A Proline-Rich N-Terminal Region of the Dengue Virus NS3 Is Crucial for Infectious Particle Production

ABSTRACT

Dengue virus is currently the most important insect-borne viral human pathogen. Viral nonstructural protein 3 (NS3) is a key component of the viral replication machinery that performs multiple functions during viral replication and participates in antiviral evasion. Using dengue virus infectious clones and reporter systems to dissect each step of the viral life cycle, we examined the requirements of different domains of NS3 on viral particle assembly. A thorough site-directed mutagenesis study based on solvent-accessible surface areas of NS3 revealed that, in addition to being essential for RNA replication, different domains of dengue virus NS3 are critically required for production of infectious viral particles. Unexpectedly, point mutations in the protease, interdomain linker, or helicase domain were sufficient to abolish infectious particle formation without affecting translation, polyprotein processing, or RNA replication. In particular, we identified a novel proline-rich N-terminal unstructured region of NS3 that contains several amino acid residues involved in infectious particle formation. We also showed a new role for the interdomain linker of NS3 in virion assembly. In conclusion, we present a comprehensive genetic map of novel NS3 determinants for viral particle assembly. Importantly, our results provide evidence of a central role of NS3 in the coordination of both dengue virus RNA replication and particle formation.

IMPORTANCE Dengue virus is an important human pathogen, and its prominence is expanding globally; however, basic aspects of its biology are still unclear, hindering the development of effective therapeutic and prophylactic treatments. Little is known about the initial steps of dengue and other flavivirus particle assembly. This process involves a complex interplay between viral and cellular components, making it an attractive antiviral target. Unpredictably, we identified spatially separated regions of the large NS3 viral protein as determinants for dengue virus particle assembly. NS3 is a multifunctional enzyme that participates in different steps of the viral life cycle. Using reporter systems to dissect different viral processes, we identified a novel N-terminal unstructured region of the NS3 protein as crucial for production of viral particles. Based on our findings, we propose new ideas that include NS3 as a possible scaffold for the viral assembly process.

INTRODUCTION

The dengue virus (DENV) belongs to the Flavivirus genus of the Flaviviridae family, together with other important emerging and reemerging human pathogens, such as Zika virus (ZKV), West Nile virus (WNV), Japanese encephalitis virus (JAEV), Saint Louis encephalitis virus (SLEV), yellow fever virus (YFV), and tick-borne encephalitis virus (TBEV) (1–3). DENV comprises four closely related serotypes (DENV-1 to -4), and all of them can produce disease ranging from mild dengue fever to severe dengue, a potentially lethal hemorrhagic and capillary leak syndrome. Found in tropical and subtropical regions of the world, dengue is the most significant viral disease transmitted by arthropods. It is estimated to cause 390 million infections per year and places over 3 billion people at risk of infection (4). In spite of this great burden, fundamental aspects of its viral biology remain elusive, and effective therapeutics are still unavailable (5).

The flavivirus genome is a single RNA molecule of about 11 kb that carries one open reading frame (ORF), flanked by highly structured 5′ and 3′ untranslated regions (UTRs). The single polyprotein is processed co- and posttranslationally to give three structural proteins (capsid [C], premembrane protein [prM], and envelope [E]) and seven nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5). The structural proteins integrate the viral particle, while the nonstructural proteins support replication of the viral RNA and virion assembly and limit the host antiviral response (6–11). Although great advances have been made in recent years to understand the mechanism of viral genome replication (12–16), less is known about the process that leads to genome encapsidation during viral assembly (17–20).

Viral particle formation requires the assembly of a nucleocapsid integrated by the viral genome and multiple copies of the C protein. The process of genome recruitment by C has been proposed to take place near RNA replication complexes associated with endoplasmic reticulum (ER) membranes (18, 21); however, the specifics of the viral and cellular components of the assembly machinery are still unclear, as are the mechanistic details of how it works. A number of reports have involved the NS1, NS2A, and NS3 flavivirus proteins as components of this machinery (22–27), but how the interplay between these components and the host cell occurs remains uncertain. Using YFV, a genetic link was first uncovered between NS2A and NS3 for viral particle assembly. A mutation in NS2A that impairs particle formation was rescued by a second site mutation in NS3 (22). Further analysis confirmed an assembly requirement of specific residues in the helicase domain of YFV NS3 protein (23). NS3 is a multifunctional protein with activities involved in polyprotein processing, viral RNA replication, and host immune evasion (20, 28–39). The protein bears an N-terminal serine protease domain (residues 1 to 169) and a C-terminal region containing RNA helicase, nucleoside triphosphatase (NTPase), 5′-RNA triphosphatase (RTPase), and RNA annealing activities (residues 179 to 618) (32, 40–42). Because of its essential role early in polyprotein processing and RNA replication, genetic studies thus far have provided limited information on the molecular determinants of NS3 responsible for assembly and release of infectious virus particles.

