Secretory glycoprotein NS1 plays a crucial role in the particle formation of flaviviruses

Flaviviruses, which are globally distributed and cause a spectrum of potentially severe illnesses, pose a major threat to public health. Although Flaviviridae viruses, including flaviviruses, possess similar genome structures, only the flaviviruses encode the non-structural protein NS1, which resides in the endoplasmic reticulum (ER) and is secreted from cells after oligomerization. The ER-resident NS1 is known to be involved in viral genome replication, but the essential roles of secretory NS1 in the virus life cycle are not fully understood. Here we characterized the roles of secretory NS1 in the particle formation of flaviviruses. We first identified an amino acid residue essential for the NS1 secretion but not for viral genome replication by using protein-protein interaction network analyses and mutagenesis scanning. By using the recombinant flaviviruses carrying the identified NS1 mutation, we clarified that the mutant flaviviruses employed viral genome replication. We then constructed a recombinant NS1 with the identified mutation and demonstrated by physicochemical assays that the mutant NS1 was unable to form a proper oligomer or associate with liposomes. Finally, we showed that the functions of NS1 that were lost by the identified mutation could be compensated for by the in trans-expression of Erns of pestiviruses and host exchangeable apolipoproteins, which participate in the infectious particle formation of pestiviruses and hepaciviruses in the family Flaviviridae, respectively. Collectively, our study suggests that secretory NS1 plays a role in the particle formation of flaviviruses through its interaction with the lipid membrane.


Introduction
The family Flaviviridae viruses consist of 4 genera: Flavivirus, Pestivirus, Pegivirus and Hepacivirus [1]. Although flaviviruses and pestiviruses can infect various species and tissues, infection of pegiviruses and hepaciviruses is observed in a strikingly restricted range of tissues and hosts. Flaviviruses can be readily transmitted by arthropod vectors to a variety of mammalian species and (re-) emerge unexpectedly in human populations [2]. Some of the mosquito-transmitted flaviviruses, including dengue virus (DENV), Japanese encephalitis virus (JEV), West Nile virus (WNV), yellow fever virus (YFV), and Zika virus (ZIKV), have expanded globally and cause febrile illness to humans. The limited availability of vaccines or specific therapeutics for the mosquito-borne flaviviruses continues to pose a threat to public health. Therefore, an improved understanding of the molecular mechanisms underlying the life cycle of these viruses is needed for the development of effective countermeasures.
Flaviviridae viruses share a common genome structure consisting of a single-stranded and positive-sense RNA with one open reading frame [1]. Interestingly, flaviviruses possess a unique non-structural protein within the Flaviviridae family. This protein is NS1, a 48-to 55-kD glycoprotein that is well-conserved among flaviviruses, exhibiting 20%-40% identity and 60%-80% similarity at the amino acid level [3,4]. NS1 is initially synthesized as a soluble monomer and becomes membrane-associated after dimerization in the lumen of the endoplasmic reticulum (ER). The higher oligomeric NS1 is secreted into extracellular environments. While the dimer of NS1 is involved in the formation of a replication complex in the ER, the secretory NS1 is responsible for dengue pathogenesis via induction of the release of vasoactive cytokines [5]. The secretory NS1 also plays a role in immune evasion by binding the complement protein C4 and triggering its degradation [6]. In addition, NS1 can directly alter the barrier function of pulmonary endothelial cell monolayers through disruption of the tissuespecific endothelial glycocalyx-like layer by triggering the activation of endothelial sialidases in vitro and in vivo [7,8]. Because DENV NS1 circulates at high levels (1-2 μg/ml) during the acute phase of infection with severe dengue disease, which correlates with viremia [9], the secretory NS1 is utilized as a marker for the diagnosis of infection [10]. Collectively, these findings demonstrate that the secretory NS1 plays multifunctional roles in the pathogenesis and tissue-specificity of flaviviruses; however, the roles of secretory NS1 in the virus life cycle have not yet been fully characterized.
In our previous studies, the amphipathic α-helices of exchangeable apolipoproteins were shown to play crucial roles in the formation of infectious particles of hepaciviruses through their interaction with viral particles [11]. We further showed that pestivirus E rns and flavivirus NS1 play functional roles similar to those of host-derived apolipoproteins in the formation of infectious particles of hepaciviruses and pestiviruses [12]. These results suggest that the host-and virus-derived secretory glycoproteins have overlapping roles in the virus life cycle of Flaviviridae viruses, especially in the maturation of infectious particles. We therefore hypothesized that NS1 plays crucial roles in the maturation of infectious particles of flaviviruses.
Here, we investigated the functional roles of secretory NS1 in the flavivirus life cycle. We showed that a loss-of-function mutant of NS1 lacking secretory ability was substantially involved in replication of viral RNA but failed to interact with the lipid membrane or to produce infectious particles. Moreover, the virus carrying the mutant NS1 could restore the particle production by the expression of the lipid associated NS1.These findings suggest that NS1 plays at least two roles in the virus life cycle. In addition, we proved that the pestivirus E rns and host-derived exchangeable apolipoproteins can compensate for the loss of function of secretory NS1 in the flavivirus particle formation. Taken together, the results of this study shed light on the functions of flavivirus NS1 in the virus life cycle.

