PLVAP and GKN3 Are Two Critical Host Cell Receptors Which Facilitate Japanese Encephalitis Virus Entry Into Neurons

Japanese Encephalitis Virus (JEV), a globally important pathogen, belongs to the family Flaviviridae, is transmitted between vertebrate hosts by mosquitoes, principally by Culex tritaeniorhynchus. The E-glycoprotein of the virus mediates its attachment to the host cell receptors. In this study, we cloned and purified JEV E-glycoprotein in pET28a vector using E. coli BL21 (DE3) cells. A pull down assay was performed using plasma membrane fraction of BALB/c mouse brain and E-glycoprotein as a bait protein. 2-Dimensional Gel Electrophoresis based separation of the interacting proteins was analyzed by mass spectrometry. Among all the identified partners of E-glycoprotein, PLVAP (Plasmalemma vesicle associated protein) and GKN3 (Gastrokine3) showed significant up-regulation in both JEV infected mouse brain and neuro2a cells. In-silico studies also predicted significant interaction of these receptors with E-glycoprotein. Additionally, overexperssion and silencing of these receptors resulted in increase and reduction in viral load respectively, suggesting them as two critical cellular receptors governing JEV entry and propagation in neurons. In support, we observed significant expression of PLVAP but not GKN3 in post-mortem autopsied human brain tissue. Our results establish two novel receptor proteins in neurons in case of JEV infection, thus providing potential targets for antiviral research.

a E-glycoprotein interactor in neuro2a cells 11 . Endoplasmic reticulum chaperone GRP78 is also an important molecule responsible for JEV internalization into neurons 12 . The other proteins taking part in JEV entry into host cells are laminin receptors and α5β3 integrin [13][14][15] . Although, JEV infects a broad spectrum of host cells, the major membrane receptors on the brain epithelium critical for viral entry still remain largely unknown. Hence we aimed to apply a pull down based proteomic approach to identify mouse brain membrane receptor proteins of JEV E-glycoprotein.
In the present study, JEV E-glycoprotein gene was cloned and expressed in E. coli BL21 (DE3) strain followed by purification through Ni-NTA beads. Simultaneously, plasma membrane fraction of 3-4 week old BALB/c mouse brain was extracted and a pull down analysis was performed using JEV E-glycoprotein as a bait protein which was then followed by 2-DE (2-dimensional gel electrophoresis) separation and mass spectrometry. Amongst the identified proteins, PLVAP (Plasmalemma vesicle-associated protein) and GKN3 (Gastrokine 3) receptor proteins were found to be significantly present in the membrane fraction of mice brain following JEV infection. We also found their presence in mouse neuro2a cell membrane, primary cortical neurons and SH-SY5Y cells at earlier time points of viral infection. Furthermore, silencing these proteins in mouse neuro2a cells prevented the viral RNA production as well as translation of viral proteins. Upon their overexperssion, viral RNA replication and protein translation were increased. In a parallel study, we found higher expression of PLVAP in basal ganglia region of autopsied human brain tissue of JE cases when compared to age matched controls of accidental injury cases. Together, our findings suggest PLVAP and GKN3 receptor proteins to be critical host factors governing JEV internalization into neurons.

Results
JEV E-glycoprotein interacting partners in the mouse brain epithelium. E-glycoprotein induction was standardized at different concentrations of IPTG and at different temperatures (data not shown). Protein expression was finally induced at 25 °C with 0.2 mM IPTG for 6 hrs. (Fig. 1A, Fig. S1). Affinity pull down analysis was performed using JEV E-glycoprotein of GP78 strain (mouse adapted) as a bait protein to identify the interacting proteins in the mouse brain membrane. Briefly 1 mg of membrane protein was incubated with 5 mg of purified His-tagged E-glycoprotein. The purity of the membrane fraction was tested by immunoblot using Caveolin and lactate dehydrogenase before proceeding with the pull down experiment (Fig. S2). After separation of the proteins by 2-DE, both silver staining ( Fig. 2A,B) and coomassie staining (Fig. 2C) were done to cover a broad range of host proteins interacting with JEV E-glycoprotein. Spots that were common in biological replicate sets were identified and excised for identification by mass spectrometry. Identified proteins are enlisted in Table 1.