Here, we used a DENV reporter system that allows the dissecting of each step of the viral life cycle to investigate the involvement of DENV NS3 in viral particle assembly. A systematic mutagenesis study taking into account three-dimensional structures of NS3 and its individual domains, as well as solvent-accessible surface areas, identified novel determinants for DENV particle assembly.

MATERIALS AND METHODS

Cell culture.

Baby hamster kidney cells (BHK-21) were maintained in minimum essential medium (MEM) alpha supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 0.10 mg/ml streptomycin (Gibco). Cells were grown in an incubator at 37°C and 5% CO₂.

Plasmids.

Monocistronic dengue virus reporter construct (mDV-R) containing the Renilla luciferase (Rluc) coding sequence was previously described (43). This construct contains the Rluc reporter gene just downstream of the minimal cis-acting elements (MCAE) of the viral ORF, followed by an FMDV-2A autoproteolytic protease sequence in order to release the Rluc enzyme from the viral polyprotein. A plasmid containing the DENV-2 strain 16681 infectious cDNA clone also was used (GenBank accession number U87411) (44).

Cloning and site-directed mutagenesis.

The desired mutations within the coding sequence of NS3 were introduced into the mDV-R cDNA clone by replacing the KpnI-NsiI, NsiI-XhoI, or XhoI-NheI fragment of the wild-type (WT) plasmid with the respective fragment derived from overlapping PCR using primers listed in Table S1 in the supplemental material. Selected NS3 mutations were also introduced in the plasmid containing the DENV-2 16681 infectious clone by double digestion and T4 DNA ligation of the respective mDV-R clone. Pfu DNA polymerase was obtained from FIL. Restriction enzymes, T4 DNA ligase, and Antarctic phosphatase were purchased from New England BioLabs Inc. DNA sequences were confirmed by automatic DNA sequencing using a 3130 genetic analyzer (Applied Biosystems).

RNA transcription.

Plasmids were linearized with XbaI and used as transcription templates. DENV genomic RNAs were obtained by in vitro transcription using T7 RNA polymerase (Ambion) in the presence of an m⁷G(5′)ppp(5′)A RNA Cap structure analog (S1405L; New England BioLabs Inc.) as previously described (45). In cases where transfected cells were used in quantitative real-time reverse transcription-PCR (RT-qPCR), the DNA templates were removed from the reaction mix by digestion with Turbo DNase (Ambion). RNA integrity was confirmed by agarose gel electrophoresis.

RNA transfections with reporter virus RNAs.

RNA transfections were performed with Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen). BHK cells were seeded in 24-well plates at a density of ∼4 × 10⁴ cells per well. The next day, the semiconfluent (∼70%) cellular monolayers were transfected with 500 ng per well of mDV-R RNA transcripts. After 3 h of incubation, RNA/Lipofectamine fluids were removed and replaced with 0.5 ml of fresh culture medium. The Rluc activity present in mDV-R-transfected cells was analyzed at 6, 24, 48, and 72 h posttransfection (hpt) from cell extracts using the Renilla luciferase assay system according to the manufacturer's instructions (Promega). Each reporter virus was assayed in two separated wells for every time point measured. At least three independent experiments were performed for each virus. The supernatant fluids from respective wells at 72 hpt were collected and stored at −70°C for subsequent release infectivity analysis. WT, replication-defective (NS5), and encapsidation-defective (ΔC) reporter viruses were added as controls to these assays (43, 46).

Release infectivity assay of reporter virus.

In order to quantify the release infectivity for each NS3 mutant of the reporter virus, these supernatants were thawed and applied (0.2 ml/well) to naive BHK cells grown under the same conditions described above. The Rluc activity present in the infected cells was analyzed from cell extracts at 1 and 3 days postinfection (dpi) using the Renilla luciferase assay system (Promega).

IFA.

For indirect immunofluorescence assay (IFA), BHK cells were seeded into 24-well plates containing glass coverslips. The next day, the cells grown at ∼70% confluence were transfected as described above with WT DENV-2 infectious clone RNA and selected NS3 mutants that solely impaired virus propagation in the reporter virus. At 2 days posttransfection, supernatant fluids were harvested and stored at −70°C, while coverslips were collected and the cells were fixed with 4% paraformaldehyde, 4% sucrose in phosphate-buffered saline (PBS), pH 7.4, at room temperature for 15 min. Coverslips were washed with PBS, and then fixed cells were permeated with 0.1% Triton X-100 for 4 min at room temperature. For the detection of DENV-infected cells, mouse monoclonal anti-E antibody E18 (dilution of 1:200 in PBS with 0.2% gelatin) was used (47). Secondary staining was carried out using a 1:500 dilution of Alexa 488-conjugated anti-mouse IgG antibody (Invitrogen Inc.). Nuclear DNA was stained with a 1:1,000 dilution of 4′,6-diamidino-2-phenylindole (DAPI) (Molecular Probes, Karlsruhe, Germany). Coverslips were mounted with Mowiol 4-88 (Sigma). Images were obtained with a Carl Zeiss AXIO Imager.A2 fluorescence microscope at ×100 magnification. The supernatant fluids collected at 2 days posttransfection were thawed and applied to naive BHK cells. After 1 and 4 days postincubation, DENV-infected cells (green fluorescence) were analyzed by IFA in order to estimate the release infectivity for each virus.

