A Novel Mechanism for Zika Virus Host-Cell Binding

Zika virus (ZIKV) recently emerged in the Western Hemisphere with previously unrecognized or unreported clinical presentations. Here, we identify two putative binding mechanisms of ancestral and emergent ZIKV strains featuring the envelope (E) protein residue asparagine 154 (ASN154) and viral phosphatidylserine (PS). Synthetic peptides representing the region containing ASN154 from strains PRVABC59 (Puerto Rico 2015) and MR_766 (Uganda 1947) were exposed to neuronal cells and fibroblasts to model ZIKV E protein/cell interactions and bound MDCK or Vero cells and primary neurons significantly. Peptides significantly inhibited Vero cell infectivity by ZIKV strains MR_766 and PRVABC59, indicating that this region represents a putative binding mechanism of ancestral African ZIKV strains and emergent Western Hemisphere strains. Pretreatment of ZIKV strains MR_766 and PRVABC59 with the PS-binding protein annexin V significantly inhibited replication of PRVABC59 but not MR_766, suggesting that Western hemisphere strains may additionally be capable of utilizing PS-mediated entry to infect host cells. These data indicate that the region surrounding E protein ASN154 is capable of binding fibroblasts and primary neuronal cells and that PS-mediated entry may be a secondary mechanism for infectivity utilized by Western Hemisphere strains.

Changes in clinical presentation during ZIKV infection from strains in the Asian/American lineage relative to the ancestral African lineage are likely the result of divergent nucleotide and amino acid sites conferring new phenotypes [36][37][38][39], and the relative contributions of various changes are still being explored. The structure of the ZIKV strain H/PF/2013 (Asian/American lineage) was resolved by cryo-electron microscopy and illustrated both unique features of this lineage and commonalities to other flaviviruses. A key difference between ZIKV lineages predicted computationally and detected structurally is the modification of asparagine at position 154 (ASN154) of the envelope protein (E) with N-acetylglucosamine (NAG). This modification is also seen across many flaviviruses but is often variable across strains [40]. Reports have implicated the ASN154 counterpart in host cell interactions by the four dengue virus serotypes [41] and in neuroinvasion of West Nile virus and St. Louis encephalitis virus [42,43]. Additionally, a recent study by Yuan et al. demonstrated that a single amino acid substitution within the PrM protein of Western Hemisphere strains conferred increased virulence and resulted in exacerbated pathology in vivo [33]. While this change confers an increased capacity for cell death and correlates with the clinical findings suggesting more severe and invasive disease, it cannot explain the newly emerged ability to directly invade the central nervous system (CNS). We sought to build upon our recent informatics analysis [36] by utilizing the findings to identify additional binding mechanisms of ZIKV.

Virus Isolates and Culture Conditions
African Green monkey kidney (Vero) cells and Madin-Darby canine kidney (MDCK) cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA). Cells were routinely propagated in Earle's Minimum Essential Medium (EMEM) with Earle's Balanced Salt Solution (BD Biosciences, San Jose, CA), supplemented with 10% fetal bovine serum, L-glutamine and Penicillin/Streptomycin. Cell cultures were incubated at 37 • C, with 5% CO 2 and a relative humidity (RH) of 90%. Low-passage isolates of ZIKV strains MR_766 (ATCC VR-84, Uganda) and PRVABC59 (ATCC VR-1843, Puerto Rico) were obtained from the American Type Culture Collection. Virus stocks were propagated on monolayers of Vero cells. Harvested virus lysates were clarified by low-speed centrifugation (500× g/10 min) and stored in 1-ml aliquots at −80 • C.

Protein Analysis and Peptide Design
The Envelope protein structure was visualized using Jmol via the Protein Data Bank (PDB ID 5JHM) [44,45], and the PDB Ligand Explorer was used to visualize the structure of NAG on ASN154. Probabilities of protein disorder at each amino acid site was estimated using PrDOS [46]. This analysis indicated that the region surrounding ASN154 constitutes a highly disordered linear epitope. Synthetic peptides representing this linear epitope including the differentially glycosylated ASN154 were generated (see Table 1) by Bachem (Bubendorf, Switzerland). The aminoterminal and carboxyterminal domains from PRVABC59 were also synthesized. Peptides were modified by the addition of an aminoterminal FITC label to allow detection and visualization.

