Exposure of Platelets to Dengue Virus and Envelope Protein Domain III Induces Nlrp3 Inflammasome-Dependent Platelet Cell Death and Thrombocytopenia in Mice

In tropical and subtropical regions, mosquito-borne dengue virus (DENV) infections can lead to severe dengue, also known as dengue hemorrhage fever, which causes bleeding, thrombocytopenia, and blood plasma leakage and increases mortality. Although DENV-induced platelet cell death was linked to disease severity, the role of responsible viral factors and the elicitation mechanism of abnormal platelet activation and cell death remain unclear. DENV and virion-surface envelope protein domain III (EIII), a cellular binding moiety of the virus particle, highly increase during the viremia stage. Our previous report suggested that exposure to such viremia EIII levels can lead to cell death of endothelial cells, neutrophils, and megakaryocytes. Here we found that both DENV and EIII could induce abnormal platelet activation and predominantly necrotic cell death pyroptosis. Blockages of EIII-induced platelet signaling using the competitive inhibitor chondroitin sulfate B or selective Nlrp3 inflammasome inhibitors OLT1177 and Z-WHED-FMK markedly ameliorated DENV- and EIII-induced thrombocytopenia, platelet activation, and cell death. These results suggest that EIII could be considered as a virulence factor of DENV, and that Nlrp3 inflammasome is a feasible target for developing therapeutic approaches against dengue-induced platelet defects.


INTRODUCTION
Dengue, caused by dengue virus (DENV), is a commonly observed mosquito-borne infectious disease in the tropical and subtropical areas of the world (1,2). Currently, no specific antiviral treatment and reliable vaccine are available against dengue (3)(4)(5)(6)(7)(8). Over half of the global population lives in areas with DENV infection risk, resulting in 96 million patients with symptomatic dengue every year (9). The primary infection of DENV is a self-limiting dengue fever (DF), whereas secondary infections are epidemiologically associated with an increased risk of life-threatening severe dengue [also known as dengue hemorrhage fever (DHF)]. Thrombocytopenia and platelet defects are the frequently observed symptoms in both primary and secondary dengue infection; however, the responsible pathogenic factor remains unclear. Here we hypothesize that the DENV envelope protein (E) domain III (EIII) can be one of the viral factors responsible for pathological changes in platelets.
The DENV EIII is an Ig-like domain involved in host cell receptor binding for viral entry (10). However, whether cellular signaling occurs after EIII binds to target cells remains unclear. Pieces of evidence have shown that DENV-induces activation of inflammasome complexes (11,12). Our previous study revealed that challenges with the DHF-viral-load-equivalent levels of EIII can induce cell death in endothelial cells (13) and neutrophils (14), and suppress megakaryopoiesis through autophagy impairment and apoptosis (15). These results suggest that EIII binding to platelets could be cytotoxic, particularly when circulating virionsurface EIII reaches the peak of the viremia stage. In addition, because gene deficiencies in Nlrp3 inflammasome components display protective effects on the hemorrhage fever mouse model (13,16), we further investigated whether Nlrp3 inflammasome contributes to EIII-induced platelet defects. In this study, EIII treatments directly caused the activation and death of platelets in vitro and thrombocytopenia in mice. Intriguingly, pyroptosis is the major regulated cell death (RCD) pathway of DENV-and EIIItreated platelets. In addition, treatment with the Nlrp3 inflammasome selective inhibitors OLT1177 and Z-WHED-FMK markedly ameliorated DENV-and EIII-induced platelet defects in vitro and in vivo, suggesting that the Nlrp3 inflammasome can be a feasible target for treating dengue-associated thrombocytopenia and platelet defects. Related implications and applications are discussed.

Ethics Statement
Animal experiments in this study were conducted in agreement with the National (Taiwan Animal Protection Act, 2008) directive for the protection of laboratory animals. All experimental protocols for examining experimental animals were approved by the Animal Care and Use Committee of Tzu-Chi University, Hualien, Taiwan (approval ID: 101019).

