Brain Microvascular Endothelial Cells-Derived HMGB1 Facilitates Monocyte Transendothelial Migration Favoring JEV Neuroinvasion

Song-Song Zou Huazhong Angricultural University Qing-Cui Zou Huazhong Angricultural University Wen-Jing Xiong Huazhong Angricultural University Ning-Yi Cui Huazhong Angricultural University Ke Wang Huazhong Angricultural University Hao-Xuan Liu Huazhong Agricultural University Wen-Juan Lou Huazhong Angricultural University Doaa Higazy Huazhong Angricultural University Hao-Wei Chen Huazhong Angricultural University Ya-Ge Zhang Huazhong Angricultural University Min Cui (  cuimin@mail.hzau.edu.cn ) Huazhong Agriculture University https://orcid.org/0000-0001-9691-033X

. Evidence has shown the contribution of extracellular HMGB1 in the migration of numerous cell types including stem cells, endothelial cells, monocytes and dendritic cells [38,40,41,44,45]. Furthermore, extracellular HMGB1 induces the activation of immune cells and the production of multiple in ammatory factors such as TNF-, IL-6, and IL-1β [37,39,[46][47][48][49], which may contribute to the disruption of the BBB. In vitro study revealed that HMGB1 is associated with BBB breakdown [50], facilitating immune cells transmigration across the BBB into the brain. Correctly, HMGB1 receptor RAGE (receptor for advanced glycation end-products), a high-a nity receptor of HMGB1, plays a restrictive and essential role during the transendothelial migration of immune cells [38]. Therefore, it is essential to determine the role of HMGB1 in immune cells across the BBB into the CNS during JEV infection.
Until now, the pathways of JEV neuroinvasion are not completely clari ed, and this investigation provides evidence of the JEV neuroinvasion pathway. However, there are limited studies on HMGB1 mediating transendothelial migration of monocyte, which expands JEV infection in CNS. JEV-infected monocytes transmit JEV into the CNS suggested that the "Trojan horse" pathway may be employed in JEV invasion which was facilitated by the HMGB1. These data provide insights into the mechanisms of JEV invasion into the CNS and assisted the JE treatment.

Materials And Methods
Mice and virus C57BL/6 mice were supplied by the Laboratory Animal Center of Huazhong Agricultural University, Wuhan, China. All work obeyed the Committee for Protection, supervision, and the Control of Experiments on Animals guidelines of Huazhong Agricultural University. JEV-P3 strain was employed in our previous research [8] and 15 μl (5×10 4 PFU) of the viral inoculum was injected into the brain of 1-day suckling mice. After euthanizing mice, brains were removed. Homogenized brains were suspended in Dulbecco modi ed eagle medium (DMEM) forming of 10% (wt/vol). After centrifugation, the debris was discarded and the supernatant was stored at -80°C until use. Baby hamster kidney broblast cell line (BHK-21) was used for viral titration by plaque assay.

