An ultrapotent human antibody neutralizes all maturation states of Zika virus Authors: Bo Shu, Thiam-Seng Ng, Xin-Ni Lim, Shuijun Zhang, Victor A. Kostyuchenko, Guntur Fibriansah, Benjamin D. McElvany, Valerie Shyn-Yun Chew, Aravinda M. de Silva, Sean A. Diehl, Shee-Mei Lok Published: 2025 ABSTRACT SignificanceZika virus causes microcephaly in fetuses and no vaccines or therapeutics currently exist against it. Mature and immature flavivirus particles are infectious. Here, we showed the cryoelectron microscopy (cryoEM) structures of an ultrapotent A9E human antibody, complexed with both mature (mZIKV) and immature (immZIKV) Zika virus, and the antibody neutralization mechanism. One important characteristic is that Fab A9E can distort both mZIKV and ImmZIKV particle structures. Additionally, Fab A9E or IgG A9E LALA mutant can abolish or reduce the overall infection to myeloid cells when added to other infection enhancing antibody DV62.5:immZIKV complexes. Thus, antibody A9E represents a promising potential prophylactic and therapeutic candidate, as it is effective against all maturation states of Zika virus. Zika virus (ZIKV) is a member of the Flaviviridae family, other members of which are also important human pathogens such as dengue virus, yellow fever virus, and West Nile virus. From 2015 to 2016, 3,670 microcephaly cases were recorded in South America, correlated with the spread of ZIKV (1). 404 microcephaly cases were also confirmed in fetuses of ZIKV infected pregnant women (2). In the fetus of pregnant mice, ZIKV can replicate in the embryonic brain and infect neural stem cells, consequently resulting in microcephaly (3). Currently, there are no treatments or approved vaccines against ZIKV. The cryoEM structure (4, 5) of the mature Zika virus (mZIKV) surface shows 180 copies of envelope (E) protein organized in icosahedral symmetry with 60 asymmetric units. Within an asymmetric unit, there are three individual E proteins arranged around fivefold, twofold, and threefold vertices, hereafter named as molecules A, B, and C, respectively (SI Appendix, Fig. S1A). The E proteins exist as dimers, three of which lie parallel to each other forming a raft, and with each raft containing two asymmetric units. Thirty rafts are organized in a herringbone pattern on the virus surface. The mZIKV has been shown to have higher structural thermal stability compared to other flaviviruses such as dengue, as the interactions between their surface E proteins are tighter (4). E protein is the major antigenic structure on the viral surface. Its ectodomain is composed of three domains: EDI, EDII, and EDIII (SI Appendix, Fig. S1A). EDI contains one glycosylation site, and at the tip of EDII is the fusion loop, which participates in viral:endosomal membrane fusion. EDIII is an immunoglobulin-like domain and is thought to participate in receptor binding. The primary function of E protein is to facilitate viral entry into cells by promoting attachment to cell surface receptors, internalization, and subsequent low pH mediated fusion with the endosomal membrane. The E protein postfusion crystal structure (6) has a trimeric arrangement, with the E proteins positioned approximately parallel to each other and with their fusion peptides exposed at the tip of the trimer. There is a major conformational change in the positioning of EDIII relative to the prefusion form, with it folding toward EDII. In cell culture, flavivirus particles are often observed to be released from infected cells as a mixture of both mature and immature virus particles, likely due to the inefficiency of the virus maturation process occurring inside the cell. Immature viruses are also likely produced during a natural infection, as there are strong anti-prM antibodies elicited in infected individuals (7). Studies have shown the normally noninfectious immature flavivirus can become infectious when complexed with anti-prM antibodies and certain anti-E protein antibodies (8). This antibody-immature virus complex then infects Fcγ receptor positive myeloid cells. Thus, although immature viruses are not inherently infectious, they can contribute to pathogenicity in the presence of specific antibodies. The surface of immature ZIKV (immZIKV), contains 180 copies of the prM:E heterodimer (SI Appendix, Fig. S1B). The prM molecule is the precursor molecule of the M protein. The prM:E surface proteins are arranged differently on the surface of the immZIKV compared to the mZIKV, as although both the immature and mature viruses have icosahedral symmetry, the prM-E heteroproteins are organized as trimeric spikes (SI Appendix, Fig. S1B) in immZIKV. Therefore, immZIKV has different antigenic characteristics compared to mZIKV. A human monoclonal antibody (HMAb) A9E specific to ZIKV has been shown to be highly neutralizing (IC50 = 5 to 