The role of endothelial activation in dengue hemorrhagic fever and hantavirus pulmonary syndrome

Abstract

The loss of the endothelium barrier and vascular leakage play a central role in the pathogenesis of hemorrhagic fever viruses. This can be caused either directly by the viral infection and damage of the vascular endothelium, or indirectly by a dysregulated immune response resulting in an excessive activation of the endothelium. This article briefly reviews our knowledge of the importance of the disruption of the vascular endothelial barrier in two severe disease syndromes, dengue hemorrhagic fever and hantavirus pulmonary syndrome. Both viruses cause changes in vascular permeability without damaging the endothelium. Here we focus on our understanding of the virus interaction with the endothelium, the role of the endothelium in the induced pathogenesis, and the possible mechanisms by which each virus causes vascular leakage. Understanding the dynamics between viral infection and the dysregulation of the endothelial cell barrier will help us to define potential therapeutic targets for reducing disease severity.

Keywords: :

• viral hemorrhagic fevers
• dengue
• dengue hemorrhagic fever
• hantavirus
• hantavirus pulmonary syndrome

Introduction

Viral hemorrhagic fevers (VHFs) are a group of illnesses caused by viruses belonging to 4 different families: Flaviviridae, Bunyaviridae, Arenaviridae, and Filoviridae. VHFs are zoonotic diseases transmitted to humans from nonhuman primates, rodents, bats, or arthropod reservoirs. Viruses causing hemorrhagic fevers are globally distributed. While some hemorrhagic fever viruses can cause relatively mild illnesses, many of these viruses cause severe, life-threatening disease with case fatality rates up to 80%. Although VHFs have diverse clinical presentations, they share common characteristic symptoms including fever, hemorrhage, and shock.

One of main target tissues in VHF infections is the microvascular endothelium. The most common clinical finding in severe VHF cases is endothelial dysregulation leading to increased vascular permeability. Therefore the endothelium plays a key role in the development of the disease. The malfunction of the endothelium can be induced either direct by the virus replication or indirectly by virus induced secreted cellular factors that can cause endothelial activation and increase vascular permeability. For example, some of the VHF-causing viruses, such as Ebola, Marburg, and Rift Valley fever viruses, produce a strong cytopathic effect on the infected endothelium, directly affecting permeability. In contrast, arenaviruses, hantavirus, and dengue viruses (DENV) produce little to no cytopathic effect on endothelium. Although induction of cytokines is a contributing factor in increasing vascular permeability in all hemorrhagic fevers, cytokines play a central role in DENV- and hantavirus-induced endothelial dysfunction. Immune activation by the viruses contributes not only to viral clearance but also to the observed immunopathology.

The causes of increased vascular permeability in VHFs have been investigated in previous studies, particularly involving Ebola and Lassa viruses.^1-3 While the mechanisms underlying VHFs in which endothelial cells are directly injured by the viruses are better understood, the mechanisms of endothelial cell dysfunction VHFs in which minimal direct endothelial cell injury are unclear. Although dengue and hantavirus infection are quite distinct in the vascular beds involved in the diseases, they share some pathophysiological features including the lack of overt endothelial injury and the associated complete clinical recovery suggestive of transient functional disturbance of endothelium. Comparing pathology and immunology of these VHFs thus may provide some insights into the underlying disease mechanisms (see Table 1). In this review we will particularly focus on the interaction of these viruses with the vascular endothelium, emphasizing the mechanisms by which they likely cause endothelial cell leakage.

Table 1. Similarities and differences between dengue hemorrhagic fever and hantavirus pulmonary syndrome

	Dengue hemorrhagic fever (DHF)	Hantavirus pulmonary syndrome (HPS)
Agent	Dengue virus (10 kb positive sense RNA genome)	New World hantavirus (11 kb negative sense RNA genome)
Distribution	Asia, Africa, Pacific, Americas	Americas
Transmission	Mosquito bite	Aerosol from rodent excreta; Limited human to human
Mortality rate	1–5%	Up to 40%
Incubation time	4–10 d	14–39 d
Clinical findings	Fever, thrombocytopenia, coagulopathy, plasma leakage	Fever, thrombocytopenia, coagulopathy, plasma leakage
Sites of plasma leakage	Pleural and abdominal cavities	Lungs
Endothelial infection	Unclear (in vivo)	Severe
Endothelial pathology	Minimal	Minimal
Cytokine upregulation	IFN-γ, IL-1, IL-6, IL-8, IL-10, IP-10, MCP-1, VEGF, Angiopoietin-2	IFN-γ, IL-6, IP-10, MCP-1, VEGF
Candidate permeability factors	TNF-α, VEGF	TNF-α, VEGF
Correlation between T cell activation and disease severity	Yes	Yes

Dengue Fever and DENV

Dengue is one of the most important vector-borne diseases worldwide, causing an estimated 5 million cases and over 20 000 fatalities annually. Current endemic areas include Southeast Asia, Central and South America, and parts of Africa.⁴ Recent reports have described major outbreaks of dengue in South Asia and the Middle East.^5-7 A few indigenous dengue cases have been reported in Europe and the United States.^8,9 The expansion of dengue, both geographically and demographically, underscores its importance as a major global emerging public health problem.

