key: cord-0289434-82m71vt0
authors: Singh, Tulika; Hwang, Kwan-Ki; Miller, Andrew S.; Jones, Rebecca L.; Lopez, Cesar A.; Giuberti, Camila; Gladden, Morgan A.; Miller, Itzayana; Webster, Helen S.; Eudailey, Joshua A.; Luo, Kan; Holle, Tarra Von; Edwards, Robert J.; Valencia, Sarah; Burgomaster, Katherine E.; Zhang, Summer; Mangold, Jesse F.; Tu, Joshua J.; Dennis, Maria; Alam, S. Munir; Premkumar, Lakshmanane; Dietze, Reynaldo; Pierson, Theodore C.; Ooi, Eng Eong; Lazear, Helen M.; Kuhn, Richard J.; Permar, Sallie R.; Bonsignori, Mattia
title: Zika virus-specific IgM elicited during pregnancy exhibits ultrapotent neutralization
date: 2021-11-24
journal: bioRxiv
DOI: 10.1101/2021.11.23.469700
sha: 48659098cc0cfb26572b563b14ae39642e19667c
doc_id: 289434
cord_uid: 82m71vt0

Congenital Zika virus (ZIKV) infection results in neurodevelopmental deficits in up to 14% of infants born to ZIKV-infected mothers. Neutralizing antibodies are a critical component of protective immunity. Here, we demonstrate that plasma IgM responses contribute to ZIKV immunity in pregnancy, mediating neutralization up to three months post symptoms. From a ZIKV-infected pregnant woman, we established a B cell line secreting a pentameric ZIKV-specific IgM (DH1017.IgM) that exhibited ultrapotent ZIKV neutralization dependent on the IgM isotype. DH1017.IgM targets a novel envelope dimer epitope within Domain II. The arrangement of this epitope on the virion is compatible with concurrent engagement of all ten antigen-binding sites of DH1017.IgM, a solution not achievable by IgG antibodies. DH1017.IgM protected against lethal ZIKV challenge in mice. Our findings identify a unique role of antibodies of the IgM isotype in protection against ZIKV and posit DH1017.IgM as a safe and effective candidate immunoprophylactic, particularly during pregnancy. Key points Plasma IgM contributes to early ZIKV neutralization during pregnancy Ultrapotent neutralization by pentameric DH1017.IgM mAb depends on isotype DH1017.IgM can engage all binding sites concurrently through different angles of approach DH1017.IgM protects mice against lethal ZIKV challenge

Zika virus (ZIKV) emergence in the Americas revealed it to be congenitally transmitted, causing 51 microcephaly and other birth defects in up to 14% of infants born to ZIKV-infected pregnancies 52 (Reynolds et al., 2017) . Even infants born seemingly healthy following maternal ZIKV infection 53 may develop neurodevelopmental defects years later (Nielsen-Saines et al., 2019). As ZIKV 54 mostly results in mild febrile disease in healthy adults, its greatest disease burden arises through 55 infections in pregnancy. In the 2015-2016 epidemic, ZIKV re-emergence in a susceptible 56 population led to 11,000 cases of microcephaly in Brazil alone (Campos et al., 2018) . There is no 57

licensed vaccine for ZIKV, and the vaccine development and testing pipeline has stalled due to 58

limited ZIKV circulation (Morabito and Graham, 2017) . lasting up to multiple years, when the typical half-life of IgM is 5 days, suggests that ZIKV-reactive 76

IgM expressing B cells are specifically activated and expanded upon ZIKV infection (Lobo et al., 77 2004) . While neutralizing activity is primarily attributed to IgG isotype antibodies, IgM may have 78 an underappreciated role in ZIKV immunity. 79

The role of ZIKV IgM and IgM-producing B cells is especially understudied in pregnancy, a period 80 of differential immunomodulation where ZIKV infections lead to their greatest disease burden. 81

Early in pregnancy B cells are stimulated to produce IL-10, and subsequently B cell lymphopoiesis 82

is suppressed, which promotes survival of mature B cells and decreases circulating naive B cell 83 frequency (Christiansen et al., 1976; Lima et al., 2016; Nguyen et al., 2013) . Retention of the IgM 84 isotype diminishes as B cells differentiate from naïve to memory and antibody-secreting cells, 85

which shapes B cell clonal selection in response to infection and long-lasting protective immunity 86 (King et al., 2021) . These factors may impact the magnitude and quality of ZIKV-specific IgM and 87

IgG immunity in pregnancy. 88

In this study, we sought to define the kinetics and role of IgM in the control of ZIKV infection during 89

pregnancy. We evaluated the contribution of plasma IgM to ZIKV neutralization in pregnant 90 women over time and demonstrated that, in pregnancy, plasma IgM contributes to ZIKV 91 neutralization primarily within the first 3 months of infection, regardless of prior exposure to other 92 flaviviruses. We probed the B cell repertoire from peripheral blood of mothers with primary and 93 secondary ZIKV infections and established 9 ZIKV-binding B lymphoblastoid cell lines (B-LCL). 94