10 ng/mL) to mZIKV in vitro and has been demonstrated to have strong protective effect against ZIKV infection in mice (9). Here, we show that HMAb A9E mainly prevents infection by blocking mZIKV from attaching to cells. The reduced attachment is likely due to the ability of the antibody to aggregate virus particles, distort its surface structure and block virus from binding to cellular receptors. We determined the cryoelectron microscopy (cryoEM) structures of the Fab fragment of this antibody complexed to both mZIKV and immZIKV to resolutions of 2.8 Å and 7.5 Å, respectively. These structures show it binds largely to EDI, EDIII, and their linker. We observed that of the 180 potential epitopes on the surface of ZIKV, only 140 sites were occupied by A9E on mZIKV and 60 sites on immZIKV. Addition of higher concentrations of antibodies to both immZIKV and mZIKV resulted in breakage of viral particles, with immZIKV being particularly susceptible to it. While most antibodies can cause ADE at some concentrations, potently neutralizing Abs only support ADE at low, narrow antibody concentration ranges, whereas poorly neutralizing antibodies support ADE at a much wider window of antibody concentrations (10). ImmZIKV uses ADE to infect cells. ADE assays of the immZIKV infection of gamma-Fc receptor positive myeloid cell show IgG A9E antibody causes ADE in a much narrower concentration range than another antibody (IgG DV62.5). When Fab A9E (the Fab form is unable to cause ADE) or HMAb A9E LALA mutant are added to IgG DV62.5/immZIKV complex, they can reduce or eliminate the ADE effect caused by IgG DV62.5. This suggests that the LALA mutant of HMAb A9E, which is unable to bind to Fc receptors on myeloid cells, can be a potential prophylactic and therapeutic antibody candidate effective against all maturation states of ZIKV. Results Neutralization Mechanism of HMAb A9E against mZIKV. Antibodies can neutralize viruses by various mechanisms including the aggregation of virus particles, blockade of virus binding to the host receptor and inhibition of virus:endosomal membrane fusion. To investigate whether the HMAb A9E could aggregate virus particles, different molar ratios of IgG:Virion were mixed and then the hydrodynamic size of particles was measured by dynamic light scattering. Nonaggregated viruses are similar in size to single virus particles, and they were detected at low antibody concentrations (≤10 antibodies per virus). Aggregated virus particles were detected starting at an IgG:virion molar ratio of 9:1, gradually peaked at 180:1 and then decreased at higher IgG concentrations. There was a sharp increase in neutralization activity between IgG:virion molar ratios of 60:1 to 180:1, which corresponds to the peak of virus aggregation. However, neutralization activity continued to increase at higher IgG concentrations, even though aggregation had decreased (Fig. 1A). Neutralization mechanism of antibody A9E. (A) HMAb A9E can induce virus aggregation at certain IgG:virion concentrations. HMAb A9E was detected to induce viral particle aggregation at IgG:virion molar ratios between 60:1 and 180:1. At these concentrations, we also observed increased virus neutralization activity. At higher IgG:virion molar ratios, when virus aggregation was reduced, HMAb neutralization capability continued to increase. This may suggest that at different IgG:virion molar ratios, the HMAb might employ different neutralization mechanisms. The hydrodynamic size (Left y-axis) of ZIKV:A9E IgG complexes in different IgG:virion molar ratio (x-axis) were measured by dynamic light scattering (black curve). The neutralization profile of these ZIKV:IgG complexes (red curve) were shown as neutralization index (Right y-axis) calculated as the log10 fold reduction of the virus titer compared to the virus-only control. Data are the mean ± SEM from three independent experiments. (B) Pre- and postattachment neutralization assays results showed antibody A9E are highly neutralizing (PRNT50 in nanogram/mL range), although their neutralization activities are more potent in the preattachment (when HMAb is added to virus before exposure to cells) than postattachment (HMAb added after virus has already attached to cells) neutralization tests. PRNT50 values indicate the antibody concentration that neutralized 50% of the plaque forming units were determined by nonlinear regression using GraphPad Prism v5.0. Data are the mean ± SEM from three independent experiments. (C) Antibody A9E inhibits virus attachment to cells. Preattachment RT-qPCR assay results showed that A9E prevents ZIKV from attaching to cells. Isotype IgG and no antibody controls were included. (D) Virus:liposomal membrane fusion assays showed that HMAb A9E has some virus:liposomal membrane fusion inhibition activity. The percentage of fusion was calculated by (1) normalizing the signal Em = emission of sample before adding Triton/emission of the sample after adding Triton, (2) fusion(%) = (Em − EmpH8)/(EmpH5.5 − EmpH8), (3) The extent of fusion was normalized against virus without antibody. Data are the mean ± SEM from three independent experiments. Significance was determined by one-way ANOVA with Dunnett’s post test compared to isotype control. