Acidity/Alkalinity of Japanese Encephalitis Virus E Protein Residue 138 Alters Neurovirulence in Mice

The E protein is the only glycoprotein in mature JEV, and it plays an important role in viral neurovirulence. E protein mutations attenuate JEV neurovirulence through unclear mechanisms. Here, we discovered that E138 is a predominant determinant of JEV neurovirulence. We demonstrated that the alkalinity/acidity of E138 determines JEV neurovirulence. These data contribute to the characterization of the E protein and the rational development of novel JEV vaccines.

ABSTRACT

The Japanese encephalitis virus (JEV) envelope (E) protein, as one of mediators of virus entry into host cells, plays a critical role in determining virulence. The Glu-to-Lys mutation of residue 138 in E protein (E138) plays an important role in attenuating JEV vaccine strain SA14-14-2. However, it is not clear how E138 attenuates JEV. Here, we demonstrate that the Glu-to-Arg mutation of E138 also determines the attenuation of JEV strain 10S3. Likewise, for its parent strain (HEN0701), a virulence strain, the mutations of E138 are responsible for virulence alteration. Furthermore, we demonstrated that mutations of alkaline residues in E138 contributed to the attenuation of neurovirulence; in contrast, mutations of acidic residues enhanced the neurovirulence of the strains. Moreover, acidity in residue E47 had a similar effect on neurovirulence. Furthermore, the alkaline E138 residue enhanced susceptibility to heparin inhibition in vitro and limited JEV diffusion in mouse brain. These results suggest that the acidity/alkalinity of the E138 residue plays an important role in neurovirulence determination.

IMPORTANCE The E protein is the only glycoprotein in mature JEV, and it plays an important role in viral neurovirulence. E protein mutations attenuate JEV neurovirulence through unclear mechanisms. Here, we discovered that E138 is a predominant determinant of JEV neurovirulence. We demonstrated that the alkalinity/acidity of E138 determines JEV neurovirulence. These data contribute to the characterization of the E protein and the rational development of novel JEV vaccines.

INTRODUCTION

Japanese encephalitis (JE), caused by Japanese encephalitis virus (JEV), is the main form of viral encephalitis in East, Southeast, and South Asia (1). Almost 60% of the world's population, over 3 billion people, live in JEV enzootic areas. Children under the age of 15 are at a high risk of JE. Every year, more than 68,000 JE cases are reported, of which about 20 to 30% are fatal, and up to 50% of survivors may develop long-term neurological sequelae (2, 3).

JEV is a small enveloped virus belonging to the genus Flavivirus, which includes dengue virus (DENV), yellow fever virus (YFV), and tick-borne encephalitis virus (TBEV). The JEV genome is a single-stranded, positive-sense RNA about 11 kb in length and encodes a single polyprotein that is processed co- and posttranslationally into structural (C, prM/M, and E) and nonstructural (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5) proteins by cellular and viral proteases (4, 5). Among those viral proteins, the JEV E protein plays a key role in virus binding and entry into host cells (6, 7). As a typical class II membrane fusion protein, the E protein, together with the M protein, forms the outer shell of viral particles by anchoring to the lipid bilayer envelope through transmembrane helices (6, 8). Furthermore, according to the in situ structure of the E protein, domains I, II, and III (DI, DII, and DIII), out of five domains, form the ectodomain, which is crucial to virus binding and entry into host cells (9). DIII is thought to interact with cellular receptors. Both the recombinant E DIII and the loop 3 peptide derived from E DIII inhibit JEV entry into cells (10). In 2017, however, Wang et al. reported the cryo-electron microscopy structure of JEV and revealed that an unusual “hole” on the surface surrounded by five encephalitis-specific motifs, which were located in DI and DII, may be implicated in receptor binding (8). JEV can infect multiple cell types with different molecules as receptors. How the E protein interacts with these receptors remains unclear.

E protein is profoundly involved in flavivirus entry, assembly, and release. During entry into and release from host cells, E protein undergoes extensive conformational changes. In TBEV, a conserved histidine (His323) at the interface between DI and DIII was identified as the critical pH sensor (11). The protonation of His323 triggers large-scale conformational rearrangement of E protein and the fusion of the viral and endosomal membranes. By structure-based mutational analysis of JEV E protein, the roles of some motifs and amino acid sites in virus assembly, release, binding, and entry were determined. The mutation of T410 and Q258 residues, which participate in the zippering reaction, resulted in a significant reduction in JEV membrane fusion activity (12). During the passage of some mutants in BHK-21 cells, positive-charge mutations occurred and recovered the reduced entry activity (12). However, some wild-type (WT) JEV passaged in vitro also easily gained positive-charge mutations in E protein. Some mutations of negatively charged amino acids to positive charge may attenuate JEV. The positively charged residues can improve E protein affinity for glycosaminoglycans (GAGs) ubiquitously present on the cell surface and extracellular matrix, which prevents JEV spread from extraneural sites of replication into the brain (13). However, it is not clear how mutations producing a positive charge affect JEV neurovirulence.

It has been shown that mutations in the E protein are responsible for JEV attenuation, such as attenuated strains SA14-14-2, RP-2ms, CH2195LA, and CJN-S1 (14–19). Among these changes, mutations in residue 138 of the E protein (E138) occurred at a high frequency and have been shown to affect neurovirulence. However, the mechanism of E138 attenuation of JEV virulence is unclear. Previously, we reported that an attenuated JEV strain, 10S3, was derived from the wild-type JEV strain HEN0701 by 100 passages in BHK-21 cells (20). Here, we show that strain 10S3 has an E138 mutation and that the acidity/alkalinity of E138 determines neurovirulence. These data provide evidence that E138 mutation is a hallmark of JEV neurovirulence and that the acidity/alkalinity of E138 is an important mechanism that contributes to shifting of JEV neurovirulence.

RESULTS

E138 plays an important role in JEV neurovirulence.

