Comparative analysis of neuroinvasion by Japanese encephalitis virulent and vaccine viral strains in an in vitro model of human blood-brain barrier

Japanese encephalitis virus (JEV) is the major cause of viral encephalitis in South East Asia. It has been suggested that, as a consequence of the inflammatory process during JEV infection, there is disruption of the blood-brain barrier (BBB) tight junctions that in turn allows the virus access to the central nervous system (CNS). However, what happens at early times of JEV contact with the BBB is poorly understood. In the present work, we evaluated the ability of both a virulent and a vaccine strain of JEV (JEV RP9 and SA14-14-2, respectively) to cross an in vitro human BBB model. Using this system, we demonstrated that both JEV RP9 and SA14-14-2 are able to cross the BBB without disrupting it at early times post viral addition. Furthermore, we find that almost 10 times more RP9 infectious particles than SA14-14 cross the model BBB, indicating this BBB model discriminates between the virulent RP9 and the vaccine SA14-14-2 strains of JEV. Beyond contributing to the understanding of early events in JEV neuroinvasion, we demonstrate this in vitro BBB model can be used as a system to study the viral determinants of JEV neuroinvasiveness and the molecular mechanisms by which this flavivirus crosses the BBB during early times of neuroinvasion.


Introduction
Flaviviruses such as Japanese encephalitis virus (JEV) are arthropod-borne viruses (arbovirus) that are transmitted through the bite of an infected mosquito and may cause serious human diseases [1]. JEV is the main causative agent of viral encephalitis in South East Asia, with an and easy-to-grow BBB in vitro. hCMEC/D3 monolayers displays good restricted permeability to paracellular tracers and retains most of the transporters and receptors present on in vivo BBB [24]. Accordingly, hCMEC/D3 cells have been used to investigate host-pathogen interactions with human pathogens that affect the CNS [25,26].
In the present study, we used an in vitro human BBB model consisting of hCMEC/D3 human endothelial cells cultivated on permeable supports above SK-N-SH human neuroblastoma cells to evaluate and compare the ability of both a virulent and a vaccine strain of JEV (JEV RP9 and SA14-14-2, respectively) to cross this BBB model.
A molecular cDNA clone of JEV genotype 3 strain RP9 was kindly provided by Dr. Yi-Ling Lin [27]. This plasmid was modified in our laboratory as previously described [28], generating pBR322(CMV)-JEV-RP9, and used by transfecting C6/36 cells with Lipofectamine 2000 (Life Technologies; catalog no. 11668-019) to produce infectious virus. Once a cytopathic effect was visible, viral supernatant was collected and used to infect C6/36 cells. Because we found hCMEC/D3 monolayers very sensitive to any change of medium, we ensured that viruses were produced from cells grown in the same medium as the one used to grow endothelial cells (EndoGro medium). CD.JEVAX 1 (JEV SA14-14-2) vaccine was kindly provided by Dr. Philippe Dussart (Institut Pasteur of Phnom Penh, Cambodia), and reconstituted with 500μL of DMEM. Two hundred and fifty μL of reconstituted vaccine were used to infect Vero cells for 7 days. Viral supernatants were collected and used to infect C6/36 cells cultivated in EndoGro medium supplemented with 2% FBS. Both JEV RP9 and SA14-14-2 viral supernatant stocks were collected 3 days after infection and the infectious titer was determined in Vero cells by focus-forming assay (see below).

Evaluation of JEV neuroinvasive capacity
hCMEC/D3 cells (5.10 4 /well) were seeded on 12-well Transwell 1 permeable inserts (Corning; catalog no. 3460) in EndoGro medium supplemented with 5% FBS and placed at 37˚C for 5 days. SK-N-SH cells (2.10 5 /well) were seeded in 12-well tissue culture plates in EndoGro supplemented with 2% FBS. Permeable inserts containing hCMEC/D3 cells were then transferred in these culture plates and medium was replaced by EndoGro medium supplemented with 2% FBS. Aliquots of virus were diluted the next day in 50μL of EndoGro medium supplemented with 2% FBS, heated at 37˚C and then added to the cells. Cells were incubated at 37˚C until collection.

