In vitro enhancement of Zika virus infection by preexisting West Nile virus antibodies in human plasma-derived immunoglobulins revealed after P2 binding site-specific enrichment

ABSTRACT Human immunoglobulin preparations contain a diverse range of polyclonal antibodies that reflect past immune responses against pathogens encountered by the blood donor population. In this study, we examined a panel of intravenous immunoglobulins (IGIVs) manufactured over the past two decades (1998–2020) for their capacity to neutralize or enhance Zika virus (ZIKV) infection in vitro. These IGIVs were selected specifically based on their production dates in relation to the occurrences of two flavivirus outbreaks in the U.S.: the West Nile virus (WNV) outbreak in 1999 and the ZIKV outbreak in 2015. As demonstrated by enzyme-linked immunosorbent assay (ELISA) experiments, IGIVs made before the ZIKV outbreak already harbored antibodies that bind to various peptides across the envelope protein of ZIKV because of the WNV outbreak. Using phage display, the most dominant binding site was mapped precisely to the P2 peptide between residues 211 and 230 within domain II, where BF1176-56, an anti-ZIKV monoclonal antibody, also binds. When tested in permissive Vero E6 cells for ZIKV neutralization, the IGIVs, even after undergoing rigorous enrichment for P2 binding specificity, failed, as did BF1176-56. Meanwhile, BF1176-56 enhanced ZIKV infection in both FcγRII-expressing K562 cells and human peripheral blood mononuclear cells. However, for enhancement by the IGIVs to be detected in these cells, a substantial increase in their P2 binding specificity was required, thus linking the P2 site with ZIKV enhancement in vitro. Our findings warrant further study of the significance of elevated levels of anti-WNV antibodies in IGIVs, considering that various mechanisms operating in vivo may modulate ZIKV infection outcomes. IMPORTANCE We investigated the capacity of intravenous immunoglobulins manufactured previously over two decades (1998–2020) to neutralize or enhance Zika virus infection in vitro. West Nile virus antibodies in IGIVs could not neutralize Zika virus initially; however, once the IGIVs were concentrated further, they enhanced its infection. These findings lay the groundwork for exploring how preexisting WNV antibodies in IGIVs could impact Zika infection, both in vitro and in vivo. Our observations are historically significant, since we tested a panel of IGIV lots that were carefully selected based on their production dates which covered two major flavivirus outbreaks in the U.S.: the WNV outbreak in 1999 and the ZIKV outbreak in 2015. These findings will facilitate our understanding of the interplay among closely related viral pathogens, particularly from a historical perspective regarding large blood donor populations. They should remain relevant for future outbreaks of emerging flaviviruses that may potentially affect vulnerable populations.

KEYWORDS intravenous immunoglobulin (IGIV), Zika virus, anti-WNV antibodies, cross-reactive antibodies, antibody-dependent enhancement (ADE) I mmunoglobulin preparations are highly purified liquid concentrates of immunoglobu lin G, manufactured on a large scale from pools of human plasma collected from thousands of healthy blood donors.Consequently, these preparations encompass a broad range of antibody specificities capable of providing effective protection against various infections commonly experienced by patients with primary immunodeficiency diseases (1,2).Essentially, each batch of immunoglobulins serves as a repository of polyclonal antibodies produced in response to previous infections within the donor populations (3).This historical perspective raises a key question: can antibodies initially developed against one pathogen in an immunoglobulin preparation affect a patient's response to a closely related but different pathogen in future encounters?
Zika virus (ZIKV), along with other clinically significant viruses closely related to it, such as West Nile virus (WNV), dengue fever virus (DENV), Japanese encephalitis virus (JEV), Yellow fever virus (YFV), and tick-borne encephalitis virus (TBEV), belongs to the Flavivirus genus within the Flaviviridae family (4,5).These single-stranded positive RNA viruses are best known for their extensive cross-reactivity, arising likely from substantial similarities in both genetic and immunological makeup (6)(7)(8).This assertion is strongly supported by evidence indicating that ZIKV can be clustered with DENV as a super serogroup in phylogenetic analyses of the main human pathogenic flaviviruses when comparing the amino acid sequences corresponding to the E protein, an envelope glycoprotein facilitating virus entry.This analysis reveals an amino acid conservation of 54%-57.8%when compared with DENV, whereas it shows a slightly lower 53% amino acid conservation when compared with WNV E protein (6).
Cross-reactive antibodies are often produced in humans during heterologous flavivirus infections.These antibodies not only complicate epidemiological surveillance and serological diagnosis in the context of past and present flavivirus infections but can also, paradoxically, lead to more severe pathogenesis in the setting of subsequent natural infection, a phenomenon known as antibody-dependent enhancement (ADE) (9)(10)(11)(12).
The risk of exacerbating disease severity through ADE during heterologous flavivi rus infections has raised significant public health concerns.For instance, infants born to mothers infected with DENV often experience more severe disease in heterotypic secondary DENV infections due to the presence of cross-reactive or enhancing antibod ies (13).The increased severity of DENV disease in seronegative individuals is attributed mainly to the dengue vaccine, which acts as a silent primary DENV infection and subsequently heightens the risk of severe disease upon natural infection with a DENV (14,15).Furthermore, a recent study on pediatric cohorts in Nicaragua revealed that prior ZIKV infection could significantly increase the risk of dengue diseases in humans, depending specifically on the infecting serotypes of DENV (16).
The role of ADE in the pathogenesis of ZIKV, particularly through anti-WNV antibod ies, appears relatively inconclusive.In mouse models, the exacerbation of ZIKV infection due to preexisting WNV immunity was observed (11).In addition, enhancement of ZIKV infection was also detected in vitro with anti-WNV antibodies in human serum samples collected from individuals with asymptomatic or symptomatic WNV infection (17).In contrast, preexisting WNV immunity did not seem to lead to detectable pathology during ZIKV infection in a mouse pregnancy model, suggesting that preexisting WNV immunity may not significantly impact the pathogenesis of ZIKV infection during pregnancy (18).
Because immunoglobulin preparations are derived from human plasma pools, concern has arisen regarding the potential risk of enhanced ZIKV infection if cross-reac tive antibodies to other flaviviruses are present in these products.This concern is particularly relevant considering the endemic nature of WNV infections in many parts of the continental U.S. (19).An estimated 7 million people have been infected with WNV since its introduction into the U.S. in 1999 (20).Given that approximately 80% of WNV infections in humans are subclinical or asymptomatic, it is not surprising that individuals who have recovered from WNV infections could still be eligible to donate blood for further manufacturing of plasma derivatives, such as immunoglobulins (17,21,22).Indeed, the average titers of antibodies to WNV in U.S. plasma-derived immuno globulins have already increased since the 1999 WNV outbreak (23,24).Interestingly, a recent study explored the effectiveness of ZIKVspecific immunoglobulin preparations in neutralizing or enhancing ZIKV and DENV infection in vitro.It revealed that these immunoglobulin preparations contained both cross-reactive neutralizing and enhancing antibodies, potentially influencing the severity of ZIKV disease in a mouse model (25).
In this study, we examined a panel of intravenous immunoglobulins (IGIVs) manufac tured over the past two decades to assess their ability to neutralize or enhance ZIKV infection in vitro.The selection of these IGIVs was based on the timing of their produc tion, coinciding with two significant flavivirus outbreaks in the U.S.: the WNV outbreak in 1999 and the ZIKV outbreak in 2015 (26).Our study revealed that antibodies elicited by WNV in IGIVs, although unable to cross-neutralize ZIKV, enhanced ZIKV infection upon specific enrichment for binding to the P2 site, located in the domain II of flavivirus E proteins.