The genome of strain 10S3 was determined. Compared with the genomic sequence of HEN0701, 10S3 had 19 mutant nucleotides, which led to 7 amino acid mutations (Table 1). RNA transcripts synthesized from the 10S3 cDNA clone pAHEN in vitro were transfected into BHK-21 cells, generating vAHEN (Fig. 1A). The recovered vAHEN exhibited plaque morphology and growth characteristics in vitro similar to those of the parental 10S3 strain (Fig. 2A and B). To detect the virulence of vAHEN, seven mice were inoculated intracerebrally (i.c.) with 1 × 10⁶ 50% tissue culture infective doses (TCID₅₀) of vAHEN and 10S3. All seven mice in the vAHEN group survived and showed no clinical symptoms, which was the same as the 10S3 group. These results indicated that vAHEN has biological properties similar to those of 10S3 in vitro and in vivo.

TABLE 1.

Nucleotide and amino acid residues differ between the JEV wild-type strain HEN0701 and the attenuated strain 10S3

Nucleotide position(s)a	Region or gene	Nucleotide (amino acid) in JEV strains		Amino acid positionb
		HEN0701	AHEN
427	C	T	C
540	prM	T	C
1390–1391	E	GA (Glu)	AG (Agr)	138
2685	NS1	G	A
2757		C	T
3188		A (Glu)	G (Gly)	237
3489		T	C
3570	NS2a	C	T
4913	NS3	A (Gln)	C (Ala)	102
6864	NS4a	G	A
7101	NS4b	A (Glu)	C (Asp)	63
7487		T (Val)	C (Ala)	192
8506	NS5	A (Lys)	G (Glu)	277
8615		T (Val)	C (Ala)	313
9267		T	A
9618		T	C
10705	3′-UTR	C	T
10960		G	A

^aThe nucleotide numbering refers to the start of the genome.

^bThe amino acid numbering refers to the start of the viral protein listed in the second column of the table.

FIG 1.

Construction of mutant JEVs. (A) The chimeric JEVs vH/5′CprME(A) and vA/5′CprME(H) were constructed by exchanging 5′-UTR and structural genes between the virulent strain vJEHEN and the attenuated strain vAHEN. (B) The mutant JEV vAR138E was constructed by revertant mutation of E138 (codon nt 1390 to 1392) in the vAHEN genome. (C) Mutant JEVs in E138 derived from the virulent strain vJEHEN. Site-directed mutations of nt 1390 to 1392 (coding for E138) in the vJEHEN genome were generated. The gag codon was mutated to agg, gac, aag, cag, ttc, gcc, and cac. Then, seven mutant JEVs, vHE138R, vHE138D, vHE138K, vHE138Q, vHE138F, vHE138A, and vHE138H, were rescued. (D) E47 mutant JEVs derived from virulent strain vJEHEN. Site-directed mutations of nt 1117 to 1119 (coding for E47) in the vJEHEN genome were generated. The aac codon was mutated into gac, gcc, and aag. Then, three mutant JEVs, vHN47D, vHN47A, and vHN47K, were recovered. The data are shown as means and SD.

FIG 2.

Biological properties of chimeric viruses and mutant viruses, vHE138R and vAR138E. (A) Plaque morphology of 10S3 and its recovered virus, vAHEN. BHK-21 monolayers were infected with 10S3 and vAHEN and then stained with crystal violet at 4 dpi. Ten plaques for each strain were chosen at random to calculate plaque size. The data are presented as means and SD. (B) One-step growth curves of vAHEN and 10S3 in BHK-21 cells. BHK-21 cells were inoculated with JEVs at an MOI of 10. The supernatant of infected cells was harvested at the indicated time points and used to determine viral titers in BHK-21 cells. The data are presented as means ± SD from three independent experiments. (C) Plaque morphology of the two chimeric strains [vA/5′CprME(H) and vH/5′CprME(A)] and the mutant viruses carrying a Glu-to-Arg (vHE138R) or an Arg-to-Glu (vAR138E) mutation in E138. Ten plaques for each strain were chosen at random to calculate plaque size. The data are presented as means and SD and were tested for significance using Student's t test. *, P < 0.05; ****, P < 0.0001. (D) Multistep growth curves of vH/5′CprME(A) and vHE138R in BHK-21 cells. (E) Multistep growth curves of vA/5′CprME(H) and vAR138E in BHK-21 cells. BHK-21 cells were inoculated with JEVs at an MOI of 0.01. The supernatant of infected cells was harvested at the indicated time points and used to determine the viral titers in BHK-21 cells. The data are presented as means ± SD from three independent experiments. * indicates a significant difference between vJEHEN and vHE138R or between vAHEN and vA/5′CprME(H) (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001). φ indicates a significant difference between vJEHEN and vH/5′CprME(A) or between vAHEN and vAR138E (φφ, P < 0.01; φφφ, P < 0.001; φφφφ, P < 0.0001).

To determine the genetic basis of the attenuated vAHEN neurovirulence, structural genes were exchanged between pAHEN and pJEHEN, and two chimeric viruses, vH/5′CprME(A) and vA/5′CprME(H), were rescued (Fig. 1A). The virological characteristics of vH/5′CprME(A) and vA/5′CprME(H), together with vAHEN and vJEHEN, in BHK-21 were investigated. The highly virulent strain vJEHEN exhibited plaques similar to those of the attenuated strain vAHEN (Fig. 2C). The plaques of the two chimeric viruses were significantly smaller than those of their parent strains. The vH/5′CprME(A) plaques were slightly bigger than the vA/5′CprME(H) plaques. Compared to their parent strains, vAHEN and vJEHEN, the chimeric strains vH/5′CprME(A) and vA/5′CprME(H) grew slowly (Fig. 2D and E).

Next, the neurovirulence of vH/5′CprME(A) and vA/5′CprME(H) in Kunming mice was examined. Seven 3-week-old mice were inoculated i.c. with 5 × 10⁴, 50, and 0.05 TCID₅₀ of viruses. All the mice inoculated with vH/5′CprME(A) or vAHEN survived throughout the experiment. Fourteen mice infected with vA/5′CprME(H) died between 4 and 9 days postinfection (dpi), similar to those infected with vJEHEN. The mortality rate of mice infected with vA/5′CprME(H) was almost identical to that of those infected with vJEHEN (Table 2). These results indicate that the structural genes play a crucial role in JEV neurovirulence.

TABLE 2.