Virus infection
Female C57BL/6 mice aged 6-8 weeks were classi ed into two groups: the control group, which was injected with 50 μl volumes of DMEM. And the JEV infected group, mice were injected in the footpad with a volume of 50 μl that contains 10 5 PFU. Control and JEV-infected mice group was maintained for 10 days with tissue sectioning and protein/RNA extraction.
Mice splenocytes and BEMCs (HBMECs, bEnd.3 cells) were exposed to JEV at MOI of 1 and incubated in DMEM at 37°C with 5% CO 2 for 2 h. Washed by PBS, then grown in a maintenance medium. JEV-free cells were serving as a control.
Cell culture and co-culture HBMECs (Human Brain Microvascular Endothelial Cells) were kept in our laboratory, which was grown in DMEM medium including 10% FBS (fetal bovine serum, Gibco, Grand Island, NY) and endothelial cell growth supplement nonessential amino acids (NEAA), minimum essential medium (MEM) vitamins, sodium pyruvate 100 U/ml penicillin and 100 mg/ml streptomycin sulfate at 37°C with 5% CO 2 . BHK-21 cells and bEnd.3 cells were kept in our laboratory, which was grown in DMEM medium including 10% FBS (fetal bovine serum, Gibco, Grand Island, NY), 100 U/ml penicillin and 100 mg/ml streptomycin sulfate at 37°C with 5% CO 2 . C6/36 cells were obtained from the Institute of Virology, Chinese Academy of Sciences, cultured at 28°C with 5% CO 2 , and the culture medium was consistent with BHK-21 cells.
Primary splenocytes were collected from healthy adult mice. After spinning down the red blood cells, then cultured at a density of 1×10 6 cells/ml in DMEM medium including 10% FBS, 100 U/ml penicillin and 100 mg/ml streptomycin sulfate, splenocytes were treated with recombinant HMGB1 (100 ng/ml) or infected with JEV(MOI=1). Then co-cultured with the JEV-infected BMECs monolayer in the 12-well plate for realtime PCR and immuno uorescence. JEV-free or HMGB1-free cells were serving as a control.

Western Blotting
Cells were lysed in RIPA buffer including the protease inhibitor cocktail, homogenized, and centrifuged at 12, 000 × g and 4°C for 5 min. Protein concentration was monitored by the BCA protein assay kit (Beyotime, China). Protein samples were isolated by performing SDS-PAGE using 12% polyacrylamide gel electrophoresis. Proteins were transferred to polyvinylidene di uoride membranes (Bio-Rad, Richmond, CA, USA). Then blocked for 2 h at room temperature in Tris-buffered saline with Tween 20 (TBST) containing 5% nonfat dry milk, after incubated with JEV-E protein (preserved in the laboratory previously) monoclonal antibody, ICAM-2, beta-catenin, E-Selectin (Proteintech, China), HMGB1 (Novus Biotechnology, NA, USA) and VE-Cadherin, VCAM-1 (Abcam, USA) overnight at 4°C. Washed by TBST, then incubated members with horseradish peroxidase-conjugated (HRP) antibodies. Enhanced chemiluminescence reagents (Bio-Rad, Richmond, CA, USA) was utilized to develop the HRP. All the data are presented as the means ± SEMs (the standard errors of the mean).

Immune cellsbrain injection
Incubated JEV-infected CD3 + T cells, CD19 + B cells and Ly6C + monocytes with the JEV antiserum, and injected into normal mice brain. JE onset time data were registered, and the JE mice brain was removed for virus detection.

Immunohistochemical and Immunofluorescence (IF)
Ketamine-xylazine (0.01 ml/g) and PBS were used for the anesthetization and perfusion of the symptomatic mice. Immediately the collected brains, olfactory bulbs and spinal cords were xed with 4% paraformaldehyde in an aseptic environment. All the xed tissues were embedded by para n for sections. And the antigen was repaired in 0.01 M sodium citrate solution buffer thermally.
Tissues sections were blocked in 5% BSA sealing uid for 30 min at room temperature and incubated with JEV-E protein monoclonal antibody overnight at 4℃. Washed by PBS, incubated with the second antibody for 1 h, and nuclear stained with DAPI for 3 min at room temperature. The tissue sections were sealed with glycerin. The uorescence microscope was used for the observation of the sealing section.

BBBmonolayer transwell model
The BBB monolayer transwell model (Corning, USA) was adopted in this study. 200 μl (50 μg/ml) rat tail collagen (Sigma, USA) was used to enclose an upper chamber at room temperature for 1 h. After washing, 500 μl DMEM (no phenol red, Sigma, USA) was used for the upper chamber pre-equilibration at 37°C for 1 h. BMEC (HBMECs, bEnd.3 cells) were cultured in the upper chamber with a total volume of 500 μl culture medium (no phenol red, Sigma, USA) at 37°C, 5% CO 2 for about 24 h. The leakage examination was continued for 4 h, and 10 kD and 70 kD FITC-dextran (Sigma-Aldrich, USA) and employed for the permeability measurement of the BBB monolayer model. TEER (transendothelial electrical resistance) (Millipore, Billerica, MA) belonging to the evaluation criteria of the monolayer models in vitro, which were monitored and recorded at indicated times (0 h, 6 h, 12 h, 18 h, 24 h).