Generation of a reporter ZIKV that is unable to replicate but is rescued by the exogenous expression of NS1
First, to investigate the role of NS1 in virus propagation, we generated cDNA clones having a deletion in the NS1 gene of flaviviruses JEV and DENV. We tried to conduct a complementation assay by transfecting the in vitro-transcribed RNAs into Vero E6 cells expressing the NS1 protein (S1A Fig). We collected the supernatants of the transfected cells at 96 h post-transfection (hpt) and measured infectious titers. However, no infectious virus was recovered, suggesting that trans-complementation of NS1 cannot be achieved by using cDNA clones possessing deletion in the NS1 genes of JEV and DENV (S1B and S1C Fig). We therefore took the alternative approach of inserting the high-affinity NanoBiT (HiBiT) luciferase gene [13,14] into the β-ladder region of NS1 ( Fig 1A). The HiBiT luciferase is a split luciferase gene encoding 11 amino acids and is one of the components of the NanoBiT. When HiBiT and its counterpart large NanoBiT (LgBiT) associate in cells or in vitro, the complex regains its NanoLuc enzymatic activity, which correlates with viral replication of ZIKV. The resulting cDNA was cloned into a vector under the control of a CMV promoter and hepatitis delta virus ribozyme cassette (ZIKV-HiBiT) and transfected into BHK-21 cells (Fig 1B). Although no RNA replication was observed in the BHK-21 cells (Mock in Fig 1B), both RNA replication and infectious particles were rescued by the expression of wild type NS1 of ZIKV (NS1 in Fig 1B). In addition, when we similarly expressed the NS1 protein of other flaviviruses, including DENV and JEV, in the BHK-21 cells, infectious particles were recovered in the supernatants of the ZIKV-HiBiT clone-transfected cells at levels comparable to those from the wild type ZIKV infection ( Fig  1C, middle), suggesting that the loss of the functional role of NS1 of ZIKV can be compensated for by the expression of NS1 derived from other flaviviruses. Moreover, significant luciferase activity was observed in the culture supernatants of cells transfected with the ZIKV-HiBiT clone by the exogenous expression of NS1 of flaviviruses (Fig 1C, right), suggesting the possibility of identifying the amino acid residue(s) in NS1 that is crucial for RNA replication and infectious particle formation by measuring luciferase activities.

Identification of an amino acid residue involved in higher oligomerization of NS1 through in silico and mutagenesis analyses
Having established a complementation assays for evaluating the roles of NS1 in the virus life cycle, we next attempted to conduct mutagenesis scanning using ZIKV-HiBiT. NS1 of flaviviruses has been reported to form a dimer and higher oligomer in the infected cells. The dimer is

