Abstract

Author Summary

Introduction

Flaviviruses are a group of ~70 RNA viruses that cause morbidity and mortality on a global scale, with greater than 100 million human infections annually [1]. Viruses within this genus of medical concern include yellow fever virus, tick-borne encephalitis virus, Japanese encephalitis virus, dengue virus (DENV) and West Nile virus (WNV). WNV is a mosquito-borne flavivirus maintained in nature in an enzootic cycle with birds. WNV infections of humans result in a spectrum of clinical symptoms depending, in part, on the age and immune status of the individual. While most infections are sub-clinical, symptomatic cases range from self-limiting fever to acute flaccid paralysis and encephalitis [2]. Since its introduction into North America in 1999, as many as three million people have been infected by WNV [3], with ~1000 severe infections occurring in the United States annually (www.cdc.gov). To date, there are no WNV-specific treatments or vaccines licensed for use in humans.

Flaviviruses are small spherical virions that encapsidate an ~11 kb genomic RNA of positive-sense polarity [1]. This RNA is translated as a single polyprotein that is processed by viral and host cell proteases into ten functionally distinct proteins. Flaviviruses encode three structural proteins that comprise the virus particle and seven non-structural proteins that function to process the viral polyprotein, replicate the viral genome, and antagonize the host's protective response to infection [1], [4], [5], [6], [7], [8], [9], [10], [11]. Flaviviruses bud into the endoplasmic reticulum as immature viruses that incorporate 60 heterotrimeric spikes of the envelope (E) and precursor to membrane (prM) proteins [12], [13]. Maturation of virions during egress from the cell is associated with a pH-dependent change in the arrangement and oligomeric state of the E protein and cleavage of prM by a host cell furin-like serine protease [14], [15], [16]. While prM cleavage is a required step in the formation of an infectious virion [17], several lines of evidence suggest that a significant fraction of infectious virions retain some uncleaved prM [18], [19], [20], [21], [22]. The E protein is a class II fusion protein composed of three structurally distinct domains (domains I-III; E-DI-III) connected to the viral membrane by a helical stem region [23]. On mature virions, 180 copies of the E protein are arranged as anti-parallel dimers in an unusual herringbone T=3 pseudo-icosahedral array [24], [25], [26]. In this configuration, E proteins are not quasi-equivalent, but exist in three distinct chemical environments defined by their proximity to the two-, three-, or five-fold symmetry axis. This arrangement adds complexity to the antigenic surface of virus particles as antibodies may not be able to bind all 180 E proteins due to steric constraints on antibody binding or occlusion of the epitope itself [27], [28].

Humoral immunity is a critical aspect of protection against flavivirus infection (reviewed in [29]). Eliciting a protective antibody response is a primary goal in the development of safe and effective flavivirus vaccines [30]. The majority of neutralizing antibodies against flaviviruses are directed against the E protein; neutralizing monoclonal antibodies (MAbs) have been identified that bind to all three E protein domains (reviewed in [29], [31]). While antibodies that bind the prM protein have been detected in humans, they possess limited neutralizing activity [32], [33], [34], [35], [36].

Antibody-mediated neutralization of WNV is a “multiple-hit” phenomenon that occurs when virions are engaged by antibody with a stoichiometry that exceeds a required threshold [29], [37]. Our estimate of this threshold is ~30 antibody molecules per virion [38]. Two factors principally govern the neutralizing activity of an antibody. Antibody affinity (or avidity) controls the fraction of accessible epitopes on the intact virion occupied at a particular concentration of antibody. From a vaccination standpoint, eliciting antibodies that bind viral antigens with high affinity is desirable because they can reach the stoichiometric threshold required for neutralization at lower concentrations. However, high-affinity interactions do not always translate into significant functional potency. Because of steric constraints arising from the dense icosahedral arrangement of E proteins on the mature virion, not all epitopes are displayed on intact virions equivalently. Epitope accessibility is a second independent parameter that defines the circumstances that allow for neutralization, and differs markedly between antibodies that recognize structurally distinct epitopes [21], [28], [38], [39], [40]. In fact, many epitopes, including those recognized by antibodies commonly elicited in vivo, are not displayed on the surface of the mature virion enough times to allow for neutralization even when fully occupied [21], [38], [41].