DENV consist of 4 serologically distinct, positive-strand RNA viruses belonging to the Flaviviridae family: DENV1, DENV2, DENV3, and DENV4. The 10 kilobase viral genome encodes 10 gene products, including structural proteins capsid (C), matrix (prM), and envelope (E), and nonstructural proteins NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5.¹⁰ The envelope protein plays an important role in viral binding and entry into host cells. The serologic reactivity to the envelope proteins defines the serotypes of DENV. In addition to functions in genome replication, protein processing, and viral particle assembly, some nonstructural proteins have been reported to possess immune evasion properties that may be relevant to DENV pathogenesis.^11,12

Clinical dengue disease

In endemic areas, where multiple DENV co-circulate, primary DENV infection occurs early in childhood, and is usually asymptomatic or causes a mild undifferentiated febrile illness.¹³ Primary infections in older children or adults, as well as secondary infections in all ages, are associated with more severe symptoms. Following an incubation period of 4–10 d after DENV infected mosquito bites, patients develop a high, sustained fever along with other symptoms including anorexia, nausea, headache, joint pain, and skin rash. Common laboratory findings include leucopenia and decreased platelet counts. The febrile period typically lasts 2–7 d. In the majority of cases, defervescence is followed by an uneventful recovery. In a minority of cases, plasma leakage occurs around the time of defervescence, resulting in pleural effusion or ascites. Plasma leakage is the main risk factor associated with disease severity, and the main feature distinguishing dengue fever (DF) from dengue hemorrhagic fever (DHF).^14,15 Hemorrhagic manifestations are common in both DF and DHF, but tend to be more severe in DHF, particularly in cases with severe plasma leakage and shock. Severe hemorrhage occasionally occurs in DF (DF with hemorrhage). Isolated organ involvement in dengue, including encephalopathy, cardiac, and hepatic failure, has also been reported.^14-17 However, the real incidence of isolated organ involvement in the absence of plasma leakage among all dengue cases is not known, since severe organ involvement can also be a consequence of poor organ perfusion associated with shock secondary to severe leakage.

Since the 1960s, clinical dengue has been classified into DF and DHF.¹⁵ Dengue fever is defined as a febrile illness with 2 or more of the following: myalgia, arthralgia, retro-orbital pain, skin rash, hemorrhagic manifestation, and leukopenia. The diagnosis of DF requires laboratory or epidemiological evidence supportive of a dengue virus infection. The case definition of DHF is a febrile illness with thrombocytopenia (<100 000/mm³), hemorrhagic tendency, and plasma leakage. This classification scheme has been criticized for its applicability, usefulness, and reflection of dengue’s natural history. A new classification scheme has been suggested in the 2009 WHO guidelines based on a multicenter study conducted in Asia and the Americas.¹⁸ In this new scheme, dengue is classified as non-severe dengue and severe dengue. The 2009 classification is currently being evaluated for its applicability. Since most studies have classified dengue based on the DF/DHF classification scheme until recently, this review will be based on the earlier classification.

Currently there are no specific treatments for dengue. Supportive treatment including fluid management to correct volume depletion and blood transfusion are the primary treatments. Since plasma leakage is the major cause of disease severity close monitoring for signs of plasma leakage and prompt fluid replacement is key to dengue care. Corticosteroids have been used without clear benefits in most studies in children.^19-21

Risk factors for severe disease

Several epidemiological studies have identified previous exposure to DENV as an important risk factor for DHF when an individual experiences a secondary infection.^13,22,23 It has been postulated that antibodies specific to one DENV serotype fail to provide protection against the second DENV serotype. Rather, they enhance viral uptake and replication though an Fc receptor mediated mechanism. This notion is supported by in vitro and in vivo studies: the levels of virus in tissue culture supernatant and in circulation increase when infection occurs in the presence of DENV-specific antibodies.^24,25 In addition, exaggerated, cross-reactive, cell-mediated immunity during a secondary infection has been documented, and likely contributes to disease severity.⁴

Several host genetic polymorphisms in loci implicated in pathogen entry and host immune responses have been reported. These include DC-SIGN (CD209), HLA class I and class II, transporters associated with antigen presentation (TAPs), cytotoxic lymphocyte antigen 4 (CTLA4), Fc receptor, and soluble factors including TNF-α, TGF-β, and mannose binding lectin-2 (MBL-2).^26,27 It is likely that each genetic locus may be associated with distinct aspects of dengue pathogenesis and the relevance of these polymorphisms to disease severity may vary in different populations.

