Inhibition of the Membrane Attack Complex by Dengue Virus NS1 through Interaction with Vitronectin and Terminal Complement Proteins

Proteins and antibodies.

Human vitronectin (VN), human hemopexin (HPX), bovine serum albumin (BSA), and heparin were purchased from Sigma-Aldrich. Human plasminogen (PLG) was purchased from Millipore. Recombinant active plasminogen activator inhibitor 1 (PAI-1) was kindly provided by Cynthia Peterson (University of Tennessee, TN), and the purified complement proteins C2, C5, C6, C7, C8, C9, and C5b6, factor H, and normal human serum (NHS) were purchased from Complement Technology.

Hexameric complex-type glycosylated NS1 proteins (NS1_mam) from DENV2 (strain Thailand/16681/84), WNV (strain New York NY99), and ZIKV (strain Uganda MR 766), expressed in HEK 293 cells, were purchased from The Native Antigen (Oxfordshire, United Kingdom).

Polyclonal antiserum against purified NS1 was raised in mice and rabbits and purified using protein G HP Spin Trap columns (GE Healthcare) (19, 20). Mouse anti-human VN and PLG monoclonal antibodies were purchased from Abcam, and goat anti-human C2, C5, C6, C7, C8, and C9 and rabbit anti-VN were purchased from Complement Technology. Anti-mouse and anti-goat IgGs conjugated to horseradish peroxidase were purchased from Promega.

Production of NS1 protein.

Recombinant dimeric NS1 was expressed in Escherichia coli (NS1_e). Briefly, the DENV2 ns1 gene from the New Guinea strain was inserted into the pET23b plasmid by GenScript (CA). A nucleotide sequence encoding a 6×His tag was added at the 3′ end of the ns1 gene. The recombinant plasmids were inserted into E. coli BL21(λDE3)pLysS cells, and the expression and purification steps were performed as described previously (19). Hexameric polymannose-glycosylated NS1 (NS1_dros) was produced and isolated from the supernatant of a DENV2 NS1-expressing Drosophila Schneider 2 (S2) stable cell line. S2 cells were cultured in Schneider's Drosophila medium (Gibco) for 5 days at 28°C, and NS1_dros expression was induced with 500 mM ZnCl₂ in serum-free Sf-900 II SFM medium (Gibco). The supernatant was harvested, and NS1_dros was purified using size exclusion chromatography on a Superdex 200 column (GE Healthcare).

Y2H screening.

Yeast two-hybrid (Y2H) screening was performed against a human liver cDNA library of sequences fused to the GAL4 activation domain by use of the pACT2 vector (Clontech, CA). Saccharomyces cerevisiae strain AH109 was transformed with the pGBKT7-NS1 plasmid, which encoded full-length DENV2 NS1 as bait, by using the lithium acetate method and then was grown in synthetic defined medium lacking tryptophan (SD–Trp). Autoactivation of the HIS3 reporter gene was verified by the growth of colonies in SD medium lacking histidine, leucine, and tryptophan (SD–His–Leu–Trp). The transformed cultures were then plated onto SD–His–Leu–Trp and SD–Ade–His–Leu–Trp media to select putative positive clones. The activity of the lacZ reporter gene was evaluated by a β-galactosidase assay (colony lift filter assay), using the substrate X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) on nitrocellulose membranes. To eliminate false-positive results, a plasmid linkage assay was also performed. The positive plasmids were sequenced, and their gene sequences were analyzed using the BLASTN and BLASTX software available at the NCBI website (www.ncbi.nlm.nih.gov).

Coimmunoprecipitation.