One of them produced an IgM antibody, DH1017.IgM, in its native pentameric conformation. 95

DH1017.IgM was somatically mutated, did not cross-react with other flaviviruses, and displayed 96 ultrapotent ZIKV neutralization that depended on its isotype. Structural studies identified a mode 97 of antigen recognition on the ZIKV virion surface compatible with the concurrent engagement of 98 all ten antigen-binding sites, a solution that is not available to antibodies of the IgG isotype, which 99

have only two antigen-binding sites. The ultrapotent activity and protection mediated by 100 DH1017.IgM in mice suggest that DH1017.IgM is a candidate for anti-ZIKV immunoprophylaxis. To evaluate the contribution of plasma IgM to ZIKV neutralization during pregnancy, we measured 119

total IgM and IgG concentrations, ZIKV binding IgG, and ZIKV neutralization of paired samples 120 after mock and IgM depletion (Table S1 ). Subject P73 displayed the highest IgM-mediated ZIKV 121 plasma neutralization, which started modestly at 8 DPS (7%), then rapidly peaked to 78% at 14 122 DPS ranging from 3% to 52% up to 112 DPS following a multimodal distribution and waned 123 thereafter to undetectable levels. (Figure 1 ). (3.8% for the heavy chain and 3.4% for the light chain) with 7 nucleotide mutations in the VH and 169 9 in the VL. All but two of these nucleotide changes were substitution mutations. Nine of the 16 170 nucleotide substitutions (56%) occurred in canonical activation-induced cytidine deaminase (AID) 171 hotspot motifs with high mutability rates (Yaari et al., 2013) . The remaining 7 substitutions 172 occurred in neutral or cold-spot motifs (Table S2) . 173

MAbs were produced and purified from the 9 B-LCLs for functional characterization. In a native 174 gel, mAb DH1017.IgM yielded two bands at ~970 kDa and ~1048 KDa (Figure 2A) , which is 175 compatible with pentameric and hexameric IgM isoforms (Keyt et al., 2020; Wiersma et al., 1998) .

Negative stain electron microscopy class average analysis supported the presence of both 177 pentameric and hexameric isoforms, with hexamers representing 18% and pentamers 178

representing 73% of all observed images (Figure 2A) . 179

All purified mAbs confirmed binding to ZIKV. While 126-1-D11.IgG bound weakly even at high 180 concentrations (LogAUC = 2.0), all other mAbs bound with LogAUC ranging from 2.7 to 5.9, with 181 the strongest ZIKV-binding mAb being DH1017.IgM ( Figure 2B ). Since plasma antibody 182 displayed substantial cross-reactivity with DENV in all subjects except primary infection subject 183 P54 ( Figure S2A ), we determined cross-reactivity of the nine mAbs with DENV 1-4 strains. Based 184 on their binding profile, the mAbs segregated into two clusters ( Figure 2C ). Cluster I comprised 185 mAbs 123-3-G2.IgG, 124-4-C8.IgG, 124-1-C2.IgG, and 124-2-G3.IgG that bound to one or more 186 DENV 1-4 strains better than to ZIKV. These mAbs were isolated from memory B cells of mothers 187

with secondary ZIKV-infection (P73 and P56), which parallels the plasma cross-reactivity profile 188

and further implies re-engagement of pre-existing DENV-reactive memory B cells. Conversely, 189

cluster II mAbs bound more strongly to ZIKV than to DENV, for which cross-reactivity was either 190

at or below limit of detection, which suggests that these mAbs may constitute a de novo immune 191 response to ZIKV. Notably, DH1017.IgM did not cross-react with or neutralize any of the four 192 DENV serotypes ( Figure 2C and Figure S2B ). 193 DH1017.IgG (EC50 = 22 pM) ( Figure 3C) . Remarkably, the functional difference between 228 DH1017.IgM, DH1017.IgG and DH1017.Fab was even greater for ZIKV neutralization.

DH1017.IgM neutralization potency (FRNT50 = 9 pM) was >40-fold higher than DH1017.IgG 230 (FRNT50 = 366 pM; Figure 3D ), whereas DH1017.Fab yielded a shallow neutralization curve 231 (FRNT50 = 93 nM) with 10,000-fold worse potency than DH1017.IgM ( Figure 3D) . These data 232 demonstrate that DH1017.IgM increased binding and ultrapotent neutralization depended on the 233

IgM isotype. 234

Since antibody-mediated complement deposition can reduce the amount of mAb required to 235 neutralize virus, we tested the effect of complement from normal human serum (NHS) on 236 DH1017.IgG and DH1017.IgM neutralizing activities. Exogenous human complement from NHS 237 enhanced ZIKV neutralization potency of both mAbs in a dose-dependent manner ( Figure 3E ).

