DENV NS1 and MMP-9 cooperate to induce vascular leakage by altering endothelial cell adhesion and tight junction

Dengue virus (DENV) is a mosquito-borne pathogen that causes a spectrum of diseases including life-threatening dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS). Vascular leakage is a common clinical crisis in DHF/DSS patients and highly associated with increased endothelial permeability. The presence of vascular leakage causes hypotension, circulatory failure, and disseminated intravascular coagulation as the disease progresses of DHF/DSS patients, which can lead to the death of patients. However, the mechanisms by which DENV infection caused the vascular leakage are not fully understood. This study reveals a distinct mechanism by which DENV induces endothelial permeability and vascular leakage in human endothelial cells and mice tissues. We initially show that DENV2 promotes the matrix metalloproteinase-9 (MMP-9) expression and secretion in DHF patients’ sera, peripheral blood mononuclear cells (PBMCs), and macrophages. This study further reveals that DENV non-structural protein 1 (NS1) induces MMP-9 expression through activating the nuclear factor κB (NF-κB) signaling pathway. Additionally, NS1 facilitates the MMP-9 enzymatic activity, which alters the adhesion and tight junction and vascular leakage in human endothelial cells and mouse tissues. Moreover, NS1 recruits MMP-9 to interact with β-catenin and Zona occludens protein-1/2 (ZO-1 and ZO-2) and to degrade the important adhesion and tight junction proteins, thereby inducing endothelial hyperpermeability and vascular leakage in human endothelial cells and mouse tissues. Thus, we reveal that DENV NS1 and MMP-9 cooperatively induce vascular leakage by impairing endothelial cell adhesion and tight junction, and suggest that MMP-9 may serve as a potential target for the treatment of hypovolemia in DSS/DHF patients.


Introduction
Dengue virus (DENV) is the most common mosquito-transmitted viral pathogen in humans. As reported by the World Health Organization (WHO), an estimated 40% of the world population is at risk of DENV infection, and approximately 390 million people worldwide are infected with DENV every year [1][2][3]. As mosquitoes are moving to new areas because of a climate change, the disease is spreading to less tropical and more temperate countries. WHO has named dengue as one of the world's top 10 threats to global health in 2019 [4]. In general, DENV-infected patients are asymptomatic or have flu-like symptoms with fever and rash. However, in severe cases of DENV infection, the disease may progress to dengue hemorrhagic fever (DHF) or dengue shock syndrome (DSS), the leading causes of morbidity and mortality in school-age children in tropical and subtropical regions [1,5]. According to the latest WHO classification, dengue severity is divided into dengue without warning signs, dengue with warning signs, and severe dengue. Vascular leakage, as one of the key features of DHF/DSS and severe dengue, is closely associated with increased vascular permeability in DENVinfected patients [6]. The presence of vascular leakage causes hypotension, circulatory failure, and disseminated intravascular coagulation as the disease progresses, which can lead to the death of DHF/DSS patients. So far, there is no licensed antiviral treatment, but only has supportive therapy, including fluid management, available for patients with vascular hyperpermeability as the mechanism underlying the phenomenon remains unclear.
DENV-induced vascular leakage is characterized by enhanced vascular permeability without morphological damage to the capillary endothelium. Despite the findings on DENV replication in some human endothelial cells (ECs), results from the postmortem analysis of DENVinfected human tissues indicate no generalized DENV ECs infection, which is supported by the fact that patients with severe DENV infection manage to fully recover in a short time [7]. All evidence that the loss of vascular integrity and function in DENV infection in vivo is not caused by extensive damage to the endothelium. Instead, vasoactive endothelial factors released from DENV-infected cells appear to play a major role in this phenomenon. During DENV infection, viruses mainly target at monocytes/macrophages and dendritic cells (DCs) for replication in vivo. Changes in production of interleukin-1 (IL-1), interleukin-6 (IL-6), macrophage inhibitory factor (MIF), tumor necrosis factor α (TNF-α), and metalloproteinases are noted in macrophages and DCs infected with DENV in vitro [8,9]. A study of EC barrier function using in vitro models to describe movement of labeled macromolecules or changes in cell electrical resistance demonstrated that soluble factors released from DENV-infected macrophages could change the permeability of an EC monolayer in the absence of relevant viralinduced cytopathic effect [10]. Taken together, altered production of factors released from circulating monocytes, macrophages, and DCs in human tissues occurs upon DENV infection, and these factors may coordinate to induce functional changes in endothelial cells.
The maintenance of endothelial cell permeability is mainly determined by two factors. One is the polyglycoprotein complex that forms a protective membrane on the cell surface to ensure the integrity of endothelial cells [11,12]. The other is the adhesion and tight junction between endothelial cells, which play important roles in maintaining the integrity of endothelial cells [13]. In addition to promoting cell adhesion, the adhesion and tight junctions can also regulate cell growth, apoptosis, gene expression, and cardiovascular formation by altering intracellular signals [13]. The adhesion and tight junction proteins play key roles in maintaining homeostasis. There are two major types of junction, adhesion junction and tight junction [13]. The changes in adhesion and tight junction structures are regulated by matrix metalloproteinases (MMPs), which are destructive to the integrity of endothelial cells [14]. In the MMP family, the MMP-9 protein promotes tumor migration by degrading extracellular matrix [15,16]. Although previous study reported that DENV induces vascular leakage by up-regulating the expression of MMP-9 in Dendritic cells (DCs) [17], the specific mechanism underlying this regulation is not clear.
Endothelial glycocalyx has been shown to increase endothelial permeability through degradation. Under normal physiological conditions, the glycocalyx as a barrier controls a number of physiological processes, which particularly prevents leukocytes and platelets from adhering to vessel walls [18]. In addition, the degradation of the glycocalyx is closely related to severe vascular leakage in DENV infections. However, it is not fully understood the causes of glycocalyx degradation upon DENV infection. DENV non-structural protein 1 (NS1) is an established early diagnostic marker for DENV infection. The serum concentration of NS1 can reach up to 50 μg/ml in a DHF/DSS case, indicating that NS1 is positively correlated with the disease severity. DENV NS1-induced vascular leakage has been extensively discussed since 2015 [19]. A previous study reported that NS1 induced vascular leakage in mice and anti-NS1 antibodies played a role in reducing NS1-induced vascular leakage and the mortality rate. A study suggested that NS1 induced vascular leakage via Toll-like receptor 4 (TLR4) [20]. Another study reported that autophagy-mediated junction disruption was associated with DENV NS1-induced vascular leakage, which may explain why vascular leakage in DENV-infected patients is a rapid and reversible pathogenic change [21]. NS1-induced MIF secretion is involved in NS1-induced EC autophagy. In addition, an in vitro study showed that DENV-infected cells increased endothelial permeability by inducing MIF secretion. NS1 not only disrupts endothelial junctions but also causes vascular leakage through HPA-1-mediated glycocalyx degradation. In short, there is growing evidence indicating that NS1 plays a critical role in dengue pathogenesis as it causes vascular leakage and hemorrhage during DENV infection [22]. Vascular permeability changes can be induced by destroying the thin (about 500 nm) and gel-like endothelial glycocalyx layer (EGL) that coats the luminal surface of blood vessels [23,24]. Although vascular permeability changes are the main research focus, the specific mechanism underlying DENV pathogenesis needs to be further investigated.
In the present study, we reveal a distinct mechanism by which DENV induces endothelial permeability and vascular leakage in human endothelial cells and mouse tissues. DENV2 infection induces MMP-9 expression and secretion in human peripheral blood mononuclear cells (PBMCs) and macrophages through NS1-induced activation of the NF-κB signaling pathway.
More interestingly, NS1 also interacts with MMP-9, resulting in the degradation of important adhesion and tight junction proteins, impairing the adhesion and tight junctions, and consequently inducing endothelial hyperpermeability and increasing vascular leakage in human endothelial cells and mouse tissues. Collectively, these findings demonstrate that NS1 and MMP-9 cooperate to cause endothelial hyperpermeability and vascular leakage by impairing endothelial cell adhesion and tight junctions.