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation considered to participate in the viral genome replication and the higher oligomer is secreted into the culture supernatants [15]. With the aim of identifying amino acid residue(s) involved in the secretory form of NS1, we utilized a computational coevolution-based method, Blocks in Sequences (BIS) [16,17]. BIS is able to calculate the protein-protein interaction network within a protein. Based on the BIS analyses of 66 different strains of Japanese encephalitis serocomplex, 10 clusters were identified (Fig 2A and S1 Table) and 9 clusters (clusters 1, 3-10) exhibited different p-values ranging between 4 x 10 −4 and 2 x 10 −19 , and cluster 2 was conserved (p-value = 1). To investigate which clusters or amino acid residues are involved in the formation of the higher oligomer, we changed the amino acid at each position to an amino acid with characteristics (S2 Table) in terms of destroying or altering NS1 oligomer formation. In total, 50 BHK-21 cells expressing the respective NS1 mutants were generated (S2A Fig) and subjected to our complementation assay. Cells expressing the respective NS1 mutants were infected with ZIKV-HiBiT, and the luciferase activity of the supernatants was measured at 96 h post-infection (hpi). In addition, we fixed the infected cells and used an immunofluorescence assay to detect dsRNA as an index of active RNA replication ( indicating that these NS1 mutants were able to support viral replication in trans. Among the 10 NS1 mutants possessing replication activity, the NS1 carrying histidine instead of isoleucine at position 273 (NS1 I273H ) exhibited luciferase activity but at significantly lower levels compared to the wild type NS1 (Fig 2B), as represented by the red dots in cluster 8 (shown in the red square in Fig 2B). Also, an active RNA replication was impaired by this substitution (S2B Fig). Collectively, these data suggest that the mutant NS1 I273H remains a dimer but might have lost the ability to form the higher oligomer.

Basic amino acid residues at position 273 of NS1 play a crucial role in secretion
Next, Huh7 cells expressing the wild type or mutant (NS1 I273H ) of JEV were infected with ZIKV-HiBiT and the synthesis of dsRNA and expression of NS1 were examined at 48 hpi by confocal microscopy (Fig 3A). The immunofluorescent signals for NS1 (green) and dsRNA (red) were observed in the cytoplasm of both cells, and some of them were colocalized. Next, secretion of NS1 was detected by immunoblotting of culture supernatants of Huh7 cells expressing either the wild type or mutant NS1 of JEV collected at 96 hpt. Although both NS1 proteins were expressed in cells, the wild type NS1 but not the NS1 I273H mutant was secreted into the culture supernatants ( Fig 3B). These results suggest that the isoleucine residue at position 273 in NS1 plays a crucial role in the secretion of JEV NS1. In addition, we found that the property of the amino acid at position 273 of NS1 in JEV is conserved among flaviviruses (Fig 3C). To investigate the effect of amino acid substitution at this position, we conducted a complementation experiment as shown in Fig 2B using NS1 possessing an amino acid mutation at position 273. NS1 mutants carrying an amino acid at position 273 with the same property as isoleucine (valine or leucine) were secreted into the supernatants, but those with a different property (serine, glutamine, or glutamic acid) were not (Fig 3D), suggesting that the basic amino acids residues at position 273 play a crucial role in the secretion of NS1 and the subsequent infectious particle production.

The non-secretory NS1 I273H sustains the role of RNA replication and is compensated the role of the infectious particle formation by the replication-defective NS1 F160D
To further characterize the NS1 mutant (NS1 I273H ), we generated a mutant JEV RNA carrying the histidine substitution in NS1 (JEV NS1 I273H ), because we were unable to obtain in vitro The proteinprotein network within the NS1 protein was analyzed by using co-evolving amino-acid pairs [16,17] with 66 strains belonging to the Japanese encephalitis serocomplex group. The amino acid residues possessing interaction within clusters are shown as spheres in the three-dimensional structures of NS1 dimers (Protein Data Bank accession number: 4TPL). (B) An illustration shows the experimental workflow. Upon infection with ZIKV-HiBiT at MOI = 0.1, the luciferase activity of cells expressing NS1 mutants carrying the amino acids with different phenotypes (S2 Table) were measured at 36 hpi. The relative luciferase activity was normalized by the activity of cells expressing GFP. The production of double-stranded RNA (dsRNA) was determined by staining with antibody against dsRNA and is indicated by red bars. The locus of amino acid position 273 on the NS1 structure is indicated as red spheres. The data set of the quantified fluorescent signal of dsRNA are found in S2B Fig

