Humoral Immune Responses Against Zika Virus Infection and the Importance of Preexisting Flavivirus Immunity

Abstract

The recent emergence of Zika virus (ZIKV) in the western hemisphere has been linked to Guillain-Barre syndrome, congenital microcephaly, and devastating ophthalmologic and neurologic developmental abnormalities. The vast geographic spread and adverse disease outcomes of the 2015–2016 epidemic have elevated ZIKV from a previously understudied virus to one of substantial public health interest worldwide. Recent efforts to dissect immunological responses to ZIKV have provided significant insights into the functional quality and antigenic targets of ZIKV-induced B-cell responses. Several groups have demonstrated immunological cross-reactivity between ZIKV and other flaviviruses and have identified antibodies capable of both cross-neutralization, as well as antibody-dependent enhancement (ADE) of ZIKV infection. However, the impact of preexisting flavivirus immunity on ZIKV pathogenesis, the generation of protective responses, and in utero transmission of ZIKV infection remain unclear. Given the widespread endemicity of DENV in the areas most effected by the current ZIKV outbreak, the possibility of ADE is especially concerning and may pose unique challenges to the development and deployment of safe and immunogenic ZIKV vaccines. Here, we review current literature pertaining to ZIKV-induced B-cell responses and humoral cross-reactivity and discuss relevant considerations for the development of vaccines and therapeutics against ZIKV.

Immense scientific efforts have been directed toward Zika virus (ZIKV) over the last 2 years, leading to deeper insights into ZIKV parthenogenesis and innate and adaptive immune responses to ZIKV infection, the identification of host factors involved in transplacental ZIKV transmission, the establishment of novel mouse and primate disease models, and the development and early evaluation of several different ZIKV vaccine platforms [1–6]. Major advances in our understanding of humoral immunity against ZIKV have been achieved, with an ever-growing body of work focused on characterizing B-cell responses, both at the serological level and the cellular level [7–11]. Extensive analyses of human and murine monoclonal antibodies (mAbs) have defined key neutralizing epitopes on the ZIKV envelope (E) protein [7, 11–13]. This has led to the identification of numerous mAbs with in vivo therapeutic potential [7, 11, 14, 15]. A number of studies have dissected immunological cross-reactivity between ZIKV and other flaviviruses, providing valuable information about the potential impact of this cross-recognition on ZIKV immunity and infection outcome [7, 11, 12, 14, 16]. In addition, a number of ZIKV vaccine candidates have been developed; these include candidates that show great preclinical promise and others that are currently being evaluated in human clinical trials [5, 17]. In view of the significant recent developments made in the characterization of human B-cell responses to ZIKV infection, here we will review our current understanding of humoral immunity to ZIKV and discuss literature of interest and relevance to the design and delivery of effective vaccines and therapies.

HUMORAL IMMUNITY TO ZIKV INFECTION

Serology and Early Cellular Responses to ZIKV Infection

After ZIKV infection, virus-specific antibodies can appear in serum as early as 2–3 days after fever onset [8, 10, 18]. The rise in ZIKV-specific serum antibody titers early after infection temporally overlaps with and is caused by the induction of plasmablasts [8, 10], an antibody-secreting B-cell subset that appears in peripheral blood transiently following antigen exposure. As demonstrated by Rogers et al, plasmablasts can constitute as much as 63% of the peripheral B-cell compartment a week after symptom onset in patients with ZIKV infection. Lai et al showed that these antibody-secreting B cells secrete ZIKV-specific immunoglobulin M, immunoglobulin G, and immunoglobulin A and become virtually undetectable in circulation by 20 days after fever onset [10]. Interestingly, while Rogers et al observed that ZIKV-infected patients from dengue-endemic areas experienced a more robust plasmablast expansion as compared to a dengue-naive case, no such correlation between flavivirus exposure history and magnitude of plasmablast responses was noted by Lai et al.

