Differential Infectivity of Human Neural Cell Lines by a Dengue Virus Serotype-3 Genotype-III with a Distinct Nonstructural Protein 2A (NS2A) Amino Acid Substitution Isolated from the Cerebrospinal Fluid of a Dengue Encephalitis Patient

Dengue encephalitis is considered as a severe but unusual clinical presentation of dengue infection. Limited molecular information is available on the neurotropism of dengue virus (DENV), highlighting the need for further research. During a dengue outbreak in Vietnam in 2013, two DENV-3 strains were isolated, in which one was isolated from cerebrospinal fluid (CSF) samples from a dengue encephalitis patient and another strain was isolated from a patient with classical dengue fever in Hai Phong, Vietnam. DENV serotype-3 (DENV-3) isolated from these samples belonged to genotype III, marking the first report of this genotype in the country at that time. Genetic variation between both strains was elucidated by using a full genome sequencing by next-generation sequencing (NGS). The infectivity of the isolated DENV-3 strains was further characterized using human and mouse neuronal cell lines. Phylogenetic analysis of the isolates demonstrated high homogeneity between the CSF-derived and serum-derived DENV-3, in which the full genome sequences of the CSF-derived DENV-3 presented a Thr-1339-Ile mutation in the nonstructural 2A (NS2A) protein. The CSF-derived DENV-3 isolate grew preferentially in human neuronal cells, with a significant proportion of cells that were positive for nonstructural 1 (NS1), nonstructural 4B (NS4B), and nonstructural 5 (NS5) antigens. These results suggest that NS2A may be a crucial region in the neuropathogenesis of DENV-3 and its growth in human neuronal cells. Taken together, our results demonstrate that a CSF-derived DENV-3 has unique infectivity characteristics for human neuronal cells, which might play a crucial role in the neuropathogenesis of DENV infection.


Introduction
Dengue is an acute mosquito-borne viral disease, and its global incidence has grown dramatically in recent decades. Te disease places a notable socioeconomic burden on many tropical and subtropical regions of the world [1]. It is estimated that more than 390 million dengue virus (DENV) infections occur annually, and approximately 96 million cases are symptomatic, accounting for 2.5% of the deaths among hospitalized cases [1]. DENV is a single positivestranded RNA virus that belongs to the family Flaviviridae and genus Flavivirus. Tere are four DENV serotypes (DENV-1, DENV-2, DENV-3, and DENV-4), all of which have been individually found to be responsible for dengue epidemics and associated with severe dengue cases [2]. Te virus is transmitted by female mosquitoes, mainly Aedes aegypti, and to a lesser extent by Aedes albopictus. Te resurgence of the mosquito vector Aedes aegypti, overcrowding, and increasing travel have been related to the expansion of dengue infection throughout the world [2,3].
Although the incidence of dengue cases presenting with neurological impairments continues to increase during dengue epidemics worldwide, the neuropathogenesis of DENV infection is still poorly understood. It is known that virus and host factors play an important role in the neurological disorders associated with dengue, as more evidence strongly supports the notion that the dengue virus is directly neurovirulent, although it is not classically categorized as a neurotropic pathogen. Here, we determined the in vitro infectivity of the two DENV-3 strains using human neural cell lines. One strain was isolated from the cerebrospinal fuid (CSF) of a patient who was diagnosed with dengue encephalitis, and another strain was isolated from serum of a patient with classical dengue fever in Hai Phong, Vietnam, in 2013.

Patient Samples.
Cerebrospinal fuid (CSF) specimens were obtained from a single dengue encephalitis patient in Hai Phong, Vietnam, in 2013. Two serum samples from two diferent patients with classical dengue fever were also obtained during the dengue endemic season in Hai Phong, Vietnam, in 2013 ( Supplementary Figures 1 and 2). Te virus strains isolated from clinical samples were used in the subsequent analysis.

