Introduction

Dengue virus (DENV) is the leading cause of arthropod-borne human disease worldwide, with an estimated 50–100 million individuals infected annually and with 2.5 billion people at risk. The virus is endemic to over 100 countries, mostly in tropical and subtropical urban areas, where the primary vector Aedes aegypti is found [1]. The virus is spread to a lesser extent by Aedes albopictus mosquito (which has a broader temperature and geographical range than Aedes aegypti). Members of the family Flaviviridae, dengue viruses are comprised of four genetically related serotypes of viruses, DENV 1–4. The viral genome is approximately 10 kB in length and is composed of 10 genes, three that encode the structural proteins capsid (C), premembrane (prM) that is later cleaved during viral development to membrane (M) and envelope (E). The remaining seven genes code for the nonstructural proteins NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5. Asymptomatic in most infected individuals, dengue can present as dengue fever (DF), the more severe dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS). The World Health Organization (WHO) is currently working to recategorize the spectrum of dengue disease to dengue with or without warning signs and severe dengue to facilitate easier clinical diagnosis [1]. This recategorization has not been fully adopted by the scientists performing basic research [2, 3], and as the majority of the literature uses the older classification system, we will be using DF, DHF and DSS in this review. DF is a febrile illness characterized by high fever, rash, severe joint paint, retro-orbital pain and nausea [4], comprising the vast majority of dengue infection-related disease. DHF and DSS usually present in a secondary infection and are characterized by more severe manifestations including hemorrhage, thrombocytopenia and/or shock [4] and can result in death in approximately 2.5 % of the cases according to the WHO [5].

DENV has been shown to infect many cells of immune system (mainly of myeloid origin) and related organs and is introduced to the human body upon a blood meal by an infected mosquito. The first cells to encounter the virus are believed to be Langerhans cell in the skin [6], macrophages [7] and dendritic cells (DCs) [8] and causes viremia in the patients, allowing transmission to the next host by mosquitoes. We have data showing that all PBMCs are capable of being productively infected by DENV and produce infectious particles, but show different kinetics of infection, suggesting a specific hierarchy and chronology of infection for different cells in humans (unpublished data).

DENV blocks the type I IFN system

The signaling system of type I interferon (IFN) is integral to the innate immune system’s ability to create an antiviral state. Viral pathogen-associated molecular patterns (PAMPs) are recognized by intracellular molecules including Toll-like receptors (TLRs), NOD-like receptors (NLRs) and RIG-I like receptors (RLRs), leading to a downward signal transduction cascade that results in the transport of NFκB, IRF3 and ATF2/JUN to the nucleus to activate the gene of IFN beta (Fig. 1a). Type I IFN is able to signal in an autocrine and paracrine manner. IFNα/β binds to the IFNα receptor leading to the activation of the Janus kinases, Jak1 and Tyk2, through tyrosine phosphorylation. These kinases then phosphorylate STAT1 and STAT2 that dimerize to form a transcription factor that turns IFN-stimulated genes (ISGs) by binding to ISREs (IFN-stimulated response elements; Fig. 1b). ISGs are responsible for turning on many antiviral properties (as reviewed in [9]). Treatment of cells with IFN alpha, beta or gamma before infection with dengue virus protects the cells from infection, though IFN gamma’s protective abilities were variable [10]. Thus, in order to mount a successful infection, dengue virus must circumvent the type I interferon system. At MSSM, our group is focused on looking at prevention of interferon production in primary cells and the García-Sastre group, also in the department of Microbiology, investigates dengue suppression of IFN signaling.

Fig. 1
figure 1

DENV blocks type I IFN production and signaling in infected cells. a Type I IFN production. In MDDCs dsRNA, a replication product of DENV, is detected either by TLR3 present in the endosomal membrane or by RIG-I or MDA5, helicases present in the cytoplasm. The detection of dsRNA activates of signal transduction pathways that lead to the translocation of NFκB, IRF3, p38 and JNK to the nucleus where they form a transcription factor complex that activates the transcription of IFNβ. Work from the Fernandez-Sesma laboratory suggests that DENV NS2B3 blocks IFN production at the level of IRF3 phosphorylation. b Type I IFN signaling. IFNα/β binds to the receptors IFNaR1 and IFNaR2 on the surface of cells that activate the Janus kinases Jak1 and Tyk2 which in turn phosphorylate STAT1 and STAT2. Phosphorylated STAT1 and STAT2 form a transcription factor complex with IRF9, which travels to the nucleus where it binds to an ISRE to turn on the transcription of ISGs. Dengue NS5 causes the degradation of STAT2, which prevents the formation of the complex, transcription of ISGs and establishment of an antiviral state

