Protection against dengue virus requires a sustained balance of antibody and T cell responses

Abstract

Pre-existing immunity to dengue virus (DENV) can either protect against or exacerbate, a phenomenon known as antibody dependent enhancement (ADE), a secondary DENV infection. DENV, as an escalating health problem worldwide, has increased the urgency to understand the precise parameters shaping the anti-DENV antibody (Ab) and T cell responses, thereby tipping the balance towards protection versus pathogenesis. Herein, we present the current state of knowledge of about the interplay between the Ab and T cell responses that dictate the outcome of DENV infection and discuss how this newfound knowledge is reshaping strategies for developing safe and effective DENV vaccines.

Introduction

Dengue virus (DENV) is a member of the Flaviviridae family of positive-sense single-stranded RNA viruses that includes Zika virus (ZIKV), yellow fever virus (YFV), and Japanese encephalitis virus (JEV). The flavivirus RNA genome encodes a large polyprotein that is cleaved into three structural proteins-envelope (E), membrane precursor (prM), and capsid (C), and seven non-structural proteins (NS1, NS2a, NS2b, NS3, NS4a, NS4b, and NS5). Flavivirus polyproteins exhibit a high degree of sequence homology-68% to 78% among the DENV serotypes and 45% to 56% between DENV and other flaviviruses. As a result, flaviviruses readily elicit cross-reactive antibody (Ab)- and T cell responses. However, whether those immune responses protect against or exacerbate subsequent infections is the subject of intense research.

What is clear is that generation of a subneutralizing Ab response against flaviviruses can facilitate viral entry into Fc receptor-positive cells during a subsequent infection, thereby exacerbating disease. This process, known as Ab-dependent enhancement (ADE), is best illustrated by patients who develop severe dengue disease after recovery from an earlier DENV infection [1,2]. Most cases of severe dengue result from either secondary infection in older children and adults or primary infection of infants born to DENV-immune mothers. Consequently, countries with both a prevalence of flaviviruses and DENV seropositive populations are at increased risk of ADE [3–5]. The DENV vaccines developed to date may at least in theory exacerbate the public health situation, given that most were designed to generate an Ab response-heightening the potential for ADE reactions, if individuals develop poorly neutralizing Abs (nAbs) to the vaccine. Recent studies using non-human primates (NHP) and type 1 interferon receptor (Ifnar)-deficient mouse models validate human studies that suggest that cross-reactive nAb and T cell responses are crucial factors in driving protection vs ADE immune responses. In this review, we highlight recent research on these factors, and how understanding these mechanisms is reshaping DENV vaccine development and testing.

Protective and pathogenic role of antibodies during DENV infection

During a primary DENV infection, the Ab response is predominantly serotype-specific. Of the cross-reactive Abs that are generated during primary DENV infection, the ones that are also neutralizing closely resemble germline-encoded Ab sequences-requiring few somatic mutations to neutralize all four DENV serotypes ex-vivo [6]. Cross-reactive DENV Abs that do not meet the threshold affinity for neutralization need to undergo affinity maturation during secondary infection to increase the breadth of Ab reactivity [7] and enable efficient neutralization [8]. Indeed, after sequential homotypic or heterologous infections (eg, DENV1-DENV1 and JEV-DENV1), cross-neutralizing Abs can be detected against more than one DENV serotype or flavivirus [5].

Anti-flavivirus humoral immune response is largely targeted to various epitopes on the viral envelope (E) protein. These E protein-specific Abs can suppress, enhance, or have no effect on a subsequent infection by a heterologous flavivirus. Flaviviral E proteins consist of three dimer domains (EDI, EDII, EDIII). Highly potent nAbs can be elicited by complex quaternary epitopes, termed E dimer epitopes (EDE) that span EDI and EDIII on mature virions [9]. These EDE-specific Abs neutralize mature virus even at low Ab concentration, by cross-linking E monomers within and between E dimers to fully coat the virion and prevent conformational changes in E protein [10,11]. On the other hand, simple immunodominant epitopes on E monomers that are largely exposed on the immature virion, such as the highly conserved fusion loop (FL) of EDII, drive development of Abs that are highly cross-reactive but subneutralizing [9,12,13]. These subneutralizing Abs can instead increase the likelihood of ADE [1,14,15]. Thus, epitope specificity is one key factor that may impact the protective vs pathogenic potential of the DENV-reactive Ab response.

