Dengue Virus NS1 Uses Scavenger Receptor B1 as a Cell Receptor in Cultured Cells

ABSTRACT

The dengue virus NS1 is a multifunctional protein that forms part of replication complexes. NS1 is also secreted, as a hexamer, to the extracellular milieu. Circulating NS1 has been associated with dengue pathogenesis by several mechanisms. Cell binding and internalization of soluble NS1 result in endothelial hyperpermeability and in the downregulation of the innate immune response. In this work, we report that the HDL scavenger receptor B1 (SRB1) in human hepatic cells and a scavenger receptor B1-like in mosquito C6/36 cells act as cell surface binding receptors for dengue virus NS1. The presence of the SRB1 on the plasma membrane of C6/36 cells, as well as in Huh7 cells, was demonstrated by confocal microscopy. The internalization of NS1 can be efficiently blocked by anti-SRB1 antibodies, and previous incubation of the cells with HDL significantly reduces NS1 internalization. Significant reduction in NS1 internalization was observed in C6/36 cells transfected with siRNAs specific for SRB1. In addition, the transient expression of SRB1 in Vero cells, which lacks the receptor, allows NS1 internalization in these cells. Direct interaction between soluble NS1 and the SRB1 in Huh7 and C6/36 cells was demonstrated in situ by proximity ligation assays and in vitro by surface plasmon resonance. Finally, results are presented indicating that the SRB1 also acts as a cell receptor for Zika virus NS1. These results demonstrate that dengue virus NS1, a bona fide lipoprotein, usurps the HDL receptor for cell entry and offers explanations for the altered serum lipoprotein homeostasis observed in dengue patients.

IMPORTANCE Dengue is the most common viral disease transmitted to humans by mosquitoes. The dengue virus NS1 is a multifunctional glycoprotein necessary for viral replication. NS1 is also secreted as a hexameric lipoprotein and circulates in high concentrations in the sera of patients. Circulating NS1 has been associated with dengue pathogenesis by several mechanisms, including favoring of virus replication in hepatocytes and dendritic cells and disruption of the endothelial glycocalyx leading to hyperpermeability. Those last actions require NS1 internalization. Here, we identify the scavenger cell receptor B1, as the cell-binding receptor for dengue and Zika virus NS1, in cultured liver and in mosquito cells. The results indicate that flavivirus NS1, a bona fide lipoprotein, usurps the human HDL receptor and may offer explanations for the alterations in serum lipoprotein homeostasis observed in dengue patients.

INTRODUCTION

The Flavivirus genus, belonging to the Flaviviridae family, includes a group of vector-borne viruses of importance in human public health: dengue virus (DENV), Zika virus (ZIKV), yellow fever virus (YFV), Japanese encephalitis virus (JEV), and West Nile virus (WNV). DENV and ZIKV are transmitted by the same arthropods of the Aedes genus, and both viruses share a similar geographical distribution (1). It is estimated that almost half of the world population, distributed in tropical and subtropical zones, is at risk of contracting dengue or Zika diseases (2). Both dengue and Zika usually manifest as a mild febrile syndrome. However, infections with DENV can evolve to hemorrhagic syndrome accompanied by hypovolemic shock, plasma leakage, and death. In turn, ZIKV infections, when acquired during pregnancy, are associated with the occurrence of microcephaly and other neurological severe injuries in the fetus's brain (3). In adults, ZIKV has been associated with the occurrence of Guillain-Barré syndrome, which is highly disabling (4). DENV and ZIKV virions are spherical particles of about 50 nm in diameter. The positive RNA genome is about 11Kb in length, encoding for 3 structural (Membrane, Capsid, and Envelope) and for 7 nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5). The gene encoding for the NS1 is 1,056 nucleotides in length, rendering a 48–54 kDa protein depending on its glycosylation status. The NS1 can be found in monomeric, dimeric, or hexameric forms (5). Structurally, the monomeric form of NS1 is composed of three domains: α/β wing (RIG-I-like folded, from amino acids 38–151), β-roll (hydrophobic, from amino acids 1–29), and β-ladder (composed of a β-sheet and a spaghetti loop, from amino acids 181–352) (6). Once the NS1 is translated from the viral genome, it is located at the lumen of the ER in a monomeric form that rapidly dimerizes (5). The intracellular NS1 is mainly dimeric and is associated with intra and extracellular membranes participating in different processes interacting with cellular and viral proteins during the viral replication, virion assembly, signaling transduction, and evasion of the cellular immune response (7, 8). The hexameric form of NS1 is an association of three dimers assembled in a hollow barrel-shaped structure, with a central channel rich in lipids acquired from cellular membranes (9). NS1 is the only flavivirus nonstructural protein secreted together with the viral particles from infected vertebrate and mosquito cells (10, 11). In fact, NS1 is used as an early marker for DENV and ZIKV detection from patient serum samples (12). The DENV soluble circulating NS1 protein has been involved in several pathogenic mechanisms during the infection process. It has been reported that NS1 can affect the complement pathway by promoting the degradation of C4 protein and can alter the coagulation system by inhibiting the activation of prothrombin (13). NS1 can also disrupt endothelium integrity, promoting an increase in the vascular permeability after internalization by endocytosis, which may be related to plasma leakage (14–16). Moreover, pre-exposure of human liver or dendritic cells to NS1 renders these cells more susceptible to DENV replication (17, 18). Thus, a better understanding of the DENV soluble NS1-cell interactions may result in strategies to combat DENV replication and pathogenesis.