An important concept of the “multiple-hit” model of neutralization is that infectious virions may be decorated with non-neutralizing quantities of antibody. Evidence in support of this includes increases in the neutralizing activity of antibodies observed in the presence of anti-IgG secondary antibodies or complement [37], [42], [43], and the phenomenon of antibody-dependent enhancement (ADE) of infection [44]. Because the rate of antibody-virion interactions is orders of magnitude faster than the rate of virus binding to cells, viruses are likely engaged by antibody for significant periods prior to productive interactions with target cells. Here, we investigated the fate of virions decorated by antibody with a stoichiometry that does not exceed the threshold required for neutralization. Our data revealed time- and temperature-dependent increases in neutralization of WNV that would not be predicted for static virions bound by antibodies under steady-state conditions. Because both viral proteins and intact virions are dynamic [45], [46], [47], [48], the increased neutralization can be explained by changes in epitope accessibility occurring as virions sample different states of a dynamic ensemble of conformations. That kinetic aspects of neutralization were observed for every WNV- and DENV-reactive MAb examined, as well as WNV-polyclonal immune sera, suggests a widespread impact of the dynamic motion of virions on epitope accessibility and antibody-mediated neutralization.

Results

Antibody modulates the infectious half-life of WNV in solution

To investigate whether prolonged exposure of virions to “non-neutralizing” quantities of antibody has functional consequences, we performed a series of kinetic experiments with the WNV E-DIII-lateral ridge specific MAb E16 at a concentration sufficient to neutralize the infectivity of half the virus particles (the EC₅₀). At this concentration, half the virions in a population are bound by antibody with a stoichiometry that exceeds the neutralization threshold, whereas the other half are bound by fewer antibody molecules. WNV reporter virus particles (RVPs) encoding a GFP reporter gene were incubated with 10 ng/ml E16 for one hour at room temperature to allow for steady-state binding (Figure S1). After the room temperature incubation, RVP-antibody complexes were shifted to 37°C for the indicated periods prior to the addition of target Raji-DC-SIGNR cells. The infectivity of RVPs in the presence and absence of antibody was determined by flow cytometric analysis, and is expressed relative to the level of infectivity observed when cells were added to viral immune complexes immediately after the initial room temperature incubation ( Figure 1a ). Using this approach, we observed an antibody-independent decrease in RVP infectivity with prolonged incubation at 37°C (intrinsic decay), consistent with our previously published findings [49]. Incubation in the presence of E16 resulted in a 4.7-fold reduction in the apparent half-life of WNV RVPs relative to the intrinsic rate of decay observed in the absence of antibody (n=23; p<0.0001) ( Figure 1b ). A similar 5.6-fold change in the rate with which virions lost infectivity in the presence of E16 was observed using fully-infectious WNV ( Figure 1c ). Incubation of RVPs or virus with high concentrations of the DENV-specific MAb 3H5 did not impact WNV infectivity, indicating a requirement for virus-reactive antibody ( Figure 1b and c ). These results were not explained by increased binding of virions to tissue-culture plastic during prolonged incubation with antibody (p=.32, n=5; Figure S2a), the action of complement (p=.17, n=5; Figure S2b), or antibody-mediated aggregation (E16 Fab experiments discussed below; Figure 2b ). Additionally, the rapid reduction in RVP infectivity in the presence of virus-specific antibody was still observed in the presence of a broad spectrum of protease inhibitors (p=.74, n=2; Figure S2c), suggesting that the loss of infectivity cannot be explained by physical destruction of RVPs due to contaminating proteases.

Open in a separate window

Figure 1

The functional half-life of WNV decreases in the presence of virus-specific antibody.

