Immunogenicity and Protective Ability of Genotype I-Based Recombinant Japanese Encephalitis Virus (JEV) with Attenuation Mutations in E Protein against Genotype V JEV

Genotype V (GV) Japanese encephalitis virus (JEV) has emerged in Korea and China since 2009. Recent findings suggest that current Japanese encephalitis (JE) vaccines may reduce the ability to induce neutralizing antibodies against GV JEV compared to other genotypes. This study sought to produce a novel live attenuated JE vaccine with a high efficacy against GV JEV. Genotype I (GI)-GV intertypic recombinant strain rJEV-EXZ0934-M41 (EXZ0934), in which the E region of the GI Mie/41/2002 strain was replaced with that of GV strain XZ0934, was introduced with the same 10 attenuation substitutions in the E region found in the live attenuated JE vaccine strain SA 14-14-2 to produce a novel mutant virus rJEV-EXZ/SA14142m-M41 (EXZ/SA14142m). In addition, another mutant rJEV-EM41/SA14142m-M41 (EM41/SA14142m), which has the same substitutions in the Mie/41/2002, was also produced. The neuroinvasiveness and neurovirulence of the two mutant viruses were significantly reduced in mice. The mutant viruses induced neutralizing antibodies against GV JEV in mice. The growth of EXZ/SA14142m was lower than that of EM41/SA14142m. In mouse challenge tests, a single inoculation with a high dose of the mutants blocked lethal GV JEV infections; however, the protective efficacy of EXZ/SA14142m was weaker than that of EM41/SA14142m in low-dose inoculations. The lower protection potency of EXZ/SA14142m may be ascribed to the reduced growth ability caused by the attenuation mutations.


Introduction
Japanese encephalitis (JE) is a severe neurological disorder caused by infection with a mosquito-borne arbovirus. Japanese encephalitis virus (JEV) is a critical public health problem in Asian countries. There are an estimated 68,000 cases of JE per year, occurring mainly in China, India, and Southeast Asian countries, resulting in 15,000 fatalities, mostly in children [1][2][3].
JEV belongs to the genus Flavivirus in the family Flaviviridae and is amplified in a bird/pig-mosquito transmission cycle [4]. The mosquitoes also transmit JEV to humans and horses, which are 'dead-end' hosts. JEV has a single-stranded positive-sense RNA genome with a single open reading frame that encodes three structural proteins (C, prM, and E) and seven non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5). The genome also contains non-coding regions (NCRs) at its 5 -and 3 -terminal ends. JEV is classified into five genotypes (GI, GII, GIII, GIV, and GV) based on genome sequence [5,6]. GIII strains were widely distributed and were most frequently identified in JE endemic areas until the 1990s. However, the major genotype has begun to change from the GIII to GI strain since the early 1990s in most JE endemic areas [7][8][9]. Although the reason for the broad shift from GIII to GI remains unclear, current findings suggest that GI

Viruses
The GI JEV strain Mie/41/2002 (GenBank accession No. AB241119), which was isolated from pig serum in Japan in 2002 [38,39], and the GV JEV strain Muar (GenBank accession no. HM59272), which was isolated from a patient with encephalitis in Malaysia in 1952, were used [19]. The working virus stocks were prepared through amplification in Vero cells.

Recombinant Viruses
The GI-GV intertypic virus E XZ0934 was used [19]. In addition, we constructed additional mutant viruses, rJEV-E M41/SA14142m -M41 (E M41/SA14142m ) and rJEV-E XZ/SA14142m -M41 (E XZ/SA14142m ) in the Mie/41/2002 backbone, as described previously [19,21]. Briefly, the E region of the infectious cDNA clone rJEV(Mie/41/2002)/pMW119 was replaced with the in vitro-synthesized E region DNA of Mie/41/2002 and XZ0934 strains with ten amino acid mutations (Eurofins Genomics, Tokyo, Japan) (Figures 1 and S1). Each synthesized DNA fragment was inserted into the corresponding region using conventional molecular cloning methods and the In-Fusion Cloning system (Takara Bio, Shiga, Japan). The nucleotide sequences of the viral genome regions of the recombinant clones were determined after amplification of the plasmids in Escherichia coli STBL2 (Thermo Fisher Scientific, Waltham, MA, USA). Recombinant viruses were recovered by transfecting Vero cells with in vitro-transcribed recombinant viral RNA, as previously described [39]. An aliquot of the culture supernatant of the transfected Vero cells was passaged once in Vero cells, and the culture fluid was used as the recombinant virus solution (v1). The nucleotide sequences of the recombinant viruses were also determined, and no additional nucleotide mutations were detected.