Expression, modeling and protein-protein docking of the identified membrane proteins in mouse brain post JE virus infection. After proteomic identification of the E-glycoprotein interactome, we intended to identify the ones having differential expression in viral infection. We experimented on both infant (10 day old) and adult (3-4 weeks old) age groups of mock infected and JEV infected BALB/c mice. Animals were either mock infected with PBS or infected with virus. Animals showed encephalitic symptoms after day 5 post infection. On day 7, brain samples were collected. Expressions of identified receptor proteins were validated through qRT-PCR. Five membrane proteins namely PLVAP, LRAT, SRC8, GKN3 and EXOC8 showed notable up-regulation in both the age groups of infected animals when compared to mock (Fig. 3A,B). The three dimensional structures of these membrane proteins PLVAP, GKN3, LRAT, EXOC8, SRC8 were successfully predicted using homology modelling and ab initio method. The stereo-chemical quality of each receptor was validated by Ramachandran plot which shows above 96% residues present in the allowed region. The molecular docking of JEV-E with its corresponding five interactors PLVAP, GKN3, LRAT, EXOC8, and SRC8 was performed by using ZDOCK and RDOCK programs. The structure with lowest E_RDock score was selected to identify important  (Table S4). The interactions are depicted in Fig. 3C-H. Residues of PLVAP protein involved in interaction with JEV-E are present in domain III region with lowest E_RDock score −29.27 kcal/ mol. The binding energy was calculated using Poisson-Boltzmann with non-polar surface area (PBSA) method 16 . The PLVAP -JEV-E complex formed seven hydrogen bonds and one salt bridge with binding energy of −162.99 kcal/mol (Table S5). The binding modes of LRAT on JEV-E protein were engaged with seven hydrogen bonds and three pi cation interactions. The best pose of LRAT is present at domain III of JEV envelope protein with binding energy of −156.29 kcal/mol (Table S6). The protein-protein interactions between JEV-E and SRC8 revealed six hydrogen bonds and five salt bridges with lowest E_RDock score of −30.47 kcal/mol. It was observed that the lowest energy pose of SRC8 is present at domain I region of envelope protein of JEV (Table S7)   Silver staining of interacting proteins on a 12% polyacrylamide gel on an IPG strip of pH 5-8. (C) Coomassie Blue staining of interacting proteins on a 12% polyacrylamide gel on an IPG strip of pH 5-8. Spots on biological replicate experiments were marked, excised and analyzed by MALDI/TOF followed by database searches. Spots are labeled on the gel according to the numbers mentioned in Table 1 value of −87.14 kcal/mol. Similarly, for GKN3, the best pose with −27.85 kcal/mol E_RDock value was selected for analyzing the binding interaction network at the interface. It was packed at domain II of JEV by forming eleven hydrogen bonds, three salt bridges and one pi cation interaction with binding energy of −88.43 kcal/mol (Table S8). It was observed that binding of EXOC8 with JEV-E occurs at domain I forming seven hydrogen bonds and two salt bridges (Table S9) with a binding energy of −130.06 kcal/mol. Study of the interaction network present at the protein-protein interface of JEV-E and its corresponding interactors indicate that the two membrane proteins PLVAP and LRAT packed at domain III region bind more tightly compared to the others.

Elevated expression of PLVAP and GKN3 in brain membrane fraction of BALB/c mice.