Prior flavivirus experience was, however, strongly associated with greater serological cross-reactivity after ZIKV infection in both studies [8, 10]. While individuals with no evidence of previous dengue virus (DENV) or yellow fever virus exposure generally display ZIKV-specific serum titers, sera from subjects with known or suspected prior flaviviral infections can more abundantly cross-react in binding and/or neutralization to other flaviviruses besides ZIKV [7, 8, 10, 18]. This cross-reactivity has complicated the serology-based diagnosis of ZIKV infections in populations that are DENV or yellow fever virus experienced [19]. As a result, the issue of cross-reactivity between ZIKV and other flaviviruses has recently reemerged as an area of interest and focused research. Several groups have tested acute- and convalescent-phase sera from DENV-infected donors against ZIKV to identify factors that modulate immunological cross-recognition between these closely related viruses and to better understand how it may influence immune responses to future ZIKV infections [12, 14, 20, 21]. These studies have used sera obtained at different time points after dengue exposure (acute vs convalescent phase) from patients with varied dengue serostatus (primary vs secondary dengue exposure). Two groups have reported that acute-phase sera (collected within 10 days after fever onset) and early convalescent-phase sera (collected 1–3 months after infection) from patients with secondary dengue exposure potently bind and neutralize ZIKV [12, 20]. Conversely, limited cross-reactivity to ZIKV was observed by Swanstrom et al and Collins et al, who tested late convalescent-phase sera (collected >6 months to several years after infection) from patients with secondary dengue exposure in their studies [14, 21]. Additionally, sera collected from patients with primary dengue exposure, regardless of the time of sampling, appeared poorly cross-reactive and neutralizing against ZIKV [14]. These data complement the findings of Lai et al and others [7, 8, 10, 18] and, taken together, suggest that prior flavivirus exposures may boost cross-reactive titers to ZIKV because of the reactivation of preexisting memory B cells that target conserved epitopes. They also suggest that the magnitude of these cross-reactive immune responses may additionally depend on the length of time separating the 2 infections and the frequency of previous flavivirus exposures (epidemic vs endemic exposures).

Single-Cell Analyses Provide Insight Into ZIKV-Induced Plasmablast Responses

A few recent studies have provided valuable information regarding the cellular origin and functional properties of ZIKV-induced antibody responses by characterizing mAbs derived from plasmablasts collected from ZIKV-infected patients [8, 9, 22]. As an example, Rogers et al compared plasmablasts and plasmablast-derived mAbs from 3 dengue-experienced donors to those from a dengue-naive donor [8]. Plasmablasts from the dengue-experienced donors were found to have high levels of somatic hypermutation, comparable to the somatic hypermutation frequencies of plasmablasts induced by influenza vaccination [23] or DENV infection [24]. In addition, a large number of plasmablasts were clonally related, presumably derived from the differentiation and proliferation of a common ancestor. Together, these data suggest that a significant proportion of the acute-phase B-cell response in these donors was memory derived. In contrast, plasmablasts from the dengue-naive donor exhibited little to no somatic hypermutation, akin to a classical primary response [25, 26], and were characterized by limited clonal expansion. Additionally, a larger majority of mAbs isolated from dengue-experienced donors were ZIKV reactive, binding whole virions and/or recombinant E protein and nonstructural protein 1, whereas <20% of mAbs from the dengue-naive donor bound ZIKV antigens. Congruent with the serological analyses of cross-reactivity presented earlier, the mAbs isolated from the plasmablasts of the dengue-experienced donors were also more cross-reactive to DENV1–4, compared with those from the dengue-naive donor, although most were poorly neutralizing against ZIKV [8].

To understand the changes in the repertoire and functional quality of ZIKV-induced B cells over time, Rogers et al and Yu et al generated mAbs from ZIKV-specific memory B cells isolated at later time points after infection. While early responses were highly cross-reactive and poorly neutralizing, both groups found that the memory B-cell repertoire after convalescence contained a larger number of clones that were ZIKV specific. Postconvalescence ZIKV-specific mAbs neutralized ZIKV more potently than acute-phase mAbs [9] and bound to ZIKV antigens with a higher affinity than cross-reactive memory B-cell–derived and plasmablast-derived mAbs [8]. These findings suggest that ZIKV-induced antibody responses mature with time and that B-cell clones that produce higher affinity and more potently ZIKV-neutralizing mAbs are selected and maintained in the memory B-cell pool after infection. Whether these superior clones are maintained over the long term is unclear, since the memory B cells used for the analyses above were isolated 3–5 months after infection without additional follow-up samples.

