Characterization of two engineered dimeric Zika virus envelope proteins as immunogens for neutralizing antibody selection and vaccine design

The envelope protein of Zika virus (ZIKV) exists as a dimer on the mature viral surface and is an attractive antiviral target because it mediates viral entry. However, recombinant soluble wild-type ZIKV envelope (wtZE) might preferentially exist as monomer (monZE). Recently, it has been shown that the A264C substitution could promote formation of dimeric ZIKV envelope protein (ZEA264C), requiring further characterization of purified ZEA264C for its potential applications in vaccine development. We also noted that ZEA264C, connected by disulfide bond, might be different from the noncovalent native envelope dimer on the virion surface. Because the antibody Fc fragment exists as dimer and is widely used for fusion protein construction, here we fused wtZE to human immunoglobulin G1 (IgG1) Fc fragment (ZE-Fc) for noncovalent wtZE dimerization. Using a multistep purification procedure, we separated dimeric ZEA264C and ZE-Fc, revealing that they both exhibit typical β-sheet–rich secondary structures and stabilities similar to those of monZE. The binding activities of monZE, ZEA264C, and ZE-Fc to neutralizing antibodies targeting different epitopes indicated that ZEA264C and ZE-Fc could better mimic the native dimeric status, especially in terms of the formation of tertiary and quaternary epitopes. Both ZEA264C and ZE-Fc recognize a ZIKV-sensitive cell line as does monZE, indicating that the two constructs are still functional. Furthermore, a murine immunization assay disclose that ZEA264C and ZE-Fc elicit more neutralizing antibody responses than monZE does. These results suggest that the two immunogen candidates ZEA264C and ZE-Fc have potential utility for neutralizing antibody selection and vaccine design against ZIKV.