Full size image

Antagonism of type I interferon production

Upon infection with dengue virus, monocyte-derived dendritic cells (MDDCs) express many proinflammatory cytokines including IFI56K, IL-8, MIP-1 beta, RIG-I, TNF alpha, CD86 and STAT-1 [8]. The Fernandez-Sesma laboratory was the first to show that type I IFN is not induced during infection by DENV in MDDCs [8]. Interestingly, our group also showed that even plasmacytoid dendritic cells (pDCs), an exquisitely sensitive cell type to viruses that produce large quantities of IFN alpha upon stimulation, do not express IFN alpha when infected with DENV [8]. Furthermore, we showed that once infected by DENV, DCs are also unable to produce IFN upon stimulation by strong inducers of type I IFN such as infections by Newcastle disease virus, Semliki Forrest virus, and Sendai virus as well as treatment with PolyI:C (an inducer of Toll-like receptor) suggesting that DENV is a strong inhibitor of IFN. We also showed that the inhibition of type I IFN production by DENV in DCs makes them very inefficient at priming T cells to a Th1 type cell (as measured by IFN gamma ELISA) [8]. These data from our laboratory strongly suggest that early events induced in human DCs by DENV may be determinant for the cross-reactive and inefficient adaptive immunity observed in DENV infected patients. Co-expressing individual dengue proteins and protein complexes in 293-T cells that express luciferase behind an IFN alpha/beta promoter, our group also demonstrated that the DENV protease (NS2B3) is required for the inhibition of type I IFN production in infected cells [11]. We are currently investigating the targets of the NS2B3 protease, paying specific attention to molecules involved in the IFN pathway that could possibly be involved in the antagonism of IFN (Fig. 1a).

In addition to DENV actively inhibiting IFN production, our laboratory is interested in how dengue can passively evade the immune system. DENV induces intracellular membrane structures such as vesicle pockets (VP), which are bound to the ER and have a pore through to the cytoplasm [12–14]. The interior of the VPs contain products of replication including nonstructural dengue proteins, de novo synthesized RNA and double-stranded RNA. Not only do these membrane compartments concentrate viral products, they may also allow for sufficient amounts of replication to occur before any PAMPs are detected by the host cells PRRs. Other flaviviruses also utilize this process for immune evasion [12, 14]. Our laboratory is currently investigating how much of the inhibition of the IFN production is through active inhibition by the NS2B3 complex or through passive inhibition by hiding replication products in membrane vesicles.

Dengue inhibits type I interferon signaling

In addition to antagonizing the production of type I IFN, DENV also inhibits its signaling. A report by Ho et al. [15] demonstrated that DENV can block signaling of IFNα/β, but it cannot block IFNγ signaling through STAT2 down-regulation. Treatment with type I IFN prior to infection with DENV prevents infection with the virus [10, 16], suggesting that IFNs are very efficient at controlling this pathogen and that the virus is likely to dedicate some of its proteins to antagonize this important innate immune system. Extensive work has been done by the García-Sastre group at Mount Sinai School of Medicine to elucidate how DENV is able to thwart the host IFN signaling. They have identified NS2A, NS4A, NS4B and NS5 as able to block IFN signaling (Fig. 1b); expression of these proteins in cells that were then infected with IFN-sensitive viruses rescued replication of those viruses, showing that these nonstructural proteins were responsible for preventing IFN signaling [17]. Interestingly, expression of NS4B alone resulted in down-regulation in the expression of interferon-stimulated genes (ISG), suggesting that this protein by itself is a potent IFN antagonist [17, 18]. They went on to show that NS4B is able to block IFN signaling by down-regulating STAT1 expression [18].