DENV enhances MMP9 production in severe dengue patients
Previous studies have found that NS1 protein produced during dengue virus infection closely correlated with the onset of disease by promoting vascular leakage. In our study, we found that the concentrations of NS1 protein in the serum samples of severe dengue patients continued increase with the prolongation of the infection time (Fig 1A and 1B). Previous studies have reported that matrix metalloprotein-9 (MMP-9, also known as Gelatinase B, GelB) produced in DENV-infected dendritic cells could induce vascular leakage. We also noticed that the concentrations of MMP-9 protein in the serum samples of severe dengue patients increased over the course of DENV infection (Fig 1C and 1D). Statistical analysis showed a close correlation between NS1 and MMP-9 ( Fig 1E). These results suggest that MMP9 production is enhanced in severe dengue patients and is correlated with NS1 production.

NS1 induces expression and proteolytic activity of MMP-9
The molecular mechanism by which NS1 regulates MMP-9 expression was investigated. HEK293T cells, Hela cells, THP-1 differentiated macrophages, and human umbilical vein endothelial cells (HUVECs) were transfected with plasmid pHA-NS1 at different concentrations. ELISA results indicated that the levels of NS1 protein in the supernatants were increased in pHA-NS1-concentration dependent fashions (S2A- S2D Fig) and similarly, Western-blot results showed that the levels of NS1 protein in the cell lysates were increased in pHA-NS1 concentration-dependent manners (S2E- S2H Fig), demonstrating that NS1 protein is secreted in the supernatants and expressed in the cell of these cell lines. Next, THP-1 differentiated macrophages were transfected with pFlag-NS1 at varying amounts. MMP-9 protein production (Fig 3A, top), MMP-9 enzyme activity (Fig 3A, middle), and MMP-9 mRNA transcription (Fig 3A, bottom) were enhanced by NS1 in dose-dependent manners. However, MMP-9 enzyme activity and MMP-9 protein production were not influenced by DENV-E or DENV-NS3 in THP-1 differentiated macrophages (S2I and S2J Fig). We noticed that unlike NS1 protein, the DENV NS3 and E proteins had no effect on the regulation of MMP-9 enzyme activity and MMP-9 protein production (S2K Fig). MMP-9 protein production and MMP-9 enzyme activity were also facilitated by NS1 in dose-dependent manners in HEK293T cells ( Fig 3B). We also noticed that level of MMP-9 protein was induced by NS1 in THP-1 differentiated macrophages ( We speculated that one of the reasons for this phenomenal is the expression level of MMP-9 in or His-DENV2-NS1 (5 μg) was incubated with purified no-tagged MMP-9 protein (3 μg) for 24 h, Mixtures were incubated with Ni-NTA Agarose beads. Mixtures were analyzed by immunoblotting using anti-MMP9, anti-NS1, anti-His antibody. Untreated protein including His-SARS-CoV-2-N (1 μg), His-DENV2-NS1 (1 μg), or no-tagged MMP-9 protein (1 μg) were analyzed by immunoblotting using anti-MMP9, anti-NS1, and anti-His antibody (as input). (F) Yeast strain AH109 were co-transformed with combination of binding domain (BD-p53, BD-MMP-9, and BD-Lam) and activation domain (AD-T, AD-NS1) plasmid. Transfected yeast cells were grown on SD-minus Trp/Leu double dropout plates, and colonies were replicated on to SD-minus Trp/Leu/Ade/His fourth dropout plates to check for the expression of reporter genes. (G, H) Schematic diagram of wild-type MMP-9 protein and truncated mutants MMP-9 protein (D1 to D9) (G). HEK293T cells were co-transfected with HA-NS1 and Flag-MMP-9 truncated mutants (D1 to D9). Cell lysates were immunoprecipitated using anti-Flag antibody, and analyzed using anti-Flag and anti-HA antibody. Cell lysates (40 μg) was used as Input (H). Dates were representative of three independent experiments.  HUVECs was much lower than that in THP-1 (S2G and S2H Fig). Taken together, the results suggest that DENV NS1 activates MMP-9 production, secretion, and enzyme activity.
MMPs are secreted as inactive proenzymes and are activated subsequently by the cleavage of the propeptide domain by proteolytic enzymes. Results from gelatinase activity assays showed that gelatinase activity was only detected in the presence of both NS1 protein and pro-MMP9 protein, but the enzyme activity was not detected in the presence of BSA protein, NS1 protein, or pro-MMP9 protein alone, or in the presence of BSA and pro-MMP9 (Fig 3H), demonstrating that NS1 regulates the protease activity of MMP9 by interacting with MMP9 to promote the maturation of pro-MMP-9. Collectively, these results suggest that NS1 promotes MMP-9 expression through activating the NF-κB signaling pathway and modulates the proteolytic activity through interacting with MMP-9.
https://doi.org/10.1371/journal.ppat.1008603.g003 DENV2 infection in DENV-infected THP-1 macrophages in a time- (Fig 4B, top) and dosedependent fashions (Fig 4C, top). MMP-9 enzyme activity was also enhanced upon DENV2 infection (Fig 4B and 4C, middle). Viral E mRNA increased in proportion to infection time and inoculum (Fig 4B and 4C, bottom). Moreover, the levels of MMP-9 protein and DENV2 NS3 protein increased over the course of DENV2 infection (Fig 4B, middle) and correlated with virus inoculum (Fig 4C, middle). Interestingly, the levels of MMP-9 mRNA (Fig 4D and  4E, top), MMP-9 enzyme activity, and MMP-9 protein (Fig 4D and 4E, middle) remained unchanged in endothelial HUVECs upon DENV infection (Fig 4D and 4E, bottom). The ability of DENV replication was initially analyzed and compared in macrophages and endothelial cells. DENV2 viral copies was significantly higher in infected THP-1 differentiated macrophages as compared to that of human umbilical vein endothelial cells (HUVECs) ( Fig 4F). Additionally, secreted MMP-9 protein increased in the supernatants of infected THP-1 differentiated macrophages but was undetectable in the supernatants of infected endothelial HUVECs (Fig 4G, top). Similarly, MMP-9 protein was highly expressed in the infected THP-1 differentiated macrophages, while modestly elevated in the infected HUVECs ( Fig 4G, bottom). Taken together, these results indicate that DENV infection induced MMP-9 secretion in THP-1 differentiated macrophages but not in endothelial cells.