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation transcribed RNA from ZIKV possessing the NS1 I273H mutation. The JEV mutant RNA was transfected into Huh7 cells, and the replication of viral RNA was determined by RT-qPCR (Fig 4A, left). Substantial viral RNA replication was observed in Huh7 cells from 24 hpt without expressing NS1, suggesting that JEV NS1 I273H is capable of replicating in cells. Upon transfection of the JEV NS1 I273H RNA, the recovery of infectious particles was only observed in BHK-21 cells at 96 hpi by the expression of NS1, suggesting that I273H substitution kept the role of RNA replication but led to a loss of the capability for infectious virus production ( Fig  4A, right). The genome replication of flaviviruses occurs in the intracellular membrane compartments known as replication complexes [18]. To investigate the formation of replication complexes, cells infected with either the wild type JEV or JEV NS1 I273H were analyzed by electron microscopy at 48 hpi. Representative images are shown in Fig 4B. Replication complexes consisting of spherules were observed in the cytoplasm of Huh7 cells infected with either the wild type or mutant JEV (denoted by red arrows in Fig 4B). Viral particles were observed in cells infected with the wild type JEV (denoted by red arrowheads in Fig 4B), but not in those infected with JEV NS1 I273H . These data suggest that JEV NS1 I273H is capable of replicating via the formation of RNA replication complexes but is incapable of producing viral particles. Next, a mutant JEV RNA carrying aspartic acid at position 160 (JEV NS1 F160D ) was generated and served as a replication-dead control, because it has been demonstrated that substitution of F160D in the hydrophobic protrusion of DENV NS1 abolishes viral RNA replication [15]. We examined whether the expression of NS1 F160D can compensate for the infectious virus production lost in JEV NS1 I273H (Fig 4C). Upon infection with the JEV NS1 I273H , infectious particles were recovered in the supernatants of the cells expressing wild type NS1 and NS1 F160D but not in those expressing NS1 I273H at 96 hpi, indicating that NS1 F160D retains the particle-production function lost in the NS1 I273H mutant. When we generated a mutant JEV NS1 F160D , infectious particles were recovered at 96 hpi in the supernatants of cells expressing the wild type NS1 or NS1 I273H , but not in those expressing NS1 F160D upon infection with JEV NS1 F160D (Fig 4D). These results suggest that flavivirus NS1 has at least dual functions in the virus life cycle-i.e., a viral RNA replication function and a particle formation function. In support of this conclusion, we confirmed that NS1 is not involved in viral RNA translation (S3C Fig), and NS1 F160D but not NS1 I273H was secreted into the culture supernatants of cells expressing these proteins ( Fig 4E).

Higher oligomerization of NS1 is responsible for the membrane association and secretion into culture supernatants
Since the BIS analyses showed that the amino acid residue at position 273 of NS1 is involved in protein-protein interaction, we hypothesized that the amino acid residue at 273 participates in the correct oligomer formation of NS1. Thus, we synthesized recombinant MBP-tagged NS1 proteins (approximately 90 kDa) and subjected them to fast protein liquid chromatography (FPLC) to analyze their oligomerization properties (Fig 5A). The peak around 150 kDa (fractions 19 and 20) was observed in both the wild type NS1 and NS1 I273H mutant samples, whereas the peak around 300 kDa (fraction 16) was observed only in the wild type NS1. The expression of MBP-NS1 in each fraction was confirmed by immunoblotting using anti-NS1 amino acid (252-280 aa of NS1) alignments of the 13 viruses belonging to the Flavivirus genus are shown. (D) Relative luciferase activities of the culture supernatants of BHK-21 cells expressing the wild type and mutant NS1 substituted at position 273 (valine, leucine, serine, glutamine, and glutamic acid) upon infection with the ZIKV HiBiT were determined. Secretion of NS1 into the supernatants was determined by immunoblotting. Statistical significance was assessed using one-way ANOVA with Dunnett's test (D) and is indicated by asterisks ( � ) (versus control cells). https://doi.org/10.1371/journal.ppat.1010593.g003