The scavenger receptor B type 1 (SRB1) or its human homolog CLA-1 is a 509 amino acid transmembrane glycoprotein containing an extracellular domain as a unique loop of 408 amino acids with multiple sites of N-glycosylation and a cysteine-rich region, two transmembrane domains of 22 and 23 amino acids and two cytoplasmatic N and C terminal domains of 9 and 44 amino acids, respectively (19). SRB1 is the physiologically relevant cell-surface high density lipoprotein (HDL) receptor responsible for selective HDL-cholesteryl esters (HDL-CE) uptake mainly in the liver, where it is most abundantly expressed, providing cholesterol for bile acid synthesis, and controlling plasma HDL levels (20, 21). In insects, such as Drosophila melanogaster and Aedes aegypti, among others, homolog genes for SRB1 have been identified; however, their functions in fatty acid transport have not been studied in these organisms (22).

The SRB1 has been described as the principal receptor of the hepatitis C virus (HCV) in hepatocytes, through the recognition of the E2 hypervariable region 1 of the HCV envelope lipoprotein (23). Recently, it was reported that the SRB1 facilitates ACE2 dependent entry of SARS-CoV-2 in various cell types by augmenting virion attachment (24). In addition, it has been reported that apolipoprotein AI (ApoA-I), the principal carrier of HDL molecules, bridges DENV particles and cell receptor SRB1 and facilitates the entry of DENV virions into various cell types, enhancing the infection (25). DENV shows tropism for liver tissue, where it replicates efficiently in hepatocytes, and extensive damage has been observed in the liver of patients with fatal severe dengue (26, 27). DENV NS1 can enter liver tissue cells, both in vitro and in vivo, in the absence of viral particles. In addition, an increase in virus yield is observed from hepatic cells preincubated with DENV NS1, and it has been postulated that DENV NS1 conditions the hepatic cells to increase viral particle production (17). Recently, it was reported that endocytosis of NS1 is required for NS1-mediated endothelial hyperpermeability (16). However, the host factors favoring NS1 binding and entry in hepatocytes or other cells are unknown.

Since NS1 must be internalized to exert its functions and the DENV NS1 is a bona fide glyco-lipoprotein, with a lipid cargo composed of triglycerides, cholesteryl esters, and phospholipids, that resembles in composition the plasma lipoprotein HDL (9), this suggested to us that DENV NS1 could mimic or hijack some of the lipid metabolic pathways to enter the cell. In this work, we present evidence demonstrating that the HDL SRB1 in human hepatic cells, and an SRB1-like molecule in mosquito cells, is also used by DENV and ZIKV soluble NS1 as cell receptors for binding and internalization.

RESULTS

A scavenger-like receptor is expressed in insect cells plasma membrane.

The expression of SRB1 in liver cells has been widely reported (21). To evaluate the expression of SRB1 in C6/36 cells, nonpermeabilized C6/36 monolayers were immunolabeled with an antibody specific for human SRB1 and analyzed by confocal microscopy, using Huh7 and Vero cells as positive and negative controls, respectively. The results indicated that the SRB1 is expressed in the cell membrane of Huh7 and C6/36 cells, but not in kidney derived Vero cells (Fig. 1A). To further corroborate that C6/36 cells indeed express the SRB1 on the cell surface, the location of the SRB1 on C6/36 was determined by TIRF microscopy, using Huh7 as positive controls. As shown in Fig. 1B, a clear signal for the SRB1 was obtained in both cell lines, indicating that C6/36 cells express an SRB1 molecule on the plasma membrane.

FIG 1.

Detection of the scavenger receptor B1 (SRB1) in liver, mosquito, and kidney derived cell lines. (A) Confluent cell monolayers were fixed and stained for SRB1 (red) and nuclei counterstained with DAPI (blue). Bars = 10 μm. (B) The presence of the SRB1 on the cell plasma membrane was corroborated by total internal reflection fluorescence (TIRF) microscopy, containing cells for SRB1 (red) and wheat germ agglutinin (green) to label the cell membrane.

DENV and ZIKV NS1 proteins are efficiently internalized in hepatic and insect cells.