Peptide Binding Assays
Vero cells and MDCK cells, both of which are permissive for all Zika strains, were grown to 80% confluency in 48-well plates. Following the removal of medium, wells were blocked with 10% fetal bovine serum for 30 min at 37 • C. Peptides (0.2 µmol) were incubated with Vero or MDCK cells for 1 hour at 37 • C. Unbound peptides were removed by washing with 1× PBS, and mammalian cells were counterstained with 4 ,6-diamidino-2-phenylindole (DAPI, diluted 1:300) to control for minor variations in monolayer populations. Bound peptides (FITC) were detected at 485/490 (excitation/emission), and cells were quantified at 350/460. Data are presented as FITC:DAPI ratios to correct for minor variance in cell numbers between wells. Statistical significance was measured by analyses of variance, and by Fisher's Protected Least Significant Difference test for posthoc comparisons when main effects were significant (GraphPad Prism v. 6.0).
Primary DRG neurons grown on coverslips were incubated with 100 µg peptide for 1 hour at 37 • C to qualitatively assess ZVBM binding potential. Unbound peptides were removed by washing with HBSS, and DRG neurons were counterstained with DAPI (diluted 1:300). Bound peptides were visualized using Keyence BZX-700 inverted widefield digital microscope. Binding was evaluated qualitatively to avoid ambiguities from the potential for greater or lower affinities for certain peptides by heterogeneous populations of neurons that may vary slightly from well to well, and from the formation of cellular islands as opposed to confluent monolayers. To avoid ambiguities with the impact of glycosylation, peptides PRV, MR, and PRV Scr were used.

Viral Inhibition Assays by ZVBM Peptides
Vero cells were propagated in 48-well plates (seed concentration-1e5 cells/well) for 24-h. Resulting monolayers (85% confluence) were washed twice with warmed PBS and incubated for 2 hours (37 • C, 5% CO 2 , 90% RH) with 0.1 ml volumes of either PBS (Negative Control), or PBS containing 0.8 µmol of the selected ZVBM peptide. Following treatment, PBS or peptide was decanted, and monolayers washed twice with warmed PBS. ZIKV stocks (stock concentrations: Strain MR_766-10 7 TCID 50 /ml; Strain PRVABC59-10 7 TCID 50 /mL) were serially diluted in serum-free Dulbecco's Minimum Essential Medium (DMEM). Host cell monolayers in treated, or untreated plates were inoculated with either strain MR_766 or PRVABC59 (0.1 mL/well, 5 wells per dilution [MOI = 1], N = 3 replicates each) and incubated for 2 hr. After inocula were removed, wells were supplemented with 0.5 ml EMEM growth medium and returned to the incubator. Virus cytopathogenic effects (CPEs) were monitored and scored over a period of 10-12 days and the resulting virus titers calculated as TCID 50 /mL. Statistical significance of changes in virus titer as a result of peptide pretreatment versus untreated control was measured by Student's T-test (GraphPad Prism v. 6.0).

Virus Treatment with Annexin V
Annexin V (AbCam, Cambridge, MA) was dissolved in PBS (2335 µg/mL) and filter-sterilized (0.2 µm). ZIKV stocks MR_766 and PRVABC59 were then serially diluted in either PBS, or PBS-Annexin V. Dilutions were incubated for 2 hours (37 • C, 5% CO 2 , 90% RH). Host cell monolayers, prepared in 48-well plates as previously described, were inoculated with respective virus dilutions (0.1 mL/well, 5 wells per dilution [MOI = 1], n = 3 replicates each). Virus CPE were scored over a period of 10-12 days and the resulting virus titers calculated as TCID 50 /ml. Statistical significance of changes in virus titer as a result of Annexin V pretreatment versus untreated control was measured by student's T test (GraphPad Prism v. 6.0) for each ZIKV strain.

