Dengue virus and antiplatelet autoantibodies

4 downloads 0 Views 2MB Size Report
cepted that antibody-dependent enhancement is the primary reason why patients with ... Autoantibody, dengue virus, Fc receptor III, Nlrp3 inflammasome, Toll.
1

Blood Coagulation, Fibrinolysis and Cellular Haemostasis

Dengue virus and antiplatelet autoantibodies synergistically induce haemorrhage through Nlrp3-inflammasome and FcγRIII Te-Sheng Lien1†; Der-Shan Sun1,2†; Chia-Ming Chang3; Cheng-Yeu Wu4; Ming-Shen Dai5; Hao Chan2; Wen-Sheng Wu6; Shu-Hui Su2; You-Yen Lin2; Hsin-Hou Chang1,2,7 1Department

of Molecular Biology and Human Genetics, Tzu-Chi University, Hualien, Taiwan; 2Institute of Medical Sciences, Tzu-Chi University, Hualien, Taiwan; 3Department of Biochemistry, Tzu-Chi University, Hualien, Taiwan; 4Center for Molecular and Clinical Immunology, Chang Gung University, Gueishan, Taoyuan, Taiwan; 5Division of Hematology/ Oncology, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan; 6Department of Laboratory Medicine and Biotechnology, Tzu-Chi University, Hualien, Taiwan; 7Center for Vascular Medicine, Tzu-Chi University, Hualien, Taiwan

Summary Dengue haemorrhagic fever (DHF) typically occurs during secondary infections with dengue viruses (DENVs). Although it is generally accepted that antibody-dependent enhancement is the primary reason why patients with secondary infection are at an increased risk of developing DHF, a growing body of evidence shows that other mechanisms, such as the elicitation of antiplatelet autoantibodies by DENV nonstructural protein NS1, also play crucial roles in the pathogenesis of DHF. In this study, we developed a “two-hit” model of secondary DENV infection to examine the respective roles of DENV (first hit) and antiplatelet Igs (second hit) on the induction of haemorrhage. Mice were first exposed to DENV and then exposed to antiplatelet or antiCorrespondence to: Hsin-Hou Chang, PhD Department of Molecular Biology and Human Genetics Tzu-Chi University, Hualien 970, Taiwan, Republic of China Tel.: +886 3 8565301 ext 2667, Fax: +886 3 8578386 E-mail: [email protected]

NS1 Igs 24 hours later. The two-hit treatment induced substantial haemorrhage, coagulopathy, and cytokine surge, and additional treatment with antagonists of TNF-α, IL-1, caspase-1, and FcγRIII ameliorated such effects. In addition, knockout mice lacking the Fcγ receptor III, Toll-like receptor 3, and inflammasome components Nlrp3 and caspase-1 exhibited considerably fewer pathological alterations than did wild type controls. These findings may provide new perspectives for developing feasible approaches to treat patients with DHF.

Keywords Autoantibody, dengue virus, Fcγ receptor III, Nlrp3 inflammasome, Toll like receptor 3, Shwartzman reaction Financial support: This work is supported by research funding from National Science Council (98–2320-B-320–004MY3, 101–2320-B-320–004-MY3 to HHC) and Tzu-Chi University (TCIRP98001, TCIRP101001 to HHC, DSS, WSW, and SHS) and Chang Gung Memorial Hospital (CLRPD1A0012 to CYW). Received: July 29, 2014 Accepted after major revision: January 6, 2015 Epub ahead of print: #####



These authors share equal contribution.

Introduction Dengue is an infectious disease caused by any one of four serotypes of the dengue virus (DENV), a mosquito-transmitted virus in the flavivirus genus. Although the disease is typically self-limiting and most patients recover without any sequelae, patients with secondary DENV infections and infants with primary DENV infections born to DENV-immune mothers can develop lifethreatening dengue haemorrhagic fever (DHF) (1, 2). Although antibody-dependent enhancement (ADE) is generally accepted to be the primary reason patients with secondary infection are at an increased risk of developing DHF (3), a growing body of evidence has shown that other mechanisms also play crucial roles in the pathogenesis of DHF (4–9). Patients with DHF exhibit elevated levels of antiplatelet and platelet-associated IgM and IgG (8, 9). Results from animal studies have shown that immunisation with DENV recombinant nonstructural protein 1 (NS1) elicits the cross-reactivity of antibodies against self-epitopes on platelets such as inte© Schattauer 2015

http://dx.doi.org/TH14-07-0637 Thromb Haemost 2015; 113: ############

grin αIIbβ3 (CD41/CD61) (4–6). However, challenges with antiNS1 Ig alone have been shown to lead to thrombocytopenia, clotting defects, and liver damage without inducing severe haemorrhage in experimental mice (4, 7), suggesting synergism with additional factors may be required to elicit haemorrhage. DENV viremia suppresses the expression of various anticoagulant proteins in plasma and induces hypercoagulation in both humans (11) and mice (Suppl. Figure 1, available online at www. thrombosis-online.com). Using a warfarin-induced hypercoagulable mouse model, we previously demonstrated that exposure to anti-NS1 Ig caused severe thrombocytopenia and disseminated intravascular coagulation (4). However, it is unclear whether this model encapsulates the complexity of DHF. Studies have shown that DHF typically occurs when viremia coincides with the induction of anti-DENV Igs (▶ Figure 1 A) (12, 13). Therefore, in this study, we developed a “two-hit” mouse model to examine the respective roles of DENV (first hit) and antiplatelet Igs (second hit) in inducting haemorrhage. The experiThrombosis and Haemostasis 113.5/2015

2

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

ment protocol was similar to that used to induce local Shwartzman reaction-mediated haemorrhage (14, 15). Two pathways were investigated, namely the inflammasome and the IgG Fc receptor (FcγR). The Nlrp3 inflammasome, a multiprotein oligomer comprising adaptor protein ASC, Nlrp3, and caspase-1, promotes the maturation of inflammatory cytokines interleukin-1 (IL-1) in response to infection or stress (16). In vitro evidence suggests that DENV stimulates the C-type lectin domain family 5 member A (CLEC5A)-mediated macrophage activation of Nlrp3 inflammasome (17). In addition, evidence indicates that DENV can activate Nlrp3-inflammasomes in platelets, resulting in the secretion of IL-1β, which in turn alters vascular permeability (18). However, the role of the Nlrp3 inflammasome in the pathogenesis of DENV-elicited haemorrhage remains unclear. The Fcγ receptor (FcγR) system comprises two opposing receptors: the activating FcγRIII and the inhibitory FcγRIIB (19). These opposing signalling pathways act in concert, determining the magnitude of the effector cell responses during inflammation and autoimmune diseases (20). FcγRs are also essential for clearing platelet-antiplatelet Ig complexes (19). However, the roles that FcγRs play in the pathogenesis of DHF have not been well-characterised. Using sequential injection protocols, we investigated the respective roles of DENV (first hit) and antiplatelet Igs (second hit) on the induction of haemorrhage.