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). To investigate whether HMAb A9E blocks virus infection at a pre– or post–virus attachment step, we compared the neutralization when HMAb was added before or after viruses were allowed to bind to precooled cells at 4 °C, that is, pre- and postattachment neutralization assays, respectively. Our results show that HMAb A9E exhibited strong postattachment neutralization activity (PRNT50 = 55.1 ng/mL), although it was a sevenfold poorer than that of the preattachment neutralization test (PRNT50 = 8.8 ng/mL) (Fig. 1B and SI Appendix, Table S1). This suggests that HMAb A9E can neutralize mZIKV infection both prior to and after virus attachment to cells. To further test whether the antibody prevents virus attachment to cells, we quantitated the amount of antibody A9E:virus complex particles attached to the cell surface. Antibody and virus were mixed and then incubated with the BHK cells at 4 °C to prevent endocytosis. Reverse-transcription quantitative polymerase chain reaction (RT-qPCR) was used to quantify the amount of virus bound to the cells. Uncomplexed viruses or those exposed to isotype control antibodies served as negative controls and were considered to represent 100% virus bound to cell surface. Our results showed that HMAb A9E at the concentrations tested (0.1 to 10 μg/mL) inhibited ~70 to 80% of virus attachment to cells (Fig. 1C). This suggests that antibody A9E could strongly inhibit viral attachment to cells. Next, we tested whether the A9E IgG could block fusion of ZIKV to liposomes. Prepared liposomes were first labeled with SRB (Sulforhodamine B). During virus-liposome fusion, the SRB-labeled liposomes were diluted, leading to an increase in fluorescence emission. Uncomplexed viruses or viruses exposed to an isotype antibody virus served as negative controls representing 100% fusion. Overall, HMAb A9E showed poor virus fusion inhibitory activity. At the highest antibody concentration tested (100 μg/mL), fusion was inhibited by about 40% and fusion inhibition activity decreased with decreasing concentrations of HMAb (Fig. 1D). In conclusion, these results suggest that antibody A9E can prevent viral infection mainly by inhibiting steps prior to virus binding to cells, predominantly by inhibition of virus attachment to cells. CryoEM Structural Studies of Fab A9E Complexed with mZIKV. To determine the cryoEM structure of the mZIKV complexed with Fab A9E, we mixed mZIKV strain H/PF/2013 with Fab A9E at molar ratios of 3Fab:3E protein at 4 °C. We collected 8,595 micrographs and obtained 183,648 intact spherical Fab:mZIKV particles for single particle cryoEM image reconstruction. The icosahedrally averaged cryoEM map of the intact Fab:mZIKV particles was determined to an overall resolution of 2.8 Å (Fig. 2A and SI Appendix, Fig. S2), as measured by the gold standard FSC curve cutoff at 0.143. Superimposition of an E protein dimer from the uncomplexed mZIKV cryoEM structure (PDB:5IZ7) onto our Fab:mZIKV cryoEM structure suggests that Fab A9E did not induce any significant structural rearrangement to the E proteins dimer (RMSD = 0.87 Å) (SI Appendix, Fig. S3A). However, the Fab molecules were not present at full occupancy on the virus particle, as there were only 140 copies of Fab bound to the 180 copies of E proteins on the virus surface. We observed strong Fab densities bound to the E protein molecules A and B while that on molecule C (around the threefold vertices) was much weaker (SI Appendix, Fig. S3B), consistent with our occupancy calculation (Fig. 2B). CryoEM Maps of the mZIKV complexed with Fab A9E (molar ratio of 3Fab:3E). (A) Surface of the mature ZIKV complexed with Fab A9E at 2.8 Å resolution. Densities corresponding to the E protein layer and Fabs are colored in cyan and magenta, respectively. The white triangle indicates an asymmetric unit with the fivefold, threefold, twofold vertices labeled. (B) Occupancy of Fab A9E as estimated by the program UCSF Chimera. (C) Localized reconstruction of regions around the threefold vertex shows three structural classes, each with only one clear Fab density (gold). (D) mZIKV particles become distorted or broken when exposed to high concentrations of Fab A9E. Micrographs of the ZIKV:Fab A9E complex at Fab:E molar ratios of 2Fab:3E and 3Fab:3E showed that most of the particles are still round and intact. At a ratio of 4Fab:3E, we observed more broken and