Binding Motif Prediction
Structural modeling predictions of strain PRVABC59 (Puerto Rico, 2015) indicated that asparagine 154 (ASN154) is part of a linear β strand ( Figure 1A). The disorder probability of this region peaks  Figure 1B), suggesting that this portion of the E protein is particularly dynamic and flexible. Structural and disorder probability predictions of the African type strain MR_766 (Uganda 1947) exhibit similar characteristics ( Figure 1C). This region was termed the (putative) Zika virus binding motif (ZVBM). ZVBM sequences from strains PRVABC59 and MR_766 were synthesized and N-terminally labelled with fluorescein isothiocyanate (FITC) in order to assess their capacity to bind ZIKV-susceptible and -permissive cell lines, disrupt ZIKV adsorption, and to interact with dorsal root ganglia (DRG) neurons ex vivo. The PRVABC59 sequence was used to generate a peptide that was modified with an N-acetyl glucosamine (NAG) molecule at position 8 (equivalent to ASN154), as it natively occurs in this strain, and without carbohydrate modification (Table 1). Structural modeling predictions of strain PRVABC59 (Puerto Rico, 2015) indicated that asparagine 154 (ASN154) is part of a linear β strand (Figure1A). The disorder probability of this region peaks at 0.72 ( Figure 1B), suggesting that this portion of the E protein is particularly dynamic and flexible. Structural and disorder probability predictions of the African type strain MR_766 (Uganda 1947) exhibit similar characteristics ( Figure 1C). This region was termed the (putative) Zika virus binding motif (ZVBM). ZVBM sequences from strains PRVABC59 and MR_766 were synthesized and N-terminally labelled with fluorescein isothiocyanate (FITC) in order to assess their capacity to bind ZIKV-susceptible and -permissive cell lines, disrupt ZIKV adsorption, and to interact with dorsal root ganglia (DRG) neurons ex vivo. The PRVABC59 sequence was used to generate a peptide that was modified with an N-acetyl glucosamine (NAG) molecule at position 8 (equivalent to ASN154), as it natively occurs in this strain, and without carbohydrate modification (Table 1).

ZVBM Binding to ZIKV Replication-Permissive Fibroblast Lines
ZVBM peptides from strains MR_766 ("MR") and PRVABC59 (NAGylated and unglycosylated, "PRV" and "PRV-N") all bound Vero cells at levels significantly above those of scrambled (NAGylated and unglycosylated, "PRV Scr " and "PRV-N Scr ") controls (Figure 2A). Unglycosylated ZVBM peptides MR and PRV both bound Madin-Darby canine kidney (MDCK) cells significantly (p < 0.05) above scrambled controls PRV Scr and PRV-N Scr , though the signal generated by PRV was significantly higher than MR. Interestingly, NAGylated PRV-N did not bind MDCK cells above the scrambled controls, ( Figure 2B). Two-fold dilutions of adherent ZVBM peptides resulted in proportional reductions in FITC signal, indicating that titration of binding to both Vero and MDCK cells is readily achieved (Figure 2C,D).
proportional reductions in FITC signal, indicating that titration of binding to both Vero and MDCK cells is readily achieved (Figure 2C,D).

ZVBM Binding to Primary Neuronal Cells Ex Vivo
We collected dorsal root ganglia (DRG) from C57/black mice and cultured DRG neurons on coverslips to qualitatively assess interactions with ZVBM peptides. Peptides PRV and MR were visualized in association with 24-hour DRG neuron cultures by fluorescence microscopy ( Figure 3A-B). Cell association was not detected for the scrambled PRVABC59 control peptide ( Figure 3C).

ZVBM Binding to Primary Neuronal Cells Ex Vivo
We collected dorsal root ganglia (DRG) from C57/black mice and cultured DRG neurons on coverslips to qualitatively assess interactions with ZVBM peptides. Peptides PRV and MR were visualized in association with 24-hour DRG neuron cultures by fluorescence microscopy ( Figure 3A-B). Cell association was not detected for the scrambled PRVABC59 control peptide ( Figure 3C).

ZVBM Binding to Primary Neuronal Cells Ex Vivo
We collected dorsal root ganglia (DRG) from C57/black mice and cultured DRG neurons on coverslips to qualitatively assess interactions with ZVBM peptides. Peptides PRV and MR were visualized in association with 24-hour DRG neuron cultures by fluorescence microscopy ( Figure 3A-B). Cell association was not detected for the scrambled PRVABC59 control peptide ( Figure 3C).