WB.

For Western blotting (WB) assays, the following antibodies and antisera were used. The immunodetection of E was performed using the specific monoclonal antibody E18 (47). Rabbit polyclonal anti-capsid serum was obtained previously (43). Rabbit polyclonal anti-NS3 serum was obtained in our laboratory. The specificity of this antiserum was evaluated by enzyme-linked immunosorbent assay (ELISA) and Western blotting employing DENV-infected and noninfected BHK cell extracts as controls. Mouse monoclonal anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 6C5) was used for a loading control (Abcam). Horseradish peroxidase (HRP)-conjugated donkey anti-mouse IgG and HRP-conjugated goat anti-rabbit IgG sera were used as secondary antibodies (Sigma).

For the detection of intracellular C, NS3, E, and GAPDH, BHK cells were grown in 24-well plates and transfected with the WT and mutant DENV-2 infectious clone as described above. At 2 and 3 days posttransfection, supernatant fluids were recovered and store on ice, while transfected cells were suspended by trypsinization, washed once with PBS supplemented with 2% FBS, and then washed twice with PBS. Cells were pelleted and stored at −70°C. Cells were thawed and denatured in buffer S (1% sodium dodecyl sulfate [SDS], 10 mM Tris, pH 6.8, and 5% glycerol) at 95°C for 5 min. Samples were analyzed under denaturing conditions by SDS-PAGE, transferred to Amersham Hybond P polyvinylidene difluoride (PVDF) membranes (0.45-μm pore size; GE-Healthcare Life Sciences), and blocked overnight with 5% low-fat milk in TBS–0.1% Tween 20 at 4°C, and Western blot analyses were performed using the specific antibodies mentioned above. For the detection of secreted E and C protein in transfected cells, supernatant fluids obtained 2 and 3 dpt were centrifuged at 4°C for 15 min at 15,000 × g to clear cellular debris. Aliquots of 50 μl were denatured and analyzed by WB as described above.

RNA extraction, reverse transcription, and real-time PCR.

The intracellular extracts and supernatant fluids of cells transfected with WT or NS3-mutated viruses were obtained as described above. Total RNA from samples was extracted with TRIzol by following the manufacturer's recommendations (Invitrogen). RNA extracted from experimental samples was reverse transcribed in 20-μl reaction volumes using 160 U Moloney murine leukemia virus (M-MLV) RT (Promega), 4 U of RNasin RNase inhibitors (Promega), 500 nM primer AVG1118 (5′-GCCGCACCATTGGTCTTCTC-3′), and 0.5 mM each deoxynucleoside triphosphate (dNTP), all in 1× M-MLV RT buffer. Primer and RNA template were preincubated for 5 min at 70°C. The RT reaction was allowed to proceed for 1 h at 42°C and then for 15 min at 65°C to denature the RT enzyme. For quantitative real-time PCR, an Mx3005P qPCR system (Agilent Technologies Inc.) was employed. Reactions were performed in triplicate in 96-well plates using 2 μl of the RT reaction as the template (1:10 dilution in RNase-free water), 5 μl FastStart Universal SYBR green master (Rox) 2× mix (Roche), 300 nM each primer, and RNase-free water to 10 μl. The primers AVG1117 (5′-ACAAGTCGAACAACCTGGTCCAT-3′) and AVG1118 were targeted to amplify nucleotides 9937 to 10113 within the NS5 coding sequence. Reactions were run with the following parameters: 95°C for 15 min and then 40 cycles of 95°C for 10 s and 60°C for 30 s. Fluorescence detection was acquired during the elongation step of each cycle. For each reaction the initial amount of template was calculated from the threshold cycle (C_T) value using a standard curve generated from serial dilutions of reverse-transcribed and purified genomic RNA.

Virus plaque- and focus-forming assays.

