Glutamic Acid at Residue 125 of the prM Helix Domain Interacts with Positively Charged Amino Acids in E Protein Domain II for Japanese Encephalitis Virus-Like-Particle Production

ABSTRACT

Interaction between E and prM proteins in flavivirus-infected cells is a major factor for virus-like particle (VLP) production. The prM helical (prM-H) domain is topologically close to and may interact with domain II of the E protein (EDII). In this study, we investigated prM-H domain amino acid residues facing Japanese encephalitis virus EDII using site-directed mutagenesis to determine their roles in prM-E interaction and VLP production. Our results indicate that negatively charged prM-E125 residue at the prM-H domain affected VLP production via one or more interactions with positively charged E-K93 and E-H246 residues at EDII. Exchanges of oppositely charged residue side chains at prM-E125/E-K93 and prM-E125/E-H246 are recoverable for VLP production. The prM-E125 and E-H246 residues are conserved and that the positive charge of the E-K93 residue is preserved in different flavivirus groups. These findings suggest that the electrostatic attractions of prM-E125, E-K93, and E-H246 residues are important to flavivirus VLP production and that inhibiting these interactions is a potential strategy for blocking flavivirus infections.

IMPORTANCE Molecular interaction between E and prM proteins of Japanese encephalitis virus is a major driving force for virus-like particle (VLP) production. The current high-resolution structures available for prM-E complexes do not include the membrane proximal stem region of prM. The prM stem region contains an N-terminal loop and a helix domain (prM-H). Since the prM-H domain is topologically close to domain II of the E protein (EDII), this study was to determine molecular interactions between the prM-H domain and EDII. We found that the molecular interactions between prM-E125 residue and positively charged E-K93 and E-H246 residues at EDII are critical for VLP production. More importantly, the prM-E125 and E-H246 residues are conserved and the positive charge of the E-K93 residue is preserved in different flavivirus groups. Our findings help refine the structure and molecular interactions on the flavivirus surface and reveal a potential strategy for blocking flavivirus infections by inhibiting these electrostatic interactions.

INTRODUCTION

Japanese encephalitis virus (JEV) is a small enveloped positive-strand RNA virus belonging to the Flavivirus genus of the Flaviviridae family (1, 2). In all flaviviruses, the RNA genome encodes three structural (core C, membrane precursor prM, and envelope E) and seven nonstructural (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5) genes, with flanking untranslated region (UTR) genes at the 5′ and 3′ ends (1, 2). The flavivirus particle assembly process entails (i) the interaction of prM and E proteins in the form of heterodimer formation in the endoplasmic reticulum (ER), (ii) genomic RNA encapsulation by the C protein and enclosure by cell membrane-derived lipid bilayers containing prM and E proteins to form immature virions, and (iii) the cleavage of prM proteins to M proteins via furin or a furin-like protease in the trans-Golgi network, thus triggering the release of viral particles. Virus-like particles (VLPs) that do not contain genomic RNA have been found in flavivirus-infected cells (3). The coexpression of prM and E envelope proteins leading to VLP formation and secretion in cell cultures has been reported for the tick-borne encephalitis virus (TBEV) (4), dengue virus (DENV) (5), JEV (6, 7), Murray Valley encephalitis virus (MVEV) (8), St. Louis encephalitis virus (SLEV) (9), and West Nile virus (WNV) (10). The interaction of prM and E proteins in infected cells has been described as a primary factor in flavivirus VLP production (11).

prM is a type I transmembrane (TM) protein with stem and anchor regions at the C-terminal ends (12). The stem region, which contains an N-terminal loop and helix domain (prM-H), undergoes a significant conformational change during the immature-to-mature structural transformation of virus particles (13–15). High-resolution cryo-electron microscopy (cryo-EM) images of DENV indicate that the N-terminal loop of the prM stem region is close to the “hole” between the two monomers that form the homodimer of the E ectodomain (15, 16). Topologically, the prM-H domain is close to domain II (DII) of homodimeric E proteins and therefore may contribute to additional prM-E interactions during virus maturation (16). The prM-H domain is partially buried at an angle to the membrane surface in the outer lipid leaflet of the C terminus (15, 16). According to one report, proline substitutions for C terminus residues of the DENV prM-H domain at I123 and V127 result in significant impairments to VLP production (17), suggesting that the prM-H domain may play an important role in flavivirus particle assembly.

For this study, we investigated the prM-H domain involvement of the CH2195LA JEV strain in insect and mammalian cell VLP production. Specifically, we investigated prM-H domain amino acid residues facing DII of the E protein (EDII), using site-directed mutagenesis to determine their roles in prM-E interaction and VLP production. According to our results, negatively charged prM-E125 residues at the prM-H domain affected VLP production via one or more interactions with positively charged E-K93 and E-H246 residues at EDII. Impaired charge-charge interactions between the prM-H domain and EDII suggest that the electrostatic attractions of charged residues represent a major factor in flavivirus VLP production.