Monoclonal antibodies, immunoglobulins, and peptides
Mouse anti-ZIKV monoclonal antibody, clone ZV-54, which binds to the domain III of the envelope protein (E) and neutralizes infection of African, Asian, and American strains of ZIKV to varying degrees (27), was purchased from Millipore Sigma (St. Louis, MO).Mouse anti-ZIKV monoclonal antibody BF1176-56, which binds to the recombinant E protein of ZIKV, was obtained from BioFront Technologies (Tallahassee, FL).IGIV lots derived from U.S. plasma were collected from various sources, including purchases from the National Institutes of Health (NIH) Clinical Center Pharmacy in Bethesda, MD; sample aliquots were stored at −80°C before use.All peptides used in this study were chemically synthesized by GenScript, Inc. (Piscataway, NJ) based on the sequences of ZIKV_H/PF/ 2013 and WNV_NY99 (28,29).These peptides covered the entire ZIKV E protein and the WNV E protein partially, as outlined in Tables 1 and 2. Each peptide was biotinylated at the N-terminus with the sequence SGSG as a linker between the peptide and the biotin tag.

Enrichment for WNV P2-binding antibodies
Affinity chromatography to enrich an IGIV sample for its antibody binding specifically toward the WNV P2 peptide was done using the BioLogic DuoFlow chromatography system (Bio-Rad Laboratories, Hercules, CA).Briefly, a total of 10 mg of the N-terminally biotinylated WNV P2 peptide was diluted in PBS to prepare a peptide solution with a concentration of 100 µg/mL.This peptide solution was then loaded onto a 5 mL HiTrap Streptavidin Sepharose column (Millipore Sigma) at a flow rate of 2 mL/min.The column was then washed to remove excess peptide.Affinity binding occurred by loading up to 50 g of an individual lot of IGIV manufactured after 2007, such as 2007 IGIV 1, 2007 IGIV 2, 2007 IGIV3, and 2019 IGIV, at a concentration of 5 mg/mL in phosphatebuffered saline (PBS) (pH 7.4), with a flow rate of 2 mL/min.Following extensive washing with approximately 200 mL of PBS, the antibodies were eluted from the column using a 0.1 M glycine buffer (pH 2.5).The eluates were then immediately neutralized by adding 1 M Tris-HCl buffer (pH 8.0).Buffer exchange was performed by dialysis of the eluates against an excess volume of PBS (approximately 1 L) three times at 4°C, with each dialysis session spanning more than 3 hours.Fractions containing WNV P2 peptide-binding antibodies, termed P2 enriched, were collected and stored at −80°C before determining their specific activity in binding to the WNV peptide using an enzyme-linked immunosorbent assay (ELISA).Additionally, the flowthrough samples post-enrichment, termed FT, were also collected and stored at −80°C for further analysis.

ELISA
The WNV E IgG ELISA kit, purchased from EUROIMMUN (Luebeck, Germany), was used following the manufacturer's protocol.In brief, the anti-WNV IgG ELISA measured total  IgG antibodies to WNV in both IGIV and IGIV enriched for WNV P2 peptide binding specificity, using recombinant WNV glycoprotein E as the antigen source.After adding IGIV samples at 200 ng/well, anti-human IgG HRP conjugates were used as secondary antibodies at a dilution of 1:3,000 for detection in a colorimetric reaction with the addition of a 3,3' ,5,5'-tetramethylbenzidine (TMB) substrate.Each assay included controls and a calibrator from the kit.The data were analyzed by calculating the ratio of the optical density (O.D.) at 450 nm (OD450) of the IGIV sample to that of the calibrator.A ratio of ≥1.1 was considered positive for WNV IgG antibodies.Similarly, a peptide-based ELISA was conducted to assess the presence of anti-WNV antibodies in both IGIV and IGIV enriched for WNV P2 peptide binding specificity.Briefly, streptavidin pre-coated 96-well plates were loaded with biotin-conjugated peptides encompassing the E protein of either ZIKV or WNV at 500 ng/well, followed by the addition of a diluted IGIV sample at 200 ng/well.Anti-human IgG HRP conjugates were then used as secondary antibodies at a dilution of 1:3,000 for detection in a colorimetric reaction with the addition of TMB substrate.Each assay included unrelated peptides as negative controls, and the data were analyzed by determining the ratio of the OD450 of the IGIV sample to that of the native control.A ratio of ≥3.0 was considered positive for anti-WNV antibody binding.