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation antibody ( Fig 5B). These data suggest that the amino acid residue at position 273 in NS1 participates in higher oligomerization but not dimerization of NS1. In addition, to examine the effect of mutation in NS1 I273H on the interaction of NS1 with membranes, the purified NS1 proteins were incubated with liposomes. Conversion of large liposomes (more than 500 nm in diameter) into small heterogeneous lipid particles was observed by the incubation with wild type NS1 and NS1 F160D as previously reported [15] (Fig 5C). In addition, NS1 I273L , which was secreted into culture supernatants (Fig 3D), induced conversion of large liposomes into small lipid particles, but NS1 I273H did not alter the liposome morphologies (Fig 5C). These results suggest that the higher oligomerization of NS1 is responsible for the lipid association and secretion into culture supernatants.
The secretory glycoproteins, including exchangeable apolipoproteins, pestivirus E rns , and flavivirus NS1, play a common role in the infectious particle formation of Flaviviridae viruses Previously, we have shown that host-and virus-derived secretory glycoproteins, including exchangeable apolipoproteins, pestivirus E rns and flavivirus NS1, play comparable roles in the particle formation of hepaciviruses and pestiviruses in the Flaviviridae family [12]. Therefore, we hypothesized that these heterogenous secretory glycoproteins might play common roles in the particle formation of flaviviruses. To examine this possibility, HA-tagged NS1of JEV, pestivirus E rns , and apolipoprotein E (ApoE) were expressed in BHK-21 cells (Fig 6A) and infected with JEV NS1 I273H . As we expected, although intracellular viral RNA replication and infectious titers in the culture supernatants of cells expressing either E rns or ApoE were lower than those in the culture supernatants of cells expressing NS1, substantial viral replication and infectivity were observed in cells expressing E rns or ApoE upon infection with JEV NS1 I273H at 96 hpi ( Fig 6B). When we used a mutant ZIKV carrying the same mutation in NS1 (ZIKV NS1 I273H ), intracellular viral RNA replication and infectious titers in the culture supernatants were again observed in cells expressing either E rns , or ApoE (Fig 6C), suggesting that exchangeable apolipoproteins and E rns play the same role as secretory NS1 in the infectious particle formation of flaviviruses. Moreover, either ApoE, wild type NS1, NS1 I273H or NS1 F160D was expressed in ApoE and ApoB double-knockout Huh7 cells and infected with hepatitis C virus (HCV), a member of the family Flaviviridae (Fig 6D). The levels of intracellular viral RNA replication were similar among the cells expressing either ApoE, wild type NS1, NS1 I273H or NS1 F160D upon HCV infection (Fig 6D, middle). However, the infectious titer in the culture supernatants and specific infectivity were impaired in cells expressing NS1 I273H (Fig 6D, left and right, respectively) at 72 hpi, suggesting that the secretory function of NS1 participates in the particle formation of HCV. These results suggest that secretory glycoproteins, including exchangeable apolipoproteins, pestivirus E rns , and flavivirus NS1, play a common role in the infectious particle formation of Flaviviridae viruses.

Discussion
Flavivirus NS1 is a unique viral protein within the family Flaviviridae and is considered to be multifunctional; however, the role of NS1 in the virus life cycle has been less fully elucidated compared with the roles of other viral proteins, such as core, envelope, protease and polymerase. In this study, we investigated whether secretory NS1 plays a role in the virus particle formation of flaviviruses. By using in silico analysis of the protein-protein network and an in vitro complementation assay, we identified the amino acid residue involved in proper oligomerization for secretion and demonstrated that the lipid membrane-binding oligomer of NS1 is

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation involved in the infectious particle formation. We also elucidated that the secretory glycoproteins commonly participate in the infectious particle formation of the Flaviviridae viruses.
In the early years of flavivirus research, NS1 was considered to participate in assembly and release of flavivirus particles due to its localization in the ER lumen and secretion profile [19]. Meanwhile, later studies using reverse genetics techniques to create deletion mutants of NS1 revealed that NS1 plays a crucial role in the formation of the replication complex through its interaction with NS4A and NS4B in the ER lumen [20,21]. To further clarify the roles of NS1, in this study we tried to carry out the trans-complementation of NS1 by using DENV and JEV mutants lacking the NS1 gene. However, we were unable to rescue infectious viruses by the trans-complementation. Our attempt suggests that NS1plays crucial roles in the formation of the virus RNA replication complex that affects particle formation, and that complementation by the expression of NS1 depends on subtle structural differences in association with host and/ or viral proteins [22]. Interestingly, by using the ZIKV carrying the dysfunctional NS1 by insertion of the HiBiT luciferase, infectious particles were successfully rescued by the exogenous expression of NS1 of other flaviviruses, suggesting that these different NS1 proteins can