At all doses of NHS tested (5-25% v/v), DH1017.IgM retained more potent neutralizing activity 239 compared to DH1017.IgG. The largest improvement in neutralization potency was observed in 240 the 25% NHS condition with 4.3-fold and 3.7-fold increase in potency for DH1017.IgM and 241

DH1017.IgG, respectively ( Figure 3E ). These data demonstrate that neutralization potency of the 242 DH1017 mAb in its IgG and IgM isoforms is enhanced in presence of complement. whereas DH1017.IgM did not (Figures 3F, G) . DH1017.IgM also did not mediate ADE in THP-248

1.2S cells, a subclone of THP-1 cells with increased sensitivity to ADE (Chan et al., 2014) ( Figure  249 3H). Similarly, DH1017.IgM did not mediate ADE in primary monocytes ( Figure S3A) . Finally, to 250 further characterize the safety profile of DH1017.IgM, DH1017.IgG and DH1017.Fab, we 251 measured binding to nine autoantigens associated with autoimmune diseases. All isotypes tested 252 negative for all antigens ( Figure S3B) . Further, they did not demonstrate intracellular 253 immunofluorescent staining of Hep-2 cells ( Figure S3C) . 254 255 Structural characterization of the epitope of the DH1017 clone on the Zika virion 256

Using cryo-electron microscopy (cryo-EM) and single particle reconstruction, we identified a set 257

of 4104 ZIKV H/PF/2013 particles bound with DH1017.Fab and generated a cryo-EM density 258 map. A surface shaded view of the virus-Fab complex is shown in Figure 4A and demonstrates 259 the sites of Fab binding to the surface of the particle. The cryo-EM density map resolution is 5.3 260 Å as determined from the Fourier Shell Correlation co-efficient 0.143 ( Figure S4A ).The Cα 261

backbone of the E glycoprotein ectodomains from the ZIKV asymmetric unit (PDB 6CO8; Sevvana 262 et al., 2018) is composed of three E monomers, chains A, C, and E ( Figure 4B ). Their position 263 in the density map is shown in Figure S4B . In this asymmetric unit, Chain A lies alongside the 264 antiparallel homodimer formed by chains C and E ( Figure 4B ). The fitted asymmetric unit is 265

repeated 60 times within the density map demonstrating icosahedral symmetry like previous Zika The Cα backbone of the homology model of the DH1017.Fab was fit into the cryo-EM density map 268

( Figure S4B ). It shows that the DH1017 variable domain interacts primarily with DII of all three E 269 monomers in the asymmetric unit, and at the interface of DII and DI on chains A and C ( Figure  270 4B, S4B). The DH1017.Fab epitope footprint was defined by residues of the E ectodomain 271

glycoprotein on the surface of the virus within 6 Å from the Cα backbone of the DH1017 variable 272 domain structure (Table S3, Figure 4B , C, D). The DH1017.Fab paratope footprint was defined 273

by Fab residues within 6 Å from the Cα backbone of the E ectodomain glycoprotein of the virus 274

surface and included all three heavy chain CDRs and FRH3 (Table S3) . 275

There are two 2-fold axes of symmetry on the virus surface: the icosahedral 2-fold (i2f) axis 276 delineates symmetry at the juncture of two antiparallel asymmetric units on the repetitive virion 277 surface between A and A', whereas the quasi 2-fold (q2f) axis describes symmetry within the 278 antiparallel E dimer between chains C and E ( Figure 4C ). Each asymmetric unit contains two 279 epitope footprints, at the i2f and q2f symmetry axes, respectively ( Figure 4C ). At the i2f axis one 280 half of the epitope footprint is located on chain A' and the other half of on chain A ( Figure 4C ), 281

whereas at the q2f axis, most epitope residues are on chain C, with some on chain E. ( Figure  282 4C, S4B). At the i2f axis, one bound Fab will exclude the other i2f site related by two-fold 283 symmetry from being bound. Whereas, at the q2f axis, the position on chain C and E is fully 284 occupied. The result is an occupancy of 1.5 Fab per asymmetric unit. The flexibility at Cμ2 domain suggests it is possible the arms of the DH1017.IgM pentamer can 300

bend toward the surface of the virus and each arm can contact the epitopes at q2f and i2f of 301 neighboring asymmetric units ( Figure 5A ). Such arrangement of the epitopes on the virion 302 surface may allow Fab pairs of a pentamer arm to contact epitope pairs on the virus surface either 303

at an i2f and q2f axes between neighboring asymmetric units or at the i2f and q2f axes within the 304 same asymmetric unit simultaneously, allowing DH1017.IgM a decavalent mode of epitope 305 recognition. 306

In the umbrella-like conformation, a theoretical maximum of three pentameric IgM can bind 307

concurrently to the same virion, occupying 30 of the 90 epitopes present on the virion surface 308

( Figure 5B ). In this bent conformation, the IgM pentamer can contact five epitope pairs compared 309

to a single epitope pair for the recombinant DH1017.IgG due to the decavalent versus bivalent 310 mode of antigen recognition. Alternatively, the less bent and planar pentamer conformations may 311 bind one or more pairs of epitopes on one virus particle and bind one or more pairs of epitopes 312 on a second virus particle ( Figure 5C ). This mode of binding effectively cross-links multiple virus 313 particles into aggregates. Simultaneous engagement of five E dimers at the 5-fold axis of viral 314 symmetry demonstrates a novel mechanism of IgM-mediated neutralization of ZIKV that is not 315 available to an IgG molecule and may contribute to the dramatically enhanced IgM potency 316 compared to the IgG. 317 318