Immune cells transmigration
Splenocytes were provided from three individual mice at least. 500 μL (5×10 5 ) cells were added into the upper chamber (PC le) of the transwell model (Corning, USA), which were already fully covered with JEVinfected BMEC (HBMEC or bEnd.3 cell) and passed permeability detection. After co-cultured for 24 hours, transmigrated immune cells were collected from the lower chamber, and stained by uorescence antibody labelling for the detection of ow cytometry.

Flow Cytometry and Quantitative real-time PCR analysis
Suspension of splenocytes was stained with the combination of mAbs conjugated with FITC, PE, PE-Cy7, APC-Cy7, PB and APC. For the cell surface marker staining, spleen cells suspensions were incubated with the subjacent Abs: CD3, CD11b, Ly6C, CD19 in PBS buffer (pH=7.4) containing 0.2% BSA (BioSharp, USA) at 4°C for 30 min. PBS was provided for double washing (400×g, 5 min, 4°C), and cell suspension. Cells determination and separation were achieved by FACS Calibur (BD Biosciences, USA) system Beckman cytoFlex (Beckman Coulter, USA) and CytExport 2.0 CellQuest Pro software were for data analyzing. The utilizing of EGFP-JEV has a critical contribution to intracellular JEV detection.
Total RNA was extracted with TRIZOL reagent (Invitrogen, Grand Island, NY, USA). 1 μg RNA served to synthesize cDNA using ReverTra Ace RT-PCR RT kit (Toyobo, Japan), followed the manufacturer's instructions. SYBR green (Invitrogen, Grand Island, NY, USA) was employed to quantitative real-time PCR using the StepOne Plus binding StepOne Software v2.2.2 (Applied Biosystems, Foster City, CA, USA). The relative expression of the JEV-C gene in the spleen or brain was normalized to the level of control β-actin. The pcDNA3.0-HA/JEV-C gene plasmid was served as a template for generating a standard curve to quantify the copy numbers of JEV. Following real-time PCR primers were list in supplementary Table 1.

Statistical analysis.
All experiments currently were repeated for several times. Fig. 9 was made on the website of https://app.biorender.com/. Data were expressed as the means ± SEMs, and the signi cance of differences between groups was followed by Tukey's post hoc tests. Graphs were plotted and analyzed using GraphPad Prism (v7.0; GraphPad, La Jolla, CA, USA).