Refinement of ZVBM Functional Elements
Peptides representing the NTD and the CTD of strain PRVABC59 s ZVBM sequence ("PRV-NTD" and "PRV-CTD") were synthesized and N-terminally labelled with FITC ( Figure 4). To avoid ambiguities with the impact of glycosylation, neither peptide was NAGylated, and Vero cells were used to assess binding. PRV-NTD was unable to bind Vero cells above the scrambled control PRV Scr , whereas PRV-CTD bound significantly (p < 0.05) above PRV Scr . Additionally, PRV-CTD bound at equivalent levels to full-length PRV and to MR.
Viruses 2019, 11, x FOR PEER REVIEW 6 of 13 PRVABC59. Punctate green staining around the DRG nuclei was observed in panels A and B, but was largely absent from panel C.

Refinement of ZVBM Functional Elements
Peptides representing the NTD and the CTD of strain PRVABC59′s ZVBM sequence ("PRV-NTD" and "PRV-CTD") were synthesized and N-terminally labelled with FITC. To avoid ambiguities with the impact of glycosylation, neither peptide was NAGylated, and Vero cells were used to assess binding. PRV-NTD was unable to bind Vero cells above the scrambled control PRV Scr , whereas PRV-CTD bound significantly (p < 0.05) above PRV Scr . Additionally, PRV-CTD bound at equivalent levels to full-length PRV and to MR.

Disruption of ZIKV Infection and CPE Generation
Host cell monolayers were pre-treated with PRV, PRV-N, MR, and scrambled controls PRV Scr and PRV-N Scr . To avoid ambiguities from the impact of glycosylation, Vero cells were used to for this

Disruption of ZIKV Infection and CPE Generation
Host cell monolayers were pre-treated with PRV, PRV-N, MR, and scrambled controls PRV Scr and PRV-N Scr . To avoid ambiguities from the impact of glycosylation, Vero cells were used to for this analysis. All peptides were used to inhibit both strains MR_766 and PRVABC59. Pretreatment of Vero cell monolayers with PRV, PRV-N, and MR significantly (p < 0.01) inhibited infectivity and CPE generation by both ZIKV MR_766 and ZIKV PRVABC59 relative to scrambled control peptides and Vero cells pretreated with PBS alone ( Figure 5A). MDCK cells are still permissive for ZIKV replication [48], suggesting that additional mechanisms facilitate infection when ZVBM is NAGylated. We pretreated ZIKV strains MR_766 and PRVABC59 with the PS-binding protein annexin V prior to infection of Vero cell monolayers. Annexin V significantly (p < 0.05) inhibited infectivity of PRVABC59 relative to untreated controls but did not inhibit MR_766 ( Figure 5B). analysis. All peptides were used to inhibit both strains MR_766 and PRVABC59. Pretreatment of Vero cell monolayers with PRV, PRV-N, and MR significantly (p < 0.01) inhibited infectivity and CPE generation by both ZIKV MR_766 and ZIKV PRVABC59 relative to scrambled control peptides and Vero cells pretreated with PBS alone ( Figure 5A). MDCK cells are still permissive for ZIKV replication [48], suggesting that additional mechanisms facilitate infection when ZVBM is NAGylated. We pretreated ZIKV strains MR_766 and PRVABC59 with the PS-binding protein annexin V prior to infection of Vero cell monolayers. Annexin V significantly (p < 0.05) inhibited infectivity of PRVABC59 relative to untreated controls but did not inhibit MR_766 ( Figure 5B).