Finally, we sought to evaluate whether DH1017.IgM could protect against ZIKV infection in mice. 320

We administered 100μg of DH1017.IgM or a non-ZIKV binding human IgM to Ifnar1 -/mice 1 day 321

prior and 1 day following infection with 1000 FFU of ZIKV H/PF/2013 and measured viremia and 322 lethality. We found that DH1017.IgM conferred protection from lethal challenge in all mice and 323 controlled viremia to the limit of detection (3.2-3.6 Log10 viral copies/mL) in comparison to mice 324 receiving control IgM (6.3-6.9 Log10 viral copies/mL). In contrast, all mice in the control IgM group 325 succumbed to infection ( Figure 6A , B). Human IgM was maintained in vivo at detectable levels 326

up to 4 days post challenge in both groups, or 3 days after the last administration ( Figure 6C ). 327

Thus, administration of ultrapotent DH1017.IgM protects against ZIKV disease in mice. 328 329

In this study, we evaluated the contribution of plasma IgM responses to ZIKV neutralization in 331 pregnancy, defined the frequency of ZIKV-specific B cells isolated from peripheral blood, and 332 isolated a ZIKV-specific IgM monoclonal antibody, DH1017.IgM, in its native multimeric form.

Unlike previous plasma-depleted and whole plasma investigation of IgM, the isolation of a human 334

IgM mAb in its native multimeric conformation enabled direct characterization of the impact of the 335

IgM isotype on antibody function against ZIKV. DH1017.IgM did not cross-react with DENV 1-4 336 and mediated ultrapotent ZIKV neutralization. DH1017.IgM ultrapotency depended on its isotype, 337

as it neutralized >40-fold more potently than when expressed as an IgG. We mapped the footprint 338 of its discontinuous epitope within the asymmetric unit and predicted a mode of antigen 339 recognition compatible with the concurrent engagement of all the ten antigen-binding sites present 340 on the IgM pentamer, a solution not available to smaller IgG molecules with only two antigen- such temporal association in the setting of pregnancy. 360

We note that subject P73, from whom we isolated DH1017.IgM at day 71 DPS, developed a 361 prolonged viremia lasting 42 days despite an early robust peak in plasma IgM neutralization. Since the IgM half-life is considerably shorter than IgG, antibody engineering to prolong the half-441 life would be likely required for effective prophylactic countermeasures. However, the potential 442

issue of short half-life would be less relevant for a therapeutic intervention administered at the 443 time of diagnosis, which will be aimed at rapid clearance of viremia and reduction in the time of 444 fetal exposure to circulating virus. Further, our findings support the development and investigation 445

of engineered multimeric antibody formulations as a prophylactic and therapeutic strategy. 446

In summary, we demonstrated a large contribution of plasma IgM in ZIKV neutralization in both 447 primary and DENV pre-exposed ZIKV infections in pregnancy. We isolated a novel ultrapotent 448

ZIKV-neutralizing IgM mAb that protected mice from lethal challenge and demonstrated the 449 impact of isotype on its antiviral function. We defined a conceptual framework by which the spatial 450 arrangement of quaternary epitopes on the virion can modulate neutralization in an isotype- NaCl, and 0.02% ruthenium red. A 5 µl drop of diluted sample was applied to a glow-discharged 613 carbon-coated grid for 8-10 seconds, blotted, then rinsed with two drops of buffer containing 1 614 mM HEPES, pH 7.4 and 7.5 mM NaCl, and finally stained with one drop of 2% uranyl formate for 615 60 s, then blotted and air dried. Images were obtained with a Philips EM420 electron microscope 616 at 120 kV, 82,000x magnification, and captured with a 2k x 2k CCD camera at 4.02 Å/pixel. The 617

RELION program (Scheres, 2016) was used for particle picking and 2D class averaging. 618 ZIKV and DENV virion capture ELISA. The virion capture ELISA methods were previously 619 described (Singh et al., 2019) . Briefly, high-binding 96-well plates (Greiner) were coated with 40 620 ng/well of 4G2 antibody (clone D1-4G2-4-15) in 0.1 M carbonate buffer, pH 9.6 overnight at 4°C. 621

Plates were blocked in Tris-buffered saline containing 0.05% Tween-20 and 5% normal goat 622 serum for 1 hour at 37°C, followed by an incubation with either ZIKV ( negative control condition was sample diluent alone. Respiratory syncytial virus specific IgG 639

Palivizumab was used as a negative control for testing ZIKV-binding of purified mAbs, and diluent 640 served as negative control for IgM purified mAb assays. For purified mAbs that were serially 641 diluted, the magnitude of virion binding was evaluated as an ED50, which was calculated with a 642 sigmoidal dose-response (variable slope) curve in Prism 7 (GraphPad) using a least squares fit. 643

The ED50 value for serially diluted mAbs was considered valid if the OD450 at 100ug/mL was 5-644

fold higher than the no sample condition. at 20ug/mL with 2-fold dilutions to 12-spots were added to the plate for an hour at 37°C. This was 654 followed by 0.5ug/mL anti-his HRP (Sigma), and binding was detected after incubation with 655

substrate at an absorbance of 450nm. EC50 was obtained by a sigmoidal dose-response (variable 656 slope) curve in Prism 7 (GraphPad) using least squares fit. The negative control was no antigen 657

and positive control were previously reported mAbs, including 1M7, ZV-2, ZV-64 and ZKA190.

