Fig 1.
DENV2 but not WNV NS1 binding to endothelial cells induces endothelial barrier dysfunction.
(A-B) Binding of DENV2 NS1 (A) and WNV NS1 (B) proteins, examined by confocal microscopy. NS1 is stained in green, and nuclei are stained with Hoechst (blue). Images (20X) are representative of three individual experiments. Scale bars, 10 μM. A trace of the mean fluorescence intensity (MFI) of one representative field is shown below each image. (C) The amount of NS1 bound to HPMEC monolayers in (A) and (B) is quantified and expressed as MFI. Graph shows average of quantification from three independent experiments. DENV2 NS1 binding to HPMEC monolayers is significantly higher than WNV NS1 binding at all concentrations (p<0.0001). (D) Experimental schematic of trans-endothelial electrical resistance (TEER) experiments. (E) TEER assay to evaluate the effect of DENV2 and WNV NS1 proteins on HPMEC endothelial permeability at indicated concentrations. Relative TEER values from three independent experiments performed in duplicate are plotted at the indicated time points. Error bars indicate standard error of the mean (SEM). All DENV2 NS1 concentrations induce statistically significant decreases in TEER (p<0.0001).
Fig 2.
DENV2 but not WNV NS1 modulates the expression of Sia in the EGL of HPMEC.
(A) Sia expression on HPMEC monolayers after treatment with DENV2 or WNV NS1 (5 μg/ml), examined by confocal microscopy. Sia was stained with WGA-A647 (red) at indicated time points (hpt). Nuclei stained with Hoechst (blue). Images (20X) and MFI values are representative of three independent experiments. Scale bar, 10 μM. (B) Quantification of MFI in (A) from three independent experiments. Sia expression in DENV2 NS1-treated monolayers is significantly different from WNV-treated and untreated controls from 0.5 to 12 hpt (p<0.0001). (C) DENV2 NS1 (green) induces dose-dependent reduction of Sia staining (red) on HPMEC after 3 hpt. Untreated cells were used as a positive control for Sia expression. See also S1 Fig. (D) ELISA to detect free Sia released into culture supernatant of HPMEC over time (hpt) under indicated experimental conditions. DENV2 NS1 shedding of Sia is significantly lower than WNV-treated and untreated controls at 6 and 12 hpt (p<0.0005). (E) Endothelial sialidase (Neu1, Neu2, Neu3) expression in HPMEC monolayers after treatment with DENV2 or WNV NS1 (5 μg/ml) at 3 hpt. Neu1, Neu2, and Neu3 are stained in green, red, and yellow, respectively. (F) Effect of Zanamivir and 2,3-didehydro-2-deoxy-N-acetylneuraminic acid (DANA) on DENV2 NS1-mediated endothelial hyperpermeability (TEER) in HPMEC monolayers. Relative TEER values from three independent experiments performed in duplicate are plotted at indicated time points (hpt). TEER values for monolayers treated with DENV2 NS1 combined with Zanamivir (50, 100 μM) or DANA (50 μg/ml) are significantly different from monolayers treated with DENV2 NS1 alone (p<0.0001). Error bars indicate SEM throughout.
Fig 3.
DENV2 NS1 increases the surface staining of syndecan-1 in the EGL of HPMEC.
(A) Staining of syndecan-1 (green) on the surface of HPMEC monolayers over time (hpt) after treatment with DENV2 or WNV NS1 proteins (5 μg/ml), examined by confocal microscopy. Untreated cells were used as a control for basal syndecan-1 detection. Nuclei are stained with Hoechst (blue). Images are representative of three individual experiments (20X). Scale bar, 10 μM. A trace of the MFI for DENV2 NS1 of one representative field is shown below each image. See also S4A Fig. (B) Expression of syndecan-1 in total protein extracts at indicated time points (hpt) in the presence of DENV2 or WNV NS1 proteins (5 μg/ml), as measured by Western blot. Ten μg/ml of total protein was loaded, and GAPDH expression was used as protein loading control. (C) Densitometry data normalized to control untreated cells. (D) Levels of syndecan-1 shed from the surface of HPMEC after treatment with DENV2 or WNV NS1 proteins (5 μg/ml) as measured by ELISA from three independent experiments. Surface staining of syndecan-1 is significantly higher in monolayers treated with DENV2 NS1 than in WNV-treated and untreated controls (p<0.0005). (E) Effect of recombinant syndecan-1 on TEER of HPMEC monolayers. Relative TEER values from three independent experiments performed in duplicate are plotted at indicated time points. All concentrations of syndecan-1 induce significant decreases in TEER value (10 ng/ml: p <0.05; 20 ng/ml, 50 ng/ml: p<0.0001). Error bars indicate SEM throughout.