Although there are currently no studies that precisely delineate viral factors that confer virulence, several outbreaks occurred when a new strain of DENV had been introduced into a new geographical area or when there was an emergence of a new viral strain.^28,29 These outbreaks might have been the result of the lack of host immunity to the virus as well as the intrinsic virulence of the pathogen. Southeast Asian DENV2 that caused a major outbreak in the Americas in 1980–1990 replicated more efficiently in vitro compared with the preexisting American DENV2, and this has been attributed to the differences in the 3′ UTR and the coding regions of the viral genomes.³⁰ Specific mutations in various non-structural genes, such as NS-1 and NS4B, have also been associated with changes in viral replication in vitro and virulence in vivo.³¹

Pathology and pathogenesis

DENV enters the host through the bites of infected mosquitoes. Experimental infection in non-human primates and mice, as well as ex vivo infection of human skin explants, has shown that different types of cells may be infected near the sites of inoculation, including keratinocytes, Langerhans cells, and dendritic cells (DC).^32-34 Various host molecules that are involved in viral binding and uptake have been reported to bind DENV. In particular, C-type lectins, DC-sign (CD209), and CLEC 5A have been shown to interact with the envelope protein and play a role in viral entry.^35-37 Human autopsy studies have demonstrated that the cells infected by DENV (as defined by detectable DENV antigen and/or genome) are mostly cells of the lymphoid and reticuloendothelial system, including monocytes, tissue macrophages, and lymphocytes.^38,39 Other cells reported to express DENV antigen include lung endothelial cells, liver and spleen sinusoidal endothelial cells, and hepatocytes.^38,39

There have been few human autopsy studies that provide systemic, detailed analyses of tissue pathology.⁴⁰ The most common pathological findings in DENV infection are tissue edema and mild perivascular mononuclear infiltration. Hemorrhage is a common autopsy finding in fatal cases, ranging from superficial petechiae of the skin, serosal, and mucosal surfaces, to focal hemorrhage of the gastrointestinal tract, lungs, and central nervous system. Perivascular edema and red blood cell diapedesis are observed, but structural changes of capillaries and postcapillary venules are not observed under routine histology.⁴⁰ Endothelial cell apoptosis has been detected using a molecular technique to detect DNA breakage.⁴¹ However, the extent of the apoptosis appears to be quite limited. Fatty changes of the liver parenchyma to various extent have been reported, but studies are conflicting regarding infection of hepatocytes by DENV.^39,41,42 Periportal inflammation and focal areas of necrosis and apoptosis of liver cells have also been reported.^40,41,43 These findings are consistent with elevated liver enzyme levels commonly observed in dengue patients.⁴⁴

The roles of adaptive and innate immunity

The increased risk of DHF during secondary DENV infection underscores the important role of pre-existing DENV-specific immunity in disease severity. Antibodies specific to DENV from a previous exposure may enhance the uptake and replication of DENV of another serotype, leading to higher viral burden and severity. Consistent with this hypothesis, studies have shown higher viral burden (as measured by levels of viremia, viral RNA, or NS-1 antigen) in plasma of DHF cases compared with DF.^45-47 These enhancing antibodies were recently shown to be largely specific to DENV prM protein.⁴⁸ During an acute secondary DENV infection, a massive expansion of DENV-specific B cells occurs at the time of defervescence; B cells have been shown to harbor the highest levels of DENV among peripheral blood mononuclear cells. Thus, DENV-specific B cells may be an important viral target during secondary infection.⁴⁹ DENV-specific antibodies may also enhance viral replication by an FcR-mediated immune suppression. A monocyte cell line, THP-1, produced IL-10 when infected with DENV immune complexes.⁵⁰ In addition, DENV immune complexes also disrupted RIG-I, MDA-5, and TLR signaling in these cells.^50,51 Overall, B cells and antibodies likely participate in DENV pathogenesis through various mechanisms that lead to increased viral replication and viral burden.

Several studies have demonstrated more intense cellular immune activation in DHF compared with DF cases.^4,52 Increased expression of activation markers and higher frequencies of DENV-specific T cells quantified using tetramer staining and functional assays have been demonstrated in DHF. Similar to B cell responses, memory T cells specific to the previously exposed DENV are preferentially expanded during a secondary infection.⁴ A number of peptide epitopes have been identified and analyses of the responses to these epitopes have shown functional differences in cytolytic activity and patterns of cytokines produced by DENV specific T cells activated with peptide variants.^53,54 Ex vivo analyses of T cells from dengue patients revealed decreased degranulation marker expression (a marker for cytolytic activity) and increased cytokine production when cross-reactive memory T cells were stimulated with epitope peptides of the infecting DENV serotype.⁵⁵ The enhanced cytokine response may play a critical role in the loss of vascular integrity, leading to plasma leakage.

