Third Helical Domain of the Nipah Virus Fusion Glycoprotein Modulates both Early and Late Steps in the Membrane Fusion Cascade

The Paramyxoviridae family includes important human and animal pathogens, such as measles, mumps, and parainfluenza viruses and the deadly henipaviruses Nipah (NiV) and Hendra (HeV) viruses. Paramyxoviruses infect the respiratory tract and the central nervous system (CNS) and can be highly infectious. Most paramyxoviruses have a limited host range. However, the biosafety level 4 NiV and HeV are highly pathogenic and have a wide mammalian host range. Nipah viral infections result in acute respiratory syndrome and severe encephalitis in humans, leading to 40 to 100% mortality rates. The lack of licensed vaccines or therapeutic approaches against NiV and other important paramyxoviruses underscores the need to understand viral entry mechanisms. In this study, we uncovered a novel role of a third helical region (HR3) of the NiV fusion glycoprotein in the membrane fusion process that leads to viral entry. This discovery sets HR3 as a new candidate target for antiviral strategies for NiV and likely for related viruses.

ABSTRACT

Medically important paramyxoviruses, such as measles, mumps, parainfluenza, Nipah, and Hendra viruses, infect host cells by directing fusion of the viral and cellular plasma membranes. Upon infection, paramyxoviruses cause a second type of membrane fusion, cell-cell fusion (syncytium formation), which is linked to pathogenicity. Host cell receptor binding causes conformational changes in the attachment glycoprotein (HN, H, or G) that trigger a conformational cascade in the fusion (F) glycoprotein that mediates membrane fusion. F, a class I fusion protein, contains the archetypal heptad repeat regions 1 (HR1) and 2 (HR2). It is well established that binding of HR1 and HR2 is key to fusing viral and cellular membranes. In this study, we uncovered a novel fusion-modulatory role of a third structurally conserved helical region (HR3) in F. Based on its location within the F structure, and structural differences between its prefusion and postfusion conformations, we hypothesized that the HR3 modulates triggering of the F conformational cascade (still requiring G). We used the deadly Nipah virus (NiV) as an important paramyxoviral model to perform alanine scan mutagenesis and a series of multidisciplinary structural/functional analyses that dissect the various states of the membrane fusion cascade. Remarkably, we found that specific residues within the HR3 modulate not only early F-triggering but also late extensive fusion pore expansion steps in the membrane fusion cascade. Our results characterize these novel fusion-modulatory roles of the F HR3, improving our understanding of the membrane fusion process for NiV and likely for the related Henipavirus genus and possibly Paramyxoviridae family members.

IMPORTANCE The Paramyxoviridae family includes important human and animal pathogens, such as measles, mumps, and parainfluenza viruses and the deadly henipaviruses Nipah (NiV) and Hendra (HeV) viruses. Paramyxoviruses infect the respiratory tract and the central nervous system (CNS) and can be highly infectious. Most paramyxoviruses have a limited host range. However, the biosafety level 4 NiV and HeV are highly pathogenic and have a wide mammalian host range. Nipah viral infections result in acute respiratory syndrome and severe encephalitis in humans, leading to 40 to 100% mortality rates. The lack of licensed vaccines or therapeutic approaches against NiV and other important paramyxoviruses underscores the need to understand viral entry mechanisms. In this study, we uncovered a novel role of a third helical region (HR3) of the NiV fusion glycoprotein in the membrane fusion process that leads to viral entry. This discovery sets HR3 as a new candidate target for antiviral strategies for NiV and likely for related viruses.

INTRODUCTION

The Paramyxoviridae family contains viruses important to human and animal health, such as measles (MeV), mumps (MuV), parainfluenza, and canine distemper viruses, avian paramyxovirus (also known as Newcastle disease virus), and the zoonotic and deadly henipaviruses Nipah (NiV) and Hendra (HeV) viruses. Henipaviruses are unique among the paramyxoviruses in that henipaviruses can infect a large repertoire of mammalian hosts. Henipaviruses include NiV, HeV, Cedar virus, Kumasi virus, Mojiang virus, and nearly 20 new henipaviruses recently discovered by recent fruit bat population surveillance and sampling (1, 2). Between 1998 and 2019, basically yearly contained outbreaks of NiV have occurred in Southeast Asia, particularly in Bangladesh. Other countries with NiV outbreaks include Malaysia, Singapore, and, more recently, the Philippines and the Kozhikode district of Kerala, India (2). Importantly, NiV outbreaks have had a high mortality rate in humans, ranging from 40 to 100% (2). NiV infections result in severe respiratory syndrome, encephalitis, vasculitis, and virally induced syncytium formation (multinucleated cells) via cell-cell fusion (3–5). Given the distribution and pathogenesis of the paramyxoviruses, and the lack of approved vaccines or therapeutic approaches for many of them, it is imperative to understand the mechanisms of viral entry (viral-cell membrane fusion) and syncytium formation (cell-cell membrane fusion) mediated by the viral glycoproteins. Such understanding may help in the design of therapeutic approaches against these viruses.