Membrane proteins, whose mRNA expression was up-regulated in mouse brain post JEV infection, were also validated in the brain membrane fraction by western-blot. Although five membrane proteins were showing increased mRNA expression, we only found enhanced expression of PLVAP and GKN3 in brain membrane fraction of both adult and 10 day old JEV infected mice (Fig. 4A). Ponceau profile of brain membrane proteins indicate equal loading (Fig. 4B). Histograms show densitometric normalization of the result using transferrin receptor as loading control (Fig. 4C). PLVAP and GKN3 were shown to be co-localized with JE virus in mouse brain immuno-histochemistry (Figs S3A, S4B). Their expression in mouse brain was detected in qRT-PCR at day 3 post infection till the appearance of encephalitic symptoms at day 7 ( Fig. S3C,D). Expression of PLVAP and GKN3 in mouse primary cortical neuron culture. Mouse primary cortical neurons were infected with JEV at an MOI of 5 for 15 and 30 minutes when E glycoprotein interaction with the membrane receptors is supposed to occur. After the treatment, cells were extensively washed with PBS to remove any unbound virus. qRT-PCR was performed with the mock and virus infected samples to check the expression of PLVAP and GKN3 proteins. Both of the proteins showed significant up-regulation at their mRNA level (Fig. 8A). At this particular time point of virus treatment, co-localization of virus with these receptors was clearly seen by immuno-fluorescence (Fig. 8B,C).

Expression of PLVAP and GKN3 in autopsied human tissues of Japanese encephalitis.
Parallel study was conducted in autopsied human tissue of JE-cases to validate the presence of E-glycoprotein interactors previously identified in mouse brain. Presence of JE virus was confirmed in the basal ganglia or frontal cortex autopsy tissue through RT-PCR (Fig. S17A). Age matched control non-JE tissue (accidental cases with minimum trauma to brain) was used for comparison. qRT-PCR experiment showed a significant presence of PLVAP in post-mortem tissue whereas GKN3 was un-detected (Fig. S17B). The study was conducted on three different age groups of JEV infected autopsy cases. In each age group, 2 patient samples were taken.

Discussion
Viral tropism inside host cell primarily depends on attachment to the cell surface proteins followed by its entry. Therefore, identification of host factors required for virus entry can be extremely important for drug development. Since JEV is a major threat to public health in the Asian subcontinent, identification of its cellular interactors is of immense importance. In the present study we aimed to identify the JEV E-glycoprotein interacting cellular membrane receptors in a BALB/c mice model of viral infection through a pull down based proteomic analysis. We identified two novel neuronal membrane receptors namely PLVAP and GKN3 which play an important role in viral entry. Previous studies have identified many receptors such as HSP70 and GRP78; playing important role in JE virus entry in in vitro systems 11,12,15,17-19 , but our study is the pioneer one identifying JE virus receptors in BALB/c mouse model. Through mass spectrometry, we identified 42 proteins from mouse brain which interact with JEV E-glycoprotein. Differential expression pattern of these proteins after viral infection was analyzed through qRT-PCR experiments in mock infected and JEV infected mouse brain. Amongst the identified, 5 proteins (PLVAP, GKN3, EXOC8, SRC8, LRAT) were found to be up-regulated at mRNA level in both adult and P10 mouse brain post viral infection. Since, bio-informatics based analysis of protein-protein interaction in disease biology is a growing area of research; we wanted to validate the interaction of these membrane proteins with JEV-E glycoprotein by in-silico structural modeling approach using ZDOCK and RDOCK programmes. In-silico analyses were carefully validated in our experimental in vivo and in vitro models of JEV infection. Although, in silico predictions indicated favorable binding of PLVAP and LRAT with E-glycoprotein; when analyzed in membrane protein fraction of infected animal brain, only PLVAP and GKN3 receptor proteins were found to be up-regulated. These receptor proteins were also co-localized with JEV in mouse brain (Fig. S3). Therefore, we conducted our further validation experiments in these receptor proteins only. PLVAP was found to interact with E glycoprotein domain III whereas; GKN3 interaction is mediated through the domain II (DII). Unlike DIII, DII does not directly take part in membrane protein binding; but its interaction with GKN3 might be advantageous for the virus to gain entry inside host system.