1M7 mAb (Smith et al., 2013) was produced from a hybridoma. ZKA190 was generated 659 recombinantly using the sequence for PDB entry 5Y0A ( dilutions. MAbs were tested at 5ug/mL or 10ug/mL with 5-fold dilution series. However, 668

DH1017.Fab, was tested at 1mg/mL with a 5-fold dilution series. Negative control was media 669 alone, and positive controls were known ZIKV-neutralizing mAbs and plasma from ZIKV-infected 670

subjects. Virus and plasma or mAb samples were co-incubated for 1 hour at 37°C, then 671 transferred to a 96-well plate (Greiner Bio One) with confluent Vero-81 cells and incubated for 1 672 hour at 37°C. Plates were overlayed with 1% methylcellulose and incubated at 37 o C for 40-42 673

hours (ZIKV and DENV4), 51-53 hours (DENV1), or 48 hours (DENV2 and DENV3). Cells were 674 fixed with 2% paraformaldehyde for 30 minutes and stained with 0.5 μg/mL of 4G2 or E60 mouse 675 monoclonal antibody. Foci were detected with an anti-mouse IgG conjugated to horseradish 676 peroxidase at a 1:5000 dilution (Sigma), followed by True Blue substrate (KPL). Foci were 677 counted using the CTL ImmunoSpot plate reader (Cellular Technology Limited). FRNT50 values 678

were calculated with the sigmoidal dose-response (variable slope) curve in Prism 8.3.0 679

(GraphPad), constraining values between 0 and 100% relative infection. Percent relative infection 680 curves were considered to pass quality control criteria for FRNT50 determination if R 2 > 0.65, 681

absolute value of hill slope >0.5, and curve crossed 50% relative infection within the range of the 682 plasma dilutions in the assay. Samples were repeated up to 3 times to quantify a valid FRNT50 in 683

accordance with the quality control criteria. Briefly, the WNV replicon and C-prM-E plasmids were co-transfected into HEK-293T cells using 689

Lipofectamine 3000 TM transfection reagent (Invitrogen). Transfected cells were incubated at 30°C 690 and RVP-containing supernatants were harvested 3-6 days post-transfection and pooled. RVP-691

containing supernatants were passed through a 0.2 µm filter (Millipore) and stored at -80°C until 692

use. sample was split into 2 aliquots, with one portion depleted of IgM isotype antibodies and another 726 portion mock depleted. First, each sample was heat inactivated for 30 minutes at 56ºC, diluted 727 1:1 with sterile PBS, and centrifuged at 10,000 x G for 10 minutes to remove debris. Depletion 728 beads were packed into sterile 0.5mL centrifugal filter devices with a 0.22µm pore PVDF 729 membrane (Millipore) and equilibrated with 3x sterile PBS washes (pH 7.2). 200mg of POROS TM 730

CaptureSelect TM IgM affinity beads (ThermoFisher Scientific) were used for IgM depletion, and 731

66mg of corresponding beads of the same size (200-400 mesh) and material (polystyrene 732 divinylbenzene 1% cross linked beads; Alfa Aesar) were used for mock depletion. Samples were 733 co-incubated with beads for 10 minutes at room temperature with gentle inversions, and then the 734 depleted fraction was centrifuged at 10,000 x g for 10 minutes. 735

Depletion of IgM was confirmed by total IgM ELISA, and non-specific losses to ZIKV-binding IgG 736

were quantified through virion binding ELISA; both assays are described in other sections of this 737

Method supplement. Limit of detection (LOD) of 0.12 μg/mL was based on the linear range of a 738 sigmoidal standard curve of human IgM (Jackson ImmunoResearch Laboratories). Magnitude of 739 ZIKV-binding IgG was assessed by virion-binding ELISA and neutralization potency was 740 assessed using the Focus Reduction Neutralization Test. Due to slight differences in ZIKV-741

binding IgG across IgM and mock depleted fractions, each fraction was adjusted to the magnitude 742 of ZIKV-binding IgG in the same sample such that differences in neutralization activity could be 743 attributed to differences in IgM isotype antibodies. Thus, the percent neutralization attributable to 744

IgM that is reported in this study was calculated as follows: 745 serum albumin and 0.05% Tween-20. Monoclonal antibodies were tested using 3-fold serial 887 dilutions starting at 10μg/ml. 10μl of primary antibodies were added to each well and incubated 888

for 1 hour at room temperature. The following positive control antibodies from Immunovision were 889 all tested at a 1:25 starting dilution with 3-fold serial dilutions: Anti-Centromere B (HCT-0100), 890