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation share the functional roles in the virus life cycle. In fact, NS1 is well conserved in flaviviruses and exhibits up to 80% amino acid similarity among them [3].
After translation, NS1 is considered to form a dimer and a higher oligomer including a putative hexamer. The dimer is located in the ER and is involved in RNA replication as previously reported [21,23]. The secretory NS1 forms the higher oligomer [15]. Therefore, we hypothesized that the ER-resident dimer NS1 and the secretory NS1 play distinct roles in the virus life cycle. To investigate this hypothesis, we needed to identify the amino acid residues involved in the oligomerization of NS1; we therefore estimated the protein-protein interaction network within NS1 by using co-evolving amino-acid pairs [16,17,24]. We attempted to subject the data set of ZIKV to the co-evolution analysis, but the co-evolution value within ZIKV strains was too small to investigate further. We therefore employed sequences of the JEV serocomplex including JEV and WNV, and the data set shown in S1 Table was used for the in vitro analyses. Although the crystal structure of the dimer formation of NS1 was uncovered previously, the oligomerization of NS1 in vitro was conducted in an artificial manner [15,25]. The co-evolving amino-acid pairs within NS1 were classified into 10 clusters, and the amino acid residue at position 273 was classified into cluster 8 (Fig 2A). The amino acid loci in cluster 8 are located on the β-ladder region which participates in neither lipid binding nor dimer formation [26]. In addition, FPLC analysis revealed that the NS1 I273H mutant was capable of forming the dimer but not a higher oligomer, suggesting that this computational analysis was useful for deducing co-evolving amino-acid pairs for estimation of the protein-protein interactions in a single viral protein.
The previous study indicated that the secretory NS1 forms a lipoprotein particle with an open-barrel protein shell [25]. Thus, the formation of secretory NS1 is required for amphipathic properties and remodeling of the lipid membrane. Our previous studies have shown that amphipathic helices of exchangeable apolipoproteins play important roles in the infectious particle formation of hepaciviruses through their interaction with the viral membrane [11]. The amphipathic properties of viral secretory glycoproteins E rns and NS1 participate in the infectious particle formation of hepaciviruses and pestiviruses [12]. In this study, we showed that JEV NS1 I273H retained its ability for RNA replication but lost its ability for infectious particle formation, and that NS1 I273H was incapable of being secreted into extracellular environments via the alteration of lipid membranes (second panel in Fig 7). These data suggest that the secretory property of NS1 is involved in the infectious particle formation. Future work to further investigate involvement of NS1 in later step of virus life cycle will be required. A study on DENV showed that mutations in NS1 impaired the production of infectious particles but had no effect on viral RNA replication [27]. On the other hand, Akey et al. showed that NS1 mutants incapable of generating replication complexes are nonetheless capable of remodeling the membranes of liposomes [15]. By exogenous expression of the NS1 F160D [15], production of the infectious particle of the JEV NS1 I273H was restored (Fig 4C). In accordance with these studies, we proved that NS1 possesses at least dual roles in the life cycle of flavivirus. In addition, we clearly showed that the exchangeable apolipoproteins and E rns of pestivirus are capable of compensating for the particle-formation function lost in the NS1 I273H mutant. Furthermore, the expression of wild type NS1 and NS1 F160D but not of NS1 I273H was able to enhance the infectious particle formation of HCV in ApoB and ApoE double-knockout cells.
These results indicate that the viruses belonging to the family Flaviviridae commonly utilize the amphipathic properties of secretory glycoproteins in the particle formation. To the best of our knowledge, our studies are the first to demonstrate the conversion of different viral proteins and host factors that share the same role and compensate for each other in the virus life cycle of the same virus family.
In summary, by means of computational analyses of the protein-protein network, evidence-based mutagenesis and physicochemical characterization, we have shown that the secretory NS1 protein participates in infectious particle formation (Fig 7). Further investigation of NS1 will be needed to clarify the role of this protein in the virus life cycle of Flaviviridae viruses.

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation previously described [32]. Then, the constructed plasmids were applied to a Bac-to-Bac system (Thermo Fisher Scientific) using DH10Bac to generate recombinant bacmids carrying the NS1 gene according to the manufacturer's protocol. All of the plasmids used in this study were confirmed by sequencing with an ABI 3130 genetic analyzer (Thermo Fisher Scientific).

Cell lines
All mammalian cell lines were cultured at 37˚C under the conditions of a humidified atmosphere and 5% CO 2 . The human hepatocellular carcinoma-derived Huh7 cells, human embryonic kidney-derived 293T cells, African green monkey kidney-derived Vero E6 cells and baby hamster kidney fibroblasts-derived BHK-21 cells were maintained in DMEM (Nakalai) supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% fetal bovine serum. The Huh7-derived cell line Huh7.5.1 was kindly provided by Dr. F. Chisari. The Huh7 cells with double-knockout of apolipoproteins B and E (BE-KO cells) were established previously [11]. Spodoptera frugiperda-derived Sf-9 cells were maintained in Sf-900II insect medium (Thermo Fisher Scientific) supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin and cultured at 28˚C.