Analysis of Plaque Morphology and Growth Kinetics
Infectious viral titers for each sample were determined using plaque-formation assays. Vero cells (~3 × 10 5 /well) were plated in 12-well culture plates and inoculated with each virus for 1 h at 35-37 • C. Next, a MEM-based overlay medium containing 1% methylcellulose and 2% FBS was added to the wells, and the cells were incubated for 4-5 days at 35-37 • C, after which the cells were fixed using a 10% formalin-PBS solution and stained with methylene blue, as described previously [39].
Ten amino acid substitutions, L107F, E138K, I176V, T177A, E244G, Q264H, K279M, A315V, K439R, and G447D, were introduced into the E region of E XZ0934 to obtain a new recombinant JEV strain, E XZ/SA14142m (Figure 1). Another recombinant virus, E M41/SA14142m , which has the same mutations involving the E region of the GI Mie/41/2002 strain, was also produced to compare growth properties, virulence, and neutralization/protection efficacy against GV JEV with E XZ/SA14142m .  The in vitro growth ability of the JEV strains was analyzed as described previously [39]. Briefly, cells were cultured in 6-well culture plates and infected with each JEV strain in 3 mL MEM supplemented with 2% FBS (2F/MEM) at a multiplicity of infection (MOI) of 0.05 plaque-forming units (PFU)/cell. Small aliquots (200 µL) of the media were collected at one-day intervals, and infectious viral titers were determined using plaque-formation assays in Vero cells, as described above. Infectious titers of the parental and mutant viruses at 3 days post infection were statistically compared using BellCurve for Excel (Social Survey Research Information, Tokyo, Japan), using student's t-test. The statistical significance was set at p < 0.05.

Mouse Challenge Experiment and Sample Collection
Female ddY mice (Japan SLC, Inc., Shizuoka, Japan) were used for mouse challenge tests. For the neuroinvasiveness analysis, groups of mice (3 weeks old, n = 10) were inoculated intraperitoneally (i.p.) with 100 µL (1 × 10 5 PFU) of virus solution diluted in 0.9% NaCl solution. The mice were observed, and the bodyweight of the mice was measured every day for 20 days after inoculation to determine survival rates. Survival curves were compared using BellCurve for Excel and the log-rank (Mantel-Cox) test. Statistical significance was set at p < 0.05. The surviving mice were sacrificed, and their sera were collected for further immunological analyses, as described below. For neurovirulence analysis, groups of mice (4 weeks old, n = 10 or 5) were inoculated intracerebrally (i.c.) with 30 µL (3 × 10 3 PFU) of virus solution, and then the mice were observed for 18 days to determine survival rates, as described above.
For protection ability analysis, groups of mice (3 weeks old, n = 10) were inoculated i.p. with 100 µL (1 × 10 5 , 1 × 10 4 , or 1 × 10 3 PFU) of virus (E M41/SA14142m or E XZ/SA14142m ) solution. In some groups, mice were inoculated again with the recombinant virus solution two weeks after the initial infection. Two or five weeks after the initial inoculation, blood samples of the inoculated mice were collected, and two or three days after collection, the mice were inoculated i.p. with 100 µL (1 × 10 4 PFU) of virus (Muar) solution. The mice Vaccines 2021, 9, 1077 5 of 18 were observed, and the bodyweight of mice was measured every day for 21 days after inoculation to determine survival rates, as described above.
For growth analysis in mice, groups of mice (n = 5) were inoculated i.p. with 100 µL (1 × 10 4 PFU) JEV solution. Serum, brain, and spleen were collected from the mice at two and five days post-infection, and the titer and RNA levels of the infectious virus in the samples were measured, as described below. Tissue weights were determined, and the tissues were homogenized in 500 µL of MEM with 2% FBS, and the homogenate was used for further analyses.

Measurement of Infectious Viral Titer
Viral titers for each sample were determined by plaque-formation assays, as described above, and then statistically compared using BellCurve for Excel, employing the Kruskal-Wallis (Steel-Dwass) test. The statistical significance was set at p < 0.05.