Paramyxoviral entry, infection, and formation of the pathognomonic syncytia characteristic of paramyxoviral infections rely on the cooperation between the two surface glycoproteins: the receptor-binding attachment glycoprotein (HN, H, or G depending on the virus genus) and the fusion glycoprotein (F). The attachment glycoprotein can bind a sialic acid receptor (HN) or a proteinaceous receptor (H or G). F is a class I fusion protein, containing the characteristic trimeric alpha-helical heptad repeat 1 and 2 regions (HR1 and HR2, also known as HRA and HRB or HRN and HRC, respectively). F requires proteolytic processing prior to becoming a fusion-primed glycoprotein. In the case of NiV and HeV, F is initially synthesized as an inactive trimeric precursor (F₀) that is transported to the cell surface and then endocytosed and cleaved in early endosomal compartments by cathepsin L or B, depending on the cell type (6–9). The resulting fusion-primed protein contains two subunits, the N-terminal F₂ and the C-terminal F₁ subunits, linked by a disulfide bond. Upon protease processing, the mature F₁ subunit displays a hydrophobic fusion peptide (FP) at its new N terminus (Fig. 1C). This mature processed F protein is primed for proper spatiotemporal triggering by the receptor-binding attachment glycoprotein. Upon receptor binding, HN/H/G undergoes recently discovered conformational changes that in turn trigger major conformational changes in the mature F (10–14). However, many steps between receptor binding and the completion of the late steps of the membrane fusion process remain fairly elusive.

FIG 1.

NiV-F structural analysis and HR3 region mutagenesis. (A) Depiction of published prefusion NiV-F crystal structure (PDB accession code 5EVM) showing HR1 in blue, HR2 in orange, HR3 in magenta, and the fusion peptide (FP) in black. A homology-modeled structure of postfusion NiV-F was generated from the existing X-ray crystal structure of hPIV3 using SWISS-MODEL (right). Both NiV-F monomer and trimeric forms are depicted. (B) Expansion of insets of the HR3 region in panel A. (C) Schematic representation of full-length NiV-F protein, and scanning alanine mutant sequences for amino acids 84 to 104, with the exception of position 93, which naturally contains an alanine. TM and CT, transmembrane and cytoplasmic tail domains, respectively.

After receiving the F-triggering signal from the attachment glycoprotein, F ultimately executes the viral-cell or cell-cell membrane fusion processes by undergoing several conformational changes. These changes span from the relatively metastable prefusion (PF) conformation to the transient prehairpin intermediate (PHI) conformation to the final, energetically favorable postfusion 6-helix-bundle conformation (6HB) (15–19). The classical HR1 and HR2 regions and their roles in the formation of the 6HB have been studied extensively for many class I fusion proteins and are widely accepted (20). Recently, we collaboratively solved the NiV fusion protein (NiV-F) prefusion crystal structure (21). Analysis of this structure revealed a third heptad repeat helical domain that we designate HR3, in close physical association with the HR1 and fusion peptide regions. In the F prefusion conformation, HR3 is composed of a helical region ending with a beta-strand (Fig. 1A and B, left). However, in a homology-modeled postfusion conformation of NiV-F, the HR3 region is far removed from the fusion peptide, and several of the amino acids that were previously in a beta-strand configuration now form an additional alpha-helical turn (Fig. 1A and B, right). Due to changes in its proximity to the fusion peptide and the observed secondary structural changes, we hypothesized that the HR3 region may modulate the membrane fusion process, particularly at the F-triggering step.