distorted particles (red circle). The Upper Right corner inset shows a zoom-in view of a typical intact (black box)/distorted particle (red box). (E, Left) Comparison of reconstructed maps of the intact and distorted particles. A quarter of the cross-section of each cryoEM map is shown. (Right) 2D average of the intact and distorted particles. Both maps and 2D averages show the lipid bilayer membranes have changed from a polygonal shape (in intact particles) to a circular shape (in distorted particles). In addition, the E protein layer of the distorted particles has moved outward. We interpreted the Fab densities by fitting a Fab A9E homology model. We first fitted the stronger Fab densities bound to the E protein molecules A and B. Comparison of the structures of Fab bound to E protein molecules A and B showed that they are highly similar (RMSD of 0.1 Å) (SI Appendix, Fig. S3C). We then fitted this Fab:E protein complex onto the Fab density around the threefold vertex (corresponding to E protein molecule C). We observed only a weak scattered Fab density in the middle of the three E protein molecules C around this vertex (Fig. 2 A and B and SI Appendix, Fig. S3B). Localized reconstruction of the region around the threefold vertex and classification detected three structural classes, each showing only one Fab binding around the vertex (Fig. 2C). This confirms that, at any given time, only one Fab can bind to this vertex. We also observed that the bound Fab would partially shield the other two available epitopes on the neighboring E proteins (SI Appendix, Fig. S4A). We superimposed the Fab:E protein structures onto all three E protein molecule positions around this vertex and observed that the Fabs would clash with each other (SI Appendix, Fig. S4B). Thus, the incomplete occupancy is due to steric hindrance caused by the bound Fab, inhibiting other Fabs from binding. At high Fab concentration (4.5Fab:3 E molar ratio), we observed more mZIKV particles with a distorted surface. We speculate that the excess Fabs might cause more binding to the molecules C around this vertex. However, this would require the virus surface to undergo “breathing.” In the 3 Fab:3 E protein molar ratio sample, 2D classification revealed that, while the majority of particles had an intact surface with bound antibody, there were some particles exhibiting distorted surfaces. When we further increased the molar ratio to 4.5 Fab:3E proteins, examination of the cryoEM micrographs showed even more distorted virus particles (Fig. 2D). From both 2D averages and also the reconstructed low-resolution map of the distorted particles, their E protein layer was observed to be located at a higher radius and, together with the bilayer lipid membrane, they had lost their typical polygonal shape normally observed in intact particles (Fig. 2E). These results suggest that higher concentrations of Fabs can distort mZIKV mature structure, whereas this was not observed with Fab C10 bound to mZIKV at the same concentration (SI Appendix, Fig. S5). A9E Epitope on Mature ZIKV. We used the 2.8 Å resolution mZIKV:Fab complex cryoEM structure for the identification of A9E epitope residues. The local resolution of the Fab-to-E protein interface densities is about 3.1 to 3.3 Å resolution (Fig. 3A) and the bulky side chain densities are clearly resolved (Fig. 3B). A cutoff of 3.5 Å distance between Fab and E protein side chains to side/main chains was used to identify hydrogen bonds/electrostatic interactions, and <4 Å for hydrophobic interactions. The epitope spans EDI, EDIII, and their connecting EDI:EDIII linker. The majority of the Fab footprint is on EDI comprising the glycosylation loop, the G0-H0 loop, and the EDI:EDIII linker region (Fig. 3C and SI Appendix, Fig. S6 and Table S2). This is consistent with previous findings on A9E escape mutants and shotgun mutation studies which showed the EDI-III linker and lateral ridge of EDIII may form part of the epitope (9). An open book presentation of the charges of the Fab paratope and epitope shows charge complementarity (Fig. 3D). Fab A9E bound to an epitope that spans across EDI and EDIII locking EDI-EDIII linker. (A) CryoEM map of A9E complexed with mZIKV colored according to local resolution. (B) Zoom-in view of the E protein and Fab A9E interface showing that cryoEM densities of the large side chains are well resolved. The main chain of the Fab heavy and light chains and E protein are colored in cyan, magenta, and pink, respectively. Nitrogen and oxygen atoms are colored in blue and red, respectively. (C) A9E epitope (green spheres) on an E protein dimer. The Right panel shows a zoom-in view of the epitope (green) with its strands labeled. The E protein EDI, EDII, and EDIII are colored red, yellow, and blue, respectively. (D) Open book representation showing the