before body weight recovery was evident after about day 14 (Figure 2d). Mice that ultimately needed to be euthanized generally showed precipitous weight loss days 3–6 post-challenge (Figure 2d, †).

Kaplan–Meier survival plots illustrated that survival of most or all of the mice required two 5 µg doses of the mRNA vaccines or Imojev vaccination (Figure 2e). Two 1 µg doses of the mRNA vaccines protected only one out of five of the vaccinated animals, with a survivor (1 µg Native) losing ≈ 14% body weight before recovering (Figure 2d). This mouse had neutralizing antibody levels just above the level of detection after the first vaccination (Figure 2c, 1 µg Native, 1st).

3.5. Viremias in Vaccinated Mice Post-Challenge

Viremia was monitored over 5 days post-challenge, with two 1 µg doses of the mRNA vaccines providing a significant mean ≈ 2 log_10_CCID_50_ reduction in viremia on day 2 (Figure 2f, p = 0.008). Two 5 µg doses of the mRNA vaccines provided a large and significant ≈5–6 log reduction in viremia days 2 and 3 (Figure 2f, p = 0.008), although for both Shorter and Native vaccine groups, two out of five mice showed detectable low-level viremias (Figure 2f and Figure S2). No viremia was detected on any day in any mice after the Imojev vaccination (Figure 2f). Virus titers in the spleen and brain largely paralleled the viremia results (Figure S2).

3.6. Disease Manifestations in Vaccinated Mice Post-Challenge

After the challenge, mice were monitored for the three dominant overt disease manifestations seen in this model; loss of normal posture, activity loss, and fur ruffling (Figure 3). The LNP control group rapidly developed loss of normal posture and activity but no fur ruffling. Mortality occurred on day 3 in these animals, with fur ruffling only seen after day 3 in the current setting (Figure 3). Disease was evident in all mice (five out of five) that had received the 1 µg doses of the mRNA vaccines (Figure 3). In mice that had been vaccinated with the 5 µg doses of the mRNA vaccines, mild disease manifestations were seen in one out of five mice (Figure 3). However, in the 5 µg Native group, the one mouse with symptoms also lost up to 14% of its body weight and then recovered (Figure 2d). In the 5 µg Shorter group, the one mouse lost > 20% of its body weight, reaching the criteria for euthanasia (Figure 2d). Imogev-vaccinated animals did not manifest any overt disease manifestations post-challenge. In summary, two doses of 5 µg of the mRNA vaccines protected 80% of mice against overt disease manifestations.

4. Discussion

Herein, we illustrate the construction of JEV GIV mRNA vaccines and show that they can generate protective immunity in Ifnar-/- mice. Neutralizing antibody responses generated by approved live attenuated arboviral vaccines, such as Imojev, provide robust benchmarks for evaluating new mRNA vaccines, given the established abilities of the former to induce effective protective immunity in humans [1]. The neutralizing antibody responses against JEV_NSW2022_ after two vaccinations with 5 µg of the Shorter vaccine were significantly lower but overlapped with those generated by Imojev vaccination (Figure 2c). The overlap may, at least in part, be due to the mRNA vaccine targeting GIV, whereas Imojev is based in a GIII virus [18]. However, a similar result was obtained for a JEV GIII mRNA vaccine with a native signal sequence. After mice received two 15 µg doses of this JEV GIII mRNA vaccine, lower-but-overlapping neutralizing antibody responses were generated, when compared with those obtained after vaccination with another live attenuated GIII JEV vaccine [6]. An mRNA vaccine for chikungunya virus (CHIKV) also provided slightly lower-but-overlapping neutralizing antibody responses in human trials [50] when compared with responses after vaccination with IXCHIQ, a licensed live attenuated CHIKV vaccine [51]; although it should be noted the latter two vaccines were evaluated in separate studies. Overall, our data argues that after two vaccinations, the performance of the Shorter mRNA vaccine approached that of a licensed live attenuated vaccine.

The Shorter vaccine would appear to generate higher levels of E protein (Figure 1c), consistent with the significantly higher ELISA titers induced by this vaccine when compared to the Native vaccine (Figure 2b, p = 0.016). Mean neutralization titers were also 4.4-fold higher, although this did not reach significance (Figure 2c, n.s.). The Shorter vaccine’s signal peptide sequence is missing the n-region (GGNGGT) (Figure 1a and Figure S1b), which, during viral infection, may facilitate cleavage by the viral protease NS2B/NS3 to release capsid (Figure S1c) [46]. One might speculate that the multiple glycine residues may hinder folding (for example, into alpha helical structures), thereby facilitating NS2B/NS3 protease access [52]. Such access is clearly dispensable in the current context as capsid is not present, and NS2B/NS3-mediated cleavage is not required. Conceivably, the overall polar GGNGGT peptide sequence may inhibit insertion of the hydrophobic h-region (VLWLMSLTIAAV) into the lipid membrane [41,44], thereby reducing prME expression from the Native mRNA vaccine.