Antibody-binding sites mapped by screening phage display libraries
The residuespecific binding of an antibody to peptides was analyzed by screening random peptide phage display libraries following the manufacturer's protocols with modifications (New England BioLabs, Beverly, MA) (30).Briefly, approximately 200 ng of BF1176-56 anti-ZIKV monoclonal antibody or IGIV samples after WNV P2-peptide affinity enrichment were loaded onto 50 µL protein G-coated magnetic beads (Dynabeads™ Protein G, Thermo Fisher Scientific) in 0.05 M Tris-HCl buffer (pH 7.5) containing 0.15 M NaCl and 0.05% Tween 20 (TBS-T) at room temperature (RT).A total of 10 10 plaqueforming units (pfu) of phage from Ph.D. 7-and 12-Phage Libraries were added to the antibody-coated beads in a volume of 100 µL of TBS-T.The mixture was then incubated at room temperature for 20 minutes with gentle shaking.Following 6-8 washes with TBS-T, the phages bound to the beads were eluted with 100 µL of 0.2 M glycine-HCl solution (pH 2.2) and immediately pH-neutralized with 15 µL of 1 M Tris-HCl buffer (pH 9.0).The eluted phages were then amplified in E. coli K12 ER2738 (New England Biolabs).After repeating this process three times, phage DNA from a single plaque was prepared for DNA sequencing, and the corresponding peptide sequence was deduced from the DNA sequence.Subsequently, the sequences of the phage-displayed peptides and the proteins of relevant flaviviruses were aligned to determine the extent of similarity.

Infectious clones and viruses
An infectious clone, nLuc-ZIKA, derived from ZIKV PRVABC59 and constructed to express the nanoluciferase gene in the duplicated capsid gene region, was cultured in Vero E6 cells (an African green monkey kidney epithelial cell line) and purified as previously reported (31,32).Viral stock titers were determined in a series of 4-fold dilutions with 5 × 10 3 cells/well of Vero E6 cells in 96-well plates.The presence of infection was determined by measuring the intracellular nanoluciferase signal at 48 hours post-infection.Wells exhibiting a signal 10-fold higher than the sham infection background were considered positive for infection.An apparent Median Tissue Culture Infectious Dose (TCID 50 ) titer was calculated using a Spearman-Kärber algorithm.

Antibody binding of the P2 site displayed on the nLuc-ZIKA virion
To perform an ELISA-based virion capture assay, 100 µL of mouse anti-ZIKV antibody (ZV-54) was initially added to each well of a 96-well plate at a final concentration of 10 µg/mL.The plate was then incubated at room temperature for 1 hour to allow the wells to be coated with the antibody.Afterward, the wells were blocked with 5% skim milk in phosphatebuffered saline containing 0.1% Tween 20 (PBS-T) (pH 7.4) at RT for 30 minutes.Prior to use in this experiment, the nLuc-ZIKA virus stock was inactivated with a UVC dose of 3.0 J/cm², allowing its safe use in Biosafety Level 2 (BSL-2) facilities.The inactivation of the virus was confirmed by testing on permissive Vero E6 cells.Subsequently, 100 µL of the inactivated virus was added to the antibody-coated wells at a virion concentration equivalent to 9.97 × 10 6 TCID 50 /mL.Following incubation at 37°C for 1 hour to capture the virion, the wells were washed three times with PBS-T.IGIV samples, with or without P2 enrichment, were then added to the wells as primary antibodies at different final concentrations ranging from 0 to 1 mg/mL.These IGIV samples were diluted with 5% skim milk in PBS-T.The 96-well plates were incubated at 37°C for 1 hour.After incubation, the wells were washed four times with PBS-T.Next, 100 µL of anti-human IgG HRP conjugates, diluted at a ratio of 1:3,000, was added as the secondary antibody to the wells.The plates were then incubated at room temperature for 30 minutes to allow for subsequent detection in a colorimetric reaction upon addition of the TMB substrate.Relative binding was measured at OD450 and plotted against the concentration of IGIV (µg/mL).