The envelope protein of Zika virus (ZIKV) exists as a dimer on the mature viral surface and is an attractive antiviral target because it mediates viral entry. However, recombinant soluble wild-type ZIKV envelope (wtZE) might preferentially exist as monomer (monZE). Recently, it has been shown that the A264C substitution could promote formation of dimeric ZIKV envelope protein (ZE A264C ), requiring further characterization of purified ZE A264C for its potential applications in vaccine development. We also noted that ZE A264C , connected by disulfide bond, might be different from the noncovalent native envelope dimer on the virion surface. Because the antibody Fc fragment exists as dimer and is widely used for fusion protein construction, here we fused wtZE to human immunoglobulin G1 (IgG1) Fc fragment (ZE-Fc) for noncovalent wtZE dimerization. Using a multistep purification procedure, we separated dimeric ZE A264C and ZE-Fc, revealing that they both exhibit typical ␤-sheet-rich secondary structures and stabilities similar to those of monZE. The binding activities of monZE, ZE A264C , and ZE-Fc to neutralizing antibodies targeting different epitopes indicated that ZE A264C and ZE-Fc could better mimic the native dimeric status, especially in terms of the formation of tertiary and quaternary epitopes. Both ZE A264C and ZE-Fc recognize a ZIKV-sensitive cell line as does monZE, indicating that the two constructs are still functional. Furthermore, a murine immunization assay disclose that ZE A264C and ZE-Fc elicit more neutralizing antibody responses than monZE does. These results suggest that the two immunogen candidates ZE A264C and ZE-Fc have potential utility for neutralizing antibody selection and vaccine design against ZIKV.
Zika virus (ZIKV), 2 as a re-emerging viral pathogen, belongs to the Flaviviridae family including dengue virus, West Nile virus, Japanese encephalitis virus, yellow fever virus, and tickborne encephalitis virus (1,2). It can be transmitted by Aedes mosquitoes and cause severe neurological diseases including Guillain-Barré syndrome in the adult (3,4), and congenital Zika syndrome in the infant that includes microcephaly, brain abnormalities, and other severe birth defects (5)(6)(7). Because of the huge threat of ZIKV to the public health, it has raised worldwide attention and lots of work on the development of drugs and vaccines against ZIKV is in progress (8 -10). However, there is no approved anti-ZIKV reagents for clinical use, which needs continuous efforts.
Like other flaviviruses, the genome of ZIKV encodes a single polyprotein which can be cleaved into three structural proteins (capsid, pre-membrane, and envelope (E)) and seven nonstructural proteins (NS1, NS2a, NS2b, NS3, NS4a, NS4b, and NS5) (11)(12)(13). Among these proteins, E protein plays a very important role in viral entry (14 -16). Therefore it is an effective target for inhibition of the virus. As a typical class II viral envelope protein, E proteins on the surface of mature ZIKV particle form antiparallel homodimers in a herringbone pattern (13,17,18). As for other flavivirus E proteins, the monomeric ZIKV E protein also has three distinct domains: a central ␤-barrel-shaped domain I (DI), an extended dimerization domain II (DII), and a C-terminal immunoglobulin-like domain III (DIII). The fusion loop (FL) of E protein is located in the distal end of DII and consists of hydrophobic residues that can insert into endosomal membranes during pH-dependent conformational changes and drive fusion (19).
Various monoclonal antibodies (mAbs) with different neutralizing activities against ZIKV have been identified that could bind to different epitopes in ZIKV E DI, DII, and DIII. For example, the antibodies that bind to DIII (e.g. ZV-67 and ZKA64) have high neutralizing activity without crossreactivity and might slightly cause antibody-dependent enhancement of infection (ADE) (20,21), whereas those binding to FL epitope (FLE) (e.g. 2A10G6) are crossreactive but relatively modest in neutralizing, and tend to cause ADE (22,23). Furthermore, a panel of antibodies recognizing quaternary epitope formed by E dimer (E-dimer-dependent epitope, EDE) were identified that are more potent and crossreactive, and have less ADE (21, 24 -26). In addition, several neutralizing antibodies (e.g. Z3L1 and Z20) with strong neutralizing activity but no crossreactivity that target conformational epitope composed of residues in DI, DII, and DIII (tertiary epitope) were also identified (27). Therefore, preparation of dimeric E as its native status is a key point for development of effective immunogen for selection of powerful neutralizing antibodies and design of effective vaccines. However, expression of soluble WT ZIKV E protein (wtZE) might only lead to generation of monomeric E protein (monZE) (28,29). Recently, it has been reported that introduction of a single Cys substitution (A259C) in E protein of dengue virus could lead to formation of dimeric E through an intersubunit disulfide bond (30). By the similar strategy, a ZIKV E protein mutant by replacement of Ala-264 to one Cys in ZIKV E protein (ZE A264C ) was also designed to show that folding and dimerization of secretory ZIKV E proteins are strongly dependent on temperature (28). However, the conformational and functional information of covalent dimeric ZE A264C were still unclear and should be well-characterized for its potential use as an ideal immunogen. Because ZE A264C exists as covalent dimer connected by a disulfide bond, it also raises a question whether E dimer could form noncovalent linkage in vitro because native E dimer on virion surface is noncovalent. As a stable dimer, antibody Fc fragment has been widely used in construction of fusion protein for therapeutic purpose and vaccine design because it could make the fused protein bivalent (31). Hence, we proposed that wtZE could be noncovalently dimerized by fusing it with Fc fragment, and then also made Fc-fusion protein (ZE-Fc) as a candidate. Here, we combined different methods to structurally and functionally characterize monZE, ZE A264C , and ZE-Fc in vitro and in vivo. Our results disclose that major neutralizing epitopes including EDE were still maintained in ZE A264C and ZE-Fc. Moreover, immunization of these three proteins in mice shows that ZE A264C and ZE-Fc are more effective in eliciting antisera against ZIKV than monZE. Hence, both ZE A264C and ZE-Fc have potentials as promising immunogens for development of neutralizing antibodies including those target tertiary/ quaternary epitopes, and potent vaccines.