NS1 promotes MMP-9-mediated endothelial hyperpermeability in human cells and mouse tissues
The biological effect of NS1 and MMP-9 in the regulation of endothelial cell permeability was evaluated. Firstly, the role of MMP-9 in the induction of endothelial cell permeability was determined. PMA-differentiated THP-1 macrophages were infected with DENV2 for different times, as indicated. The results showed that the secretion of NS1 in supernatants (Fig 5A, top) the production of MMP-9 protein and NS1 protein in cell lysates (Fig 5A, bottom) were upregulated in time-dependent fashions. HUVECs grown on polycarbonate membrane system were incubated with the supernatants of DENV2-infected HUVEC cells or DENV2-infected THP-1 differentiated macrophages or pre-incubated with SB-3CT (a specific inhibitor of MMP-9) or irrelevant chemical inhibitor SC75741 (a specific inhibitor of NF-κB). Endothelial permeability was evaluated by measuring trans-endothelial electrical resistance (TEER) (ohm) using EVOM2 epithelial volt ohm meter. The level of TEER was not affected by the supernatants of DENV-infected HUVECs; significantly attenuated by the supernatants of DENVinfected THP-1 differentiated macrophages from 3 h to 15 h post-treatment; and however, such reduction was recovered by the treatment of SB-3CT but not recovered by the treatment of SC75741 ( Fig 5B); suggesting that MMP-9 plays an important role in the induction of endothelial hyperpermeability mediated by DENV infection. Next, we further determined whether MMP-9 plays an important role in the induction of vascular permeability in mice after DENV infection. IFNAR -/-C57BL/6 mice were treated with PBS as a control group (n = 4), infected with DENV2 (n = 6), and intravenously treated with MMP-9 specific inhibitor SB-3CT and then infected with DENV2(NGC) (n = 6). DENV2 E and NS5 RNA were detected at high levels in the blood of DENV2-infected mice or SB-3CT-treated and DENV2-infected mice at 2 days Intracellular MMP-9 RNA (top) and DENV2 E RNA (bottom) was determined by qRT-PCR analysis, MMP-9 proteinase activity in the supernatants was determined by gelatin zymography assays and proteins in cell extract (middle) were analyzed by Western blotting. (F) HUVEC cells or PMA-differentiated THP-1 macrophages were infected with DENV2 at MOI = 5 for 24 h. Viral copies were quantified by RT-PCR. (G) HUVEC cells or PMA-differentiated THP-1 macrophages were equally distributed to four 12-hole plates and infected with DENV2 at MOI = 5 for 24 h. MMP-9 protein in cell supernatants were measured by ELISA (top) and indicated proteins in cell extract were analyzed by WB (bottom). Dates were representative of three independent experiments. ns means not significant. Values are mean ± SEM, P �0.05 ( � ), P �0.01 ( �� ), P �0.001 ( ��� ).
https://doi.org/10.1371/journal.ppat.1008603.g004 . Cell lysates were analyzed by immunoblotting (bottom). (B) Confluent monolayers of HUVEC cells were grown on polycarbonate membrane system and treated with the supernatants came from DENV2 infected HUVEC cells or THP-1 cells for 24 h or pre-incubated with 600nM SB-3CT (a specific inhibitor of MMP-9 protein) or 600 nM SC75741 for 1h. Endothelial permeability was evaluated by measuring trans-endothelial electrical resistance (TEER) (ohm) using EVOM2 epithelial voltohmmeter. (C-I) IFNAR -/-C57BL/6 mice were intravenously injected with 300 μl DENV2 at a dose of 1×10 6 PFU/mouse (n = 6), pre-treated with 300 μl PBS containing MMP-9 specific inhibitor SB-3CT (5 mg/kg per mice) by intraperitoneal injection for 90 min and then treated with DENV2 (1×10 6 PFU/mouse), repeat treated with SB-3CT (5 mg/kg per mice) on the fourth day after DENV2 (NGC) infection (n = 6), or 300 μl PBS containing the same volume DMSO as a control group (n = 4). 