electrostatic potential of the interface between the epitope and paratope of mature ZIKV:A9E Fab structures. Positive, negative, and neutral charged residues are colored in blue, red, and white, respectively. Residues at the interface are indicated. There are five interacting patches between the Fab and the E protein (Fig. 3D and SI Appendix, Table S2): i) CDR2 of the Fab heavy chain with the EDI glycosylation loop; ii) The CDR3 of Fab heavy chain with EDI G0-H0 loop; iii) The CDR3 of the Fab light chain also with EDI G0-H0 loop; iv) CDR1 of the Fab light chain and EDI-EDIII hinge; v) The CDR3 of the Fab light chain and the Dx in the EDIII. Interaction of Fab A9E with ImmZIKV Particles. Fully immature and partially immature virus particles are often found in standard flavivirus preparations, as the virus maturation process inside the cell is inefficient (11). Since these viruses with different maturation states can be infectious (8), we examined the interaction of Fab A9E with fully immZIKV particles. The cryoEM micrographs of immZIKV complexed at different molar ratios showed that at a molar ratio of 1.5 Fab:3 E protein, most viral particles are structurally intact while others become larger deformed particles (Fig. 4A). At a molar ratio of 3 Fab:3 E protein, most particles become larger and are deformed. Therefore, we determined the cryoEM structure of the intact immZIKV:Fab complex at a molar ratio of 1.5 Fab:3 E proteins, to 7.5 Å resolution (Fig. 4B and SI Appendix, Fig. S7 A and B). The cryoEM map shows that 60 of the 180 copies of the E proteins on the immZIKV surface are bound by Fabs (Fig. 4B). Superimposition of a prM-E heterodimer of uncomplexed immZIKV (PDB code:6LNU) (12) onto the Fab:prM-E complex in our cryoEM structure shows that Fab binding did not induce large conformational changes of prM and E proteins (SI Appendix, Fig. S7C). Of three individual E proteins (molecules A-C) within an asymmetric unit, only molecule A is bound by Fab (Figs. 4 B and C and 5A). CryoEM map of the immZIKV complexed with Fab A9E. (A) CryoEM micrographs of immZIKV virus incubated with Fab A9E at different molar ratios. (Scale bars: 50 nm.) In cryoEM micrographs of immZIKV:Fab A9E at a molar ratio of 1.5:3, some particles are distorted (red arrows) while most particles are distorted at a molar ratio of 3:3. (B) 7.5 Å resolution cryoEM map showing the surface of intact immZIKV: A9E Fab complex particles (1.5Fab:3E molar ratio). E proteins, prM proteins, and Fab A9E densities are colored in cyan, blue, and magenta, respectively. The black triangle indicates an asymmetric unit and the fivefold, threefold, twofold vertices are labeled. (C) Localized reconstruction of the region in (B) circled in white, showing a 6.5 Å resolution cryoEM density covering one prM-E trimeric spike. (D) Superimposition of the Fabs within E:Fab complexes from the mZIKV and immZIKV samples show that the Fab epitopes largely overlap. The E proteins of mZIKV and immZIKV are colored gray and tan, respectively. The heavy chain of the Fab on mZIKV and immZIKV is colored blue and orange-red, respectively. Residues bound by Fabs on mZIKV and immZIKV are indicated by magenta and green spheres, respectively. (E) Amino acid residues forming the A9E epitopes on the mZIKV and immZIKV. Fab A9E binding destabilizes immZIKV particles and could prevent ADE caused by other antibodies. (A) Superimposition of our Fab (blue ribbon):E complex structure onto E protein molecules B and C shows that, if bound, the Fabs will clash with neighboring E proteins (gray surfaces). Densities of E protein molecules A, B, and C are colored in light blue, light green, and tan, respectively. The observed Fab density bound to molecule A is colored orange. (B) HMAb A9E supports ADE infection immZIKV of myeloid K562 cells within a much narrower range of IgG concentrations than HMAb DV62.5. (C) Addition of different Fab or IgG A9E LALA mutant concentrations can either reduce or abolish the ADE activity of the immZIKV:IgG DV62.5 (1,000 ng/mL) complex. IgG A9E (LALA mutant) shows very high potency. The immZIKV:IgG DV62.5 (1,000 ng/mL) complex causes ADE activity as shown in (B). We added various concentrations of Fab A9E or IgG A9E (LALA mutant) to this complex and determined their % inhibition of ADE activity. Data in panels B and C are the mean ± SEM from three independent experiments. To further improve the resolution of the region around the threefold vertex, we applied localized reconstruction to this vertex, resulting in only one structural class with one Fab bound in this region, in which the final resolution was improved to 6.5 Å (Fig. 4C). We then fitted the Fab and E protein structures into the cryoEM densities. Because the immZIKV:Fab complex structure was determined at a lower resolution (6.5 Å) compared to the mZIKV structure (2.8 Å), we identified potential Fab-E protein interactions by analyzing Cα atom pairs within an 8 Å cutoff distance (13) (SI Appendix, Table S3). The precision of the mapped epitope is thus limited by the resolution constraint. Superimposition of the Fabs bound to the E proteins of mZIKV and immZIKV, and also the comparison of the residues forming the epitopes, showed that they largely overlapped (Fig. 4 D and E). 