Dengue virus-elicited tryptase induces endothelial permeability and shock

Abstract

Dengue virus (DENV) infection causes a characteristic pathology in humans involving dysregulation of the vascular system. In some patients with dengue hemorrhagic fever (DHF), vascular pathology can become severe, resulting in extensive microvascular permeability and plasma leakage into tissues and organs. Mast cells (MCs), which line blood vessels and regulate vascular function, are able to detect DENV in vivo and promote vascular leakage. Here, we identified that a MC-derived protease, tryptase, is consequential for promoting vascular permeability during DENV infection, through inducing breakdown of endothelial cell tight junctions. Injected tryptase alone was sufficient to induce plasma loss from the circulation and hypovolemic shock in animals. A potent tryptase inhibitor, nafamostat mesylate, blocked DENV-induced vascular leakage in vivo. Importantly, in two independent human dengue cohorts, tryptase levels correlated with the grade of DHF severity. This study defines an immune mechanism by which DENV can induce vascular pathology and shock.

Keywords: Infectious disease; Mast cells; Tight junctions; Vascular Biology.

Conflict of interest statement

Conflict of interest: The authors have declared that no conflict of interest exists.

Figures

Figure 1. Tryptase and chymase break tight junctions to disrupt endothelial cell contact sites.

(A) DENV-stimulated MC supernatant was transferred onto huMEC monolayers or separated into soluble and particulate fractions, followed by transfer onto huMEC monolayers. For controls, DENV-alone or media-alone groups were used. TER of huMEC monolayers was measured 24 hours after treatment. Whole supernatant or isolated MC particles each significantly reduced the TER of huMEC monolayers (P < 0.05 by 1-way ANOVA with Dunnett’s post test). (B and C) TER of huMEC monolayers after treatment for 24 hours with either purified, recombinant tryptase (B) or chymase (C). *P < 0.05, decrease in TER over controls by 1-way ANOVA with Dunnett’s post test (A–C). (D–F) huMECs were treated with low (0.1 μM) or high (1 μM) concentrations of either tryptase or chymase for 24 hours, followed by fixation and staining against tubulin (green), nuclei using DAPI (blue), and tight junctions (ZO-1, red). (D) In control cells, tight junctions were intact in between cells, visualized by ZO-1 staining. (E) Tryptase induced a concentration-dependent reduction in ZO-1 staining that appeared disjunctive at low concentrations and absent at high concentrations. Lifting of cells forming gaps was also observed after high-concentration tryptase treatment. (F) Low-concentration chymase had no apparent effect on tight junctions, while staining grew more punctate at high concentrations. For A–F, data are representative of 3 independent repeats. Scale bars: 25 μm. (G and H) Levels of CD31 on endothelial cells were measured by flow cytometry on cells isolated from mouse footpads 6 hours after injection of 100 ng of tryptase, chymase, or saline vehicle control. (G) CD31+ cells showed reduced levels of staining after injection of tryptase (representative histogram plots). (H) Comparison of MFI of CD31 staining in mouse footpads (n = 5–6 each group) showed that tryptase, but not chymase, is sufficient to induce a significant decrease in CD31 staining in vivo (right panel, P < 0.05, 1-ANOVA with Dunnett’s post test). For graphs, error bars represent SEM.

Figure 2. MC proteases promote vascular leakage and shock in vivo.

(A) Hematocrit values were obtained 6 hours after injection with saline alone or 30 ng of either tryptase, chymase, or OVA. Means differ significantly by 1-way ANOVA (P < 0.0001). Bonferroni’s multiple comparison test was used to determine significance among groups. Control, n = 15; tryptase and chymase, n = 10; OVA, n = 5. Data were added from 2 independent experiments. (B and C) Mice (n = 3–4) were injected with 30 ng each of tryptase, chymase, or OVA i.v., or an equivalent volume of saline was injected for controls. To measure shock, the body temperature of animals was recorded every 5 minutes for the first 15 minutes and subsequently at 10-minute intervals. (B) Both tryptase and chymase caused sudden drops in body temperature, indicative of shock, compared with both OVA and saline control groups. Data were analyzed by 2-way ANOVA with Holm-Šidák multiple comparison test to compare temperatures at each time point. (C) The maximal difference in temperature during the time course is presented, suggesting that tryptase treatment causes significantly higher plasma loss in animals compared with chymase, OVA, and saline control groups, determined by 1-way ANOVA with Holm-Šidák multiple comparison test. Data are shown as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Figure 3. Tryptase inhibition therapeutically blocks vascular leakage during DENV infection.