Blood and Platelet Isolation and Parameter Analyses
Mouse blood samples were collected through the retro-orbital venous plexus technique by using plain capillary tubes (Thermo Fisher Scientific, Waltham, MA, USA) and were transferred into polypropylene tubes containing anticoagulant acid-citrate-dextrose solution (38 mM citric acid, 75 mM sodium citrate, and 100 mM dextrose) (29,30). Washed platelets were prepared as previously described (31,32). Platelet-poor plasma was prepared through centrifugation at 1,500 × g for 20 min. Platelet-poor plasma was centrifuged again for 3 min at 15,000 × g to remove contaminant cells from the plasma. According to manufacturer's instructions, the activated partial thromboplastin time (aPTT) experiment was performed using a coagulometer (ACL-Futura Plus, Instrumentation Laboratory, Milan, Italy; (29,33). The red blood cell and platelet counts of mice were measured using a hematology analyzer (KX-21N; Sysmex, Kobe, Japan) (26).

Analyses of Platelet Response on Viral-Protein Exposure
To analyze the binding properties of rEIII proteins on platelets, rEIII proteins (laboratory-prepared 20-kDa domain III, 45-kDa domains I + II + III, 32-kDa domains I + II, and 25-kDa domains II cterminal + III purchased from ProSpec-Tany TechnoGene) were conjugated with biotin by using an EZ-Link ™ Sulfo-NHS-Biotinylation kit (Thermo Fisher Scientific). The levels of biotinlabeled rEIII proteins bound to mouse platelets (50 mg/ml protein + 1 × 10 7 platelets/ml in Tyrodes buffer for 30 min) were determined through flow cytometry by using PE/Cy5 avidin (Biolegend, San Diego, CA, USA) staining. The rEIII-competitive inhibitor chondroitin sulfate B (CSB, 10 µg/ml; Sigma-Aldrich, St. Louis, MO, USA) was used to suppress rEIII-induced platelet activation. The fluorescence intensities of rEIII-bound platelets were analyzed using a fluorescent microplate reader (Varioskan ™ Flash Multimode reader; Thermo Fisher Scientific) (26,29) at Excitation/Emission spectrum = 480/530 nm. A flow cytometer (FACScalibur, BD Biosciences) (29,30) was used for analyzing platelet surface P-selectin expression and biotin-rEIII binding by using anti-mouse P-selectin Ig-phycoerythrin (PE; eBioscience, Thermo Fisher Scientific) and biotin-conjugated rEIII, followed by avidin-PE/Cy5 (Biolegend) labeling. Scanning electron microscopy analysis for platelet morphology was conducted as previously described (31). The washed mouse platelets were incubated for 15 min under coverslips coated with various amounts of rEIII or rGST. After washing with phosphate-buffered saline (PBS), these cells were fixed with glutaraldehyde and subjected to a series of alcohol dehydrations, critical point drying procedures, and gold coating. The platelet morphology of the cells was then observed under a scanning electron microscope at 15 kV (Hitachi S-4700, Hitachi, Japan). Morphological scoring and quantification of platelet activation levels were conducted using previously described methods (34). Six different areas were randomly selected for photography at each magnification for quantitative analyses. Mouse bleeding time was measured 24 h after rEIII injections with or without CSB rescues (0.5 mg/kg).

Platelet Cell Death and Mitochondrial Analyses
To analyze DENV-or rEIII-induced platelet cell death, washed mouse platelets were incubated with DENV or rEIII for 1 h and then subjected to flow cytometry analyses after washing with PBS. Various RCD responses, namely including apoptosis

Western Blotting
Western blotting analysis was performed following previously described methods (28,(35)(36)(37)(38). A mini trans-blot electrophoretic transfer cell, a vertical electrophoresis gel tank, and a power pac 200 electrophoresis power supply (Bio-Rad Laboratories, Hercules, CA, USA) were used to separate and transfer the platelet protein samples to a polyvinylidene fluoride membrane (Immobilon-P, 0.45 µm, Merck Millipore, Burlington, MA, USA). Pre-stained protein ladders (PageRuler, Thermo Fisher Scientific) and an antibody against GSDMD (Abcam) were used to mark the molecular weight position and to stain the GSDMD, respectively. The intensities of Western-blotting image were measured using ImageJ software (version 1.32; National Institutes of Health, USA).