Discussion
The change in clinical spectrum of ZIKV disease caused by Asian/American lineage strains suggests fundamental changes in key protein functions relative to African lineage strains. Comparative analyses between strains of both lineages identified areas of interest with potential to contribute to the changes in clinical presentations, including ASN154 [36,40]. ASN154 is part of an intrinsically disordered region of the E protein. Disordered regions lack fixed tertiary structures, which affords them flexibility in spatial arrangement. The array of structures expands further if the disordered regions are post-translationally modified, including via glycosylation [49]. The diversity that this provides in terms of function is increasingly reported for many organisms, including flaviviruses [50,51]. The region surrounding ASN154 was thus an attractive target for mechanistic evaluation to explore the changes in clinical presentation.
ZVBM peptides MR, PRV, and PRV-N all bound Vero cells at levels significantly above those of scrambled controls (Figure 2A), and peptides MR and PRV bound MDCK cells significantly above scrambled controls ( Figure 2B). Interestingly, NAGylated PRV-N did not bind MDCK cells above the scrambled controls, (Figure 2B). Two-fold dilutions of adherent ZVBM peptides resulted in proportional reductions in FITC signal, indicating that titration of binding to both Vero and MDCK cells is readily achieved (Figure 2C, 2D). These findings suggest that this motif binds to fibroblasts and has the potential to mediate ZIKV infectivity, but the difference in avidity between PRV and MR and the inability of NAGylated PRV-N to bind MDCK cells indicates that the functionality of ZVBM may be host cell and strain dependent. Pre-treatment of host cell monolayers with PRV, PRV-N, MR significantly inhibited infectivity and CPE generation by both ZIKV MR_766 and ZIKV PRVABC59 relative to scrambled control peptides and Vero cells pretreated with PBS alone (Figure 5A),