(A) WNV RVPs were incubated in the absence or presence of sub-neutralizing concentrations (the EC₅₀) of the MAb E16 for one hour at room temperature to allow binding to reach equilibrium. RVP-antibody complexes were then incubated at 37°C for the indicated times. The infectivity at each point was determined following infection of Raji-DC-SIGNR cells and monitored by flow cytometry 48 h post-infection. The data is presented normalized to levels obtained prior to incubation at 37°C (but after steady-state binding was reached) and fitted to a single-phase exponential decay to obtain the half-life. For the representative experiment shown, dotted lines represent 95% confidence intervals and error bars represent the standard error of duplicate measurements. The fold-reduction in the half-life of WNV RVPs (B) or infectious WNV (C) in the presence of sub-neutralizing quantities of MAb E16 (n=23 and n=2, respectively) or high concentrations of the DENV-specific MAb 3H5 (n=2), as compared to the absence of antibody, are shown. Error bars represent the standard error.

Open in a separate window

Figure 2

Kinetic changes in antibody-mediated neutralization of WNV infection by the DIII-specific MAb E16.

Nine serial four-fold dilutions of MAb E16 (A) or E16 Fab fragments (B) were incubated with WNV RVPs for one hour at room temperature to allow binding to reach equilibrium. RVP-antibody complexes were then incubated at 37°C for incremental lengths of time before infecting Raji-DC-SIGNR cells. Infectivity was monitored by flow cytometry at 48 h post-infection. The reference curve represents RVP-antibody complexes added to target cells immediately after the room temperature incubation. For A, dose-response curves from a representative experiment are expressed relative to the infectivity of RVPs in the absence of antibody that were added to cells immediately after the room temperature incubation (left panel), or relative to the infectivity of RVPs in the absence of antibody at each individual time point (right panel). For B, the E16 Fab fragment results from a representative experiment are displayed relative to the infectivity of RVPs in the absence of antibody at each individual time point. Error bars display the standard error of duplicate infections. Results are representative of 19 and two independent experiments for A and B, respectively.

Kinetic aspects of the antibody dependent enhancement (ADE) profile of MAb E16

ADE describes the dramatic increase in infection of Fcγ-receptor-expressing cells in the presence of non- or weakly-neutralizing quantities of antibody and has been linked to severe clinical outcomes following secondary DENV infections of humans (reviewed in [50]). Antibody-mediated neutralization and ADE are two phenomena related by the number of antibodies bound to the virion [38], [51], therefore kinetic changes in the neutralization activity of an antibody should have a corresponding impact on the ADE profile as well. To explore this, kinetic dose-response curves were set up as described above, except using FcγIIb-expressing K652 target cells. Infection of K562 cells occurs through Fcγ-receptor-mediated uptake of antibody-virus immune complexes; in the absence of antibody, these cells are refractory to infection due to an inability to efficiently bind WNV. When immune complexes were added to cells immediately following the room temperature incubation required to achieve steady-state binding ( Figure 3 ; reference curve), the expected bell shaped dose-response profile was observed. Incubation of antibody-virion complexes at 37°C prior to the addition of cells resulted in a marked change in the shape of the ADE profile. A reduction in the magnitude of ADE was observed over time, consistent with a decrease in the number of infectious virions in solution due to intrinsic decay ( Figure 3 ; left panel). Notably, prolonged incubation resulted in a reduction in the concentration of E16 at which maximal ADE was observed ( Figure 3 ; right panel), corresponding to the reduction in the EC₅₀ observed above ( Figure 2a ; right panel).

Open in a separate window

Figure 3

Kinetic changes in antibody dependent enhancement (ADE) of WNV infection by the MAb E16.

Nine serial four-fold dilutions of MAb E16 were incubated with WNV RVPs for one hour at room temperature to allow binding to reach equilibrium. RVP-antibody complexes were then incubated at 37°C for incremental lengths of time before infecting K562 cells. Infectivity was monitored by flow cytometry at 48 h post-infection. The reference curve represents RVP-antibody complexes added to target cells immediately after the room temperature incubation. Dose-response curves from a representative experiment are expressed relative to the maximum infectivity of RVPs that were added to cells immediately after the room temperature incubation (left panel), or relative to the maximum infectivity of RVPs at each individual time point (right panel). Error bars display the standard error of duplicate infections. Results are representative of four independent experiments.