Up-regulation of PLVAP occurs in glioma signaling and it was identified as a Glioma Endothelial Marker (GEM) 20,21 . Increased expression of PLVAP is also an indication of blood brain barrier damage 22 which in turn is a key event in JEV pathogenesis 23 . In our experiments also, we observed elevated expression of PLVAP in both adult and 10 day old BLAB/c mouse brain post JEV infection. To find out which cell type of brain is responsible for this response, we infected three major cell types of brain namely neuron, astrocytes and microglia in vitro in    On the other hand, Gastrokine family of proteins is mainly stomach specific. While GKN1 and GKN2 are functional in human, GKN3 is an inactive pseudo gene 24 having a premature stop codon 25 . Although it is actively functional in other mammals including mice. GKN3 mRNA and protein are primarily found in mucus neck cells and Brunner's gland of duodenum 26 . In case of several gastric diseases in mouse, GKN3 has been found to be up-regulated like in cases of fundic atrophy/mucus metaplasia and antral tumorigenesis 27 , profound GKN3 expression is observed. It is also up-regulated in genetically susceptible sheep after nematode infection 28 . However, till date there is no report available showing GKN3 expression in mouse brain post viral challenge. In our studies, we found elevated GKN3 expression in both adult and 10 day old BALB/c mouse brain post JEV infection. This protein is found to be localized in mouse neuro2a cell membrane after 15 and 30 minutes of JEV infection and its expression is also increased in the plasma membrane protein fraction. No change in GKN3 expression was observed in microglia or astrocyte cell lines post early JEV infection. Purified and dialyzed E-glycoprotein treatment on mouse neuro2a cells and primary cortical neurons elevated PLVAP and GKN3 expression levels compared to untreated control, indicating no direct role of glycosylation of the E-protein expressed in bacterial system instead of mammalian system; thus being devoid of post translational modifications.
In support of our observations, we infected human neuronal cells (SH-SY5Y) with JEV for 15 and 30 minutes and checked expression of PLVAP and GKN3 at mRNA level (Fig. S5A). As in case of humans, GKN3 is a pseudo gene, we did not observe its amplification although PLVAP receptor was up-regulated manifold when compared to mock. Immunocytochemistry and Co-IP experiments also support interaction of JE virus with PLVAP receptor in human neuronal cells (Fig. S5C-E). Similar data was obtained in autopsied human tissue samples of JEV infection (Fig. S17). We isolated RNA from the paraffin embedded basal ganglia region of JEV infected patients and found out elevated expression of PLVAP receptor when compared with the age matched control cases.
JEV also infects another major cell population in the brain which are neural stem/progenitor cells and decreases their proliferation 29 . The virus induces endoplasmic reticulum stress and apoptotic response in human neural stem cells 30 . To the best of our knowledge, no previous studies have been done to identify E-glycoprotein interacting membrane proteins in this cell type. We therefore aimed to validate the presence of identified membrane proteins in mouse sub-ventricular zone and human neural stem cells (hNS1). PLVAP was found to be elevated in the membrane fraction of both adult and 10 day old infected BALB/c mice sub-ventricular zone (Fig. S15). There was up-regulation of PLVAP in hNS1 (human neural stem cell line) cells after 15and 30 min of virus infection (Fig. S5B). Virus infection in animals. [3][4] week and 10 day old BALB/c mice of either sex were divided into two groups. Food and water supply in the cages were ad libitum. Range of body weight of 3-4 week old animals was 12-15 g and for P10 animals it was 4-5 g. No of cage companion in each experiment were variable according to the litter size. Minimum no of animals per experimental group used was at least 10(5 mock, 5 virus infected). In one group of animals of either age group, 3 × 10 5 pfu of JEV was administered intraperitoneally; while the other group received equal volume of PBS. Both mock infected and virus infected animal brain tissue were collected after the appearance of encephalitic symptoms by perfusion using chilled PBS and stored at −80 °C until used for membrane protein or RNA extraction. In other set of experiments, brains were collected in 4% PFA for immunohistochemistry work. Until cryosectioning brain samples were stored at 4 °C.

Virus infection in cells.
Neuro2A cells were cultured till 80% confluency followed by serum free media addition. After 2 hours of serum free addition, cells were either mock infected or infected with 5 MOI of JE virus (GP78 strain) for 15 or 30 min. The cells were then harvested for membrane protein or RNA isolation. Viral infection in SHSY-5Y, hNS1, N9 and C8-D1A cells were performed in the same way.