Focus-forming assay (FFA)
Vero cells were seeded in 24-well plates (10 5 /well). Ten-fold dilutions of virus samples were prepared in DMEM and 200μL of each dilution was added to the cells. The plates were incubated for 1h at 37˚C. Unabsorbed virus was removed and 800μL of DMEM supplemented with 0.8% carboxymethyl cellulose (CMC), 5 mM HEPES buffer, 36 mM sodium bicarbonate, and 2% FBS were added to each well, followed by incubation at 37˚C for 48h for JEV RP9 or for 72h for JEV SA14-14-2. The CMC overlay was aspirated, and the cells were washed with PBS and fixed with 4% paraformaldehyde for 20 min, followed by permeabilization with 0.1% Triton X-100 for 5 min. After permeabilization, the cells were washed with PBS and incubated for 1h at room temperature with anti-E antibody (4G2), followed by incubation with HRPconjugated anti-mouse IgG antibody. The assays were developed with the Vector VIP peroxidase substrate kit (Vector Laboratories; catalog no. SK-4600) according to the manufacturer's instructions. The foci were then counted in each well manually. The viral titers were expressed in focus-forming units (FFU) per milliliter.

Lucifer Yellow (LY) permeability assays
LY dye migration through the BBB monolayers was performed as previously described [25,26]. Briefly, Transwell 1 inserts containing hCMEC/D3 monolayers were transferred in culture wells containing 1.5 mL of Hanks' Buffered Salt Solution (HBSS) supplemented with 10 mM of HEPES buffer, 1 mM of sodium pyruvate and 50μM of LY (Sigma-Aldrich; catalog no. L0144). The culture medium inside the Transwell 1 inserts was replaced with 500μL of HBSS buffer containing 50μM of LY. Cells were incubated at 37˚C for 10 min. Permeable inserts were then transferred in culture well containing 1.5 mL of HBSS buffer and incubated at 37˚C for 15 min. They were then transferred in culture well containing 1.5 mL of HBSS buffer and incubated at 37˚C for 20 min. Concentrations of LY in the wells were determined using a fluorescent spectrophotometer (Berthold, TriStar 2 LB 942). The emission at 535 nm was measured with an excitation light at 485 nm. The endothelial permeability coefficient of LY was calculated in centimeters/min (cm/min), as previously described [30].

Virus infections
hCMEC/D3 cells (10 5 ) were seeded on coverslips in 24-well tissue culture plates in EndoGro medium supplemented with 5% FBS. After 5 days, cell medium was replaced with 1 mL of EndoGro medium supplemented with 2% FBS. SK-N-SH cells (10 5 ) were seeded on coverslips in 24-well tissue culture plates in DMEM supplemented with 2% FBS. Aliquots of virus were diluted in 200μL of medium and added to the cells. Plates were incubated for 1h at 37˚C. Unabsorbed virus was removed and 1mL of EndoGro or DMEM supplemented with 2% FBS was added to the cells, followed by incubation at 37˚C until collection.

Immunofluorescence analysis (IFA)
All the following steps were performed at room temperature. Cells were fixed with 4% paraformaldehyde for 20 min followed by permeabilization with 0.1% Triton X-100 for 5 min. After permeabilization, the cells were washed with PBS and incubated for 5 min with PBS containing 1% BSA. The cells were then washed with PBS and incubated for 1h with anti-JEV NS5 antibody diluted at 1:200 in PBS, followed by incubation with Alexa Fluor 488-conjugated antimouse IgG antibody diluted at 1:500 in PBS. The coverslips were mounted with ProLong gold antifade reagent with DAPI (Life Technologies; catalog no. P36931). The slides were examined using a fluorescence microscope (EVOS FL Cell Imaging System).

Gene expression studies
hCMEC/D3 cells (5.10 4 /well) were seeded on 12-well Transwell1 insert filters in EndoGro medium supplemented with 5% FBS for 5 days. SK-N-SH cells (2.10 5 /well) were seeded in 12-well tissue culture plates in EndoGro supplemented with 2% FBS. Transwell 1 containing hCMEC/D3 cells were then transferred in these culture plates and medium was replaced by EndoGro medium supplemented with 2% FBS. Cells were incubated at 37˚C. At 24h post-contact, total RNA of hCMEC/D3 cells were extracted using NucleoSpin RNA kit (Macherey-Nagel; catalog no. 740955.50) following the manufacturer's instructions. Two hundred ng of total RNA were used to produce cDNA using the SuperScript II Reverse Transcriptase (Thermo Fisher; catalog no. 18064014) according to the manufacturer's instructions. Quantitative PCR were performed on 2μL of cDNA using SYBR Green PCR Master Mix (Thermo Fisher; catalog no. 4309155) according to the manufacturer's instructions. The CFX96 real-time PCR system (Bio-Rad) was used to measure SYBR green fluorescence with the following program: an initial PCR activation at 95˚C (10 min), 40 cycles of denaturation at 95˚C (15s) and annealing-extension at 60˚C (1 min). Results were analyzed using the CFX Manager Software (Bio-Rad) gene expression analysis tool. GAPDH was used as the reference gene. Primers used in gene expression studies are listed in Table 1.