Design, expression, and separation of soluble ZIKV E dimer
As reported previously, the ectodomain of ZIKV E (1-408 amino acids) (monZE) was used for soluble expression (32) as monomeric form (Fig. 1A). The residue Ala-264 in ZIKV E protein was mutated to Cys to generate ZE A264C , which was desired to form dimer (28) (Fig. 1A).
According to the cryo-EM structure of ZIKV virion (PDB ID: 5IZ7) (17), there are two transmembrane helices (E-TM) at the C terminus of one E protein that interacts with transmembrane domains of M protein (M-TM) (Fig. 1B), which has also been illustrated well in other flaviviruses (33,34). The lacking of this interaction might result in the loss or significant reduction of dimerization of E protein. Therefore, we fused wtZE to antibody Fc fragment to construct a fusion protein ZE-Fc that was desired to form dimer noncovalently because of the strong interaction between two CH3 domains in Fc fragment (PDB ID: 1HZH) (35) (Fig. 1B). In addition, A264C in ZIKV E protein was also introduced in ZE-Fc (ZE A264C -Fc) to show whether the disulfide bond could enhance the dimerization of E protein when fused to Fc or not (Fig. 1B).
All four constructs (monZE, ZE A264C , ZE-Fc, and ZE A264C -Fc) were expressed in Drosophila S2 cells. After first-step purification by Strep-Tactin affinity chromatography column, the samples were subjected to size exclusion chromatography (SEC) for further separation. In the case of purified monZE, only one peak (Peak 1) was observed ( Fig. 2A) whereas three major peaks (Peak 1, Peak 2, and Peak 3) were observed in the case of purified ZE A264C (Fig. 2B). According to analysis of three peaks by SDS-PAGE, it could be concluded that dimeric ZIKV E A264C mainly formed in Peak 2, whereas Peak 1 represented unpaired monomeric ZIKV E A264C , and Peak 3 indicated bigger oligomer. Therefore, we collected Peak 1 in the case of monZE and Peak 2 in the case of ZIKV E A264C for following experiments. For Fc-fusion proteins, we also checked all the peaks by SDS-PAGE. Although all of peaks of ZE-Fc could migrate to position of correct dimer according to the marker, only main peak (Peak 1) was collected (Fig.  2C). However, in the case of ZE A264C -Fc, Peak 1 was the main peak and the migration of it on SDS-PAGE indicated the formation of soluble aggregation (Fig. 2D). Taken together, monZE from Peak 1, ZE A264C from Peak 2, and ZE-Fc from Peak 1 were used for further analysis.

Formation of dimer
After concentration, the purified monZE, ZE A264C , and ZE-Fc were run on SEC again and no additional peak was observed, which indicated that they were in stable status in solution (Fig. 3, A and B). Then all of them were confirmed on SDS-PAGE under different conditions. By comparison of the migration in nonreduction and reduction conditions, it could clearly find that ZE A264C forms dimer through disulfide bridge (Fig. 3C). Similarly, ZE-Fc could also form dimer, which was deduced from its migration in boiling and nonboiling conditions (Fig. 3D). To confirm molecular mass (M.M.) of ZE A264C that should be double mass of monZE, we also performed the MALDI-TOF MS assay. The M.M. values from the assay were 50.0 kDa for monZE and 100.1 kDa for ZE A264C , which matched the theoretical M.M. values of E monomer (48.6 kDa) and dimer (97.2 kDa) well. In conclusion, both ZE A264C and ZE-Fc form stable dimer in solution.