Plaque Reduction Neutralization Test (PRNT)
Neutralizing antibodies against JEV were measured by the PRNT method. Each JEV strain was combined at a 1:1 ratio with 2-fold serial dilutions (1:10 to 1:10,240) of sera from mice infected with JEV and then incubated at 35 • C for 90 min. Vero cell monolayers were inoculated with these mixtures in 12-well plates and incubated at 35 • C for 90 min. Subsequently, overlay medium containing 1% methylcellulose was added, and cells were incubated at 35 • C for 4-5 days. The cells were fixed using a 10% formalin-PBS solution and stained with methylene blue. The PRNT titer (PRNT 50 ) was defined as the reciprocal of the highest dilution, resulting in a 50% reduction relative to the mouse serum-free control. PRNT 50 was statistically compared using BellCurve for Excel, employing the Kruskal-Wallis (Steel-Dwass) test. The statistical significance was set at p < 0.05.

Indirect Immunofluorescence Assay (IFA)
Vero cells were infected with the Muar strain at an MOI of 5 and incubated for 24 h. The cells were suspended, and cell smears were prepared on 10-well glass slides and then fixed for 5 min in acetone at room temperature. The cells were exposed to JEVinfected mouse sera, followed by Alexa Fluor 488 goat anti-mouse IgG antibody (Thermo Fisher Scientific, Waltham, MA, USA). The stained cells were visualized by fluorescence microscopy (BZ-X810; Keyence Corp., Osaka, Japan).

Production of Recombinant JEV Strains with Attenuating E Gene Mutations
Ten amino acid substitutions, L107F, E138K, I176V, T177A, E244G, Q264H, K279M, A315V, K439R, and G447D, were introduced into the E region of E XZ0934 to obtain a new recombinant JEV strain, E XZ/SA14142m (Figure 1). Another recombinant virus, E M41/SA14142m , which has the same mutations involving the E region of the GI Mie/41/2002 strain, was also produced to compare growth properties, virulence, and neutralization/protection efficacy against GV JEV with E XZ/SA14142m .

Growth of E XZ/SA14142m and E M41/SA14142m Strains In Vitro
In Vero cells, plaques formed by E M41/SA14142m were slightly smaller than those formed by the parental Mie/41/2002, while plaques formed by E XZ/SA14142m were considerably smaller than those formed by the parental E XZ0934 (Figure 2A). The growth kinetics of E M41/SA14142m was slightly slower than that of the parental Mie/41/2002, whereas E XZ/SA14142m grew obviously slower than the other three strains in Vero cells ( Figure 2B). The growth rate of E M41/SA14142m was lower than that of Mie/41/2002 and E XZ0934 , but E XZ/SA14142m rarely propagated in mouse neuroblastoma Neuro-2a cells ( Figure 2C) and human neuroblastoma IMR-32 cells ( Figure 2D). These results suggested that the engineered amino acid substitutions decreased the in vitro growth properties of Mie/41/2002 and E XZ0934 and that the mutations strongly influenced the growth of E XZ0934 .

Growth of E XZ/SA14142m and E M41/SA14142m Strains In Vitro
In Vero cells, plaques formed by E M41/SA14142m were slightly smaller than those formed by the parental Mie/41/2002, while plaques formed by E XZ/SA14142m were considerably smaller than those formed by the parental E XZ0934 (Figure 2A). The growth kinetics of E M41/SA14142m was slightly slower than that of the parental Mie/41/2002, whereas E XZ/SA14142m grew obviously slower than the other three strains in Vero cells ( Figure 2B). The growth rate of E M41/SA14142m was lower than that of Mie/41/2002 and E XZ0934 , but E XZ/SA14142m rarely propagated in mouse neuroblastoma Neuro-2a cells ( Figure 2C) and human neuroblastoma IMR-32 cells ( Figure 2D). These results suggested that the engineered amino acid substitutions decreased the in vitro growth properties of Mie/41/2002 and E XZ0934 and that the mutations strongly influenced the growth of E XZ0934 .

Pathogenicity of E XZ/SA14142m and E M41/SA14142m Strains in Mice
Mice were inoculated i.p. with Mie/41/2002, E M41/SA14142m , E XZ0934 , and E XZ/SA14142m and observed for 20 days ( Figure 3A). Two of ten mice inoculated with Mie/41/2002, and seven of ten mice inoculated with E XZ0934 died. None of the mice inoculated with E M41/SA14142m, or E XZ/SA14142m died. An evident loss in body weight was observed in some Mie/41/2002-and Values: means ± standard deviation from three independent inoculations. Significance (parental virus vs. mutant virus, day 3) was analyzed using the student's t-test (** p < 0.01).