Quantification of JEV RNA copy number
Total RNA from JEV BBB-crossing samples was extracted using NucleoSpin1RNA kit (Macherey-Nagel; 740955.50) according to the manufacturer's instructions. The number of JEV RNA copies present in BBB-crossing samples was determined by RT-qPCR using Taq-Man1 Fast Virus 1-Step Master Mix kit (Applied Biosystems1, 4444432) according to the manufacturer's instructions. The forward and reverse primers (Sigma-Aldrich1) were 5'GAAGATGTCAACCTAGGGAGC3' and 5'TGGCGAATTCTTCTTTAAGC3' respectively, while [6FAM]AAGAGCCGTGGGAAAGGGAGA[BHQ1] was the probe for the assay. JEV RNA copies were calculated from a standard curve generated by amplifying known amounts of in vitro-transcribed RP9 NS5 gene region cloned and under SP6 promotor control. The in vitro transcription was performed using mMESSAGE mMACHINE™ SP6 kit (Invitrogen, Thermo Fisher Scientific, AM1340) following the manufacturer's instructions.

Statistical analysis
Unpaired two-tailed t test, Mann-Whitney test and ANOVA test corrected with Tukey method for multiple comparisons were used to compare experimental data. GraphPad Prism 7 was used for these statistical analyses. The significance level for our data was set to 5% or less (P �0.05).

hCMEC/D3 cell monolayers grown on permeable inserts form a BBB whose properties are not affected by SK-N-SH cell presence
A basic in vitro model to study JEV neuroinvasion should consist of two main components: 1) a cell monolayer mimicking the BBB, and 2) a brain tissue-derived cell line permissive to JEV.
Based on our previous work [26], we chose to use hCMEC/D3 human endothelial cells monolayers cultivated on permeable inserts and place these inserts in wells in which human neuroblastoma SK-N-SH cells were grown, in order to partly mimick the brain parenchyma. Relevant parameters of a functional BBB model, such as permeability and presence of cell transporters and receptors specific of hCMEC/D3 cells were evaluated when the endothelial cells were grown or not above SK-N-SH monolayers (Fig 1). Permeability measurement of hCMEC/D3 monolayers through evaluation of Lucifer Yellow (LY) passage showed no significant difference whether SK-N-SH cells were present or not (Fig 1A, + or-respectively). Moreover, the relative levels of mRNA for genes encoding cell receptors ( Fig 1B) and transporters ( Fig 1C) characteristic of endothelial barriers were similar in the two conditions. Together, this suggests that the culture of neuroblastoma cells under the inserts on which hCMEC/D3 were grown did not disturb the intrinsic BBB endothelial cell properties and affirms that this in vitro BBB model can be used as a tool to study the neuroinvasiveness of JEV.

JEV SA14-14-2 is less replication efficient than JEV RP9 in SK-N-SH cells
Independent reports have found that neuroblastoma-derived SK-N-SH cells are susceptible to both the virulent JEV RP9 strain and the SA14-14-2 attenuated strain [28,35]. To directly compare replication of these two JEV strains in SK-N-SH cells, however, we evaluated replication of each JEV strain in these cells at 24 and 48 hpi (Fig 2). As demonstrated by the detection of a viral antigen (NS5 protein) through immunofluorescence assays, both JEV strains infected  the SK-N-SH cells but the viral progeny of JEV SA14-14-2 vaccine strain produced in SK-N-SH cells at 24 and 48 hpi was significantly lower than that of JEV RP9 (1.7 and 1.2 log 10 less at 24 and 48 hpi respectively, Fig 2B), suggesting that JEV SA14-14-2 is less neurovirulent than JEV RP9 in human cell cultures.