Secondary structure and stability
Circular dichroism (CD) spectra of these three proteins exhibited a maximum negative peak between 216 and 218 nm, illustrating that these proteins adopted rich ␤-sheet secondary structures (Fig. 4A). The highly overlapping structural profiles

Engineered dimer of ZIKV E protein
showed that introduced Cys mutation does not significantly change the overall structure (ZE A264C versus monZE). As the temperature increased, the structure was destroyed with obviously S-shaped curve (two-state) in the case of monZE and ZE A264C (Fig. 4B). The melting temperature (T m ) values of both proteins were calculated by Boltzman fitting equation (T m of monZE: 42.1°C and T m of ZE A264C : 44.5°C). Although it seemed that ZE A264C was slightly more thermostable, the dimerization might not obviously alter the stability of each domain in E protein. These results demonstrated the successful formation of dimeric ZIKV E protein by engineered disulfide bond without change of overall structure. The maximum negative peak in the case of ZE-Fc located between 216 and 218 nm as desired because Fc is also mainly composed of ␤-strands (Fig. 4A). The thermo-induced unfolding curve of ZE-Fc exhibited three sections including unfolding of E protein and CH2 and CH3 domains in Fc fragment. The unfolding process of E protein in Fc-fusion protein was quite similar to that of monZE/ZE A264C , whereas the unfolding of CH2 and CH3 domains in Fc fragment were similar to our previous result (Fig. 4B) (36). Therefore, it is reasonable to believe that fusion of wtZE with Fc has no obvious influence on secondary structure of E protein.

Recognition by neutralizing antibodies targeting nonquaternary epitopes
The ability of these three proteins to be recognized by neutralizing antibodies scFv-ZV-67 (targeting DIII) and scFv-2A10G6 (targeting FL) was firstly tested by ELISA. The EC 50 values of binding of scFv-ZV-67 to ZE A264C and ZE-Fc were 145 and 68 nM, respectively, whereas the value was 405 nM in the cases of monZE (Fig. 5A). In contrast, scFv-2A10G6 bound to ZE A264C and ZE-Fc with EC 50 values of 170 and 114 nM, whereas the value decreased to 65 nM in the cases of wtZE (Fig.  5B). In comparison of these two sets of results, the bindings of 2A10G6 to ZE A264C and ZE-Fc were relatively reduced, possibly because of burying of the fusion loop region after dimerization. These observations provide the possibility that dimeric E protein would elicit fewer antibodies targeting FLE as immunogen than does monomeric E, and therefore reduced ADE could be desired (30).

Engineered dimer of ZIKV E protein Recognition by neutralizing antibodies targeting tertiary/quaternary epitopes
We tested the binding of IgG-Z20, IgG-Z3L1 (targeting tertiary epitopes), IgG-B7, IgG-C10, and IgG-A11 (targeting quaternary epitopes) to coated monZE, ZE A264C , and ZE-Fc. However, no binding signal was observed because the conformational epitopes might be affected (e.g. hidden or destroyed) after coating on ELISA plate (Fig. 6A). So we performed capture ELISA by coating antibodies on the plate. In general, binding of most of these antibodies (e.g. IgG-B7 and IgG-C10) to ZE A264C and ZE-Fc were stronger than that to monZE (Fig. 6B). The IgG-Z3L1 and IgG-Z20 targeting tertiary epitope recognized monZE very weakly, whereas they could strongly bind to ZE A264C and ZE-Fc. It indicates recognition of both quaternary and tertiary epitopes is highly dependent on the formation of E dimer. We also noticed that IgG-A11 does not show strong binding to monZE, ZE A264C , and ZE-Fc. In previous study, the formation of complex of E protein and Fab format IgG-A11 in vitro was also difficult (25). Probably this antibody somehow prefers to recognize epitope relying on membrane-associated E protein. According to the reports, these antibodies, isolated from convalescent patients, could neutralize ZIKV with very high potency. So, they should be able to bind to the E protein on viral surface very well. However, when monomeric E protein was used as antigen for testing, the binding was very low, indicating monZE could not present tertiary and quaternary epitopes sufficiently (27). Hence, it might be very difficult to use monomeric E protein to select antibodies that could recognize the high-order epitopes. Based on these results, it could be highly desired that use of dimeric E protein for antibodies selection would result in much higher opportunity in isolation of antibodies targeting tertiary or quaternary epitope compared with use of monomeric E protein.