Statistical Analyses
Using SigmaPlot 10 (Systat Software, San Jose, CA, USA) and SPSS 17 software packages (International Business Machines Corporation, Armonk, NY, USA), the means, standard deviations, and statistics of quantifiable data presented in this report were calculated. Using one-way analysis of variance and the post-hoc Bonferroni-corrected t test, the significance of data was examined. The probability of type-1 error a = 0.05 was set as the threshold of statistical significance.

Soluble Recombinant EIII Protein Binds to Platelets and Induces Coagulation Defects, Platelet Activation, and Thrombocytopenia in Mice
Soluble recombinant EIII (rEIII; Figure S1) can bind to and suppress the function of megakaryocytes (15). As the precursor cells of platelets, megakaryocytes express various platelet surface proteins (39). In this study, flow cytometry analysis results revealed that biotinylated DENV and various recombinant E fragments (domains I + II, II + III, and full length) could bind to platelets (Figure 1 and Figure S2). Probably because of the protein folding structure, among all rE protein fragments, rEIII displayed the strongest platelet-binding property (Figure 1). The binding of both DENV and rEIII to platelets could be suppressed by neutralizing Igs against rEIII ( Figure S3), indicating that rEIII and DENV share common and functional platelet-binding epitopes. Recombinant EIII bound to mouse platelets in a dosedependent manner (Figures S4A-D). Compared with the negative control recombinant proteins (glutathione-S transferase and isotype control Ig), rEIII induced markedly higher levels of surface expression of P-selectin ( Figures S4E-H, S5), which is a marker of platelet activation (18,40). In addition, platelets preferentially bound to rEIII-coated but not rGST-coated cover slides, with higher percentages of activation-associated morphologies ( Figures S4I-P). Mouse experiments further revealed that similar to challenges with DENV, injections of rEIII but not those of rGST induced thrombocytopenia and shortened plasma clotting time in vivo ( Figure 2). These results suggest that exposure of DENV and virion-surface EIII to platelets and plasma is sufficient, leading to an abnormal coagulation response.

Treatment With rEIII Abnormally Affect Plasma Clotting Time: Implications on Hemorrhage
In addition to platelets, rEIII can affect plasma clotting time because rEIII is a glycosaminoglycan (GAG)-binding protein.
Experiments on aPTT revealed that because of the neutralizing effect on heparin, the addition of the heparin-binding proteins rEIII and protamine but not that of the control proteins, rGST and bovine serum albumin shortened heparinized aPTT plasma clotting time (Figures S6A, B). In addition to heparin, EIII binds to other GAGs (10), which may be used as antidotes to suppress any interaction between rEIII and heparin or the cell surface glycoprotein (10). Using this GAG binding property of EIII, here we developed aPTT-based assays for (I) screening selective rEIIIbinding GAGs and (II) determining viremia equivalent doses of EIII. These two technologies (I and II) are useful for the EIII characterizations, while not relevant to the dengue pathogenesis. First, among the tested GAGs, we found that CSB most effectively blocked rEIII-heparin and rEIII-platelet binding ( Figures S6C, D, S7). Second, based on this aPTT-based functional analysis, we measured and compared the heparinbinding properties of DENV and rEIII ( Figure S8). For example, 0.3-1.2 mM of rEIII, cell-activating and cytotoxic dose for platelets (Figures 3-6) was estimated to be functionally equivalent to a virion titer within a previously published DHF viral load (41) range ( Figure S8). Mouse experiments further revealed that CSB treatment not only inhibited rEIII-platelet binding in vitro but also ameliorated rEIII-induced thrombocytopenia and rEIII-prolonged bleeding time in mice ( Figure S6).