Here, we present a comprehensive study of the HR3 region of NiV-F, yielding mutants that not only affected early F-triggering steps in the membrane fusion cascade, as we hypothesized, but also affected a late fusion pore expansion step, still requiring the presence and the primary triggering signal of NiV-G. Our results identify and characterize two novel fusion-modulatory roles for HR3 in the henipaviral F protein. Based on sequence homology between the NiV HR3 and the HR3 domains of other paramyxoviruses, these findings may have implications for membrane fusion of other members of the Henipavirus genus and the paramyxoviral family.

RESULTS

Mutational analysis of the NiV-F HR3 domain.

In a collaborative study we reported the NiV-F prefusion structure via X-ray crystallography (PDB accession code 5EVM) (Fig. 1A) (21). It is well established that the trimeric class I fusion proteins contain two canonical helical heptad-repeat regions (HR1, in blue, and HR2, in orange) that upon F-triggering bind to each other to form a six-helix bundle, mediating membrane fusion and viral entry. Notably, the structure of the prefusion NiV-F monomer revealed a third helical region, HR3, in magenta, in close proximity with the fusion peptide, in black, and with a portion of the HR1 region (Fig. 1A). In the prefusion conformation, the HR3 domain includes an alpha-helical region that spans amino acids 74 to 99 and a small beta-strand and linker that span amino acids 100 to 104. The HR3 beta-strand forms a short antiparallel beta-sheet with the beta-strand of the fusion peptide (in black in Fig. 1A and B).

Although the postfusion NiV-F structure remains unsolved, the X-ray crystal structure of the related paramyxovirus human parainfluenza virus 3 fusion protein (hPIV3-F) in the postfusion conformation has been determined (22). NiV-F and hPIV3-F structures share ∼30% sequence identity. We thus generated a homology-modeled postfusion NiV-F structure using the SWISS-MODEL program based on the existing postfusion hPIV3-F structure (PDB accession code 1ZTM) (21–26), showing analogous secondary structural architecture (Fig. 1A). Interestingly, the modeled NiV-F postfusion conformation shows that the beta-strand of the HR3 region present in the prefusion conformation is replaced by an extra alpha-helix turn (Fig. 1B). Furthermore, the postfusion HR3 is no longer in close association with the fusion peptide and is now in the inverted orientation (Fig. 1A).

Thus, we hypothesized that the HR3 domain would be important for modulating the release of the fusion peptide and HR1 region, facilitating the transition from prefusion to postfusion F. Importantly, the relatively high level of conservation of the HR3 sequence among henipaviruses and paramyxoviruses (see Discussion and Fig. 8 for details) suggests that if our hypothesis is correct, the potential role(s) of the HR3 region may expand beyond NiV to other henipaviruses and perhaps to other paramyxoviruses. To elucidate the NiV HR3 structure-function relationship, we performed site-directed mutagenesis. Twenty scanning single-point alanine mutations were introduced in the region of the HR3 domain closest to the fusion peptide: residues Asn 84 to Leu 104 (Fig. 1C). Notably, an Ala naturally occurs at position 93, so no mutant was generated at this position (wild-type NiV-F). Additionally, we previously reported that there is an N-linked glycan (termed F3) effectively added N99 of the NNT glycosylation motif at residues 99 to 101; thus, Ala substitutions at N99 and T101 should abrogate glycosylation at this site (27).

FIG 8.

HR3 sequence and structural homology. Shown is sequence alignment of the portion of the HR3 domain in close association with the FP and HR1 regions between NiV and other henipaviruses (A) or between NiV and other paramyxoviruses (B). The most homologous residues are colored, and the percent identities are shown. Identical, very similar, and similar residue side chains are represented by asterisks, colons, and periods, respectively. PIV5-F was from the W3 strain, and MeV-F was from the Ichinose-B95a strain. NDV, Newcastle disease virus; CDV, canine distemper virus. (C) Structural homology of the HR3 among paramyxoviruses. Shown are prefusion NiV-F, HeV-F, MeV-F, and PIV5-F structures with the HR1 domain in blue, the HR2 domain in orange, the FP domain in black, and the entire HR3 domain in magenta (PDB accession codes 5EVM, 5EJB, 5YXW, and 4WSG).

Mutants in the HR3 region affect fusion.