Another major factor that influences neutralization vs enhancement capacity of the flaviviral Ab response is the concentration of virus-reactive Abs, which is a function of the robustness of the initial Ab response and subsequent decay. A longitudinal pediatric cohort study revealed that an intermediate titer of DENV-specific Abs (vs low and high titers) associates with an increased probability of DENV ADE [1]. Thus, high titers of cross-reactive anti-DENV nAb can protect against secondary DENV infections that occur shortly after primary infection (Figure 1A). As these nAb titers decay over time, they reach an intermediate concentration that can drive ADE and eventually drop below a functional range to have no effect on infection outcome (Figure 1B). In line with this schematic, a recent Ab transfer study demonstrated that low concentrations of Abs from both asymptomatic and symptomatic human ZIKV infection but not naïve patients could induce ADE in mice infected with ZIKV [16]. This evidence implies that ADE can occur in the context of sequential infection with the same flavivirus. Consistent with this implication, reinfection with the same DENV serotype has been documented in humans [17,18]. However, an exact timeline for the Ab decay after a flaviviral infection likely varies by individual. A recent study established a diagnostic timepoint 1-year after DENV infection to define a minimum nAb titer that would correlate with continued protection against DENV ADE in humans [19].

Figure 1. Effects of nAb decay on secondary infection outcome.

Primary DENV infection induces a nAb responses (red line) that rises and then decays, eventually reaching a titer range that can induce ADE (orange rectangle). The clinical outcome of a secondary DENV infection (grey histograms) depends on whether the challenge occurs A) prior to the ADE-inducing titer range, leading to a protective response, or B) after, leading to a large spike in virus and potentially ADE.

The role of nAb titers in determining flavivirus ADE vs protection has been most clearly demonstrated in parent-offspring studies. Infants born to DENV-exposed mothers acquire the maternal anti-DENV Abs via the placenta and are protected for first several months (typically 6 months), then become at risk to severe dengue disease for the next several months (between 6–9 months of age) but are immunologically naïve by 12 months of age [20]. Accordingly, some of the strongest data for the role of DENV Abs in dengue pathogenesis comes from the observations that 1) severe dengue disease has rarely been observed in infants born to DENV-naïve mothers, and 2) as the maternally transferred DENV Ab levels decrease, there is an increase in the probability of ADE in the offspring [21,22]. Consistent with these human findings, our recent work demonstrated that maternal Abs induced DENV ADE in mouse pups born to ZIKV-immune mothers and infected with DENV 4 to 5 weeks post-birth (ie, post-weaning) [23]. Reciprocally, two different groups showed that DENV-immune mothers infected with ZIKV during pregnancy increased ZIKV infection and resorption of fetuses [24,25]. More recently, we reported that ZIKV ADE could occur in mouse pups born to JEV-immune mothers [26]. In line with these mouse studies, A formaldehyde and UV-inactivated tetravalent dengue vaccine sensitized monkeys to enhanced viremia during challenge with DENV [27]. These recent studies lend further support to the concept that ADE may be observed when the nAb titer wanes and reaches below a protective threshold concentration--this feature of the flaviviral nAb response highlights a limitation in Ab-centric vaccines against DENV.