It is well established that DENV NS1 is internalized in human hepatic and dendritic cells in culture, and also by the liver of mice inoculated with purified NS1 (17,18), yet internalization of DENV NS1 in mosquito cells or of ZIKV NS1 in any cell type have not been reported. Thus, using Huh7 liver cells as positive controls, we evaluated if DENV and ZIKV NS1 were also internalized by C6/36 mosquito cells after 1 or 2 h of incubation. As shown in Fig. 2, permeabilized cells were readily stained with anti-NS1 antibodies indicating that DENV and ZIKV NS1 were efficiently internalized by mosquito cell lines. However, different fluorescent patterns were observed for the internalized DENV NS1 in liver and mosquito cells. While a punctate pattern was observed in liver cells, a diffuse cytoplasm distribution for NS1 was observed in the mosquito cells. Likewise, different fluorescent patterns were observed between internalized DENV and ZIKV NS1 in liver cells, with ZIKV NS1 showing a fiber-like pattern. These observations, in agreement with previous data (15), suggest that there may be cell and virus dependent differences in the way NS1 is internalized or located in the cytoplasm after internalization. Once the entry of NS1 into C6/36 and Huh7 cells was assessed, we evaluated whether preincubation of mosquito cells with DENV or ZIKV NS1 will render the cells more susceptible to virus replication, just as has been observed for human liver and dendritic cells exposed to DENV NS1. For this, C6/36 confluent monolayers were incubated with 0, 3.5, or 7 μg/well of DENV or ZIKV NS1 for 1.5 h. Afterward, cells were infected with DENV or ZIKV at MOI = 1 and at 48 hpi, and the supernatants were collected to determine the viral titers. A significant increase in the DENV and ZIKV yield, of approximately half a log, was observed from cells preincubated with 3.5 μg/well in comparison with those not preincubated (Fig. 2F and G). Finally, no internalization or increase in virus titer was observed when heat denatured NS1 was used in these experiments as a control (Fig. 2). This result suggests that exposure and internalization of NS1 by C6/36 cells favors DENV and ZIKV replication, in agreement with previous results reported for liver and dendritic cells with DENV (17, 18).

FIG 2.

Internalization of DENV and ZIKV rNS1 in Huh7 (A), C6/36 (B), and Vero cells (C). Cells were preincubated with recombinant NS1 (3.5 μg/well), and at the indicated times cells were washed, fixed, permeabilized, and stained for intracellular NS1 (red). Cell nuclei were counterstained with DAPI. Panels A.1 and B.1 correspond to cells incubated with heat denatured NS1. Bars = 10 μm. (D, E). Quantification of the NS1 signal. Results are the mean ± SD of at least 100 different cells per condition. (F) DENV and (G) ZIKV titers after C6/36 cells preincubation or not with recombinant NS1 (light gray bars) or heat denatured NS1 (dark gray bars). Monolayers of C6/36 cells were preincubated for 1.5 h with the indicated concentrations of either DENV or ZIKV rNS1. After washing, the cells were infected with the corresponding virus at MOI = 3 for 1 h, the infection allowed to proceed, and the supernatants, collected at 24 hpi, titrated by plaque assay in Vero cells. PFU, plaque forming units. Results are the mean ± SD of at least 2 independent experiments. *, P = 0.05.

SRB1 participates in the entry of NS1 in hepatic and insect cells.

To evaluate the participation of the SRB1 in the entry of NS1, monolayers of C6/36 or Huh7 cells were treated with 0, 1, and 10 μg/mL of anti-SRB1 monoclonal antibody to block the SRB1. After removal of excess antibody, cells were incubated with DENV NS1. Cells were fixed, permeabilized, and immunolabeled to detect internalized DENV NS1 as described above. Fluorescent signals were quantified to compare the internalization of the NS1 in the presence or absence of the antibody. A significant decrease of approximately 30%, in the intracellular specific signal for NS1 was observed in C6/36 cells preincubated with the specific monoclonal antibody compared to controls, not preincubated cells. No significant difference in the signal was observed when the cells were preincubated with 1 or 10 μg/mL of the antibody (Fig. 3A and B). Also, a significant reduction in the internalization of DENV NS1 was observed in Huh7 cells treated with the anti-SRB1 antibody (Fig. 3C and D). The specific intracellular fluorescent signal for DENV NS1 in Huh7 cells showed a decrease of approximately 38% and 63% when preincubated with 1 or 10 μg/mL of the anti-SRB1 antibody, respectively, in comparison with the control. The significant decrease in the intracellular NS1 specific signal observed in the preincubated cells suggests that blocking SRB1 with specific antibodies affects the entry of NS1 in both cell lines.

FIG 3.

Inhibition of DENV NS1 internalization by antibodies against SRB1. Huh7 cells (A) and C6/36 cells (C) were incubated with anti-SRB1 antibodies at the indicated concentrations for 1 h; after extensive washing, cells were incubated with DENV NS1 (3.5 μg/well) for 1.5 h. Finally, cells were fixed, permeabilized, and labeled for NS1. Cells were analyzed by confocal microscopy. (B, D). The levels of NS1 inside the cells were quantified by analyzing at least 10 different fields and expressed as the mean ± SD of fluoresce arbitrary units (FAU). Three independent experiments were carried out. Differences in FAU were compared for significance using an ANOVA test. *, P = 0.05. Bars = 10 μm.

HDL is the natural ligand of SRB1 in human liver cells. To evaluate if the natural ligand of the SRB1 will compete with NS1, monolayers of Huh7 and C6/36 cells were preincubated with human HDL at 0, 75 or 150 μg/mL. After several washes to remove unbound HDL, the monolayers were incubated with 3.5 μg/well of DENV NS1, and for this competition study, ZIKV NS1 was also included, at the same concentration. Huh7 cells showed a reduction of approximately 28% and 35% in the amount of internalized DENV NS1. In addition, the incubation of Huh7 cells with HDL also reduced the amount of internalized ZIKV NS1 in approximately 22% and 45%, depending on the HDL concentration (Fig. 4A and B). Interestingly, the incubation of C6/36 cells with human HDL also caused significant reductions in the amount of internalized DENV and ZIKV NS1. With DENV NS1, reductions of approximately 64% and 76% were observed, and with ZIKV NS1, reductions of approximately 81% and 86% were observed, depending on the HDL concentration (Fig. 4C and D). These results agree with the results obtained with the anti-SRB1 antibodies and indicate that SRB1 acts as the cell receptor for both DENV and ZIKV NS1 in vertebrate and mosquito cells.

