Human umbilical vein endothelial cells (HUVECs) were obtained from the American Type Culture Collection and cultured in Dulbecco’s modified Eagle’s medium (DMEM; Corning, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% antibiotics (penicillin-streptomycin) in a humidified atmosphere containing 5% CO₂ at 37°C.

The CV-A16-G20 strain (subgenotype B, GenBank: JN590244.1), which was isolated from an HFMD patient in Guangxi, China in 2010, was used for these experiments. Monolayers of HUVECs were plated into 6-well plates overnight and then infected with CV-A16 at a multiplicity of infection (MOI) of 0.1. After 2 h of adsorption, the virus inoculum was removed, and the cells were maintained in DMEM supplemented with 2% FBS at 37°C. The cells and supernatants were harvested at the indicated time points.

Next, to investigate the effect of NF-κB on the production of IP-10 and the effect of IP-10 on BBB integrity, HUVECs were transfected with NF-κB-knockdown or NF-κB-overexpression plasmids and treated with Eldelumab (a monoclonal antibody against IP-10), respectively.

For the quantification of IP-10 mRNA, the following primers were used: (forward: 5’-AGTTAGCAAGGAAAGGTCT-3’, reverse: 5’-ACATTATAGTGCCAGGT-3’) were used in this study. Total RNA was extracted from cells via TRIzol reagent (TIANGEN, China) according to the manufacturer’s protocol. The quality and concentration of RNA were measured by the ratio of absorbance at 260/280 nm via a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA). The RNA was subjected to cDNA synthesis with a PrimeScript™ RT reagent Kit (TAKARA, Japan), followed by qRT-PCR with SYBR Premix Ex Taq (TAKARA, Japan) on an ABI7500 real-time PCR system (Applied Biosystems, USA). The data were normalized to the level of GAPDH expression in each individual sample, and the 2^−ΔΔCt method was used to calculate relative expression changes.

WB analysis was performed as described below. Briefly, total proteins were extracted from cell lines or tissues with cell lysis buffer containing 1 mM PMSF (Beyotime, China) at 4°C for 30 min, followed by centrifugation at 12,000 rpm for 10 min to collect the supernatant. The protein concentrations of the lysates were quantified with a BCA protein assay kit (Beyotime, China). Then, the lysates were resuspended in 2 × loading buffer and boiled at 100°C in boiling water for 10 min. Equal protein samples were electrophoretically separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) at a concentration of 10% and transferred to transferred onto a polyvinylidene difluoride (PVDF) membranes (Millipore, USA). The membranes blocked with 5% skim milk for 2 h at room temperature, reacted with the indicated primary antibodies, including IP-10 (1:1,000 dilution; ABclonal, China; Cat NO. A21986), TLR3 (1:1,000 dilution; ABclonal, China; Cat NO. A24885), TRIF (1:1,500 dilution; Affinity, USA; Cat NO. DF6289), RIG-I (1:1,000 dilution; ABclonal, China; Cat NO. A23179), MDA5 (1:1,000 dilution; ABclonal, China; Cat NO. A2419), MAVS (1:1,500 dilution; Affinity, USA; Cat NO. DF12211), TRAF3 (1:1,500 dilution; Affinity, USA; Cat NO. AF5380), TBK1 (1:1,500 dilution; Affinity, USA; Cat NO.DF7026), NF-κB (1:1,000 dilution; Affinity, USA; Cat NO. AF5006), Claudin5 (1:1,500 dilution; Affinity, USA; Cat NO. AF5216), Occludin (1:1,500 dilution; Affinity, USA; Cat NO. DF7504), ZO-1 (1:1,000 dilution; ABclonal, China; Cat NO. A11417), VE-Cadherin (1:1,500 dilution; Affinity, USA; Cat NO. AF6265), VP1 (1:1,000 dilution; Millipore, USA; Lot Cat 3822821) and GAPDH (1:5,000 dilution; Abbinke, China; Cat NO. ABL1025), overnight at 4°C, followed by the corresponding horseradish peroxidase (HRP)-conjugated secondary antibodies (1:1,0000 dilution; Abbinke, China; Cat NO. A21010 and A21020) for 2 h at room temperature. The membranes were washed with 0.1% Tween 20/PBS three times after each step. The immunoreactive bands were visualized using an enhanced chemiluminescence (ECL) system (Biosharp, China).