Neither JEV RP9 nor JEV SA14-14-2 infects hCMEC/D3 cells after they form a BBB
In order to examine the susceptibility of our hCMEC/D3 BBB model to JEV infection, the cells were grown 6 days on coverslips to allow the BBB to form, and then inoculated with either the RP9 or SA14-14-2 JEV strain (Fig 3). As evidence of infection we assessed the expression of  none was observed at either 24 or 48 hpi (Fig 3A). On the other hand, hCMEC/D3 cells could be infected by either JEV strains when they were inoculated after only one day of culture (ie not forming of a BBB), as detected through the same immunofluorescence approach (Fig 3B). Moreover, in this condition, both JEV strains produced infectious viral progeny in hCMEC/D3, although the RP9 viral titer was significant higher by around 2 log than that observed for SA14-14-2 ( Fig 3C). These results suggest that hCMEC/D3 cells are not susceptible to JEV infection when they already have formed a barrier, but they are JEV permissive before tight junction formation.

Neither JEV RP9 nor JEV SA14-14-2 disrupt the BBB when added for 6h
It has been suggested that JEV infects brain tissue cells as a consequence of a preceding inflammatory process which in turn leads to disruption of the BBB and viral neuroinvasion [36,37]. Our knowledge of the very early events of JEV crossing the BBB is, however, still scant. In order to shed some light on these early times of viral exposure, we evaluated the neuroinvasive ability of JEV in our BBB model at early times post viral addition. hCMEC/D3 cells were cultivated on permeable inserts to form a BBB above a SK-N-SH cell monolayer and exposed to either JEV RP9 or SA14-14-2 viruses at MOIs of 1 or 10 (Fig 4). As assayed by Lucifer Yellow permeability, we found the BBB integrity was not significantly compromised by either JEV strain when compared to mock-infected conditions (Fig 4A), suggesting that the BBB model was not disturbed either by the JEV strains or the MOIs used.

More JEV RP9 infectious particles may cross the in vitro BBB model than JEV SA14-14-2
Since the BBB permeability was not affected by the addition of either virus, we quantitated the viral crossing of each strain by assaying the quantity of viral RNA and infectious particles in the supernatants under the inserts (Fig 4B and 4C). The number of viral RNA copies detected for both viruses was 1.7 log 10 higher when a MOI of 10 was used in comparison to a MOI of 1 (Fig 4B), suggesting that the higher the JEV viral load, the greater the number of viral particles crossing the BBB. Of note, there was no significant difference in the viral RNA copy number between the JEV strains for each MOI (MOI = 1 or = 10, Fig 4B). However, the infectious titers of the JEV particles that crossed the BBB was notably different between the RP9 and SA14-14-2 strains, as about 3 times more RP9 infectious particles compared to SA14-14-2 where found in the supernatants under the inserts when an MOI of 1 was used, and close to 10 times more for a MOI of 10 ( Fig 4C). Calculation of the specific infectivity for JEV RP9 and SA14-14-2 strains as the ratio between the detected JEV RNA copy number per infectious focus-forming unit did not show a significant difference between the 2 viral stocks (Fig 5A). Interestingly, the specific infectivity for the RP9 BBB-crossing samples was significantly lower than that observed for the vaccine strain SA14-14-2 with a 3 to 10 fold decrease for MOI of 1 and 10 respectively (Fig 5B). These results indicate that more JEV RP9 infectious particles may cross our BBB model than SA14-14, and demonstrate that this in vitro barrier is capable of discriminating between 2 viruses with different neuroinvasive capabilities. stained by DAPI (in blue). C) Supernatants from non-forming BBB hCMEC/D3 cells infected by JEV RP9 (black bar) or JEV SA14-14-2 (white bar) were collected at 24 and 48 hours post-infection and their viral titer was determined as described in Material and Methods. The arithmetic means ± standard deviation of three independent experiments performed in triplicate is shown. Asterisks indicate a significant difference between RP9 and SA14-14-2 for each time post-infection evaluated ( �� , P = 0.0056, ��� , P < 0.001).
https://doi.org/10.1371/journal.pone.0252595.g003 In vitro BBBs were generated as indicated above and either JEV strains were added at MOI = 1 or = 10. After 6 h, total RNA was extracted from media under the inserts and the number of JEV RNA copies was determined by RT-qPCR as