Neurovirulence of chimeric and mutant viruses in mice

Inoculation dose (TCID50)	Mortality rate (%)
	vJEHEN	vAHEN	vA/5'CPrME(H)	vH/5'CPrME(A)	vAR138E	vHE138R
5 × 104	7/7 (100)	0/7 (0)	7/7 (100)	0/7 (0)	7/7 (100)	0/7 (0)
50	7/7 (100)	0/7 (0)	6/7 (86)	0/7 (0)	7/7 (100)	0/7 (0)
0.05	1/7 (14)	0/7 (0)	1/7 (14)	0/7 (0)	1/7 (14)	0/7 (0)

There are 4 nucleotides (nt) in the 5′ untranslated region (UTR) and structural genes that differ between vAHEN and vJEHEN, and only two neighboring nucleotide mutations resulted in amino acid changes (Table 1). The different amino acids were located at E138. The amino acid at E138 was Glu in vJEHEN and Arg in vAHEN. We investigated what role the E138 residue plays in JEV neurovirulence.

Two JEV mutants, vHE138R and vAR138E, were constructed using site-directed mutagenesis on pJEHEN and pAHEN (Fig. 1B and C). Compared to vJEHEN, only the Glu in E138 was replaced with Arg in vHE138R. Conversely, compared to vAHEN, the Arg in E138 was replaced with Glu in vAR138E. The mutant strain vHE138R had growth kinetics similar to those of its parent strain, vJEHEN. However, the mutant strain vAR138E grew slowly compared to the parent strain but similarly to vA/5′CPrME(H) (Fig. 2D and E). The plaques of vHE138R and vAE138E were also significantly smaller than those of their parent strains (Fig. 2C). When mice were infected with vHE138R and vAR138E, they exhibited differences in neurovirulence. vHE138R lost neurovirulence in mice, identically to vAHEN. All seven mice in the group infected with 5 × 10⁴ TCID₅₀ of vHE138R survived. Conversely, the neurovirulence of vAR138E was high. All seven mice in the group infected with 50 TCID₅₀ of vAR138E died. Even in the group infected with 0.05 TCID₅₀, one mouse died (Table 2). These results indicated that vAR138E, like vJEHEN, has high neurovirulence in mice. Thus, it is the E138 residue that determines the different neurovirulences of vAHEN and vJEHEN.

The acidity of the side chain of the E138 residue is associated with JEV neurovirulence.

The Glu in E138 is highly conserved in wild-type JEVs. However, the Glu in E138 is replaced with Lys in many attenuated strains, such as SA14-14-2, at222, and RP-2ms (Fig. 3). Glu is acidic, while Arg and Lys are alkaline. We investigated whether the acidity/alkalinity of E138 is related to JEV neurovirulence. Six mutant JEVs were constructed using site-directed mutagenesis on pJEHEN (Fig. 1C). The Glu in E138 was replaced by Asp, Lys, His, Gln, Phe, and Ala. All the corresponding mutants, vHE138D, vHE138K, vHE138H, vHE138Q, vHE138F, and vHE138A, were rescued successfully. As shown in Fig. 4, almost all the mutants reached peak titers at 48 h postinfection (hpi), except that vHE138K reached peak titer at 60 hpi in BHK-21 cells. The titer of the parent strain, vJEHEN, was significantly higher than those of vHE138D, vHE138K, vHE138F, and vHE138Q at 24 hpi; those of vHE138K, vHE138F, vHE138Q, and vHE138A at 36 hpi; and those of vHE138D, vHE138K, vHE138F, and vHE138H at 48 hpi (Table 3). This suggested that these E138 mutations, especially E-to-D, E-to-K, and E-to-F mutations in E138, decreased the viral growth rate in BHK-21 cells. After infecting mice i.c. at 5 × 10⁴, 50, or 0.05 TCID₅₀, the mutants displayed different neurovirulences (Table 4). Compared to that of vJEHEN, the neurovirulences of six mutants decreased. All the mice infected with 0.05 TCID₅₀ of the mutants survived throughout the experiment. The mouse mortality was 7/7, 7/7, 6/7, 4/7, 1/7, and 0/7, caused by vHE138D, vHE138H, vHE138A, vHE138Q, vHE138F, and vHE138K at 5 × 10⁴ TCID₅₀, and it was 4/7, 5/7, 2/7, 2/7, 0/7, and 0/7 at 50 TCID₅₀. Like vHE138R, vHE138K lost neurovirulence, indicating that the E-to-K mutation attenuated vJEHEN. Except for vHE138H, the viral neurovirulence was closely related to the alkalinity/acidity of the E138 residue. Viruses such as vHE138R and vHE138K, with highly alkaline residues at E138, showed no neurovirulence. On the other hand, viruses such as vJEHEN and vHE138D, with highly acidic residues at E138, showed high neurovirulence. The mutant viruses vHE138A, vHE138Q, and vHE138F, with weakly acidic residues at E138, showed partly attenuated neurovirulence. These results suggest that the acidity of a side chain of residue 138 has an important influence on JEV neurovirulence.

FIG 3.

Sequence alignment of amino acids 111 to 160 of JEV E protein among wild-type strains and four attenuated strains. The amino acid alignment was conducted with MEGA version 5.0. The E gene sequences were derived from the GenBank database, and the accession numbers are as follows: SA14 (M55506), Beijing-1 (L48961), P3 (U47032), Nakayama (EF571853), JaOArS982 (M18370), KV1899 (AY316157), JEV/sw/Mie/40/2004 (AB241118), JEV/eq/India/H225/2009 (JX131374), JEV/SW/IVRI/395A/2014 (KP164498), AT31 (AB196923), NT109 (U44967), HEN0701 (FJ495189), at222 (AB196924), RP-2ms (AF014160), and SA14-14-2 (AF315119). SA14-14-2 was attenuated from the WT strain SA14, at222 from AT31, RP-2ms from NT109, and 10S3 from HEN0701.

FIG 4.

Multistep growth curves of E138 mutant viruses. BHK-21 cells were inoculated with JEVs at an MOI of 0.01. The supernatant of infected cells was harvested at the indicated time points and used to determine viral titers in BHK-21 cells. The data are presented as means ± SD from three independent experiments. * indicates a significant difference between vJEHEN and vHE138Q (*, P < 0.05; **, P < 0.01; ****, P < 0.0001), # between vJEHEN and vHE138H (#, P < 0.05; ##, P < 0.01), ω between vJEHEN and vHE138D (ω, P < 0.05; ωω, P < 0.01; ωωω, P < 0.001), φ between vJEHEN and vHE138F (φφ, P < 0.01; φφφφ, P < 0.0001), δ between vJEHEN and vHE138K (δ, P < 0.05; δδδδ, P < 0.0001), and ψ between vJEHEN and vHE138AN (ψ, P < 0.05; ψψψψ, P < 0.0001).