MATERIALS AND METHODS

Cells and antibodies.

Sf9 cells were cultured in Sf900II serum-free medium (Invitrogen) at 28°C. Human embryonic kidney 293 (HEK293) cells were grown in Dulbecco modified Eagle medium (Gibco) supplemented with 5% (vol/vol) fetal bovine serum and 1% penicillin-streptomycin at 37°C. Mouse hybridoma cells producing monoclonal antibodies (MAbs) 4G2 (anti-flavivirus E fusion loop), E3.3 (anti-JEV E), and 5B1 (anti-JEV prM) were grown in Iscove modified Dulbecco medium (Invitrogen) containing 10% fetal bovine serum and 1% penicillin-streptomycin, also at 37°C.

Generation of plasmids containing JEV prM and E.

prM-E and EDI/II gene fragments of the E protein of the CH2195LA JEV strain (GenBank accession no. AF221499) were amplified by PCR and cloned into pJET1.2/blunt cloning vectors (Fermentas). Substitution mutations were introduced by site-directed mutagenesis. All gene fragment constructs were confirmed by sequence analysis performed at Mission Biotech (Taipei).

For the insect cell/baculovirus expression system, prM-E genes were subcloned into Bac-to-Bac pFastBac1 vectors (Invitrogen) to obtain recombinant bacmids. HEK293 cell transfecting plasmids were generated by subcloning prM-E genes into pcDNA3.1 vectors (Invitrogen). For Escherichia coli expression, the EDI/II gene fragments of the E protein were subcloned into pET22b(+) vectors.

Recombinant protein expression and purification.

Recombinant bacmids were transfected into Sf9 cells with Lipofectamine 2000 (Invitrogen) for 5 days according to the manufacturer's instructions. Baculoviruses were further amplified (two passages) in Sf9 cells to create sufficient virus titer for Sf9 infection. Sf9 cells were infected at a multiplicity of infection of 3.0 and harvested at 48 h postinfection. HEK293 cells were transfected with Turbofect (Fermentas) according to the manufacturer's instructions and harvested 48 h posttransfection.

EDI/II fragments containing pET22b plasmids were transformed into E. coli BL21(DE3), grown in Luria-Bertani (LB) broth at 37°C, and shaken at 200 rpm overnight. The resulting culture was diluted 100-fold in fresh LB broth and grown at 37°C with shaking to an optical density at 600 nm of 0.6, followed by high-level expression induced with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) and incubation with shaking for an additional 12 h. Cells were harvested by centrifugation (6,000 × g for 15 min) and homogenized using a French press at 15,000 lb/in². Pelleting was performed at 12,000 × g for 10 min; pellets were solubilized in 8 M urea at 4°C for 12 h. Solubilized EDI/II proteins were purified by using nickel-affinity chromatography in 50 mM Tris-HCl, 300 mM NaCl, and 5% glycerol using a linear imidazole gradient from 5 to 300 mM. Peak fractions were collected and concentrated using an Amicon concentrator (Millipore).

SDS-PAGE and Western blotting.

Harvested Sf9 and 293 cells were lysed with 0.5% Triton X-100 (vol/vol) in 50 mM Tris-HCl, 300 mM NaCl, and 2 mM EDTA, followed by centrifugation at 12,000 × g for 5 min at 4°C to remove cell debris. Lysate and purified proteins were heated at 95°C for 5 min in reducing sample buffer and resolved using 12% SDS-PAGE gel. For Western blotting, nitrocellulose paper (Millipore) was blocked in 5% (wt/vol) skim milk for 1 h at room temperature, followed by three washes with Tris buffer saline containing 0.05% Tween 20 (TBST). prM and E proteins were probed with 5B1, E3.3, and 4G2 mouse MAbs for 1 h at room temperature and detected with horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (KPL) for 1 h at room temperature. HRP-catalyzed enhanced chemiluminescence (Millipore) was captured with X-ray film.

Immunoprecipitation.

Cell lysates were precleared with protein A-magnetic beads (Millipore) for 2 h at 4°C and then reacted with rabbit polyclonal anti-myc IgG (Abcam catalog no. ab9106) for 5 h at 4°C, followed by another precipitation with protein A-magnetic beads. Immunoprecipitated samples were washed three times with TBST and solubilized by heating at 95°C for 5 min in reducing sample buffer for Western blot analysis as described above.

VLP production and purification.