Materials and methods DENV, recombinant protein and antibodies The DENVs [DENV2 strains PL046 and New Guinea C (NGC), DENV4 H241] and soluble recombinant proteins DENV NS1 (rNS1) were obtained and purified as previously described (4, 6). Mosquito C6/36 cells (ATCC®CRL-1660TM) were employed for DENV amplification. Sucrose density gradation was used for DENV purification. QIAamp viral RNA minikits (Qiagen, Hilden, Germany), Maxima SYBR Green qPCR Master Mixes (Thermo Scientific, Waltham, MA, USA), a Lightcycler 480 (Roche, Basel, Switzerland) and plaque-forming assay on BHK-21 cells (ATCC®CCL-10TM) (20), were used to quantify DENV copies and titres. The pre-immune Ig and the anti-NS1 Ig from rabbits were obtained before and after rNS1-immunisations, respectively (4, 6). Rabbit anti-NS1 sera were collected 7–10 days after the 5th immunisation cycle. Anti-sera with the highest anti-platelet IgG titres were used. The IgG fractions were obtained and purified using a protein-A column connected to a peristaltic pump (Amersham Biosciences, Little Chalfont, UK) that operated at a flow rate of 0.5–1 mL min-1. Rabbit polyclonal anti-DENV2 envelope protein IgG was purchased from GeneTex (Irvine, CA, USA).

Experimental mice Wild type C57BL/6J mice aged 8–12 weeks were purchased from the National Laboratory Animal Center (Taipei, Taiwan). Mutant mice with a C57BL/6J background, including Tlr3-/-, Nlrp3-/-, and Thrombosis and Haemostasis 113.5/2015

Casp1-/- (21–23) were obtained from the Centre National de Recherche Scientifique (Orléans, France). The Tlr4-/- (B6.B10ScNTlr4lps-del/JthJ), Fcgr2b-/- (CD32; B6. 129S-Fcgr2btm1Ttk/J), and Fcgr3-/- (CD16; B6.129P2-Fcgr3tm1Sjv/SjvJ) mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA). All animals were housed in the Tzu-Chi University Animal Center in a specific-pathogen-free, temperature and lighting controlled environment with free access to filtered water and food. All of the genetic knockout strains were backcrossed with the wild-type C57Bl/6J mice for at least six generations. The experimental procedures were approved by the Animal Care and Use Committee of Tzu-Chi University (approval ID: 101019).

Two-hit challenges using DENV plus antiplatelet Ig to induce a local Shwartzman reaction-like response To induce local Shwartzman reaction-mediated haemorrhage in mice using traditional methods, two sequential subcutaneous lipopolysaccharide (LPS) injections are required over a period of two consecutive days (14). In our two-hit mouse model, mice first received a subcutaneous injection of DENV (3 × 105 PFU/mouse; DHF viral load) (24) followed by a subcutaneous injection of antiplatelet IgGs [using either (I) anti-CD41, rat monoclonal MWReg30, Pharmingen; 0.2 mg/kg, well-established for immune thrombocytopenia (ITP) induction (25); or (II) anti-NS1 Ig, rabbit polyclonal, 8.5 mg/kg, ITP-inducible (4)] 24 hours (h) later at the same site of DENV injection. Anesthesia was induced 5 minutes (min) before each injection (e. g. vehicle, DENV and Igs) by intraperitoneal injection of 2.5 % Avertin solution (in saline, 10 ml/kg). Because isotype control rat Ig (0.2 mg/kg body weight; Biolegend, San Diego, CA, USA) does not induce abnormal responses in mice (▶ Figure 1), Ig from preimmune rabbits (8.5 mg/kg) was used as the control Ig. At 24 h after the injection of antiplatelet Ig, we analysed platelet counts (Analyser KX-21N, Sysmex; blood samples) (25), degree of haemorrhage (as described in the next section; skin samples), the level of D-dimer expression (ELISA, Sekisui Diagnostica, Lexington, MA, USA), level of cytokines TNF-α (ELISA, e-Bioscience), IL-1β (ELISA, e-Bioscience, San Diego, CA, USA), IL-6 (ELISA, Biolegend) and IL-10 (ELISA, Biolegend), and the level of expression of anticoagulant proteins antithrombin III (chromogenic, Sekisui Diagnostica) and protein C (chromogenic, Sekisui Diagnostica) using plasma samples. Inhibitors and drugs, such as IVIg (Bayer, Leverkusen, Germany; 1 g/kg), caspase inhibitor Z-VAD-FMK (R&D Systems, Indianapolis, IN, USA, 10 mg/kg), IL-1RA (recombinant IL-1 receptor antagonist, Peprotech, 0.8 mg/kg), and etanercept (Pfizer, New York, NY, USA; 10 mg/kg), were subcutaneously administered before (10 min post DENV injection) or after (10 min post Ig injection) injections with anti-CD41/anti-NS1 Ig (Suppl. Table 1, available online at www.thrombosis-online.com). The conditions of primary DENV-infected infants, those who previously obtained maternally-derived DENV-elicited autoantibodies, were also simulated using a “reverse” two-hit mouse model. Two sequential subcutaneous injections of anti-platelet IgG (first-hit, anti-CD41, MWReg30) and DENV (second-hit, 3 × 105 PFU/mouse) were © Schattauer 2015