ELISA was performed to detect the concentrations of IP-10, IL-1β, IL-6 or TNF-α via the double-antibody sandwich method. Commercial kits, namely the Human IP-10 ELISA Kit (Abcam, USA; Cat NO. ab173194), the Human IL-1β ELISA Kit (NeoBioscience, China; Cat NO. EHC002bQT.96), the Human IL-6 ELISA Kit (NeoBioscience, China; Cat NO. EHC007QT.96) and the Human TNF-α ELISA Kit (NeoBioscience, China; Cat NO. EHC103aQT.96), were used following the manufacturers’ instructions. The concentrations of IP-10, IL-1β, IL-6 or TNF-α were calculated according to their standard curves.

HUVECs plated on coverslips in 24-well plates were infected with CV-A16 for 24 h, washed with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde (Biosharp, China) for 30 min. The cells were then incubated with 0.1% Triton X-100 for 10 min and blocked with 2% bovine serum albumin (BSA; Biofroxx, China) for 1 h at room temperature. Next, the cells were incubated with primary antibodies at a dilution of 1:200 in PBS, including NF-κB, Claudin5, Occludin, ZO-1 and VE-Cadherin, overnight at 4°C, followed by immobilization with Alexa Fluor^® 488-conjugated goat-anti-mouse IgG (1:300 dilution; CST, USA; Cat NO.4408S) and Alexa Fluor^® 594-conjugated goat-anti-rabbit IgG (1:300 dilution; CST, USA; Cat NO.8889S) for 2 h in the dark. The cells were washed three times with PBS after each incubation, and the nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI). Finally, photographs were taken with a laser confocal microscope (Leica, Germany).

Detection of inflammatory cytokines was performed via flow cytometry. Briefly, 25 μl of sample, 25 μl of trapped microspheres and 25 μl of experimental buffer were mixed and incubated for 2 hours with shaking. Then, 25 μl of SA-PE antibody was added, and the mixture was incubated for 30 min. Finally, after cleaning with 100 μl of washing buffer and centrifuging for 5 min at 1,500 rpm/min, 300 μl of washing buffer was added, and the mixture was analyzed on a flow cytometer. The levels of inflammatory cytokines were calculated using a LEGENDplex v8.0 software.

The animal experiments were conducted according to the protocol (number: DWSP202312004) approved by the Animal Ethics Committee of Institute of the Medical Biology, Chinese Academy of Medical Sciences. All procedures performed in studies involving animals were in accordance with the Guide for the Care and Use of Laboratory Animals.

Pregnant BALB/C mice obtained from the Experimental Animal Center of the Institute of Medical Biology, Chinese Academy of Medical Sciences, were randomly divided into the following 10 groups (N=12): Control, CV-A16, IP-10, Anti-IP-10, IP-10+CV-A16, Anti-IP-10+CV-A16, TNF-α, Anti-TNF-α, TNF-α+CV-A16 and Anti-TNF-α+CV-A16. The groups of suckling mice were treated as follows: mice in the IP-10-related treatment groups were intraperitoneally injected with 50 μl of 100 ng/ml IP-10 recombinant protein (ABclonal, China); mice in the Anti-IP-10-related treatment groups were intraperitoneally injected with 50 μl of 100 ng/ml IP-10 antibody (Abmart, China); mice in the TNF-α-related treatment groups were intraperitoneally injected with 50 μl of 50 ng/ml TNF-α recombinant protein (MCE, USA); and mice in the Anti-TNF-α-related treatment groups were intraperitoneally injected with 50 μl of 5 mg/kg TNF-α recombinant protein (MCE, USA).

Suckling mice were clinically examined daily, including for survival and clinical symptoms. At Day 4 and Day 8, blood samples were collected, and brain tissues were removed immediately after all the animals were sacrificed painlessly. Blood samples were used for the examination of IP-10, IL-1β, IL-6 or TNF-α via ELISA. The brain tissues were stored at -80°C prior to testing the viral dissemination of CV-A16 via qRT-PCR and inflammatory cytokines by flow cytometry. The brain tissues were also collected, fixed with 10% formalin, and then embedded in paraffin before cutting 5 μm coronal sections. The sections were then used for hematoxylin and eosin (H&E) and immunohistochemical (IHC) staining.