FIG 4.

Competition assays between human HDL and recombinant DENV and ZIKV NS1. (A) Huh7 and (C) C6/36 cells were incubated with HDL at the indicated concentrations for 1.5 h; after extensive washing, cells were incubated with DENV and ZIKV NS1 (3.5 μg/well) for 1.5 h. Finally, cells were fixed, permeabilized, and labeled for NS1. Cells were analyzed by confocal microscopy. (B, D) The levels of NS1 inside the cells were quantified by analyzing at least 10 different fields and expressed as the mean ± SD of fluoresce arbitrary units (FAU). Three independent experiments were carried out. Differences in FAU were compared for significance using an ANOVA test. *, P = 0.05. Bars = 10 μm.

siRNA knockdown expression of SRB1 in culture cells.

Attempts to knock down the expression of the SRB1 in Huh7 cells using siRNAs specific for the SRB1 resulted in high toxicity for these cells in our hands; therefore, the silencing experiments were attempted in C6/36 cells. Western blot and fluorescence analysis indicated that mosquito cells transfected with siRNAs specific for SBR1 showed approximately 50% reduction in the expression of the SRB1-like receptor, compared to cells transfected with an irrelevant siRNA (Fig. 5A, B, and E). The internalization of DENV NS1 was diminished to approximately half (P < 0.001) in C6/36 cells silenced for SRB1, compared to controls (Fig. 5C and D). These results suggest the SRB1-like receptor of C6/36 cells participates in the entry of DENV NS1.

FIG 5.

Silencing SRB1 expression in C6/36 cell. Confluent C6/36 monolayers were transfected with siRNAs targeting the SRB1, or an irrelevant siRNA, as a control. Silencing efficiency was assayed by immunofluorescence (A, B) and Western blot (E). (C) Transfected cells were incubated with DENV NS1 and 1.5 h, washed, and processed for immunofluorescence. (D) The levels of NS1 inside the cells were quantified by analyzing at least 10 different fields and expressed as the mean ± SD of fluoresce arbitrary units (FAU). Three independent experiments were carried out. Differences in FAU were compared for significance using an ANOVA test. *, P < 0.001. Bars = 10 μm.

Overexpression of the SRB1 in Vero cells.

We observed previously that Vero cells do not express the SRB1 (Fig. 1) and at the concentrations tested, DENV NS1 was not internalized by these cells (Fig. 2). To further demonstrate the participation of the SRB1 in the DENV NS1 internalization process, Vero cells were transfected with a commercial plasmid expressing the human SRB1 and assayed for the capacity to internalize soluble DENV NS1. The expression of the SRB1 in the transfected cells was determined at 48 hpt by Western blot and confocal microscopy (Fig. 6A and B). Immunofluorescence staining of transfected, nonpermeabilized cell monolayers indicated that transfection efficiency was about 10% and that the SRB1 was expressed at the cell surface (Fig. 6A). Next, monolayers of transfected Vero cells were incubated with DENV NS1, fixed, permeabilized, and simultaneously labeled for the SRB1 and DENV NS1. The presence of DENV NS1 was observed in the cytoplasm of all Vero cells expressing the SRB1 (Fig. 6C); meanwhile, no signal for NS1 was observed in Vero cells not transfected for the SRB1 (Fig. 6D). These data suggest that expression of the SRB1 on the cell surface renders Vero cells capable of internalizing DENV NS1.

FIG 6.

Expression of the SRB1 receptor in Vero cells. (A) Immunofluorescence of transfected, nonpermeabilized cells stained for the SRB1 at 48 hpt. (B) Cells lysates from transfected and nontransfected cells were analyzed for the expression of SRB1 by Western blot. *, MWM corresponding to 60 kDa. (C) Transfected cells incubated with DENV NS1 for 1.5 h, fixed, permeabilized, and labeled with anti-NS1 (green) and anti-SRB1 (red) antibodies. (D) Nontransfected cells incubated with NS1 and both primary antibodies. Cell were analyzed by confocal microscopy. Nuclei were counterstained with DAPI. At least 3 independent experiments were carried out. Bars = 20 μm.

The SRB1 interacts directly with DENV and ZIKV NS1 protein.

To determine if there is an actual physical interaction between DENV and ZIKV NS1 and the SRB1, the interactions between them on Huh7 and C6/36 cells were analyzed in situ by proximity ligation assay (PLA). The assay clearly indicated that the added DENV and ZIKV NS1 interact with the SRB1 in Huh 7 (Fig. 7A and E) as well as in mosquito cells (Fig. 7C and G). In contrast, no signal was observed in any of the 2 technical negative controls run in parallel (Fig. 7B, D, F, and H), showing the specificity of the red dot signals observed.

FIG 7.