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

Figure 1: Two-hit-induced pathogenesis. Changes in clinical parameters during DHF (A) and the experimental outline (B) are shown. C–F) Mice with (panels 8–14) or without (panels 1–7) DENV (3 × 105 PFU/mouse; 1st-injection; DENV2 PL046) injections were then challenged by various treatments (2nd-injections). The changes in (C) platelet counts, (D) haemorrhagic lesions, (E) haemorrhage score, and (F) proinflammatory cytokines were recorded. Results are shown as mean ± standard deviation (SD). * P < 0.05, **P < 0.01

indicate significantly worse conditions vs DENV + control Ig-1 groups; #P < 0.05, ##P < 0.01 vs vehicle groups. Vehicle, DENV, DENV + anti-CD41 Ig, n = 20; anti-CD41 Ig, DENV + control Ig, DENV + anti-NS1 Ig, n = 15; DENV + anti-envelope Ig, n = 10; other groups n = 6 (3 independent experiments with two replicates). Control Ig-1: rabbit preimmune control IgG; control Ig-2: rat isotype control Ig.

given over a period of two consecutive days. At 24 h after DENV treatment (48 h after MWReg30 injection), platelet counts, degree of haemorrhage, and levels of cytokines interleukin (IL)-1β and tumour necrosis factor (TNF)-α, levels of protein C and antithrombin III, and D-dimer were analysed in accordance with the aforementioned methods.

To analyse aforementioned platelet counts, cytokines and coagulant parameters, whole blood (100 µl) was collected from retroorbital venous plexus of mice and mixed with anticoagulant citrate dextrose solution (38 mM citric acid, 75 mM sodium citrate, 100 mM dextrose) in Eppendorf tubes (25). Plasma samples could be obtained after additional centrifugation 15 min at 2,000 × g to re-

© Schattauer 2015

Thrombosis and Haemostasis 113.5/2015

3

4

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

move blood cells and platelets. To collect the skin samples for imaging haemorrhage lesions in two-hit-induced local Shwartzman reaction, and at the end of other experiments, mice were euthanised using CO2 following a NIH guideline (http://oacu.od.nih.gov/ARAC/documents/Rodent_Euthanasia_Adult.pdf).

Quantification for the degree of haemorrhage using digitised images In the two-hit mouse model, DENV and antiplatelet Ig were injected at a same skin location of each mouse. The haemorrhage is a localised response; we did not observe haemorrhage elsewhere of the mouse body. As a result, we collected the skin samples just around the injection sites. The grading of haemorrhage in the local Shwartzman reaction can be measured using an arbitrary scaling of 0 to 4 (13, 26). In this study, however, we developed a quantification protocol to obtain a relatively more objective measurement (Suppl. Figure 2, available online at www.thrombosis-online.com). Images of the haemorrhagic lesions were taken under standard conditions (elimination density 200 lux, 20W Phillips fluorescent lamp; Canon IXUS-860IS camera; with a sample to camera distance of 7 cm). Red and green signals for the digitised images (RGB mode, 0.75 × 0.6 cm2, 600 dpi) were then obtained using Photoshop software (Adobe, San Jose, CA, USA) without any brightness or contrast adjustments. The red and green intensities in a particular image of the haemorrhagic lesion were further measured using Image J software (v1.46, NIH, Bethesda, MD, USA). An approximate value for the severity of haemorrhage was calculated by subtracting the image intensity of the red images from that of the green images (Suppl. Figure 2, available online at www.thrombosis-online.com).

Detection of DENV-elicited FcγRIII+ and tissue factor+ leukocytes At 24 h after the injection of DENV, peripheral blood mononuclear cells of the infected mice were collected, incubated with phycoerythrin (PE)–conjugated anti-CD14 Ig (CD14: monocyte/macrophage marker; Sa14–2, Biolegend) and anti-FcγR Igs, and then analysed using flow cytometry. Because anti-mouse CD32 (FcγRIIB) Ig is not commercially available, we used fluorescein isothiocyanate (FITC)–conjugated anti-CD16/CD32 Ig (Biolegend) and allophycocyanin (APC)–conjugated anti-CD16 Ig (United States Biological, Salem, MA, USA). For tissue factor staining, rabbit anti-mouse tissue factor IgG (Sekisui Diagnostica), and goat anti-rabbit IgG-FITC (Cappel) were used.

Statistical analyses The means, standard deviations, and statistics for the quantifiable data were calculated using Microsoft Office Excel 2003, SigmaPlot 10 and SPSS 17. Significance of data was examined by one-way ANOVA followed by the post-hoc Bonferroni-corrected t-test. The Thrombosis and Haemostasis 113.5/2015

probability of type 1 error α = 0.05 was recognised to be the threshold of statistical significance.

Results Concurrence of viremia and antiplatelet autoantibodies exacerbated haemorrhage and inflammation Several pathophysiological parameters, including thrombocytopenia, plasma leakage or hypoproteinaemia, and high levels of circulating aspartate amino transferase (AST), are recognised by the World Health Organisation as standard measurements for evaluating the severity of DHF (13, 28). The results of studies that used two-hit mouse models of DHF comprising DENV plus anti-NS1 Igs (containing antiplatelet Igs) (4) or anti-CD41 Igs (CD41 is a putative platelet target of anti-NS1 Ig) (5) have revealed considerable exacerbation of thrombocytopenia (▶ Figure 1 C; Suppl. Figure 3A, B, available online at www.thrombosis-online.com) and plasma leakage (hypoproteinaemia; Suppl. Figure 3C, available online at www.thrombosis-online.com) as well as substantially elevated levels of circulating AST (Suppl. Figure 3D, available online at www.thrombosis-online.com). Those studies also showed that death occurred within 24 hours in mice with severe DHF (Suppl. Figure 3E, available online at www.thrombosis-online.com). Because two higher doses (1.2 × 106 and 2.4 × 106 PFU/mouse) of DENV were shown to induce high mortality in mice when cotreated with antiplatelet or anti-NS1 Igs (Suppl. Figure 3, available online at www.thrombosis-online.com), we chose a DENV dose of 3 × 105 PFU/mouse for further analyses. Data on haemorrhage and inflammation severity revealed that mice that received combined treatments of DENV plus either anti-NS1 Igs or anti-CD41 Igs, but not control Igs (preimmune and anti-DENV-envelope Igs), exhibited considerably lower platelet counts (▶ Figure 1 C), more severe haemorrhaging (▶ Figure 1 D, E), and higher levels of proinflammatory cytokines TNF-α, IL-1β, and IL-6 (▶ Figure 1 F) and anti-inflammatory cytokine IL-10 (Suppl. Figure 4A, B models; C, available online at www.thrombosis-online.com) than mice in the single-injection groups (▶ Figure 1 C-F, Suppl. Figure 4C, panels 12–13 vs the other single-treatment groups, available online at www.thrombosis-online.com). Inductions of thrombocytopenia, haemorrhage, and the four aforementioned cytokines have been associated with the severity of DENV infection (11, 13, 28–32). Similarly, mice that received the two-hit treatment had markedly lower levels of circulating anticoagulant protein C and higher levels of circulating D-dimer (Suppl. Figure 4D-E, panels 12–13 vs the other single-treatment groups, available online at www.thrombosis-online.com). Although evaluating the severity of disease in mice is difficult when using guidelines for humans, the pathophysiological changes observed in our two-hit mouse model are similar to those observed in DHF patients (13, 28). Because the pathogenic response occurs approximately 24–48 h after DENV challenge, we analysed the viral load according to this time course. The results indicated that after injections, the DENV titre rapidly declined without further rebound within 24–48 h (Suppl. Figure 5, © Schattauer 2015