Human hepatocellular carcinoma (HuH-7.5.1) cell lines were cultured in high-glucose Dulbecco's modified Eagle's medium (DMEM) (Gibco) supplemented with 10% fetal bovine serum (Invitrogen, NY), 2 mM l-glutamine (Gibco), penicillin and streptomycin (Gibco), 0.22% sodium bicarbonate, and 0.2% HEPES, pH 7.4, in a humidified CO₂ incubation chamber at 37°C. After 2 days, HuH-7.5.1 cells were mock infected or infected with DENV2 strain 16681 at a multiplicity of infection (MOI) of 1. The HuH-7.5.1 culture supernatants were collected at 48 h postinfection. The coimmunoprecipitation assays were performed using a Pierce coimmunoprecipitation kit (Pierce). Purified anti-NS1 polyclonal antibody was attached to an N-terminus-binding resin, and the supernatant (S) of mock-infected or DENV2-infected HuH-7.5.1 cells was added to the resin. The coimmunoprecipitation flowthrough (F) and eluted (E) samples were separated by 12% SDS-PAGE and transferred to a Hybond ECL nitrocellulose membrane (GE Healthcare). The membrane was then blocked with 5% nonfat milk in TBST (0.1% Tween 20 in Tris-buffered saline [25 mM Tris-HCl, pH 7.6, 3 mM KCl, and 140 mM NaCl]) for 2 h, followed by overnight incubation with an anti-NS1 or anti-VN antibody in blocking solution. The membrane was then washed three times with TBST and incubated with horseradish peroxidase-conjugated anti-mouse IgG in blocking solution for 2 h. The membrane was washed again and developed with a Supersignal West Pico kit (Pierce, IL), and the image was acquired using ImageQuant LAS-4000 (GE Healthcare).

Direct binding assays.

Microtiter plates were coated with 100 nM VN, PAI-1, HPX, or BSA in phosphate-buffered saline (PBS; 8.06 mM Na₂HPO₄, 1.94 mM KH₂PO₄, 2.7 mM KCl, and 137 mM NaCl; pH 7.4) overnight at 4°C. After five washes with PBS, the wells were blocked with 200 μl of 2% BSA in PBS containing 0.05% Tween 20 (PBST) for 1 h at 37°C, followed by five washes with PBST. NS1_e at specific concentrations was added to each well and incubated for 2 h at 37°C, followed by a 1-h incubation with anti-NS1. Peroxidase-conjugated anti-mouse IgG was then added for 1 h at 37°C. The plates were washed five times with PBST between each step.

To investigate the effect of heparin or PAI-1 on the NS1-VN interaction, the binding buffer was supplemented with 0 to 500 μg/ml heparin or with 0 to 5 μg/ml PAI-1, and bound NS1_e (10 or 1 μg/ml, respectively) was detected using anti-NS1 followed by a peroxidase-conjugated antibody. To investigate the effect of PLG on NS1_e binding to immobilized VN, a mixture with a constant amount of NS1_e (100 nM) and PLG at different molar ratios was added to VN-coated wells (100 nM). Similarly, NS1_e was used at different molar concentrations combined with a constant amount of PLG (100 nM) and added to immobilized VN. Bound NS1 and PLG were detected using anti-NS1 and anti-PLG, respectively, followed by the addition of peroxidase-conjugated antibodies.

To assess binding of complement components to NS1_e-coated wells (10 μg/ml), the proteins C2, C5, C6, C7, C8, and C9 (0 to 1,000 nM [each]) were incubated for 1 h at 37°C. Bound components were detected using specific polyclonal antibodies followed by a peroxidase-conjugated anti-goat IgG antibody. To investigate whether C9 would bind to DENV2, WNV, or ZIKV NS1_mam-coated wells (10 μg/ml), C9 (0 to 500 nM) was added to each well and incubated for 1 h at 37°C. Wells were then washed with PBS and incubated with a polyclonal anti-C9 antibody followed by incubation with a peroxidase-conjugated anti-goat IgG antibody for 1 h at 37°C. The reactions from all binding experiments were developed by adding 100 μl of 0.4 mg/ml o-phenylenediamine and 50 μl of 9 N H₂SO₄. The optical density at 490 nm (OD₄₉₀) was determined by use of a 96-well plate reader.