Protective role of T cells against DENV infection

Increasing evidence suggests that T cells provide protection against DENV infection, even in situations where Ab titers wane to levels conducive for ADE. DENV- or ZIKV-specific CD4 and CD8 T cells target different immunodominant epitopes: CD4 T cells largely target epitopes in structural proteins C and E and some variable NS epitopes, while CD8 T cells predominately target NS proteins including NS3 and NS5 and some additional E epitopes [28–32]. Recent reviews also describe a number of factors that might contribute to the formation of immunodominance patterns, including T cell subsets analyzed, the infecting virus type, and whether primary or secondary immune responses [33,34]. Early evidence of T cell activation, and especially IFN-γ production, are inversely associated with viremia and disease severity in primary DENV infection [35]. In secondary infections, DENV and ZIKV viral burden can be limited by cross-reactive CD4 and CD8 T cells specific for a broad range of conserved NS epitopes [36,37]. Then, type-specific and cross-reactive T cells can persist through convalescence after DENV infection in humans [38]. These memory T cells can be reactivated by DENV or ZIKV to promote target cell killing [38]. Thus, the presence of functional flavivirus-specific T cells during acute, convalescent, and resolved stages of DENV or ZIKV infections illustrate a potentially important protective role for T cells during natural infection or vaccination.

CD4 T cells:

T cells expand during acute flavivirus viral infection in humans [37,39], and with either primary DENV or ZIKV infections they adopt a T helper (Th) 1 phenotype and produce the associated cytokines IFN-γ, TNF and IL-2 [40,41]. In addition to Th1 responses, DENV and ZIKV lead to the expansion of populations of CXCR5+ T follicular helper (Tfh) cells and Foxp3+ T regulatory (Treg) cells [29,39,40]. Tfh cells are important for germinal center and neutralizing Ab development [29,42]. The contribution of Treg during DENV or ZIKV clearance has not been fully characterized, but they may be involved in limiting T cell responses [34].

During primary viral infections, CD4 T cells play a non-redundant indirect role to promote Ab production and CD8 T cell function in controlling systemic and tissue specific viral load [31]. Recent studies using mouse models of ZIKV infection have shown that, in the absence of CD4 T cells at specific infection sites, such as intravaginal ZIKV infections, viral infection is uncontrolled [29]. Although CD4 T cells may mediate some direct viral control, reduced Ab titers in the absence of CD4 T cells likely contributes to increased ZIKV burden [29]. However, in another study, memory CD4 T cells optimally protected Ifnar1^−/− mice against ZIKV in combination with CD8 T cells [42], highlighting CD4 T cells as potent support for cytotoxic immune responses to aid control of infection.

Cross-reactive CD4 T cell responses that can promote neutralizing Abs and help CD8 T cells then represent a promising vaccine target that may address complications due to ADE. In humans, JEV-vaccination generates limited cross-reactive CD4 T cells, mostly against ZIKV and to a lesser extent DENV and YFV [5]. Ideally, robust cross-reactive CD4 T cells with broad epitope specificity would be elicited to convey lasting protection against multiple DENV serotypes and flaviviruses. To this end, our lab immunized Ifnar1^−/− HLA-DRB1*0101 mice with DENV/ZIKV cross-reactive CD4 T cell epitopes in E, NS2A, NS4B and NS5. Immunization with these cross-reactive CD4 T cell peptides enhanced CD4 T cell responses and reduced viral burden after ZIKV challenge [41]. Thus, vaccine strategies that combine epitopes enhancing CD4 T cell help with epitopes driving cross-reactive B cells or CD8 T cells may provide robust cross-protection against different DENV serotypes and flaviviruses.

CD8 T cells:

Eliciting CD8 T cells that directly clear viral infection independent of Ab is likely important for lasting protection against reoccurring flavivirus infections. CD8 T cells isolated from patients during acute DENV or ZIKV infections express IFN-γ and adopt an activated cytolytic phenotype after ex vivo stimulation [32,43]. DENV- and ZIKV-specific CD8 T cells are frequently polyfunctional and when stimulated by viral peptides can co-produce IFN-γ with CD107a, granzyme B, or TNF [28,30,37,44,45]. In some individuals, DENV-specific CD8 T cells can persist with T effector memory (Tem) or T effector memory expressing CD45RA (Temra) phenotypes that display robust activation ex vivo after stimulation with DENV peptide [46]. In animal models, activated DENV- or ZIKV-specific CD8 T cells are essential for control of primary infection [44,47]. CD8 T cell depletion prior to primary DENV or ZIKV infection dramatically reduces host survival, with evidence suggesting direct CD8 T cell lysis of infected targets is largely responsible for viral clearance [28,30,42,48]. Even in the absence of CD4 T cell help, CD8 T cells can be induced and control the severity of DENV and ZIKV infection in Ifnar1^−/− and LysMCre⁺Ifnar1^fl/fl mouse models [29,42,49].