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

available online at www.thrombosis-online.com), suggesting DENV amplification is not involved in the development of twohit-induced complications in mice. Furthermore, in addition to using DENV2 PL046, we also obtained similar results using strains DENV2 NGC and DENV4 H241 (Suppl. Figures 6 and 7, available online at www.thrombosis-online.com). This implies that the aforementioned two-hit-induced pathogenesis is not a serotype or strain specific response.

Involvement of Nlrp3 inflammasome in the haemorrhage pathogenesis After establishing the validity of our mouse model (▶ Figure 1), we investigated the involvement of specific molecular pathways using pharmacological antagonists and gene knockout mice. TLR3 is a well-studied pattern recognition receptor that recognises viral RNA, including DENV (33). TLR3-deficient mice were firstly used to test our model. As expected, the DENV and antiplatelet Ig twohit-induced pathological alterations, which included haemorrhage, cytokine surge, and coagulant disturbances, were all considerably less affected in Tlr3-/- mutant mice (Suppl. Figure 8, available online at www.thrombosis-online.com). A previous in vitro analysis revealed that DENV activates Nlrp3 inflammasomes through CLEC5A (17); however, the pathogenic role of inflammasomes in DHF remains unclear. Our results show that DENV treatments resulted in a substantial increase in IL-1β levels (▶ Figure 1 F, panel 2), a downstream response to inflammasome-mediated activation of caspase-1. Exposure to the caspase inhibitor Z-VAD-FMK (Z-VAD) considerably ameliorated all of the aforementioned pathological changes elicited by the two-hit protocol, suggesting that a caspase pathway is involved (▶ Figure 2 A, experiment outline; ▶ Figure 2 B-E, and Suppl. Figure 9A-D, available online at www.thrombosis-online.com). To clarify the involvement of the Nlrp3 inflammasome, Nlrp3 (Nlrp3-/-) and caspase-1 (Casp1-/-) knockout mice were exposed to the two-hit treatment. Consistently, all of the aforementioned pathological changes were greatly reduced in both Nlrp3-/- and Casp1-/- mice when compared with the wild-type controls after the two-hit treatments (▶ Figure 2 F-I, and Suppl. Figure 9E-G, available online at www. thrombosis-online.com). Additional analyses revealed that treatment with the caspase inhibitor Z-VAD resulted in more pronounced amelioration of the aforementioned pathological changes when the mice were treated before (10 min after DENV injection; ▶Figure 2A), but not after (10 min after antiplatelet Ig injection), injection of antiplatelet Igs (Suppl. Table 1, available online at www.thrombosis-online.com). This result suggests that the Nlrp3 inflammasome plays a crucial role in the initial stage of the two-hit mediated pathogenesis.

Involvement of FcγRIII pathway in the haemorrhage pathogenesis FcγRIII (CD16) is essential for the clearance of platelet-antiplatelet Ig complexes (19). We found that DENV challenge resulted in a moderate increase in FcγRIII+ leukocytes, an increase that was © Schattauer 2015

even greater after exposure to anti-CD41 Ig (Suppl. Figure 10, available online at www.thrombosis-online.com). DENV-mediated elicitation of FcγRIII+ leukocytes indicates an induction of proinflammatory phenotype, which may exacerbate inflammation without a corresponding antibody (33, 34). Anti-inflammatory intravenous immunoglobulin (IVIg) treatments blocked the access of immunocomplexes to FcγRIII and modulated the expression of FcγRs on leukocytes (35). Human-originated IVIg ameliorated mouse ITP through FcγRs, and rat antiplatelet Ig induced less-severe ITP in FcγRIII deficient mice (26, 36), suggesting a cross-reorganisation of mouse FcγRs toward xenotransfusion Igs. Our data indicated that IVIg treatment markedly reduced the level of circulating FcγRIII+ leukocytes (Suppl. Figure 10, available online at www.thrombosis-online.com) and ameliorated all of the aforementioned two-hit-mediated pathological changes (▶ Figure 3 A-E; Suppl. Figure 11A-D, available online at www.thrombosisonline.com). Verification using genetically deficient mice showed that two-hit-mediated pathogenic responses were substantially diminished in Fcgr3-/- mice but not in Fcgr2b-/- mice (▶ Figure 3 F-I; Suppl. Figure 11E-G, available online at www.thrombosis-on line.com), suggesting that FcγRIII was involved in exacerbating the two-hit mediated pathogeneses.

Treatments with IL-1 and TNF inhibitors This two-hit mouse model may be used to screen and investigate the therapeutic agents for treating patients with dengue in whom antiplatelet autoantibodies are elicited. IL-1β release depends on inflammasome activation. Our results showed that treatment with an IL-1 receptor antagonist (IL-1RA) and TNF-α inhibitor etanercept markedly ameliorated all of the aforementioned two-hit-mediated pathogenic changes (▶ Figure 4; Suppl. Figure 12, available online at www.thrombosis-online.com). These results suggest that the proinflammatory inflammasome/IL-1 and TNF-α pathways play critical roles in the development of two-hit-induced haemorrhage complications.