Antibodies
Mouse monoclonal antibodies to β-actin, dsRNA and NS1 were obtained from Millipore-Sigma, English & Scientific Consulting Kft and Bio Matrix research, respectively. Alexa Fluor (AF) 488-conjugated anti-rat IgG and 647-conjugated anti-mouse IgG antibodies were purchased from Thermo Fisher Scientific. Rat anti-HA and rabbit anti-NS1 antibodies were purchased from Roche Diagnostics and GeneTex, respectively. Rabbit anti-HCV NS5A was generated as described previously [33].

Immunoblotting
Cells lysed on ice in lysis buffer , 135 mM NaCl, 1% Triton-X 100, 10% glycerol) supplemented with a protease inhibitor cocktail, cOmplete mini (Millipore-Sigma), were boiled in loading buffer and subjected to 5%-20% gradient SDS-PAGE. The proteins were transferred to polyvinylidene difluoride membranes (MilliporeSigma) and incubated with the appropriate antibodies. The immune complexes were visualized with SuperSignal West Femto substrate (Thermo Fisher Scientific). The signals were detected by use of an LAS-4000 image analyzer system (Fujifilm)/an Amersham Imager 600 (GE Healthcare) or an WSE-LuminoGraph I (ATTO)/ImageSaver6 (ATTO).

Immunofluorescence
Cells were fixed with 4% paraformaldehyde (Nacalai Tesque). After fixation, the cells were permeabilized with PBS containing 0.2% Triton X-100 (Nacalai Tesque) for 10 min, blocked with 1% BSA fraction V (Millipore Sigma) in PBS, and then reacted with the indicated primary antibodies in PBS for 1 h at room temperature. After washing with PBS three times, cells were incubated with goat anti-mouse IgG Alexa Fluor 647-conjugated (1:1,500 dilution) or antirabbit IgG Alexa Fluor 488-conjugated (1:1,000 dilution) secondary antibodies in PBS for 1 h at room temperature. The cells were then counterstained with Hoechest 33342 (Dojindo) (1:2,000 dilution) for 10 min.

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation

Preparation of viruses
To obtain ZIKV and the mutants, BHK-21 cells were transfected with the cDNA clones by Trans IT LT-1 (Mirus), the culture supernatants were harvested, and infectious titers were determined by a focus-forming assay and expressed in focus-forming units (FFU). DENV, JEV and their mutants were generated as described previously [28]. In brief, the linearized infectious clones were transcribed by using an mMESSAGE mMACHINE T7 Ultra Kit, and the in vitro-transcribed RNA was electroporated. The supernatants were collected from the electroporated cells and infectious titers were determined by a focus-forming assay. pHH-JFH1-E2p7NS2mt was introduced into Huh7.5.1 cells, HCVcc in the supernatant was collected after serial passages, and infectious titers were determined by a focus-forming assay. All of the viruses were kept at -80˚C until use. To generate the baculoviruses carrying the NS1, recombinant bacmids were transfected into Sf-9 cells by using X-tremeGENE HP DNA Transfection Reagent (MilliporeSigma). After incubating the cells for 4 days, the recombinant baculoviruses in the culture supernatants from the bacmid-transfected Sf-9 cells were collected and centrifuged at 1,500 g for 3 min to remove cells and debris. Then, the harvested supernatants were inoculated into naïve Sf-9 cells for 4 days. The passage 1 (P1) viruses were collected in a tube wrapped with foil and stored at 4˚C until use.