The culture media of baculovirus-infected Sf9 cells were harvested at 72 h postinfection and cleared by centrifugation at 150 × g for 5 min. VLPs were precipitated using 10% (wt/vol) polyethylene glycol 8000, followed by centrifugation at 12,000 × g for 10 min. Resuspended pellets were ultracentrifuged through a 20% (wt/vol) sucrose buffer (20% sucrose in 50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 2 mM EDTA) at 33,000 rpm for 2 h. VLPs were subjected to further ultracentrifugation through a 0 to 60% (wt/vol) continuous sucrose gradient at 33,000 rpm for 2 h at 4°C. Sucrose gradient buffer was distributed throughout 12 top-to-bottom fractions.

ELISAs.

Purified proteins, lysates, or sucrose gradient fraction samples diluted in 0.05 M carbonate-bicarbonate coating buffer (pH 9.6) were added to 96-well microtiter enzyme-linked immunosorbent assay (ELISA) plates (Costar 9018; Corning) and incubated overnight at 4°C. After blocking with 3% bovine serum albumin (BSA) for 2 h at room temperature, prM and E proteins were probed using MAbs 5B1 and E3.3, respectively, for 1 h at room temperature. Bound antibodies were detected by HRP-conjugated goat anti-mouse IgG (KPL). After 1 h of incubation at room temperature, TMB (3,3′,5,5′-tetramethylbenzidine; BioLegend) substrate was added, followed by 10 min of additional incubation at room temperature in darkness. Color development was stopped by the addition of 2 M H₂SO₄ prior to measuring the absorbance at 450 nm.

Custom biotinylated oligopeptides purchased from Mission Biotech (Taipei) were subjected to oligopeptide ELISAs. Briefly, individual wells of 96-well microtiter plates were coated with 0.2 μg of purified recombinant EDI/II proteins in 0.05 M carbonate-bicarbonate coating buffer (pH 9.6), held overnight at 4°C, and then blocked with 3% BSA for 2 h at room temperature. Fourfold serial dilutions of biotinylated oligopeptides in potassium phosphate with 120 mM NaCl (pH 6.2) were added to the coated wells, held for 1 h at room temperature, and detected with HRP-conjugated streptavidin (BioLegend) for 1 h at room temperature. Absorbance was measured at 450 nm following TMB color development.

The ELISA binding curves of oligopeptides to recombinant EDI/II protein were further transformed to the Eadie-Hofstee plots to calculate the binding dissociation constant (K_d) and the maximum binding values (B_max) based on the following equation:

$$\frac{1}{(B-B_{\text{scrambled}})}=\frac{K_d}{B_{\max \hspace{0.17em}}}\times \frac{1}{[P]}+\frac{1}{B_{\max \hspace{0.17em}}}$$

where B is the amount of the bound oligopeptide, B_scrambled is the amount of nonspecific bound scrambled oligopeptide, and [P] is the total peptide concentration (18).

Statistical analysis.

A minimum of three experiments were performed for calculating mean values and standard deviations. The data represent the means ± the standard deviations (SD). The results were analyzed by using a two-tailed Student t test. A P value of <0.05 was considered statistically significant.

RESULTS

JEV prM-H domain residues face toward EDII.

According to high-resolution cryo-EM images of DENV, the prM-H structure is partially buried in the outer lipid leaflet, and the E-H1 and E-H2 domains are either angled or lie flat on the outer lipid leaflet (Fig. 1A) (15, 16). Topologically, the prM-H domain interacts with EDII in close proximity (Fig. 1A). The results from a PSIpred analysis of the amino acid sequences of the prM-H domain indicate prM amino acids at residues S114, A117, L121, E125, and I128 facing toward EDII (Fig. 1B).

FIG 1.

(A) Schematic diagram of prM and E proteins with helical (H) domains and anchor (TM) regions located at their C-terminal ends. Bars represent likely interacting residues at the prM-H domain (gray) and EDII (black). (B) Helical wheel projection of the prM-H domain.

Recombinant baculoviruses encoding the full-length prM-E gene of the CH2195LA JEV strain were constructed with each site-directed mutation (prM-S114A, prM-A117V, prM-L121A, prM-E125A, or prM-I128A). Sf9 insect cells were infected with each recombinant baculovirus and analyzed for VLP production in culture supernatant. The intracellular expression levels of prM and E proteins in Sf9 cells infected with wild-type (WT) and mutant prM-S114A, prM-A117V, prM-L121A, prM-E125A and prM-I128 baculoviruses were identical to those determined by Western blotting (Fig. 2A) and quantitated ELISAs (Fig. 2D). In contrast, the prM-E125A mutant triggered significantly lower VLP production (as determined by E protein content) in culture supernatant via 20% sucrose gradient sedimentation (Fig. 2B). According to our data for E protein content in VLPs, prM-E125A was the only mutant that significantly impaired VLP production in Sf9 cells (P < 0.05) (Fig. 2E). For further confirmation, culture supernatants were subjected to additional purification in the form of 0 to 60% continuous sucrose gradient sedimentation. Our results indicate that with the exception of the prM-E125A mutant, all VLP groups had similar sedimentation velocities and peaks between fractions 5 and 9 (Fig. 2C). The amount of VLPs produced by the prM-E125A mutant was ∼10-fold lower compared to the WT and other prM-H mutant viruses (P < 0.05) (Fig. 2F).