Two-hit mouse model for infant DHF Previous studies have indicated that maternal DENV-elicited antibodies are essential for DHF developing in infants with primary DENV infections (37). In these cases, DENV-elicited antibodies appeared before DENV-viremia. Libraty et al. showed that ADE was not involved in the elicitation of DHF in infants (38). This suggests that, in addition to ADE, the maternal-derived DENV-elicited antibodies exerted other pathogenic roles. Notably, a revised two-hit model designed to simulate the infant DHF through a reversed injection protocol (initial antiplatelet-Ig injection followed by secondary DENV treatment) led to substantial haemorrhage, cytokine surge, and disturbed coagulation system in mice (Suppl. Figure 13, available online at www.thrombosis-online.com). This suggests that the two-hit model can also be applied to infants with DHF.

Thrombosis and Haemostasis 113.5/2015

5

6

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

Figure 2: Involvement of the Nlrp3 inflammasome pathway. Experimental outline (A) and the pathophysiological changes in mouse platelet counts (B, F), haemorrhagic lesions (C–D, G–H), and levels of proinflammatory cytokines TNF-α, IL-1 and IL-6 (E, I) are shown. B–E) Treatment with caspase inhibitor Z-VAD reduced haemorrhage and inflammation of two-hit challenged wild-type (WT) mice (panels 3, 5 vs 2, 4, respectively). F–I) Haemorrhagic and inflammatory manifestations of Nlrp3-/- (panels 4–6) and Casp1-/- (panels 7–9) mutants were compared to those of WT (panels 1–3) mice after DENV (PL046) and antibody treatments (n = 6; F, H, I: panels 6, 9, n = 4). Data are represented as mean ± SD. #P < 0.05, ##P < 0.01 indicate significantly ameliorated conditions compared with the conditions of respective DENV + antiCD41 (α-CD41)/anti-NS1 (α-NS1) Ig groups (B, D–E). *P < 0.05, **P < 0.01 vs respective DENV + control (Ctrl) Ig groups; *P < 0.05, **P < 0.01 vs respective WT groups (F, H–I).

Thrombosis and Haemostasis 113.5/2015

© Schattauer 2015

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

Figure 3: Involvement of the FcγRIII pathway. Experimental outline (A) and changes in mouse platelet counts (B, F), severity of haemorrhagic lesions (C-D, G-H), and (E, I) plasma levels of proinflammatory cytokines TNF-α, IL-1β, and IL-6. (B–E) IVIg treatments reduced haemorrhage and inflammation of two-hitchallenged wild-type (WT) mice (panels 3, 5 vs 2, 4). F–I) Haemorrhagic and inflammatory manifestations of Fcgr2b-/- (panels 4–6) and Fcgr3-/- (panels 7–9) mutants were compared to those of WT (panels 1–3) mice after two-hit challenges (n = 6; B–E panel 5, n = 4). Data are represented as mean ± SD. #P < 0.05, ##P < 0.01 indicate significantly ameliorated conditions compared with the conditions of respective DENV (PL046) + antiCD41 (α-CD41)/anti-NS1 (α-NS1) Ig groups (B, D–E). *P < 0.05, **P < 0.01 vs respective DENV + control (Ctrl) Ig groups; *P < 0.05, **P < 0.01 vs respective WT groups (F, H–I).

© Schattauer 2015

Thrombosis and Haemostasis 113.5/2015

7

8

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

Figure 4: Involvement of TNF-α and IL-1β pathways. Experimental outline (A) and changes in mouse platelet counts (B), severity of haemorrhagic lesions (C-D), and levels of proinflammatory cytokines TNF-α, IL-1β and IL-6 (E) are shown. B-E) Treatments with IL1RA or etanercept (IL-1 and TNF-α antagonist, respectively) reduced haemorrhage and inflammation in two-hit challenged wild-type (WT) mice (n = 6; B–E panels 3, 6, 7, n = 4). Data are represented as mean ± SD. #P < 0.05, ##P < 0.01 indicate significantly ameliorated conditions compared with those of respective DENV (PL046) + anti-CD41 (α-CD41)/ anti-NS1 (α-NS1) Ig groups.

Figure 5: The two-hit model. Hypothetical twohit model of DHF pathogenesis (A) and the timing and parameter changes (B) in circulating DENV (first-hit events [a]), and anti-DENV Ig (second-hit events [b]) in DHF patients are illustrated. ↓ Activation processes. ⊥ Inhibition processes.