A codon-optimization approach, widely used for mRNA vaccines [53], was not used in this study, as prME proteins are already efficiently produced in mice, pigs, and humans during JEV infections [12,18,54,55]. The natural hosts of JEV are primarily ardeid wading birds and Culex mosquitoes [56], with codon usage shaped by conflicting evolutionary pressures in avian hosts and mosquito vectors [55,57]. Codon optimization for translation in mammals might thus be expected to improve vaccine performance. However, non-optimal codons have been shown to increase viral fitness for dengue virus [58], with RNA structural elements often important for viral fitness [59]. Optimizing mRNA secondary structure, by enhancing mRNA stability, has emerged as an important factor for mRNA vaccine performance [60,61]. Future endeavors might apply some of the new algorithms [61,62] to JEV mRNA vaccine design and test whether they are able to improve performance. However, algorithms that increase CpG dinucleotide content might be avoided as this may increase sensitivity to ZAP, an anti-viral protein that promotes degradation and inhibits translation. JEV, but not dengue or ZIKV, is sensitive to ZAP due to a CpG-rich region in the 3′ UTR [63], which in the vaccine is replaced by the HBB UTR (Figure 1a). The role of T cells was not evaluated in this study, with CD8 T cells in mouse models of JEV [64] and ZIKV [65] shown to play only a minor role in protection. Neutralizing antibodies are also the key readout in clinical evaluations of orthoflavivirus vaccines encoding prME, with a licensed inactivated JEV vaccine, IXIARO/JESPECT [1], which is also unlikely to induce CD8 T cells efficiently. Nevertheless, anti-NS1 T-cell responses have been shown to be protective against ZIKV in mice [66] and T-cell-based vaccination approaches are being pursued for dengue in human clinical trials, primarily because they would cross react against all four serotypes [67]. That CD8 T cells are induced by a JEV prME mRNA vaccine has been reported previously [6], and would be entirely expected for the mRNA vaccine described herein.

A theoretical concern for orthoflavivirus vaccines is Antibody Dependent Enhancement (ADE), with a number of approaches developed to avoid ADE, for instance, modifying the fusion loop of the JEV envelop protein [7]. However, although ADE can be shown to occur in vitro for many viruses, significant, clinical relevant enhancement of disease or infection, outside of dengue, has yet to emerge [1]. Existing JEV vaccines have also not been associated with ADE problems, despite widespread geographic overlap with other orthoflaviviruses. A study in pigs also illustrated that a JEV vaccine, which induced antibody responses that were strongly enhancing in vitro, nevertheless protected pigs against the challenge [68].

Another limitation of our study is the use of the Ifnar-/- model, which does not recapitulate the infection of brain cells or encephalitis [18]. Despite several attempts, we have been unable to detect clear histopathological alterations or JEV_NSW2022_-infected brain cells in these mice by immunohistochemistry. The likely reason is that the Ifnar-/- mice succumb to infection before brain infection and pathology can manifest. JEV_NSW2022_ replicates poorly in C57BL/6J mice, with a low proportion (≈10%) of mice showing infection of the brain cells [18]. Very large cohorts of C57BL/6J mice would thus be needed before significant protection against brain infection could be demonstrated for this JEV GIV isolate.

Two previous studies of protective JEV mRNA vaccines have been reported, one for GI [7] and one for GIII [6]. Figure S3 provides comparisons for several key features for these two studies and the data reported herein. Zhu et al. 2024 was focused on avoiding ADE of ZIKV, which used envelope rather than the more commonly used prME [1] as the antigen and vaccinated 1 week old C57BL/6 mice [7]. The Chen at al. 2022 study was based on a historically dominant isolate from 1949, and was able to show complete protection in adult C57BL/6 mice against detectable brain immuno- and histo-pathology [6]. Chen at al. 2022 used a full-length Native 24 a.a. signal peptide from JEV P3. A shorter version, used herein (based on Moderna ZIKV vaccine), provided enhanced protein expression when compared to the Native JEV_NSW2022_ signal peptide.

A major hurdle for new vaccines against epidemic arboviral diseases is the unpredictable nature of outbreaks. This often makes it difficult to identify when and where to set up phase III efficacy field trials [1], with such trials representing a major cost on the pathway to approval and licensing. However, in 2023, IXCHIQ was approved via the FDA’s Accelerated Approval pathway without such field trials [1]. Instead, sera taken from vaccine recipients in a phase III trial were adoptively transferred to non-human primates, which were then shown to be protected after challenged with CHIKV [69]. A similar approach was also used for a Ross River virus vaccine, with human sera from a phase I/II trial transferred to mice that were then shown to be protected after a Ross River virus challenge [70]. The latter approach could clearly also be used for JEV mRNA vaccine trials, using mouse models such as the one described herein. Although such adoptive transfers of human sera do not evaluate the contribution of T cells [71,72], they illustrate attainment of neutralizing antibody titers that correlate with protection [1,73].

It remains unclear whether an mRNA vaccine would be able to compete commercially with the existing JEV vaccines, the live attenuated Imojev, and the inactivated IXIARO/JESPECT vaccines [1]. One might argue that mRNA vaccines can be made in-country, reducing the reliance on a supply from overseas, which may become limiting if the supply is compromised or demand is high [73]. Manufacturing would also not require a high biocontainment facility that would ordinarily be required for the culture of live virus, a feature that contributes to the relatively high costs of current JEV vaccines. Ultimately, mRNA vaccines may thus emerge to be cheaper [74]; this would clearly be an advantage given many populations at risk reside in resource-poor settings [73]. In addition, the robust safety data, including safety data for immunocompromised vaccine recipients [75], together with the considerable global experience with mRNA vaccines during the COVID-19 epidemic [76], should also facilitate the entry of new mRNA vaccines into local and regional markets.