JEV infection induced the translocation and secretion of HMGB1 in HBMEC
The HBMECs were infected with JEV at 1 MOI, and the expression of JEV-E protein was measured by Western blotting. JEV replicates in HBMEC in a time-dependent manner (Fig. 1A), abundant JEV-E protein presented intracellularly at 24 h and 48 h (Fig. 1B). JEV-infection induced a dramatical increase of HMGB1 both in mRNA (Fig. 1C), and protein level (Fig. 1D). Meanwhile, HMGB1 was also highly expressed in mouse brain microvascular endothelial cell line bEnd.3 during JEV infection (S.1). These results demonstrate that brain microvascular endothelial cells are susceptible to JEV that particularly induces HMGB1 production in brain microvascular endothelial cells.
The biological functions of HMGB1 are dominated by its expression and subcellular location [51]. Thus, the HMGB1 intracellular distribution and release after JEV infection on HBMEC was determined. HMGB1 was predominantly located in the nucleus of uninfected cells at a low level detected by confocal immuno uorescence microscopy ( Fig. 1E). At 6 h, 12 h and 24 h, there was a signi cant increase of HMGB1 at the cytoplasm in HBMEC, which suggested the translocation of HMGB1 from the nucleus to the cytoplasm (Fig. 1E). To further con rm this phenomenon, the protein was extracted from the different cell compartment, the nuclear and the cytoplasmic. And then, the expression of HMGB1 was measured with JEV infection at 1 MOI. The results showed a signi cant increase of HMGB1 in the cytoplasm after JEV infection, and reached a peak at 12 h, and then gradually slowed-down from 24 h to 48 h. The expression of HMGB1 in the nucleus concomitantly increased at 6 h and slipped to the lower level from 12 h to 48 h after JEV infection (Fig. 1F). It further con rmed that JEV-infection stimulated the cellular expression of HMGB1, and HMGB1 translocated from the nucleus to the cytoplasm. The accumulation of HMGB1 in the cytoplasm could actively trigger the autocrine, which is governed by post-translational modi cations. Brefeldin A, an inhibitor of intracellular protein transport, was applied to HBMEC after 12 h postinfection. The intracellular distribution of HMGB1 was observed by the confocal laser scanning microscope. Inhibition of vesicles by Brefeldin A dramatically suppressed HMGB1 release accompanied with an increase of the cytoplasmic HMGB1 at 24 h (Fig. 1G), which suggested that the increased intracellular expression of HMGB1 could be released to the extracellular space.
Taken together, these data suggested that JEV induced HMGB1 upregulation and translocation from the nucleus to the cytoplasm in brain microvascular endothelial cells, and then subsequently released from the cells.

JEV infection induced activation of BMECs and increased adhesion molecules
Brain microvascular endothelial cells are critical in forming BBB and maintaining the barrier function. High expression of adhesion molecules and integrin ligands is necessary for immune cells adhesion to BBB endothelium, which facilitates the immune cell in ltration into the CNS. To identify the underlying mechanism of leukocytes crossing BBB, yeast cells which highly express GFP-LFA-1 (ICAM-1 ligand) were used to detect the interaction with JEV-activated endothelial cells [52]. It was found that more GFP-LFA-1 yeast cells were attached to JEV-infected bEnd.3 monolayer than control cells ( Fig. 2A All these results suggested that JEV infection induced the upregulation of adhesion molecules on BMEC, and HMGB1 also increased the integrin ligands of immune cells during early JEV infection, which could facilitate the interaction between immune cells and the BBB endothelium.