Because the kinetic neutralization and enhancement experiments presented above were designed to allow for steady-state binding of the virion by antibody prior to incubation at 37°C (validated in Figure S1), changes in virus infectivity over time should not reflect continued engagement of individual virions by increasing numbers of antibody molecules. However, an important assumption of this model is that the number of epitopes on the virion does not change.

Dynamic motion changes epitope accessibility in a time- and temperature-dependent manner

To explore the role of dynamic motion of the virion in WNV neutralization, we took advantage of the fact that the neutralizing activity of several classes of antibodies is significantly limited by epitope accessibility [21], [39], [40]. MAb E53 is a high affinity DII-fusion loop-reactive antibody that has a limited capacity to neutralize mature virions because its epitope is buried on the surface of the virus particle. Even in the presence of saturating concentrations of antibody, E53 does not bind mature virions enough times to exceed the stoichiometric requirements for neutralization [21]. In support, structural studies revealed that E53 Fab fragments were only capable of binding E proteins of the heterotrimeric spikes present on immature virions [52]. For E53 to engage a mature virion with a stoichiometry that exceeds the neutralization threshold, changes in the accessibility of its otherwise cryptic epitopes would be required.

In agreement with our previous findings [21], MAb E53 failed to significantly neutralize a homogeneous population of mature WNV RVPs when antibody-virion complexes were added to cells following a one hour incubation at room temperature ( Figure 4a and b , reference curves). Incubation at 37°C for increasing time intervals prior to infection revealed a gradual increase in antibody potency, with more than half the virions incubated at 37°C for 26 hours becoming susceptible to neutralization ( Figure 4a ). Because the E53 epitope is not accessible on the highly ordered surface of the mature virion [52], the increase in potency requires changes in exposure of the fusion loop epitope over time as the E proteins on the surface of the virion sample different conformations at equilibrium. Binding of E53 has the potential to stabilize the epitope in an exposed conformation that occurs only transiently in the absence of antibody. That significant neutralization requires considerable time at 37°C likely reflects the relatively large number of epitopes that must become accessible on the average virion in order to exceed the neutralization threshold (Figure S3). If increased neutralization by E53 reflects dynamic movement of the virion, one would predict it would be both time- and temperature-dependent. To test this, we performed additional experiments with longer incubation times and a range of temperatures. Increasing the temperature to 40°C enhanced neutralization as compared to 37°C ( Figure 4b , compare green to orange curves), whereas decreasing the temperature to 33°C reduced neutralization capacity ( Figure 4b , compare green to blue curves). Similar results were observed for the E-DI specific MAb E121 shown previously to bind an epitope that is poorly accessible on the mature virion ( Figure 4c ) [21], [39].

Open in a separate window

Figure 4

Time- and temperature-dependent increases in neutralization by antibodies that bind cryptic determinants on the mature virion.

Nine serial four-fold dilutions of the DII-fusion loop (fl)-reactive MAb E53 (A and B) or the DI-reactive MAb E121 (C) were incubated with mature WNV RVPs for one hour at room temperature to allow binding to reach equilibrium. RVP-antibody complexes were then incubated at 37°C only (A) or at 33°C, 37°C, and 40°C (B and C) for incremental lengths of time before infecting Raji-DC-SIGNR cells. Infectivity was carried out at 37°C and monitored by flow cytometry at 48 h post-infection. The reference curve represents RVP-antibody complexes added to Raji-DC-SIGNR cells immediately after the room temperature incubation. Dose-response curves from representative experiments are expressed relative to the infectivity of RVPs in the absence of antibody at each individual time point. Error bars display the standard error of duplicate infections. Results are representative of five and three independent experiments for MAb E53 and E121, respectively.