Pyroptosis Is a Major Cell Death Pathway of Platelets Treated With DENV and rEIII
After the characterization of EIII-mediated effects on platelet binding and plasma clotting, we further investigated whether rEIII treatment is sufficient for inducing platelet cell death. Annexin V and caspase results suggested the involvement of platelet apoptosis in DENV-induced pathogenesis and its association with disease severity (42)(43)(44). Because annexin V and caspase signals can be detected in both apoptotic and nonapoptotic cell death (45)(46)(47)(48)(49), we investigated whether nonapoptotic, regulated cell death (RCD) pathways are also involved in DENV-and rEIII-induced platelet death. Accordingly, RCD pathways, including pyroptosis, necroptosis, ferroptosis, apoptosis, and autophagy were analyzed using flow cytometry analysis. Here we found that in agreement with platelet activation data (Figures 1 and S3-S5), treatments with DENV and rEIII induced platelet cell death in a dose-dependent manner ( Figure 3A). Various cell death inducers, including nigericin (pyroptosis), doxorubicin (apoptosis), rapamycin (autophagy), TNF-a (necroptosis), and erastin (ferroptosis) roughly induced the respective cell death pathway of tested platelets ( Figure 3B, RCD % of total cells; 3C, RCD % of total dead cells; Figure S9, flow cytometry gating and calculations) (13,14). Intriguingly, when compared with these cell-death agonists, treatments with DENV and rEIII induced considerable pyroptosis, necroptosis, and ferroptosis responses of platelets but only minor levels of apoptosis ( Figures 3B, C). In addition, cell death patterns were similar in DENV-and rEIIItreated groups in which pyroptosis displayed the highest levels in both groups among all tested RCD pathways ( Figure 3C; >60%), suggesting that DENV-induced RCD is mediated through EIII moieties on virus particles. In addition, treatment with selective inhibitors of pyroptosis and necroptosis exerted ameliorative effects against DENV-and rEIII-induced platelet cell death ( Figure S10) and further showed the involvement of these two major RCD pathways. Accordingly, the Nlrp3 inhibitor OLT1177 and the inflammasome/caspase1 inhibitor Z-WHED-FMK were used to further characterize whether Nlrp3 inflammasome is involved in such pyroptosis response. Analysis results indicated that both Nlrp3 (OLT1177) and inflammasome/caspase 1 inhibitor (Z-WHED-FMK) treatments reduced total platelet death ( Figures 4A, B), in which almost all types of RCDs seem to be suppressed ( Figures 4B, C, F, G, J, K). While intriguingly, if we normalized various RCD cell population with the respective dead cell population (dead cell of each group normalized to 100%), (Figures 4D, E, H, I, L, M) we found that inhibitors of OLT1177 and Z-WHED-FMK preferentially suppressed pyroptosis ( Figure 4E), but not the other RCD pathways of platelets ( Figures 4H, I, L, M). To verify these pyroptosis results using different methods, we performed additional caspase-1 colorimetric assay, detections of cleaved GSDMD and cellsurface GSDMD levels, and the releases of IL-1b and LDH. Consistently, inflammasome inhibitors Z-WHED-FMK, OLT1177, and pyroptosis/GSDMD inhibitor DMF (14,50) markedly rescued rEIII-induced pyroptosis and damages of  Figure S8]. (B) RCD inducers, doxorubicin (DOX; apoptosis) (2.5 mg/ml), rapamycin (autophagy) (0.5 mM), erastin (ferroptosis) (10 mM), TNF-a (necroptosis) (2.5 ng/ml), and nigericin (pyroptosis) (3.5 mM) induced relatively simple RCD patterns. By contrast, DENV and rEIII induced multiple RCD pathways, in which pyroptosis is the major RCD response, with counts approximately 60% of total RCD (B). If we normalize the respective RCD by the population of death cells (dead cell population normalized to 100%), we can obtained a more similar RCD pattern in DENV and rEIII groups (C). *P < 0.05, **P < 0.01 vs. vehicle groups. platelets ( Figures S11, S12). These results collectively suggest that EIII induced an Nlrp3 inflammasome-dependent pyroptosis in platelets. In addition to EIII, soluble DENV NS1 levels also markedly increased in the circulation of DHF patients (51). Because DENV NS1 can activate inflammation through Toll-like receptor 4 (TLR4), NS1 is considered as one of the viral factors on the induction of dengue associated inflammation (52)(53)(54). Flow cytometry and platelet count analyses revealed that rEIII induced more pronounced platelet cell death, pyroptosis levels, and severer thrombocytopenia in mice, as compared to the rNS1 treatments ( Figure S13). This suggests that, compared to NS1, EIII may be a more potent viral factor on the induction of platelet damage.
In addition to the use of Nlrp3 inhibitors, Nlrp3 and caspase 1 null mice were compared with wild-type mice. In agreement with inhibitor experiments, intravenous rEIII injections induced more pronounced platelet count reduction ( Figure S14) and platelet activation ( Figure S15) in wild type, compared to the Nlrp3 and caspase 1 mutant mice. Furthermore, we found that Nlrp3 and caspase 1 deficiencies protected mouse platelet from DENV-and rEIII-induced cell death ( Figure 5). These results collectively suggested that Nlrp3 inflammasome is involved in  DENV-and rEIII-induced platelet pyroptosis, and Nlrp3 inflammasome inhibitors display a protective role.