Recombinant E proteins can bind to sensitive cell lines
The E protein on the surface of ZIKV is essential for virus to bind and enter the host cell. Therefore, the binding ability of recombinant E proteins to sensitive cell lines is an important parameter to verify the conformation. As shown in Fig. 7A, monZE, ZE A264C , and ZE-Fc were able to bind to sensitive cell lines (e.g. bind to the cellular receptor). Although it seemed that the relative host-cell binding of ZE A264C and ZE-Fc was comparable with that of monZE (Fig. 7B), further evidence is necessary to know how the virus interacts with the host through E proteins.

Comparison of immunogenicity of recombinant E proteins in mice
To retain the natural conformation of antigens, water-soluble adjuvant was used here. As shown in Fig. 8A, all three recombinant E proteins were able to elicit high antibody responses in mice. To be specific, the antibody titer of ZE A264Cimmunized or ZE-Fc-immunized mice was 4-fold higher than

Engineered dimer of ZIKV E protein
that of monZE-immunized mice. The data indicate that mice immunized with dimeric E protein (ZE A264C or ZE-Fc) could produce higher antibody titer than that of monomeric E in vivo. The results of immunization in mice further disclose that dimerization of E protein could increase the immunogenicity of E protein.

Neutralization potency of antisera
The plaque reduction assay was employed to determine PRNT 50 values of pooled sera collected from each group. As shown in Fig. 8B, all of three group antisera were able to neu-tralize the ZIKV/SZ-WIV01 strain, but the potency was different. The PRNT 50 titers of anti-ZE A264C sera and anti-ZE-Fc sera are 896 and 867, respectively, whereas the PRNT 50 value of antisera from monZE-immunized group is 269, which shows a relative low neutralizing potency.

Discussion
To control ZIKV infection, mAbs and vaccines are very promising for treatment and prophylaxis. However, how to design an ideal immunogen for selection of mAbs with high potency and development of vaccine that could induce effective humoral
As mentioned in the introduction, mAbs targeting FLE normally cause serious ADE and their neutralizing activity is relatively modest. The mAbs that bind to DIII have reduced ADE but their neutralizing spectra are narrow. For now, it has been shown that mAbs recognizing quaternary conformation exhibit the most potent crossreactive neutralization and no obvious ADE. However, most of mAbs targeting quaternary structure E protein were isolated from convalescent patients, which is source limited. Meanwhile, because of the lack of native-like E protein, the antibodies selected from naïve phage display library or immunized animal might be modest or biased to certain epitope (20,39). Hence, preparation of native-like E protein (E dimer) is one of the key steps to overcome these limitations. Currently, only covalent linked ZIKV E mutant (ZE A264C ) has been reported to exist as dimer in solution (28). However, the detailed information on structure and function of it was still not very clear. Additionally, because native E dimer on viral surface forms noncovalently, a noncovalent dimeric E protein by fusion to antibody Fc fragment (ZE-Fc) was also designed and evaluated here.
A panel of positive antibodies was used to comprehensively characterize the recombinant protein ZE A264C and ZE-Fc. According to our results, the dimeric E could be efficiently formed under covalent or noncovalent condition and both of them could be recognized well by most of tested mAbs targeting tertiary or quaternary structure epitopes, which showed that major neutralizing epitopes including tertiary/quaternary structure epitopes are maintained in both ZE A264C and ZE-Fc.
The binding activities of each tested antibody to them are comparable in general.
Because EDIII contains important neutralizing epitopes, the domain itself is a superior ZIKV vaccine candidate that is better than monZE for immunization in mice (32). The explanation for this conclusion might be that monZE could elicit relatively more antibodies targeting FLE and further generate severe ADE, which drastically decreases the protection of such kind of immunogen. However, it might be effectively improved when using dimeric E protein as an immunogen because the FL region in E dimer could be buried more than that in E monomer. In our case, ZE A264C and ZE-Fc are recognized by EDIII-specific antibody better than monZE, whereas FLE-specific antibody could bind to monZE more strongly than ZE A264C and ZE-Fc.