DENV activates innate immunity and induces a cytokine cascade that likely plays both protective and pathogenic roles. Early cytokine response in DENV infection is characterized by the production of type I interferon (IFN). Multiple receptors, including TLR3, MDA5, and RIG-I, are involved in DENV activation of type I IFN response.⁵⁶ Several non-structural DENV proteins interfere with type I IFN signaling through various mechanisms, including degrading STAT2 by NS5, interfering with STAT1 phosphorylation by NS4B, and inhibiting type I IFN production by NS2B3 complex through STING degradation.^11,57-60 The impaired initial control of viral replication may lead to the higher viral burden observed in DHF cases.

Cellular mediators involved in disease severity

Several pro-inflammatory cytokines are produced by DENV-infected cells. DENV-infected monocytes and DC produce a number of mediators, such as interleukin 6 (IL-6), IL-8, and IFN-γ-inducible chemokines CXCL9, 10, and 11, with permeability-enhancing and chemoattractant properties.^61,62 Elevated levels of these mediators have been documented in plasma/serum or pleural effusions of DHF cases.^63,64 Immunomodulatory cytokine IL-10 can be produced by DENV infected monocytes/DC, and elevated IL-10 levels have been reported in DHF patients.^50,52 TNF-α, a permeability-enhancing and pro-coagulation factor, has been extensively studied, but reports on TNF-α levels in severe DENV infection are conflicting.^65,66 Enhanced activation of DENV-specific T cells likely contributes to the elevated cytokine levels in DHF. Multiple cytokines are secreted by DENV-specific T cells, including Th1 type cytokines IL-2, IFN-γ, TNF-α, and various chemokines.^53,67 Elevated levels of both Th1- (IFN-γ) and Th2-type cytokines (IL-4, IL-13, and IL-10) have been reported in DHF.^68,69 The potential contribution of these cytokines in the pathogenesis of DHF can only be speculated from their known biological activities.

The difficulties in interpreting cytokine level data in DENV infection are due to the heterogeneity in study designs, times of sample collection, differences in patient populations, and clinical classification. Further, the lack of animal models that completely reproduce the full spectrum of clinical DENV infection has been a major obstacle in understanding the relative roles of various mediators in DENV pathogenesis. Recent progress has been made using immunocompromised mice and/or mouse-adapted DENV. Some of these models exhibit clinical features similar to human disease, including fever, thrombocytopenia, rash, hemorrhage, and vascular leakage.^70,71 However, pattern of plasma leakage in these models differ from human DHF. Studies using these models have implicated TNF-α in mediating plasma leakage and hemorrhage, possibly through the production of reactive oxygen species.^71,72

Endothelial cells in severe DENV infection

Since plasma leakage is the central feature of severe DENV infection, endothelial cells have been speculated to play a key role in DENV pathogenesis. Autopsy studies in humans have reported perivascular edema with some mononuclear cell infiltration. Changes in endothelial cells, such as edema and pyknosis, were also found.⁴⁰ However, the degree and extent of these changes appeared to be limited. Although DENV antigen has been identified in endothelial cells in the lungs, liver, and spleen in humans and experimental mouse models, conclusive evidence of DENV infection in human endothelial cells is still lacking.^38,39 In vitro, endothelial cells can be infected with DENV, leading to the release of chemokines, including IL-8, RANTES, and monocyte chemoattractant protein-1 (MCP-1).^61,64 Increased cell surface expression of intercellular adhesion molecule-1 (ICAM-1) and β3-integrin, and cytokine receptor VEGF-R2 has been reported.^73,74 Notably, the increase in VEGF-R2 expression was associated with an enhanced response to VEGF by DENV infected endothelial cells.⁷³ In vivo evidence supporting endothelial cell activation includes elevated circulating levels of soluble surface molecules expressed by endothelial cells, such as soluble intercellular adhesion molecule (sICAM-1) and vascular cell adhesion molecule (sVCAM-1).^75,76 Changes in circulating levels of angiogenic factors, including VEGF, its soluble receptors VEGF-R1 and VEGF-R2, and angiopoietin-1 and angiopoietin-2 have been reported to correlate with DENV infection severity.^73,77 In support of this notion, similar “pro-leakage” angiogenic profile; low levels of angiopoietin-1 and elevated levels of angiopoietin-2, have been shown in other conditions with plasma leakage such as sepsis and have been found to correlate with severity and outcomes.^78,79 These changes may reflect the effects of DENV or the immune response on vascular and hemopoietic cells, such as endothelial cells and platelets, which are important sources of these factors. A number of studies have shown that endothelial cell permeability could be altered by interaction with DENV-infected cells, particularly monocytes. In these models, TNF-α, IL-8, and MCP-1 released from monocytes have been implicated as the key mediator in some studies.^80-82 DENV-infected DCs have been demonstrated to secrete matrix metalloproteases (MMPs) that disrupt endothelial cell integrity and enhance permeability.⁸³ Taken together, endothelial cells may be directly affected by DENV or indirectly through the interaction with DENV-infected cells and/or activated immune cells, resulting in the loss in endothelial integrity. A proposed model for dengue pathogenesis is shown in Figure 1.