Extracellular HMGB1 promoted immune cells adhesion to endothelium
Leukocyte-endothelium adhesion is indispensable for immune cell CNS in ltration. 293 T cells were used to overexpress HMGB1, and the supernatant was collected to treat the THP-1 cells. The results showed an increase of THP-1 cell adhesion to JEV-infected HBMEC monolayer with the treatment of supernatant containing HMGB1 (Fig. 3A). Trichostatin A (TSA), a deacetylase inhibitor, have been shown to stimulate the hyperacetylation of HMGB1 as well as histones which facilitates the release of HMGB1 from chromatin. Besides, the secreted HMGB1 (sHMGB1) dramatically increased THP-1 cells adhering to the JEV-activated HBMEC monolayer than that of the uninfected group in vitro (Fig. 3A). Moreover, the adhesion was measured between GFP + -leukocytes from JEV-infected mice and bEnd.3 cell preincubated with JEV-P3 virus. More GFP + cells were found adhering on bEnd.3 cells primed with alive JEV-P3 virus, but not found on the cells treated with UV-deactivated virus (Fig. 3B). To further study the function of HMGB1, mouse splenocytes were treated with recombinant HMGB1 (rHMGB1), and co-cultured with bEnd.3 monolayers. It was found that rHMGB1 increased the adhesion of Ly6C + CD11b + monocytes to the JEV-primed bEnd.3 monolayer (Fig. 3C, D). However, there was no signi cant change of the HMGB1treated adhering CD3 + T cells and CD19 + B cells to the JEV-primed endothelial monolayer (Fig. 3C, D).
Furthermore, JEV infection enhanced the puri ed Ly6C + monocyte adhesion to the BMEC monolayer in vitro (Fig. 3E). These results indicated that JEV-activated endothelium promoted the adhesion of immune cells and the presentation of extracellular HMGB1 enhanced monocytes adhering to BMEC monolayer, which may potentiate immune cells crossing BBB entry into the CNS. Extracellular HMGB1 facilitated transendothelial migration of JEV-infected monocytes Data above suggested that the presence of HMGB1 increased the transmigration of THP-1 cells in the JEV-infected monolayer model. To emphasize, the BMEC monolayer was employed to elucidate the role of extracellular HMGB1 in leucocytes migration during JEV infection. We used 10 kD and 70 kD dextrans to visualize the spatially variable, size-dependent permeability of the monolayer (Fig. 4A). Without JEV stimulation, the TEER value, which re ects the integrity of the BBB, were kept stable over 24 hours (>200 Ω×cm 2 ), which implied the integrity of the BBB monolayer in vitro (Fig. 4B). Accompanied with the decrease of TEERs, more Ly6C + CD11b + monocytes, CD3 + T cells, and CD19 + B cells transmigrated from the upper chamber to the lower chamber in the JEV-infected monolayer models (Fig. 4B, C, D). A high concentration of rHMGB1 exacerbated the destruction of the monolayer models during the early infection ( Fig. 4B). Moreover, rHMGB1 increased mice Ly6C + CD11b + monocytes transmigrating from the upper chamber to the lower chamber, while there was no signi cant change of CD3 + T cells, and CD19 + B cells compared to JEV-infected group (Fig. 4B, C, D). These results indicated that JEV infection caused BBB uctuation and increased immune cell CNS in ltration during early infection, which was aggravated by HMGB1. Furthermore, infection-induced HMGB1 enhanced monocyte transendothelial migration, which was based on leukocyte-endothelium adhesion, resulting in BBB breakdown during early infection.

Page 11/39
To discover whether transmigrated immune cells act as virus carriers, JEV with EGFP tag (EGFP-JEV) was applied to the bEnd.3 monolayer to visualize the transmigration of immune cells (Fig. 5A). On the contrary, during JEV infection, the TEER value decreased with time, which means that the integrity of the monolayer was compromised, and the permeability accordingly increased (Fig. 5B). More importantly, the decrease of TEER coincided with an increased number of immune cells transmigrating into the lower chamber. The migrated cells, including Ly6C + CD11b + monocytes, CD3 + T cells and CD19 + B cells, were measured in the absence or presence of the virus. The migratory capability was enhanced in the presence of EGFP-JEV (Fig. 5E). In addition, more EGFP + Ly6C + CD11b + monocytes infected transmigrated from the upper chamber to the lower chamber than CD3 + T cells with EGFP-JEV infection, and only a few transmigrated CD19 + B cells were detected in the lower chamber (Fig. 5F). These results showed that the disruption of the BBB monolayer model increases transmigration of immune cells, especially Ly6C + CD11b + monocytes after JEV infection.
These data suggested that extracellular HMGB1 promotes the leukocytes transendothelial migration, especially for monocyte during early infection. Meanwhile, the infected monocytes, T cells and B cells carried JEV participating in transendothelial migration, which acts as "Trojan horse", may enhance JEV neuroinvasion and aggravate JE.