Supernatant fluids from WT and NS3 mutant DENV-transfected cells were harvested at 3 days posttransfection and stored at −70°C. These fluids were used to infect fresh cells in order to quantify viral titers. Viral titers were quantified by plaque titration assay as described previously (45). Briefly, supernatant fluids were serially diluted, added to fresh cells in 24-well plates, and incubated for 1 h. Afterwards, 1 ml of overlay medium (culture medium supplemented with 0.8% methyl cellulose) was added to each well. Cells were fixed 7 days postincubation by addition of 0.5 ml 10% formaldehyde per well, incubated for 1 h, thoroughly washed with water, and stained with crystal violet. For the focus-forming assay, supernatant fluids were serially diluted, and 0.20 ml was added to fresh cells grown in 24-well plates and incubated for 1 h. Afterwards, 1 ml of overlay medium was added to each well. At 5 days postincubation, overlays were washed away and cells were fixed with 0.5 ml methanol for 20 min at 4°C. Fixed cells were washed with Tris-buffered saline (TBS), pH 7.4, and incubated for 2 h in blocking buffer (TBS supplemented with 0.25 mM EDTA, 1% BSA, and 0.1% Tween 20). Two hundred microliters of mouse monoclonal anti-E antibody E18 (1:500 dilution in blocking buffer) then was added to each well and incubated for 1 h at room temperature. After three washes with blocking buffer, 200 μl alkaline phosphate-conjugated anti-mouse IgG antibody (1:2,500 dilution in blocking buffer) was added and incubated for 1 h. After thorough washes with TBS, BCIP-NBT (5-bromo-4-chloro-3-indolylphosphate–nitroblue tetrazolium) color development substrate was added as indicated by the manufacturer (Promega).

RESULTS AND DISCUSSION

Based on the X-ray crystal structure determined for the DENV-2 NS3hel domain (PDB code 2BMF [48]), the solvent-accessible surface area for each amino acid residue was computed using the GetArea program (49). Residues with side chains that point away from the NS3hel core domain were identified. Amino acids within known functional/conserved motifs for enzymatic activities were excluded (34). Double or triple mutations to alanine of contiguous (or very close) solvent-exposed residues were tested. Additionally, mutant W₃₄₄A, corresponding to the homologue W₃₄₉A substitution previously identified in the NS3hel of YFV as relevant for viral assembly (23), was included in the analysis (Fig. 1A). A ribbon diagram representation of the NS3 protein structure indicating the 3D location of the selected residues was generated using Pymol (DeLano Scientific) (Fig. 1B).

FIG 1.

DENV helicase domain of NS3 is involved in infectious particle production. (A) Sequence of DENV-2 NS3 helicase domain (residues 180 to 618) of strain 16681 is shown, with the helicase superfamily 2 conserved motifs in boxes and the amino acids involved in the mutagenesis analysis in red. (B) Ribbon diagram representation of the X-ray crystal structure of the DENV NS3 protein (PDB accession number 2VBC) (52). The helicase domain is formed by three subdomains (shown in bluish tones). The protease domain is formed by the N-terminal third of the NS3 polypeptide with part of NS2B cofactor (shown in orange and yellow, respectively). The interdomain linker region (residues 170 to 179) is shown in green. Residues targeted by site-directed mutagenesis are indicated in red. (C) Translation and replication levels of the Renilla luciferase (Rluc) reporter system mDV-R containing the engineered mutations in the helicase domain of NS3. Rluc activity was measured at the indicated times in transfected cells with viral RNAs corresponding to the WT reporter virus, the replication-impaired NS5 mutant, the propagation-impaired ΔC mutant, and the engineered NS3 mutant. Data correspond to the averages from three independent experiments. Error bars point out standard deviations. The red dashed line indicates background levels of the signal. au, arbitrary units. (D) Released infectivity from mDV-R-transfected cells. Culture supernatants (0.25 ml) collected at 72 hpt from transfected cells shown in panel C were used to inoculate fresh cells, and then luciferase activity in these cells was measured at 1 and 3 days postincubation. (E) Multiple-sequence alignment of N₅₇₆ and surrounding residues of the NS3 proteins from the four DENV serotypes and other flaviviruses. The triple mutation of the solvent-accessible conserved residues E₅₇₄, E₅₇₅, and N₅₇₆ (shaded blue) for alanine abrogates the release infectivity of reporter virus. Asterisks, colons, and dots indicate identical, conserved, and partially conserved residues, respectively. (F) Translation and replication levels in RNA-transfected BHK-21 cells of the reporter virus carrying single-residue mutations of E₅₇₄A, E₅₇₅A, and N₅₇₆A. WT and triple-mutant data are also included for ease of comparison. (G) Released infectivity assays of the culture supernatants collected at 72 h posttransfection from each mutant shown in panel F.