FIG 2.

Alanine/valine substitution mutagenesis of prM-H domain residues. Five alanine/valine substitution mutations were introduced into prM-H domain residues: S114A, A117V, L121A, E125A, and I128A. Western blotting was used to analyze prM and E protein expression in Sf9 cells (with GAPDH serving as an internal control) (A) and VLPs pelleted by ultracentrifugation through a 20% sucrose buffer (B). (C) Pelleted VLPs were resolved by 0 to 60% continuous sucrose gradient ultracentrifugation. Relative VLP quantities in the collected fractions were detected by ELISAs determining E protein concentrations. The results of one representative experiment of at least three performed are shown. The highest WT fraction was taken as 100%. A minimum of three ELISAs with three replicates was performed to determine the relative levels of prM and E expression in cell lysates (D) and VLP secretion into the supernatant (E). The WT quantity was taken as 100%. (F) Average areas under the curves of E protein concentrations from fractions 5 to 9 in panel C. The secreted WT VLP quantity was taken as 100%. The data are presented as the means ± the SD. The asterisk (*) indicates a statistically significance determined by Student t test at P < 0.05.

VLP production by prM-E125 mutations in HEK293 cells.

Since baculovirus-infected cells are known to hinder VLP production during late stages of lytic infection, we used HEK293 cells to produce VLPs by cotransfection with a pcDNA3.1 vector encoding the full-length prM-E gene. Four prM-E125 mutants were constructed using site-directed mutagenesis to determine whether the electrostatic characteristics of prM-E125 reduced VLP production in HEK293 cells: prM-E125D (negatively charged aspartate), prM-E125Q (noncharged glutamine), prM-E125K (positively charged lysine), and prM-E125R (positively charged arginine). Despite prM-E125 residue substitutions with other electrostatically and/or stereologically distinct amino acids, prM and E protein expression in HEK293 cells remained unchanged (Fig. 3A). The quantitated ELISA results were consistent with the Western blotting (Fig. 3D). A significant decrease in VLP production in culture supernatant was observed for the prM-E125Q, prM-E125K, and prM-E125R mutants (Fig. 3B); no significant effect was noted for the prM-E125D mutant (Fig. 3B). The amount of VLPs produced by the prM-E125Q, prM-E125K, and prM-E125R mutants was ca. 70% lower compared to the WT (P < 0.05) (Fig. 3E). The results from 0 to 60% continuous sucrose gradient ultracentrifugation indicate the presence of purified VLPs in fractions 5 to 9 (data not shown). According to an E protein ELISA, VLP amounts from the prM-E125Q, prM-E125K, and prM-E125R mutants were significantly lower compared to the WT and prM-E125D mutant (P < 0.05) (data not shown). Combined, these results demonstrate the important role of negatively charged amino acids in prM-E125 residues in VLP production.

FIG 3.

Substitution of negatively charged prM-E125 residue side chains. prM-E125 residues were substituted with negatively charged aspartic acid (prM-E125D), noncharged glutamine (prM-E125Q), and positively charged lysine (prM-E125K) and arginine (prM-E125R). Western blotting was used to analyze prM and E protein expression in Sf9 cells (A), VLPs pelleted by ultracentrifugation through 20% sucrose buffer (B), and immunoprecipitated prM and Myc epitope-fused E (Myc-E) proteins from cell lysates (C). Rabbit anti-c-Myc IgG was used to probe the Myc-E proteins. The results of one representative experiment of at least three performed are shown. A minimum of three ELISAs with three replicates was performed to determine the relative levels of prM and E expression in cell lysates (D) and VLP secretion into the supernatant (E). The WT quantity was taken as 100%. The data are presented as the means ± the SD. An asterisk (*) indicates a statistically significance determined by Student t test at P < 0.05. (F) VLPs with or without myc epitopes were pelleted by ultracentrifugation through a 20% sucrose buffer.