7 days after infection, mice were euthanasia, and the tissues were collected. MMP-9 RNA in the blood was and 4 days post-infection, but not detected in the blood of mocked-infected mice indicating that DENV2 replicated well in the mice (S4A and S4B Fig). We also noted that compared with DENV2-infected mice, the DENV2 E and NS5 RNA has no influence in SB-3CT-treated and DENV2-infected mice, indicating that SB-3CT had no effect on the replication of virus in mice (S4A and S4B Fig). It is worth noting that MMP-9 protein was significantly induced in the blood of DENV-infected mice, but not induced in the blood of mock-infected mice or SB-3CT-treated and DENV-infected both at 2 days, 4 days and 6 days post-treatment (Fig 5C, bottom); MMP9 mRNA was significantly induced in the blood of DENV-infected or SB-3CTtreated and DENV-infected, but not induced in the blood of mock-infected mice at 2 days and 4 days post-treatment (Fig 5C, top). Moreover, the intensities of Evans blue dye in the Liver, Spleen and Lung of DENV-infected mice were significantly higher than that mock-infected mice or SB-3CT-treated and DENV-infected mice tissues (Fig 5D-5F), suggesting that DENV infection induces vascular leakage in mice through promoting MMP-9 production. Meanwhile, Histopathology analysis showed that tissue injury like the spaces between the cells of tissue became larger in the Liver and Lung, the boundary between red pulp and white pulp were disrupted and the lymphatic nodules and pulping cells were increased in spleen were induced in DENV-infected mice organs compared with mock-infected mice tissues, but this phenomenon was rescued in SB-3CT-treated and DENV-infected mice tissues (Fig 5G-5I). Additionally, HUVECs grown on Transwell inserts were incubated with purified MMP-9 protein, NS1 protein, and NS1 protein plus MMP-9 protein at different concentrations, or pre-treated with SB-3CT or irrelevant chemical inhibitor SC75741 and then incubated with NS1 protein plus MMP-9 protein. The level of TEER was reduced by MMP-9 protein alone, NS1 protein alone, and MMP-9 protein plus NS1 protein from 2 to 7 h post-treatment; but the reductions were recovered by SB-3CT and not restored by SC75741 ( Fig 5J); demonstrating that MMP-9 induces endothelial hyperpermeability in human endothelial cells.

NS1 recruits MMP-9 to disrupts the junctions between endothelial cells
Changes in endothelial cell permeability can be achieved by destroying the endothelial glycocalyx layer (EGL) on the surface of endothelial cells or by altering the adhesion and tight junctions between endothelial cells [12,13,28,29]. Previous study reported that DENV NS1 disrupts the EGL, leading to hyperpermeability [23]. However, the roles of NS1 in the regulation of the adhesion and tight junctions between endothelial cells have not been reported. Here, the expression of junction molecular include E-cadherin, ZO-2, α-E-catenin, and β-catenin was detected. The results showed that E-cadherin, β-catenin, and ZO-2 proteins were down-regulated by DENV infection compared with mock-infected mice in the liver (Fig 6A), spleen (Fig 6B), and lung ( Fig 6C), but rescued in SB-3CT-treated and DENV-infected mice (Fig 6A-6C). Similarly, immunohistochemistry analyses also showed that β-catenin was attenuated by DENV infection compared with mock-infected mice in the liver (Fig 6D), spleen ( Fig  6E), and lung ( Fig 6F), but rescued in SB-3CT-treated and DENV-infected mice (Fig 6D-6F). determined by qRT-PCR (upper) and MMP-9 protein in the serum was measured by ELISA (lower). Points represent the value of each serum samples (C). Evans blue dye was intravenously injected into mice 7 days after DENV infected groups (n = 5), control groups (n = 4) and DENV+SB-3CT (n = 5) (C-E). The dye was allowed to circulate for 2 hours before mice were euthanasia, tissues include liver (D), spleen (E) and lung (F) were collected, and the value of Evans blue was measured at OD 610 . Histopathology analysis of tissues includes Liver (G), Spleen (H) and Lung (I) after DENV infection. (J) Monolayers of HUVEC cells grown on Transwell inserts were incubated for 48 h with MMP-9 protein (100 ng/ml) or NS1 protein (5 μg/ml) or NS1 (5 μg/ml) plus different concentration of MMP-9 (50 ng/ml to 100ng/ml) or pre-treated with 600 nM SB-3CT or 600 nM SC7574 for 1 h, then incubated with NS1 plus MMP-9. The TEER (ohm) was measured at indicated time points. Dates were representative of two to three independent experiments. ns means not significant. Values are mean ± SEM, P �0.05 ( � ), P �0.01 ( �� ), P �0.001 ( ��� ).