Figure 1. A model for dengue pathogenesis. Infection of target cells (monocyte, dendritic cells) is enhanced in the presence of non-neutralizing, cross-reactive antibodies. The output viruses and viral proteins such as NS-1 may binds to DENV-specific antibodies and activate complement system leading the release of vasoactive complement fragments resulting functional and structural changes in endothelial cells (EC). Infected cells and activated T lymphocytes release various cytokines with permeability enhancing activities such as IL-6, IL-8, TNF-α, MCP-1, and VEGF. It is controversial whether EC are infected with DENV in vivo. However, in vitro infected EC have been shown to upregulate VEGF-R 2 expression and secrete various cytokines. Activation of EC by VEGF and other cytokines leads to disruption of adherens junctions, resulting in increased permeability.

Many aspects of the coagulation pathway are affected during DENV infection which may predispose to bleeding tendency. Prolonged prothrombin and partial thromboplastin time, elevated levels of tissue factor, thrombomodulin, and decreased levels of Protein C and Protein S have been reported in severe dengue cases.^84,85 These changes, in conjunction with reported increase in levels of active form of von Willebrand factor (VWF) and D-dimer, are consistent with an abnormal activation of the coagulation and fibrinolysis system.^84,86 The mechanisms underlying these changes are complex and multiple. DENV-infected HUVEC upregulated tissue factor expression and downregulated thrombomodulin expression and protein C activity.^87-89 Activation of endothelial cells either by DENV or by mediators such as TNF-α is likely the mechanism associated with the changes in coagulation in dengue.

A series of studies have demonstrated that DENV induces autoantibodies against endothelial cells, platelets, and molecules in the coagulation and fibrinolytic pathways in humans and experimental animals.^90-92 The majority of these antibodies react to NS1 antigen, although some were E- and prM-specific.⁹³ These antibodies induce abnormal activation and function of platelets and endothelial cells, leading to endothelial cell apoptosis in vitro, and to mortality in experimental animals.^90,92 It is unclear how long these antibodies persist after an acute DENV infection and how clinically important these antibodies are in DENV pathogenesis, since patients usually have a complete recovery without autoimmune manifestations as long-term complications, particularly in children. More definitive studies will be needed to clarify the role of autoantibodies in DENV pathogenesis.

Hantaviruses

Another group of hemorrhagic fever viruses associated with endothelial dysfunction is the hantaviruses. Hantaviruses are distributed worldwide, and are broadly split into Old World hantaviruses, which cause hemorrhagic fever with renal syndrome (HFRS), and the New World hantaviruses, which cause HPS. HFRS-causing hantaviruses are mainly present in Asia and Europe, and have a case fatality of 0.1–15%. HPS-causing hantaviruses are found throughout the Americas, and have much higher case fatality rates of 35–40%.⁹⁴ Old World hantaviruses were discovered to cause HFRS around the time of the Korean War, although the first hantavirus was not isolated until 1978. However, hantaviruses were not known to cause human disease in the Americas until Sin Nombre virus—which means “no name virus” in Spanish—was first identified as the cause of an outbreak of acute respiratory distress syndrome in the southwestern United States in 1993.⁹⁵ Since then, HPS has been an emerging infectious disease in the Americas, and more than 20 different HPS-associated hantaviruses have been identified.⁹⁶ The virus causing the highest burden of disease in South America is Andes virus (ANDV), the only known hantavirus implicated in human-to-human transmission.^97,98 Hantaviruses are transmitted to humans by inhalation of infectious aerosols from rodent excreta.⁹⁹ Although a rare disease with a total of a few thousand cases reported, HPS remains in the spotlight because of the high case fatality rates. For example, a recent HPS outbreak in Yosemite park in California, in the summer of 2012, although modest in size with 10 confirmed cases and 3 deaths, caused international alarm, and hundreds of thousands of international visitors had to be notified by the National Park Services.¹⁰⁰ Currently, no FDA-approved immunotherapeutics, antivirals, or vaccines are available for use against HPS.¹⁰¹

The hallmarks of both HFRS and HPS are increased vascular permeability and thrombocytopenia.^96,102-104 In both syndromes, patients may have pulmonary and/or renal involvement, though pulmonary involvement dominates in HPS, and renal involvement in HFRS. For the rest of the review, we will mainly focus on HPS-causing hantaviruses.