Viral dynamics broadly impact antibody-mediated neutralization of flaviviruses

To investigate whether kinetic effects on antibody-mediated neutralization are a common phenomenon, we performed additional studies using a panel of 12 WNV MAbs. In each case, we observed time- and temperature-dependent increases in neutralization, similar to our results with E16, E53, and E121 ( Figure 5 ). Collectively, the 15 WNV MAbs used in the current study bind epitopes distributed across all three domains of the E protein and are characterized by varying degrees of functional potency [39], [53]. We next performed experiments using previously characterized immune sera from five recipients of a phase I clinical trial of a live attenuated WNV vaccine [21] (A. Durbin, S. Whitehead, and colleagues, unpublished data). After a one hour room temperature incubation to allow for steady-state binding, virus-sera mixtures either were used to immediately infect cells ( Figure 6 , reference curves) or incubated an additional 23 hours at 37°C or 40°C before infection ( Figure 6 , blue and orange curves, respectively). We found that the polyclonal mixtures of antibody present in immune sera behaved in a manner similar to MAb of defined specificity.

Open in a separate window

Figure 5

Kinetic changes in neutralization occur for WNV MAbs specific for structurally distinct epitopes.

Nine serial four-fold dilutions of various WNV MAbs were incubated with mature WNV RVPs for one hour at room temperature to allow binding to reach equilibrium. RVP-antibody complexes were then incubated at 37°C or 40°C for incremental lengths of time before infecting Raji-DC-SIGNR cells. Infectivity was monitored by flow cytometry at 48 h post-infection. The reference curve represents RVP-antibody complexes added to Raji-DC-SIGNR cells immediately after the room temperature incubation. Dose-response curves are expressed relative to the infectivity of RVPs in the absence of antibody at each time point. Error bars display the standard error of duplicate infections. Results are representative of two independent experiments. MAbs selected for study were specific for epitopes on the DII-fusion loop (DII-fl) (A), the DII-central interface, DII-dimer interface, and DII-hinge interface (E48, E100, and E113, respectively) (B), the DIII-lateral ridge (DIII-lr) (C), and epitopes within DIII that fall outside of the lateral ridge (D).

Open in a separate window

Figure 6

Kinetic effects of WNV neutralization by polyclonal sera.

Nine serial three-fold dilutions of five heat-inactivated polyclonal WNV immune sera samples were incubated with mature WNV RVPs for one hour at room temperature to allow binding to reach equilibrium. RVP-antibody complexes were then incubated at 37°C or 40°C for incremental lengths of time before infecting Raji-DC-SIGNR cells. Infectivity was carried out at 37°C and monitored by flow cytometry at 48 h post-infection. The reference curve represents RVP-antibody complexes added to Raji-DC-SIGNR cells immediately after the room temperature incubation. Dose-response curves from a representative experiment are expressed relative to the infectivity in the absence of antibody at each individual time point. Error bars display the standard error of duplicate infections. Data is representative of two independent experiments.

To extend our observations to another flavivirus, we performed analogous studies with DENV serotype 1 (DENV-1) RVPs and a panel of previously characterized DENV-1 E protein-reactive MAbs [53], [54]. After incubating RVPs and antibody at 37°C for one hour to allow steady-state binding to occur, RVP-antibody complexes were either immediately used to infect cells ( Figure 7 , reference curves) or further incubated for two or seven additional hours at 37°C and 40°C before infection ( Figure 7 , dotted curves). Because it was established previously that some antibodies do not bind DENV efficiently at low temperatures [46], incubation at 37°C was used to establish steady-state binding (Figure S1). As observed in our studies with WNV, time- and temperature-dependent increases in neutralization potency were observed ( Figure 7 ). Interestingly, shorter incubation times were necessary as compared to experiments with WNV, due to a more rapid intrinsic decay rate of DENV as compared to WNV, which has previously been reported [49]. This was also demonstrated in parallel studies of WNV and DENV-1 RVPs with the cross-reactive MAb E60; significant increases in neutralization were observed more rapidly with DENV than WNV (Figure S4). Altogether, these results suggest that DENV exists in a more dynamic state than WNV.