Virus and Cell
Line. To isolate the virus, 10 μl of CSF and serum specimens were inoculated onto C6/36 mosquito cells and Vero cells (African green monkey epithelial kidney ATCC, CCL81). Vero cells were maintained at 37°C and C6/ 36 at 28°C in Eagle's minimum essential medium (MEM), containing 10% fetal bovine serum (FBS) and 0.2 mM nonessential amino acids. Cells were observed daily for the cell cytopathic efect (CPE) for a total of one week after infection. Viral RNA was extracted from infected culture fuids on day 7, by using a QIAamp Viral RNA Mini Kit (QIAGEN), and the presence of DENV-3 RNA was confrmed by RT-PCR [19]. Undiluted virus stock (i.e., infected culture fuids) was stored at −80°C until use. Te virus isolates were used within fve passages and maintained in C6/36 cells throughout the study.
In this study, human neuroblastoma SKNSH cells (ACTT ® -HTB-11 ™ ), human glioblastoma T98G cells (ACTT ® -CRL-1690), and murine neuroblastoma N2A cells (ACTT ® -CCL-131) were used to determine the infectivity of the virus isolates from CSF specimens and serum samples. Te cells were maintained and cultured according to the manufacturer's instructions.

Sequence and Phylogenetic
Analyses. Due to the limited amount of CSF and serum specimens, samples were initially inoculated onto C6/36 mosquito and Vero cells. RNA was extracted from infected culture fuids using the QIAamp Viral RNA Mini Kit (QIAGEN), and the presence of DENV-3 RNA was confrmed by RT-PCR. Te RNA sequence was analyzed using an Ion Proton (Termo Fisher) and a conventional capillary sequencer (3100-Avant Genetic Analyzer, Life Technologies). Nucleotide sequences were aligned using the MAFFT software version 7.215 [20]. Te substitution models were selected using jModelTest-2.1.7 [21], and GTR + I + G was used as the model.   Tables 1 and 2). Reactions were incubated at 50°C for 5 min, followed by incubation at 95°C for 20 s and 40 cycles of 95°C for 3 s and 60°C for 30 s.
To generate a standard curve, a region of 545 base pairs from the envelope sequence of DENV-3 was frst amplifed by forward primer 5′-CYTGGWTGTCDRCYGARGGAG-3′ and reverse primer 5′-TGCACCACTTTTCCCTCTAT-3′, then cloned into TOPO TA cloning vector (Invitrogen). Te plasmid of a positive colony was extracted, linearized, and then transcribed using the T7 mMessage mMachine kit (Ambion). Te in vitro transcribed RNA was diluted in a series of 10-fold dilutions equivalent to 10 7 -10 3 copies/ reaction to generate the standard curve. Tis 10-fold standard curve was run in parallel with the test samples in the 96microplate. Te copy number of the DENV-3 RNA from infected cells was measured based on the standard curve.

Immunofuorescence Microscopy Assay (IFA).
An immunofuorescence microscopy assay was used to determine the infectivity of the CSF-derived DENV-3 and serumderived DENV-3 strains in human neuroblastoma (SKNSH), glioblastoma (T98G), and mouse neuroblastoma cell lines (N2A). For this purpose, the diferent cell lines were infected with either CSF-derived DENV-3 or serum-derived DENV-3 at a multiplicity of infection (MOI) of 0.1, 0.01, and 0.001. Te infected cells were harvested on day 2 postinfection (P.I) and centrifuged at 100 rpm for 5 min. Te cell pellet was washed three times with PBS (1×) and applied onto a Tefon-coated eight-multiwell glass slide (MP Biochemicals, CA, USA). After complete drying, the cells were fxed for 10 min in a fxative solution consisting of methanol and acetone at a ratio of 1 : 1 at room temperature (RT). Te slides were then washed three times in PBS (1×) with gentle shaking and then air-dried for 10 min at RT. For each test well, 15 μl of rabbit polyclonal antibody against the DENV envelope protein and nonstructural proteins 1-5, except NS2A (GeneTex, USA), was diluted 1 : 100 in PBS (1×). Samples were incubated with each of the diluted antibody mixtures for one hour in a moist chamber at 37°C. Anti-Flavivirus monoclonal antibody 12D11/7E8 [22] was used as a positive control. Te slides were washed three times in PBS (1×) and air-dried for 10 min at RT. Next, 15 μl of goat antirabbit IgG H&L (ab150077-ALexa Fluor ® 488, Japan) and 15 μl of goat anti-mouse IgG H&L (ab150113, ALexa Fluor ® 488, Japan) at a dilution of 1 : 50 were applied to each test well and to the positive control well, respectively, and incubated for one hour at 37°C in the dark. Te slides were washed three times in PBS (1×) and air-dried for 10 min and fnally examined by fuorescence microscopy.