Engineered dimer of ZIKV E protein
To further illustrate the advantages of dimeric E (ZE A264C and ZE-Fc) in vivo, immunization assay was performed in mice, and sera were obtained from different groups to assess the production of antigen-specific antibody and neutralizing potency of these antisera. First, we found ZE A264C and ZE-Fc could cause higher humoral immunity responses than monZE in vivo. Second, the antisera from ZE A264C -and ZE-Fc-immunized mice show higher neutralizing activities than that of monZE in vitro. Although we do not isolate neutralizing monoclonal antibodies from immunized mice, we could rationally speculate that stronger neutralizing potency probably derives from elicitation of tertiary/quaternary-specific antibodies, more EDIII-specific antibodies and fewer FLE-specific antibodies after immunization by ZE A264C and ZE-Fc compared with monZE.
Combined with all the results, it reasonably shows evidence that dimeric E might be a better candidate for immunogen than monomeric E. In addition, fusion with Fc fragment could bring extra benefits. It is able to enlarge molecular size and prolong the half-life of wtZE and bring other benefits in vivo after immunization, which could effectively enhance the immunogenicity and increase the reactivity of immune system (40,41).
In conclusion, two candidate immunogens, ZE A264C and ZE-Fc, could be useful for development of therapeutic mAbs and design of vaccines against ZIKV infection. Both strategies for dimerization could be expanded to study on antiviral reagents against other flaviviruses.
The Asian lineage ZIKV strain SZ-WIV01 (GenBank accession no.: KU963796) (42) was obtained from the Microorganisms & Viruses Culture Collection Center, Wuhan Institute of Virology, Chinese Academy of Sciences. The virus was amplified in Vero cells, and titers were determined by plaque assay as described previously (42). The sequence of its E protein is identical to that of the Asian strain ZIKV H/PF/2013 (GenBank accession no.: KJ776791) which has been widely used in related studies.

Production and purification of monZE, ZE A264C and Fc-fusion proteins
The gene encoding the E protein of ZIKV (the Asian strain H/ PF/2013, GenBank accession no.: KJ776791) and an enterokinase (EK) cleavage site followed by the Twin-Streptag were codon-optimized and synthesized (GENEWIZ) for Drosophila S2 expression system, yielding plasmid pUC57-ZIKV-E. For producing wtZE (1-408) (25), the gene of these fragments was amplified from pUC57-ZIKV-E and cloned into the secreted expression vector pMT/BiP/V5-His A (Invitrogen), which was named as pMT-ZE-EK-Twinstrep. The single Cys mutation E A264C was introduced by SOE (gene splicing by overlap extension) PCR with primers containing mutant sequence and the construct was named as pMT-ZE A264C -EK-Twinstrep and verified by DNA sequencing. The Fc-fusion proteins were constructed by fusing wtZE or ZE A264C to Fc fragment of human IgG1 with a flexible linker (mutate all Cys residues to Ser residues), which of them were named ZE-Fc and ZE A264C -Fc, respectively.
Stable expression S2 cell lines were created as described in Drosophila Expression System User Guide (Invitrogen). Briefly, Drosophila S2 cells were co-transfected with an expression vector (pMT-ZE-EK-Twinstrep, pMT-ZE A264C -EK-Twinstrep, pMT-ZE-Fc-EK-Twinstrep, and pMT-ZE A264C -Fc-EK-Twinstrep) and the selection vector pCoBlast (Invitrogen). Blasticidin (Invitrogen) was added into Schneider's Drosophila medium to select the stable cell lines. After 2-week culture in selective medium, stable S2 cell lines expressing wtZE, ZE A264C , ZE-Fc, and ZE A264C -Fc were obtained. Then the stable cell lines were adapted into protein-free Insect-XPRESS TM Medium (Lonza) and CuSO 4 was added at a final concentration of 500 M to induce the protein expression. The supernatants were collected 7-10 days after induction and concentrated by Vivaflow 200 (Sartorius) and purified via affinity chromatography with Strep-Tactin columns (IBA) according to the manufacturer's instructions. The elution of protein was loaded into Superdex 200 10/300 GL column (GE Healthcare) equilibrated with PBS (pH 7.4) for second-step purification. Purified proteins were concentrated by 10-kDa (wtZE) or 30-kDa (ZE A264C , ZE-Fc, and ZE A264C -Fc) cut-off membrane (EMD Millipore) and concentration was measured by NanoPhotometer 60 (Implen) with corresponding extinction coefficient.