Pathological changes in brain tissues were observed by H&E staining. Deparaffinized slides were stained with hematoxylin for 5 min and eosin for 2 min at 25°C. Then, the stained slides were sealed with a neutral resin and images were acquired using a digital pathology slide scanner (KFBIO, China).

qRT-PCR was used for the examination of the viral load and was conducted as previously described. In brief, the RNA of mouse brains was extracted with TRIzol reagent (TIANGEN, China) according to the manufacturer’s guidelines. The qRT-PCR process was subsequently performed using a One Step PrimeScript™ RT-PCR kit (TAKARA, Japan) on a GENTIER 96 instrument (TIANLONG, China). Moreover, the standards for CV-A16 at known concentrations were measured concurrently with the test samples. The viral load of each sample was subsequently calculated from the standard curve generated from the CV-A16 standards.

Paraffin-embedded brain sections (5 μm thick) were deparaffinized and rehydrated. Subsequently, the sections were subjected to antigen retrieval with citrate antigen retrieval solution (Servicebio, China) by microwaving for 10 min and then blocked with 5% BSA in PBS. CV-A16-VP1 protein monoclonal antibody (1:100 dilution) was applied overnight at 4°C. Following rinses three times with PBS, the sections were further incubated with HRP-conjugated secondary antibody (1:800 dilution) at room temperature for 1 h. Finally, diaminobenzidine (DAB; Servicebio, China) was used as a substrate for color development, and the sections were observed with a digital pathology slide scanner (KFBIO, China).

All results are representative of 3 independent experiments, and the data are presented as the means ± standard deviations (SDs) and were analyzed via GraphPad Prism 7.0 software. The statistical significance of differences between two groups was evaluated using Student’s t test, but three groups were analyzed using a one-way ANOVA with Dunnett’s test, and multiple groups were analyzed using a two-way ANOVA with Tukey’s multiple-comparisons test. Differences were considered statistically significant when P < 0.05.

IP-10 is often up-regulated in the early stage of many viral infections, making IP-10 useful as a marker of early viral infection (18). However, how IP-10 changes during CV-A16 infection is not known. As shown in Figures 1A, B , qRT-PCR and WB data revealed that CV-A16 infection of HUVECs caused a significant, time-dependent increases in the mRNA and protein expression of IP-10, respectively. Moreover, the ELISA results also revealed that the cumulative protein level of IP-10 protein in culture supernatants from CV-A16-infected cells clearly increased over time ( Figure 1C ). Thus, these results suggest that CV-A16 infection rapidly promotes IP-10 expression.

Previous studies have been reported that the IP-10 expression involves the activation of many pathways, mainly including the p38-, JNK-, ERK-, Akt- and NF-κB-mediated pathways (18). Since IP-10 is a proinflammatory factor, we focused on the NF-κB signaling pathway. Moreover, because CV-A16 is an RNA virus, we investigated 2 pattern recognition receptor pathways associated with NF-κB, namely, TLR3-TRIF-TRAF3-TBK1-NF-κB and RIG-I/MDA5-MAVS-TRAFS-TBK1-NF-κB. Therefore, we first investigated the relationship between NF-κB and IP-10. To demonstrate that the activation of NF-κB can promote IP-10 expression, we transfected NF-κB-knockdown and NF-κB-overexpression vectors into cells, and the WB results revealed that plasmid transfection of NF-κB was effective ( Figure 1D ). Then, it was found that NF-κB-knockdown decelerated IP-10 expression, and NF-κB-overexpression accelerated IP-10 expression during CV-A16 infection ( Figures 1E, F ). These data indicate that NF-κB can regulate the expression of IP-10.

Next, IF staining revealed that during CV-A16 infection, NF-κB was markedly increased in the nucleus ( Figure 1G ). Moreover, PPR-related molecules involved upstream of NF-κB were markedly elevated during CV-A16 infection with increasing infection time ( Figure 1H ). These findings suggest that CV-A16 infection may improve the NF-κB signaling pathway by activating innate immunity.