SPR measurements.

Surface plasmon resonance (SPR) measurements were performed using a Biacore X instrument (GE Healthcare). VN was covalently immobilized to the dextran matrix of a CM5 sensor chip via the primary amine groups (amine coupling kit; GE Healthcare). The carboxymethylated dextran surface was activated by the injection of a mixture of 0.2 M N-ethyl-N′-(diethylamino-propyl)carbodiimide and 0.05 M N-hydroxysuccinimide. The ligand was injected in 10 mM sodium acetate buffer (pH 4.0). The remaining N-hydroxysuccinimide esters were blocked by injection of 1 M ethanolamine hydrochloride (pH 8.5). All immobilization steps were performed at a flow rate of 10 μl/min in 10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM EDTA, and 0.005% P20 (HBS-EP; GE Healthcare). The immobilization level for VN was 8,500 response units (RU). No protein was immobilized on the control flow cell that underwent the activation and blocking steps. Binding experiments were performed at 25°C at a flow rate of 15 μl/min in HBS-EP. The NS1_e protein was injected at different concentrations ranging from 0.12 μM to 3.0 μM. The data were double referenced by subtraction of the control flow cell signal and a blank run (buffer only) signal. In all experiments, association phases ran for 120 s and dissociation phases for 280 s. The surface was regenerated with pulses of 0.01 M glycine (pH 2.0). The data were analyzed by global fitting to a 1:1 Langmuir binding model of both the association and dissociation phases for several concentrations simultaneously, using BIAevaluation 4.1 software (BIAcore). In each case, the data were obtained with a χ² value of <2. The apparent equilibrium dissociation constants (K_D) were calculated from the ratio of the dissociation and association rate constants (k_d/k_a).

Sandwich enzyme-linked immunosorbent assay (ELISA) with plasmas from DENV-infected patients.

Wells were coated at 4°C overnight with purified anti-NS1 rabbit polyclonal antibodies (20 μg/ml) in PBS. The plates were blocked for 1 h at 37°C with 2% BSA in PBST and then washed five times with PBST. This step was performed after each period of incubation. The wells were incubated with plasmas from DENV-infected patients or healthy blood donors (HD), diluted 1:2 in PBS, for 4 h at 37°C. Subsequently, the wells were incubated with an anti-VN monoclonal antibody (1:10,000) for 1 h at 37°C, followed by incubation for 1 h at 37°C with a peroxidase-conjugated anti-mouse IgG antibody (1:10,000). The reactions were developed as described above. Each sample was assayed twice.

The use of human samples was approved by the Research Ethics Committee for Clinical Studies of the University Hospital Clementino Fraga Filho, Federal University of Rio de Janeiro (UFRJ), and registered with the Brazilian Human Ethics Committee (Plataforma Brasil, CAAE; protocol 00534012.6.0000.5257; expiration date, 3 June 2017). Informed consent was not obtained because the patient records and information were anonymized and deidentified prior to analysis.

Hemolytic assay.

The MAC-inhibitory activity of NS1 and VN was analyzed in a hemolytic assay using sheep erythrocytes and purified complement proteins (18). Erythrocytes were resuspended to 1 × 10⁸ cells/ml in Veronal-buffered saline (VBS) and preincubated with 1 μg/ml C5b6 for 1 h at room temperature. In a separate preparation, a mixture of purified NS1_e (12.5 μg/ml) and VN (20 μg/ml), NS1_e (12.5 μg/ml) only, or VN (20 μg/ml) only was preincubated for 1 h at 37°C, and the complement proteins C7 (1 μg/ml), C8 (0.1 μg/ml), and C9 (1 μg/ml) (C7–9) were added to the mixture for 15 min at 37°C. Thereafter, the C5b6-coated erythrocytes were added to the NS1_e-VN-C7–9 mixture and incubated for 30 min at 37°C. Alternatively, increased amounts of NS1_e or NS1_dros (6.25, 12.5, and 25 μg/ml) were preincubated with the C7–9 proteins for 15 min at 37°C before incubation with C5b6-coated erythrocytes. Erythrocytes were centrifuged, and the amount of hemoglobin released from the lysed cells was measured by determining the absorbance at 540 nm. The relative MAC-inhibitory activity was presented as the percentage of total hemolysis. NS1 buffer and factor H (50 μg/ml) were used as negative controls.