Targets for In Vitro Neutralizing and In Vivo Protective B-Cell Responses to ZIKV

In addition to revealing the cellular and functional properties of ZIKV-induced B-cell responses, the analysis of human and murine mAbs has provided key insight into the targets of neutralizing antibodies to ZIKV (Figure 1A) [7, 9, 11, 22, 27]. Stettler et al generated a sizeable panel of memory B-cell–derived mAbs from 4 patients in the convalescent phase of ZIKV infection and found that the most potently neutralizing antibodies identified were either EDIII specific or targeted quaternary epitopes on ZIKV virions. One of the potently neutralizing EDIII-specific antibodies identified, ZKA64, when expressed as a LALA mutant (a mutant that does not bind to FcγR or complement), provided complete prophylactic protection against a lethal challenge in A129 mice. Similarly, another human memory B-cell–derived EDIII-specific mAb, Z23, generated by Wang et al, fully protected type I interferon receptor–deficient mice from weight loss and mortality upon ZIKV infection [15]. The capacity of EDIII-specific mAbs to protect from lethal challenge was also demonstrated by Zhao et al, using murine mAbs ZV-54 and ZV-67 in wild-type mice receiving antibody to type I interferon receptor prior to infection [13].

Figure 1.

Zika virus (ZIKV)–induced antibodies (Abs) and their potential roles during infection. A, Examples of neutralizing monoclonal Abs and the epitopes they bind (illustrations adapted from [44]). B, Illustrations of Ab-mediated protection through viral neutralization or Ab-dependent enhancement of viral infection. Abbreviations: DENV2, dengue virus serotype 2; EDE, envelope dimer epitope.

While EDIII is a major immunodominant target for DENV-neutralizing antibodies in mice, several studies have suggested that this is not the case in humans [28, 29]. The importance of EDIII-specific antibodies in the ZIKV immune response in humans was examined by Robbiani et al [11], who screened a large number of sera from cohorts of patients with convalescent-phase ZIKV infection in Brazil and Mexico for binding titers against ZIKV EDIII. They found that a subset of these patients displayed very high binding titers against ZIKV EDIII and that EDIII-specific titers in these donors correlated positively with high neutralization titers against ZIKV. Expression cloning of mAbs from EDIII-specific memory B cells from 6 donors in the subset showed that they shared a common rearrangement (VH3-23/Vk1-5) and potently neutralized ZIKV in vitro and in vivo [11]. Additionally, Yu et al observed that the depletion of EDIII-binding antibodies from ZIKV-infected patient serum caused a conspicuous reduction in ZIKV neutralization, with the percentage reduction in neutralization increasing between 7 days after fever onset (13%) and 188 days after fever onset (63%). It is worth noting, however, that these serum depletion experiments were performed on sera collected from 1 donor only. Nevertheless, the studies described above suggest that EDIII-specific antibodies contribute to anti-ZIKV humoral immunity.

Another class of antibodies that has gained significant interest recently comprises antibodies that target quaternary epitopes on the viral surface [7, 12, 16, 27]. These mAbs typically do not bind to recombinant E protein, as they target epitopes that span a single E protein dimer or the E protein dimer-dimer interface. Examples of such conformation-sensitive, dimer-dependent antibodies include the broadly neutralizing envelope dimer epitope (EDE)–specific antibodies [12, 30]. Originally isolated from patients with dengue, several mAbs binding the EDE1 epitope have been shown to potently neutralize ZIKV in vitro [12, 14, 31]. In addition to its high in vitro ZIKV neutralization potency, EDE1 mAb C10 (Figure 1A) has also been shown to protect AG129 mice from a lethal challenge, requiring just 2 doses of 10 μg each for complete protection against ZIKV [14]. ZIKV-117 is a dimer-dimer interface–binding mAb and was isolated by Sapparapu et al from a ZIKV-immune donor [27, 32]. ZIKV-117 was shown to protect therapeutically in both a pregnant and nonpregnant mouse model of ZIKV infection, preventing fetal infection and demise [27].