Proximity ligation assays for dengue and Zika virus NS1 and the SRB1 in human and mosquito cells. Cells incubated with recombinant DENV NS1 (A, C) or ZIKV NS1 (E, G) at a concentration of 3.5 μg/well for 45 min, then fixed and processed following the manufacturer’s instructions. Positive interactions between NS1 and SRB1 are visualized as red dots. Cell nuclei were counterstained with DAPI. B1, D1, F1, and H1 cells were incubated with DENV and ZIKV NS1, but primary anti-NS1 antibody omitted. B2, D2, F2 and H2 cells not incubated with NS1 and incubated with both primary antibodies. (I) Quantification of PLA signal. Results are the mean ± SD of at least 100 cells observed per condition. Experiments were carried out three times, and typical results are shown. Bars = 20 μm.

The second approximation to evaluate DENV NS1 and SRB1 physical interactions was surface plasmon resonance (SPR). As shown in the sensorgram in Fig. 8, the SRB1 developed a faster on-rate (0 to 200 s) and slower off-rate (200 s and on) with the interacting recombinant DENV NS1, at all the NS1 concentrations tested. The DENV NS1 showed a relatively high affinity toward the SRB1, as indicated by the calculated average equilibrium dissociation constant (KD) values, ranging in the subnanomolar range (47.02 ± 0.01 nM versus 10 mM or higher for nonspecific binding interactions). Of note, the average equilibrium dissociation constant value observed for the recombinant NS1 and the SRB1 is in the same order of magnitude as those observed for the HDL and the SRB1 (KD = 18.71 nM). The SPR results corroborate the results obtained in the infected cells by PLA, and together indicate that the SRB1 is bound by the DENV, and ZIKV NS1 protein.

FIG 8.

Surface plasmon resonance sensorgrams obtained for the binding interaction of DENV NS1 with the human SRB1. Increasing concentrations (color lines) of recombinant NS1 were tested for interaction with SRB1 immobilized on a Biacore sensor chip. Binding curves were expressed in resonance units (RU) as a function of time (seconds). Solid black lines through the curve show model fit for the calculation of interaction affinities.

DISCUSSION

NS1 is one of the most pleiotropic proteins encoded by flaviviruses described so far. The ability of DENV NS1 to be secreted concomitantly with viral particles and reach the extracellular milieu makes it a powerful “viral weapon” capable of participating in several pathogenesis processes, both intracellularly and extracellularly, thus favoring the installation and progression of the infection (28). DENV NS1 is internalized in cultured cells as well as in vivo, out of the context of the viral infection, to enhance viral production. Also, the endothelium disrupting capacity of flavivirus NS1 is dependent on cell internalization. However, little is known about the mechanisms and host molecules involved in NS1 internalization. In this work, we assessed the participation of the SRB1 as a receptor of the DENV and ZIKV NS1 in human hepatic and insect cells. Several lines of evidence were obtained, proving that the NS1 proteins of DENV and ZIKV use the SRB1 as cell-binding protein: (i) only those cells expressing the SRB1 are able to internalize NS1; (ii) entry of NS1 was affected by anti-SRB1 antibodies; (iii) NS1 internalization is competed by HDL, the natural ligand of the SRB1; (iv) NS1 internalization is reduced in C6/36 silenced for SRB1 expression; (v) the expression of the SRB1 in cells naturally lacking the receptor makes the cells susceptible to NS1 entry; (vi) physical interaction between NS1 and the SRB1 was obtained both in situ and in vitro. The observation that the DENV and ZIKV NS1 use and bind to the SRB1 is in line with the fact that NS1 is indeed a bona fide lipoprotein. Interestingly, NS1 recognizes an SRB1-like receptor in mosquito cells. Mosquito cells are unable to synthesize cholesterol de novo, and SRB1-like receptors may participate in cholesterol uptake by these cells. Moreover, and as has been previously described for mammalian cells, the preincubation of insect cells with DENV or ZIKV NS1 renders the cells more susceptible to the respective viral infection. NS1 antigenemia have been shown to determine ZIKV infectivity in Aedes aegypti (29). Thus, NS1 binding to SRB1-like receptors in mosquito cells and the promotion of viral replication after internalization may be part of the mechanisms underlying this observation. The mechanisms by which NS1 contributes to increasing the viral yield are not known. However, it has been proposed that the accumulation of DENV NS1 in the late endosomal compartment, after the entry to hepatocytes, potentializes subsequent dengue virus infection in vitro (17), and incubation of cultured RAW 264.7 cells with soluble DENV NS1 results in lipid raft accumulation and increased virion attachment (30). Recently, it was reported that endocytosis of NS1 is required for NS1-mediated endothelial hyperpermeability (16). However, it is unknown if the SRB1 also participates in the entry of NS1 in endothelial cells.