TABLE 3.

Statistical differences between JEV strains

JEV strain	P valuea
	12 h	24 h	36 h	48 h	60 h	72 h
vHE138H	0.1159	0.9999	0.0543	0.0328	0.478	0.004
vHE138Q	0.0038	0.0247	<0.0001	0.9759	0.1197	0.0183
vHE138F	0.0711	0.0074	<0.0001	<0.0001	<0.0001	<0.0001
vHE138K	0.0318	<0.0001	<0.0001	<0.0001	0.9996	0.1636
vHE138D	0.3334	0.0056	0.1187	0.0004	0.0118	0.01
vHE138A	0.9759	0.9996	0.0105	0.7921	0.094	<0.0001
vAHEN	0.706	0.0872	0.9997	0.9519	0.9921	0.1511

^aThe P values show the statistical differences between the indicated JEV strains and vJEHEN.

TABLE 4.

Amino acids located at E138 and E47 and neurovirulences of mutant JEVs

Virus	Amino acida		No. of hydrogen bonds between E138 and E47b	Mortality at TCID50c of:
	E47	E138		50,000	50	0.05
vJEHEN	Asn (5.41)	Glu (3.22)	3	7/7	7/7	1/7
vHE138D		Asp (2.77)	3	7/7	4/7	0/7
vHE138F		Phe (5.48)	2	1/7	0/7	0/7
vHE138Q		Gln (5.65)	2	4/7	2/7	0/7
vHE138A		Ala (6.00)	2	6/7	2/7	0/7
vHE138H		His (7.59)	3	7/7	5/7	0/7
vHE138K		Lys (9.74)	2	0/7	0/7	0/7
vHE138R		Arg (10.76)	2	0/7	0/7	0/7
vHN47D	Asp (2.77)	Glu (3.22)	2	7/7	7/7	1/7
vHN47A	Ala (6.00)		2	7/7	6/7	0/7
vHN47K	Lys (9.74)		3	0/7	0/7	0/7

^aThe value of the isoelectronic point, pI, for each amino acid is shown in parentheses.

^bThe hydrogen bonds between E138 and E47 were analyzed by a homology model of the crystal structure of the E ectodomain of JEV SA14-14-2 (PDB ID 3P54).

^cTCID₅₀, dose used to inoculate the mice.

The E47 residue also has a critical effect on JEV neurovirulence.

There are 23 β-strands in the JEV E protein (Protein Data Bank [PDB] identifier [ID] 3P54) structure (17). DI consists of nine β-sheets (β-sheets 1, 2, 3, 4, 9, 10, 11, 12, and 17), and β-sheet 9, containing E138, is adjacent to β-sheet 4 (Fig. 5). The E138 residue interacts with the E47 residue via hydrogen bonding. We analyzed the interaction between E138 and E47 by homology modeling of the JEV E protein structure. There are two hydrogen bonds between E138 and E47 in vHE138A, vHE138Q, vHE138F, vHE138R, and vHE138K and three hydrogen bonds between E138 and E47 in vJEHEN, vHE138D, and vHE138H, which showed high neurovirulence (Table 4).

FIG 5.

E138 residue Glu interacts with E47 residue Asn in the 3D structure of vJEHEN E protein. The 3D structure of vJEHEN E protein was generated by a homology model of the crystal structure of the E ectodomain of JEV SA14-14-2 (PDB ID 3P54) via the ExPASy Web server. The E47-resident β-sheet 4 and the E138-resident β-sheet 9 in domain I are shown. Three hydrogen bonds formed between the Glu residue of E138 and the Asn residue of E47 are shown as dotted lines.

It appeared that the hydrogen bonds between residues E138 and E47 had an effect on JEV neurovirulence. Therefore, the E47 residue was mutated to test this hypothesis in the infectious clone pJEHEN. Several amino acids, including Ala, Lys, Asp, Gln, Phe, and Arg, were selected to replace the Asn residue in E47. Three mutant JEVs, vHN47A, vHN47K, and vHN47D, were rescued (Fig. 1D). However, no viruses were rescued after replacements with CAG, TTC, or AGG in pJEHEN. The mutations in E47 had an adverse effect on virus growth in BHK-21 cells. The three mutants had the same growth rate as the parent virus, vJEHEN, before 48 hpi (Fig. 6), but the growth of the mutants deviated from that of vJEHEN after 48 hpi. The titer of vJEHEN continued to increase, reaching a peak at 72 hpi, and then gradually decreased. The three mutants reached peak titers at 48 hpi and then decreased. Although all three mutations, i.e., N47A, N47K, and N47D, attenuated virus growth in vitro, they had different effects on viral neurovirulence. After mice were inoculated i.c. with the three mutants and vJEHEN, vHN47D showed high virulence, similar to its parent strain, vJEHEN. The neurovirulence of vHN47A slightly decreased. However, the neurovirulence of vHN47K was markedly decreased, and no infected mice died (Table 4). The interactions of the mutant E47 residues with the Glu residue in E138 were analyzed by homology modeling of JEV E protein structure. There were two hydrogen bonds between E47 and E138 in vHN47D and vHN47A and three hydrogen bonds between E47 and E138 in vHN47K (Table 4). These data suggest that the hydrogen bonds between E47 and E138 may have little effect on JEV neurovirulence. In view of the acidity/alkalinity of the side chains of Asn, Ala, Asp, and Lys, the E47 residue affected JEV neurovirulence similarly to the E138 residue.

FIG 6.

Multistep growth curves of the E47 mutant virus. BHK-21 cells were inoculated with JEVs at an MOI of 0.01. The supernatant of infected cells was harvested at the indicated time points and used to determine the viral titers in BHK-21 cells. The data are presented as means ± SD from three independent experiments. * indicates a significant difference between vJEHEN and the E47 mutants vHN47A, vHN47D, and vHN47K (***, P < 0.001; ****, P < 0.0001).