We designed and constructed six recombinant viruses with mutations in the NS3hel domain and assessed replication and infectivity using a Renilla luciferase reporter DENV-2 (mDV-R), competent for replication and production of infectious particles (43, 46). Mutations were introduced in the cDNA clone by overlapping PCR (using primers listed in Table S1 in the supplemental material). RNA transcriptions and transfections were carried out as previously reported (43, 46). Measurements of luciferase activity expressed from the mDV-R RNA-transfected cells were performed at 6, 24, 48, and 72 hpt. The first rounds of RNA translation were evaluated at 6 h and RNA replication at 24, 48, and 72 h (Fig. 1C). Four controls were used, the parental viral RNA (WT), a replication-impaired RNA with a mutation in the viral polymerase (NS5), a propagation-impaired control with a deletion of the capsid protein (ΔC), and a mock transfection. To examine secreted infectious particles, the culture supernatants from each case were collected at 72 h and used to infect fresh BHK-21 cells, and luciferase was monitored as a function of time postinfection (Fig. 1D). As expected, transfection of the WT reporter virus RNA resulted in efficient viral replication and production of infectious viral particles (Fig. 1C and D, WT). The replication control (NS5) showed high levels of translation (6 h) but no RNA amplification, while the ΔC control showed WT levels of translation and RNA replication but undetectable infectious particles (Fig. 1C and D, NS5 and ΔC). All of the mutations introduced in NS3hel diminished viral replication to some extent (Fig. 1C). In contrast to that observed in YFV, alteration of the homologous and conserved tryptophan residue to alanine (W₃₄₄A) in DENV NS3hel resulted in a drastic reduction of viral replication, and consequently an undetectable production of viral particles, indicating a different requirement for the replication/assembly machinery of these two related viruses (Fig. 1C and D, W₃₄₄A). The triple-residue mutant P₅₀₁A+E₅₀₂A+I₅₀₄A showed a drastic reduction in replication, comparable to the profile obtained for the NS5 control, suggesting that these residues were crucial for RNA amplification. Mutants V₂₇₃A+R₂₇₄A+P₂₇₆A, K₂₁₃A+R₂₁₄A, and R₅₂₆A+E₅₂₈A+K₅₃₁A presented significant reduction in viral replication at different degrees (Fig. 1C). As a consequence of these defects in replication, a major reduction or impediment in virus infectivity was observed for these three mutants (Fig. 1D).

The mutation E₅₇₄A+E₅₇₅A+N₅₇₆A produced only a slight delay in viral replication. At 72 h posttransfection, the luciferase activity was similar to that observed for the WT (Fig. 1C). Notably, this substitution drastically decreased the production of infectious viral particles (>10,000-fold less than that for the WT) (Fig. 1D). Alignments using sequences from different flaviviruses revealed that these amino acids are highly conserved (Fig. 1E). To further examine the possible relevance of these residues on viral particle assembly, they were individually mutated in the reporter virus. Translation and replication of the three mutants carrying each single substitution (E₅₇₄A, E₅₇₅A, or N₅₇₆A) were indistinguishable from the WT (Fig. 1F). Interestingly, while mutant E₅₇₅A produced a moderate reduction in infectivity (10-fold reduction of Rluc activity at 3 dpi), the N₅₇₆A mutation greatly reduced (more than 1,000-fold) the production of infectious particles (Fig. 1G). These results indicate that the conserved residue N₅₇₆, located at subdomain III of NS3hel, performs an important function in viral assembly without reducing the enzymatic activities involved in the first steps of the viral life cycle.

The linker region (residues 170 to 178 in NS3 of DENV-2) that naturally tethers the protease and helicase domains is particularly interesting because it allows NS3 to acquire different conformations relevant for the diverse functions of the protein (50, 51). It has been reported that increasing the linker flexibility led to a significant reduction in the ATPase and helicase activities in vitro and a decrease in viral RNA replication in cell culture (50). Crystallographic studies carried out with the DENV-4 NS3 solved two structures for the same protein crystallized under different conditions (50, 52). In both cases the protease and helicase domains were superimposed separately as independent entities, and their structures were consistent with those previously obtained for the isolated protease and helicase domains (48, 50, 53). Interestingly, while the full-length NS3 molecule adopted elongated shapes in both conformations, they diverged in a 161° rotation between the domains, which correlated with a reorientation of segment E₁₇₇-D₁₇₉ in the linker (50). Based on the structural/functional relevance of this linker region, we performed a systematic mutational analysis to assess its possible involvement in viral assembly.

Sequence alignments of the NS3 interdomain region obtained from the four DENV serotypes was found not to be well conserved (Fig. 2A). Three triple-residue substitutions comprising the complete linker first were designed in the context of DENV-2 (Fig. 2B and C). Mutants K₁₇₀A+S₁₇₁A+I₁₇₂A and E₁₇₃A+D₁₇₄A+N₁₇₅A exhibited replication and infectivity levels similar to those of the WT, whereas mutant P₁₇₆A+E₁₇₇A+I₁₇₈A showed a very slight delay in replication (Fig. 2B) but a drastic reduction in infectivity (Fig. 2C). We further examined the cause of this phenotype by designing individual substitutions. Mutants P₁₇₆A and E₁₇₇A exhibited no defect in replication and slight reductions in infectivity. However, mutant I₁₇₈A showed a slight delay in RNA replication but a profound (>10,000-fold) reduction in infectivity (Fig. 2D and E). The absence of major defects in replication by changing each residue of the linker supports the hypothesis that this region is only marginally implicated in NS3 enzymatic activities but plays an important role in infectious particle production, presumably by modulating overall NS3 conformation. The delay in replication of the I₁₇₈A mutant is in agreement with the functional shortage observed in DENV-4 NS3 linker mutants that increased protein flexibility (50). Our results are consistent with the idea that flexibility of the linker contributes to modulating the multifunctional activities of NS3, likely through allosteric interactions.