NS1 recruits MMP-9 to interact with adhesion and tight junction proteins
The mechanism by which NS1 and MMP-9 induce endothelial hyperpermeability and vascular leakage was evaluated. The interactions of NS1 and MMP-9 with adhesion and tight junction proteins were determined. Co-IP results showed that in HEK293T cells and Hela cells, NS1 and β-catenin interacted with each other (Figs 7A, 7B, S6A and S6B), and similarly, NS1 and ZO-1 associated with each other (Figs 7C, 7D, S6C and S6D). We noted that in HEK293T cells or Hela cells, MMP-9 and β-catenin failed to interact with each other (S6E- S6H Fig) and MMP-9 and ZO-1 were also unable to interact with each other (S6I and S6J Fig). His-pulldown assays further demonstrated that purified DENV2-NS1 protein could directly interact with β-catenin and ZO-1, but purified SARS-CoV-2 N protein and MMP-9 protein failed to interact with β-catenin and ZO-1 (Fig 7E and 7F). These results indicate that NS1 protein, but not N protein, can directly interact with β-catenin and ZO-1. Interestingly, in Hela cells in the presence of NS1, the MMP-9 protein and β-catenin protein could interact with each other ( Fig  7G and 7H), and similarly MMP-9 and ZO-1 could also interact with each other (Fig 7I and  7J), indicating that NS1 facilitates the interaction of MMP-9 with β-catenin or ZO-1.
Moreover, the binding sites of NS1 involved in the NS1/MMP-9, NS1/β-catenin, NS1/ZO-1 interaction were determined by progressive truncation of NS1 (NS1-1-NS1-8) (S7A Fig). Finally, HUVECs were incubated with commercialized DENV2 NS1 protein, recombinant human MMP-9 protein, and NS1 protein plus MMP-9 protein, respectively. Immunofluorescence results revealed that when presented individually, NS1 protein (Fig 7K a-e) or MMP-9 protein (Fig 7K f-j) was distributed outside the nucleus; when precented together, extracellular NS1 protein and MMP-9 protein were colocalized with each other (Fig 7K k-p) and distributed in cell membrane (Fig 7K q-v). Notably, NS1 and MMP-9 protein also co-located with β-catenin (Fig 7K ac-ah) and ZO-1 (Fig 7K auaz). We also noticed that compared with untreated cells (Fig 7K w-ab and 7K ai-an), the structures of β-catenin (Fig 7K ac-ah) and ZO-1 (Fig 7K au-ay) were destroyed in the presence of NS1 and MMP-9 in HUVEC cells (Fig 7K and 7L). There results confirm that NS1 recruits MMP-9 to disrupt the junctions between endothelial cells. Therefore, our results reveal that NS1 is probably acting as bridge to promote the interactions of MMP-9 with adhesion and tight junction proteins.