Mice (n = 3–5 per group) were either mock infected or infected with DENV (1 × 106 PFU), followed by treatment with vehicle control (saline) or using a specific tryptase inhibitor, nafamostat mesylate, at a dose of 0.06 mg/kg. (A) Hematocrit analysis was performed using an automated hematology analyzer on whole blood at 24 hours after treatment. (B) Serum was isolated to measure tryptase activity by enzymatic assay. Only DENV-infected mice that were vehicle treated had elevated tryptase activity over uninfected control group. Nafamostat mesylate treatment reversed tryptase activity to baseline levels. (C) Platelet counts at 24 hours are presented. The data show a strong reduction in DENV-induced vascular leakage upon treatment with tryptase inhibitor, nafamostat mesylate, but no significant difference in platelet counts compared with DENV-infected and vehicle-treated mice. For A–C, statistical significance was determined using 1-way ANOVA with Bonferroni’s multiple comparison test. (D) No difference in the DENV burden in the spleen determined by real-time reverse-transcription PCR (RT-PCR) was observed between vehicle- and nafamostat mesylate–treated animals at 72 hours after infection by Student’s unpaired t test. For all panels, data are presented as mean ± SEM. *P < 0.05; ****P < 0.0001.

Figure 4. Improvement of vascular leakage in severe DENV infection models.

AG129 mice were infected with a low (1 × 106 PFU) or high (5 × 107 PFU) dose of DENV and either mock or nafamostat mesylate treated beginning 1 hour after infection. Uninfected, n = 3; DENV infected, vehicle treated, n = 4–5; and DENV infected, nafamostat mesylate treated, n = 4–5. (A) At 24 hours after infection, hematocrit was measured. A single treatment of nafamostat mesylate reversed DENV-induced vascular leakage for both infection doses. (B and C) Nafamostat mesylate effectively restored hematocrit values to baseline levels at days 2 and 3 after infection in an antibody-enhanced DENV mouse model. Treatment was initiated (B) 1 hour or delayed (C) 24 hours after infection and given at 24-hour intervals thereafter. Statistical significance was calculated using 1-way ANOVA with Bonferroni’s multiple comparison test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. For all panels, data are presented as mean ± SEM.

Figure 5. Visualization of inhibition of plasma leakage in DENV-infected mice.

AG129 mice were infected with 1 × 106 PFU of DENV i.p. and either treated with nafamostat mesylate or with vehicle. Mice (n = 3), 24 hours after infection, were injected with 70 kDa FITC-dextran. Two-photon images were acquired continuously at 2-second intervals, beginning 5 minutes after injection, for 18 minutes total. (A) Representative images from the indicated time points after FITC-dextran injection from Supplemental Videos 1–3, showing vascular leakage in the DENV-infected mock-treated ear, while control mice and DENV-infected mice treated with nafamostat mesylate showed no visually discernible vascular leakage. Scale bars: 50 μm (B) MFI in the acquired images over time is presented. Intensity was measured by averaging 10 areas in the interstitial space. Data are representative of 3 independent experiments.

Figure 6. Serum tryptase levels are correlated with DHF/DSS in humans.

Serum samples from virologically confirmed hospitalized dengue patients in Jakarta, Indonesia, in 1975–1978 (n = 9 DF; n = 25 DHF) were tested retrospectively in a blinded manner for (A) chymase and (B) tryptase. Both were significantly increased (P < 0.0001 and P = 0.0004, respectively, determined by Student’s unpaired t test) in patients with DHF compared with DF. (C) Mean serum tryptase levels for each grade of DHF were also strongly correlated with the grade of DHF based on the patient’s reported symptoms. P = 0.05; R2 = 0.89. (D and E) Tryptase was measured in a second cohort of prospectively obtained patient samples from virologically confirmed dengue patients (n = 30 DF; n = 25 DHF) in Sri Lanka, Colombo, in 2012–2013. (D) Serum tryptase levels were significantly elevated in serum samples from DHF versus DF patients (P = 0.0005). (E) Mean serum tryptase levels in Sri Lankan samples were strongly correlated with disease severity. P = 0.02; R2 = 0.86. For all panels, data are presented as mean ± SEM.