Thrombosis and Haemostasis 113.5/2015

© Schattauer 2015

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

Discussion Dengue viruses do not normally propagate or induce DHF in mice. Previous studies have indicated that single injections of DENV can lead to haemorrhaging when extremely high doses of DENV are administered (4 × 107–3 × 109 PFU/mouse) (39), equivalent to a surge of 2.3 × 107–1.7 × 109 PFU/ml DENV in the circulation of mice. This dosage is approximately 102–104-fold higher than the viral load detected in patients with DHF [105–106 PFU/ml (25)]. In this study, we demonstrated for the first time that concurrence of DHF-viral-load DENV (1.6 × 105 PFU/ml, equivalent to 3 × 105 PFU/mice, or 1.2 × 107 PFU/kg) and antiplatelet autoantibodies can cause severe haemorrhage and cytokine surge. In the traditional LPS-elicited Shwartzman reaction, TNF-α, IL-1, procoagulation activity, and haemorrhage were induced (15), reactions that were similarly prompted in our two-hit mouse model (▶ Figures 1-4) and in patients with DHF (11, 29, 30), suggesting that haemorrhage related to DHF might be a DENV-induced Shwartzman reaction-like response. According to these results, we propose a two-hit hypothesis for the development of DHF (▶ Figure 5). The first hit is DENV-induced inflammation, which involves the regulation of TLR3, Nlrp3 inflammasome, and subsequent release of IL-1β and TNF-α (▶ Figure 2, ▶ Figure 5 A, [a1-a4]; Suppl. Figure 8, available online at www.thrombosis-online.com). Hottz et al. found that activation of the Nlrp3 inflammasome by DENV triggered the release of IL-1β, resulting in enhanced vascular permeability (18). In the present study, we found that the pathological changes elicited by the two-hit model were inhibited by the caspase inhibitor Z-VAD, the IL-1β inhibitor IL-1RA, and the TNF-α inhibitor etanercept (▶ Figure 2 A-E, ▶ Figure 4, ▶ Figure 5 A, [a3-a4]). The second hit is the DENV-mediated generation of autoantibodies (▶ Figure 5 A [b1-b4]). DENV-exposed mice had considerably elevated levels of FcγRIII+ leukocytes after exposure to antiplatelet Igs (Suppl. Figure 10, available online at www.thrombosisonline.com), revealing a proinflammatory phenotype. In Fcgr3-/--deficient mice, both single- and two-hit treatments resulted in lower levels of circulating TNF-α, IL-1β, IL-6 and IL-10 than those in wild type mice (▶ Figure 3 and Suppl. Figure 11, available online at www.thrombosis-online.com). These data suggest that the involvement of FcγRIII pathway (▶ Figure 5 A [b2]). Anti-inflammatory IVIg treatments substantially attenuated all of the aforementioned pathogenic changes (▶ Figure 3 B-E; Suppl. Figure 11, available online at www.thrombosis-online.com) associated with the suppression of FcγRIII on leukocytes (Suppl. Figure 10, IVIg-groups, available online at www.thrombosis-online.com; and ▶ Figure 5 A [b2-b3]). Because FcγRIII+ leukocytes play a critical role in modulating proinflammatory response even without a corresponding antibody (33, 34), the ameliorative effect of IVIg in this model was expected. Additionally, because both the first and second hits involve FcγRIII, and IVIg suppresses FcγRIII+ leukocytes, that IVIg-treatments were equally potent before and after the administration of second-hit antiplatelet IgG is reasonable (Suppl. Table 1, available online at www.thrombosis-online. com). When both hits achieved a pathogenic threshold, haemor© Schattauer 2015

rhage occurs (▶ Figure 5 A [c], B [c, c2]). Consistently, the coagulation disturbances (suppression of protein C and antithrombin III, Suppl. Figures 6–9 and 12–13, available online at www.throm bosis-online.com) were highly correlated with the induction of haemorrhage and inflammation (▶ Figures 2–4) in this two-hit mouse model. DENV infection elicits serotype-specific protective immunity (1). This may explain why DHF most commonly occurs after consecutive infections with DENVs of two different serotypes. According to the two-hit model, we postulate that protective immunity can ameliorate the impact from the first-hit and reduce the rates of DHF in single-serotype epidemics. Nonetheless, if such protective-immunity cannot be properly elicited for certain reasons [e. g. original antigenic sin (40), intrinsic ADE (41)], singleserotype-induced DHF may occur. This is probably the reason why DHF cases caused by single-serotype DENV infection have been reported mainly in remote areas and islands (29, 42–44). Organ tropism studies have isolated DENVs from skin and gastrointestinal tissue (45, 46). Because antibodies theoretically circulate and reach most organs, concurrent DENV and antiplatelet Ig in these tissues during secondary DENV infections may explain why skin haemorrhages, including petechiae and purpura, as well as gum bleeding, epistaxis, menorrhagia, and gastrointestinal haemorrhage, are common in patients with DHF (46). When DENV-elicited autoantibodies are considered a major pathogenic factor, why chronic vascular permeability and thrombocytopenia are not induced remain a question (47). Through our two-hit model, we posit that neither DENV (first hit) nor autoantibodies (second hit) alone are sufficient to induce the development of DHF. During the febrile phase in which the autoantibody titre has not reached a pathogenic threshold, theoretically, no DHF should occur after the first hit (▶ Figure 5 B [c1]). Regarding the acute (DHF) phase, we hypothesise that only patients who were twice exposed to viral antigens (secondary DENV-infection) have circulating autoantibodies that can potentially reach a pathogenic threshold (▶ Figure 5 B [b1, b3]). These autoantibodies may exacerbate thrombocytopenia, coagulopathy, and inflammation (▶ Figure 5 A [a3, b4]) and contribute to the development of DHF (▶ Figure 5 A [c], B [c, c2]). For unknown reasons, the titre of such autoantibodies (e. g. platelet-associated Ig) decreased considerably in the recovery phase (9) (▶ Figure 5 B [b1]). Thus, during convalescence, DENV-induced inflammation is diminished, and the titre of autoantibodies is reduced, no longer potent enough to induce a pathogenic response (second hit alone; ▶ Figure 5 B [c3]). Thus, no subsequent chronic vascular permeability or thrombocytopenia occurs. According to the ADE hypothesis (3), the viremia titre of DENV can be elicited by infecting leukocytes though virusIg-complexes. Such a response theoretically enhances the impact of the first hit. Consequently, the ADE and two-hit hypotheses are not mutually exclusive. Future clinical studies based on these concepts and involving effectively designed protocols are warranted to further clarify the respective roles of ADE- and DENV-elicited autoantibodies in the pathogenesis of DHF. In the present study, treatments with IVIg and blockers of IL-1 (IL-1RA) and TNF (etanercept) diminished the pathogenic proThrombosis and Haemostasis 113.5/2015

9

10

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis

What is known about this topic?

• • •

Dengue haemorrhage fever (DHF) occurs in a paradoxical phase, in which viremia titres are decreasing and dengue virus (DENV)elicited Igs are elevating. Patients with DHF display antiplatelet and/or platelet-associated Igs and the levels of these antibodies are elevated in severe cases. Antibody-dependent enhancement (ADE) of infection is involved in a number of viral infections. However, only DENVs cause a more severe disease during secondary infections, suggesting that DHF is not completely explained by ADE.

What does this paper add?

• • •

Using a two-hit model of DENV infection, we found that concurrence of DENV and antiplatelet antibodies was sufficient to induce haemorrhage and cytokine surge in mice. Haemorrhage induced by this two-hit model involved activation of the Nlrp3 inflammasome, FcγRIII (CD 16), and Toll-like receptor-3 (TLR3). Interventions targeting the Nlrp3 inflammasome, FcγRIII and TLR3 pathways markedly reduced the severity of haemorrhage.