We then searched for fusion mutants, defined as mutants that significantly increased or decreased the levels of cell-cell fusion without drastically affecting the levels of glycoprotein cell surface expression (CSE), as glycoprotein CSE levels have been shown to affect cell-cell fusion levels in our prior work with the NiV glycoproteins (Fig. 2A) (27). Thus, we transfected 293T cells with NiV-G and NiV-F expression plasmids at a 1:1 ratio as previously described and evaluated them for syncytium formation, taking into account F CSE levels (Fig. 2A) (27–30). CSE levels were derived from mean fluorescence intensity (MFI) values after using our polyclonal 834/835 anti-F rabbit serum in a flow cytometric assay. Levels of fusion were determined by counting the number of nuclei inside syncytia or fused cells, considering syncytia as cell compartments containing 4 or more nuclei, to fully discard cell division events. A total of five fields (at ×200) per sample were counted. Both cell-cell fusion and CSE levels were normalized to those of wild-type NiV-F (syncytium formation and CSE levels set to 100%) (Fig. 2A). Then, a relative fusion index of the NiV-F HR3 mutants was determined as the ratio of normalized syncytium formation (relative number of fused cells) to normalized F cell surface expression levels (Fig. 2B) (27, 31). This yields a wild-type NiV-F fusion index (FI) of 1, with hyperfusogenic mutants having an FI of >1 and hypofusogenic mutants an FI of <1. I86A, I90A, I96A, Y97A, N99A, and T101A mutants, most of which are labeled in red text in Fig. 2A and B, yielded statistically significantly higher fusion indexes and were termed hyperfusogenic (Fig. 2B). Notably, the I86A mutant yielded a significantly higher fusion index score (Fig. 2B); however, due to poor cell surface expression levels (Fig. 2A) and total cell expression levels (Fig. 3A), we did not include this mutant in subsequent functional analyses of the fusion mutants. Examples of hyperfusogenic and hypofusogenic syncytium formation images are shown in Fig. 2C. In contrast, the K98A and L104A mutants had wild-type levels of CSE or slightly increased levels, respectively, yet reduced or completely abrogated syncytium levels, yielding fusion indexes statistically significantly decreased compared to that of wild-type NiV-F (Fig. 2B and C). Asterisks depict levels of statistical significance (with P values between <0.05 and <0.001). These hypofusogenic mutants are marked in blue text. To assess whether hypofusogenic mutants may display reduced syncytium formation due to a less effective membrane fusion process, as opposed to a too-rapid and thus nonproductive refolding to the postfusion conformation at 37°C, we performed syncytial assays at the reduced temperature of 32°C. We observed that the mutants still exhibited similarly reduced levels of syncytium formation at 32°C relative to wild-type NiV-F compared to those at 37°C (Fig. 2D). These data suggest that the hypofusogenic mutants are less effective at executing membrane fusion, as opposed to executing too-rapid and thus nonproductive refolding to the postfusion conformation. Only mutants that had significantly higher or lower fusion indices were further analyzed in our subsequent functional assays, described below.

FIG 2.

Several mutants within the HR3 region affect the levels of cell-cell fusion observed. (A) Relative levels of cell surface expression (CSE) measured using polyclonal antibody 834 or 835 by flow cytometry. Mean fluorescence intensity (MFI) was normalized to that of wild-type (WT) NiV-F. Levels of cell-cell fusion were quantified by counting nuclei inside syncytia and normalizing to the levels induced by wild-type NiV-F. Wild-type NiV-F had between 50 and 100 nuclei counts per ×200 field. Averages ± standard errors of the means are shown. (B) Fusion indexes were calculated by dividing percent normalized syncytia by percent normalized CSE. Data are averages ± standard errors. One-sample Student t tests were used to determine statistical significance in fusion indices (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001). Green lines depict WT F levels (C) Examples of light microscope images of wild-type NiV-F, the hyperfusogenic Y97A mutant, and the hypofusogenic L104A mutant at a magnification of ×200. (D) Quantification of syncytium formation for the K98A and L104A mutants performed at 32°C versus 37°C, normalized to wild-type syncytium levels.

FIG 3.

NiV-F HR3 mutant fusion phenotypes are not accounted for by changes in F processing or conformational states. (A) Western blot analysis of wild-type and HR3 mutant NiV-F proteins (bottom) and percent processing by densitometry, calculated as F₁/(F₀ +F₁) (top). (B) Relative binding ratios of conformational monoclonal antibody (MAb) 92 and MAb 66, which bind prefusion F of HR3 region mutants of interest. Polyclonal antibody 835 was used to detect levels of overall F expression, to correct for transfection discrepancies or inherent differences in protein expression levels. Data are averages from three or more independent experiments. Error bars designate standard errors of the means. Green lines depict WT F levels. Statistical significance is indicated by asterisks (*, P ≤ 0.05).