In vitro neutralization and ADE assays
Vero E6 cells and Fcγ receptor II (FcγRII, CD32A)-expressing human lymphoblast K562 cells were obtained from the American Type Culture Collection (ATCC).These cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) and Roswell Park Memorial Institute (RPMI) Medium, respectively, supplemented with 10% fetal bovine serum (FBS).Cell-based experiments were conducted under BSL-2 conditions, following proper safety protocols and guidelines.
An in vitro antibody-mediated neutralization assay was performed according to a previously reported study (27).Briefly, IGIVs or IGIVs enriched for P2 peptide binding specificity were serially diluted in DMEM and incubated with nLuc-ZIKA for 1 hour at 37°C.Cell culture media with an equivalent volume were used as negative controls for the experiments.The mixture of virus with immunoglobulin samples was then added to Vero E6 cells, and viral infectivity was assessed for the presence of infection using the Nano-Glo luciferase assay system (Promega, Madison, WI).Neutralizing activity was determined as a percentage after normalizing the data to cells infected without an immunoglobulin sample or a culture media control.
Similarly, the effects of ADE were measured 48 hours post-infection with nLuc-ZIKA using FcγRII-bearing K562 cells in 96-well plates, following the procedure previously reported (27,33).Data analysis involved normalizing the wells that showed maximum enhancement, as detected by relative luciferase activity using the Nano-Glo assay system.
ADE assay with human peripheral blood mononuclear cells (PBMCs) was performed according to a previous report (34).Briefly, PBMCs, purchased from Lonza (Cohasset, MN), were cultured in RPMI-1640 medium supplemented with 10% FCS.Therefore, 2 to 4 × 10 5 PBMCs in 100 µL were seeded in 96-well plates in the presence or absence of 4,200 U/mL human macrophage-colony stimulating factor (M-CSF) (Thermo Fisher Scientific) to drive macrophage differentiation from monocytes.The cells, after M-CSF treatment, were then cultured for an additional 5 days for subsequent virus infection.nLuc-ZIKA was incubated with serially diluted IGIVs or IGIV samples enriched for P2 peptide-binding specificity for 1 hour at 37°C before adding this mixture to the cultured PBMCs.For PBMCs without M-CSF treatment, the cells were infected with ZIKV at a multiplicity of infection (MOI) of 2, whereas for PBMCs treated with M-CSF, an MOI of 1 was used.The infection was allowed to continue for a period of 2 hours at 37°C.The infected cells were washed twice with RPMI-1640 medium.After 2 days, viral infectivity was assessed using the Nano-Glo luciferase assay system.

Statistical analysis
Statistical analysis was conducted using GraphPad Prism (GraphPad Software, San Diego, CA).The comparison between multiple groups was performed using the Kruskal-Wallis test, and Spearman's correlation with linear regression was used for all correlation determinations using GraphPad Prism Software.P values below 0.05 were considered significant.

Detection of ZIKV cross-reactive antibodies in IGIVs manufactured after the WNV outbreak
We quantitatively measured the presence of WNVspecific antibodies in eight lots of IGIV manufactured from 1998 to 2020 using a commercially available ELISA kit (Fig. 1).We observed that the three lots produced prior to the 1999 WNV outbreak contained minimal, if any, detectable anti-WNV activity.In contrast, the other five lots manufactured thereafter consistently showed anti-WNV activity (Fig. 1).This result aligns with previous observations indicating that US plasma-derived IGIV lots released during the period from 2003 to 2008 exhibited WNV neutralization titers ranging from 2.8 to 69.8 (23,24).
We then investigated the potential cross-reactivity of these anti-WNV antibodies with ZIKV.A series of overlapping peptides covering both the entire ZIKV E protein and partially the WNV E protein was chemically synthesized (Tables 1 and 2).The binding of antibodies to these ZIKV peptides was measured in 10 IGIV lots using an ELISA (Fig. 2).Of these lots, seven were manufactured between 2007 and 2019 and exhibited varying degrees of binding to ZIKV peptides, with the most dominant binding observed at the P2 peptide, spanning residues 211-230, within domain II.In contrast, the three remaining lots made before the 1999 WNV outbreak failed to bind significantly to these peptides.
To investigate the potential contribution of P2 sitespecific antibodies, most likely generated in response to the earlier WNV outbreak, to cross-reactivity with ZIKV, we conducted an ELISA comparing the binding capacity of these anti-WNV antibodies to the P2 peptide within domain II of the E protein of ZIKV_H/PF/2013 and WNV_NY99.The potential cross-reactivity was attributed to the 50% homology between their amino acid sequences (Fig. 3A).Although all seven IGIV lots produced after the WNV out break retained their binding capacity to the P2 peptide derived from WNV_NY99, they exhibited a similar pattern of binding concurrently with the P2 peptide derived from ZIKV_H/PF/2013.The three lots manufactured before the WNV outbreak did not show significant binding to the P2 peptides (Fig. 3B).These findings confirmed that antibodies FIG 2 Presence of cross-reactive antibodies to ZIKV in IGIV lots.The levels of cross-reactive antibodies to ZIKV in 10 IGIV lots manufactured between 1998 and 2019 were measured in an ELISA, with the antigenic targets being a specific set of ZIKV-E peptides as described in Table 1.The X-axis denotes the locations of these peptides within the domains of ZIKV-E, whereas the Y-axis represents the average antibody levels to ZIKV-E peptides, expressed as O.D. at 450 nm.Each assay was repeated at least three times.
reactive to WNV, elicited from the earlier response to the WNV outbreak, constituted the primary source of cross-reactivity observed with the IGIV lots against ZIKV.