2D classification of the Fab:immZIKV sample at a molar ratio of 0.5 Fab:1 E shows there are some virus particles with their spiky surface bound by Fabs and they have a tight viral surface, similar to the uncomplexed immZIKV. However, there are also other Fab-bound particles that exhibit a distorted lipid bilayer membrane (SI Appendix, Fig. S8). In addition, Fabs at increased concentrations (3 Fab:3 E protein molar ratio) (Fig. 4A) appear to distort most particles. We speculate that high concentrations of antibodies might cause distortion of the overall immZIKV structure by binding to potential epitopes on Mol A or Mol B (Fig. 5A). The ADE Activity of HMAb A9E to ImmZIKV. Immature flaviviruses normally exhibit poor infectivity, but can become more infectious when complexed with antibodies, enabling the IgG:immZIKV to infect Fcγ receptor-positive receptor myeloid cells by antibody-dependent enhancement (ADE). Neutralizing or nonneutralizing antibodies are capable of ADE but with different windows of antibody concentrations. Highly neutralizing antibodies cause ADE at a narrower window of concentrations compared to poorly neutralizing or nonneutralizing antibodies. To test for ADE activity of HMAb A9E on the immZIKV, a series of concentrations of antibody was incubated with immZIKV and the mixture was added to Fcγ receptor positive K562 cells for infection. Three days postinfection, we harvested the supernatant and determined the virus titer. Our results showed that IgG A9E supports ADE at a much narrower window of concentrations (Fig. 5B) than the positive control IgG DV62.5. IgG DV62.5 binds to an epitope that spans across prM and E proteins (12), a distant site from the A9E epitope (SI Appendix, Fig. S9). IgG DV62.5 supports ADE at all tested concentrations starting from 0.64 ng/mL, while IgG A9E only supports ADE activity between 0.64 to 400 ng/mL. To determine whether IgG A9E could suppress or reduce the antibody-dependent enhancement (ADE) effect of IgG DV62.5 on immature Zika virus (immZIKV), we incubated immZIKV with a series of concentrations of either Fab A9E (which cannot bind to K562 cells) or the LALA mutant IgG A9E (which ablates Fcγ receptor binding) together with a constant enhancing concentration (1 μg/mL) of DV62.5. The ADE assay demonstrated that both Fab A9E and LALA-mutant IgG completely abolished ADE activity mediated by the IgG DV62.5:immZIKV complex at concentrations from 8 nM (Fig. 5C). The LALA mutant IgG A9E is especially potent as it has 50% inhibitory effect even at 0.064 nM concentration. Discussion To determine why HMAb A9E is ZIKV-specific, we compared the sequence of epitope to other flaviviruses [West Nile (WNV), Japanese encephalitis (JEV), and dengue serotypes 1-4 (DENV1-4)] (SI Appendix, Fig. S6). The glycosylation loop on EDI forms a part of the Fab A9E epitope and the length and sequence of this region is highly diverse among flaviviruses. This part of the epitope thus confers specificity of the Fab binding activity to ZIKV. HMAb A9E is highly inhibitory at steps prior to and at virus attachment stages of the cell infection cycle (Fig. 1C). Since EDIII forms part of its epitope and it has been thought to contain a flavivirus receptor binding site (14), antibody binding might directly inhibit virus attachment to host receptors. We also observed that at certain IgG concentrations, HMAb A9E can cause virus aggregation, which might reduce the number of viruses capable of binding to cells. Third, at high antibody concentrations, HMAb A9E distorted the structure of both mature and immature virus particles which might also reduce the amount of virus capable of infecting cells. Other potent antibodies such as C10 (15) and 1A1D-2 (16) against DENV2 and ZKA190 (17) against ZIKV also showed distortion of either E protein dimer conformation or the quaternary structural arrangement of E proteins on the viral surface. Our findings suggest antibody-induced virus particles aggregation may have a moderate neutralizing effect. At virus:antibody ratios between 60:1 and 180:1, we observed that virus particles had aggregated, and the same samples showed a moderate increase in the neutralization index (Fig. 1A). At ratios above 180:1, we detected less aggregated virus but the neutralization effect continued to increase. There are two possible interpretations for our results: 1) at low antibody concentrations, antibodies could aggregate