Data Analysis.
In this study, all data from the focus forming assay and real-time PCR were transformed to a base-10 logarithm for analysis. In descriptive analyses, numbers and percentages were used for categorical variables (percentage of infected cells in diferent protein expressions from an envelope, nonstructural 1-5). Te mean and standard deviation (SD) were used for continuous variables (viral load at 0, 24, 48, 96, and 120 h). For comparison in specifc groups, the chi-squared test, Fisher's exact test, Wilcoxon rank-sum (Mann-Whitney) test, and t-test were used appropriately. Statistical tests were performed using Stata 14.1 (StataCorp LP, College Station, Texas 77845, USA) and GraphPad Prism version 7.0a (GraphPad software, La Jolla, California, USA) with a 5% level of signifcance and two-tailed p values.

Phylogenetic Analysis and Comparison of Amino Acid
Sequences. Te complete envelope nucleotide sequences of the CSF-derived DENV-3 isolate (accession no. KP893717) and two serum-derived DENV-3 isolates from patients with classical dengue fever in Hai Phong, Vietnam, in 2013 (accession nos. KP893718 and KP893719) were determined and deposited in GenBank [19]. Te envelope sequences available in GenBank corresponding to strains that belong to the major branches of DENV-3 phylogenies were included in the analysis. Envelope nucleotide sequences were initially aligned using MAFFT version 7.215 [20], and the substitution models were selected by jModelTest-2.1.7 [21]. GTR + I + G was used as the model to construct the phylogenetic trees using the maximum likelihood method using PHYML 3.0.1 [23]. Phylogenetic analysis showed a close relationship between CSF-derived DENV-3 and the two serum-derived DENV-3 strains. All of the isolates belonged to DENV-3 genotype-III ( Figure S1). DENV-3 genotype II was found to circulate in Vietnam in 2003, 2006, 2007, and 2009. DENV-3 genotype-III was frst reported in the Indian subcontinent since the 1960s [24] and then in Sri Lanka in the 1980s and the 1990s. Since the 2000s, the number of dengue cases caused by this genotype has substantially increased in many countries, including Cambodia, Bhutan, Tailand, Laos, Pakistan, China, Senegal, and Singapore [24][25][26][27][28]. According to the national data report in 2013, two other dengue cases isolated in the Ha Tinh province were determined to be caused by DENV-3 genotype-III ( Figure S1). Terefore, it is possible that the emergence of DENV-3 genotype-III in Vietnam occurred before 2013 and spread throughout the region, including Hai Phong, but also the neighboring areas in the north of Vietnam by 2013.
Next, we determined the full-length genome sequence of the DENV-3 strains isolated from serum and CSF samples and mapped the amino acid diferences between the strains [19]. Te three strains did not have any signifcant diferences in the nucleotide and amino acid substitutions in the envelope sequence [19]. However, full genome analyses demonstrated a Tr-1339-Ile mutation (threonine was substituted by isoleucine) in the nonstructural region 2A of the CSF-derived DENV-3 ( Figure S2) and an Ala-3018-Tr mutation (alanine was substituted by threonine) in the nonstructural region 5 of the serum-derived strain (HP-5528) ( Figure S2). In this study, the infectivity and phenotypes of the two DENV-3 strains, a serum-derived DENV-3 strain (HP-5528) and a CSF-derived DENV-3 strain (CSF-11098), were determined in vitro using mosquito cell C6/36 and human and mouse neuronal cell lines.

DENV Infectivity in Neuronal Cell Lines as Determined by Focus Forming Assay (FFA).
To determine the infectivity of the isolates, a focus forming assay on the Vero cell was performed to determine the virus titer in each cell line.