In contrast, other EDI/EDII-specific mAbs isolated from patients with ZIKV infection were shown by Stettler et al to be largely moderate-to-poor neutralizers and more promiscuously bound to DENV1–4 than the EDIII-specific mAbs [7], akin to what has been described in the past in the dengue literature [28, 29]. The fusion loop represents a portion of the EDII and is highly conserved between various flaviviruses [16]. Despite their highly cross-reactive nature, antibodies against the fusion loop are typically poorly neutralizing [12, 27, 31], possibly because their epitopes are not easily accessible on the viral surface [33]. An exception is the fusion loop–specific antibody originally characterized by Deng et al [34], named 2A10G6 (Figure 1A), which potently neutralized multiple flaviviruses in vitro, including ZIKV, DENV1–4, yellow fever virus, and Japanese encephalitis virus, and protected mice from a lethal ZIKV challenge [35].

These descriptions of mAbs with ZIKV-specific or broadly cross-reactive phenotypes have helped identify critical targets for neutralizing antibody responses in ZIKV infections, which may prove invaluable for ongoing and future rational epitope-based vaccine strategies. In addition, mAbs targeting virus-specific epitopes have aided in the development of diagnostic tests that differentiate ZIKV from other flavivirus infections [36]. Given the health risks associated with the vertical transmission of ZIKV, including congenital malformations and pregnancy-related complications, the development of safe and effective immunization strategies for women of reproductive age, especially pregnant women, is a high public health priority. The success of gamma globulin treatments in safely controlling varicella, rubella, and human cytomegalovirus infections in pregnant women and neonates has set a strong precedent for the use of antibody therapy in these immunosuppressed populations [37–41]. Consequently, the administration of human or humanized mAbs with potent prophylactic or therapeutic potential may also be a viable approach to tackle ZIKV infections. Several antibodies described above, including ZIKV-117, Z23, C10, and ZKA64, appear ideal candidates for such antibody-based interventions. However, given that the protective efficacy of these mAbs has primarily been evaluated in mice with immune deficiencies, a more physiologically relevant assessment of their protective potential and safety warrants further investigation.

Antibody-Dependent Enhancement (ADE)

The development of DENV vaccines has been hampered in part by the epidemiological link between secondary DENV exposure and exacerbated disease outcome [42, 43]. One of several hypotheses put forward to explain this linkage is the ADE of viral infection. ADE is said to occur when subneutralizing, cross-reactive antibodies bind to infectious virus and mediate FcγR-dependent endocytosis of the virus-antibody complex, leading to subsequent infection of FcγR-expressing cells (Figure 1B) [44]. In this way, preexisting immunity to 1 serotype of DENV is thought to increase the risk for developing severe disease after subsequent exposure to a heterologous DENV serotype [44]. Since the first in vitro descriptions of ADE of DENV infection in the 1970s [45, 46], numerous laboratories worldwide have studied the phenomenon in vitro, in cells of the myeloid lineage [28, 47–49]. Groups have demonstrated ADE of DENV infection in mice [50–52] and macaques [53] to understand the mechanism and factors governing ADE in vivo. Importantly, compelling evidence in support of ADE has been provided by studies of infants who acquired DENV antibody from their DENV-immune mothers [54–56]. These studies showed that, although younger infants are protected from dengue by maternal antibody, infants aged 6–12 months experience the greatest risk for severe disease, likely due to a decay in maternal antibody to subneutralizing, disease-enhancing levels [54–56]. More recently, long-term follow-up studies of the phase 3 trials of Dengvaxia, Sanofi Pasteur’s tetravalent live attenuated vaccine candidate, indicated poorer vaccine efficacy and an elevated risk of hospitalization and severe dengue among participants aged 2–5 years at vaccination, compared with older vaccine recipients [57, 58]. This age-related imbalance in vaccine efficacy has been linked to the dengue serostatus of vaccinees at baseline, among other factors. These studies suggest that, while vaccination boosted preexisting DENV1–4 titers in seropositive individuals, the incomplete tetravalent protection provided by the vaccine may have left seronaive vaccinees more susceptible to exacerbated disease upon exposure to DENV [57, 58].