Fig 4.
DENV2 NS1 triggers the activation/increased expression of endothelial heparanase.
(A) Heparanase expression (red) in HPMEC monolayers over time (hpt) after treatment with DENV2 or WNV NS1 proteins (5 μg/ml), examined by confocal microscopy. Untreated cells were used as a control for basal heparanase expression. Nuclei stained with Hoechst (blue). Images are representative of three individual experiments (20X). Scale bar, 10 μM. A trace of the mean fluorescence intensity (MFI) for DENV2 NS1 of one representative field is shown below each image. (B) Quantification of MFI in (A) from three independent experiments. Staining of heparanase in monolayers treated with DENV2 NS1 is significantly different from WNV-treated and untreated controls from 0.5–12 hpt (p<0.0001). (C) Heparanase activation in HPMEC monolayers at indicated time points (hpt) after interaction with DENV2 and WNV NS1 proteins (5 μg/ml) via Western blot. Ten μg/ml of total protein was loaded, and GAPDH expression was used as a protein loading control. Upper (~60 kDa) and lower bands (~50 kDa) correspond to inactive and active forms of human heparanase I, respectively. (D) Densitometry of Western blot from (C). Graph shows lower band densitometries corresponding to active heparanase (~50 kDa), normalized to GAPDH at each time-point. Heparanase activation is significantly higher in monolayers treated with DENV2 NS1 than in WNV-treated and untreated controls from 0.5–12 hpt (p<0.0001). (E) Co-staining of human heparanase I (red) and syndecan-1 (green) in HPMEC treated with DENV2 NS1 (5 μg/ml) after 30 min, 1 and 3 hpt. Manders’ Overlapping Coefficient (MOC) value for the overlapping fraction (merge) is listed to the right of each time point.
Fig 5.
DENV2 NS1 increases the activity of cathepsin L protease.
(A) Cathepsin L proteolytic activity (Magic Red assay, in red) in HPMEC monolayers over time (hpt) after treatment with DENV2 or WNV NS1 (5 μg/ml) and cathepsin L or B inhibitors (10 μM). Nuclei are stained with Hoechst (blue). Untreated cells were used as control for basal cathepsin L expression. Images are representative of three individual experiments (20X). Scale bars, 10 μM. A trace of the MFI for DENV2 NS1 of one representative field is shown below each image. (B) Quantification of MFI in (A) from three independent experiments. Staining for cathepsin L activity is significantly higher in monolayers treated with DENV2 NS1 than in WNV-treated and untreated controls at 0.5–12 hpt (p<0.0001). Staining for cathepsin L is significantly lower in monolayers treated with DENV2 NS1 and cathepsin L inhibitor compared to monolayers treated with DENV2 NS1 alone at 0.5–12 hpt (p<0.0001). Error bars indicate SEM throughout.
Fig 6.
DENV2 NS1-mediated endothelial dysfunction is blocked by inhibition of heparanase activation and cathepsin L activity.
(A) Effect of cathepsin L inhibitor (10 μM) on DENV2 NS1-mediated disruption of the EGL components Sia (red, upper panels) and syndecan-1 (green, middle panels), and on DENV2 NS1-induced increase of heparanase expression (red, lower panels). Nuclei stained with Hoechst (blue). Untreated cells were used as a control for basal expression of EGL components and heparanase. Cathepsin B inhibitor (10 μM) is used as a negative control. Images are representative of three individual experiments (20X). Scale bar, 10 μM. (B) Effect of cathepsin L inhibitor and OGT 2115 (human heparanase I inhibitor) on DENV2 NS1-induced release of Sia into the supernatant, as measured by ELISA from three independent experiments. Shedding of Sia is significantly different in monolayers treated with DENV2 NS1 and cathepsin L inhibitor or OGT 2115 than in monolayers treated with DENV2 NS1 alone at 6 and 12 hpt (p<0.0001). (C) Effect of cathepsin L inhibitor on DENV2 NS1-induced shedding of syndecan-1, as measured by ELISA from three independent experiments. Shedding of syndecan-1 is significantly lower in monolayers treated with DENV2 NS1 and cathepsin L inhibitor compared to monolayers treated with DENV2 NS1 alone at 1–12 hpt (p<0.0001). (D) Effect of OGT 2115 (heparanase I inhibitor) on DENV2 NS1-triggered endothelial hyperpermeability (TEER) in HPMEC monolayers. TEER values of monolayers treated with DENV2 NS1 and OGT 2115 (0.5, 1.0 μM) are significantly different than values of monolayers treated with DENV2 NS1 alone (p<0.0001). (E) Effect of cathepsin L inhibitor on DENV2 NS1-induced endothelial hyperpermeability (TEER) in HPMEC monolayers. TEER values of monolayers treated with DENV2 NS1 and cathepsin L inhibitor (5, 10 μM) are significantly different than values of monolayers treated with DENV2 NS1 alone (p<0.0001). In (C) and (D), relative TEER values from three independent experiments performed in duplicate are plotted at indicated time points. In (A) and (C), Cathepsin B inhibitor (CA704, 10 μM) was included as a control for specific cysteine protease inhibition. Error bars indicate SEM throughout.