Anti-single stranded DNA (HSS-0100), Anti-Histone (HIS-0100), Anti-Jo 1 (HJO-0100), Anti-SRC 891

(HRN-0100), Anti-Scl 70 (HSC-0100), Anti-Sm (HSM-0100), Anti-SSA (HSA-0100), and Anti-SSB 892

(HSB-0100). The negative control was assay diluent alone. Plates were developed using 15μl/well 893 of combination of horseradish peroxidase-conjugated antibodies in assay diluent comprising) 894

goat anti-human IgG (Jackson ImmunoResearch Laboratories,109-035-098) at 1:10,000 dilution; 895

2) goat anti-human IgM (Jackson ImmunoResearch Laboratories,109-035-129) at 1:10,000 896 dilution; 3) goat anti-human H+L (Promega, W403B) at 1:3,000 dilution. to HEp-2 cells was performed as previously described (Bonsignori et al., 2014) . IgG1 mAbs 2F5 902 and 17B were used as positive and negative controls, respectively. All mAbs were tested at 903 25ug/ml and 50ug/mL. Images were acquired for 8 seconds with a 40x objective. 904 Table 1 . Immunogenetics of monoclonal antibodies isolated from B-LCLs. See also Figure S1 and . DH1017 clone interacts with envelope dimer. A. Surface-shaded view of the Zika virion bound with the DH1017 Fab fragment at a resolution of 5.3 Å acquired via cryo-EM. The color represents the distance to the center and shown by the colored scale bar in Å. The black triangle represents the asymmetric unit, a pentagon represents the five-fold axis, the triangle the three-fold axis, and the oval the two-fold axis. B. Surface representation of the E ectodomain of Zika virus asymmetric unit (PDB 6CO8) shown in top view unbound, then bound with the variable domain of DH1017.Fab (cyan) and the constant domain (green), and a side view of the asymmetric ectodomain bound the DH1017 Fab fragment (left to right). The E ectodomains are colored with DI in red, DII in yellow, and DIII in blue. The residues of the footprint are colored, magenta. See also Fig. S4 and Table S3 . C. A radially colored roadmap (scale bar in Å). The icosahedral and quasi two-fold axes are labelled i2f and q2f, respectively. The monomer chains of two E dimers are labeled A' and A at the i2f, and chains C and E at the q2f. Residues on the surface of the virus within 6 Å of the variable domain structure fit to the density map are colored yellow and magenta. Yellow residues are on chain C at the q2f axis and magenta residues are at the i2f axis on chain A' and A. D. Fab DH1017 epitope shown on the primary sequence of E ectodomain. The epitope residues are colored magenta and the domains DI, DII, and DIII are indicated by line color red, yellow and blue, respectively. A B C Figure 5 : Models of IgM pentamer bound to Zika virus particle. A. A top down and side view of a model of the pentamer binding the Zika virus particle in an umbrella like conformation. The density map is shown in gray. The Fc domain is shown in yellow, domains Cμ3 and Cμ4 are labeled. The Cμ2 domain is labeled and colored purple. The DH1017 Fab structure is fit to the density map, at the i2f it is cyan and at the q2f, green. B. The density map of the Fab bound virus particle in an icosahedral cage with top-down views at each icosahedral axis, i2f, i5f, and i3f. The green asterisk indicates a position where the pentamer is bound as in A. The red X indicates the position of five-fold axes related to the green asterisk by two-fold symmetry that do not have the i2f position available for Fab binding to form the umbrella like conformation. The i3f view shows positions of complete occupancy for the particle bound as in A. C. Schematic representation of virus particles crosslinked by pentamers. The Fc portion of the molecule is shown in red crosslinking virus. A. Viral load in serum was assessed by qRT-PCR and limit of detection was 1000 copies/mL. B. Survival curves for each IgM intervention group. C. Total human IgM concentrations were measured in mouse sera by ELISA over days post challenge. Limit of detection for this assay was 0.08-0.03 μg/mL across assays.

Protective efficacy of multiple vaccine 907 platforms against Zika virus challenge in rhesus monkeys

Analysis of a Clonal Lineage of HIV-1 Envelope 910

V2/V3 Conformational Epitope-Specific Broadly Neutralizing Antibodies and Their Inferred 911 Unmutated Common Ancestors

An autoreactive antibody from an SLE/HIV-1 914 individual broadly neutralizes HIV-1

Maturation Pathway from Germline to Broad HIV-917 1 Neutralizer of a CD4-Mimic Antibody

Inference of the HIV-1 VRC01 Antibody Lineage 920 Unmutated Common Ancestor Reveals Alternative Pathways to Overcome a Key Glycan 921 Barrier

The Specificity of the Persistent IgM

Neutralizing Antibody Response in Zika Viral Infections Among Individuals with Prior Dengue 925 Virus Exposure