C9 polymerization assay.

The effects of NS1_e and NS1_dros on C9 polymerization were assessed according to a previously published protocol (21). Briefly, NS1_e (10 to 100 μg/ml), NS1_dros (10 to 100 μg/ml), or NS1 buffer was preincubated with 3 μg of C9 at 37°C in 20 mM Tris-HCl (pH 7.2). After 40 min of incubation, 50 μM ZnCl₂ in 20 mM Tris-HCl (pH 7.2) was added for 2 h at 37°C. In a separate experiment, NS1_e (10 or 25 μg/ml) was preincubated with VN (10 or 25 μg/ml) for 30 min at 37°C in 20 mM Tris-HCl (pH 7.2), and 3 μg of C9 was added for 40 min at 37°C. Polymerization was induced by addition of 50 μM ZnCl₂ in 20 mM Tris-HCl (pH 7.2) for 2 h at 37°C. The samples were separated in precast 4 to 20% gradient polyacrylamide gels (Bio-Rad) under nonreducing conditions, and C9 polymerization was visualized by silver staining. The C9 monomer band intensity was evaluated by densitometry using ImageJ software, version 1.49 (National Institutes of Health).

The effect of NS1_mam on C9 polymerization was assessed by using proteins from DENV2, WNV, and ZIKV. Initially, BHK-21 cells were cultivated in a 24-well microplate in alpha minimal essential medium (α-MEM; Gibco) supplemented with 10% fetal bovine serum, 0.22% sodium bicarbonate, and 0.2% HEPES, pH 7.4, in a humidified CO₂ incubation chamber at 37°C. After 2 days, cells were washed three times with VBS and preincubated with 5 μg/ml purified C5b6 for 2 h at 37°C in VBS. The NS1_mam protein from DENV2, WNV, or ZIKV was preincubated with 5% NHS for 30 min at 37°C. Rabbit polyclonal anti-NS1 (50 μg/ml), which cross-reacts with the WNV and ZIKV NS1 proteins (data not shown), was mixed with 50 μg/ml NS1_mam to serve as a control. Cells were washed with VBS three times and then incubated with NS1_mam mixed with NHS for 50 min at 37°C. Thereafter, cells were washed three times with VBS and lysed with cell lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% NP-40). The samples were separated in a 10% polyacrylamide gel under nonreducing conditions and then transferred to a 0.45-μm Immobilon-P polyvinylidene difluoride (PVDF) membrane (GE Healthcare) at 15 V for 16 h. The membrane was blocked with 5% nonfat milk in TBST for 2 h, followed by overnight incubation with an anti-C9 antibody in blocking solution. The membrane was then washed three times with TBST and incubated with anti-goat IgG conjugated to horseradish peroxidase in blocking solution for 2 h. The membrane was washed again and developed with a Supersignal West Pico kit (Pierce, IL), and the image was acquired using ImageQuant LAS-4000 (GE Healthcare).

To evaluate the effect of VN preincubation with NS1_mam on C9 polymerization, VN (0 to 25 μg/ml) was preincubated with the NS1_mam protein from DENV2, WNV, or ZIKV (25 μg/ml) for 30 min at 37°C, incubated with 5% NHS for 30 min at 37°C, and added to C5b6-treated BHK-21 cells for 50 min at 37°C. Thereafter, cells were washed three times with VBS and lysed with cell lysis buffer, and the C9 polymer was analyzed by Western blotting using anti-C9.

Statistical analysis.