Open in a separate window

Figure 7

A kinetic aspect of neutralization is also observed with DENV.

Nine serial four-fold dilutions of various DENV-1-reactive MAbs were incubated with DENV-1 RVPs for one hour at 37°C to allow binding to reach equilibrium. RVP-antibody complexes were then incubated at 37°C or 40°C for incremental lengths of time before infecting Raji-DC-SIGNR cells. Infectivity was carried out at 37°C and monitored by flow cytometry at 48 h post-infection. The reference curve represents RVP-antibody complexes added to Raji-DC-SIGNR cells immediately after the one hour incubation at 37°C. Dose-response curves from a representative experiment are expressed relative to the infectivity in the absence of antibody at each individual time point. Error bars display the standard error of duplicate infections. Data is representative of three (E95, E105) or two (E60, E102, E113) independent experiments. E95, E102, E105, and E113 are DIII-reactive MAbs specific for the G-strand, BC-loop, BC/DE/FG loops, and A-strand, respectively. E60 is a WNV-specific MAb that is cross-reactive for the DENV DII-fusion loop.

Discussion

The existence of high-resolution structures of the E proteins of flaviviruses, and their arrangement on the surface of the virion at different stages of the viral lifecycle, have been a powerful tool for studying how these viruses interact with the host (reviewed in [6], [26]). Interpretation of these structures in the context of biological systems is complicated by the fact that they represent the average state of what is likely a very dynamic ensemble of conformations sampled by virions at equilibrium. The structural dynamics of non-enveloped viruses have been studied extensively (reviewed in [45], [55]). Limited proteolysis studies of the Flock house virus strongly suggest the capsid proteins “breathe”, allowing proteases access to internal structures predicted to be inaccessible on an intact and non-dynamic virion [47]. Furthermore, drugs that prevent the dynamic movement of the picornavirus human rhinovirus 14 have been shown to effectively inhibit infectivity [48]. While the dynamics of enveloped viruses have been studied less extensively, examples of temperature- and time-dependent antibody reactivity have been described for several viruses [46], [56], [57]. The DENV group-reactive neutralizing MAb 1A1D-2 binds an epitope on the A-strand of E-DIII that is poorly accessible on proteins proximal to each of the three symmetry axes of the mature virion. Binding of this MAb is temperature-dependent (does not occur at 4°C) and appears to trap E proteins on the virion surface in a conformation distinct from the herringbone icosahedral arrangement of the mature virion [46].

Overall, prior studies implicating a kinetic component of antibody-mediated neutralization have focused on either a single MAb (DENV MAb 1A1D-2, influenza MAb Y8-10C2) or a panel of MAbs that recognizes a similar epitope (MAbs specific for the membrane-proximal external region (MPER) of HIV gp41), with the implication that such antibodies were atypical in their ability to bind buried or inaccessible epitopes [57], [58], [59]. In this study we provide functional evidence identifying the widespread impact of the dynamic movement of flaviviruses on neutralization by all antibodies that bind the E protein; a kinetic aspect of neutralization appears to be the rule rather than the exception.

Our data suggest that the structural dynamics of virions has the potential to modulate the potency of all antibodies that bind E proteins arrayed on the surface of the virion through changes in epitope accessibility. This is illustrated most dramatically by antibodies that bind poorly accessible determinants on the mature virion, such as the WNV DII-fusion loop-reactive MAb E53. Neutralization of mature WNV by E53 is restricted by the number of times the antibody can bind the virion. E53 binds the E protein only when associated with prM as heterotrimeric spikes that project off the surface of the immature virus particle [21], [52]. While homogeneous populations of mature WNV are not efficiently neutralized by E53 when assayed using conventional approaches, increasing the time the virus is incubated with antibody resulted in a marked increase in neutralization activity (>100-fold) ( Figure 4a and b ). Because the virus cannot revert to an immature configuration once prM cleavage occurs and the virion is released from the cell [15], this dramatic increase in neutralization can only be explained by exposure of the DII-fusion loop epitope through dynamic motion of viral E proteins.