Discussion
DENV infection may cause life-threatening diseases such as DHF, DSS, and ADE [30,31]. The clinical symptoms of DENV infection include hypotension, reduced blood volume, and vascular permeability changes [32]. Therefore, it is important to investigate the mechanism by which DENV infection increases vascular permeability. The present study revealed a distinct mechanism by which DENV induces endothelial permeability and vascular leakage in human endothelial cells and mice tissues.
Our initial results show that DENV2 promotes MMP-9 expression and secretion in human PBMCs and macrophages. These results are consistent with previous reports that MMP-9 protein is highly expressed in immune cells [15], but expressed at a very low level in endothelial cells [22]. More significantly, our findings demonstrate that DENV NS1 enhanced MMP-9 expression through the activation of the NF-κB signaling pathway and modulate the proteolytic activity through interacting with MMP-9. Other groups also proved that ZIKV NS1 protein bound to MMP-9 and facilitated K63-linked polyubiquitination of MMP-9 to stabilize the expression of MMP9, leading to the destruction of the blood-testis barrier [33]. Previous studies revealed that the surface of the endothelial cells (ECs) is coated with a glycocalyx of membrane-bound macromolecules comprised of sulfated proteoglycans, glycoproteins, and plasma proteins that adhere to the surface matrix [12], MMP-9 is destructive to the integrity of endothelial cells and specifically degrades extracellular matrix [14,16], and DENV induces vascular leakage by up-regulating the expression of MMP-9 in Dendritic cells (DCs) [17]. Here, we further demonstrate that NS1 interacts with the Zinc-binding catalytic domain and the Fibronectin type-like domain of MMP-9 and facilitate the enzyme to alter the adhesion and tight junctions, and thereby promoting vascular leakage, in human endothelial cells and mice tissues including liver, spleen, and lung.
The maintenance of endothelial cell permeability is determined by two factors: (1) EGLs form protective membranes on the surfaces of endothelial cells [11,12], and (2) the adhesion and tight junctions between endothelial cells maintain the integrity of endothelial cells [13]. It was reported that NS1 induces vascular endothelial cell permeability leading to vascular leakage by disrupting the extracellular polysaccharide-protein complexes [19,20,24], and NS1 acts as a pathogen-associated molecular pattern (PAMP) by activating the Toll-like receptor 4 (TLR4) signaling pathways to promote the release of inflammatory factors IL-6 and IL-8 and induce vascular leakage [20]. Recent studies have further confirmed that macrophage migration inhibitory factor (MIF) plays a key role in regulating NS1-induced degradation of extracellular polysaccharide proteins [22].
Interestingly, here we demonstrate that NS1 induces endothelial hyperpermeability in human endothelial cells and mice tissues through activating MMP-9. The levels of TEER are reduced by MMP-9 and NS1, but the reductions are recovered by the specific inhibitor of MMP-9 (SB-3CT), demonstrating that MMP-9 induces endothelial hyperpermeability in human endothelial cells. Additionally, the intensities of Evans blue dye are enhanced by NS1 in the tissues of C57BL/6 mice; relatively unaffected by NS1 in the tissues of MMP-9 -/mice; and significantly facilitated by NS1 in the tissues of MMP-9 -/mice treated with MMP-9 protein; revealing that NS1 induces vascular leakage in mice tissues, and demonstrating that MMP-9 is required for NS1-induced hyperpermeability in mice tissues.
More interestingly, the mechanism by which NS1 and MMP-9 induce endothelial hyperpermeability and vascular leakage is further revealed. The adhesion junction proteins N-cadherin and β-catenin as well as the tight junction factors ZO-1, ZO-2, and ZO-3 are not affected by NS1, but ZO-1, ZO-2, N-cadherin, and β-catenin are reduced by MMP-9, and such reductions are eliminated by SB-3CT. Similarly, the endogenous ZO-1, ZO-2, N-cadherin, and βcatenin are reduced by NS1 and MMP-9, and such reductions are recovered by SB-3CT. Additionally, β-catenin and ZO-2 are down-regulated by NS1 in of WT C57BL/6 mice tissues, relatively unaffected by NS1 in MMP9 -/mice tissues, while significantly repressed by NS1 in the tissues of MMP9 -/mice supplemented with MMP-9 protein. Moreover, we further reveal that NS1 can interact with β-catenin and ZO-1 through different domains; MMP-9 fails to interact with β-catenin and ZO-1; however, in the presence of NS1, MMP-9 associates with β-catenin and ZO-1; indicating that NS1 facilitates MMP-9 to interacting with β-catenin or ZO-1; and thereby degrading the adhesion and tight junction proteins. However, the detailed functional relevance of the interaction between NS1 and MMP-9 needs further investigations in the future. Taken together, these results demonstrate that NS1 induces hyperpermeability and vascular leakages in endothelial cells and mice tissues, and reveal that MMP-9 is required for NS1-induced endothelial hyperpermeability and vascular leakage through degrade ting the adhesion and tight junction proteins. In our previous work, we found that DENV M protein induced vascular leakage in mice by activating NLRP3 inflammasome [9,34], these two different molecular mechanisms are induced in DENV infection maybe play an equally important role in vascular leakage.
In summary, we reveal a distinct molecular mechanism by which the viral NS1 protein coordinates with the host factor MMP-9 to induce endothelial hyperpermeability and vascular leakage in human endothelial cells and mice tissues through disrupting the adhesion and tight junctions between endothelial cells. This study would provide new in signs into the pathogenesis caused by DENV infection, and suggest that MMP-9 may acts as a drug target for the prevention and treatment of DENV-associated diseases.

Ethics statement
All human subjects were adult. The study was conducted according to the principles of the Declaration of Helsinki and approved by the Institutional Review Board of the College of Life Sciences, Wuhan University in accordance with its guidelines for the protection of human subjects. The Institutional Review Board of the College of Life Sciences, Wuhan University, approved the collection of blood samples for this study, and it was conducted in accordance with the guidelines for the protection of human subjects. Written informed consent was obtained from each participant.
All animal studies were performed in accordance with the principles described by the Animal Welfare Act and the National Institutes of Health Guidelines for the care and use of laboratory animals in biomedical research. All procedures involving mice and experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of the College of Life Sciences, Wuhan University (Permit numbers: WDSKY0201901).

Clinical sample analysis
In this study, severe dengue patient samples were collected at Eight people's Hospital of Guangzhou during a DENV outbreak in Guangzhou, China, in 2014. Patients were categorized as having severe dengue according to the 2009 WHO criteria for dengue severity. The characteristics of sever dengue patients were listed in S1 Table. All the dengue patient samples were assessed by anti-dengue IgM and IgG enzyme-linked immunosorbent assay (ELSIA) and qRT-PCR to DENV nucleic acid test. Serum samples from 8 patients with severe dengue were collected for ELISA analysis on day 2, day 5, day 8, and day 11 after hospitalization. The expression of NS1 and MMP-9 in individual dengue patient were listed in S2 Table. In addition, 8 serum samples from healthy donors were include as the negative control. Informed consent was obtained from each person.