Given the structural and antigenic similarities between DENV and ZIKV, it is perhaps no surprise that ADE can be readily demonstrated for ZIKV in vitro [7, 16, 27, 59, 60]. ZIKV only exists as 1 serotype [61]; therefore, these descriptions of ADE have been made using antibodies generated during heterologous flavivirus infections, with a large majority of studies focusing on DENV-induced antibodies and their impact on ZIKV infection. DENV immunity–linked ADE of ZIKV infection has also been reported in mice by Bardina et al, who showed that Stat2^−/− mice receiving pooled DENV immune human sera prior to ZIKV infection exhibited exacerbated disease, increased viremia levels, and increased viral titers in immune-privileged sites such as the testes and eyes [62]. Interestingly, the severity of disease after ZIKV infection was highly dependent on antibody concentration, as the transfer of a 10-fold higher amount of serum caused no viral enhancement but rather complete protection. However, a notable limitation of this work is the use of immunodeficient mice to demonstrate ADE. The physiological relevance of Bardina et al’s findings is also brought into question by the absence of clinical/epidemiological data linking preexisting flavivirus immunity with enhanced ZIKV infection in humans and by the lack of in vivo observations of ADE in nonhuman primate models. Pantoja et al demonstrated that preexisting DENV antibodies did not increase disease severity in macaques infected with ZIKV. Interestingly, while no difference in peak viremia levels was observed, macaques that were infected with DENV 2.8 years prior to ZIKV infection had a shorter viremic period than DENV-naive macaques. Differences were also observed in B-cell and T-cell activation and in cytokine and chemokine profiles, suggesting that DENV immunity may modulate immune responses to ZIKV [59]. Shortly thereafter, McCraken et al also reported no significant differences in viremia level after ZIKV infection between naive macaques and macaques previously infected with DENV and yellow fever virus [63]. In both studies, macaques with previous flavivirus exposures had cross-reactive serum antibody titers that enhanced ZIKV infection in vitro. These findings clearly illustrate that in vitro descriptions of ADE are not predictive of ADE in vivo. As such, additional macaque studies and extensive analyses of the disease outcomes of flavivirus-preexposed versus flavivirus-naive patients with ZIKV infection are needed to fully clarify the relevance of ADE to human ZIKV infections.

In conclusion, our understanding of humoral immune responses to ZIKV has greatly benefitted from the massive efforts recently directed toward this previously understudied virus. Preliminary insights have been made into the kinetics and functional quality of early B-cell responses to ZIKV infection, and extensive analyses of mAbs and sera have revealed key binding and neutralization targets for ZIKV-induced antibodies. The cross-reactivity between ZIKV and other flaviviruses has been a topic of focused research, leading to a better grasp of the immunological interplay between flaviviruses. Despite this rapid and tremendous progress, important questions remain regarding the magnitude and durability of ZIKV-specific versus cross-reactive antibodies generated after ZIKV infection, as well as the effect of flavivirus serostatus on the induction of protective immune responses to ZIKV. In addition, the role of ADE in disease severity and viral transmission through the placental barrier remains unclear. Clarifying these issues requires supplementing existing and future in vitro and in vivo data with human cohort studies in areas of endemicity and nonendemicity. This information will be essential for the development, testing, and eventual delivery of efficient and safe ZIKV vaccines.

Notes

Supplement sponsorship. This work is part of a supplement sponsored by the National Institute of Allergy and Infectious Diseases (NIAID), part of the National Institutes of Health (NIH).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.