To further determine whether prM-E interactions and the VLP assembly process are affected by prM-E125 mutations, we separately cloned the following prM and E genes into pcDNA3.1 vectors: the prM gene of WT; prM-E125A; prM-E125D; prM-E125; prM-E125Q; prM-E125K; prM-E125R; and the WT E gene fused with a myc epitope at its N terminus. VLP production was reduced by prM-E125A, prM-E125Q, prM-E125K and prM-E125R but not by either the WT or prM-E125D mutant (data not shown). Further, after coimmunoprecipitation with an anti-Myc antibody, the prM-E125A, prM-E125D, prM-E125Q, prM-E125K, and prM-E125R mutants exerted no effects on prM-E heterodimer formation (Fig. 3C). We also confirmed that the myc epitope tag did not interfere with VLP formation (Fig. 3F). Therefore, substituting positively charged amino acids or alanine for prM-E125 significantly impaired VLP production in nonlytic mammalian cells.

prM-H oligopeptides bind with recombinant EDI/II.

To further characterize prM-H domain binding to EDII, we synthesized a series of biotinylated 15-mer oligopeptides (from S114 to I128) comprising the amino acid sequences of the prM-H domain, prM-E125A, prM-E125D, prM-E125Q, prM-E125K, and prM-E125R, plus a scrambled sequence for use as a negative control. Since EDI and EDII are interlinked, we expressed and purified recombinant E domain I-domain II protein (EDI/II) from E. coli (Fig. 4A) to react with each of these seven oligopeptides for ELISA binding. Dose-response binding curves are shown in Fig. 4B. Each of the six prM-H oligopeptides had stronger bindings to EDI/II compared to the scrambled control. These binding curves were further transformed to the Eadie-Hofstee plots (Fig. 5A to F) to calculate the binding dissociation constant (K_d) and the maximum binding value (B_max). The binding affinity of the prM-H125Q, prM-H125K, prM-H125R, and prM-H125A oligopeptides to EDI/II was significantly lower than that of the prM-H and prM-H125D oligopeptides, as shown by a 60% increase in the K_d values (Fig. 5G). The B_max values of prM-H and prM-H125D oligopeptides bound to EDI/II are also 50% higher than the B_max values of prM-H125Q, prM-H125K, prM-H125R, and prM-H125A (Fig. 5G). Therefore, the negatively charged residues prM-H and prM-H125D had stronger interactions to EDI/II than the noncharged prM-H125Q and prM-H125A or the positively charged residues prM-H125K and prM-H125R, suggesting interaction between negatively charged amino acids such as E and D and EDI/II.

FIG 4.

prM-H oligopeptide binding with recombinant EDI/II proteins. (A) Recombinant EDI/II proteins purified from E. coli were resolved using SDS–12% PAGE with Coomassie blue staining (left) or Western blotting using MAb 4G2 (right). (B) A minimum of three ELISAs with three replicates were performed to examine prM-H, prM-H125D, prM-H125Q, prM-H125K, prM-H125R, and prM-H125A oligopeptide binding to recombinant EDI/II. Absorbance at 450 nm is plotted against oligopeptide concentration (μg/ml). Scrambled oligopeptide binding served as background.

FIG 5.

Eadie-Hofstee plots of the ELISA binding curves of prM-H (A), prM-H125D (B), prM-H125Q (C), prM-H125K (D), prM-H125R (E), and prM-H125A (F) oligopeptide to the EDI/II protein shown in Fig. 4 were plotted as the reciprocal of the amount of specific bound oligopeptides (B − B_scrambled) against the reciprocal of oligopeptide concentration: “[P]”. (G) The K_d and B_max values (means ± the SD) were calculated using the Eadie-Hofstee plot.

Positively charged residues at EDII are critical to prM-H binding and VLP production.

To map EDII-binding residues associated with prM-E125, we analyzed the cryo-EM M-E structure of DENV and identified eight positively charged residues at EDII that were conformationally close to the prM-H domain: R57, K93, H214, R215, H219, R234, H246, and H263. Recombinant EDI/II mutant proteins containing alanine substitutions of each positively charged residue (R57A, K93A, H214A, R215A, H219A, R234A, H246A, and H263A) were expressed and purified, and ELISA bindings of prM-H and scrambled oligopeptides to these mutant proteins were measured (Fig. 6). Compared to the scrambled oligopeptide, the prM-H oligopeptide strongly interacted with the WT, H214A, R234A, and H263A EDI/II proteins (Fig. 6A, D, G, and I). No significant differences in ELISA bindings were observed for R57A, K93A, R215A, H219A or H246A (Fig. 6B, C, E, F, and H). To confirm these results, we transfected WT and mutant pcDNA3.1-prM-E plasmids in HEK293 cells, and found that VLP production was impaired by the E-R57A, E-K93A, E-R215A, and E-H246A mutants (data not shown). Further, these four mutants did not affect prM and E expression or prM-E heterodimeric interactions in HEK293 cells (data not shown). These findings indicate the presence of prM-H binding determinants at residues E-R57, E-K93, E-R215, and E-H246 in EDII.