Hantaviruses have a single-stranded, negative-sense RNA genome consisting of 3 RNA segments: small (S), medium (M), and large (L). The term “negative-sense” refers to an RNA genome complementary to the mRNA encoding the viral proteins. The S segment varies from 1600–2004 nucleotides in length and encodes the viral nucleoprotein (N). The M RNA segment of hantaviruses is ~3600 nucleotides long, and encodes the mature Gn and Gc glycoproteins. The hantavirus L segment is approximately 6500 nucleotides, and encodes the virus L polymerase protein.^94,105 The naked RNA genome is not infectious; an infectious virus unit requires the viral RNA to be encapsidated by the N protein, forming the nucleocapsid core associated with the viral L polymerase protein. To form a mature virus, this unit is then enveloped by the host cell membrane. Embedded in the membrane via the cytoplasmic tail of Gn are the two viral surface glycoproteins, Gn and Gc, which mediate viral attachment to cells.^106,107

Hantavirus pulmonary syndrome

Like many acute viral infections, the initial symptoms for HPS are reported as a flu-like illness, with fever, myalgia, and headache, followed by severe gastrointestinal symptoms, such as abdominal pain, vomiting and diarrhea. This prodromal phase of HPS lasts around 5 d, during which these cases are difficult to distinguish from other viral infections. The second phase of the disease is very rapid and can progress over a few hours to acute pulmonary edema, severe respiratory syndrome, and shock. Patients require mechanical ventilation within the first 24 h. Extracorporeal membrane oxygenation support has been shown to improve the survival rates for patients with severe HPS.¹⁰⁸

The severe clinical impact of HPS is on the lungs, and is typically manifested as rapidly developing, diffuse non-cardiogenic pulmonary edema, followed in fatal cases by cardiogenic shock.^102,109 The pulmonary manifestations are similar among hantaviruses, but differ subtly in presentation among diseases caused by the North and South American hantaviruses.^102,110-114 In addition to the pulmonary symptoms, South American hantaviruses often present with some hemorrhagic and renal manifistations,^{111,112,114-116} and viral antigens and infectious virus can also be detected in urine of HPS patients infected with ANDV.¹¹⁷ In contrast to HPS caused by ANDV, however, hemorrhagic and renal symptoms are not a common characteristic of HPS caused by Sin Nombre virus.

Immune modulation in hantavirus pulmonary syndrome

In general, increased vascular permeability is an important component of severe disease progression in VHFs.¹¹⁸ A number of studies have investigated the cause of increased vascular permeability in VHFs induced by viruses such as DENV or Ebola virus.^2,3,119-123 The major hantavirus target is the microvascular endothelium, and severe human disease has been attributed to microvascular leakage. Initially, the innate immune system recognizes the virus, and proinflammatory cytokines and chemokines are induced. This is followed by the adaptive immune system, which consists of cytotoxic T cells and B cells that produce virus-specific antibodies. Although these responses are crucial for viral clearance, they can also damage the host if not properly regulated. It is a long-standing belief that the induction of an uncontrolled immune response to hantavirus infection and the generation of a cytokine storm, rather than the viral infection per se, causes microvascular leakage and HPS.^103,124-126 Evidence supporting the hypothesis that HPS is an immune-modulated disease is outlined below.

Histopathological examination of multiple organs has revealed the presence of small and enlarged mononuclear cell infiltrates, but no evidence of cell destruction despite prominent accumulation of viral antigen in the infected vascular endothelium. The major histopathological features in fatal HPS cases are observed mainly in the lung, and the infected lung macrovascular endothelium is uniformly infected with high viral titers.¹⁰³ Hantaviruses infect endothelial cells in various tissues throughout the body in addition to the lung endothelium.^95,103 Large immunoblasts are in circulation at the onset of HPS, and while they are primarily concentrated in the lungs, they are also observed within the red pulp of the spleen and other lymphoid organs.^103,127 Hantaviruses can also infect other human cell types, including monocytes and macrophages^128,129 and DC¹³⁰ without apparent cytopathic effects in vitro.

Hantaviruses induce a long-lasting humoral immune response. Initially, antibodies against hantavirus antigens are present in HPS patient sera at the time of disease onset, followed by neutralizing antibodies directed against the viral Gn glycoprotein.^131,132 Production of IgM-specific antibodies occurs early during the clinical course of the disease followed by an increase in IgG antibody titers. However, unlike in DENV infection, the presence of anti-hantavirus antibodies does not correlate with an increased risk of severe disease. On the contrary, the presence of high-titer neutralizing antibodies in patient sera correlates with favorable disease outcomes.^133,134

High levels of cytokines have been detected in HPS patients.^126,135 Based on early immunocytochemical studies, several cytokines are detected, especially in lung tissues from patients with fatal HPS,¹³⁵ including those potentially produced by T cells, such as TNF-α, IL-2, IL-6, and IFN-γ. Both TNF-α and IL-2 can increase vascular permeability. In vitro studies have also shown the induction of chemokines, such as MCP-1, RANTES, and IP-10 in infected vascular endothelium.^136-139 These chemokines can recruit immune cell infiltrates into the lung and other organs. Primary human DC infected with hantaviruses in vitro are quickly activated and secrete pro-inflammatory cytokines like TNF-α and an active form of MMP9.¹⁴⁰ These immune cells can then release more cytokines, intensifying the cytokine storm, further increasing vascular permeability, and inducing pulmonary edema.