The entry of DENV NS1 both in hepatic and insect cells was hampered by the preincubation with a specific monoclonal antibody against the SRB1. In both cells, there was a significant reduction of the NS1 internalization when 1 or 10 μg/mL of the antibody was used. However, the antibody only partially abolishes the NS1 binding and internalization, especially in hepatic cells, even at the higher concentrations used. It has been reported that position C323 plays a critical role in HDL binding and that the exoplasmic C384 plays a role in the selective HDL-cholesteryl esters (CE) uptake and in SRB1-mediated selective HDL-CE transport activity (31, 32). The antibody used in this work was developed against an epitope near the amino-terminal region of the SRB1, away from positions C323 and C384. Thus, the antibody may only partially cause a steric impediment for NS1 internalization. Also, a significant but incomplete competition between NS1 and HDL, the natural ligand of the SRB1, was observed. However, it is unknown if HDL and NS1 bind to the same domains on the SRB1 molecule. The entry of a protein into a cell is a multifactorial process since the plasmatic membrane is a plethora of different kinds of molecules determined by the origin of the cell, which in turn modulate the nature of proteins to be internalized (33). Thus, as is the case for HCV entrance, where the successful virion binding to hepatic cells requires in addition to the SRB1, claudin-1, the human cluster differentiation CD81, and occludin as part of a minimal set of receptors working orchestrated as a multimolecular complex (34), SRB1 may not be the only receptor molecule for NS1. Even so, the participation of the SRB1 as a key cell-binding molecule for NS1 is supported by the results obtained with transfected Vero cells where the sole expression of SRB1 is enough to render the cell susceptible to NS1 entry. In the absence of the SRB1, Vero cells did not internalize NS1 under the experimental conditions used. This observation agrees with previous results where the presence of NS1 was detected in the liver but not in the kidneys of mice exposed intravenously to DENV NS1 (17).

To point to a given molecule as a receptor for a ligand, evidence of direct interaction between the molecule and the ligand is required. In this work, direct physical interactions between the SRB1 expressed in Huh7 and C6/36 cells and the added soluble DENV and ZIKV NS1 were demonstrated by PLA. Several negative controls assure the specificity of the PLA reaction. In addition, the PLA results were corroborated in vitro by SPR. The results obtained by SPR clearly indicate physical interactions between recombinant, soluble DENV NS1 protein and a recombinant human SRB1, in a dose-dependent manner. Interestingly, the use of HDL in the SPR assays as positive control showed that the affinity of the DENV NS1 and the human HDL for SRB1 does not differ significantly. In hepatic cells, SRB1 is the main receptor of the HDL, playing an important role in the homeostasis of lipoproteins circulating in serum (21). The main function of the SRB1 is to participate in the influx and efflux of the HDL hauled molecules contributing to lipid homeostasis. The observation that HDL and DENV NS1 bind to the same receptor with comparable affinities, together with the high levels of circulating NS1 seen in patients during the acute phase of the disease, offers an explanation for the alterations in lipid homeostasis seen in dengue patients, as, for example, the decrease in HDL levels observed in patients with severe dengue (25).

In summary, in this work evidence is presented that the SRB1 in hepatic cells and an SRB1-like receptor in mosquito cells act as the cell-binding protein for DENV and ZIKV NS1. Moreover, evidence is presented indicating that the C6/36 receptor is functional. It had been described that the selective uptake of cholesteryl esters by the cells through the SRB1, either in vitro and/or in vivo, does not show any specificity toward a single class of lipoproteins (19). Thus, we hypothesize that NS1 hijacks classic lipid routes for cell entry through the SRB1, but it is unknown if the lipids carried out by NS1 are transferred to the cell. Once inside the cell, NS1 promotes viral replication and causes the delocalization of tight junction proteins. Thus, future experiments should include the elucidation of the entry steps followed by NS1 and the signaling pathways that are triggered after the SRB1 and NS1 interactions.

MATERIALS AND METHODS

Virus and cells.

Dengue 4 virus (kindly donated by Dr. Juan Salas, Instituto Politécnico Nacional, Mexico) was propagated in suckling mouse brains as previously described (35). Zika virus strain M7366 (kindly donated by Prof. Susana López, Instituto de Biotecnología, UNAM, Mexico) was propagated in C6/36 cells. Virus titers from mice brain homogenates or cell culture supernatant were determined by plaque assay as described by Ludert et al. (36) but using Vero cells grown in the cell culture conditions described in this work. C6/36 cells (ATCC, USA) were grown in EMEM medium at 28°C and 5% CO₂ atmosphere. Huh7 and Vero cells were grown in DMEM medium (Sigma, USA) at 37°C and 8% CO₂ atmosphere. All cell culture media were supplemented with 5% FBS.

Cell surface localization of SBR1.