Treatment With the Nlrp3 Inflammasome Inhibitor OLT1177 Ameliorate DENV-and rEIII-Induced Platelet Activation and Metabolic Burden of Mitochondria
Because Nlrp3 inflammasome-mediated pyroptosis is a major RCD involved in DENV-and rEIII-induced platelet defect, we further investigated whether the suppression of platelet Nlrp3 inflammasome through inhibitor treatment is sufficient to ameliorate DENV-and rEIII-induced abnormal platelet activation. Here we found that DENV and rEIII induced platelet aggregation, surface P-selectin expression, and increased mitochondrial superoxide levels in a dose-dependent manner (Figures 6A, C, E; gating Figure S16), whereas treatments with the Nlrp3 inflammasome inhibitors OLT1177 and Z-WHED-FMK ameliorated such abnormal platelet responses ( Figures 6B, D, F). Animal experiments further revealed that treatment with the Nlrp3 inflammasome inhibitor OLT1177 markedly ameliorated DENV-and rEIIIinduced thrombocytopenia in wild-type mice (Figure 7). These results collectively suggested that EIII is a virulence factor that induces thrombocytopenia, and Nlrp3 inflammasome is a critical target for DENV and EIII to induce platelet defects.

DISCUSSION
According to the World Health Organization guidelines, thrombocytopenia, a commonly observed symptom in both DF and DHF (55,56), is one of the crucial indicators of clinical severity (57). Mechanisms underlying thrombocytopenia and bleeding during DENV infection are not completely understood. Because patients with DF show thrombocytopenia and coagulant parameter changes (55,56), DENV alone, without secondary infection-eliciting DHF pathogenesis, is sufficient to cause platelet and coagulation defects. Hypotheses have been proposed to explain the mechanism involved. For example, studies have suggested that the increased destruction of platelets through an ongoing coagulopathy process or increased peripheral sequestration causes thrombocytopenia in dengue (55,56). However, the exact reasons and viral factors that lead to thrombocytopenia remain elusive. DENV EIII has shown to have cytotoxic effect on megakaryocytes (15,58). At the same time, DENV NS1 was demonstrated to enhance endothelial permeability and vascular leaks through a toll-like receptor 4 (52, 53). Both DENV virus particle-associated EIII and soluble NS1 could be detected at high levels prior to the acute phase of DHF (51,59) and may be considered as two virulence factors. Using clinically detected doses, here we found that intravenous treatments of both rEIII and rNS1 can activate platelets, induce platelet cell death and thrombocytopenia in mice. Despite these, compared to NS1, rEIII induced much higher levels of platelet pyroptosis and thrombocytopenia in mice ( Figure  S13). Because the induction of virion-surface EIII and soluble NS1 is induced in a similar, but not a same time course (51), the respective pathogenic role of EIII and NS1 on the elicitation of DHF-related pathogenesis remains to be further studied. However, data obtained in this present study suggested that virion-surface EIII is a candidate virulence factor that contributes to dengue-elicited thrombocytopenia. In addition, similar to soluble NS1, virionsurface EIII is a virulence factor that contributes to the first-hit in our two-hit DHF model (13,16). Primarily based on annexin V and caspase data, reports have suggested that platelet apoptosis is involved in DENV-induced pathogenesis and is associated with disease severity (42)(43)(44). However, because non-apoptotic cell death signaling pathways have recently been discovered, annexin V and caspase signals can be detected in various other types of RCD (45)(46)(47)(48)(49). Therefore, after the re-examination of RCD pathways, in this present study, we found that pyroptosis and necroptosis are the two major types of DENVand rEIII-induced platelet RCDs, with approximately 55-60% and  20-25% RCD-marker (+) population ( Figure 3C), respectively. Pyroptosis and necroptosis belong to necrotic RCD pathways (60,61). Inflammasomes