FIG 2.

Identification of the linker region that tethers the protease and the helicase domains of NS3 as a critical determinant for DENV propagation. (A) Multiple-sequence alignment of the NS3 interdomain region from the four DENV serotypes and other flaviviruses. Based on the available structure of full-length NS3 comprising both protease and helicase domains from DENV-4, we demarcate the interdomain linker region (highlighted in light brown). (B) Translation and replication capabilities of the reporter viruses containing mutations in the linker region of the NS3 protein were assayed as described for Fig. 1. The WT was also included for ease of comparison. The dashed line indicates background levels of luciferase activity. (C) Released infectivity measurements of the culture supernatants collected at 72 h posttransfection from DV-R RNA-transfected cells shown in panel B. (D) Translation and replication capabilities of the reporter viruses carrying single-residue mutations of P₁₇₆A, E₁₇₇A, and I₁₇₈A in DV-R-transfected cells. (E) Released infectivity measurements at the indicated time after transfection of the culture supernatants collected from cell cultures shown in panel D at 72 h posttransfection.

To complete a systematic analysis, we also examined a possible role of the NS3 protease domain in viral infectivity, which has never been explored before for viral assembly. Structure-based mutations were designed to substitute alanine for solvent-exposed side chain residues, avoiding those included in catalytic motifs (Fig. 3A). The solvent-accessible surface area for each residue within the NS3 protease domain was calculated from the X-ray crystal structure of the NS2B-NS3 protease domain (PDB code 2FOM). Seven double-residue mutants were designed and constructed in the context of the mDV-R system. Overall, the mutated sites were distributed evenly over the surface of NS3pro, as depicted in Fig. 3B. In addition, a mutation that replaced two proline residues in a proline-rich region at the intrinsically unstructured N-terminal region of the NS3 protein was also included (53).

FIG 3.

Proline-rich N-terminal region of the NS3 protease domain is a novel determinant for DENV infectious particle formation. (A) Sequence of the N-terminal region of DENV-2 NS3 comprising the protease domain (residues 1 to 169). Conserved motifs of chymotrypsin-like protease are indicated in boxes. The mutated residues are shown in red. The interdomain linker region is highlighted in green. (B) Ribbon representation of the X-ray crystal structure of the protease domain of DENV-2 NS3 protein generated using Pymol (DeLano Scientific). The N-terminal third of the NS3 polypeptide, in orange, together with part of NS2B, in yellow, form the protease domain. The helicase domain is partially shown (in blue). The interdomain region (residues 170 to 178) is indicated in green. Double-residue mutations are indicated in red. The 18 residues of the unstructured N-terminal region of the protease domain are depicted as a dashed line. (C) Translation and replication capabilities of the luciferase reporter system mDV-R containing mutations in the protease domain of NS3 protein. Rluc activity was measured as a function of time after transfection in BHK-21 cells of genomic RNAs. WT reporter virus and controls were also included as indicated in the legend to Fig. 1. Data correspond to averages from three experiments. Error bars point out standard deviations. The dashed line indicates background levels of signal. (D) Released infectivity from mDV-R-transfected cells. Culture supernatants collected at 72 h posttransfection from DV-R RNA-transfected cells shown in panel C were used to infect fresh cells, and then luciferase activity in these cells was measured at 1 and 3 days postinfection. (E) Multiple-sequence alignment of the NS3 protease from flaviviruses showing the N-terminal region (left, residues 1 to 25) and a beta-hairpin region (right, residues 53 to 75). The double mutation of P₁₀ and P₁₂ as well as K₆₃ and R₆₄ (highlighted in gray) for alanine residues abrogated the release infectivity of reporter virus. (F) Translation and replication capabilities of the reporter viruses that carry a single-residue replacement of L₄A, W₅A, D₆A, V₇A, P₈A, S₉A, P₁₀A, P₁₁A, P₁₂A, K₆₃A, and R₆₄A in RNA-transfected BHK-21 cells. (G) Release infectivity of the culture supernatants collected at 72 h posttransfection from DV-R RNA-transfected cells shown in panel F.