Effects of E138 mutations on JEV binding to neuronal cells.

Neuro-2a cells, SK-N-SH cells, and primary neuronal cells were infected with vJEHEN and its mutants and were subjected to indirect immunofluorescence analysis (IFA) at 2 dpi. The primary neuronal cells infected with different mutants showed green fluorescence with a monoclonal antibody (MAb) against JEV NS3, and Neuro-2a cells and SK-N-SH cells showed red fluorescence, which indicated that all the mutants may infect the primary and passaged neuronal cells (Fig. 7A to C). Next, we investigated the abilities of the JEV mutants to bind to the primary and passaged neuronal cells. After the neuronal cells were incubated with JEV at equal copy numbers and incubated at 4°C for 2 h, the total RNAs of the cells and the bound virions were isolated and quantified using quantitative reverse transcription (qRT)-PCR. On different neuronal cells, the binding capability of vHE138H was the highest among these JEVs and significantly higher than that of wild-type vJEHEN (Fig. 7D). The binding capability of vHE138D varied on different neuronal cells and was significantly higher than that of vJEHEN on SK-N-SH cells. The binding capabilities of vHE138A, vHE138R, and vHE138K were similar to that of vJEHEN on the three types of neuronal cells (Fig. 7D). These results suggested that the acidity/alkalinity of the E138 residue had little effect on JEV binding to neuronal cells.

FIG 7.

E138 mutations had effects on JEV binding to neuronal cells in vitro. (A to C) IFA detection of JEV infection in primary mouse neuronal cells (A), Neuro-2a cells (B), and SK-N-SH cells (C). The primary neuronal cells, Neuro-2a cells, and SK-N-SH cells were infected with JEVs and then subjected to IFA at 2 dpi. The primary cells were detected with anti-JEV NS3 protein MAb and rabbit polyclonal anti-GFAP antibodies. The bound MAbs were detected by goat anti-mouse IgG conjugated to Alexa Fluor 488 and the bound polyclonal Abs by goat anti-rabbit IgG conjugated to Alexa Fluor 594. The bound MAbs in the infected Neuro-2a cells and SK-N-SH cells were detected with goat anti-mouse IgG conjugated to Alexa Fluor 594. (D) Capability of E138 mutant strains to bind to primary mouse neuronal cells, Neuro-2a cells, and SK-N-SH cells in vitro. The neuronal cells were inoculated with JEV mutants at equal copy numbers and incubated at 4°C for 2 h. Then, the numbers of copies of bound viruses were determined by qRT-PCR. The percentages of bound copies of each mutant relative to those of vJEHEN are shown. The data are presented as means and SD from three independent experiments (n = 3).

The alkaline E138 residue enhanced susceptibility to heparin inhibition of JEV binding.

It has been reported that GAGs have an inhibitory effect on some attenuated JEVs infecting cells. To assess the effect of GAGs on the E138 mutants binding to neuronal cells, a heparin inhibition assay was performed for vJEHEN, vHE138D, and vHE138R on primary neuronal cells, Neuro-2a cells, SK-N-SH cells, and Vero cells. As shown in Fig. 8, heparin at 1,000 μg/ml significantly decreased vHE138R binding to the primary neuronal cells, SK-N-SH cells, and Vero cells and also had an inhibitory effect, to some extent, on Neuro-2a cells. As the dose of heparin decreased, the inhibitory effect on vHE138R decreased. The high dose of heparin also prohibited vJEHEN and vHE138D from binding neuronal cells and Vero cells, while at a low dose it had little inhibitory effect. These results showed that the mutant vHE138R was more susceptible to heparin inhibition, which suggested that GAGs on the cell surface could contribute to the attenuated JEV binding to cells.

FIG 8.

Inhibition effects of heparin on JEV binding to neuronal cells. vHEJEN, vHE138D, and vHE138R (10⁶ copies) were treated with heparin and then inoculated onto primary neuronal cells, Neuro-2a cells, and SK-N-SH cells. The numbers of RNA copies of bound viruses were determined by qRT-PCR, using GAPDH as an internal control. The percentages of copies of each virus with heparin treatment relative to those of virus without heparin treatment are shown. The data are presented as means and SD from three independent experiments (n = 3). *, P < 0.05.

E-to-R mutation in E138 decreases diffusion in mouse brain.

Mice infected i.c. with the attenuated JEV at 5 × 10⁴ TCID₅₀ did not show any clinical symptoms of encephalitis. To reveal the infection characteristics of the attenuated JEV in mouse brain, mice were inoculated i.c. with vHE138R and vJEHEN. The genomic copies of both vJEHEN and vHE138R increased between 6 and 12 hpi, but the numbers of genomic copies of vJEHEN were 1,000-fold greater than those of vHE138R at 12 hpi. The genomic copies of vJEHEN increased progressively and peaked at 72 hpi. However, the genomic copies of vHE138R remained constant between 12 and 96 hpi (Fig. 9A). The right hemispheres of the infected mice were subjected to immunohistochemical staining at 96 hpi. Almost all neuronal cells stained positive with a MAb against JEV NS3 protein in the mice infected with vJEHEN. However, only a few neuronal cells in small areas around the injection site of inoculation were stained positive in the mice infected with vHE138R (Fig. 9B). These results revealed that both the virulent and attenuated JEVs were able to infect neuronal cells in vivo by i.c. inoculation, but the virulent JEV was able to spread in mouse brain tissues while the attenuated JEV did not diffuse.

FIG 9.

E-to-R mutation of the E138 residue limited JEV dissemination in mouse brain. (A) Genomic copies of vJEHEN and vHE138R in infected mouse brain tissues. Mice were inoculated i.c. with vJEHEN and vHE138R and sacrificed at the indicated time points. The numbers of JEV copies were determined by qRT-PCR. The data are presented as means ± SD from five infected mouse brains. *, P < 0.05; **, P < 0.01; ***, P < 0.001. (B) Distribution of vJEHEN and vHE138R in infected mouse brain tissues. Mice were injected with vJEHEN and vHE138R into the right cerebral hemisphere and sacrificed at 96 hpi. The right hemispheres were immunohistochemically stained with anti-JEV NS3 protein MAb. Positive neuronal cells infected by JEV are circled in red.