Discussion
Lines of research from both in vivo and in vitro systems have suggested JEV infects brain tissue cells as a consequence of a preceding inflammatory process that in turn may facilitate BBB described in Material and Methods. C) Samples were collected 6 h post-addition (JEV RP9, black bars or SA14-14-2, white bars) and their viral titer was determined as described in Material and Methods. The arithmetic means ± standard deviation of at least two independent experiments performed in triplicate is shown. Asterisks indicate a significant difference between the RP9 and SA14-14-2 titers for each MOI evaluated in the BBB-crossing experiments ( ���� , P < 0.0001).
https://doi.org/10.1371/journal.pone.0252595.g004 disruption and viral neuroinvasion [36,37]. While in vivo approaches primarily give insights to systemic viral disease, in vitro models tend to allow examination and manipulation of the molecular mechanisms that govern viral pathogenesis. In this regard, previous approaches have generally focused on characterizing JEV neuroinvasive properties at late times of infection, mainly 24 hpi or later [35,[38][39][40], leaving our knowledge of events at early times of JEV contact with the BBB poor, if not null.
In this study, we have used an in vitro human BBB model to compare the ability of two JEV strains (the virulent RP9 strain and the SA14-14-2 vaccine strain) to cross the BBB at early times post-addition. We have shown that both JEV RP9 and SA14-14-2 are able to cross the BBB without disrupting it at 6 hpi. Our finding corroborates in vivo studies that have demonstrated that JEV is able to get access to the CNS and establish a primary infection without the preceding need of BBB leakage [36,37].
Moreover, the fact that both JEV RP9 and SA14-14-2 strains crossed the BBB without infecting BBB endothelial cells, or disrupting the barrier, also suggests that the pathway JEV uses to cross the BBB is either a transcellular one, through the endothelial cells, or paracellular, between the endothelial cells. These observations are consistent with other studies conducted in vivo in mice and monkeys [11,16,41]. Electron-microscopic studies of brains from JEVinfected suckling mice have suggested that viruses cross the BBB endothelial cells by transcytosis [14]. In spite of these observations, to date there are no published data from biochemical, genetics or functional approaches to support or refute this hypothesis. The combination of these approaches, together with the use of our in vitro BBB model and JEV strains with different neuroinvasive capabilities(such as the ones used in this work) would be useful to identify which cellular mechanisms might be "hijacked" by these pathogens to cross the BBB.
Interestingly, although our specific infectivity data suggest that JEV RP9 infectious particles crossed the BBB more efficiently than those of the vaccine strain JEV SA14-14-2, comparison of hCMEC/D3 cell transcriptomes from BBBs that were exposed for 6h to either JEV RP9, SA14-14-2 or no virus showed no significant difference in the levels of gene expression (foldchange threshold of 2, data not shown). This suggests that an immediate or early cellular response is unlikely to be responsible for the differential BBB crossing of JEV RP9 versus JEV SA-14-14-2 particles we observed. Instead, we suspect specific viral factors to be at play, for example, interaction of the viral particle with a strain-specific cellular surface receptor for viral entry. Other considerations to pursue such as full characterization of the viral particles that are able to cross the BBB including, by deep-sequencing of their RNA genomes, and an electron microscopic examination of the endothelial cells forming the BBB after contact with either virus, could help to shed significant light on this intriguing difference.
Interestingly, we found that hCMEC/D3 were permissive to both RP9 and SA14-14-2 strains only when the BBB formation was not completed. BBB formation induces changes in cell conformation, which can then lead to the relocation of cell receptors between BBB cells [42]. Differences in hCMEC/D3 cells permissiveness could be due to differential accessibility of cell receptors when BBB is formed. Based on our data, and considering the current model of JEV neuroinvasion that suggests disruption of the BBB following CNS viral infection [11], endothelial cells from a disrupted barrier might become permissive to JEV because of better accessibility to cell entry receptor(s), and these cells, upon infection, could in turn become a new source of viral production contributing to JEV infection of the CNS.
In conclusion, our study demonstrates that both a virulent and a vaccine strain of JEV are able to cross a BBB model without disruption at early times post viral addition. This BBB formed by human endothelial cells represents a useful discriminant in vitro model to characterize viral determinants of JEV neuroinvasiveness as well as a tool to study the molecular mechanisms by which these pathogens cross the BBB.

Acknowledgments
Transcriptomic analysis was performed by the Pôle Biomics of the Institut Pasteur Center for Technological Resources and Research (C2RT). We thank Dr. Philippe Dussart for providing the JEV SA14-14-2 vaccine, Dr. Yi-Lin Ling for providing the JEV-RP9 cDNA clone, Dr. Yoshiharu Matsuura for providing the anti-JEV NS5 antibody and Dr. Jonathan Bradley for text edition.