Pathogenicity of E XZ/SA14142m and E M41/SA14142m Strains in Mice
Mice were inoculated i.p. with Mie/41/2002, E M41/SA14142m , E XZ0934 , and E XZ/SA14142m and observed for 20 days ( Figure 3A). Two of ten mice inoculated with Mie/41/2002, and seven of ten mice inoculated with E XZ0934 died. None of the mice inoculated with E M41/SA14142m, or E XZ/SA14142m died. An evident loss in body weight was observed in some Mie/41/2002-and E XZ0934 -inoculated surviving mice, but this was not observed for E M41/SA14142m -or E XZ/SA14142m -inoculated mice ( Figure S2). Mice were inoculated i.c. with the four strains and observed for 18 days ( Figure 3B). All ten mice inoculated with Mie/41/2002 and E XZ0934 died by day six after inoculation, whereas all ten mice in the E M41/SA14142m -and E XZ/SA14142m -inoculated groups survived. These results indicated that both the neuroinvasiveness and neurovirulence of Mie/41/2002 and E XZ0934 strains were significantly reduced by engineered amino acid substitutions involving the E protein.
Vaccines 2021, 9, x FOR PEER REVIEW 7 of 18 E XZ0934 -inoculated surviving mice, but this was not observed for E M41/SA14142m -or E XZ/SA14142minoculated mice ( Figure S2). Mice were inoculated i.c. with the four strains and observed for 18 days ( Figure 3B). All ten mice inoculated with Mie/41/2002 and E XZ0934 died by day six after inoculation, whereas all ten mice in the E M41/SA14142m -and E XZ/SA14142m -inoculated groups survived. These results indicated that both the neuroinvasiveness and neurovirulence of Mie/41/2002 and E XZ0934 strains were significantly reduced by engineered amino acid substitutions involving the E protein.

Neutralizing Ability of Sera of mice Inoculated with E XZ/SA14142m and E M41/SA14142m Strains against JEV Strains
Serum samples were collected from the surviving mice at 20 days post-inoculation in the neuroinvasiveness experiment described in Figure 3A. The samples were provided for PRNT assays against JEV GV Muar, GI Mie/41/2002, recombinant E XZ0934 , E XZ/SA14142m , and E M41/SA14142m (Figure 4 and Table 1). Sera from mice inoculated with E XZ0934 showed higher PRNT50 (1:320-640) against the Muar strain, and sera from Mie/41/2002-inoculated mice also exhibited titers between 1:40 and 1:320 ( Figure 4A). PRNT50 in the E XZ/SA14142m -inoculated mice group resembled that in the E M41/SA14142m -inoculated group, but the titer distributions of the two groups were significantly lower than those of Mie/41/2002-and E XZ0934inoculated groups. Similar results were also observed when E XZ0934 was used as the challenge virus for neutralization assays (Table 1)

Neutralizing Ability of Sera of Mice Inoculated with E XZ/SA14142m and E M41/SA14142m Strains against JEV Strains
Serum samples were collected from the surviving mice at 20 days post-inoculation in the neuroinvasiveness experiment described in Figure 3A. The samples were provided for PRNT assays against JEV GV Muar, GI Mie/41/2002, recombinant E XZ0934 , E XZ/SA14142m , and E M41/SA14142m (Figure 4 and Table 1). Sera from mice inoculated with E XZ0934 showed higher PRNT 50 (1:320-640) against the Muar strain, and sera from Mie/41/2002-inoculated mice also exhibited titers between 1:40 and 1:320 ( Figure 4A). PRNT 50 in the E XZ/SA14142minoculated mice group resembled that in the E M41/SA14142m -inoculated group, but the titer distributions of the two groups were significantly lower than those of Mie/41/2002-and E XZ0934 -inoculated groups. Similar results were also observed when E XZ0934 was used as the challenge virus for neutralization assays (Table 1). PRNT 50 against Mie/41/2002 was significantly higher in the Mie/41/2002-inoculated group (1:160-1280) ( Figure 4B). Sera from E M41/SA14142m -and E XZ0934 -inoculated mice exhibited titers between 1:40 and 1:160, but no obvious neutralizing activity was detected in E XZ/SA14142m -inoculated mice sera (<1:10) against Mie/41/2002. The neutralizing ability of the sera from E M41/SA14142m -and E XZ/SA14142m -inoculated mice against E M41/SA14142m and E XZ/SA14142m strains, respectively, were also examined, and the titers in the E M41/SA14142m group (1:80-1280) were higher than  (Table 1). Thus, our data indicated that E XZ/SA14142m and E M41/SA14142m could induce neutralizing antibodies against GV JEV in mice. In contrast, the induction ability of E XZ/SA14142m was slightly lower than that of E M41/SA14142m . E M41/SA14142m -and E XZ0934 -inoculated mice exhibited titers between 1:40 and 1:160, but no obvious neutralizing activity was detected in E XZ/SA14142m -inoculated mice sera (<1:10) against Mie/41/2002. The neutralizing ability of the sera from E M41/SA14142m -and E XZ/SA14142m -inoculated mice against E M41/SA14142m and E XZ/SA14142m strains, respectively, were also examined, and the titers in the E M41/SA14142m group (1:80-1280) were higher than those in the E XZ/SA14142minoculated group (<1:10-1:160) (Table 1). Thus, our data indicated that E XZ/SA14142m and E M41/SA14142m could induce neutralizing antibodies against GV JEV in mice. In contrast, the induction ability of E XZ/SA14142m was slightly lower than that of E M41/SA14142m .