Zika might not be acting alone: Using an ecological study approach to 928 investigate potential co-acting risk factors for an unusual pattern of microcephaly in Brazil

Envelope Protein Glycosylation Mediates 931 Zika Virus Pathogenesis

Antibody-Dependent Dengue Virus Entry Modulates Cell 934 Intrinsic Responses for Enhanced Infection

Ligation of Fc gamma receptor IIB inhibits antibody-937 dependent enhancement of dengue virus infection

Leukocyte immunoglobulin-like receptor B1 is 940 critical for antibody-dependent dengue

Fc receptors and their influence on 942 efficacy of therapeutic antibodies for treatment of viral diseases

The neutralizing role of IgM 948 during early Chikungunya virus infection

Human antibody response to Zika 951 targets type-specific quaternary structure epitopes

Coronavirus disease 2019 vaccines in 953 pregnancy

A 955 Critical Role for Induced IgM in the Protection against West Nile Virus Infection

Major role of IgM in the neutralizing activity of 959 convalescent plasma against SARS-CoV-2

Detection of anti-yellow fever virus immunoglobulin 962 M antibodies at 3-4 years following yellow fever vaccination

Integrated pipeline for the accelerated discovery 966 of antiviral antibody therapeutics

Protected to 968 death: Systematic exclusion of pregnant women from Ebola virus disease trials

CisTEM, user-friendly software for single-particle 971 image processing

Zika virus IGM 25 months after symptom 974 onset

Single particle cryo-electron microscopy and 3-D reconstruction 976 of viruses

Pathogenesis of dengue: Challenges to molecular biology. Science (80-. 978 )

Antibody-enhanced dengue virus infection in primate 980 leukocytes

A human antibody against Zika virus crosslinks 983 the E protein to prevent infection

A human antibody against Zika virus crosslinks 986 the E protein to prevent infection

Epstein-Barr Virus 989 Infection of Naïve B Cells In Vitro Frequently Selects Clones with Mutated Immunoglobulin 990 Genotypes: Implications for Virus Biology

Ten-year follow-up of a prospective trial for the targeted therapy of gastric cancer with 993 the human monoclonal antibody PAT-SC1

Report on a WHO 995 consultation on immunological endpoints for evaluation of new Japanese encephalitis vaccines, 996 WHO

Immune correlates of protection for 999 dengue: State of the art and research Agenda

Therapeutic Use of IgM Antibodies. Antibodies 9

Single-cell analysis of human B cell maturation predicts how antibody class switching 1004 shapes selection dynamics

Antibodies protect mice 1006 against challenge with tick-borne encephalitis virus (TBEV)-infected macrophages

A systematic approach to the development 1010 of a safe live attenuated Zika vaccine

Vaccine protection 1013 against Zika virus from Brazil

A Mouse Model of Zika Virus Pathogenesis

Times to key events in Zika virus 1018 infection and implications for blood donation: A systematic review

High-throughput isolation of immunoglobulin 1022 genes from single human B cells and expression as monoclonal antibodies

Characterization of B cells in healthy pregnant women from late pregnancy to post-1026 partum: A prospective observational study

Antibody pharmacokinetics and 1028 pharmacodynamics

Structural basis of a potent human monoclonal antibody 1031 against Zika virus targeting a quaternary epitope

Distinct neutralizing 1035 antibody correlates of protection among related Zika virus vaccines identify a role for antibody 1036 quality

Impact of flavivirus vaccine-induced 1039 immunity on primary Zika virus antibody response in humans

Yellow fever vaccine: direct 1042 challenge of monkeys given graded doses of 17D vaccine

Complement Protein C1q Reduces the Stoichiometric Threshold for 1045

Antibody-Mediated Neutralization of West Nile Virus

Oligomeric state of the ZIKV E protein defines protective immune 1048 responses

Neutralizing antibody responses in the major immunoglobulin classes to 1050 Yellow Fever 17D vaccination of humans

Zika Virus Vaccine Development

Epstein-Barr virus reprograms human B lymphocytes immediately 1055 in the prelatent phase of infection

Enhancing dengue virus maturation using a stable furin over-expressing cell line

High-resolution structures of the IgM Fc domains reveal principles of its hexamer 1061 formation

To B or not to B cells-mediate a healthy 1063 start to life

Delayed childhood 1066 neurodevelopment and neurosensory alterations in the second year of life in a prospective 1067 cohort of ZIKV-exposed children

Persistence of Zika Virus 1070 in Body Fluids -Final Report

Solution structure of human 1072 and mouse immunoglobulin M by synchrotron X-ray scattering and molecular graphics 1073 modelling. A possible mechanism for complement activation

UCSF Chimera -A visualization system for exploratory research and 1076 analysis

Development of Envelope Protein 1079

Antigens To Serologically Differentiate Zika Virus Infection from Dengue Virus Infection

Crystal structure of a 1083 glycosylated Fab from an IgM cryoglobulin with properties of a natural proteolytic antibody