Virus-specific CD8⁺ T cells correlate with disease severity. A robust T cell response is generated in humans during hantavirus infection, and is followed by a long-lived memory T cell response.¹⁴¹ Several CD4⁺ and CD8⁺ T cell clones obtained from the blood of acute and convalescent hantavirus HPS patients recognize epitopes identified in hantavirus N, Gn, and Gc proteins.^141-143 Surprisingly, the most immunodominant epitopes are located in the Gn glycoprotein.¹⁴¹ Higher frequencies of virus-specific CD8⁺ T cells were detected in PBMCs from HPS patients than in patients with other virus infections, like DENV or influenza virus. Most importantly, these frequencies were higher in patients with severe HPS (hospitalized patients that required mechanical ventilation) than in patients with moderate disease (hospitalized patients that did not require mechanical ventilation).¹⁴³ These are strong indications that virus-specific CD8⁺ T cells play a role in virus-induced immunopathology.

Genetic factors associated with clinical outcomes in hantavirus infection have also been investigated in a few studies. Specific HLA alleles have been correlated with either mild or severe disease. In HPS patients infected with Sin Nombre virus, the HLA-B*3501 allele has been associated with severe disease.¹⁴³ However, in a more recent study of ANDV-infected patients, the HLA-B*35 allele was detected more frequently in patients with rather mild disease than those with severe HPS.¹⁴¹ In addition, the polymorphic TNF-2 allele (308G/A) has been associated as a high-risk factor for developing HPS, but no connection has been found so far regarding this allele and disease severity.¹⁴⁴

Direct effects of hantavirus infection on the vascular endothelium

So far we have highlighted that hantavirus-infected vascular endothelium orchestrates the induced immune pathology by secreting chemokines that attract the monocytes, macrophages, and T cells to the vicinity of the virus. However, more recent studies show that hantavirus replication alone can compromise the integrity of the endothelial cell barrier.¹⁴⁵ Most importantly, hantavirus infection can sensitize the endothelium and cause vascular hyper-permeability in response to permeability factors like vascular endothelial growth factor (VEGF).^146-148

Endothelial cell permeability is a highly regulated process and is maintained by both tight and adherens junctions. A transient disruption of adherens junctions is sufficient to disturb endothelial barrier function and increase permeability, leading to edema.^149-151 Vascular endothelial cadherin (VE-cadherin) is an adherens junction adhesion protein that plays a central role in maintaining the vascular barrier.¹⁵² In order to maintain the endothelial cell barrier, VE-cadherin interacts with the tyrosine kinase receptor VEGF-R2. However, when VEGF, a potent inducer of vascular permeability, is present, it binds to VEGF-R2, disassociating VE-cadherin and VEGF-R2.^153,154 This can initiate a junction-dependent increase in permeability by a clathrin-dependent internalization and degradation of VE-cadherin and disruption of the adherens junctions. Infection of human primary lung endothelial cells with HPS-associated hantaviruses revealed a moderate early increase in secreted VEGF and a concomitant decrease in VE-cadherin.¹⁴⁵ This resulted in increased vascular permeability and loss of integrity of the endothelial cell barrier. Consistent with this process, antibody blockage of VEGF-R2 activation inhibited hantavirus-induced VE-cadherin reduction.¹⁴⁵

The most prominent cellular response to hantavirus infection, however, is the induction of the hyper-permeability of the microvascular endothelium. In vitro experiments have shown that adding VEGF to primary endothelial cell monolayers infected with pathogenic hantaviruses activated VEGF-R2 and Src kinase, and led to VE-cadherin internalization and degradation.^147,155,156 Adding factors such as Angiopoietin-1 and S1P, which inhibit VEGF-R2 induced permeability, blocked VE-cadherin internalization in response to VEGF.^147,155-157 In vitro studies have also shown that small molecule inhibitors which block the VEGFR2 or Src kinase, such as pazopanib and dasatinib, can block ANDV-induced hyper-permeability.¹⁵⁶ Increased amounts of VEGF in the serum of HPS patients have been reported.^145,148

The results of the above studies suggest that agents that can block the activation of VEGF-R2 and Src kinases may be efficacious HPS treatments. Currently, FDA-approved small molecule inhibitors, particularly tyrosine kinase inhibitors such as vandetanib, are being used in studies utilizing the Syrian golden hamster HPS disease model to test such hypotheses.¹⁰¹

A model of the activation of vascular endothelium by hantavirus infection

Increased vascular permeability observed in infections with pathogenic hantaviruses is certainly a multifactorial event. A strong immune response and release of permeability factors can contribute to increased vascular permeability and HPS induction. Viral infection of the endothelium can also initiate vascular leakage and the cascade of cell signaling events leading to hyper-permeability and disease. The following is our current model of what happens in the lung of an HPS patient based on the synthesis of our current knowledge of in vitro and animal studies (see Fig. 2).