Animal studies
Wide-type (WT) C57BL/6 mice were purchased from Hubei Research Center of Laboratory Animals (Wuhan, Hubei, China). MMP9 -/mice were purchased from Model Animal Research Center of Nanjing University. IFNAR -/ -C57BL/6 mice were bred in our laboratory. All mice were bred and maintained under specific pathogen-free conditions at Jinan University. For DENV2 infection assays, 6-week-old IFNAR -/-C57BL/6 mice were tail vein injected with PBS (mock infection), pre-treated with 300 μl PBS containing MMP-9 specific inhibitor SB-3CT (5 mg/kg per mice) by intraperitoneal injection for 90 min and then treated with DENV2 (1×10 6 PFU/mouse), repeat treated with SB-3CT (5 mg/kg per mice) on the fourth day after DENV2 (NGC) infection, or 300 μl PBS containing the same volume DMSO as a control group. One week after the DENV2 injection, mice were sacrificed, and tissues were collected for immunohistochemical and histopathological analyses. For other animal assays, sixweek-old and sex-matched of MMP9 -/mice and Wide-type C57 BL/6 mice were randomly chosen to injection with DENV2-NS1 protein or recombinant mouse MMP-9 protein or DENV2-NS1 protein plus MMP-9 protein. The equivalent volumes of PBS-injected mice were used as negative controls. At 24 h post-injection, mice were sacrificed, and tissue were collected for Histopathology analysis.

Cell culture and transfection
Human umbilical vein endothelial cells (HUCEV) were purchased form Obio Technology (Shanghai, China). Human monocytic cell lines (THP-1), human embryonic kidney cell lines (HEK-293T), Hela cells, African green monkey cell lines (Vero) and Aedes albopictus gut cell lines (C6/36) were purchased from the American Type culture Collection (ATCC). THP-1 was grown in RPMI 1640 medium supplemented with 10% fetal calf serum, 100 U/ml penicillin, and 100 μg/ml streptomycin sulphate. HEK-293T, HUVEC and Vero were grown in DMEM medium with 10% fetal calf serum, 100 U/ml penicillin, and 100 μg/ml streptomycin sulphate. C6/36 was grown in MEM medium supplemented with 10% fetal calf serum, 100 U/ml penicillin, and 100 μg/ml streptomycin sulphate. THP-1, HEK-293T, HUVEC and Vero cells were maintained at 37˚C in a 5% CO 2 incubator. C6/36 were maintained at 30˚C in a 5% CO 2 incubator. In order to differentiation of macrophages. THP-1 were stimulated with Phorbol-12-myristate-13-acetate (PMA) for 12 h. Afterwards, cells were incubated for 24 h without PMA. Among those cells, HEK293T and Hela cells are easy to be transfected with plasmids, and the main purpose of whose are to verify the protein expression and the interaction between proteins. THP-1 and HUVEC cells are mainly used to study DENV virus infection and changes in endothelial cell permeability. Vero and C6/36 cells are used to amplification of DENV virus.
Peripheral blood mononuclear cells (PBMCs) were separated by density centrifugation of fresh peripheral venous blood samples that they were diluted 1:1 in pyrogen-free PBS over Histopaque (Haoyang Biotech). Then the cells were washed twice with PBS and resuspended in medium (RPMI 1640) supplemented with 10% FBS, penicillin (100 U/ml), streptomycin (100 μg/ml) in 6 well plates and 12 well plates. Lipo2000 (Invitrogen) were used for cell transfection in HEK293T cells, Hela cells, THP-1 cells, and HUVECs. Take HUVECs transfection as an example. The operation was as follows: when HUVEC cells grew to 80-90% of the volume of the cell culture dish, the media were replaced with serum-free medium, and transfected with the ratio of plasmid concentration (1 μg) and Lipo2000 dosage 1:3. After transfection, HUVEC cells were replaced with complete medium until the predetermined time point.

Virus
All experiments used DENV-2 strain NGC (GenBank accession number KM204118.1) was kindly provided by Dr. Xulin Chen of Wuhan Institute of Virology, Chinese Academy of Sciences. To generate large stocks of dengue virus for experiments. C6/36 cells or Vero cells were incubated with DENV-2 at MOI of 0.5 for 2 h, then unbound dengue virus was washed away. The infected cells were cultured sequentially in fresh medium with 2% FBS until seven days. Supernatants were harvested and centrifuged at 4000 rpm for 10 min to remove cellular debris; then they were filtrated by 0.22 μm filter membrane. All dengue virus was aliquoted into tubes for freezing at -70˚C. Virus tiles were determined by plaque assay using Vero cells.

RNA extract and Real-time PCR
Trizol reagent (Invitrogen, Carlsbad, CA) was used for total cellular RNA extracted according to the manufacturer's instructions. The RNAs (1 μg) were then reverse transcribed to cDNA with 0.5 μl oligo (dT) and 0.5 μl Random primer at 37˚C for 60 min and 72˚C for 10 min. The cDNA then was used as templates for real-time PCR analysis. Real-time PCR was performed in a LightCycler 480 thermal cycler (Roche) by the following procedure: heat activate polymerase at 95˚C for 5 min, afterwards, 45 cycles of 95˚C for 15s, 58˚C for 15s and 72˚C for the 30s, the fluorescence was collected and analyzed at the 72˚C step. A final melting curve step from 50˚C to 95˚C was used to test the specificity of the primer. The primers used in real-time PCR detection were listed S3 Table

Enzyme-linked immunosorbent assay
The concentration of culture supernatants and MMP-9 or NS1 were measured by Human MMP-9 ELISA Kit (Invitrogen) or DENV2-NS1 ELISA Kit (Arigo biolaboratories) according to manufacturer's instructions.