Discussion
The change in clinical spectrum of ZIKV disease caused by Asian/American lineage strains suggests fundamental changes in key protein functions relative to African lineage strains. Comparative analyses between strains of both lineages identified areas of interest with potential to contribute to the changes in clinical presentations, including ASN154 [36,40]. ASN154 is part of an intrinsically disordered region of the E protein. Disordered regions lack fixed tertiary structures, which affords them flexibility in spatial arrangement. The array of structures expands further if the disordered regions are post-translationally modified, including via glycosylation [49]. The diversity that this provides in terms of function is increasingly reported for many organisms, including flaviviruses [50,51]. The region surrounding ASN154 was thus an attractive target for mechanistic evaluation to explore the changes in clinical presentation.
ZVBM peptides MR, PRV, and PRV-N all bound Vero cells at levels significantly above those of scrambled controls (Figure 2A), and peptides MR and PRV bound MDCK cells significantly above scrambled controls ( Figure 2B). Interestingly, NAGylated PRV-N did not bind MDCK cells above the scrambled controls, ( Figure 2B). Two-fold dilutions of adherent ZVBM peptides resulted in proportional reductions in FITC signal, indicating that titration of binding to both Vero and MDCK cells is readily achieved (Figure 2C,D). These findings suggest that this motif binds to fibroblasts and has the potential to mediate ZIKV infectivity, but the difference in avidity between PRV and MR and the inability of NAGylated PRV-N to bind MDCK cells indicates that the functionality of ZVBM may be host cell and strain dependent. Pre-treatment of host cell monolayers with PRV, PRV-N, MR significantly inhibited infectivity and CPE generation by both ZIKV MR_766 and ZIKV PRVABC59 relative to scrambled control peptides and Vero cells pretreated with PBS alone (Figure 5A), suggesting that ZIKV and ZVBM target the same host cell receptor. These findings demonstrate that adherence of ZVBM peptides to Vero cells has functional relevance, and that this motif likely mediates at least some host cell infection.
Binding of the MR_766 ZVBM peptides to Vero cells and MDCK cells, despite a four-amino acid deletion relative to PRVABC59, suggests that the critical portion of the ZVBM is potentially contained entirely on the aminoterminus (NTD) or the carboxyterminus (CTD). PRV-NTD was unable to bind Vero cells, whereas PRV-CTD bound significantly (p < 0.05) above scrambled controls. Consistent with the notion that binding is facilitated by either the NTD or the CTD, PRV-CTD bound at equivalent levels to full-length PRV and to MR indicating that the functional element of ZVBM that mediates binding is contained entirely on the CTD (Figure 4). Interestingly, PRV-CTD does not contain ASN154, suggesting that any impacts of this residue on host cell binding and/or specificity stem from its proximity to the critical binding mediators rather than its direct involvement.
Given that MDCK cells are still permissive for ZIKV replication [48], we hypothesized that a more generalized mechanism may be contributing to viral adsorption to some strains. The association of human AXL with ZIKV adsorption [27][28][29] suggests that viral PS may facilitate entry into certain host cells by binding Gas6, which in turn binds AXL, as is seen with multiple viruses [52][53][54][55][56]. Pretreatment ZIKV strain PRVABC59 with the PS-binding protein annexin V prior to infection of Vero cell monolayers significantly inhibited infectivity relative to untreated controls. Conversely, pretreatment of ZIKV strain MR_766 with annexin V did not impact infectivity ( Figure 5B). This finding suggests that PS-mediated ZIKV adsorption is possible for at least some strains, although a relationship between PS-mediated infectivity and ZVBM-NAGylation is unclear, and it is plausible that strains differing in maturation kinetics will differ in membrane PS concentration. While at least one additional mechanism has been described for the greater infectivity of Asian and American strains [33], PS-mediated host cell entry has high potential to contribute to this phenotype as well. Additionally, these findings also support previous studies that both implicate AXL as a host cell receptor for Asian/American ZIKV strains and those that show genetically ablated animals are still susceptible to infection by establishing two distinct binding mechanisms for this clade [27][28][29][57][58][59].
CNS disease or infectivity with MR_766 following intrathecal or intracerebral inoculation in vivo, or neuronal cell infectivity in vitro, has been reported [60][61][62][63][64]. Observation of the MR peptide binding to DRG neurons is consistent with these findings; however, both stand in conflict with a lack of evidence for CNS involvement during human disease caused by African ZIKV strains. We hypothesized that exposure to neuronal cells ex vivo would result in MR_766 ZVBM peptide binding, and the lack of neurological complications during Zika virus disease caused by African strains stems from an inability of these strains to penetrate into the CNS. These findings were consistent with those of Annamalai et al., who demonstrated a lack of neurological disease with strains lacking NAGylation at ASN154 when injected intravenously, and overt disease when the same strain was injected intracranially [65]. This suggests that the ability of Asian/American lineage strains to cause disease in the CNS is not exclusively due to a novel binding mechanism but is more likely due to a newfound ability to access the CNS and other privileged body sites. The ZVBM peptide MR binding to neuronal cells ex vivo is consistent with this suggestion. While additional mechanisms facilitate novel interactions between Asian/American lineage ZIKV strains and neuronal cells [33], ZVBM-mediated binding to neuronal cells is not unique to this lineage. N-linked glycosylation of E in other flaviviruses is proposed to facilitate blood-brain barrier (BBB) transit by allowing interaction with DC-SIGN on microvascular endothelial cells followed by transcytosis [66], and ASN154 is notably absent from strain MR_766. Though not required for direct binding to neuronal cells, ASN154 may very well facilitate BBB passage in Asian/American lineage strains. In vivo inhibition of Asian/American ZIKV BBB transit by peptides would demonstrate this.
The outcomes of our in vitro and ex vivo studies suggest that the disordered region of the ZIKV E protein surrounding ASN154, termed ZVBM, binds fibroblasts and primary neuronal cells, indicating its potential to contribute to host cell infection. The carboxyterminal portion of the ZVBM (i.e., ENRAKV) is capable of binding fibroblasts at the same levels as full-length ZVBM, suggesting that it mediates the binding. Despite retaining the sequence ENRAKV, glycosylated peptides have differing abilities to bind MDCK cells and ZIKV strains lacking ASN154 feature different tissue tropism, suggesting that glycosylation of this site could either sterically hinder binding to some cell types, provide proximity to other cell types by facilitating entry into the privileged body sites, or both. This model is consistent with both the clinical disparity between ZIKV lineages and the generation of neurological disease by the African strain MR_766, which lacks ASN154, when introduced directly into the CNS as previously described [61,65]. These findings demonstrate the impact of NAGylation of a pathogen surface protein in the vicinity of a binding motif on its potential host cell targets. This change in post-translational modification can therefore instantly alter the potential target tissues of infectious agents and can be expected to similarly alter the array of clinical presentations they cause in turn. This concept has widespread implications for parallel changes in other pathogens, and therefore applies broadly to the field of emerging infectious diseases.