Confluent monolayers grown in LAB-TEK (Nalgene Nunc International, USA) were fixed (4% paraformaldehyde-PBS) for 10 min at room temperature (RT) and then incubated with blocking buffer (10% FBS, 3% BSA, 100 mM glycine in PBS) for 45 min at RT. An anti-SRB1 rabbit polyclonal antibody (Abcam, USA, cat# ab 106572), was used as primary antibody for the detection of the SRB1 molecule in Huh7, Vero, and C6/36 cells. This polyclonal antibody was raised using a 15-amino-acid peptide located within amino acids 70–100 of the human SRB1 molecule. BLAST analysis of this region of the human SRB1 (NP_005496.4) shows 59.26% identity with the scavenger receptor class B member 1 (NCBI protein database code XP_019931364.2) of Aedes albopictus, and an E value of 2e-04. By Western blot, the commercial serum recognizes a band of approximately 60 Kd in mosquito cells; meanwhile, a band of approximately 60 Kd was recognized in Huh cell lysates, and no band was recognized in Vero cells (data not shown). For immunofluorescent assays, the anti-SRB1 sera was diluted 1:300 (10% FBS, 1% BSA, 100 mM glycine in PBS) and incubated with the cells for 1 h, 37°C. After 5 washes, cells were incubated with an antirabbit Alexa Fluor 594 (Invitrogen, USA) diluted 1:1000 in PBS. After 10 washes, the cells were incubated with DAPI (4′,6-Diamidine-2′-phenylindole dihydrochloride) in mounting medium (Molecular Probes). Afterward, a coverslip was added and sealed with nail polish. The immunolabeled cells were observed in the inverted confocal microscope (Olympus MPhot), and the images were analyzed in the Image J/FIJI software. To capture images with a higher resolution of the molecules located on the plasma membranes, monolayers of Huh7 and C6/36 were cells grown in Fluorodish Cell Cultures (WPI, USA) and the SRB-1 coimmunolabeled with wheat germ agglutinin conjugated with Alexa Fluor488 (Thermofisher, USA) to observe the silhouette of the cells and the integrity of the plasmatic membrane. Colabeled cells were examined by total internal reflection fluorescence microscopy (TIRF) (Olympus IX81 TIRF). The antibodies and conditions used in TIRF were as described for the confocal microscopy.

NS1 internalization assays.

Huh7 and C6/36 cells were grown in 8-well chambers (LAB-TEK, Nalgene Nunc International, USA) for 24 h, until confluent monolayers were formed. Monolayers were incubated with 3.5 μg/well of DENV serotype 2 (Aalto Bio Reagents, Dublin, cat# CL 6243) or ZIKV (Aalto Bio Reagents, Dublin, cat# AZ 6903) NS1 recombinant protein, produced in insect cells. As a control, cells were also incubated with recombinant NS1 heat denatured for 60 min in boiling water. After 1.5 h of incubation, at the cell-specific growth temperature, cells were washed extensively, fixed with 4% paraformaldehyde-PBS for 10 min at RT, and permeabilized with 0.1% Triton-PBS, for 5 min at RT and finally incubated with blocking buffer (10% FBS, 3% BSA, 100 mM glycine in PBS) for 45 min at RT. Internalized NS1 was revealed by immunolabeling using an anti-NS1 Mab, clone 7E11 (kindly donated by Prof. Eva Harris, University of California, Berkeley; 16) diluted 1:300 (10% FBS, 1% BSA, 100 mM glycine in PBS) as primary antibody incubated for 1 h at 37°C. After 5 washes, cells were incubated with an antimouse Alexa fluor 568 (Invitrogen, USA) diluted 1:1000 in PBS as secondary antibody. After 10 washes, the cells were incubated with DAPI in mounting medium (Molecular Probes). Afterward, a coverslip was added and sealed with nail polish. The immunolabeled cells were observed in an inverted confocal microscope (Olympus MPhot) and images analyzed with Image J/FIJI software. For the quantification of NS1 levels inside the cells, 10 cells by field on 3 fields by condition were selected as regions of interest, and arbitrary fluorescent units were measured using the Image J/Fiji software. The means for each condition were calculated, and an analysis of variance (ANOVA) one-tailed test was used to compare the mean of fluorescent arbitrary units (FAU) from each measure.

DENV and ZIKV infection in insect cells preincubated with recombinant NS1.

C6/36 cell monolayers grown in 24-well plates were preincubated or not with 3.,5 and 7.0 μg/well of recombinant DENV or ZIKV NS1 protein for 1.5 h. As a control, heat inactivated NS1 (7.0 μg/well) was also included. After extensive washing, cells were inoculated with DENV or ZIKV at MOI = 1. Infection was left to proceed for 48 h, the supernatants collected, and the virus yield titrated by plaque assay using Vero cells, as previously described (36).

Inhibition of NS1 entry by preincubation with anti-SRB1 antibodies.

Confluent monolayers of Huh7 and C6/36 cells, grown in 8-well chambers LAB-TEK (Nalgene Nunc International, USA) for 24 h, were preincubated or not with an anti-SRB1 rabbit polyclonal antibody at concentrations of 1 and 10 μg/mL for 1 h at 37°C. After extensive washing, cells were incubated with recombinant DENV NS1 protein at a concentration of 3.5 μg/well for 1.5 h, and the cells fixed and processed for detection and quantification of internalized NS1 as described above.

NS1 and HDL competition assays.

C6/36 or Huh7 was seeded in 8-well chambers, and confluent monolayers were incubated with 150 or 300 μg/mL of HDL (Merck, USA, cat# LP3) in specific growth medium without FBS for 1.5 h, at the appropriate cell growth conditions. After extensive washing, 3.5 μg/well of DENV or ZIKV recombinant NS1 protein were added in growth medium without FBS. After incubation for 1.5 h at the cell specific growth conditions, cells were fixed, permeabilized, and immunolabeled for detection of internalized DENV or ZIKV NS1 as described above.

Silencing SRB1 expression in C6/36 cells.