Growth of E XZ/SA14142m and E M41/SA14142m Strains in Mice
Infectious virus and viral RNA levels in mice inoculated with the recombinant strains by i.p. were investigated ( Figure 5). Two days after inoculation, no infectious virus was seen in the brain in any four groups, whereas infectious viruses were detected in serum and spleen samples in most groups ( Figure 5A). High levels of viremia were observed in sera from mice inoculated with the Mie/41/2002 and E XZ0934 strains. Viremia was also observed in four of five E M41/SA14142m -inoculated mice and one of five E XZ/SA14142minoculated mice. Infectious viruses were detected in spleen samples of some mice inoculated with Mie/41/2002, E XZ0934 , and E M41/SA14142m strains, but not in those inoculated with E XZ/SA14142m . Levels of viral RNA in serum and brain samples were also significantly higher in Mie/41/2002-and E XZ0934 -inoculated mice than in E M41/SA14142m -and E XZ/SA14142m -inoculated animals ( Figure 5B). Five days after inoculation, infectious viruses were not seen in most serum and spleen samples in all four groups. However, a high titer of infectious virus was detected in the brain in two of five Mie/41/2002-and three E XZ0934 -inoculated groups ( Figure 5C). No infectious virus was detected in the brains of all five E XZ/SA14142m -infected mice. Levels of viral RNA in the brain and spleen samples were also significantly higher in the Mie/41/2002-and E XZ0934 -inoculated groups than in E M41/SA14142m -and E XZ/SA14142m -inoculated animals ( Figure 5D). These data indicated that the mutant viruses E M41/SA14142m and E XZ/SA14142m had a lower growth ability than the parental Mie/41/2002 and E XZ0934 viruses in mice and E XZ/SA14142m exhibited the lowest replication capacity of the four strains. , or E XZ/SA14142m (E_XZ/SA14142m, n = 5) were euthanized at two or five days after inoculation, and serum, brain, and spleen samples were collected. Sera and tissue homogenates were used to quantify the infectious virus titer (PFU/mL or g) (A,C) and viral genome (genome copies/mL or g) (B,D). Dotted line: detection limit. Significance was analyzed using the Kruskal-Wallis test (* 0.01 < p < 0.05).

Protective Efficacy of E XZ/SA14142m and E M41.SA14142m Strains against GV Muar in Mice
Our results indicated that the mutant viruses E M41/SA14142m and E XZ/SA14142m display little or no virulence in mice, implying that these strains can be used as live attenuated vaccines against JEV infections. To investigate this possibility, mice inoculated once or twice with 1 × 10 5 PFU of E M41/SA14142m and E XZ/SA14142m were inoculated i.p. with the GV Muar strain ( Figure 6A). Before the Muar strain challenge, blood samples were collected from the mice, and PRNT50 against Muar was measured ( Figure 6B and Table 2). Neutralizing antibodies against Muar were induced in most mice inoculated with the strains. However, PRNT50 was significantly higher in mice inoculated twice with E XZ/SA14142m . No obvious differences were observed in terms of neutralizing titers among the other three groups. No neutralization abilities (>1:10) were detected in three mouse sera, but antibodies against Muar were detected in the sera of the mice by IFAs ( Table 2). All mice immunized with , or E XZ/SA14142m (E_XZ/SA14142m, n = 5) were euthanized at two or five days after inoculation, and serum, brain, and spleen samples were collected. Sera and tissue homogenates were used to quantify the infectious virus titer (PFU/mL or g) (A,C) and viral genome (genome copies/mL or g) (B,D). Dotted line: detection limit. Significance was analyzed using the Kruskal-Wallis test (* 0.01 < p < 0.05).