Purification of MBP-tagged NS1 and gel filtration
After inoculation of the recombinant baculoviruses P1 for 4 days in the Sf-9 cells, the cellular pellets were collected and resuspended with PBS containing EDTA and cOmplete mini. The

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation cellular lysate was subjected to 2 freeze-thaw cycles and then centrifuged at 18,000 g for 10 min to remove cells and debris. The supernatants were transferred into Dextrin Sepharose High Performance (GE Healthcare) and purification of the MBP-tagged NS1 was performed according to the manufacturer's protocol. Briefly, samples were poured into a column, washed with the binding buffer [20 mM TrisHCl (pH 7.4), 0.2 M NaCl, 1 mM EDTA and 1 mM DTT] and eluted with the elution buffer [0.5 M maltose, 20 mM TrisHCl (pH 7.4), 0.2 M NaCl, 1 mM EDTA and 1 mM DTT]. The size of the purified NS1 protein was examined by gel filtration using a Superose 6 column (Cytiva). Reaction was carried out at a flow rate of 0.5 ml/min with a buffer [50 mM phosphate buffer (pH 7.4) and 150 mM NaCl]. The loaded samples were fractionated into 39 fractions (500 μl each), and all fractions (20 μl) were analyzed by SDS-PAGE and immunoblotting using anti-NS1 antibody.

Synthesis of liposome and interaction with NS1
Liposome solution was prepared as previously described [15]. The 1,2-Distearoyl-sn-glycero-3-phosphocholine (DSPC) and cholesterol (CHOL) were purchased from Avanti Polar Lipids and mixed at ratio of 1:9 CHOL:PC in chloroform. The lipid solutions were evaporated under a nitrogen gas stream to form a lipid film on the surface of glass tubes. Then, 400 μl of buffer [50 mM Bis-Tris (pH 5.5), 50 mM (NH 4 ) 2 SO 4 , 10% glycerol] was added to the dried lipids and incubated in a water bath at 60˚C for 10 min, followed by vortexing for 30 sec. Liposome solution (150 μl) and 150 μg of the purified NS1 protein were mixed and incubated for 2 h at 37˚C with shaking. After centrifugation at 13,000 g for 20 min, the supernatant samples were subjected to negative staining.

Negative staining and electron microscopy (EM)
The virus-infected cells were cultured on a Cell Desk polystyrene coverslip (Sumitomo Bakelite) and fixed with 4% paraformaldehyde at room temperature for 30 min. Cells were postfixed for 1 h with 1% osmium tetroxide and 1% potassium ferrocyanide in 0.1 M sodium-phosphate buffer (pH 7.4), dehydrated in a graded series of ethanol and embedded in Epon812 (Taab Laboratory Equipment). 80 nm ultra-thin sections were stained with saturated uranyl acetate and lead citrate solution. Electron micrographs were obtained with a JEM-1400plus transmission electron microscope (JEOL). Liposomes containing the recombinant NS1 proteins were put on hydrophilic carbon-collodion grids and stained with 2% uranyl acetate. The grids were wick dried with filter paper, air dried for 10 min, and examined on a JEOL JEM-1400plus electron microscope at 80 kV. Images were captured on an Olympus Veleta 2K x 2K side-mounted TEM CCD camera.

Luciferase assay
Luciferase activity was measured by using a Bright-Glo luciferase assay system (Promega) and Nano-Glo HiBiT Lytic assay system (Promega) according to the protocol provided by the manufacturer.

Quantitative RT-PCR
For quantification of viral RNA copies, total RNA was extracted from cells by using a PureLink RNA Mini Kit (Thermo Fisher Scientific), and then first-strand cDNA synthesis and qRT-PCR were performed by using a TaqMan RNA-to-C T 1-step Kit and ViiA7 system (Thermo Fisher Scientific), respectively, according to the manufacturer's protocols. For quantification of viral

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation RNA, the primer sets for the detection of the noncoding region reported in the previous studies [34][35][36] were used. Fluorescent signals were determined by the ViiA7 system.

Computational intra-protein network analysis
The intra-protein interaction was analyzed by the BIS. The methodology and related co-evolution signal analysis are described in previous studies [16,17].

Statistical analysis
All assays were performed in triplicate and independently repeated at least two times. Results were expressed as the means±standard deviations or standard errors. Statistical significance was determined by the two-tailed Student's t-test, one-way ANOVA with Dunnett's test, or Kruskal-Wallis test performed by GraphPad Prism (Software ver. 9.2.0). Significantly different values (p<0.05) are indicated by an asterisk.

PLOS PATHOGENS
Role of flavivirus NS1 in virus particle formation S2 Table. Amino acid substitutions of NS1 used for the screening. The 50 positions within NS1 were substituted for generation of mutant NS1. Lentiviruses of the indicated mutant NS1 were subjected to the screening. (XLSX)