Levels of Viral Structural and Nonstructural Proteins in DENV-3 Infected Human Neural Cell Lines.
Te infectivity of the CSF-derived DENV-3 isolate (CSF-11098) and serumderived DENV-3 isolate (HP-5528) in diferent neural cell lines was determined by measuring the proportions of cells presenting structural (envelope) and nonstructural proteins (NS1, NS2B, NS3, NS4A, NS4B, and NS5) (Figures 4 and 5). All infected cells were harvested 48 h P.I. At MOI � 0.01, the percentage of C6/36 cells infected with CSF-derived DENV-3 was signifcantly higher than that of cells infected with serum-derived DENV-3 (Figure 4(a)) (p < 0.05). In human neural cells (SKNSH and T98G), the proportion of infected cells detected by antibodies against NS1, NS4B, and NS5 was signifcantly higher in the group infected with the CSFderived strain than in that infected with the serum-derived strain (Figures 4(b) and 4(c)) (p < 0.05). However, there were no signifcant diferences in the infected cells between the CSF-and serum-derived isolates in the mouse neural cell line (Figure 4(d)) across the panel of antibodies used. Te results indicate that while virus growth is supported in human neural cell lines, surface antigens may be expressed diferentially between CSF-and serum-derived isolates.

Discussion
Dengue virus infection is a major public health issue affecting almost every country in the tropical and subtropical regions of the world. In the past two decades, the neurological manifestation of dengue has been increasingly recognized due to its severity and sequelae. In most cases, the neurological involvement of dengue manifests through conditions such as mononeuropathies, polyneuropathies, and Guillain-Barré syndrome. Since 2009, the World Health Organization has recategorized CNS involvement as severe dengue [18]. To date, CNS involvement in DENV has been reported in 25 countries spanning almost all continents [29].
Te pathogenic mechanism by which the dengue virus causes neurological manifestations, especially dengue encephalitis, is still poorly understood. In DENV neuropathogenesis, it has been hypothesized that either DENV passively crosses the blood-brain barrier (BBB) and actively invades the CNS system or that the symptoms are the consequence of an opportunistic infection [13]. However, mounting evidence such as the detection of dengue viral RNA in CSF and DENV antigen in brain tissue strongly supports that the virus itself is neurovirulent despite not being classically categorized as a neurotropic pathogen [13,30,31]. Moreover, the simultaneous negative fndings of viral RNA in CSF and serum samples suggested that the virus might actively enter the CNS rather than passively crossing the BBB [3,32,33]. Te results of our study of the CSF-derived strain have suggested that mutations within nonstructural proteins could be associated with DENV neuropathogenesis.
In this study, the threonine residue was substituted by the isoleucine residue at position 1339 of the nonstructural 2A glycoprotein in the CSF-derived DENV-3 isolate (CSF-11098). In contrast, there was a substitution of the alanine residue by a threonine at position 3018 in the nonstructural NS5 glycoprotein in the serum-derived DENV-3 isolate (HP-5528). Mutations of three amino acids in the envelope (E) and nonstructural 3 (NS3) gene in DENV-1 produced a neurovirulent phenotype that resulted in extensive brain damage with encephalitis and lepto-meningitis in a mouse model [34]. In 1996, a substitution of alanine by a valine residue at position 173 of the envelope gene was reported in encephalopathy cases associated with a DENV-2 serotype [29]. In addition, Phe-401-Leu and Asp-390-His mutations in the E protein have also been linked as genetic determinants of DENV-2 neurovirulence [35]. Additionally, another study demonstrated that the cleavage of the N153 glycosylation site due to Tr-155-Ile substitution was Canadian Journal of Infectious Diseases and Medical Microbiology responsible for the neurovirulence of DENV-4 in mice [36]. Taken together, these studies suggest that the structural protein (envelope) plays a crucial role in the pathogenesis of neuroinvasiveness and neurotropism of the dengue virus. While the CSF-isolate of this study presented none of these reported amino acid substitutions, our results suggest that distinct regions of the NS2A (from CSF-derived-DENV-3) and NS5 proteins (from serum-derived DENV-3) may be associated with the neurovirulence of DENV-3.