Protective Efficacy of E XZ/SA14142m and E M41.SA14142m Strains against GV Muar in Mice
Our results indicated that the mutant viruses E M41/SA14142m and E XZ/SA14142m display little or no virulence in mice, implying that these strains can be used as live attenuated vaccines against JEV infections. To investigate this possibility, mice inoculated once or twice with 1 × 10 5 PFU of E M41/SA14142m and E XZ/SA14142m were inoculated i.p. with the GV Muar strain ( Figure 6A). Before the Muar strain challenge, blood samples were collected from the mice, and PRNT 50 against Muar was measured ( Figure 6B and Table 2). Neutralizing antibodies against Muar were induced in most mice inoculated with the strains. However, PRNT 50 was significantly higher in mice inoculated twice with E XZ/SA14142m . No obvious differences were observed in terms of neutralizing titers among the other three groups. No neutralization abilities (>1:10) were detected in three mouse sera, but antibodies against Muar were detected in the sera of the mice by IFAs ( Table 2). All mice immunized with one or two doses of E M41/SA14142m or E XZSA14142m survived, while nine of ten mock-immunized mice died by day 16 after Muar challenge ( Figure 6C). No evident loss in body weight was observed in the mutant-inoculated mouse groups ( Figure S3).  Figure 6C). No evident loss in body weight was observed in the mutant-inoculated mouse groups ( Figure S3).   For single-dose immunization groups, mice were inoculated i.p. with 1 × 10 5 PFU of E M41/SA14142m (n = 10) or E XZ/SA14142m (n = 10), or mock-inoculated (n = 10). Double-dose immunization group mice were inoculated i.p. with 1 × 10 5 PFU of E M41/SA14142m (n = 10) or E XZ/SA14142m (n = 10) and were immunized again two weeks after the first immunization. At five weeks after the initial immunization, mice were bled to determine serum PRNT 50 and then inoculated i.p. with 1 × 10 4 PFU of Muar strain and observed for 20 days. (B) Neutralization activity of serum taken from immunized mice against GV Muar. Dotted line: detection limit. Significance was analyzed using the Kruskal-Wallis test (* 0.01 < p < 0.05); NS: not statistically significant. (C) Survival curve of immunized mice after challenge with GV Muar strain. Significant p values by log-rank test are also indicated. 1 Serum was recovered from mice 5 weeks after the initial inoculation of viruses and then used for PRNT against GV Muar strain and for IFA to stain Muar-infected Vero cells. 2 +, positive; −, negative; ±, uncertain.
Next, mice were inoculated with lower doses (1 × 10 3 and 1 × 10 4 PFU) of E M41/SA14142m or E XZ/SA14142m , and at 18 days after immunization, the mice were challenged i.p. with Muar ( Figure 7A). One and two of ten mice immunized with 1 × 10 3 and 1 × 10 4 PFU of E M41/SA14142m , respectively, died, whereas eight and two of ten mice immunized with 1 × 10 3 and 1 × 10 4 PFU of E XZ/SA14142m , respectively, died ( Figure 7B). Seven of ten mock-immunized mice died. Some mice with rapid and severe body weight loss survived, regardless of the virus used for immunization ( Figure S4). This result sug-gested that the protective potential of E XZ/SA14142m against Muar may be weaker than that of E M41/SA14142m .
x FOR PEER REVIEW 13 of 18 of the virus used for immunization ( Figure S4). This result suggested that the protective potential of E XZ/SA14142m against Muar may be weaker than that of E M41/SA14142m .