Additionally, the Garcia-Sastre group investigated the IFN antagonistic properties of some of the other DENV nonstructural proteins identified in their earlier screenings, such as the NS5. It was shown by another group that IFN alpha signaling is decreased in DENV infected cells through down-regulation of STAT2 [19], indicating that another protein, different from NS4B, may act on STAT2. The Garcia-Sastre group showed that binding of NS5 to STAT2 is required for the observed decreased levels of STAT2 in an ubiquitin and proteasome-dependent process (Fig. 1b) [20]. Interestingly, they showed in a subsequent study that dengue NS5 is unable to degrade mouse STAT2 and can only degrade human STAT2, strongly suggesting that STAT2 could be a restriction factor for DENV infection in mice [21]. This may explain why establishing a dengue mouse model has been so difficult for the DENV research community. If NS5 is unable to cause the degradation of STAT2, then IFN signaling is able to take place, creating an antiviral state and preventing the replication of dengue virus. Currently, the majority of dengue mouse models, such as the AG129, used are immunocompromised, many with the type I IFN system knocked out [22]. Using transgenic technology, a better, immunocompetent mouse model might be created with by replacing mouse STAT2 with human STAT2 [21].

Severity of dengue disease

Our laboratory is also very interested in finding out what causes severe dengue disease manifestation in humans. DENV is unique in that the virus can induce a more severe disease upon subsequent infections, implying that the patient’s own immune response to the first infection may determine the severity of subsequent secondary infections by other serotypes of DENV. While many infections with DENV are asymptomatic [23, 24], patients can also develop DF, DHF or DSS. DF is characterized by arthralgia, rash, high fever, nausea, retro-orbital pain and headache [1, 4]. Severe DENV usually manifests upon a secondary infection of the virus, though there are exceptions. Ten percent of DHF infections are primary infections, and almost all of these cases are infants born to mothers who previously had dengue [25]. DHF is characterized as an extremely severe febrile illness with thrombocytopenia, petechiae and bruising, and dengue shock syndrome is even more severe with shock manifestations due to vasculature leakage [1, 4, 26]. Patients with DHF also have a 10–100 times higher viremia than patients with DF [27–29] as well as other markers of enhanced disease.

There are several different theories to explain how severe cases of DENV can arise including antibody-dependent enhancement and T-cell-mediated immunity and these will be discussed in this section.

Antibody-dependent enhancement

The phenomenon of antibody-dependent enhancement was first described by Sabin on work performed on the virus during World War II [30] and extensively studied by Halstead in the 1960 and 1970s in a group of small children in Bangkok [31]. Currently, many studies are being performed in vitro to fully elucidate this occurrence. During a primary infection, antibodies are generated against that particular serotype. After a brief period of time when these antibodies are able to neutralize the viruses, these antibodies remain cross-reactive for other serotypes and facilitate entry into the cell. In vitro work has demonstrated that heterotypic antibodies are able to neutralize at high titers, but as concentrations decrease, the neutralization threshold has been lost and entry is enhanced [32–34]. Cross-reactive antibodies are able to bind to the virus and also bind to Fcγ I and II receptors [32, 35], and Fc receptor-mediated endocytosis is faster and more efficient than normal entry into the host cell via the yet undescribed viral entry receptor. Faster entry into the cell allows for more rapid replication and higher production of infectious particles. Several cell types that are primary targets for DENV support enhanced entry including monocytes, macrophages and mature dendritic cells [36]. Other cells such as immature dendritic cells do not have enhanced entry in the presence of antibodies but demonstrate high infectivity by DENV without antibodies [36]. Many of the antibodies generated against DENV antigens are against the E protein, the most outwardly exposed protein in the virion, and there is mounting evidence that the antibodies involved ADE are mostly against the prM protein of immature virions that have failed to mature as they exit the host cell [37]. In vitro studies have demonstrated that anti-prM antibodies facilitate binding and entry of immature virions, and these virions are able to replicate [38]. To complicate matters, it has been proposed that neutralizing antibodies, when titers decrease, could become enhancing [39, 40]. This ADE phenomenon makes it very challenging to develop efficient vaccines against DENV, since the generation of the wrong type of antibodies in the vaccines may make them more susceptible to sever dengue disease. Several ongoing trials for vaccine development are trying to address this issue by using different strategies of vaccination [41]. The Fernandez-Sesma laboratory is currently exploring the influence of human antibodies to different DENV in the ability of other serotypes of the virus to infect primary human cells (such as monocytes, MDDCs and MDDM) and to antagonize the IFN response in those cells.