Figure 2. Proposed model for HPS pathogenesis. The exact means of entry of hantaviruses into the vascular endothelium is not known, but likely is via infected dendritic cells and/or infected alveolar macrophages (A). Infection of endothelial cells by hantavirus causes secretion of VEGF, triggering the disruption of adherens junctions and downregulation of VE-cadherin (see details in [D]). Hantavirus-infected endothelial cells also produce proinflammatory cytokines and chemokines, such as IP-10 and RANTES, and upregulate adhesion molecules on their cell surface, attracting monocytes, macrophages, and T cells. Accumulation of hantavirus infected monocytes and macrophages in the vicinity of the endothelium results in a “cytokine storm” by secreting additional chemokines/cytokines. Additional VEGF is secreted by hantavirus activated T cells, platelets, and macrophages. At this point VEGF could achieve high concentrations in the microvasculature of the lung, resulting in vascular hyper permeability and leakage (B). (C) Diagram of intact adherens junctions. (D) Diagram of disrupted adherens junctions.

First, hantaviruses infect the respiratory epithelium and/or the DC residing in the respiratory epithelium.¹³⁰ This is followed by viral dissemination, via infected DC, to monocytes and macrophages in the lymph nodes, and via the respiratory epithelium to the respiratory endothelium.^130,158 Efficient viral replication and spread may occur early after infection due to viral inhibition of innate immune responses.^159,160 Hantavirus-infected endothelial cells produce proinflammatory chemokines and upregulate adhesion molecules on their cell surface, attracting monocytes, macrophages, and T cells.^136,137,139 Infected monocytes and macrophages, in turn, produce additional chemokines, cytokines, and other proinflammatory molecules, like TNF-α and reactive oxygen/nitrogen species.^{126,128,161-164} Viral infection of the vascular endothelium and early VEGF secretion triggers the disruption of adherens junctions and downregulation of VE-cadherin. VEGF is secreted by virus-activated T cells, platelets, and macrophages in vivo.^124,165-167 Under these conditions, permeability factors (TNF-α and/or VEGF) from all these sources reach high concentrations in the microvasculature of the lung, resulting in vascular hyper-permeability and disease (Fig. 2).

Problems and Future Directions

Although DENV and hantaviruses are very distinct, they both cause VHF with notable similarities. The first similarity is the dysregulation the vascular endothelium, leading to increased macrovascular permeability and plasma extravasation. The increased vascular permeability is not based on a strong cytopathic effect on endothelial cells. Second, cell-mediated immunity contributes to disease severity; particularly, CD8⁺ T cells are likely play an important role in the pathogenesis of both infections. Finally, pro-inflammatory and permeability-promoting factors are produced secondary to a strong immune activation in infection with both viruses.

The continued expansion of the geographic and demographic range of dengue fever and HPS highlights the important global health threat posed by these viruses. The lack of specific treatments for these diseases underscores the urgent need for better insights into mechanisms underlying severe disease and protection. Additional in vitro studies are needed to identify the precise steps leading to virus-induced endothelial cell leakage. Although much progress has been made using animal studies, the applicability of the findings from these models to human disease is uncertain. In the case of DENV, better animal models that closely mimic human disease are still needed. For HPS, the Syrian golden hamster model recapitulates human disease quite well, but the disease is uniformly lethal, as is only seen with ANDV and not with other HPS viruses. Prospective studies which include well-characterized patients encompassing different genetic and demographic backgrounds are a prerequisite for further understanding the pathogenesis of both viruses. The application of novel multi-parametric assays and novel molecular platforms will facilitate more comprehensive and global analysis of the contribution of viral and host factors to disease development. Insights gained from such studies will be critical in developing vaccines and therapeutic interventions for these diseases.

Acknowledgments

The findings and conclusions in this report are those of the authors and do not necessarily represent those of the Centers for Disease Control and Prevention. The authors thank Dr T Klimova and Dr In-Kyu Yoon for critical reading of the manuscript, and Craig Manning for help with the illustration. Part of this work was supported by National Institutes of Health Grant NIH-P01AI34533. The opinions or assertions contained herein are the private ones of the authors and are not to be construed as official or reflecting the view of the US Government.

Submitted

04/15/13

Revised

06/25/13

Accepted

06/27/13

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.