GRP78-directed immunotherapy in 1087 relapsed or refractory multiple myeloma -results from a phase 1 trial with the monoclonal 1088 immunoglobulin M antibody PAT-SM6

Differential human antibody repertoires following Zika infection and the 1092 implications for serodiagnostics and disease outcome

Vital Signs: Update on Zika 1095

Virus-Associated Birth Defects and Evaluation of All U.S. Infants with Congenital Zika Virus 1096 Exposure -U.S. Zika Pregnancy Registry

Vaccine Mediated Protection Against Zika 1100 Virus-Induced Congenital Disease

Recurrent Potent Human 1103 Neutralizing Antibodies to Zika Virus in Brazil and Mexico

Zika virus activates de novo and cross-reactive 1106 memory B cell responses in dengue-experienced donors

CTFFIND4: Fast and accurate defocus estimation from 1108 electron micrographs

A combination of two human 1111 monoclonal antibodies limits fetal damage by Zika virus in macaques

Optimal determination of particle orientation, 1114 absolute hand, and contrast loss in single-particle electron cryomicroscopy

Recognition 1118 determinants of broadly neutralizing human antibodies against dengue viruses

I-TASSER: A unified platform for automated 1121 protein structure and function prediction

Neutralizing human antibodies prevent Zika virus 1124 replication and fetal disease in mice

Fast maximum-likelihood refinement of 1127 electron microscopy images

Refinement and Analysis of the Mature Zika Virus Cryo-EM Structure at 3.1 Å 1130 Resolution

Insights 1132 into IgM-mediated complement activation based on in situ structures of IgM-C1-C4b. Proc. Natl

A maximum-likelihood approach to single-particle image refinement

Classical detection theory and the cryo-EM particle selection problem

Efficient transplacental IgG transfer 1140 in women infected with Zika virus during pregnancy

Flavivirus cell entry and 1142 membrane fusion

The potent and broadly neutralizing human dengue 1145 virus-specific monoclonal antibody 1C19 reveals a unique cross-reactive epitope on the bc loop 1146 of domain II of the envelope protein

Isolation and characterization of broad 1149 and ultrapotent human monoclonal antibodies with therapeutic activity against chikungunya 1150 virus

Specificity, cross-reactivity, and function of 1153 antibodies elicited by Zika virus infection

Zika virus RNA and IgM 1156 persistence in blood compartments and body fluids: a prospective observational study

Automated molecular microscopy: The new Leginon system

Dengue virus envelope 1163 dimer epitope monoclonal antibodies isolated from dengue patients are protective against zika 1164 virus

Mechanisms of Neutralization of Influenza Virus by 1166 IgM

Evolution of the innate 1169 and adaptive immune response in women with acute Zika virus infection

A Human Bi-specific Antibody against Zika Virus 1173 with High Therapeutic Potential

Longitudinal dynamics of the 1176 human B cell response to the yellow fever 17D vaccine

Structural and functional 1179 analysis of J chain-deficient IgM

Interpretation of electron density with stereographic 1181 roadmap projections

Models of Somatic 1184

Hypermutation Targeting and Substitution Based on Synonymous Mutations from High-1185

The I-TASSER suite: 1187 Protein structure and function prediction

I-TASSER server for protein 3D structure prediction

A simple method for Alexa Fluor 1190 dye labelling of dengue virus

Structure of Acidic pH Dengue Virus Showing the Fusogenic 1193

Structural Basis of Zika Virus-Specific 1196

Anisotropic correction of beam-induced motion for improved cryo-electron 1199 microscopy

Data were collected on a Titan Krios (Thermofisher) microscope equipped with a Gatan K3 818 detector using and Leginon software package (Suloway et al., 2005) . A total of 1,929 cryo EM 819 micrographs were collected with a nominal magnification of 64000x, 0.66 Å/pixel size, and an 820 electron dose equivalent, of 35 e-/Å2. Motion correction and CTF calculations estimations were 821 performed using MotionCorr2 (Zheng et al., 2017) and CTFFIND4 (Rohou and Grigorieff, 2015) 822 respectively. 823Automated particle selection picking (Sigworth, 2004) particles were selected for further processing. 827Single particle reconstruction was performed according to the 'gold standard' method using jspr 828 (Guo and Jiang, 2014) . Briefly, the particles were divided equally into two randomly selected 829 independent particle sets. Twenty ab-initio models were generated from a random set of 700 830particles selected from the set of 4,104 particles. Two ab-initio models were selected, and one 831 ab-initio model was assigned to one independent particle set and the other ab-initio model was 832assigned to the other independent data set. Each dataset was refined iteratively assuming 833icosahedral symmetry. Refinement resulted in two independent models that converged on the 834 same structure. Following corrections for astigmatism, elliptical distortion, defocus, and the 835 masking of the disordered nucleocapsid core, the final models of each independent data set were 836combined into a single final model. The resolution of the map was calculated at 0.143 from the 837 FSC curve (Rosenthal and Henderson, 2003) . PDB 2CRJ were fit to the density map after being aligned manually to the fitted DH1017.Fab 858 structure. 859