The data sets were compared using two-tailed, unpaired Student's t test, and statistical significance was achieved when P values were <0.05. Multiple comparisons were performed using one-way or two-way analysis of variance (ANOVA) (with the Bonferroni posttest), and asterisks in figures indicate significant differences from the control, as indicated in the legends.

Identification of VN as an NS1-interacting partner.

Using full-length NS1 as bait, we performed a Y2H screen of a human liver cDNA library as previously described by our group (8). Putative DENV NS1 protein-interacting partners were identified by activation of the HIS3 and ADE2 reporter genes, and the authenticity of the interaction was confirmed by colony lift filter assays of positive clones. True positive transformants activated the β-galactosidase (lacZ) reporter gene and appeared blue, whereas false-positive clones remained colorless (Fig. 1A). One positive clone identified by the colony lift assay was subjected to nucleotide sequencing and identified as carrying the human VN gene.

To assess whether this interaction occurs in DENV-infected cells, we first infected HuH7.5.1 cells, which have been shown to express VN (22), with DENV2 strain 16681 at an MOI of 1 for 48 h. The mock- or DENV2-infected HuH7.5.1 cell supernatants (S) were then loaded onto a column containing anti-NS1 polyclonal antibodies immobilized on a resin to collect all coimmunoprecipitated proteins. The elution fractions (E) were then analyzed using anti-NS1 and anti-VN polyclonal antibodies. We observed two bands, of 75 and 65 kDa, corresponding to VN (Fig. 1B, upper panel) and one band, of approximately 50 kDa, corresponding to DENV2 NS1 (Fig. 1B, lower panel). This result indicates that VN was coimmunoprecipitated with the NS1 protein.

DENV NS1 directly binds VN.

To assess whether NS1 directly interacts with VN, purified human VN was immobilized onto microtiter plates, and serial dilutions of recombinant dimeric NS1_e (19) were added. NS1_e interacted with VN in a dose-dependent manner, whereas no binding was observed with BSA, used as a negative control (Fig. 2A). This interaction was further confirmed by SPR analysis, in which VN was immobilized on a CM5 sensor chip dextran surface and the NS1_e protein was injected at different concentrations, ranging from 0.12 μM to 3.0 μM. A concentration-dependent binding of NS1_e to immobilized VN was observed (Fig. 2B). An evaluation of the global rates of association and dissociation (k_a and k_d, respectively) revealed a k_a of 4.23 × 10⁴ ± 0.23 × 10⁴ M⁻¹ s⁻¹ and a k_d of 6.45 × 10⁻² ± 0.92 × 10⁻² s⁻¹. The global dissociation equilibrium constant (K_D) was 1.52 ± 0.17 μM, assuming a simple 1:1 Langmuir binding mode. Taken together, the ELISA and SPR data indicate that DENV2 NS1 directly binds VN.

DENV NS1-VN interaction is increased by PAI-1 and inhibited by plasminogen.

VN can bind several molecules, such as heparin, PAI-1, HPX, and PLG, and these interactions occur at distinct sites in the VN molecule (23–27). Therefore, to identify the putative binding sites of NS1 on the VN molecule, competitive ELISAs were performed in which VN ligands were used as competitors (28). VN-coated wells were incubated with fixed concentrations of NS1_e and various concentrations of heparin (15.6 to 500.0 μg/ml) or PAI-1 (0.16 to 5.00 μg/ml). The addition of increasing amounts of heparin did not affect binding (Fig. 3A), suggesting that the NS1-VN interaction does not occur through the HBDs. However, when PAI-1 was coincubated with NS1_e, an unexpected dose-dependent increase in the binding of NS1 to VN was observed (Fig. 3B), thus indicating that PAI-1 facilitates VN-NS1 interactions. To evaluate whether NS1 directly interacts with PAI-1, purified human PAI-1 was immobilized and incubated with increasing concentrations of NS1_e. A dose-dependent interaction was observed, thus confirming direct binding between these molecules (Fig. 3C).