T-cell-mediated immunity

Another immune factor contributing to disease severity is the T-cell response. There is some evidence in dengue patients indicating that T cells react quickly and most likely enhance severity of disease; DHF patients have high levels of soluble IL-2R, soluble CD8 and soluble CD4 [29, 42, 43] and as well as higher levels of activated T cells and some cytokines compared with patients with other febrile illnesses and patients with dengue fever in the case of elevated cytokine levels [44, 45]. After an infection with DENV of a particular serotype, mice and human patients have been reported to produce cross-reactive CD4+ and CD8+ T cells against the other serotypes [46–48]. Cross-reactive murine T cells were shown to be against nonstructural DENV proteins, while serotype-specific T cells were against structural DENV proteins [48–51], suggesting a role for cross-reactive T-cell responses in severity of dengue disease. T-cell epitopes have been identified in both mice and humans against almost all of the DENV proteins; however, the majority of these epitopes are against either the E protein and more abundantly against NS1, NS2A and NS3 [48–50, 52, 53]. The NS3 epitopes have been identified to be important target of CD8+ T cells in DHF patients [52]. More studies need to be performed in order to identify what epitopes are important for protective responses in DENV infections, but limitations due to the lack of an immunocompetent small animal model have made this kind of research very difficult.

Intrinsic viral properties

There is also mounting evidence suggesting that distinct viral properties can influence the severity of dengue disease; however, it can be difficult to fully ascertain specific strain properties when most areas that have endemic DENV have multiple co-circulating serotypes. Studies by Rico-Hesse et al. have demonstrated that in the US, the circulating DENV-2 American genotype, which had no link to DHF and DSS, was replaced with a Southeast Asian (SEA), which was associated with DHF [54–56]. Recent work in Cambodia demonstrates that clinical isolates from children with DSS were phenotypically distinct and replicated less efficiently in mammalian cell culture than isolates from patients presenting with DF or DHF from the same outbreak [57]. In Sri Lanka, DHF was extremely uncommon before 1989; however, in 1989, DHF emerged. At the time of the onset of DHF, there had been neither an increase in transmission of the virus nor a change in the circulating serotypes [58]. Phylogenetic analysis of clinical isolates revealed a clade replacement in the circulating DENV-3 viruses from a subtype group A virus to a group B virus, which accounted for change in disease severity [58, 59]. Our laboratory is currently working with viruses from Sri Lanka from patients with DF (before 1989) or DHF (after 1989) in collaboration with the de Silva laboratory at the University of North Carolina, Chapel Hill. We are investigating if viruses associated with DHF will induce stronger immune responses in a primary infection conditions with monocytes and MDDCs that could suggest that intrinsic viral properties can induce severe disease. Our preliminary data show that in a primary infection of MDDCs, the DHF-associated viruses induce higher levels of cytokines and chemokines (unpublished data, Fig. 2).

Fig. 2
figure 2

Proposed model of how intrinsic DENV properties may induce different immune responses in MDDCs, leading to more severe disease manifestation. a MDDCs infected with DF-associated DENV do not produce type I IFN but do express cytokines and chemokines. b MDDCs infected with a DHF-associated DENV produce type I IFN as well as increased cytokines and chemokines

Full size image

Host genetic determinants

Additionally, there are host factors that also play a role in disease severity. There are several genetic factors that have been correlated with more severe disease including HLA alleles [60–63], polymorphisms in TNF alpha [64], LT alpha [64] and the promoter of CD209 [65]. These facts may explain the differences in distribution of DHF cases in different areas of the world where all serotypes of DENV co-circulate, like the low incidence of sever dengue disease reported in Africa as compared to Asia or Central and South America [66].

Conclusions

DENV is an important pathogen to study based on the wide distribution of the disease worldwide and in addition to the spectrum of disease severity it causes. Dengue disease is multifactorial, and it has been difficult to study due to the lack of tools and immune competent small animal models. Currently, several groups, including ours, are focusing on primary human cells to try to identify the specific viral and host factors that contribute to the severity of dengue disease. Identifying such factors will facilitate the development of efficient antivirals and vaccines for DENV.

Considerations for enhanced severity of disease have been vital in the creation of a tetravalent DENV virus vaccine. If the vaccine does not provide complete, lifelong immunity to all four serotypes of the virus, the vaccine could then potentially enhance infection of the waning serotype. Thus, ongoing research is being performed by several groups to create a strong vaccine and several candidates are in different phases of clinical trials.