Cloning and expression of JEV E-glycoprotein. The full-length JEV-E glycoprotein (GP78 strain)
gene was PCR amplified using primers JEV-E F (5′-CCGGGATCCATGGGCAATCGTGACTTC-3′) and JEV-E R (5′GCAAGCTTGATGTCAATGGCACATCCAGT-3). The 1280-bp product, corresponding to the full-length mature protein, was purified by gel extraction using a commercial kit (Qiagen, Germany) and was cloned into the pET28a vector (Qiagen, Germany) according to the manufacturer's instructions. The ligated products were transformed into Escherichia coli BL21 (DE3) cells. Plasmids were isolated from 10 randomly selected clones (colonies appeared on agar plate were randomly picked up and grown overnight in 5 ml broth) and were tested for the presence of the insert by size determination on an agarose gel (1%) and PCR amplification of the target gene. Two clones positive for the insert in the correct orientation were subjected to double-pass sequencing to check for possible mismatches using the commercial services (Xcelris Labs, India) employing an automated sequencer. The expression of recombinant protein was induced for 6 hrs. at 25 °C in 250-ml LB medium culture containing 0.2 mM isopropyl-D-thiogalactopyranoside (IPTG). The over-expressed recombinant protein was purified to near-homogeneity by using Ni-NTA Sepharose resin (Qiagen, Germany) according to the manufacturer's instructions.
Isolation of plasma membrane protein from cells or mouse brain tissue. Plasma membrane proteins from neuro2a cells or BALB/c mice brain samples were isolated using Plasma Membrane Protein Extraction Kit (Abcam) according to the manufacturer's instruction. Briefly, cells were collected in ice cold PBS followed by sonication and the tissue was homogenized using homogenization buffer mix provided in the kit. Homogenates were centrifuged and the supernatant thus collected was re-centrifuged at 10000 × g for 30 min. The pellet contained total cellular membrane protein. Plasma membrane protein fraction was separated by repeated phase separations using upper and lower phase solutions provided with the kit following the manufacturer's instructions.

JEV E-glycoprotein pull down analysis with brain membrane protein.
Affinity pull down analysis was carried out as described previously 31 . 1 mg of membrane protein was pre-cleared with Ni-NTA beads to remove non-specific binding to JEV-E-glycoprotein. The pre-cleared lysate was mixed in a micro-centrifuge tube with 5 mg of purified His-tagged JEV E-glycoprotein. The mixture was incubated on a rocking platform at 4 °C for 1 hr, and then 100 μl Ni-NTA bead suspension was added, and samples were incubated another 1 hr at 4 °C with rocking. His tagged E-glycoprotein pulled-down beads were suspended in 8 M Urea and 2% CHAPS to extract the protein complexes and the protein concentration was determined by Bradford's method.

2-DE gel electrophoresis.
The pulled down proteins were separated by 2-DE gel electrophoresis using a procedure described earlier 30,32 . 200 µg of the pulled down proteins were precipitated overnight with 10% TCA. The resulting protein pellet was washed twice with chilled acetone and centrifuged at 10000 × g for 10 minutes at 4 °C. The protein pellet was air dried for 5 minutes and resuspended in a buffer containing 7 M Urea, 4% CHAPS and 50 mM DTT and IPG strips of pH range 5-8 and 3-10 were rehydrated overnight for 2DE separation of proteins. The proteins were then focused for 10000 VHr at 20°C on a protean i12 TM IEF cell (Bio-Rad, USA). After focusing, the strips were incubated for 10 min, in 2 ml of equilibration buffer I (6 M urea, 30% w/v glycerol, 2% w/v SDS and 1% w/v DTT in 50 mM Tris/HCl buffer, pH 8.8) followed by equilibration buffer II (6 M urea, 30% w/v glycerol, 2% w/v SDS and 4% w/v iodoacetamide in 375 mM Tris/HCl buffer, pH 8.8). After the equilibration steps, the strips were transferred to 12% SDS-PAGE for the second dimension. The proteins were visualized by silver staining or Coomassie Blue R-250 staining. Images of three replicate experiments were captured in LI-COR odyssey image (LI-COR Biosciences, USA). Spots were excised and identified by mass spectrometry at Institute of Life Sciences, Odisha, India.