Transmigration of JEV-infected immune cells correlated with the onset of JE in mice
The correlation between JEV-infected transmigrated splenocytes and the JE incidence in vivo needs further con rmation. Hence, Ly6C + monocytes, CD3 + T cells and CD19 + B cells were puri ed by ow cytometry, which were exposed with JEV at MOI of 1. There was a high replication of JEV in splenocytes at 24 h (Fig. 6A). Western blot demonstrated that there was inclusive JEV inside the isolated splenocytes (Fig. 6B). Then, these cells were incubated with the anti-JEV serum to neutralize the non-speci c adhesive viruses and injected into the brain of healthy adult mice at 1×10 5 cells respectively. Furthermore, the initial onset of JE was much earlier in the intracranial injection of JEV-infected Ly6C + monocyte and CD3 + T cells than the injection of JEV-infected CD19 + B cells (Fig. 6C). No doubt, all the mice delivered JEVinfected immune cells resulted in JE ultimately, which were further supported by the detection of JEV in the brain by immuno uorescence (Fig. 6D, E, F). Moreover, there was a positive association between the onset time of JE and the quantity of transformed JEV-infected immune cells. These data indicated that the transmigration of JEV-infected immune cells from the circulation to the brain were precisely associated with JE onset in the early stage.
The natural infection route was also mimicked by JEV footpad injection in C57BL/6 mice. Tissue samples were collected from the cerebrum, olfactory bulb and spinal cord after infection, and the viral loads were determined by real-time PCR. The results showed that JEV tends to accumulate early in the cerebrum than that of olfactory bulbs and spinal cords (S. 6A, B), suggesting that the hematogenous route, rather than the olfactory nerve or long-range retrograde axonal transport, is more feasible in the JEV neuroinvasion. To further con rm that the hematogenous route is one of those pathways for JEV spreading into the CNS; Immuno uorescence was conducted to determine the JEV distribution in the different neuronal tissues, including the cerebrum, olfactory bulb, and spinal cord. A high count of the JEV virus existed in cerebrum in comparison, to the fewer count entered the CNS via the olfactory bulb or retrograded axonal transportation (S. 6D). These results demonstrated that cerebrum was the primary target of the JEV and the hematogenous route was one of the major pathways in the JEV neuroinvasion.
Together, these data indicated that the hematogenous pathway could be utilized by JEV, while the transmigration of JEV-infected monocytes, T cells and B cells were relevant to the onset of JE in mice.
Discussion HMGB1 (amphoterin) is a DNA-binding, intracellular transcription regulation protein [36]. Cell activation or necrosis induces HMGB1 translocating to the cytoplasm and releasing to the extracellular space [37]. Extracellular HMGB1 has been described as a DAMP mediating the in ammatory responses and regarded as a cell migration mediator [36,40,41,53]. In this study, we demonstrated that BMEC-derived HMGB1 promoted the in ltration of immune cells in vitro. That was performed by the enhancement of leukocyteendothelium adhesion and the increase of transmigration immune cells. Altogether, this evidence indicated that HMGB1, as a critical adhesion and cell transendothelial migration mediator, was involved in the "Trojan horse" pathway in JEV neuroinvasion, contributing to BBB breakdown and JEV pathogenesis.
Previous studies have reported that RSV, H5N1, WNV, DENV, and HIV-1 induced HMGB1 production and release, which played a signi cant mediator role of in ammation during these viral infections [37,54].
Also, HMGB1 is an essential cytokine in cell migration, especially in monocyte transendothelial migration [36,38]. This investigation initially illustrated that JEV infection induced the HMGB1 release from intracellular to extracellular in the HBMEC. The appearance of intracellular vesicles inhibitor caused an increase of cytoplasmic HMGB1. Therefore, although HMGB1 belongs to classical leader peptide lacking protein, the translocation and extracellular release of HMGB1 may be via the nonclassical vesicledependent secretory manner in BEMC [42,55]. Moreover, studies have found that classical protein kinase C phosphorylation and calcium/calmodulin-dependent kinases are also included in HMGB1 secretion [56,57].