DISCUSSION

This study provides novel information regarding the role of JEV E138 in neurovirulence determinants in mice. Our data demonstrate that the acidity of E138 contributes to JEV neurovirulence in mice. In contrast, mutation with alkaline residues in E138 contributes to attenuation of neurovirulence. Collectively, our data provide a novel demonstration that the acidity/alkalinity of E138 could determine JEV neurovirulence mediated by JEV E protein.

The E protein of JEV plays an important role in neurovirulence determination. It has been shown that some encephalitic strains become avirulent through an exchange for the E proteins of avirulent strains (21). Furthermore, more recently, it has been reported that the single E-to-K mutation of E138 completely attenuated JEV neurovirulence (8). In fact, as a key residue site for virulence determination, Glu in E138 is very conserved in encephalitic strains (as shown in Fig. 3). However, during passaging in vitro, the E138 residue of encephalitic strains was prone to E-to-K mutation, consistent with virulence attenuation (15, 20). In comparison, avirulent strains, such as SA14-14-2, were stable, with Lys in E138, during passaging in PDK cells or Vero cells (22). Moreover, when SA14-14-2 was passaged in suckling mouse brains, the K-to-E reversion mutation in E138 began to appear in the fourth passage, consistent with the increased neurovirulence by the fourth passage of virus in 4-week-old BALB/c mice (23). Thus, E138 is a predominant determinant of JEV neurovirulence.

Furthermore, our data support this finding, i.e., a single mutation of E138 could attenuate the neurovirulence of the wild-type strain HEN0701; on the other hand, a single mutation of E138 in the avirulent strain 10S3 promoted its neurovirulence. Furthermore, by a series of mutations of E138, we identified an unrevealed phenomenon where acidic residues in E138 of strain HEN0701 contribute to its neurovirulence; in contrast, mutation to alkaline residues in E138 attenuates its neurovirulence. To our knowledge, this is the first time that exchange between acidic amino acids and basic amino acids in E138 has been shown to contribute to the neurovirulence of JEV strains.

The E protein of JEV mediates virus binding and entry into host cells. It has been shown that E138 maps onto a conserved encephalitis-specific motif surrounding an unusual “hole” on the surface with the other four encephalitis-specific motifs implicated in receptor binding, and it is important for binding to neuroblastoma cells (10). These results motivated us to investigate the role of the acidity/alkalinity of E138 in cell binding. Interestingly, under our experimental setup, including different strains and virus inoculations with equal copy numbers, we found that all of the viruses carrying E138 mutations, such as vHE138K, vHE138R, vHE138A, and vJEHEN, appeared to bind cells similarly (Fig. 7D). Thus, our data suggest that alteration of neurovirulence mediated by mutation of E138 may not be attributed to the ability to bind cells.

The E protein of JEV mediates virus binding and entry into cells by interacting with cellular receptors, such as DC-SIGN (dendritic cell-specific intercellular adhesion molecule-3-grabbing nonintegrin), and CLEC5A, which are member of the C-type lectin family and can trigger an innate immune response to virus infection (24, 25). Though our data show that E138 mutations have no role in virus binding, this does not exclude the possibility that the interaction of different mutations with cellular receptors may not be the same, which leads to different innate immune responses. At present, whether the different innate immune responses at the beginning are key to pronounced differenences in infections of mutants is not clear. Thus, future studies are needed to address this point.

GAGs are attachment factors for flaviviruses to enhance virion attachment at the target cell surface before their interaction with primary receptors (13, 26). The GAG-susceptible variants of JEV were rapidly removed from the bloodstream and failed to spread from extraneural sites of replication into the brain, which may illustrate the mechanism of virulence attenuation of GAG-binding variants of JEV (13). Although both E-to-K and E-to-R mutations improved JEV susceptibility to GAGs (8, 27), the GAG binding did not explain the attenuated neurovirulence of vHE138K and vHE138R when viruses were directly inoculated into mouse brains. Our results showed that vHE138R inoculated i.c. was able to replicate but did not diffuse in brain tissue (Fig. 9). When JEV was passaged in SW13 cells or BHK-21 cells, some amino acid mutations of the E protein from negatively charged amino acids to positively charged amino acids usually occurred (12, 13). The gain-of-net-positive-charge mutations of E protein in JEV passaged in vitro could be a result of virus adaptation to cells, which enhanced E protein binding to GAGs with negative charges on the cell surface and then contributed to JEV infection in vitro.

In summary, this study establishes that the acidity/alkalinity of E138 contributes to alteration of JEV neurovirulence. Furthermore, our data highlight the fact that E138, as a predominant determinant of JEV neurovirulence, could serve as an important molecular marker for assessing the virulence of different JEV strains.

MATERIALS AND METHODS

Cells and viruses.

BHK-21 cells (ATCC; CCL-10) were cultured at 37°C in 5% CO₂ in Eagle's minimum essential medium (Sigma-Aldrich) supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum (FBS) (Gibco). Neuro-2a cells (ATCC; CCL-131) and SK-N-SH cells (ATCC; HTB-11) were cultured at 37°C in 5% CO₂ in Dulbecco's modified Eagle's medium (DMEM) (Gibco) supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum. Primary neuronal cells were isolated from the brains of 1-day-old mice. The brain tissue was trypsinized and treated with DNase. Neuronal cells were suspended (5.0 × 10⁶ cells/ml) and cultured at 37°C in 5% CO₂ in DMEM containing 10% (vol/vol) fetal bovine serum and 1% (vol/vol) penicillin-streptomycin solution. Cytosine arabinoside (10 μM) was added after 48 h to prevent glial cell proliferation.

The JEV strain HEN0701 (GenBank accession no. FJ495189) was originally isolated from aborted pig fetuses in China in 2007 (28). The attenuated 10S3 strain was derived from HEN0701 by 100 passages in BHK-21 cells (18).

Construction of JEV mutants.