Determination of the residue specificity of anti-WNV antibodies within the IGIV lots that target the P2 site
We conducted a series of experiments to define the residue specificity of anti-WNV antibodies within the IGIV lots, specifically targeting the P2 site, by screening random peptide phage display libraries (Fig. 4).We initially enriched anti-WNV P2 antibodies from IGIV lots produced after the WNV outbreak using affinity chromatography, achieving a calculated enrichment factor of approximately 10 5 .This enrichment factor was deter mined based on the anti-WNV P2specific activity in ELISA experiments conducted on two 2007 IGIV lots before and after enrichment for P2 binding.The specific activity for 2007 IGIV-1 and 2007 IGIV-2 was calculated as 546 ± 168 and 493 ± 112 RU/mg IgG before enrichment, and 7605658 ± 91215 and 11549545 ± 69285 RU/mg IgG after enrichment, respectively.Subsequently, we used this unique sample as the primary source of antibodies to screen the phage display libraries.The positively selected peptides displayed by the phages were then compared with the linear P2 sequence  1 and 2. The X-axis denotes the ZIKV P2 and WNV peptides used in the assays, whereas the Y-axis represents the average antibody levels to these peptides, expressed as O.D. at 450 nm.Each assay was repeated at least three times, and standard deviations are presented as error bars.
derived from both ZIKV_H/PF/2013 and WNV_NY99.We identified a group of displayed peptides containing short motifs, such as IPLP, LNLP, MPW, and WPW.These motifs could collectively form a consensus sequence of (I/L)(P/N)LP(W/F/V), closely resembling 221 IPLPW 225 in the P2 of ZIKV and 221 LNLPW 225 of WNV (Fig. 4).Our analysis suggested that this set of residues in the P2 peptide was directly contacted by anti-WNV antibodies present in the IGIV lots.We also observed several other groups of peptides that could be recognized independently by the IGIV lots even after rigorous enrichment for their P2-binding specificity, likely due to the wide-ranging antibody diversity inherent in IGIV preparations (data not shown).
Recognizing the intricate complexity of polyclonal antibodies within the plasmaderived IGIVs, we aimed to identify a monoclonal antibody that targets specifically the P2 site to facilitate our understanding of the observed antibody cross-reactivity in IGIV lots.After mapping binding sites in a diverse pool of anti-ZIKV antibodies, previously collected in our lab, using random peptide phage display, we identified that the monoclonal antibody BF1176-56 exhibited a residuespecific binding pattern closely resembling that observed in anti-WNV P2 antibodies from the IGIV lots (Fig. 5A).Specifically, BF1176-56 recognized a motif of WxxDxxLPWH, where the residues W, D, and LPW appeared with 100% conservation across human pathogenic flaviviruses, although the residue H at the end of this motif was unique to ZIKV (Fig. 5B).Additionally, when compared with the entire WNV P2 site (aa.211-230), a range of 40%-75% amino acid sequence homology to that of other flaviviruses was revealed (Fig. 5B).Therefore, monoclonal antibody BF1176-56 was selected to serve as a reference in our study to define the residue specificity of antibody binding in IGIV lots.Furthermore, we compared the three-dimensional structures of the respective E protein monomers from WNV (PDB ID 2HG0), ZIKV (PDB ID 5IRE), and DENV3 (PDB ID 1UZG) by analyzing structural alignment using PyMol (The PyMOL Molecular Graphics System, Version 2.5.4,Schrö dinger, LLC) with a built-in global multiple alignment command (Fig. 5C).The amino acid sequences between Val206 and Thr266 (DENV3 coordinates) were selected for alignment to focus on the binding site of the anti-ZIKV antibody BF1176-56.Coinciden tally, this region also contained the epitope of the anti-DENV neutralizing monoclonal antibody d448, which was isolated from rhesus macaques after immunization with an experimental dengue vaccine (35).Our analysis revealed that the binding residues of BF1176-56 were superimposable among WNV, ZIKV, and DENV, thus suggesting the cross-reactivity among these viruses, particularly around the P2 site.Moreover, the two shared residues (D and P) crucial for the antibody binding of BF1176-56 and d448 were found to overlap precisely at the local 3-dimensional structure level (Fig. 5C).Although BF1176-56 displayed a preference for residues in the N-terminal segment of this region for typical linear binding (W217, D220, L223, P224, and W225 in WNV), d448 recognized an extended range of residues scattered throughout the entire aligned region between V206 and T266 in DENV (35).The residues in the binding site of d448 appeared to be arranged in a conformational manner, including D213, P217, L235, Q254, and G264, as evidenced by the coordinates in the PDB ID 1UZG (Fig. 5C).It is worth noting that this conformational epitope was proposed to be a part of the interface between the E protein and the M protein in DENV, thus providing structural insights into d448's broad neutralizing capacity against all four serotypes of dengue (35).

Binding of Zika virions by IGIV after WNV P2 enrichment
To evaluate the ability of IGIV samples to recognize the P2 site on ZIKV virions, we conducted an ELISA-based capture experiment using IGIV samples with or without enrichment for P2-binding (Fig. 6).We observed that IGIV lot samples known to contain anti-WNV antibodies (2007 IGIV 1 and 2007 IGIV 2) were unable to capture the virions without P2 enrichment.In contrast, following the P2-enrichment process with a specific enrichment factor of approximately 10 5 , the IGIV samples (2007 IGIV 1-P2 enriched and 2007 IGIV 2-P2 enriched) displayed dose-dependent virion capture by binding to the naturally displayed P2 site, thereby defining the half-maximal effective concentration (EC50) at 10.19 µg/mL for 2007 IGIV 1-P2 enriched and 15.68 µg/mL for 2007 IGIV 2-P2 enriched (Fig. 6).These results provide evidence that WNV P2 sitespecific enriched antibodies were capable of cross-reacting with the P2 site on the ZIKV virions.

Lack of cross-neutralizing activity against ZIKV by anti-WNV antibodies present in IGIVs
Upon discovering that antibodies within the IGIV lots manufactured before the 2015 ZIKV outbreak were cross-reactive to ZIKV due to prior WNV infections, especially at the shared P2 binding site within domain II, we examined three 2007 IGIV lot samples for their ability to cross-neutralize ZIKV via binding to the P2 site.These three 2007 IGIV lots, released in 2008, were previously determined to contain anti-WNV antibodies, with an average WNV neutralization titer of 21 ± 1 (23,24).Subsequent neutralization assays were conducted using the ZIKV-permissive Vero E6 cell line.Surprisingly, neither the IGIV lots nor the individual IGIV lot, which underwent rigorous enrichment for P2 peptide specificity, demonstrated the ability to block ZIKV infection.In contrast, under the same experimental conditions, the monoclonal antibody ZV-54, which served as a positive control, effectively neutralized the virus (Fig. 7A).Like the IGIVs, the other anti-ZIKV monoclonal antibody BF1176-56, although it bound to the P2 site in ZIKV as well, did not exhibit any detectable neutralizing activity (Fig. 7B).