Our results showed that JEV infection also indicated a cytoskeletal rearrangement in endothelial cells, and previous studies have shown that JEV infection caused the release of in ammatory factors, which were described as a sign for endothelial cell activation [12]. Furthermore, the activated endothelial cells upregulated the expression of adhesion-related molecules, such as ICAM-1, VCAM-1, and increased the level of LFA-1 and VLA-4 in vivo and in vitro during infection. It was also found that rHMGB1 upregulated the expression of LFA-1 and VLA-4 both in THP-1 cells and in mouse splenocytes in vitro. More evidence presented that the interaction of adhesion molecules and integrin ligands facilitate the transmigration of adhesion immune cells [11,[19][20][21]32], which accounts for the increase of adhesion immune cells to the BMECs monolayer during early infection. These results indicated that BMEC-derived HMGB1 mediated immune cells adhering to the BBB in a coordinated manner with the increase of adhesion-associated molecules, which might enhance leukocyte transendothelial migration across the BBB, promoting JEVinfected immune cells brain invasion.
Beside adhesion molecules and ligands, we also found that HMGB1 played an important role in leukocyte migration across the BBB during early infection. Interactions between HMGB1 and cell surface receptors participate in the migration of various kinds of cells such as dendritic cell, tumor cell, endothelial cell, neutrophil and monocyte [36,38,58]. Most invading pathogens induced the high expression of in ammatory cytokines and chemokines and caused in ammatory responses, leading to degradation of tight junction proteins and disruption of the BBB, uncontrolled transmigration of leukocytes into the CNS, and ultimately causing neuronal damage of the CNS [7,10,12,59,60]. At the same time, virus infection upregulated the levels of chemokines such as CCL5, CCL2, which plays an essential role in the recruitment of immune cells [3]. Nonetheless, we focused on the HMGB1-mediated transendothelial migration of monocytes entry into CNS during infection. Our previous results revealed that JEV itself caused no signi cant damage to the tight junctions of endothelial cells during early infection, which was different to other neurotropic Flaviviridae such as WNV [6,61,62]. On the contrary, the JEV-infected brain supernatant containing MMPs, IL-6, TNF-, CCL-2 and other soluble pro-in ammatory factors, dramatically destroyed the integrity of endothelial cells monolayer in vitro, supporting that the systemic in ammatory response breakdown the BBB during early infection [8,11,63].
Endothelial cells, coupled with pericytes, glia, and neurons constitute a relatively moderate physiological barrier to determine transportation of the CNS and maintain a relatively balanced physiological state, which is considered to form the BBB [6,7,10]. HMGB1 mediates monocyte transendothelial migration, which was already adhered to the endothelium, collaborative with the JEV-activated endothelium, enhancing leukocytes adhesion and in ammatory responses, subsequently leading the uctuation of the BBB, promoting monocyte in ltration. It has shown that during HIV/WNV infection, immune cells (monocytes, T cells) act as the virus carriers were also recruited to the BBB surface for CNS in ltration [7,17,28]. Our results revealed that HMGB1 mediated monocyte transendothelial migration into the CNS during JEV infection. It was demonstrated that JEV induced signi cant uctuation to the endothelial barrier in co-culture models, while not in the monoculture BBB models. Extracellular HMGB1 also activates immune cells to produce in ammatory cytokines [49]. HMGB1 enhances in ammatory response and leukocyte-endothelium adhesion when destroying the BBB, which was re ected by the decrease of TEERs and the increase of migration cells during early infection. Moreover, evidence showed that HMGB1 was directly related to the breakdown of the BBB in vitro [50]. However, neither the BBB monolayer model nor the in vitro transwell model can fully represent the intact BBB in vivo [64]. More comprehensive BBB models in vivo or in vitro are being developed or sought for the investigation of HMGB1-mediated JEV-infected monocytes in the early stage of viral neuroinvasion. The migration of immune cells is probably different in monocytes and T cells in response to HMGB1, which may explicate the distinction in transmigration e ciency in our model. Furthermore, our data suggested that HMGB1 is concerned with the breakdown of the BBB and immune cells in ltration into the CNS during JEV infection.