The neutralization activity of every monoclonal (n=20) and polyclonal (n=5) antibody assayed in this study was enhanced by increasing the time the virion was exposed to antibody prior to the addition of target cells, including antibodies previously demonstrated to be non- or weakly-neutralizing using standard assays. This reflects the fact that as the number of accessible epitopes on the individual virion increases as a consequence of dynamic motion, the fraction of them that must be bound in order to exceed the stoichiometric threshold (percent occupancy) is reduced; neutralization can then occur at lower concentrations of antibody. However, the magnitude of this kinetic effect was not uniform among antibodies localizing to different epitopes. This may reflect differences in the number of times an antibody binds the average state of the virion relative to the threshold number of antibodies required for neutralization, as well as the rate at which additional epitopes are made available for binding (Figure S3). In contrast to the cryptic nature of the DII-fusion loop epitope recognized by E53, the potently neutralizing MAb E16 is specific for a relatively accessible determinant displayed on the lateral ridge of DIII [53]. Cryo-electron microscopy studies indicate this antibody binds 120 out of 180 E proteins incorporated into mature virions; the remaining 60 E proteins proximal to the five-fold symmetry axis of the virion cannot be bound due to steric conflicts among the tightly clustered DIII epitopes [27], [28]. An increase in the accessibility of these additional epitopes on the virion through dynamic motion would translate into a modest reduction in the occupancy requirements for neutralization by E16, in agreement with the 4.0-fold increase in antibody potency (n=11, range 2.3–6.8) observed in our studies after ~24 hours incubation. A similar 3.8-fold increase after ~24 hours was observed when E16 Fab fragments were used (n=2, range 3.2–4.5).

The dynamic motion of virions has the potential to increase antibody potency by providing access to otherwise cryptic antibody-binding determinants. Of note, mapping studies suggest that many epitopes on the mature virion are poorly accessible for antibody binding [38], [39], [40], [46]. Antibodies do not induce viral breathing, but rather stabilize conformations of the E protein that exist as part of the ensemble of conformations sampled by the virion at equilibrium. The longer the virion remains exposed to antibody, the more opportunities exist for engagement of an otherwise inaccessible epitope, allowing for time-dependent increases in the stoichiometry with which antibodies decorate virions. If changes in epitope accessibility are the underlying mechanism of the kinetic aspects of neutralization, there should be a limit to the increase in potency observed over time. Eventually, dynamic virion structures should expose all potential epitopes, and these will become fixed in place by antibody binding, yielding a neutralization profile determined by the relationship between antibody occupancy and the stoichiometric threshold. In support of this, increases in neutralization and changes in the ADE curves for E16 no longer occurred when incubations longer than 24 hours were performed (Figure S5).

In addition to exposing more epitopes for antibody binding, time-dependent changes in the antigenic surface of the virus particle may also allow engagement of the virion with increased affinity, via bivalent interactions among E proteins in conformations not present in the average state, as well as cooperative effects during antibody binding. That the kinetic impact on neutralization by E16 was observed using both intact antibodies and Fab fragments incapable of cross-linking virions indicates that increases in antibody potency do not reflect antibody-mediated aggregation among virions. Importantly, all of the experiments included in our study were performed using conditions of antibody excess, and yielded results that were independent of the concentration of virus in the assay. In contrast, the aggregation of antigens by antibodies is dependent on the antibody-antigen ratio.