FIG 6.

ELISA analyses of oligopeptide binding to recombinant EDI/II proteins with or without alanine substitutions at EDII. prM-H oligopeptides were interacted with nine recombinant EDI/II proteins—WT (A), R57A (B), K93A (C), H214A (D), R215A (E), H219A (F), R234A (G), H246A (H), and H263A (I), including scrambled oligopeptide as background (n = 3 replicates of three ELISAs). Oligopeptide (A₄₅₀) binding is plotted against the oligopeptide concentration.

Positively charged E-K93 and E-H246 residues interact with prM-E125.

We verified interactions between negatively charged prM-E125 residue and E-R57, E-K93, E-R215, and E-H246 residues at EDII via amino acid exchanges. Five oligopeptides (prM-H125, prM-H125R, prM-H125K, prM-H125H, and scrambled) were reacted with recombinant EDI/II proteins containing E-R57E, E-K93E, E-R215E, and E-H246E mutations. Stronger ELISA binding with the prM-H125K oligopeptide was noted for EDI/II proteins containing the E-K93E mutation, and stronger ELISA binding with the prM-H125H oligopeptide was noted for EDI/II proteins containing the E-H246E mutation, in both cases compared to ELISA binding with the prM-H oligopeptide (Fig. 7B and D). In contrast, EDI/II proteins containing E-R57E and E-R215E mutations had binding affinities similar to those of prM-H and prM-H125R oligopeptides (Fig. 7A and C). This suggests that the electrostatic attraction between the prM-E125 residue and the E-K93 and E-H246 residues contributed to the binding of the prM-H domain to EDII. For confirmation, we transfected WT and mutant pcDNA3.1-prM-E plasmids in HEK293 cells for side-by-side comparison (Fig. 8). The prM and E proteins in cell lysates were not affected by amino acid substitutions or exchanges of prM-E125 and the residues E-R57, E-K93, E-R215, and E-H246 at EDII (Fig. 8A and C). Substantial reductions in VLP production were noted for the prM-E125R/E-R57E and prM-E125R/E-R215E mutants, but only moderate reductions were noted for the prM-E125K/E-K93E and prM-E125H/E-H246E mutants (Fig. 8B and D). The amount of VLPs produced by the prM-E125K/E-K93E mutant was partially rescued compared to the amount of VLP produced by the mutant prM-E125K (P < 0.05) (Fig. 8D). The amount of VLPs produced by the prM-E125H/E-H246E was also partially rescued compared to the amount of VLP produced by the mutant prM-E125H (P < 0.05) (Fig. 8D). In contrast, the prM-E125R/E-R215E mutants reduced the VLP production more severely compared to the prM-E125R (P < 0.05) (Fig. 8D). The results from 0 to 60% continuous sucrose gradient sedimentation indicate similar velocities for all VLP groups; relative VLP amounts were consistent with those shown in Fig. 8B (data not shown). The results indicate that exchanges of oppositely charged residue side chains at prM-E125/E-K93 and prM-E125/E-H246 are recoverable for VLP production, although at levels that are not as efficient as that for the WT. We also evaluated the WT VLPs by electron microscopy to give a size varying from 30 to 50 nm (data not shown). Whether the conformation of the prM-E125/E-K93 and prM-E125/E-H246 mutants is comparable to the WT VLP is still unknown.

FIG 7.

ELISA analyses of oligopeptide binding to recombinant EDI/II proteins with different side chains of charged residues. The bindings of four oligopeptides to four EDI/II proteins are plotted: prM-H/prM-H125R to EDI/II-R57E (A), prM-H/prM-H125K to EDI/II-K93E (B), prM-H/prM-H125R to EDI/II-R215E (C), and prM-H/prM-H125H to EDI/II-H246E (D), including scrambled oligopeptide as background (n = 3 replicates of three ELISAs).

FIG 8.

Dual mutations of charged residues at the prM-H domain and EDII. Western blotting was used to analyze the side-by-side comparison of prM and E protein expression levels in HEK293 cells (A) and VLPs from culture supernatant collected by ultracentrifugation through a 20% sucrose buffer (B). The results of one representative experiment of at least three performed are shown. (C) Three ELISAs with three replicates were performed to determine the relative levels of prM and E expression in cell lysates. (D) The relative amount of VLP secretion into supernatant. WT quantity was taken as 100%. The data are presented as means ± the SD. An asterisk (*) indicates a statistically significance determined by Student t test at P < 0.05.