Size exclusion chromatography (SEC)
To evaluate the purity and stability of monZE, ZE A264C , and ZE-Fc proteins in solution, purified proteins were loaded into Superdex 200 10/300 GL column again (GE Healthcare) equilibrated with PBS (pH 7.4). The UV absorbance at 280 nm was monitored.

Engineered dimer of ZIKV E protein MALDI-TOF MS analysis
MALDI-MS was acquired using 5800 MALDI-TOF/TOF (Applied Biosystems/MDS Sciex) equipped with an Nd:YAG laser with 355 nm wavelength of Ͻ500 picosecond pulse and 200 Hz repetition rate. The spectrometer was operated in positive mode and the spectra were accumulated by 1200 laser shots. The MS data were further processed by using Data Explorer 4.0 (Applied Biosystems/MDS Sciex). The 50% acetonitrile/water was used to dissolve the proteins, each 1 l of dissolved protein samples was mixed with 1 l freshly made sinapic acid (SA) solutions (10 mg/ml in 70% MeOH). Mixtures were then loaded onto the MALDI plate and the program was performed.

SDS-PAGE
Purified proteins were analyzed by SDS-PAGE. MonZE and ZE A264C protein samples were mixed with reducing-loading buffer or nonreducing-loading buffer (Sangon Biotech). For Fcfusion proteins, they were mixed with reducing-loading buffer and treated with boiling or nonboiling condition. Then all these samples were loaded on 10% SDS-PAGE gel. After electrophoresis, the gel was stained with Coomassie Brilliant Blue R-250.

Circular dichroism
The secondary structures of purified proteins were determined by CD. The proteins were dissolved in PBS at the final concentration of 0.3 mg/ml, and the CD spectra were recorded from 195 to 260 nm on an Applied Photophysics Chirascan-SF.3 spectrophotometer (Applied Photophysics Ltd.) at 25°C in a 0.1-cm pathlength cuvette (Applied Photophysics Ltd). Thermal stability was measured by recording the CD signal when temperature increased from 25 to 70°C (monZE and ZE A264C ) or from 25 to 94°C (ZE-Fc) with a ramp rate of 0.5°C/min at 216 nm. The CD data were shown as mean residue ellipticity.

Indirect ELISA
Purified monZE, ZE A264C , and ZE-Fc proteins were used as coating antigens, and BSA was negative control antigen. ELISA plates (Corning) were coated with 50 l of 4 g/ml protein overnight in PBS at 4°C and blocked with 100 l per well of 3% skim milk (Bio-Rad) in PBS at 37°C for 1 h. The plates were washed with PBS containing 0.05% Tween 20 (PBST), then 3-fold serial diluted positive antibodies were added and incubated at 37°C for 1.5 h. Plates were washed for five times with PBST and 50 l of 1:3000 HRP-conjugated anti-His antibody (Proteintech) for scFv form or 1:5000 HRP-conjugated anti-Human IgG (Fc specific) antibody (Sigma-Aldrich) for IgG form in PBS adding 1% BSA (Sangon Biotech) per well before incubation at 37°C for 1h. After washing with PBST, the binding was measured by the addition of diammonium 2,2Ј-azinobis (3-ethylbenzothiazoline-6-sulfonate) (ABTS) substrate (Invitrogen) and signal was measured at 405 nm.