Mass spectrometry analysis and database searching. Proteins were identified by mass spectrometry (MS) using an AB Sciex MALDI TOF/TOF 5800 (AB Sciex, CA, USA) at Institute of Life Sciences, Bhubaneswar, after washing and in-gel trypsin digestion of gel spots 33 . All MS and MS/MS spectra were simultaneously submitted to Protein Pilot software version 3.0 (Applied Biosystems) for database searching using Mascot search engine against UniprotKB-Swissprot database containing 544996 sequences with the taxonomy group of Mus musculus. Search parameters were as follows: trypsin digestion with one missed cleavage, variable modifications (oxidation of methionine and carbamidomethylation of cysteine), and the peptide mass tolerance of 100 ppm for precursor ion and mass tolerance of ±0.8 Da for fragment ion with +1 charge state. Results obtained from database search were further analyzed. Proteins from Mus musculus species with significant Mowse scores and more than one unique peptide were identified and used for further study as shown in Table 1.
Homology modelling and protein-protein docking. The amino acid sequences of mouse PLVAP, LRAT, GKN3, EXOC8 and crystal structure of SRC8 at 1.65 Å were retrieved from Uniprot and Protein Data Bank (PDB) databases respectively. BLAST (Basic Local Alignment Search Tool) was performed for selection of templates for homology modelling of GKN3, EXOC8 and LRAT against PDB database. Proteins that were used as templates for homology modelling, along with Uniprot IDs, PDB IDs and their identities with the targeted membrane proteins are shown in Table 2. Alignment of template-target sequences were performed by ClustalW. Based on the atomic coordinates of the identical regions of the template structures, the targeted 3D structures of the respective proteins were generated by MODELLER 9v10 34,35 . As no structural analogue of PLVAP was available, ab initio method was used for prediction of 3D structure of PLVAP. ROBETTA server was used for this prediction that uses Rosetta fragment insertion method 36 .
The crystal structure of ectodomain of Japanese encephalitis virus (JEV) envelope protein (E) at 2.1 Ǻ was retrieved from PDB (PDB ID: 3P54). The envelope protein is composed of separate structural domains. Domain I (DI) consists of 127 residues (1-51, 135-193 and 283-299); domain II (DII) consists of 172 amino acids  and domain III (DIII) consists of a continuous stretch of 100 residues (300-399). The interaction between JEV (PDB ID: 3P54) and its corresponding membrane proteins was studied with protein-protein docking by ZDOCK/RDOCK programmes 37,38 implemented in Discovery Studio 3. ZDOCK is a rigid body protein docking algorithm that searches rotational space based on Fast Fourier Transformation (FFT) algorithm. Angular step size for rotational sampling was set to 15° which resulted in 3600 poses of each protein-protein complex. The top 2000 poses were retained using ZRANK score and processed by the clustering method. RDOCK algorithm was used for optimization and refinement of docked complexes from each cluster region based on CHARMm-based energy minimization process. The structure with the lowest RDOCK scores (E_RDock) was selected for interaction analysis.
RNA isolation from mouse brain tissue and neuro2a cells. Both adult (3-4 weeks) and 10 day old mouse brain tissues were homogenised in Trizol reagent (Sigma, USA) followed by addition of chloroform and centrifuged at 12000 rpm for 15 min at 4 °C for phase separation. The aqueous phase was carefully collected and mixed with isopropanol. This was again centrifuged at 12000 rpm for 15 min at 4 °C to get RNA pellet. The pellet was washed in 75% ethanol and air dried. RNA from neuro2a cells were isolated in a similar manner.  Supplementary Table S10 and S11. Primer sequences were designed at NCBI database through primer blast in the available sequences of our genes of interest.