We further conducted a time course study to measure the VLP production of WT, prM-E125K, prM-E125H, prM-E125K/E-K93E, and prM-E125H/E-H246E. Each of the VLPs produced in culture supernatants was determined by E protein content using ELISA. The WT and mutant VLPs produced from HEK293 cells were detected at 24 h posttransfection and continued to increase at 72 h posttransfection (Fig. 9). The VLP production of prM-E125K and prM-E125H mutants was significantly reduced compared to the WT, with the amounts of secreted E protein content decreased to 40% (Fig. 9). The VLP production of E-K93E mutant was lower than the prM-E125K and prM-E125H mutants, with the amounts of secreted E protein content decreased to 30% compared to the WT (Fig. 9). The amount of VLPs produced by the E-H246E mutant more severely decreased to 10% compared to the WT (Fig. 9). In contrast, the exchanges of oppositely charged residue side chains at prM-E125/E-K93 and prM-E125/E-H246 resulted in restoring the VLP production, with the amount of secreted E protein content increased to 60 to 70% at 72 h posttransfection (Fig. 9). The statistical analysis was performed to determine whether dual mutants are significantly different from the single mutants. As shown in Fig. 9, the dual mutants (prM-E125K/E-K93E and prM-E125H/E-H246E) had significantly higher levels of VLP production than all of the single mutants (prM-E125K, prM-E125H, E-K93E, and E-H246E) at 48, 60, and 72 h posttransfection (P < 0.05). The time course study confirmed that the prM mutant with a positive charge at position 125 can be partially rescued for VLP production by compensatory mutations (positive to negative) at positions 93 or 246 on E protein.

FIG 9.

Time course VLP production of prM-H, single mutants (prM-E125K, prM-E125H, E-K93E, and E-H246E), and dual mutants (prM-E125K/E-K93E and prM-E125H/E-H246E). The VLP secretion into supernatant was pelleted by ultracentrifugation through a 20% sucrose buffer. Three ELISAs with three replicates were performed to determine the amount of pelleted VLP. The WT quantity at 72 h was taken as 100%. An asterisk (*) indicates significantly higher levels of VLP production than all of the single mutants (prM-E125K, prM-E125H, E-K93E, and E-H246E) at 48, 60, and 72 h posttransfection determined by Student t test at P < 0.05.

Sequence alignment analyses of prM-E125, E-K93, and E-H246 in different flavivirus groups indicate conserved residues.

To determine whether prM-E125, E-K93, and E-H246 residues were conserved, we analyzed the amino acid sequence alignments of the prM-H domains (Table 1) and EDIIs (Table 2) of 22 flaviviruses, including 5 JEV serocomplex viruses (JEV, KUNV, MVEV, SLEV, and WNV), 4 serotype DENVs (DENV-1, DENV-2, DENV-3, and DENV-4), TBEV, and YFV. All sequences were aligned using the CLUSTAL W online program default settings (http://www.ebi.ac.uk/Tools/msa/clustalw2/). The results indicate that the prM-E125 and E-H246 residues were completely conserved in all 22 flaviviruses (Tables 1 and 2) and that the E-K93 residue was conserved in 17. A different positively charged arginine (R) amino acid was found in WNV, DENV-1, and DENV-4 (Table 2). According to these results, electrostatic charges at the prM-E125, E-K93, and E-H246 residues are preserved in different flavivirus groups.

TABLE 1.

Multiple alignments of amino acid sequences in the prM-H domains of 22 flavivirus strains^a

^aConsensus amino acids are indicated by dashes. The prM-H region is shaded; prM-E125 is labeled with white dashes in the black band.

TABLE 2.

Multiple alignments of EDII amino acid sequences in 22 flavivirus strains^a

^aFor simplicity, only regions around the E-K93 and E-H246 residues of JEV are shown. Consensus amino acids are indicated by dashes. E-K93 and E-H246 residues are labeled with white color in black or gray bands.

DISCUSSION

Our motivation was to map interactions between prM-H domain and EDII residues associated with JEV VLP production. We used oligopeptides synthesized from prM-H sequences and investigated their bindings to EDI/II recombinant proteins and the expression of full-length prM and E genes in cis or in trans in insect and mammalian cells. Our results indicate that negatively charged prM-E125 residue interacted with positively charged E-K93 and E-H246 residues at EDII. In addition to observing that exchanges of charge-charge amino acids at the prM-E125, E-K93, and E-H246 residues resulted in the recovery of VLP production in cultured cells, we noted that prM-E125 and E-H246 residues are conserved and that the positive charge of the E-K93 residue is preserved in different flavivirus groups. Combined, these findings suggest that the electrostatic attractions of prM-E125, E-K93, and E-H246 residues are important to flavivirus VLP production.