NF-κB is a critical regulator of the immediate early pathogen response and plays an essential role in inflammatory responses (26, 27). The above results indicated that NF-κB was activated, so we further explored whether the inflammatory response under the downstream of NF-κB was also activated. It was observed that IL-1β and IL-6 were the first to be upregulated at 6 hours after CV-A16 infection, then IL-8 was upregulated beginning 12 hours after CV-A16 infection, and finally, TNF-α and IFN-α were also upregulated during the later stage of CV-A16 infection ( Figure 2 ; Supplementary Table 1 ). Therefore, these results suggest that CV-A16 infection triggers an inflammatory response.

Accumulating evidence has confirmed that proinflammatory cytokines play crucial roles in BBB dysfunction (28). However, junction complexes between cerebral endothelial cells are essential for maintaining BBB integrity and mainly contain TJ proteins (e.g., Claudin5, Occludin and ZO-1) and AJ proteins (e.g., VE-Cadherin) (29). WB results revealed that CV-A16 infection disrupted the expression of TJ and AJ proteins, but this phenomenon was reversed after the administration of a monoclonal targeting antibody with IP-10 (e.g., Eldelumab) ( Figure 3A ). Subsequently, IF staining further revealed that CV-A16 infection resulted in the downregulation of these proteins, but Eldelumab treatment sharply increased the expression of these proteins in CV-A16-infected cells ( Figure 3B ). Since HUVECs are often used to construct BBB models in vitro, combined with our results above, we hypothesized that IP-10 induced by CV-A16 infection may contribute to BBB damage.

As a chemokine, IP-10 plays an important role in neuroinflammation, primarily by attracting inflammatory cells to cross the BBB, after which it infiltrates the sites in the brain and ultimately promotes the production of chemokines and cytokines, and enhances BBB permeability (24, 30). Therefore, IP-10 is often not the direct cause of BBB damage, but the above results indicate that the inhibition of IP-10 prevents CV-A16-induced BBB damage. Thus, we hypothesized that IP-10 may cause damage to the BBB through its downstream factors. In this study, it was seen that after treatment with Eldelumab, the release of TNF-α from CV-A16-infected cells decreased, but other proinflammatory cytokines, such as IL-6, IL-8, IL-1β and IFN-α, which were enhanced during CV-A16 infection, were not affected by Eldelumab treatment in CV-A16-infected cells ( Figure 4A ). Moreover, it was also discovered that the higher the dose of Eldelumab treatment, the lower the level of TNF-α was released by CV-A16-infected cells, and vice versa ( Supplementary Table 2 ). We further tested the concentrations of IP-10 and TNF-α in CV-A16-infected cells treated with different doses of Eldelumab. Our results revealed that the protein release of IP-10 and TNF-α after CV-A16 infection was relatively low when given a high dose of Eldelumab was used, but the protein release of IP-10 and TNF-α after CV-A16 infection was relatively high when given a low dose of Eldelumab was used ( Figure 4B ), which was in line with data of the above flow cytometry data described above. On the basis of these results, we speculated that Eldelumab, as an antibody against IP-10, has a neutralizing effect on the IP-10 protein; thus, in the case of a high dose of Eldelumab, the IP-10 protein released after CV-A16 infection is neutralized by Eldelumab, and TNF-α changes with the amount of IP-10, which further proves that TNF-α may be a downstream molecule of IP-10. To further verify whether TNF-α affects the BBB, we examined the expression of junctional complexes. Our results revealed that, compared with CV-A16 infection and IP-10 treatment, TNF-α treatment markedly reduced the expression of Claudin5, Occludin, ZO-1 and VE-Cadherin. Previous studies have reported that TNF-α, as a cytokine, plays an important role in neuroinflammation mainly by inducing alterations in the organization of junctional complexes that result in BBB dysfunction (31, 32). Thus, our data illustrated that TNF-α may cause BBB damage by disrupting the expression of junctional complexes. Moreover, combined the above results suggest that IP-10 triggered by CV-A16 infection may disrupt the BBB through damage to the BBB junctional complexes caused by its downstream regulation of TNF-α.