Confluent monolayers of C6/36 cells grown in Lab-Tek II (Nalgen Nunc International, Denmark) were transfected with 30 nM SRB1 siRNA (Ambion, Lithuania, Cat AM16708), 30 nM negative control siRNA (Ambion, Lithuania, Cat 4618G), or nontransfected, using Siport Neo Fx siRNA transfection agent (Invitrogen silencer siRNA Transfection Kit, Lithuania, Cat AM4510) according to the manufacturer’s instructions. Gene knockdown efficiency was evaluated by Western blot and immunofluorescence assays 48 hpt. Final transfection conditions were chosen according to silencing efficiency and cell viability. To analyze the effect of the SRB1 knockdown on the internalization of the recombinant DENV NS1 protein, 48 hpt cells were incubated with 7.5 μg/well of the NS1 protein diluted in EMEM medium without FBS. After 90 min incubation, cells were washed, fixed with 4% paraformaldehyde in PBS, and permeabilized with PBS Triton X-100 0.5% during 5 min. Finally, cells were immunolabeled with anti-NS1 Mab as described before and analyzed by confocal microscopy. The fluorescence arbitrary units (FAU) detected from 10 cells by field in 10 fields, were used to estimate the amount of SRB1 or NS1.

Overexpression of SRB1 in Vero cells.

Monolayers, 80–90% of confluence, of Vero cells seeded in 8-well chambers, were transfected with the plasmid pCMV-SCARB1-His (Sino Biological, USA; cat# HG11069-CH) for the expression of human SRB1. Transfections were carried out with VeroFect (OZBIOSCIENCES; cat# VF60250) according to manufacturer’s instructions using 0.5 μg/well of the plasmid. Twenty-four hpt, supernatants were removed and the cells incubated with 3.5 μg/mL of recombinant DENV NS1 in DMEM medium without FBS for 1.5 h at the cell growth conditions. Cells were fixed, permeabilized, and coimmunolabeled with anti-NS1 and anti-SRB1 specific primary antibodies and reveled and visualized by confocal microscopy as described above. Nontransfected Vero cells incubated with both primary antibodies were included as controls for specificity of antibodies. In addition to immunofluorescence, SRB1 expression in transfected Vero cells was evaluated by Western blot.

Proximity ligation assays.

Interactions between the SRB1 and NS1 proteins from DENV and ZIKV in Huh7 and C6/36 cells were detected using the commercial kit Duolink (Sigma-Aldrich, cat# DUO 92008) based on in situ proximity ligation assay (PLA), and used following manufacturer´s instructions. PLA allows for the detection of direct protein-protein interactions at distances <40 nm, in intact fixed cells (37, 38). Briefly, C6/36 or Huh7 monolayers were grown on 8-well chambers, preincubated with DENV or ZIKV NS1 recombinant proteins (3.5 μg/well), and after 45 min of incubation, fixed with 4 % paraformaldehyde for 10 min at RT. After incubation with a blocking buffer for 30 min at 37°C, cells were incubated with mouse anti-NS1 MAb and rabbit anti-SRB1 polyclonal antibody (Abcam, USA) as primary antibodies. After several washes, cells were incubated with the PLA probes linked, antimouse and antirabbit secondary antibodies. Finally, the cells were subjected to ligation and amplification reactions. The specificity of PLA was evaluated using two negative controls: cells incubated with DENV NS1 and with the NS1 specific antibody only, and cells not exposed to NS1, and treated to both primary antibodies. The PLA signal was detected in a confocal microscope (Inverted Olympus MPhot) using a 60× immersion oil objective and values determined considering the mean of three images per condition in three independent experiments. Image analysis was carried out using the Fiji ImageJ software. PLA signal (dots/cell) was quantified in at least 100 cells per condition in maximum projection images.

Surface plasmon resonance.

The extracellular domain of human SRB1, expressed in HEK 293 cells (Sino Biological, USA, cat# 11069-H08H), was covalently immobilized onto carboxymethylated-dextran sensor chips (CM5 series S, General Electric Healthcare). Previously, the sensor chip surface was activated using 1-ethyl-3-(3 dimethyl aminopropyl) carbodiimide (EDC) and N-hydroxysuccinimide (NHS) reagents, according to the standard procedure for amine coupling. For direct capture-coupling to NH₂ (in immobilization buffer of 10 mM sodium acetate, pH 5.0), the SRB1 was diluted to 30 μg/mL and injected at 5 μg/min flow rate until a density of 10,000 resonance units was reached, using running buffer HBS-EP. The remaining surface-active groups were blocked with 1 M ethanolamine hydrochloride, pH 8.5. A dilution series of the DENV recombinant NS1 protein (Aalto Bio Reagents, Dublin, cat# CL6243), at six concentrations of 23.4, 46.8, 93.7, 187.5, 375, and 750 nM, were then injected at a flow rate of 30 μL/min for 200 s for association with SR-B1, followed by dissociation step adjusted to 600 s. The surface was successively regenerated between analyte injections with a single-brief (30 s) injection of glycine-HCl, pH 2.5. All analyses were carried out at 25°C using a Biacore T200 (General Electric Healthcare). Surface plasmon resonance (SPR) response data defined as sensorgrams were zeroed at the beginning of each individual injection, and then double referenced. The responses were plotted against the analyte concentration and fit to a 1:1 (A+B = AB) binding model using Biacore T200 evaluation software, version 2.0 (General Electric Healthcare).

Statistical analysis.

All results were expressed as means ± standard deviations (SD) from at least three independent experiments and analyzed for statistical significance using the ANOVA test.