Discussion
In this study, we sought to develop a recombinant live attenuated JE vaccine that is effective against GV JEV infection using a reverse genetics system. In our previous study, we generated a recombinant GI-GV intertypic JEV E XZ0934 , encoding the E protein of a recent GV isolate, XZ0934, using the GI strain Mie/41/2002 backbone [19]. We used E XZ0934 to generate a novel mutant virus E XZ/SA14142m . Ten amino acid substitutions related to the attenuation of the GIII JEV strain SA 14 were introduced in the E region of E XZ/SA14142m . In addition, we also produced a recombinant mutant virus, E M41/SA14142m , that was introduced to the same mutations in the E region of GI Mie/41/2002. The in vitro and in vivo properties of the mutant viruses were examined, and their ability to serve as a live attenuated vaccine against the GV JEV infection was evaluated. Previously, we generated a recombinant intertypic virus, rJEV-5NCME XZ0934 -M41, in which the 5′NCR-C-prM-E region was replaced with that of XZ0934 in the Mie/41/2002 backbone [21]. However, the prM protein is also involved in the high virulence of GV JEV; therefore, E XZ0934 was selected for this study.
Previous reports have shown that the introduction of multiple mutations found in the E region of SA 14-14-2 reduces the proliferative capacity of GI and GIII JEV [33][34][35]41]. Although E M41/SA14142m and E XZ/SA14142m also exhibited reduced growth capability compared with the parental viruses in Vero cells, the mutants maintained a peak infectious titer of approximately 10 7 PFU/mL ( Figure 2B), indicating that viral replication was not critically impaired. However, the mutants presented marked decreases in proliferative ability in mouse neuroblastoma Neuro-2a and human neuroblastoma IMR-32 cells (Figure 2C,D).