Direct Sandwich ELISA
The positive antibodies were coated on the 96-well ELISA plates as capture antibodies. The methods of coating and blocking were carried out as described above. Then 3-fold serial diluted antigens (monZE, ZE A264C , and ZE-Fc) were added and incubated at 37°C for 1.5 h. After washing, 50 l of 1:10000 HRP-conjugated Strep-Tactin (IBA) was added and incubated at 37°C for 1 h. The binding activity was observed with ABTS substrate and measured at 405 nm.

Cytofluorometry
Vero cells were washed and resuspended in PBS containing 1% BSA and biotin-labeled 5-fold serial diluted monZE, ZE A264C , and ZE-Fc proteins for 1 h at 4°C. After washing, Phycoerythrin (PE)-conjugated streptavidin (Invitrogen) was added and incubated for another 1 h at 4°C. Further analysis was carried out with a FACSCalibur (BD Biosciences). The data were processed by the software of FlowJo X.

Mice, immunization, and ethics statements
All mouse experiments were performed in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals in China, and the protocols were approved by the ethics committee of the Wuhan Institute of Virology, Chinese Academy of Science (permit number WIVA34201702).
Female BALB/c mice (6 -8 weeks) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. The mice were housed in a specific pathogen-free animal laboratory under standard conditions. Prior to immunization, purified monZE, ZE A264C , and ZE-Fc proteins were mixed well with QuickAntibody-Mouse5W adjuvant (Biodragon Immunotech). Each injection dose contained 10 g protein and 50 l adjuvant in a volume of 100 l. In addition, PBS was mixed with same volume adjuvant as the negative control. Mice were divided into four groups (n ϭ 6): monZE, ZE A264C , ZE-Fc, and PBS. The mixture was injected into a quadriceps muscle of each mouse. After 3 weeks, a boost immunization was performed with the same dose and method. Blood samples were collected from the mice at 2 weeks after the final immunization and sera were isolated for further evaluation.

Antibody titer in sera
The antigen-specific serum antibody titers were measured by ELISA. Briefly, ELISA plate was coated with monZE, ZE A264C , and ZE-Fc respectively overnight at 4°C. Next day, the wells were blocked and then 4-fold serial dilutions (starting at 1:1000) of pooled mouse serum were added for 1.5 h at 37°C, followed by incubation with 50 l/well of HRP-conjugated goat antimouse IgG antibody (1:50,000 diluted in 1% BSA, Abcam) for 1 h at 37°C. After color development, the absorbance was measured at 405 nm and end point titer was reported as the reciprocal of the highest serum dilution that had an absorbance Ն0.1 optical density (OD) unit above that of the pre-immune samples.

Virus neutralization assay
The neutralization activities of immunized mouse sera were determined by performing plaque reduction assay. Briefly, pooled mouse sera were 3-fold serially diluted, starting at a dilution of 1:50, and then diluted serum samples were mixed

Engineered dimer of ZIKV E protein
with 100 plaque forming units ZIKV solution, followed by incubation at 37°C for 1 h. The serum/virus mixtures were added onto pre-seeded Vero cell monolayers in 24-well plates and incubated at 37°C for 1.5 h. Then, the medium was replaced with fresh DMEM containing 2% FBS and 2% methylcellulose. The plates were then transferred to 37°C in 5% CO 2 incubator. At 80 h post infection, plaques were visualized by fixation with 4% paraformaldehyde and staining with 0.1% crystal violet. For a given serum sample, the percent reduction of plaques was calculated by comparing the plaque number obtained to that of the virus only. The 50% plaque reduction neutralization titers (PRNT 50 ) were determined by a four-parameter logistic regression.