Our results suggest that changes in the antibody-mediated neutralization of DENV occur more rapidly than with WNV. One interpretation of this result is that DENV virions are more dynamic than those of WNV, allowing more rapid access to otherwise inaccessible determinants. In the absence of antibody, preparations of DENV become less infectious at a faster rate than observed with WNV, consistent with prior studies [49]. Additionally, kinetic changes in neutralization with the cross-reactive MAb E60 occur at a faster rate with DENV-1 than WNV RVPs when compared in parallel studies (Figure S4). While we do not yet understand, in molecular terms, why viruses lose infectivity over time, one possibility is that the intrinsic decay of flaviviruses is a consequence of structural dynamics. Viruses sampling multiple conformations in dynamic equilibrium may not always return to the average state because moving backwards may no longer be the most energetically favorable path. Additional evidence of time-dependent structural changes to the virus population is demonstrated by differences in the intrinsic decay rate of WNV observed when different cell types are used to measure infectivity. The rate of decay of viruses was ~2.7-fold more rapid when assayed on Raji-DC-SIGNR cells as compared to a K562 cell line expressing the same attachment protein (n=6, p<0.0001). Thus, the observed intrinsic decay cannot be attributed solely to the physical destruction of the virus, and suggests the additional possibility that not all conformations in a heterogeneous ensemble of virions are equally infectious on different cell types. Of interest, E proteins on individual virions in conformations that may no longer contribute functionally to fusion may also stably expose a different array of epitopes.

Our data suggest that the circumstances of antibody-virion interactions may significantly impact the fate of the virion immune complex. Standard in vitro neutralization assays for flaviviruses generally include a short pre-incubation (~1 hr) of antibody and virus prior to infection of target cells; this incubation presumably allows the binding reaction between antibody and cognate epitope to reach steady-state. However, depending on the extent to which a virion is structurally dynamic (which controls the rate at which epitopes may become transiently accessible), the target cell type, and the volume of infection in vitro, this presumption may be inaccurate. Because increases in the neutralization activity of DENV-reactive antibodies that bind dynamically exposed epitopes occur rapidly (within two hours) ( Figure 7 ), the interaction of DENV with antibodies may never truly reach steady-state. From this perspective, the length of time antibody is incubated with DENV is a variable that cannot be ignored. While antibody-mediated neutralization activity measured in vitro using standard plaque reduction neutralization tests (PRNT) generally correlates with protection in vivo [39], [53], [54], [60], [61], [62], this is an imperfect relationship. Antibodies with limited neutralization activity have been shown to protect in animal models of flavivirus infection [39], [54], [62]. While this may reflect the direct contributions of effector functions of antibodies in vivo, it is also possible that existing assays of the functional properties of antibodies have limitations. Considering the contribution of the structural dynamics of the virion when designing neutralization studies merits a systematic evaluation.

The impact of the dynamic exposure of viral epitopes in vivo remains uncertain. Virtually nothing is known about the relevant concentrations and volumes that govern antibody-virion interactions in the tissues where many of the key events in the pathogenesis of these viruses occur. Kinetic changes in neutralization occur gradually over time as dynamic motion provides new opportunities for engagement of virions with a stoichiometry sufficient for neutralization (Figure S3). Thus the rate of virus entry in vivo is also an important, yet unknown, parameter that defines the extent to which this phenomenon will contribute to protection of the host. Of interest, the kinetics of WNV binding to target cells in vitro occurred rather slowly (with maximal binding requiring ~3 hours) even in the presence of the high affinity attachment factor DC-SIGNR (Figure S6). As DENV appears to be extremely dynamic, with kinetic effects on neutralization observed almost immediately, the impact of viral breathing on neutralization in vivo cannot be discounted. Because the kinetics of neutralization are increased by an elevated temperature, it is interesting to speculate that certain classes of antibodies, such as those recognizing the fusion loop epitope commonly observed in infected individuals, may function better than previously anticipated in the context of the febrile response. Resolving this question awaits the development of approaches to quantitatively and directly measure antibody-mediated neutralization in vivo.

Neither proteins, nor intact virions, are static structures. Our findings are consistent with a model in which the dynamic motion of flaviviruses provides an opportunity for antibodies to engage virions at otherwise inaccessible epitopes to reach a stoichiometry sufficient for neutralization. Given time, all of the E protein-reactive antibodies investigated were able to block virus infection, even those described originally as non-neutralizing using conventional assays [39]. These results add to the complexity of our understanding of the functional properties of antibodies and suggest new avenues of investigation and analysis into the widespread and unappreciated impact of the dynamic motion of virions as moving targets for antibody recognition.

Materials and Methods

Supporting Information

Acknowledgments

Footnotes

References