Mutants E₈₈A+K₉₀A and K₁₄₂A+K₁₄₃A showed replication and infectivity levels similar to those of the WT (Fig. 3C and D). Mutant P₁₀₂A+K₁₀₄A exhibited a drastic reduction in luciferase levels similar to that of the NS5 control, indicating that these residues are critical for DENV-2 replication. Mutant K₁₁₇A+N₁₁₉A showed a significant reduction in viral replication and, probably as a consequence of this, a major reduction in virus production/infectivity (Fig. 3C). Mutant R₁₅₇A+S₁₅₈A exhibited no defect in replication but an ∼100-fold decrease in infectivity (Fig. 3C). Unexpectedly, mutants P₁₀A+P₁₂A and K₆₃A+R₆₄A exhibited no defect in viral replication, but infectious particle formation was almost abolished (Fig. 3C). We then designed the single-residue mutants P₁₀A, P₁₂A, K₆₃A, and R₆₄A and extended the analysis to include single substitutions within the uncharacterized Pro-rich N terminus of the protease domain, residues 4 to 12 (Fig. 3E). With the exception of L₄A, none of the mutants exhibited a defect in viral replication, with luciferase activity levels similar to those of the WT at 6, 24, 48, and 72 h (Fig. 3F). Interestingly, a wide variety of phenotypes in the production of infectious particles was observed (Fig. 3G). S₉A and P₁₂A mutations showed no apparent defect in infectivity, K₆₃A displayed a minor reduction, P₈A, P₁₁A, and R₆₄A exhibited a significant decrease, and L₄A, W₅A, D₆A, V₇A and P₁₀A abolished infectious particle formation (Fig. 3G). Our results support a novel role of the N-terminal region of the NS3 protease domain on infectious viral particle release without participating in early viral replication steps. This finding supports the hypothesis that the structurally disordered N-terminal region of NS3 does not modulate protein enzymatic activities but rather participates in viral assembly.

We defined novel functions of the N-terminal region and the interdomain linker of DENV NS3 on production of infectious particles using a reporter system that allowed us to focus on mutant viruses fully competent for translation and RNA replication. Since this is the first study reporting a role of different domains of NS3 on DENV particle assembly, we further investigated phenotypic properties of these mutant viruses in the context of the infectious 16681 strain of DENV-2. We selected mutants that were competent for replication with substitutions in the N-terminal, protease, linker, and helicase domains and incorporated each one in the infectious cDNA clone.

Viruses containing the individual W₅A, D₆A, V₇A, P₁₀A, I₁₇₈A, or N₅₇₆A mutation were constructed, and transcribed vRNAs were transfected into BHK-21 cells in parallel with controls (Fig. 4A). Two days after transfection, cells were examined by immunofluorescence assay (IFA) using anti-E monoclonal antibody. A similar signal of the viral protein was observed for WT and mutant viruses at 2 days posttransfection (Fig. 4A, vRNA transfections), suggesting that the mutations did not affect early steps of viral translation/replication as expected. Subsequently, viral infectivity from the supernatant media collected 2 dpt was evaluated by IFA at 1 and 4 days postincubation of fresh BHK-21 cells (Fig. 4A, infections). Consistent with the results obtained with the reporter virus, a dramatic decrease in viral spread in all mutants as a consequence of single-point mutations in NS3 was observed. The only mutant that showed a low but detectable level of propagation was the N₅₇₆A mutation, which displayed merely 1% of the monolayer positive signal for the viral antigen (Fig. 4A).

FIG 4.

Multiple domains within NS3 are involved in assembly/secretion of DENV particles. (A) Representative images for IFA showing DENV antigen-positive BHK cells transfected with WT or mutant DENV-2 RNAs at 2 days posttransfection. Culture supernatants and cells were harvested separately and stored at −70°C until use. Culture supernatants were thawed and applied to fresh cells to determine the viral propagation capabilities. Representative IFAs from 1 and 4 days postinfection are shown. Percentages of infected cells were determined in each case. (B) WBs from cell lysates for E, NS3, and C. Viral protein production in transfected cells at 2 dpt with mutated virus impaired in secreted infectivity is shown. GAPDH was utilized as a loading control. (C) WBs for detection of secreted C released to the extracellular milieu from transfected cells at 3 dpt. (D) RT-qPCR of cell lysates at 2 dpt to determine the intracellular amount of vRNA. (E) RT-qPCR to determine the secreted amount of vRNA released to the supernatant fluids from RNA-transfected cells at 2 dpt.

To rule out that the infectivity impairment of the NS3 mutants was due to a defect in proteolytic processing of the viral polyprotein that unexpectedly impacted viral assembly or protein stability, Western blot analyses against C, E, and NS3 proteins were performed for each mutant. Cell lysates collected at 2 dpt revealed comparable levels of viral proteins (Fig. 4B), confirming unaffected viral translation and polyprotein processing in the context of the infectious DENV clone. It is possible that noninfectious viral particles were secreted that were undetectable in a functional infection assay (Fig. 4A). If this were the case, inactive particles containing nucleocapsids (C and viral genome) would be secreted. To evaluate this possibility, both C and viral RNA were evaluated in the supernatants. Western blot assays were performed to detect C released in the media of transfected cells at 2 days posttransfection. While the WT virus showed strong accumulation of C and the mutants W₅A and N₅₇₆A showed a faint band, C was undetectable in the media of cells transfected with the other mutants (Fig. 4C). To further confirm this observation, quantitative real-time RT-PCR was performed to determine the levels of intracellular and secreted viral RNA. To analyze the intracellular compartment, cells at 2 dpt were thoroughly washed with PBS and then total RNA was extracted. For the analysis of extracellular vRNA, culture fluids were collected and centrifuged for 15 min at 15,000 × g, and then supernatants were used for RNA extraction. As shown in Fig. 4D, similar amounts of viral RNA were observed in the intracellular compartments of transfected cells for the WT and all NS3 mutants, confirming that RNA replication and stability were not significantly affected by NS3 mutations. In contrast, the amounts of vRNA in the extracellular media were drastically reduced for all of the mutant viruses >1,000-fold, consistent with impaired secretion of infectious particles (Fig. 4E). Again, vRNA was detectable for the W₅A and the N₅₇₆A mutants.