Specific amino acid residues located at the pr and M proteins of flaviviruses have been shown to affect virus particle assembly. The list includes prM-Y78 at the pr protein of WNV (19), prM-H99 at the N-terminal loop of the M protein (7), prM-120, prM-123, and prM-127 at the prM-H domains of DENV-1, DENV-2, and DENV-4 (17), and prM-H130 between the prM-H and TM domains of DENV2 (20). The GXXXG motif of the JEV prM-TM domain is also involved in VLP production and prM-E interaction (7, 21). Four DENV residues are located near the C-terminal end of the prM-H domain: prM-120, prM-123, prM-127, and prM-130. For the present study, we used helical wheel projection to analyze prM-H domain amino acid sequences and made residue substitutions for prM-S114, prM-A117, prM-L121, prM-E125, and prM-I128 at the prM-H domain. According to our results, the prM-E125A, prM-E125Q, prM-E125K, and prM-E125R mutations significantly impaired VLP production. This finding is in agreement with a report indicating that C-terminal residues at the DENV prM-H domain are important for VLP production (17). Also consistent with that finding are our data indicating that none of our prM-H domain mutations affected prM-E heterodimeric interactions. C-terminal residue mutants at the prM-H domain of DENV-4 have been shown to inhibit VLP budding in the lumen of the endomembrane system instead of decreasing prM-E heterodimeric interaction (17). Our results suggest that the prM-E125 residue of JEV is involved in the assembly of membrane-bound prM-E heterodimers or complexes for forming macromolecular VLPs during ER membrane budding.

We found that negatively charged prM-E125 residues are important for interaction between prM-H oligopeptides and EDI/II proteins (Fig. 4B). Changes in the surface electrostatic charges of the DENV-2 E protein have been shown to occur as pH declines from 8.0 to 6.0, including increases in the surface positive charges of E proteins (22). This E protein change may inhibit resistance to the negatively charged prM-E125 and favor prM-H domain binding. Further, we identified prM-E125 interacting residues E-K93 and E-H246 at EDII (Fig. 6C and H) and confirmed their interactions by exchanging them with residues prM-E125/E-K93 or prM-E125/E-H246 (Fig. 7B and D). Interactions between prM-E125/E-K93 and prM-E125/E-H246 also affected VLP production (Fig. 8B and D). Negatively charged prM-E125 and positively charged E-K93 or E-H246 residues may interact via electrostatic attraction, which would be consistent with a report that the DENV-2 prM-H domain has greater binding affinity with the E protein and virus membrane under low-pH conditions (22). Based on the crystal structure of JEV E protein ectodomain (PDB 3P56) and DENV prM-H domain (PDB 1P58), the residues E-K93 and E-H246 are located close at the side facing the prM-H domain, indicating the interactions between residues prM-E125 and E-K93 and E-H246 (Fig. 10). The Rossmann group has reported the DENV E-pr peptide complex crystal structure but the structure does not contain the membrane proximal region of prM, including the N-terminal loop and prM-H domain (13). The C-terminal part of pr and the N-terminal loop of M undergo a huge conformational change at low pH during virus maturation process (13). According to a recent cryo-EM structure of the mature DENV (15), the N-terminal loop of M can act like a “drawstring” to pull down E ectodomain and latch it to the E dimer at low pH during virus maturation process, resulting in shortened distance between prM-H domain and EDII. Associations among the E protein, prM-H domain, and virus membrane have been suggested as modulating the movement of structural proteins during DENV-2 maturation (22). The electrostatic attraction between prM-E125 and either E-K93 or E-H246 residues may affect structural protein rearrangement during virus maturation. According to our multiple sequence alignment results for the prM-H domain and EDII (Tables 1 and 2), the prM-E125 and E-H246 residues are conserved and the positive charge at residue E-K93 is preserved in different flavivirus groups, suggesting that the equivalent residues of other flaviviruses may interact with each other via electrostatic attractive force, thus facilitating VLP or virion production. Since VLPs are much smaller and have only 1/3 of the number of E/prM molecules compared to a virion with a capsid (11), the effects of these mutations prM-E125, E-K93, and E-H246 on virion production require further investigation using infectious clone technology.

FIG 10.

Location of eight positively charged residues at EDII based on the crystal structure of JEV E protein ectodomain (PDB 3P56) and the DENV prM-H domain (PDB 1P58). The eight positively charged residues and prM-E125 residue are shown in the sphere. The prM-H domain is indicated in cyan. The interactions between prM-E125/E-K93 and prM-E125/E-H246 are represented as dotted red lines.

In conclusion, we noted interactions between the negatively charged prM-E125 residue at the prM-H domain and positively charged E-K93 and E-H246 residues at EDII and suggest that this interaction is vital to VLP production. These residues may play similar roles in different flavivirus groups. Inhibiting these interactions is a potential strategy for blocking flavivirus particle production.