regulate and interact with various RCDs (62)(63)(64)(65). DENV-induced Nlrp3 inflammasome activation and pyroptosis have been described in DENV-stimulated macrophages and monocytes (12,66,67). Although DENVinduced platelet pyroptosis has not been clearly characterized, Nlrp3 inflammasome-mediated inflammation has also been reported in DENV-stimulated platelets and DENV-induced thrombocytopenia and coagulopathy (12,16,68). These results suggested that although necrotic RCD of platelets need further characterization, Nlrp3 inflammasome-mediated inflammatory responses in other cell types have been recognized in dengue.
Even in one single cell type, cell population is still heterogeneous. This is likely the reason why previous reports have revealed that pathogen infections and cytotoxic agent treatments can lead to multiple types of RCDs simultaneously (69)(70)(71)(72)(73)(74)(75). For example, apoptosis, necroptosis, and pyroptosis participate simultaneously  during retinitis (69). Photodynamic therapy can induce multiple RCD pathways, including apoptosis and autophagy (70). Cellular stress can activate both receptor-induced lysosomal-dependent and mitochondrial-mediated cell death pathways, which may lead to both programmed necrosis and apoptosis (71). Similarly, as DENV and EIII have been reported to have multiple cellular targets, it is not surprising to detect multiple RCDs after DENV and EIII challenges. In this present report, we found that DENV and EIII but not the other cell death inducers, induced a similar RCD pattern in platelets ( Figures 3B, C). Although further investigations are needed, this platelet cell-type-specific RCD patterns (CTS-RCDPs) (13,14) may be also useful on the characterization of specific pathway inhibitors; as inflammasome inhibitors OLT1177 and Z-WHED-FMK preferentially blocked pyroptosis, but not ferroptosis, apoptosis, and autophagy ( Figure 4E, vs. Figures 4I, L, M). Various putative cellular DENV cellular targets are expressed by platelets. For example, recent lines of evidence have suggested that CLEC5A mediates DENV-induced inflammation (67,76), and glycoprotein Ib (GP1ba; CD42b) is involved in DENV infection (77). Both CLEC5A and CD42b are expressed by platelets (78,79). Because CLEC5A involves cell death regulations (80) and interaction of platelet CD42b with von Willebrand factor induces platelet cell death (78), competition assay was performed to investigate whether recombinant CLEC5A and CD42b can block rEIII induced RCD. Results revealed that recombinant CD42b can block both rEIII-platelet binding, and rEIII-induced platelet cell death ( Figure S17). In addition, similar to DENV and rEIII that induced platelet RCDs, pyroptosis is the major RCD of anti-CD42b antibody that induced platelet cell death ( Figure S17). These results suggested that CD42b and CLEC5A are potential cellular targets of EIII on platelets and is worthy of further investigations. Cellular stresses, including stimulation of danger-associated molecular patterns and pathogen-associated molecular patterns, can activate NLRP3 inflammasome to initiate pyroptosis and release the proinflammatory cytokines IL-1b and IL-18 (81). NLRP3 inflammasome inhibition prevents a wide range of diseases, including Alzheimer disease, metabolic diseases, and infectious diseases (64). In this study, treatment with selective inhibitors against Nlrp3 inflammasome displayed rescue effects on DENV-and rEIII-induced platelet cell death in vitro and thrombocytopenia in mice, suggesting that Nlrp3 inflammasome is a potential target for the therapeutic intervention of DENVinduced platelet defects. In addition, with further upstream blocking, treatment with an EIII-competitive inhibitor CSB displayed promising rescue effects. Our data suggest that DENV-E is a virulence factor that contributes to plateletassociated pathogenesis in dengue. This concept may be useful for developing therapeutic strategies to manage DENV-induced hemorrhage pathogenesis.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.