Adoptive transfer studies with cross-reactive CD8 T cells from DENV-immune mice have further illustrated protection against ZIKV challenge and vice versa [44,50]. Importantly, these primed CD8 T cells are potent enough to control viral burden even in the presence of immune sera that otherwise can induce ADE [51]. Even in pregnancy models of ZIKV infection, DENV-elicited CD8 T cells efficiently controlled ZIKV virus [49], negating cross-reactive Ab responses that could promote ADE [24,25,52]. DENV-immune CD8 T cells not only reduced viremia within the maternal spleen and placenta but also conveyed protection at the maternal-fetal interface and rescued fetal weight and size [49]. Although the exact mechanism of cross-reactive CD8 T cell protection is not fully defined, the efficient induction of these cross-reactive memory CD8 T cells is then an avenue for promoting lasting protection even in situations where ADE might normally result from prior flavivirus immunity (Figure 2A). This has been the focus of recent vaccine-related studies. Our lab tested a peptide-immunization strategy with DENV immunodominant CD8 T cell epitopes. In this setting, DENV2-immune CD8 T cells were efficiently induced in Ifnar^−/− HLA-B*0702 and HLA-A*0101 transgenic mice and engendered cross-reactive protection against a subsequent ZIKV infection [53]. This lends further support to the concept that cross-reactive CD8 T cell recall responses generated by emerging vaccine strategies may contribute to protection against DENV and ZIKV even when the pre-existing Abs could result in ADE (Figure 2B).

Figure 2. T cells provide lasting protection even in the presence of ADE promoting nAb.

Exploiting mouse models permissive for DENV and ZIKV infection. A) After a primary DENV infection that induces only Ab responses, a secondary challenge is likely to promote ADE. B) Whereas simultaneous Ab and T cell induction prevents ADE.

Balancing Ab and T cell responses to change DENV infection outcomes.

The effort to develop a DENV vaccine is 70 years old. Accumulating evidence indicates that traditional, nAb-inducing vaccines are likely to generate ADE complications. In fact, Dengvaxia, the only DENV vaccine approved for human use is not approved for use in children less than 9-years-old or naive individuals, since it appears to prime these patient populations for more severe DENV infection. To develop safe and effective DENV vaccines that induce sustained immunity requires understanding the factors and mechanisms driving protective vs pathogenic immune responses. In this regard, extensive longitudinal Ab surveys of endemic regions, characterization of Ab interactions, and evaluation of T cell characteristics, have begun to define potential mechanisms of immune protection against DENV.

A better approach than Ab-centric DENV vaccines is to incorporate Ab vaccination strategies with those that also capitalize on the protective effects of memory T cell responses. The NIH DENV vaccine TV005, which is a combination of attenuated DENV1, 3, and 4 with a DENV2/4 chimeric viruses, induces CD4 and CD8 T cells responses to different DENV serotypes [54]. Another live-attenuated DENV vaccine, TAK-003, is a recombinant tetravalent platform incorporating DENV2 backbone with prM and E genes from DENV1, 3, and 4, and this vaccine elicits CD8 T cells with cross-reactive cytokine and proliferative function against all four DENV serotypes [55]. Results of the ongoing phase 3 trial data on TV005 and TAK-003 vaccines should provide critical insights into the potential role of vaccine-induced T cells in conferring protection against DENV. Based on recent human and animal model data, DENV vaccines that induce robust nAb responses and robust memory CD4 and CD8 T cell responses will be key. Achieving a proper balance of these Ab and T cell responses may be challenging with live-attenuated vaccines due to potential dominant effects of one serotype over the other three serotypes. However, new advances in the development of subunit vaccine platforms offer promising avenues for designing next generation DENV vaccines that induce robust and balanced Ab and T cell responses.