Because W₅A and N₅₇₆A were mutants that displayed a highly reduced but detectable secreted vRNA (Fig. 4E), we further characterized these viruses. Growth curves were performed comparing their replication to that of the WT (Fig. 5A). The growth curve demonstrates an increase of N₅₇₆A mutant approaching the WT level from day 1 to day 3 (about 1,000-, 400-, and 100-fold lower titers in supernatant fluids collected at 1, 2, and 3 dpt, respectively). This mutant virus approaches the WT level because the WT reaches its maximum at about 3 days posttransfection while the mutant continues replicating. This observation is compatible with the reduction in vRNA detected in the supernatants by quantitative real-time RT-PCR (Fig. 4E and 5B). In addition, a reduced level of released E protein was observed for this mutant compared to the WT in supernatants at 3 dpt (Fig. 5C). Furthermore, although detectable, this mutant displayed small and diffuse plaques and small infectious focus phenotypes, as observed in virus plaque- and focus-forming assays (Fig. 5D and E, respectively). The results are consistent with the idea that mutation N₅₇₆A largely reduces the amount of particles produced but the virions released are infectious.

FIG 5.

Characterization of selected NS3 mutants defective in propagation reveals singular deficiencies. (A) Focus-forming unit (FFU) quantification by serial dilution titrations of supernatants collected for each virus at the indicated time posttransfection. (B) Secreted vRNA was analyzed by RT-qPCR. Total RNA was extracted from supernatant fluids of transfected cells collected at 3 dpt and reverse transcribed prior to being assayed by qPCR. (C) Secreted E protein was analyzed by WB. Supernatant fluids of transfected cells collected at 3 dpt were applied to SDS-PAGE followed by WB. (D) Virus PFU were analyzed from culture supernatants from WT and N₅₇₆A and W₅A mutant vRNA-transfected cells. WT virus produced well-defined plaques, whereas N₅₇₆A produced tiny and diffuse plaques. No plaques were detected for W₅A virus. (E) Viral focus-forming units were immunodetected for the WT and N₅₇₆A mutant, although those of the latter were of a smaller size. No FFU were detected for W₅A.

Mutant W₅A was particularly interesting because it did not propagate in any of the functional assays tested (IFA and virus plaque- and focus-forming assays in Fig. 4A and 5A, D, and E), but vRNA and C were detected in the media of transfected cells (Fig. 4C and E and 5B). The secreted vRNA at 3 dpi was only 20-fold lower than that of the WT, but the E protein was undetectable (Fig. 5B and C). These observations are consistent with the idea that vRNA was secreted at low levels by a noncanonical process, possibly as naked nucleocapsids that were not infectious. This was different from that observed for mutant N₅₇₆A, in which every parameter analyzed was consistent with very low levels of infectious particle secretion.

Taking these findings together, we conclude that mutants in the NS3 protein do not release significant amounts of noninfectious particles and the defects observed are associated with particle formation or release. Interestingly, we identified a novel proline-rich, N-terminal region of the DENV NS3 protease domain as a determinant for production of infectious viral particles. Sequence analyses of this region revealed that it is highly conserved among different flaviviruses, and structural analysis supports the conservation of an intrinsically unstructured region.

We did not expect to find determinants for DENV assembly in separate regions of the large NS3 protein. We speculate that NS3 serves as a platform for interactions with viral and/or host factors implicated in the production of infectious particles. In this regard, proteolytic cleavage of capsid from the viral polyprotein by NS2B3 has been proposed to be coordinated with nucleocapsid assembly and uptake of the complex into budding particles (54). Thus, it is possible that the presence of NS3, acting as a protease for capsid maturation, also performs a function in coordinating genome recruitment. This is also relevant when taking into account that NS3 is in intimate interaction with the viral genome during RNA replication, acting as an RNA helicase and RNA annealing enzyme. Here, we provide direct evidence for the involvement of different domains of NS3 in viral particle formation. Because the flavivirus machinery for the initial steps of particle assembly still is unknown, our work provides a framework to further explore NS3 participation. The definition of the NS3 protein-protein interaction network with host and viral proteins will be the next step to dissect the components and mechanism of viral particle assembly.