A canonical feature of the VN structure is the homology with hemopexin-type repeats found in human HPX. Computationally predicted models indicate that the VN central domain adopts a full four-bladed β-propeller fold similar to that described for HPX (29). Moreover, both HPX and VN hemopexin domains are recognized by surface proteins of Streptococcus pyogenes, and these interactions are important for bacterial attachment (30). Based on these findings, we assessed whether NS1 interacts directly with HPX or, alternatively, if NS1 competes with HPX for binding to VN. First, HPX-coated wells were incubated with increasing concentrations of NS1_e. No significant binding was observed (Fig. 3C). In a competitive binding assay in which VN-coated wells were incubated with fixed concentrations of NS1_e and increasing concentrations of HPX (1.8 to 30.0 μg/ml), the addition of HPX did not alter the NS1-VN interaction (data not shown). These two assays indicate that NS1 does not bind HPX and does not interact with VN via the hemopexin domains.

Because PLG binds VN, we hypothesized that PLG-binding sites on VN may overlap the NS1-binding sites. A constant amount of NS1_e (100 nM) with increasing amounts of PLG or a constant amount of PLG (100 nM) with increasing amounts of NS1_e was added to immobilized VN. NS1 binding to VN was affected in the presence of PLG and was reduced ∼80% at a 10-fold molar excess (Fig. 3D). Similarly, the presence of increasing concentrations of NS1 affected the binding of PLG to immobilized VN (Fig. 3E). Taken together, these results indicate that NS1 and PLG have overlapping binding sites on VN.

Identification of the NS1-VN complex in plasmas from DENV-infected patients.

During DENV infection, the NS1 protein is found at high concentrations in the sera of infected patients (31, 32). Because alterations in vascular homeostasis are the basis of dengue pathophysiology and severity, we evaluated whether NS1 interacts with circulating VN in the plasma in DENV-infected patients. A capture ELISA probed with an anti-VN monoclonal antibody was performed using a plasma panel composed of 14 samples from DENV-infected patients (Table 1) and 40 samples from healthy blood donors (HD). All DENV-positive samples were from DENV1 infections and were NS1 positive according to the Platelia Dengue NS1 Ag assay. The captured NS1-VN complex was significantly increased in DENV-positive samples compared to HD samples (P < 0.001) (Fig. 4A), clearly suggesting that this interaction occurs during infection in the plasma. We then evaluated whether this complex formation correlates with disease severity. DENV-positive samples were divided according to the new dengue classification guidelines (2), into samples from cases of dengue without warning signs (DWO; n = 8) and those from cases of dengue with warning signs (DW; n = 6). Although both groups showed higher mean OD values than those for HD (P < 0.001 [for each group]) (Fig. 4B), we did not observe a significant difference in these values between DWO and DW samples (P = 0.439) (Fig. 4B), indicating that the presence of the NS1-VN complex during infection does not correlate with disease progression. Nonetheless, we cannot exclude the possibility that the lack of a significant difference between dengue patients with different severities of disease may have been due to the small sample size.

DENV NS1 binds to MAC components and inhibits its formation.

To assess whether the NS1-VN interaction would contribute to subverting the complement attack as described for bacterial pathogens (28, 33), we evaluated whether this complex blocks MAC formation by using purified MAC components. NS1_e was preincubated with VN, mixed with purified C7, C8, and C9, and then added to C5b6-treated sheep erythrocytes. In the presence of both NS1 and VN, erythrocyte lysis was significantly inhibited, whereas the NS1 buffer did not block the MAC cytolytic activity (Fig. 5A). VN inhibited MAC formation, as expected (17). Interestingly, recombinant NS1_e also inhibited MAC activity, by 50%. However, erythrocyte lysis inhibition was more pronounced in the presence of both NS1_e and VN (P < 0.05) (Fig. 5A), suggesting that the NS1-VN interaction is important for avoiding MAC activation.