virus particles causing moderate neutralization, and at higher concentrations, the mechanism switches to predominantly inhibition of virus attachment to cells, or alternatively, 2) antibody induced virus aggregation does not contribute substantially to neutralization, and the predominant mechanism is by inhibition of virus attachment to cells with increasing antibody concentrations. It is also possible that the contribution of antibody induced virus aggregation and its effect on inhibition of virus attachment and neutralization is a more nuanced event than described in refs. 1 and 2. Neutralization mechanisms involving antibody induced virus particle aggregation have been described in reviews (18) and also shown in our previous studies on flaviviruses (15–17) and alphaviruses (19). We observed that aggregation of viruses by HMAb A9E occurs in an antibody concentration–dependent manner. At low concentrations where we observed virus aggregation, each arm of the IgG may bind to a different viral particle. At high/saturating antibody concentrations where no/low levels of aggregation were observed, it is possible that only one arm of IgG is bound while the other arm is not. Alternatively, they can bind to two epitopes within a single virus particle. The two Fabs on an IgG are separated by a distance of <87 Å. We measured the distance between each Fab A9E on the mZIKV surface and found that the longest Fab–Fab distance is <75 Å (SI Appendix, Fig. S10). It is therefore possible for an IgG to bind to two neighboring epitopes on the viral surface. In the postfusion E protein structure, EDIII moves by approximately 70° relative to the prefusion structure, and this movement is enabled by the EDI-EDIII linker (20). Since the A9E epitope spans EDI and EDIII including the linker, A9E may be able to interfere with virus fusion to the endosomal membrane. However, we observed poor fusion inhibition activity in our experiments (Fig. 1D). Examination of the cryoEM structure of the Fab:mZIKV complex shows incomplete Fab occupancies especially around the threefold vertex. This may explain the inability of HMAb A9E to completely block virus:fusion activity, as there are some unbound E proteins. The Fabs seem to be able to distort both mZIKV and immZIKV particles at high Fab concentrations (Figs. 2D and 4A). On the mZIKV at 3:3 molar ratio, strong Fab densities above E protein molecules A and B suggest high Fab occupancies, whereas the density on molecule C around the threefold vertex is much weaker, indicating low occupancy (Fig. 2 A and B). This is consistent with our localized reconstruction of the region around the threefold vertex showing three structural classes, each containing one Fab bound to only one of the three molecules C (Fig. 2C). The ability of the flaviviral surface to “breathe” has previously been shown in DENV (21). If all three molecules C are bound by Fabs, the Fabs could then push the E proteins apart from each other, thereby distorting the virus surface structure. Compared to mZIKV, the immZIKV is even more prone to structural distortion by Fab A9E. At 1.5Fab:3E molar ratio, some immZIKV particles already appeared distorted and this effect was exacerbated by increased Fab concentrations (3 Fab:3 E molar ratio), leading to widespread broken/distorted virus particles (Fig. 4A). Thus, for cryoEM structural determination, we focused on 1.5Fab:3E molar ratio immZIKV samples. The intact particles show Fabs bound to a third of the E proteins (molecule A) on the virus surface (Fig. 4 B and C). The bound Fab did not induce any structure changes on these intact virus particles. To understand why E protein molecules B and C were not bound by Fabs on these intact immZIKV particles, we superimposed the Fab:E complex on them (Fig. 5A). This showed that, if bound, the Fab would clash with other neighboring prM-E proteins. However, these epitopes are partially exposed, and it is therefore possible that they might become available when the virus surface undergoes breathing. Hence, we speculate that high concentration of Fabs might result in additional Fab binding to molecules B or/and C, and that this binding might push surface proteins apart, thus distorting virus surface structures. ZIKV is closely related to DENV both genetically and antigenically, and since ZIKV infections occur in regions around the world where DENV serotypes cocirculate, we cannot exclude the possibility of ADE infection caused by cross-reactive antibodies against different flaviviruses. Although little is known about ADE facilitating ZIKV infection in the context of natural human infections in epidemiologic studies, studies using in vitro cell culture systems (22), mouse (3, 23), and human placental explants (24) have shown that antibodies can facilitate enhanced ZIKV infection. A study using