Discussion
In this study, we sought to develop a recombinant live attenuated JE vaccine that is effective against GV JEV infection using a reverse genetics system. In our previous study, we generated a recombinant GI-GV intertypic JEV E XZ0934 , encoding the E protein of a recent GV isolate, XZ0934, using the GI strain Mie/41/2002 backbone [19]. We used E XZ0934 to generate a novel mutant virus E XZ/SA14142m . Ten amino acid substitutions related to the attenuation of the GIII JEV strain SA 14 were introduced in the E region of E XZ/SA14142m . In addition, we also produced a recombinant mutant virus, E M41/SA14142m , that was introduced to the same mutations in the E region of GI Mie/41/2002. The in vitro and in vivo properties of the mutant viruses were examined, and their ability to serve as a live attenuated vaccine against the GV JEV infection was evaluated. Previously, we generated a recombinant intertypic virus, rJEV-5NCME XZ0934 -M41, in which the 5 NCR-C-prM-E region was replaced with that of XZ0934 in the Mie/41/2002 backbone [21]. However, the prM protein is also involved in the high virulence of GV JEV; therefore, E XZ0934 was selected for this study.
Previous reports have shown that the introduction of multiple mutations found in the E region of SA 14-14-2 reduces the proliferative capacity of GI and GIII JEV [33][34][35]41]. Although E M41/SA14142m and E XZ/SA14142m also exhibited reduced growth capability compared with the parental viruses in Vero cells, the mutants maintained a peak infectious titer of approximately 10 7 PFU/mL ( Figure 2B), indicating that viral replication was not critically impaired. However, the mutants presented marked decreases in proliferative ability in mouse neuroblastoma Neuro-2a and human neuroblastoma IMR-32 cells (Figure 2C,D).
Moreover, no evident increase in infectious virus levels was observed in the culture supernatant of Neuro-2a cells inoculated with E XZ/SA14142m ( Figure 2C). The growth rate of E XZ0934 was also lower than that of Mie/41/2002, and these intertypic viruses may be more sensitive to additional mutations than Mie/41/2002 in virus replication.
The mutants E XZ/SA14142m and E M41/SA14142m showed lower neuroinvasiveness than the parental strains in mice ( Figure 3A). Infectious virus levels and genomic RNA levels in peripheral tissues and the brain were also lower in mice infected with the mutant viruses than in those infected with the parental viruses ( Figure 5), suggesting that the low neuroinvasiveness of the mutants is due to the low proliferative potential of the mutants. In the neurovirulence analysis, all mice inoculated with the parental strains died within 6 days. In contrast, all mice inoculated with the mutant viruses survived until the end of the observation period without any symptoms ( Figure 3B). This result indicated that the mutations substantially reduced the replicative ability and cytotoxicity of the parental viruses in central neurons.
Our results demonstrated that a single inoculation with 1 × 10 5 PFU of E XZ/SA14142m or E M41/SA14142m could block lethal GV Muar infection in mice ( Figure 6). It is generally accepted that neutralizing antibodies with a PRNT 50 of 1:10 or greater are sufficient to protect against the onset of JE [42]. However, the PRNT 50 of sera from three mice inoculated with 1 × 10 5 PFU of the mutant viruses against Muar were lower than 1:10 ( Table 2). All mutant virus-inoculated mouse sera, including <1:10 of PRNT 50 against Muar, were positive for antibodies against Muar in IFA ( Table 2), indicating that all of the mice were immunized with the mutant viruses. Cellular immunity to non-structural proteins of JEV may also be induced in mutant virus-inoculated mice and have a suppressive effect on the pathogenesis caused by Muar infection [43][44][45]. Twenty percent of mice inoculated with 1 × 10 4 PFU E M41/SA14142m in vitro and in of E M41/SA14142m or E XZ/SA14142m died; however, 10% of mice in the E M41/SA14142m group and 80% of mice in E XZ/SA14142m group died when 1 × 10 3 PFU of the mutant viruses were inoculated ( Figure 7B). As mentioned above, the growth ability of E XZ/SA14142m was lower than that of vivo, suggesting that E XZ/SA14142m does not amplify sufficiently to induce humoral and cellular immunity in mice infected with the virus at lower doses. It might be possible to recover the growth capacity and immunogenicity of E XZ/SA14142m by reducing the number of amino acid substitutions. Previous reports suggest that four amino acid residues at positions 107, 138, 176, and 244 of the 10 substitution sites play essential roles in the attenuation of neurovirulence and neuroinvasiveness of the SA 14 strain, but at least two at positions 177 and 264 seem to be dispensable for such attenuation [33,34,36,41]. A recombinant JEV with a minimum of four amino acid substitutions in the GV E protein may be a safe and highly efficacious live vaccine against GV JEV infection. The reverse genetics of GV JEV have already been established [18,46]. This method could be used to generate mutant GV JEV with the same amino acid substitutions in the E protein, and the viruses may represent highly proliferative but safe and effective live vaccine candidates. However, we previously showed that the growth of Muar is deficient in mouse neuroblastoma cells [19,22], indicating that the GVbased mutants would have low replicative potential in mouse-derived cells. Increasing the growth capacity of GVs by introducing new mutations might be required for GV-based live attenuated vaccine development. Our previous data showed that Muar could grow efficiently in human neuroblastoma-derived IMR-32 cells [21]. Thus the low growth ability of GV JEV in mouse-derived cells may not be critical for the development of live attenuated recombinant GV JE vaccines in humans.
Most amino acid residues in the 10 sites we introduced specific amino acid substitutions in this study are also conserved in JE serocomplex flaviviruses, the West Nile virus, Usutu virus, and Murray Valley encephalitis virus (Table S2). A previous report indicated that introducing all 10 substitutions into the attenuated West Nile virus strain W1806 resulted in defective virus propagation in mammalian cells [47]. ChimeriVax-WN, which has the prM-E region of West Nile virus in the live attenuated yellow fever vaccine strain 17D backbone, has three amino acid substitutions L107F, A316V (A315V in JEV), and K440R (K439R in JEV), and the mutations are critical for the reduction in neurovirulence of the chimeric virus [48]. However, a single E138K mutation in the E protein of West Nile virus does not attenuate its virulence [49]. Thus, the "multiple substitution method" could apply to the development of live attenuated vaccines for JE serocomplex flavivirus infections, but would also require careful selection of the positions introducing amino acid substitutions.
In this study, we used a mouse infection model to evaluate the protection and immunogenicity potency of the mutant viruses. However, a more comprehensive analysis using mice and non-human primates is needed to assess the safety and efficacy of the mutant virus against JEV infection before considering its application to humans.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/vaccines9101077/s1, Table S1: Comparison of the amino acid residues on 10 sites in JEV E protein. Table S2: Comparison of the amino acid residues at 10 sites in E protein of JEV, West Nile virus, Usutu virus, Murray valley encephalitis virus, and St. Louis encephalitis virus. Figure S1: Alignment of the nucleotide sequences of E region of recombinant JEV strains. Figure S2: Body weight of mice inoculated with recombinant JEV strains as shown in Figure 3A. Figure S3: Body weight of mice inoculated with recombinant JEV strains as shown in Figure 6C. Figure S4: Body weight of mice inoculated with recombinant JEV strains as shown in Figure 7B. Figure S5: Comparison of amino acid sequences of E protein of GV JEV Muar and XZ0934 strains.