Acknowledgements

The authors wish to thank Prof. David M Ojcius, University of California, USA, and Chang Gung University, Taiwan; and Prof. Bernhard Ryffel, CNRS, France, for kindly providing the Tlr3-/-, Nlrp3-/- and Casp1-/- mutant mice. The authors are grateful to Dr. Yi-Ling Lin, Institute of Biomedical Sciences, Academia Sinica, Taipei, Taiwan, for her kindly providing DENV strains including DENV2 NGC and DENV4 H241. The authors also want to thank Experimental Animal Center, Tzu-Chi University, for the help on the animal care and experimental consultant. This work was supported by the National Science Council, Republic of China (NSC 101–2320-B-320–004-MY3) and Tzu-Chi University (TCIRP 101001–01–01; TCRPP103001–01). Author contribution

TSL – performed majority of experiments, and data analyses; DSS – designed and performed part of experiments and analysed data; CMC, CYW, MSD, HC, WSW, SHS and YYL– performed part of experiments and analysed data; HHC-original experimental design, data analyses and manuscript drafting. Conflicts of interest

None declared. gression of haemorrhage before and after the second-hit treatment (Suppl. Table 1, available online at www.thrombosis-online.com; ▶ Figure 5 A [a4] and B [b3, b5]). Those reagents, however, can suppress the immune system (36, 48). Unlike in humans, DENV cannot propagate in mice, and therefore, immunosuppression does not exacerbate the outcome of the disease. In contrast, antiviral immunity is crucial for patients. Clinicians must be extremely careful when testing these drugs in a clinical setting. The therapeutic effect of IVIg treatments on the rescue of thrombocytopenia during DENV infections remains controversial, not to mention the role of IVIg on haemorrhage, cytokine surge and the mortality in DHF (49, 50). Theoretically, therapy should be administered to patients after the viral load has declined and before DENV-elicited autoantibodies have increased to a pathogenic threshold (▶ Figure 5 B [c1, b3]). At this stage, the immunity of the patient overpowers the viral amplification, and thus, immunosuppressive treatments might have less impact. In addition, DENV-elicited autoantibodies can serve as measureable risk factors for DHF. With such surveillance, DHF might be treated at an early stage. In summary, this study demonstrates a novel two-hit model for DHF. This model enables us to dissect the pathogenic progression of dengue infection in patients who harbour DENV-elicited autoantibodies. The analyses indicated that the Nlrp3 inflammasome and FcγRIII are involved in the induction of haemorrhage and cytokine surge. Therefore, preventing autoantibodies from emerging can be crucial in developing a feasible DENV vaccine. In addition, DENV-elicited autoantibodies should be considered key parameters for estimating the risk of DHF. These findings may facilitate developing feasible approaches for treating and preventing DHF.

Thrombosis and Haemostasis 113.5/2015

References 1. Simmons CP, Farrar JJ, Nguyen VV, et al. Dengue. N Engl J Med 2012; 366: 1423–1432. 2. Gubler DJ. Dengue and dengue haemorrhagic fever. Clin Microbiol Rev 1998; 11: 480–496. 3. Halstead SB. Pathogenesis of dengue: challenges to molecular biology. Science 1988; 239: 476–481. 4. Sun DS, King CC, Huang HS, et al. Antiplatelet autoantibodies elicited by dengue virus non-structural protein 1 cause thrombocytopenia and mortality in mice. J Thromb Haemost 2007; 5: 2291–2299. 5. Falconar AKI. The dengue virus nonstructural-1 protein (NS1) generates antibodies to common epitopes on human blood clotting, integrin/adhesion proteins and binds to human endothelium cells: potential implications in haemorrhagic fever pathogenesis. Arch Virol 1997; 142: 897–916. 6. Chang HH, Shyu HF, Wang YM, et al. Facilitation of Cell Adhesion by Immobilized Dengue Viral Nonstructural Protein 1 (NS1): Arginine-Glycine-Aspartic Acid Structural Mimicry within the Dengue Viral NS1 Antigen. J Infect Dis 2002; 186: 743–751. 7. Lin CF, Wan SW, Chen MC, et al. Liver injury caused by antibodies against dengue virus nonstructural protein 1 in a murine model. Lab Invest 2008; 88: 1079–1089. 8. Oishi K, Inoue S, Cinco MT, et al. Correlation between increased platelet-associated IgG and thrombocytopenia in secondary dengue virus infections. J Med Virol 2003; 71: 259–264. 9. Saito M, Oishi K, Inoue S, et al. Association of increased platelet-associated immunoglobulins with thrombocytopenia and the severity of disease in secondary dengue virus infections. Clin Exp Immunol 2004; 138: 299–303. 10. Wills BA, Oragui EE, Stephens AC, et al. Coagulation abnormalities in dengue haemorrhagic Fever: serial investigations in 167 Vietnamese children with Dengue shock syndrome. Clin Infect Dis 2002; 35: 277–285. 11. Halstead SB. Dengue. Lancet 2007; 370: 1644–1652. 12. WHO. Dengue: guidelines for diagnosis, treatment, prevention and control. New ed. Geneva, World Health Organisation; 2009, Chapter 1 and 2; pp. 11–28. 13. Andre P, Hartwell D, Hrachovinova I, et al. Pro-coagulant state resulting from high levels of soluble P-selectin in blood. Proc Natl Acad Sci USA 2000; 97: 13835–13840.