To investigate whether NS1 glycosylation and oligomerization have similar inhibitory effects on MAC formation, purified NS1_e expressed in E. coli, which is in its dimeric conformation and is not glycosylated, and purified NS1 expressed in Drosophila S2 cells (NS1_dros), which is in its hexameric conformation and is polymannose glycosylated, were incubated with purified C7, C8, and C9 and then added to C5b6-treated sheep erythrocytes. Both proteins significantly inhibited erythrocyte lysis in a dose-dependent manner, whereas factor H and NS1 buffer, included as negative controls, did not inhibit MAC-dependent lysis (Fig. 5B). Therefore, it seems that NS1 glycosylation and oligomerization are not required for the protein to inhibit MAC activity.

We also evaluated whether NS1 interacts with purified MAC components. Microtiter plates coated with NS1_e were incubated separately with purified C5, C6, C7, C8, and C9. NS1 strongly bound to C5, C6, and C9, whereas a weaker interaction was observed between NS1 and C7. Furthermore, no interaction was detected with C8. C2 was included as a negative control because it does not interact with NS1 (Fig. 5C). From these data, it is plausible to conclude that NS1 may inhibit MAC formation through direct interaction with MAC components.

The DENV2 NS1-VN complex inhibits C9 polymerization in vitro.

Because DENV NS1 interacts with C9, we next investigated the effects of NS1_e and DENV2 NS1_dros on C9 polymerization. Both NS1_e (Fig. 6A) and NS1_dros (Fig. 6B) inhibited C9 polymerization in a concentration-dependent manner, indicating that the NS1-C9 complex affects MAC formation. We also evaluated whether the NS1-VN complex affects C9 polymerization. A more pronounced inhibition was observed in the presence of both NS1_e and VN than that with NS1_e or VN alone, suggesting that this interaction enhances the inhibition of C9 polymerization (Fig. 6C and D). We also performed a competitive ELISA using a constant amount of NS1_e in the presence of increasing amounts of VN. NS1 binding to C9 was not affected by VN even at a 10-fold molar excess (Fig. 6E), thus indicating that NS1 and VN do not compete for the same binding sites on C9. Taken together, our data strongly suggest that NS1 and VN bind C9 simultaneously and interfere with its polymerization.

Flavivirus NS1 binds C9 and inhibits C9 polymerization in serum.

To address whether NS1 secreted by mammalian cells (NS1_mam) is capable of inhibiting C9 polymerization, we used recombinant NS1_mam proteins from DENV2, WNV, and ZIKV expressed in HEK 293 cells. Initially, BHK-21 cells were preincubated with C5b6, and DENV2, WNV, or ZIKV NS1_mam was preincubated with 5% NHS in a separate preparation. C5b6-treated BHK-21 cells were then incubated with NS1_mam plus 5% NHS, washed, and lysed. Samples were analyzed by Western blotting with anti-C9 antibody. As the NS1_mam concentration increased, the band intensity corresponding to the C9 polymer decreased (Fig. 7A) for all tested flavivirus proteins. Surprisingly, the C9 polymerization showed the largest inhibitory effect when cells were incubated with the ZIKV NS1_mam protein. When anti-NS1 was preincubated with NS1_mam at 50 μg/ml, the inhibition of C9 polymerization was abolished, thus confirming a specific role of NS1 in inhibition of C9 polymerization.

Because ZIKV NS1_mam inhibited C9 polymerization more strongly than the DENV2 and WNV NS1_mam proteins did, we assessed binding of purified C9 to immobilized NS1 proteins. As shown in Fig. 7B, the DENV2 and WNV NS1_mam proteins bound C9 similarly, whereas ZIKV NS1_mam bound C9 more avidly, suggesting that the pronounced effect of ZIKV NS1 on C9 polymerization may be explained by its binding to C9. In addition, preincubation of flavivirus NS1_mam (25 μg/ml) with increasing concentrations of VN was more effective at preventing C9 polymerization in serum (Fig. 7C).