We next used animal experiments to investigate the associations between IP-10 and TNF-α and their role in the pathological injury of the central nervous system caused by CV-A16 infection, we conducted animal experiments. The TNF-α+CV-A16 group presented the most severe clinical symptoms, the lowest survival rate and the most severe pathological changes in brain tissue ( Figure 5 ), but Anti-TNF-α treatment before CV-A16 infection resulted in not only a significant reduction in the clinical symptoms and the degree of pathological injury, but also a significant reduction in the mortality of suckling mice. Although the trends of these indicators in the IP-10+CV-A16 and Anti-IP-10+CV-A16 groups were similar to those in the two groups above, the degree of change in these indicators was significantly lower than that in the two groups above. Therefore, these data suggest that both IP-10 and TNF-α could aggravate the damage caused by CV-A16 infection, and that the effect of TNF-α is more severe. Moreover, compared with IP-10 treatment alone, TNF-α treatment alone resulted in relatively severe clinical symptoms, a relatively lower survival rate and relatively severe pathological changes in brain tissue than IP-10 treatment alone. This result may further confirm our above results shown in Figure 4 . As it is downstream of IP-10, TNF-α can directly disrupt BBB permeability by destroying intercellular connections; thus, the reason that the relatively more severe effect of TNF-α alone than of IP-10 alone may be because TNF-α acts directly on the BBB, while IP-10 acts indirectly via TNF-α.

Next, we measured the concentrations of IP-10 and TNF-α in brain tissues and blood on Day 4 and Day 8 after different treatments. As shown in Figure 6A , IP-10 expression was upregulated in the brain tissue and blood of all the groups infected with CV-A16 compared with the control group on both Day 4 and Day 8. Moreover, IP-10 production in the TNF-α+CV-A16 and Anti-TNF-α+CV-A16 groups was not significantly different from that in the CV-A16 group. In addition, there were no differences between the TNF-α/Anti-TNF-α group and the Control group. These results indicate that CV-A16 infection may promote IP-10 production from the beginning of infection and that TNF-α is not associated with IP-10 production. However, TNF-α expression in the brain tissue and blood in the CV-A16 group was not altered on Day 4, but did increase on Day 8. Additionally, TNF-α expression was gradually upregulated in the IP-10, IP-10+CV-A16, TNF-α and TNF-α+CV-A16 groups. These data indicate that CV-A16 infection may increase TNF-α release at later stages of infection and IP-10 induces the release of TNF-α.

Next, the junctional complexes were analyzed. IP-10 treatment had no significant effects on the expression of Claudin5, Occludin, ZO-1 or VE-Cadherin molecules, but TNF-α treatment directly reduced the expression of these molecules ( Figure 6B ). Furthermore, these junctional complexes in the IP-10+CV-A16 and TNF-α+CV-A16 groups were markedly lower than those in the CV-A16 group, but they were markedly greater in the Anti-IP-10+CV-A16 and Anti-TNF-α+CV-A16 groups. Therefore, combined with the above results, we hypothesized that IP-10 induces TNF-α expression, which may further affect BBB integrity.

Finally, the inflammatory response and CV-A16 loads in the brain tissues were assessed. Flow cytometry revealed that the highest expression of inflammatory cytokines was in the TNF-α+CV-A16 group, followed by the IP-10+CV-A16 group ( Figure 7A ; Supplementary Table 3 ). However, treatment with Anti-TNF-α or Anti-IP-10 dramatically reduced the expression of some inflammatory cytokines, especially TNF-α, in CV-A16-infected cells. In addition, IP-10 treatment alone indeed increased TNF-α expression. These findings suggest that CV-A16 infection causes neuroinflammation, which may be activated by IP-10-induced TNF-α during CV-A16 infection and could be further aggravated via the regulation of TNF-α. On the basis of the results of BBB damage caused by CV-A16 infection, it is possible that CV-A16 is more likely to enter the central nervous system through the damaged BBB. In all groups subjected to CV-A16 infection, IP-10 or TNF-α treatment prominently increased the virus load and VP1 expression in CV-A16-infected cells, whereas Anti-IP-10 or TNF-α treatment noticeably decreased the virus load and VP1 expression in CV-A16-infected cells ( Figure 7B ). Moreover, the viral load and VP1 expression significantly greater in the TNF-α+CV-A16 group than those in the IP-10+CV-A16 group. Therefore, these data suggest that TNF-α and IP-10 may rapidly increase the transmission of CV-A16 in the brain, but the promotion of CV-A16 transmission resulting from IP-10 may be due to its regulation of TNF-α.