In this study, the mutation in the NS5 region in the serum-derived DENV-3 isolate (Supplementary Figure 2) has been hypothesized to be associated with the abrogation     of viral replication in vitro. A previous study demonstrated that a chimeric virus strain with a mutation in the NS5 gene exhibited comparatively lower growth rates [37]. While diferences in the clinical presentation were absent when a mouse model (A129) was inoculated with CSF-derived DENV-3 and serum-derived DENV-3 isolates, further studies are needed to determine the pathogenesis of DENVrelated neurological involvement in human patients. Next, in vitro assays were performed to evaluate the infectivity of CSF-derived DENV-3 (CSF-11098) and serumderived DENV-3 isolates (HP-5528) in human neural cells (SKNSH and T98G) and mouse neuroblastoma cells (N2A). Both isolates propagated well in the human neuroblastoma cell line (SKNSH) compared to the other neural cell lines (T98G and N2A), regardless of the multiplicity of infection ( Figure 3). Additionally, the human glioblastoma cell line T98G was a highly permissible cell line for the propagation of the CSF-derived DENV-3 isolate as compared to the serum-derived DENV-3 isolate (Figures 2 and 3).
Te neurovascular unit of the BBB consists of astrocytes, pericytes, neurons, and endothelial cells. All of these components function to selectively control the passage of molecules from capillaries into the brain parenchyma and vice versa [38].    virus (JEV) is known to cause dramatic changes in BBB function, mainly in permeability and selectively. Tese changes facilitate plasma leakage and the entry of viruses or infected cells into the brain parenchyma, which promotes the spread of the virus. Te brain cell type that is infected by DENV remains controversial. Many studies have demonstrated that DENV is more likely to infect microglial cells, which are macrophage-like resident immune cells in the brain [39]. Another study found that mouse brain endothelial cells (MBECs) were highly susceptible to DENV-4 and neuro-adapted DENV variant infections [38]. In this study, T98G cells were used for the in vitro experiment, and these cells exhibited the characteristics of adherent fbroblasts originating from human glioblastoma multiforme. Te cellular origin of glioblastoma is unknown, but it has been suggested to be derived from glial cells (astrocytes, oligodendrocytes, microglia, and ependymal) and other neural stem cells. In addition, T98G also expresses the surface markers CD19, CD44, CD90, CD105, and CD133, which are typical of mesenchymal cells [40]. Furthermore, this cell type demonstrated high gene expression of the main angiogenesis inductors vascular endothelial growth factor (VEGF) and fbroblast growth factor 2b (FGF2-b), as well as the matrix protein thrombospondin-1, which is involved in the activation of transforming growth factor β1 (TGFβ1) [40]. Te association between these chemokines and the pathogenesis of neurological manifestations in DENV infection requires further study. However, in JEV encephalitis, astrocyte infection results in the release of VEGF, which alters endothelial cell junctions by causing proteasomal degradation of Zonula occludens 1 (ZO-1) [41]. Interestingly, the CSF-derived DENV-3 isolate demonstrated high growth in human neural cell lines (especially in T98G cells), but limited growth was observed in a mouse neuroblastoma cell line, suggesting that the isolate preferentially infected human neural cells. Our data indicate that (1) the CSF-derived DENV-3 strain has unique virulence features potentially associated with the NS2A protein, which might play a crucial role in the neuropathogenesis of DENV infection. Te association between the amino acid substitution in the NS2A protein and neuropathogenesis in CSF-derived DENV-3 requires further investigation for elucidation.
(2) Te mutation in the NS5 protein might restrict the viral replication resulting in a low virus titer of serum-derived DENV-3 in human neural cells as compared to CSF-derived DENV-3.
(3) T98G cell line represents a permissible cell line to illuminate the pathogenesis of the DEN-3 serotype and neuro-adapted DENV-3 variants in dengue-related neurological cases.

Conclusion
We isolated a DENV-3 genotype-III strain from a CSF sample of a DENV encephalitis patient during a DENV epidemic in Vietnam in 2013. Te CSF-derived DENV-3 isolate presented enhanced growth in human neuronal cells, with a signifcant number of cells that were positive for NS1, NS4B, and NS5 antigens. Te full-length genome sequence demonstrated that a distinct amino acid substitution in the NS2A protein was unique to the CSF-derived DENV-3 strain. Tis result suggests that NS2A may be a crucial region in the neuropathogenesis of DENV-3. Taken together, our results demonstrate that a CSF-derived DENV-3 had unique infectivity characteristics for human neuronal cells, which might play a crucial role in the neuropathogenesis of DENV infection. Te association between the amino acid substitution in the NS2A protein and neuropathogenesis in CSF-derived DENV-3 requires further investigation.