The complete 10S3 genome was cloned into four overlapping cDNA segments, and an infectious cDNA clone, pAHEN, for 10S3 was constructed as described previously (28, 29). Two chimeric full-length clones, pH/5′CPrME(A), containing the 5′-C–prM–E region of 10S3 in the HEN0701 backbone, and pA/5′CPrME(H), containing the 5′-C–prM–E region of HEN0701 in the 10S3 backbone, were constructed by exchanging the 5′-C–prM–E regions between pAHEN and pJEHEN, an infectious clone of HEN0701, by digestion with NotI and AgeI restriction enzymes (Fig. 1A). The amino acid E138 was mutated using PCR-based site-directed mutagenesis on plasmid pHT7JE1 or pAT7C1 (carrying T7 sequences followed by nt 1 to 2913 of the 10S3 genome) with eight pairs of primers (Table 5). The mutated sequences were inserted into the corresponding positions in pJEHEN or pAHEN by digestion with NotI and AgeI restriction enzymes, generating the mutated full-length clones pAR138E, pHE138R, pHE138D, pHE138K, pHE138H, pHE138Q, pHE138F, and pHE138A (Fig. 1B and C). To mutate the E47 residue, the mutated segments were amplified using the forward primer PJET7 paired with three reverse primers (Table 5) with the template of pHT7JE1 and were inserted into the corresponding positions in pHT7JE1 with NotI and NheI, generating pHT7JE1-A, pHT7JE1-K, and pHT7JE1-D. Then, three mutated full-length clones, pHN47A, pHN47K, and pHN47D, were constructed by inserting the JEV sequences in pHT7JE1-A, pHT7JE1-K, and pHT7JE1-D, respectively, into the corresponding positions in pJEHEN by digestion with NotI and AgeI (Fig. 1D). Recovery of JEV mutants from in vitro transcripts and by transfection in BHK-21 cells was performed as described previously (29). The recovered viruses were passaged twice in BHK-21 cells, generating virus stocks for growth property, virulence, and sequence analyses.

TABLE 5.

Primers used for PCR and qRT-PCR

Primer	Sequence (5′–3′)a	Purpose
PAR138EF	GATCCAACCAGAGAACATCAAGTACgaGGTTGGCATATTCGTGCACG	PCR of R-to-E mutation at residue E138 in vAHEN genome
PAR138ER	CGTGCACGAATATGCCAACCtcGTACTTGATGTTCTCTGGTTGGATC
PHE138RF	GATCCAACCAGAGAACATCAAGTACagGGTTGGCATATTCGTGCACG	PCR of E-to-R, D, K, Q, F, and A mutations at residue E138 in vJEHEN genome
PHE138RR	CGTGCACGAATATGCCAACCctGTACTTGATGTTCTCTGGTTGGATC
PHE138DF	GATCCAACCAGAGAACATCAAGTACgacGTTGGCATATTCGTGCACG
PHE138DR	CGTGCACGAATATGCCAACgtcGTACTTGATGTTCTCTGGTTGGATC
PHE138KF	GATCCAACCAGAGAACATCAAGTACaagGTTGGCATATTCGTGCACG
PHE138KR	CGTGCACGAATATGCCAACcttGTACTTGATGTTCTCTGGTTGGATC
PHE138QF	GATCCAACCAGAGAACATCAAGTACcagGTTGGCATATTCGTGCACG
PHE138QR	CGTGCACGAATATGCCAACctgGTACTTGATGTTCTCTGGTTGGATC
PHE138FF	GATCCAACCAGAGAACATCAAGTACttcGTTGGCATATTCGTGCACG
PHE138FR	CGTGCACGAATATGCCAACgaaGTACTTGATGTTCTCTGGTTGGATC
PHE138AF	GATCCAACCAGAGAACATCAAGTACgccGTTGGCATATTCGTGCACG
PHE138AR	CGTGCACGAATATGCCAACggcGTACTTGATGTTCTCTGGTTGGATC
PJEF1-T7	GTAATGCGGCCGCTAATACGACTCACTATAGAGAAGTTTATCTGTGTGAACTTCTTG	PCR of N-to-D, K, and A mutations at residue E47 in vJEHEN genome
PHN47DR	CTAGCTAGCTTCAATgtcGATCATGCG
PHN47KR	CTAGCTAGCTTCAATcttGATCATGCG
PHN47AR	CTAGCTAGCTTCAATggcGATCATGCG
Mou-GAPDHF	TGCACCACCAACTGCTTAGC	qPCR for expression of GAPDH in mouse cells
Mou-GAPDHR	TGGATGCAGGGATGATGTTC
Mon-GAPDHF	CCTTCCGTGTCCCTACTGCCAAC	qPCR for expression of GAPDH in monkey cells
Mon-GAPDHR	GACGCCTGCTTCACCACCTTCT
Hum-GAPDHF	AGAAGGCTGGGGCTCATTTG	qPCR for expression of GAPDH in monkey cells
Hum-GAPDHR	AGGGGCCATCCACAGTCTTC
PqJEF	TGTGTGGTTCGTGGCAAGTG	Relative qPCR for JEV copy
PqJEF	TGGTTATGGGGGATGGGTTT
JENS2qF	AGCTGGGCCTTCTGGT	Absolute qPCR for JEV copy
NS2-probe	FAM-CTTCGCAAGAGGTGGACGGCCA-BHQ
JENS2qR	CTTCGCAAGAGGTGGACGGCCA

^aLowercase letters indicate mutated nucleotides; italic letters indicate endonuclease restriction sites. FAM, 6-carboxyfluorescein; BHQ, black hole quencher.

Viral plaque assay and growth kinetics.

The JEV mutants were serially 10-fold diluted in MEM, and 0.5 ml of the respective viruses was inoculated onto BHK-21 monolayers in 35-mm culture dishes. After 1 h at 37°C, the monolayers were washed twice with phosphate-buffered saline (PBS) and overlaid with 4 ml MEM containing 1% (wt/vol) low-melting-point agarose and 2% FBS. After incubation for a further 4 days at 37°C, the cells were fixed with 4% paraformaldehyde (PFA) and stained with 0.5% crystal violet.