immature West Nile virus, another flavivirus, showed that mice can succumb to infection when the virus is complexed with anti-prM antibody (25). ImmZIKV primarily uses the ADE mechanism to infect cells (8). HMAb A9E shows poor ADE activity—it supports immZIKV ADE infection of Fcγ receptor positive K562 myeloid cells at a much narrower window of concentrations compared to HMAb DV62.5 (Fig. 4B). We observed by cryoEM that the structural integrity of the surface of immZIKV is highly sensitive to the binding of Fab A9E, which might explain the narrow range of IgG concentrations supporting ADE, as the virus structure would have been broken at high antibody concentrations. Other studies on DENV in mouse systems showed suppression of ADE by strongly neutralizing serotype specific mAbs or LALA variants (26). This may explain why ADE does not occur with repeat DENV infections by the same serotype, as the ADE antibodies elicited from the first infection are suppressed by the highly potent antibodies against the same serotype in subsequent infections. However, in a secondary infection with a different dengue serotype, the same antibodies from the first infection become effective in facilitating ADE because of the lack of protective neutralizing antibodies against the second serotype. Due to the inefficiency of the flavivirus maturation process inside the cell (27), there is likely a continuum of different flaviviral maturation states being released during a natural infection. As these fully mature, partially mature and immature viruses are infectious (28), it is important to investigate ways to inhibit them. Our experiments used highly mature and immature ZIKV particles but we did not use partially mature virus because it is difficult to perform cryoEM reconstruction of these highly heterogenous virus particles. Despite the fact that partially mature virus were previously observed to have distinct surface patches of surface consistent with the fully mature or immature form (29), we can predict how the antibodies would react to these patches. For mature virions or the mature patches on partially mature viruses, antibodies A9E could easily bind to at least 2/3 of the E proteins and could inhibit virus attachment to cells. At higher concentrations, A9E could distort the mature virus particles or the mature patches on partially mature viruses. In the case of fully immature viruses or the immature patches on partially mature viruses, even low concentrations of Fab A9E could distort the particle structure. Indeed, we have tested the LALA mutant of the HMAb A9E in the context of ADE induced infection of immZIKV when complexed with enhancing anti-prM antibody DV62.5, and found that HMAb A9E LALA mutant potently inhibited virus infection. Thus, it is likely that the HMAb A9E LALA mutant binding can distort virus particles when bound to DV62.5. In conclusion, our work demonstrates a very potent antibody that can structurally distort all maturation states of ZIKV and block important pathways of the virus lifecycle. These characteristics potentially make this antibody a good candidate for prophylactic and therapeutic uses. Materials and Methods A detailed description of all materials and methodology is included in SI Appendix, Materials and Methods. This includes the cells and virus, mature Zika virus and immature Zika virus production, plaque reduction neutralization test, pre- and postattachment neutralization assays, dynamic light scattering assay, attachment RT-qPCR assay, SRB-labeled virus fusion assay, cryoEM sample preparation, data collection, processing, and analysis. Sample Preparation for CryoEM. The Fab A9E was mixed with mZIKV/immZIKV at different molar ratios. The mixture was incubated for >30 min on ice and then applied to cryoEM grids before plunge freezing into liquid ethane by using a Vitrobot Mark IV (FEI, the Netherlands). Data Collection and Image Processing. The immZIKV complexes were imaged using the 300KV Titan Krios electron microscope (ThermoFisher) with a FEI Falcon II direct electron detector, while the mZIKV complexes were imaged on the same microscope equipped with a Gatan K3 direct electron detector with a postcolumn energy filter operated in zero-loss mode. Images for virus_Fab complexes were collected in movie mode, with a total exposure of 1.6 s and total dose of 20 e Å−2. In total, 2,865 and 8,595 micrographs were collected for Fab_A9E_immZIKV and Fab_A9E_mZIKV, respectively. The images were acquired at underfocus in 0.5 to 2.5 µm range. The motion correction and astigmatic defocus parameters were estimated with Patch CTF estimation in CryoSPARC (30). Particle picking, 2D classification, and refinement were performed using cryoSPARC. The reconstruction process is detailed in SI Appendix, Materials and Methods. Supplementary Material Appendix 01 (PDF)