Enhancement of ZIKV infection potential by IGIVs after enrichment for P2 site binding specificity
We examined whether anti-WNV antibodies in the IGIV lots could enhance ZIKV infection in FcγRII-expressing K562 cells.As shown in Fig. 8, the enhancement by the IGIV lots was not detectable unless their P2 site binding specificity significantly increased through affinity enrichment.Similarly, BF1176-56 also demonstrated the ability to enhance ZIKV infection, with an effective dose range detected from 0.1 to 10 µg/mL, which is notably lower compared with the approximately 100 times higher effective dose range required for the P2-enriched IGIV samples (Fig. 8).The increased dosage requirement for the P2-enriched samples to exhibit ADE may be attributed to the significant heterogeneity of antibodies present due to their polyclonal nature, despite the enrichment for P2 specificity.Although these results confirm the functional equivalency of BF1176-56 with the P2-enriched IGIV samples, they suggest that a sufficient quantity of P2 sitespecific antibodies is needed to detect the effect of ADE.Indeed, it is noticeable that the range of antibody doses required to demonstrate ADE in this experiment largely overlaps with the range needed for cross-capturing the ZIKV virion, that is, 10-100 µg/mL (Fig. 6).
We also investigated whether anti-WNV antibodies present in the IGIV lots could enhance ZIKV infection in human primary PBMCs.In this experiment, the PBMCs were either induced to undergo macrophage differentiation by macrophage colonystimulating factor (M-CSF) or left untreated.We observed a significant enhancement of ZIKV infection by the IGIV lot samples, which became detectable after increasing their specificity for the P2 site through affinity enrichment (P < 0.01), regardless of M-CSF treatment (Fig. 9).Additionally, there was a more than 25-fold difference in ZIKV infectivity measured in PBMCs treated with M-CSF compared with untreated cells (Fig. 9A and B).The relatively increased infectivity observed in M-CSF-treated cells was accompanied by changes in cell adhesion and macrophage morphology.Similar to the effect observed in FcγRII-expressing K562 cells, BF1176-56 also enhanced ZIKV infection in PBMCs with a minimal effective dose of 0.1 µg/mL (data not shown).These results confirmed that enriching the P2 site specificity of preexisting anti-WNV antibodies in IGIV lots revealed their ability to enhance ZIKV infection in human primary cells in vitro.

Attenuation of P2 antibody-mediated enhancement in ZIKV infection by non-anti-WNV antibodies in the IGIVs
We investigated whether the inability of IGIV lots, in the absence of the P2 enrichment process, to enhance ZIKV infection is due to the abundance of other non-anti-WNV antibodies present in the IGIV products, given their polyclonal nature (Fig. 10).To address this question, human PBMCs were prepared by pre-treating them with M-CSF for 4 days to allow macrophage differentiation.ADE in ZIKV infection was then measured in these cells by mixing 100 µg of the 2007 IGIV 1 P2-enriched sample (labeled as P2 enriched) at ratios of 1:3 or 1:10 (weight/weight) with the IGIV flowthrough samples collected after P2 enrichment (labeled as FT).The dose of 100 µg/mL was selected based on the quantity of anti-WNV antibodies in this P2-enriched sample (approximately 10 EC50 for virion binding), which was sufficient to demonstrate cross-reactivity with the P2 site displayed on the ZIKV virion and enhancement of ZIKV infection in human PBMCs (Fig. 6  and 8).Additionally, samples of an IGIV lot (1998 IGIV 1) manufactured prior to the WNV outbreak, which contains no detectable anti-WNV antibodies, were also included in the experiment as a negative control (Fig. 10).We found that both non-anti-P2 and non-anti-WNV antibodies in the IGIVs reduced ADE at the P2 site in a dose-dependent manner, resulting in a reduction of ADE more than 60% (Fig. 10).These results demonstrate the ability of antibodies other than those targeting WNV to mitigate the phenomenon of anti-WNV antibody-mediated ADE in ZIKV infection in human primary cells.