RNA isolation from human tissue samples. RNA was isolated from paraffin embedded basal ganglia region of JEV infected human autopsy tissue (CSF positive for JEV-IgM). Age matched non-JEV control samples were accidental cases with least possible trauma to brain. Tissue samples were collected from Human Brain Bank, NIMHANS, Bangalore according to institutional ethics and confidentiality of the subjects. Tissue collection was carried out following institutional rules and guidelines. Consent was obtained from the family members of the subjects. Samples were merged in mineral oil (Sigma, USA) and heated for 2 min at 95 °C to remove paraffin according to published protocol 39 . The process was repeated until paraffin was completely removed. Then the tissue was repeatedly washed in RNAlater solution (Sigma, USA) followed by homogenization in Trizol (Sigma, USA) reagent. The aqueous phase was collected and RNA was precipitated by isopropanol. cDNA preparation and qPCR protocol were similar to that used for mouse brain tissue.
Mouse primary cortical neuronal culture. 2 day old BALB/c mouse pups were decapitated under sterile conditions and cortex was isolated in calcium-magnesium free Tyrode solution under a dissecting microscope. Cortex was digested with Trypsin and DNase enzymes to make single cell suspensions. The suspension was passed through a 127 µm pore size nylon mesh (Sefar) to eliminate cellular debris. The filtrate was centrifuged at 1000 rpm to get the cell pellet. Cells were counted in a hemocytometer and equal number of cells were seeded in poly-D-lysine (Sigma, USA) coated plates in a neurobasal medium containing 2 mM L-glutamine, 1% glucose, 5% FBS, Horse serum and penicillin-streptomycin. After two days, serum was removed from the media to inhibit glial growth and N2 and B27 supplements were added in the media. To eliminate glial presence in the culture, cells were treated with 20 µM arabinoside 1 day prior to viral infection. Neurons were infected with 5 MOI of JEV for 15 and 30 minutes followed by thorough washes with PBS to remove the unbound virus. Thereafter, cells were either immunostained for PLVAP and GKN3 or harvested for RNA isolation and qRT-PCR.
Immunostaining. Neuro2a cells were either mock infected or infected with JEV at an MOI of 5 for 15 and 30 minutes. The cells were thoroughly washed with PBS and fixed in 4% PFA which was followed by blocking with animal serum in which the secondary antibody is raised in. No permeablization step was performed. The cells were then incubated with primary antibody (PLVAP/GKN3/JEV/Caveolin) at 4 °C overnight [dilution used for each primary antibody was 1:250; sources are mentioned in Supplementary Table S1]. The cells were washed with PBXT (Triton X) and incubated with flurochrome conjugated secondary antibodies (1:500) for 1 hr at room temperature. The cells were then mounted with DAPI (Vector Laboratories, USA) following five washes with PBXT (Triton X) and were subsequently observed under a fluorescence microscope (Zeiss, Germany). Isotype controls (IgG) corresponding to the primary antibody and only secondary antibody controls were included in the experiments to ensure specific antibody binding. Similar protocol was followed in case of immunohistochemistry.
Immunoblot. Membrane proteins isolated from either mouse brain tissue or neuro2a cells were separated by SDS-PAGE with 30 µg of proteins in each well. After separation, the proteins were transferred on to nitrocellulose membrane and ponceau staining image was captured. The nitrocellulose membrane was incubated with respective primary antibodies (dilutions are provided in Table S1) overnight at 4 °C with gentle shaking. After PBST (Tween 20) washes, blots were incubated with respective secondary antibodies (1:5000, Vector Laboratories, USA) and developed in UVITECH imaging system, Cambridge using ECL reagent(Millipore, CA, USA). Transferrin receptor was used as a loading control for membrane proteins. Alternatively ponceau profile was captured to ensure equal loading.