All the results suggested that BMEC-derived HMGB1 facilitated the endothelium adhesion and transendothelial migration of monocytes during JEV infection. JEV-infected monocyte mingled with a few of other leukocytes acts as the "Trojan horse", transporting virus into the CNS, resulted in JEV neuroinvasion.
To determine whether transmigrated JEV-infected immune cells induce JE, JEV-carrying immune cells were injected into the brain of healthy adult mice. Not surprisingly, JEV-infected monocytes invaded into the CNS, leading to the advanced onset of JE, expanding virus infection in the brain. The presence of monocytes in the brain is a characteristic feather of the CNS in ammation in viral encephalitis, which contributes to reducing viral burden generally [17]. Paradoxically, studies have shown that monocytes were capable of in ltrating into the CNS during viral encephalitis, and of differentiating into a certain type of macrophage, upregulating the production of proin ammatory cytokines, such as NO, MMP and other molecules. Activated immune cells were involved in the in ammatory response, contributing to the BBB breakdown and CNS tissue damage and eventually long-term neurological sequelae or mortality [3,7,8,11,30]. The intracranial injection of JEV-infected T cells initially caused the occurrence of JE. Notably, the appearance of T cells in CNS was possibly regulated by the transmigration of monocytes [30,65,66].
Although evidence revealed that inverse neural pathway is also employed in JEV neuroinvasion [67], neurotropic viruses might spread via multiple pathways to achieve the CNS invasion [19]. Our results suggested that CNS invading immune cells such as JEV-carried monocytes and T cells, caused the occurrence of obvious clinical symptoms of JE early, which was contributed by BMEC-derived HMGB1 during the early stage of JEV infection.
Viruses invasion led to the release of HMGB1, which may play a critical chemotactic role during the early stage. BMEC-derived HMGB1 caused integrins to increase on immune cells, enhancing immune cell adhesion to the BBB and transendothelial migration, contributing to viral spread and in ammation. Studies have reported that ds-HMGB1 triggers in ammatory response may be involved in the development of JE. There is no doubt that other cells or cytokines were involved in JEV neuroinvasion in vivo. We also discovered that JEV tends to accumulate in the cerebral cortex easily during JEV infection in mice. Hence, we discovered that the hematogenous pathway might be involved in the JEV neuroinvasion and the "Trojan horse" pathway. It is possibly expanding the infection of JEV in CNS accelerating the disease, leading to serious encephalitis, whose emergence is inseparable for the monocyte transmigration.
In summary, our results provide evidence that HMGB1 promotes the peripheral monocytes carrying the JEV to sneak into the CNS as "Trojan horse" (Fig. 7). JEV infection induced the release of HMGB1 from brain microcircuit endothelial cells, which stimulates adhesion molecules upregulation on BBBendothelial cells, and corresponding ligands upregulation on monocytes. Thus, the interaction between monocytes and BBB-endothelial cells is enhanced. As a consequence, JEV-infected monocytes transmigrate more frequently from the peripheral to the CNS, leading to neuroinfection and neuroin ammation.

Conclusion
This investigation presents evidence that the JEV-infected immune cells which act as the "Trojan horse" are involved in JEV brain invasion. JEV infection induced the release of HMGB1 from brain microcircuit endothelial cells and, contributed to the in ltration of JEV-infected monocytes into the CNS, accelerating the onset of JE. Therefore, JEV-infected monocytes CNS migration and extracellular HMGB1 can be discussed as the therapeutic target of JE treatment. Expression levels of LFA-1(CD11a and CD18) (E) and VLA-4 (CD49d and CD29) (F) in recombinant HMGB1-treated (100 ng/ml) mice splenocytes, which were performed by real-time PCR at indicated times (0 h, 3 h, 6 h, 12 h, 24 h, 48 h). These data were repeated at least three times. These data were expressed as the means ± SEM. *p < 0.05; **p < 0.01, ***p < 0.001. were normalized to betaactin, which were respectively and quantitatively analyzed as the related fold uctuation to the control.