Fig 7.
DENV NS1-induced effects on the EGL leading to endothelial hyperpermeability are similar across all four DENV serotypes.
(A) TEER assay to evaluate the effect of NS1 proteins from DENV1-4 and WNV (5 μg/ml) on HPMEC endothelial permeability. Relative TEER values from three independent experiments performed in duplicate are plotted at the indicated time points. Error bars indicate standard error of the mean (SEM). NS1 from all DENV serotypes induces statistically significant decreases in TEER (p<0.0001), while NS1 from WNV does not. (B) Quantification of MFI in S9 Fig from three independent experiments. Sia expression in DENV1 and DENV2 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 1 to 12 hpt (p<0.0001), and Sia expression in DENV3 and DENV4 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 3–12 hpt (p<0.0001). (C) Quantification of MFI in S10 Fig from three independent experiments. Neu1 expression in DENV1, 2, 3, and 4 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 1 to 12 hpt (p<0.0001). (D) Quantification of MFI in S11 Fig from three independent experiments. Neu2 expression in DENV1, 2, and 4 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 1 to 12 hpt (p<0.0001), and Neu2 expression in DENV3 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 3–12 hpt (p<0.0001). (E) Quantification of MFI in S12 Fig from three independent experiments. Neu3 expression in DENV1, 2, and 3 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 1 to 12 hpt (p<0.0001), and Neu3 expression in DENV4 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 6–12 hpt (p<0.0001). (F) Quantification of MFI in S13 Fig from three independent experiments. Syndecan-1 expression in DENV1 and 2 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 1 to 12 hpt (p<0.0001); Syndecan-1 expression in DENV3 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 3–12 hpt (p<0.0001); Syndecan-1 expression in DENV4 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 3–6 hpt (p<0.0001). (G) Quantification of MFI in S14 Fig from three independent experiments. Heparanase expression in DENV1 and 2 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 1 to 12 hpt (p<0.0001), and heparanase expression in DENV3 and 4 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 3–12 hpt (p<0.0001). (H) Quantification of MFI in S15 Fig from three independent experiments. Cathepsin L activity in DENV1, 2, 3, and 4 NS1-treated monolayers is significantly different from WNV NS1-treated and untreated controls from 1 to 12 hpt (p<0.0001).
Fig 8.
Model of DENV NS1-induced endothelial hyperpermeability of human pulmonary microvascular endothelial cells (HPMEC).
(1) The endothelial glycocalyx layer (EGL), a network of membrane-bound proteoglycans and glycoproteins, lines the endothelium on the luminal side. Both endothelium- and plasma-derived soluble molecules interact with this mesh. (2) Hexameric NS1 protein secreted from flavivirus-infected cells binds to the surface of uninfected cells, including human pulmonary microvascular endothelial cells (HPMEC), upregulating the expression of lysosomal, cytosolic, and cell membrane endothelial sialidases (3,4) that translocate to the cell membrane, initiating trimming of terminal sialic acid residues expressed on EGL (5). In addition, DENV NS1 enhances the expression of the inactive precursor of the endoglycosidase (pro-heparanase) (6a) and the activity of the lysosomal cysteine protease cathepsin L (6b), potentially through internalization of NS1 (6c). Cathepsin L processes pro-heparanase into an active/mature form (7,8), leading to cleavage of heparan sulfate chains on the EGL (9,10). This results in shedding of syndecan-1, a main component of the EGL (11,12), and its accumulation after binding back to the cell surface (13). Additionally, DENV NS1 may trigger TLR4 signaling (14a), leading to the translocation of Neu1 to the cell membrane and further disruption of sialic acid in the EGL (14b). Together, these processes lead to EGL disruption on the surface of endothelial cells, resulting in endothelial barrier dysfunction and fluid extravasation (hyperpermeability) that occurs in severe dengue disease (15).