To examine the growth properties of JEV mutants in BHK-21 cells, a multiple-step growth curve analysis was conducted. BHK-21 cell monolayers grown in 35-mm culture dishes were infected with JEV at a multiplicity of infection (MOI) of 0.01. At 1 hpi, the monolayers were washed twice with PBS, and 2 ml fresh medium (MEM plus 2% FBS) was added. Infected cell supernatants (200 μl) were collected, and the same volume of fresh medium was added at different time points (12 to 96 hpi). JEV titers were determined as TCID₅₀ per milliliter in BHK-21 cells. Serial 10-fold dilutions of virus were prepared in MEM plus 2% FBS and inoculated onto BHK-21 monolayers in 96-well plates (8 replicates per dilution; 100 μl/well). The plates were incubated for 4 days at 37°C under 5% CO₂. Titers in TCID₅₀ per milliliter were calculated from the cytopathic effect (CPE) according to the Reed–Muench formula.

IFA.

BHK-21 cells, Neuro-2a cells, and SK-N-SH cells grown in 35-mm culture dishes were infected with JEV at an MOI of 0.01. After 1 h of incubation at 37°C, the monolayers were washed once with PBS before adding fresh medium. The cells were incubated at 37°C for a further 2 days, washed once with PBS, and fixed in cold methanol for 10 min at −20°C. The cells were then permeabilized with 0.1% Tween 20 in PBS at room temperature for 5 min. After blocking with 2% bovine serum albumin in PBS for 30 min at 37°C, the cells were incubated for 1 h at 37°C with anti-JEV NS3 protein MAb (kindly provided by Shengbo Cao, Huazhong Agricultural University, Wuhan, China) at 1:500 dilution in PBS with 1% bovine serum albumin. After washing, bound antibody (Ab) was detected using goat anti-mouse IgG conjugated to Alexa Fluor 594 (Invitrogen).

Primary neuronal cells were infected with JEV at an MOI of 0.01. At 2 dpi, the primary neuronal cells were subjected to IFA to detect JEV protein using a MAb against NS3. Rabbit polyclonal anti-glial fibrillary acidic protein (GFAP) antibodies were used to detect mouse GFAP. The bound MAbs were detected with goat anti-mouse IgG conjugated to Alexa Fluor 488 and the bound polyclonal Abs with goat anti-rabbit IgG conjugated to Alexa Fluor 594.

In vitro binding assays.

The primary neuronal cells and the two neuronal cell lines, Neuro-2a and SK-N-SH, grown in 12-well plates, were washed twice with precooled PBS, and 1 ml DMEM per well was added. The cells were placed at 4°C for 30 min and then inoculated with JEV mutants at 10⁶ genomic copies per well, followed by incubation at 4°C for a further 2 h. After washing three times with precooled DMEM, total RNA of viruses binding to cells was isolated using an RNeasy minikit (Qiagen) according to the manufacturer's instructions.

To assess the inhibition effect of heparin on JEV binding to neuronal cells, 10⁶ copies of JEV plus heparin (1,000, 10, or 0.1 μg/ml) were preincubated for 1 h at 37°C, cooled at 4°C for 15 min, and then inoculated onto primary neuronal cells, Neuro-2a cells, and SK-N-SH cells. Then, binding assays were performed for the JEVs treated with heparin as described above.

Mouse infections.

All mouse experiments were performed according to protocols approved by the Animal Care and Ethics Committee of Shanghai Veterinary Research Institute, Chinese Academy of Agricultural Sciences, under the number Shvri-mo-20150691. All protocols adhered to the Guide for the Care and Use of Laboratory Animals.

Three-week-old White Kunming mice were purchased from Shanghai SLAC Laboratory Animal Co. Ltd., housed in an environmentally controlled room, and maintained on standard laboratory food and water ad libitum throughout the study. To assess the virulence of the JEV mutants, seven mice per group were inoculated i.c. with 50 μl diluent virus (5 × 10⁴, 50, or 0.05 TCID₅₀) and then monitored for 3 weeks. Mouse survival was monitored daily.

To evaluate JEV replication in mouse brains, 30 mice per group were inoculated i.c. with 5 × 10⁴ TCID₅₀ vJEHEN and vHE138R. At 6, 12, 24, 48, 72, and 96 hpi, five mice from each group were randomly sacrificed, and the brains were used for isolation of total RNA with TRIzol reagent.

Immunohistochemistry.

Kunming mice (3 weeks old; n = 7) were inoculated with 50 μl JEV (5 × 10⁴ TCID₅₀) into the right cerebral hemisphere. At 4 dpi, the mice were sacrificed, and the right hemispheres were fixed in 4% PFA, embedded in paraffin, and cut into 6-mm sections. The brain sections were treated in a microwave for antigen retrieval and incubated with 1% H₂O₂ in ice-cold methanol for 30 min to block endogenous peroxidase. They were then blocked with 1% FBS and incubated with NS3 MAb (1:300) for 12 h at 4°C, followed by incubation with goat anti-mouse IgG conjugated to peroxidase (Invitrogen). Signals were visualized by staining with 3,3′-diaminobenzidine solution containing 0.003% H₂O₂ and counterstaining with hematoxylin.

Real-time reverse transcription (RT)-PCR.

Total RNA was extracted from cells with an RNeasy minikit (Qiagen, Germany) or from tissues with TRIzol reagent (Invitrogen), and viral RNA was purified using a QIAamp viral RNA minikit (Qiagen). The first-strand cDNA was synthesized using a Thermo Scientific RevertAid First Strand cDNA synthesis kit (Thermo Scientific), following the manufacturer's instructions. Relative quantitative PCRs (qPCRs) were conducted using a SYBR Premix Ex Taq kit (TaKaRa, Japan). Primers qJEF/qJER (Table 5) were used for relative analysis of JEV genomic copies. Primers mou-GAPDHF/mou-GAPDHR, hum-GAPDHF/hum-GAPDHR, and mon-GAPDHF/mon-GAPDHR (Table 5) were used for quantification of GAPDH (glyceraldehyde-3-phosphate dehydrogenase) mRNA of the primary neuronal cells and Neuro-2a, SK-N-SH, and Vero cells, which served as an internal control. The absolute qPCR was performed with the primers JENS2qF/JENS2qR and the probe NS2-probe.

Statistical analysis.

Data were analyzed by using GraphPad Prism version 6. Measured values are expressed as means with standard deviations (SD). Binding assays were analyzed by using one-way analysis of variance (ANOVA) and unpaired two-tailed t tests. Growth curves and heparin assays were analyzed by two-way ANOVA with multiple comparisons.

Accession number(s).

The genome of 10S3 was deposited into GenBank under accession number MF542268.