DISCUSSION
Guided by their production dates that align specifically with the timing of two major flavivirus outbreaks in the U.S., the WNV outbreak in 1999 and the ZIKV outbreak in 2015, we investigated a selection of U.S. plasma-derived immunoglobulins manufactured during the last two decades to gain historical insights into how our immune system responds consecutively to closely related viral pathogens.Specifically, we found that antibodies generated in response to prior WNV infections can interact with ZIKV.This cross-reactivity results from the substantial genetic and immune-related similarities shared by these flaviviruses, illustrating how past infections can shape our responses to future closely related pathogens.
WNV antibodies in the IGIV lots cross-bind to ZIKV, but they cannot neutralize it; instead, they enhance its infection in both FcγRII-bearing K562 cells and human PBMCs, with or without M-CSF treatment.However, the observed enhancement is circumstantial, occurring only after the WNV antibodies are significantly concentrated from IGIV lots.Furthermore, the enhancement can be precisely traced to the binding of non-neutraliz ing antibodies against ZIKV at the P2 site, corresponding to amino acids 211-230 in the domain II of the WNV E protein.Given the 40%-75% homology range of the P2 site between WNV and other flaviviruses, as illustrated in Fig. 5B, it is conceivable that further enrichment for anti-WNV activity through the WNV P2 site from human plasmaderived IGIV may fortuitously increase the cross-reactivity of the enriched antibodies with other flaviviruses.However, the outcome of this enrichment-whether it leads to cross-neutralization or cross-enhancement of infection by other flaviviruses-requires investigation in each individual case.With this consideration, it becomes important to systematically analyze other dominant and subdominant cross-binding sites of preexisting WNV antibodies in the IGIV lots, as described in Fig. 2, to assess their role in cross-reactivity with ZIKV and other closely related flaviviruses, such as WNV and DENV.Intriguingly, the P2 site as described in our study partially overlaps with the epitope of d448, a monoclonal antibody, which has been shown to broadly neutralize all four serotypes of DENV in a rhesus macaque model after vaccination with tetravalent recombinant E proteins (DEN-80E) (35).Despite its broadly neutralizing activity against DENV, antibody d488 exhibited weak or minimal neutralizing activity (around 30%) against ZIKV and WNV at concentrations above 1 µg/mL and above 30 µg/mL, respec tively.Our structural analysis reveals a close alignment of the P2 site in the E proteins of WNV, ZIKV, and DENV, particularly at central positions where D and P residues overlap.Through local structural comparison, we propose differences in interaction patterns between anti-WNV antibody BF1176-56, representing WNV P2-enriched antibodies in IGIVs, and anti-DENV antibody d448 at the P2 site.BF1176-56 primarily binds specific residues in the N-terminal segment of WNV domain II, indicating a linear epitope (e.g., W217, D220, L223, P224, and W225).In contrast, d448 exhibits a broader bind ing spectrum, interacting with additional residues (e.g., L235, Q254, and G264) across the entire DII of DENV.These residues overlap with the E protein-M protein interface in DENV (based on coordinates from PDB ID 1UZG).Further comparative analyses of these immune complex structures in the future could elucidate the significance of the epitope-level connection among antibodies targeting these human pathogenic flaviviruses concerning cross-reactivity, whether for neutralization or infection enhance ment.
Differences in the levels of anti-WNV antibodies between convalescent plasma from individual donors and IGIV lots should not be overlooked.Convalescent plasma, sourced from individuals who have recovered from symptomatic or asymptomatic WNV infections, typically contains higher levels of anti-WNV antibodies.In contrast, IGIV lots primarily consist of a mixture of non-anti-WNV antibodies from healthy donors, although anti-WNV antibodies can be detected in IGIVs made after WNV outbreaks.Recognizing the disparity in antibody composition between convalescent plasma and IGIVs may help elucidate the importance of enriching IGIVs with WNV P2specific antibodies while reducing non-anti-WNV antibodies to detect enhancement in ZIKV infection.Recogniz ing the differences in antibody composition between convalescent plasma and IGIVs may help elucidate the potential importance of enriching IGIVs with WNV P2specific antibodies while reducing non-anti-WNV antibodies to detect enhancement in ZIKV infection.Consequently, we addressed the question raised by our study regarding why IGIV lots, without enrichment for P2 site binding specificity, failed to exhibit ADE in ZIKV infection.In our experiment by mixing WNV P2specific antibodies with non-anti-WNV antibodies, we found that the presence of antibodies other than anti-WNV polyclonal antibodies, particularly non-P2 sitespecific antibodies, in the IGIVs reduces ZIKV ADE in human PBMCs.We speculate that this reduction occurs due to competition for Fc receptor binding, wherein non-anti-WNV antibodies outcompete the smaller pool of anti-WNV antibodies within the IGIV lots.Similar reductions in ZIKV infection have been observed in other studies when an excessive amount of anti-ZIKV monoclonal antibodies was present, irrespective of their neutralizing activity (27,33).Although the exact mechanism underlying the absence of detectable ADE with IGIV lots remains to be investigated, the competition for Fc receptor binding, if occurring in vivo, could potentially shield host cells from ZIKV ADE.
Although our observations have established enhanced ZIKV infection in vitro due to WNV antibodies enriched from the IGIV lots, interpreting the significance of this phenomenon in vivo requires caution, as various mechanisms known to operate in vivo, such as suppression by serum complement and Fc receptor availability, can modify the outcomes of flavivirus infections (36)(37)(38).Future studies are needed to determine whether our findings of ADE in ZIKV infections in vitro align with occurrences in vivo, should they indeed occur.In this context, clarifying this relationship may enhance our understanding of the pathological implications of ZIKV ADE, especially with preexist ing WNV antibodies.Additionally, epidemiological studies may be necessary to better understand the true incidence of ADE in ZIKV infection, with clear evidence of traditional signs such as increased viral load or aberrant immune responses leading to more severe disease.Nevertheless, our study lays the groundwork for exploring relationships among closely related pathogens from a historical perspective.It demonstrates the feasibility of surveilling viral pathogen antibody landscapes in large blood donor populations through comparative analysis of a few carefully selected plasma-derived IGIV products.
Finally, the past ZIKV epidemic serves as a reminder that initially obscure flaviviruses can swiftly evolve into significant public health threats within a compressed timeframe (8,39).Our study retains relevance in the event of future outbreaks involving emerging flaviviruses that may impact vulnerable populations.

FIG 1
FIG 1 Presence of anti-WNV antibodies in IGIV lots derived from plasma collected from U.S. donors.The levels of WNV E IgG in IGIV produced between 1998 and 2020 were measured using an ELISA kit purchased from EUROIMMUN, with recombinant WNV glycoprotein E as the antigen source.The X-axis represents the IGIV lots used in this experiment.The Y-axis indicates the average level of anti-WNV antibodies expressed as Relative Units (RU)/mg IgG.The assay was performed three times, and standard deviations are shown as error bars.