Biomolecular fluorescence complementation (BiFC) assay
The N-terminal truncated (VN-173) and the C-terminal truncated (VC-155) version of nonfluorescent Venus YFP fragments vectors were brought here (Addgene plasmid # 22010). Human MMP-9 gene was subcloned in VC-155, and DENV NS1, E, and NS4A genes were subcloned in VC-155 respectively. The resulting plasmids or empty vectors were co-transfected into HEK293T cells or Hela cells using Lipo2000. At 24 h post-transfection, cells were pre-culture at 4˚C for 10 min. The fusion proteins in living cells were observed by confocal microscopy.

Yeast double hybridization assay
Human MMP-9 gene was subcloned in pGBKT7, and DENV NS1, E, and NS4A were subcloned in pGADT7 respectively. control vectors pGBKT7, pGADT7, pGBKT7-p53, pGADT7-T, and pGBKT7-lam, and some reagents were purchased from Clontech Laboratories. All experimental procedures were done following the Matchmaker Gold Yeast 2-Hybrid System User Manual. Briefly, yeast strain AH109 cells were co-transformed with plasmid pGADT7 and plasmid pGBKT7. Transformed yeast cells were grown in SD-minus Trp/Leu plates (DDO) for 24-48 h, and then subcloned replica plated on SD-minus Trp/Leu/Ade/His plate (QDO) for another 48-96 h.

Co-immunoprecipitation assays
HEK-293T cells or Hela cells were spread to 6-cm-diameter dishes, and co-transfected with the purpose of plasmids for 24 h. Then the cells were lysed in 293T lyses buffer (50 mM Tris-HCl, 300 mM NaCl, 1% Triton X-100, 5 mM EDTA, and 10% glycerol, pH7.4), The lyses buffer was rotating at 4˚C for 30 min and centrifuged at 12000 rpm for 15 min to remove cellular debris. A little part of supernatants was sucked out as Input, and the others were incubated with the indicated antibodies overnight at 4˚C.Then mixed with the Protein G sepharose beads (GE Healthcare) for 2 h at 4˚C.The immunoprecipitates were washed four to six times with the 293T lyses buffer (50 mM Tris-HCl, 300 mM NaCl, 1% Triton X-100, 5 mM EDTA, and 10% glycerol, pH7.4), boiled in protein loading buffer for 10 min. Then analyzed by using SDS-PAGE and Western blotting.

Immunofluorescence
HEK-293T cells were grown on sterile cover slips were transfected with HA-MMP-9 and Flag-NS1 at 40% confluence for 24 h. HUVEC cells or Hela cells were grown on sterile cover slips at 80% confluence, then treated with NS1 protein (5 μg/ml) or MMP-9 protein (100 ng/ml) or NS1 (5 μg/ml) plus MMP-9 (100 ng/ml) or pre-incubated with 600 nM SB-3CT for 1 h, then treated with NS1 (5 μg/ml) plus MMP-9 (100 ng/ml) for 6 h. Cells were fixed with 4% paraformaldehyde for 15 min and then washed three times with wash buffer (ice-cold PBS containing 0.1% BSA), permeabilized with PBS containing 0.2% TritonX-100 for 5 min and washed three times with wash buffer, after blocking with 5% BSA for 30 min, cells were incubated overnight with anti-HA antibody and anti-Flag antibody (1:200 in wash buffer) or anti-β-catenin or anti-ZO-1 antibody (1:100 in wash buffer), followed by staining with FITC-conjugated donkey anti-mouse IgG and Daylight 649-conjugated donkey anti-rabbit IgG or just Cy3-conjugated donkey anti-mouse IgG secondary antibody (Abbkine) (1:100 in wash buffer) for 1 h. Nuclei were stained with DAPI for 5 min, and then the cells were washed three times with wash buffer. Finally, the cells were viewed using a confocal fluorescence microscope (Fluo View FV1000; Olympus, Tokyo, Japan).

Quantization of vascular leakage in vivo
The level of vascular leakage in mice was quantified through Evans blue assays as previously described. Briefly, 300 μl of 0.5% Evans blue dye was injected intravenously to five groups and allowed the dye to circulate for 2 h. Then the mice were euthanized and extensively perfused with PBS. Tissues was collected and weighed. The tubes containing tissue were added to 1 ml formamide and incubated at 37˚C for 24 h. Evans blue concentration was quantified by measuring OD 610 and comparing to the standard curve. Data was expressed as ng Evans blue dye/ mg tissue weight.

Trans-endothelial electrical resistance (TEER)
Human umbilical vein endothelial cells (HUVEC) monolayers (2 x 10 5 ) were grown on the 24-well Transwell polycarbonate membrane system (Transwell permeable support, 0.4 μm, 6.5 mm insert; Corning Inc.) until a monolayer formed (about 24 h), and then treated with different reagents. After 24 h of treatment, 50% of upper and lower chamber media was replaced by fresh endothelial cell medium. Untreated endothelial cells grown on Transwell inserts were used as negative controls and medium alone were used for blank controls. Endothelial permeability was evaluated by measuring TEER in ohms at indicated time points using EVOM2 epithelial voltohmmeter (World Precision Instruments). Relative TEER was expressed as follows: [ohm (experimental groups)-ohm (medium alone)] / [ohm (untreated endothelial cells)ohm (medium alone).