Silencing of PLVAP and GKN3 in Neuro2a cells. Neuro2a cells grown up to 60% confluency, were transfected with 30 pmol Endoribonuclease prepared short interfering RNA (esiRNA) against mouse PLVAP and GKN3 (Sigma, USA) with Lipofectamine RNAiMAX (Invitrogen, CA, USA) according to manufacturer's instruction. Mock transfection was done with scrambled siRNA. Post 48 hrs. of transfection, cells were infected with JEV at an MOI of 5 for 15 and 30 minutes and washed with PBS thoroughly to remove any unbound virus. Then fresh media was added to the cells and 24 hrs. later cells were harvested for determination of viral load checking through qRT-PCR. In another set of experiments, cells were washed thoroughly after 15 or 30 min of viral infection to remove any unbound virus particles and then again kept for 24 hrs. at 37 °C in fresh media for analyzing JEV NS3 protein level through immunoblotting.  Plaque assay. Release of virion particles from plasmid and esiRNA treated mouse neuro2a cells were monitored in porcine stable (PS) kidney cell monolayer culture. PS cells were seeded in 6 well plates. Culture supernatants collected from transfected and infected cells were subjected to serial dilutions (multiples of 10 folds) and added to PS monolayers and incubated for 2 hrs. at 37 °C. Then the inoculum was removed and PS monolayers were covered with a solution containing 1% low melting agarose, minimum essential media, FBS and antibiotics (penicillin, streptomycin). The plates were kept at 37 °C until the plaques became visible after 72-96 hrs. To count the plaques, PS cells were stained with crystal violet solution after fixation using 10% para-formaldehyde.
Antibody-blocking experiment. Neuro2a and SH-SY5Y cells were incubated with different concentrations of PLVAP and GKN3 antibodies (N2a: PLVAP and GKN3, SH-SY5Y-only PLVAP) according to the previously published protocol 14 for 1 hr. at 4 °C. To assess the role of these receptors in viral entry, cells were incubated with 5 MOI of JE virus at 37 °C for 15 and 30 minutes. Then cells were washed with acid citrate buffer (pH 3.0) to remove unbound virus and washed thoroughly with PBS. Then fresh media was added to the cells and they were incubated for 6 hrs. at 37 °C. Cells were then harvested for RNA isolation and qRT-PCR for viral RNA load check.

Co-IP of E-glycoprotein and membrane receptors.
Purification of JEV E-glycoprotein was described earlier. Purified E-glycoprotein was bound to dynabeads containing protein G (Novex, Life Technologies) for 1 hr at 4 °C. Membrane proteins were isolated form Mock infected and JEV infected (15 min and 30 min) neuro2a and SH-SY5Y cells. Equal amount of membrane protein from each samples were then incubated with bead bound E-glycoprotein for 2 hrs. at 4 °C. To ensure specific binding, IgG was incubated with protein G beads and later incubated with 30 min JEV infected membrane protein sample. Elution buffer was added to remove the immunocomplex from the bead. The complex was probed by western-blotting using PLVAP and GKN3 antibodies (stripping and reprobing) as well as E-glycoprotein antibody.
Purified E-glycoprotein treatment in neuro2a cells and mouse primary cortical neurons. JEV-E glycoprotein was purified using Ni-NTA beads as described previously. Purified proteins were dialyzed before treatment into neuronal cells using Dialysis tubing cellulose membrane (Sigma, USA). Dialyzed protein was quantified and at different concentrations (2.5, 5, 10, 15, 20 µg/ml) it was added to neuro2a cells and primary cortical neurons for 15 and 30 minutes. Same volume of dialysis buffer was added to the cells as a blank control. After treatment cells were washed and used for either RNA isolation or immunostaining.

Lead Acetate treatment in neuro2a and SH-SY5Y cells. Neuro2a and SH-SY5Y cells were incubated
with different concentrations of lead acetate (10-150 µM) for 15 and 30 minutes. Cells were then harvested for RNA isolation and qRT-PCR.
Statistical Analysis. Data is represented as mean ± SD of three independent experiments (n = 3). Statistical significance was calculated in Sigma Plot (Version 13) using Student's t-test or one way analysis of variance (ANOVA) followed by Holm-Sidak post hoc test (for multiple groups). p value < 0.05 was considered to be statistically significant.