FIG 3
FIG 3 Determination of antibodies present in IGIV lots that exhibit cross-reactivity with a specific peptide in the domain II of both ZIKV and WNV.(A) Sequence alignment of a segment of domain II (DII) between ZIKV_H/PF/2013 (GenBank Genome Accession code KJ776791) and WNV_NY99 (GenBank Genome Accession code DQ211652).P2 peptide sequence is indicated by an arrow.(B) Detection of cross-reactive antibodies to ZIKV and WNV in the IGIV lots.The levels of cross-reactive antibodies to both ZIKV and WNV were determined in 10 IGIV lots manufactured between 1998 and 2019 using an ELISA.The antigenic target was the P2 peptide, spanning amino acids 211 to 230 in ZIKV_H/PF/2013 or WNV_NY99, as described in Tables1 and 2. The X-axis denotes the ZIKV P2 and WNV

FIG 4
FIG 4 Determination of residue specificity of the antibodies in the IGIV lots after enrichment for WNV P2 binding through screening of Ph.D. 7-and 12-mer random peptide phage display libraries.The peptide sequences identified after three rounds of screening using an IGIV sample enriched for P2 peptide specificity are depicted.These sequences are compared with the linear sequences of the P2 peptide, spanning amino acids 211 to 230 within domain II of ZIKV_H/PF/2013 (GenBank Genome Accession code KJ776791) and WNV_NY99 (GenBank Genome Accession code DQ211652).The potential contact residues by the antibodies are highlighted, with the key residue or its alternative presented in parentheses as the consensus sequence.

FIG 6
FIG 6 Evaluation of the ability of IGIV samples to recognize ZIKV virions.The nLuc-ZIKA virions were first inactivated by UVC and then captured in 96-well plates through interaction with the mouse monoclonal anti-ZIKV antibody, clone ZV-54.Subsequently, the captured virions served as the antigenic target for specific binding by each of the four primary antibodies, including 2007 IGIV 1, 2007 IGIV 2, 2007 IGIthe neutralizing antibody ZV-V 1-P2 enriched, and IGIV 2-P2 enriched in an ELISA using the anti-human IgG-HRP conjugate as the secondary antibody.The X-axis represents the concentration of antibodies used in the experiment, whereas the Y-axis indicates relative binding to the captured virion.The assay was performed three times, and standard deviations are presented as error bars.

FIG 7
FIG 7 Evaluation of ZIKV neutralization in vitro by IGIVs and IGIV samples following enrichment for P2 peptide specificity.(A) Four IGIV lots, produced between 1998 and 2019, and three P2-enriched IGIV samples were examined for their ZIKV neutralization capacity using Vero E6 cell infectivity assays.The neutralizing antibody ZV-54 served as a positive control.(B) The neutralizing activity of monoclonal antibody BF1176-56, which recognizes the P2 of both ZIKV and WNV, was evaluated.The assay included the positive control antibody ZV-54 and a murine IgG1 isotype control.The X-axis represents the concentration of antibodies used in the experiment, whereas the Y-axis indicates the neutralizing capacity, expressed as a percentage of the negative control.The assay was performed three times, and standard deviations are presented as error bars.

FIG 8
FIG 8 Evaluation of ADE of ZIKV infection in FcγRII-expressing K562 cells by anti-WNV antibodies present in the IGIV lots and IGIV samples following enrichment for P2 peptide specificity.Five IGIV lots produced between 1998 and 2019, along with three P2-enriched IGIV samples, were tested in FcγRII-expressing K562 cells for ADE of nLuc-ZIKV infection.Monoclonal antibody BF1176-56, recognizing the P2 site shared by both ZIKV and WNV, was also assessed for ADE in the K562 cells with nLuc-ZIKV.The X-axis represents the concentration of antibodies used in this experiment, whereas the Y-axis indicates infectivity, expressed in relative light units (RLU).The assay was performed three times, and standard deviations are presented as error bars.

FIG 9
FIG 9 Evaluation of ADE of ZIKV infection in human PBMCs by anti-WNV antibodies present in the IGIV lots and IGIV samples following enrichment for P2 peptide specificity.One IGIV lot produced in 2007, along with its enriched sample for P2 peptide binding affinity, was tested for ADE of nLuc-ZIKV infection in PBMCs pre-treated without (A) or with M-CSF (B).The X-axis, whose descriptions shown on the left of the graph as lanes 1-5, represents the antibodies used in this experiment, whereas the Y-axis indicates infectivity, expressed in RLU.The assay was performed three times, and standard deviations are presented as error bars.Four asterisks (****) indicate P < 0.01.

FIG 10
FIG 10 Attenuation of ADE in PBMCs by non-anti-WNV antibodies present in the IGIV lots and IGIV flowthrough samples after P2-enrichment.Human PBMCs were pre-treated with macrophage colony-stimulating factor (M-CSF) for 4 days before being used to detect ADE of nLuc-ZIKV infection under the indicated treatments.The X-axis represents combinations of samples used in the experiment, whereas the Y-axis indicates infectivity expressed as a percentage of the positive control, 2007 IGIV 1 P2-enriched.Asterisks (*) and (**) indicate the ratios of 1:3 and 1:10 (wt/wt), respectively, between the P2-enriched sample (labeled as P2 enriched) and either the flowthrough sample (labeled as FT) or the IGIV lot sample.The assay for detecting ADE was performed three times, and standard deviations are presented as error bars.

TABLE 1
Peptides derived from the E protein of Zika virus used in this study

TABLE 2
Peptides derived from EDII of West Nile virus used in this study