© Schattauer 2015

Lien, Sun et al. Two-hit dengue haemorrhage pathogenesis 14. Brozna JP. Shwartzman reaction. Semin Thromb Hemost 1990; 16: 326–332. 15. Gross O, Thomas CJ, Guarda G, et al. The inflammasome: an integrated view. Immunol Rev 2011; 243: 136–151. 16. Wu MF, Chen ST, Yang AH, et al. CLEC5A is critical for dengue virus-induced inflammasome activation in human macrophages. Blood 2013; 121: 95–106. 17. Hottz ED, Lopes JF, Freitas C, et al. Platelets mediate increased endothelium permeability in dengue through NLRP3-inflammasome activation. Blood 2013; 122: 3405–3414. 18. Nimmerjahn F, Ravetch JV. Fcgamma receptors as regulators of immune responses. Nat Rev Immunol 2008; 8: 34–47. 19. Schmidt RE, Gessner JE. Fc receptors and their interaction with complement in autoimmunity. Immunol Lett 2005; 100: 56–67. 20. Chang HH, Kau JH, Lo SJ, et al. Cell-adhesion and morphological changes are not sufficient to support anchorage-dependent cell growth via non-integrin-mediated attachment. Cell Biol Int 2003; 27: 123–133. 21. Kuida K, Lippke JA, Ku G, et al. Altered cytokine export and apoptosis in mice deficient in interleukin-1 beta converting enzyme. Science 1995; 267: 2000–2003. 22. Alexopoulou L, Holt AC, Medzhitov R, et al. Recognition of double-stranded RNA and activation of NF-kappaB by Toll-like receptor 3. Nature 2001; 413: 732–738. 23. Martinon F, Petrilli V, Mayor A, et al. Gout-associated uric acid crystals activate the NALP3 inflammasome. Nature 2006; 440: 237–241. 24. Thomas L, Verlaeten O, Cabie A, et al. Influence of the dengue serotype, previous dengue infection, and plasma viral load on clinical presentation and outcome during a dengue-2 and dengue-4 co-epidemic. Am J Trop Med Hyg 2008; 78: 990–998. 25. Huang HS, Sun DS, Lien TS, et al. Dendritic cells modulate platelet activity in IVIg-mediated amelioration of ITP in mice. Blood 2010; 116: 5002–5009. 26. Subramaniam M, Frenette PS, Saffaripour S, et al. Defects in hemostasis in P-selectin-deficient mice. Blood 1996; 87: 1238–1242. 27. WHO. Dengue haemorrhagic fever: diagnosis, treatment, prevention and control. 2 ed. World Health Organisation 1997, Chapter 2; pp. 12–23. 28. Iyngkaran N, Yadav M, Sinniah M. Augmented inflammatory cytokines in primary dengue infection progressing to shock. Singapore Med J 1995; 36: 218–221. 29. Arias J, Valero N, Mosquera J, et al. Increased expression of cytokines, soluble cytokine receptors, soluble apoptosis ligand and apoptosis in dengue. Virology 2014; 452–453: 42–51. 30. Tsai TT, Chuang YJ, Lin YS, et al. An emerging role for the anti-inflammatory cytokine interleukin-10 in dengue virus infection. J Biomed Sci 2013; 20: 40. 31. Rachman A, Rinaldi I. Coagulopathy in dengue infection and the role of interleukin-6. Acta Med Indon 2006; 38: 105–108. 32. Tsai YT, Chang SY, Lee CN, et al. Human TLR3 recognizes dengue virus and modulates viral replication in vitro. Cell Microbiol 2009; 11: 604–615.

© Schattauer 2015

33. Belge KU, Dayyani F, Horelt A, et al. The proinflammatory CD14+CD16+DR++ monocytes are a major source of TNF. J Immunol 2002; 168: 3536–3542. 34. Ziegler-Heitbrock L. The CD14+ CD16+ blood monocytes: their role in infection and inflammation. J Leukocyte Biol 2007; 81: 584–592. 35. Nimmerjahn F, Ravetch JV. Anti-inflammatory actions of intravenous immunoglobulin. Annu Rev Immunol 2008; 26: 513–533. 36. Park-Min KH, Serbina NV, Yang W, et al. FcgammaRIII-dependent inhibition of interferon-gamma responses mediates suppressive effects of intravenous immune globulin. Immunity 2007; 26: 67–78. 37. Kliks SC, Nimmanitya S, Nisalak A, et al. Evidence that maternal dengue antibodies are important in the development of dengue haemorrhagic fever in infants. Am J Trop Med Hyg 1988; 38: 411–419. 38. Libraty DH, Acosta LP, Tallo V, et al. A prospective nested case-control study of Dengue in infants: rethinking and refining the antibody-dependent enhancement dengue haemorrhagic fever model. PLoS Med 2009; 6: e1000171. 39. Chen HC, Hofman FM, Kung JT, et al. Both virus and tumor necrosis factor alpha are critical for endothelium damage in a mouse model of dengue virus-induced haemorrhage. J Virol 2007; 81: 5518–5526. 40. Rothman AL. Immunity to dengue virus: a tale of original antigenic sin and tropical cytokine storms. Nat Rev Immunol 2011; 11: 532–543. 41. Halstead SB, Mahalingam S, Marovich MA, et al. Intrinsic antibody-dependent enhancement of microbial infection in macrophages: disease regulation by immune complexes. Lancet Infect Dis 2010; 10: 712–722. 42. Scott RM, Nimmannitya S, Bancroft WH, et al. Shock syndrome in primary dengue infections. Am J Trop Med Hyg 1976; 25: 866–874. 43. Chao DY, Lin TH, Hwang KP, et al. 1998 dengue haemorrhagic fever epidemic in Taiwan. Emerg Infect Dis 2004; 10: 552–554. 44. Barnes WJ, Rosen L. Fatal haemorrhagic disease and shock associated with primary dengue infection on a Pacific island. Am J Trop Med Hyg 1974; 23: 495–506. 45. Marchette NJ, Halstead SB, Nash DR, et al. Recovery of dengue viruses from tissues of experimentally infected rhesus monkeys. Appl Microbiol 1972; 24: 328–333. 46. Martina BE, Koraka P, Osterhaus AD. Dengue virus pathogenesis: an integrated view. Clin Microbiol Rev 2009; 22: 564–581. 47. Halstead SB. Controversies in dengue pathogenesis. Paediat Internat Child Health 2012; 32 (Suppl 1): 5–9. 48. Irikura VM, Hirsch E, Hirsh D. Effects of interleukin-1 receptor antagonist overexpression on infection by Listeria monocytogenes. Infect Immun 1999; 67: 1901–1909. 49. Dimaano EM, Saito M, Honda S, et al. Lack of efficacy of high-dose intravenous immunoglobulin treatment of severe thrombocytopenia in patients with secondary dengue virus infection. Am J Trop Med Hyg 2007; 77: 1135–1138. 50. Rajapakse S. Intravenous immunoglobulins in the treatment of dengue illness. Transact Royal Soc Trop Med Hyg 2009; 103: 867–870.

Thrombosis and Haemostasis 113.5/2015

11