Additionally, several mutants had highly significantly lower levels (P ≤ 0.01 to 0.0001) of NiV-F on the cell surface than wild-type NiV-F: the G85A, I86A, L87A, P89A, and L94A mutants (Fig. 2A). This indicates that some residues in the HR3 region, in addition to modulating membrane fusion, are important for correct synthesis and folding of stable NiV-F.

Proteolytic processing and gross conformational aberrancies do not account for the fusion phenotypes of the HR3 region mutants.

NiV-F is initially synthesized as an inactive precursor (F₀), proteolytically cleaved by cathepsin L upon endocytosis into F₁ and F₂ subunits (6, 7, 32, 33), and then recycled back to the cell surface. F₁ and F₂ remain linked by a covalent disulfide bond (Fig. 1C). To analyze potential effects of defects in processing on fusion capabilities, especially because of the proximity of the HR3 region to the cleavage site between amino acids 109 and 110, we analyzed the level of F cleavage for the HR3 mutants. Percent cleavage was determined using a semiquantitative fluorescence-based Western blot (WB) densitometric analysis and by calculating the percentage of processed F (F₁) from total F (F₀ + F₁), as we previously performed (27, 29). F processing for the majority of the hyperfusogenic mutants (Y97A, N99A, N100A, and T101A mutants) was not statistically significantly different from the wild-type NiV-F cleavage level of 32%. Notably, as expected, both the N99A and T101A mutations displayed slightly higher gel migration rates due to the lack of an N-glycan, which confirms the removal of the F3 N-glycan we previously reported (27). Of the hyperfusogenic mutants, only the I90A mutant had a significant increase in processing, to 42% (P ≤ 0.05); however, this slight increase in processing likely does not account for the 4-fold increase in syncytium formation levels observed for this mutant. In agreement with this, HR3 region G85A and P89A mutants that also exhibited statistically significant increases in F processing levels compared to that of wild-type NiV-F did not yield increased syncytium levels (Fig. 2A and Fig. 3A). Similarly, a statistically significant decrease in F processing for the K91A mutant did not correlate with a significant decrease in the cell-cell fusion levels for this F mutant. Moreover, in a previous study, cytoplasmic tail HeV-F mutants that exhibited increases in F cleavage did not translate to significant increases in fusion (34). Again, all these data combined suggest that levels of F processing did not significantly account for how the HR3 affected the fusion properties of the F protein.

F processing for the hypofusogenic mutants K98A and L104A was not statistically significantly different from wild-type NiV-F processing. Interestingly, the L104A mutant, which displayed the lowest level of cell-cell fusion and was closest to the cleavage site (R109/L110), had no significant decrease in F processing (Fig. 3A). Therefore, the fusion phenotypes observed with the fusion mutants of interest (I90A, I96A, Y97A, K98A, N99A, T101A, and L104A mutants) have likely little to do with F processing effects and more to do with steps of the fusion cascade subsequent to F-processing (Fig. 3A).

We thus selected hyperfusogenic I90A, I96A, Y97A, N99A, and T101A mutants and hypofusogenic K98A and L104A mutants for further study. These mutants were selected because they displayed a cell-cell fusion phenotype but did not significantly affect CSE levels. For these mutants, we also tested, using several of our conformational antibodies, whether the F protein undergoes gross conformational changes. For example, hypofusogenic mutants may not be in a prefusion conformation and therefore account for the fusion phenotype observed. We previously reported that conformational monoclonal antibody (MAb) 92 and MAb 66 differentially bound NiV-F hyper- or hypofusogenic mutants and did not bind HeV-F (29). Moreover, MAb 66 was recently shown to bind the apex portion of the prefusion NiV-F (35). We used these conformational monoclonal antibodies to confirm by flow cytometric analyses that the HR3 NiV-F mutants maintain relatively wild-type-like NiV-F prefusion conformations. We accounted for transfection efficiency and/or CSE differences by analyzing binding of the mutants not only to MAb 66 and MAb 92 but also to polyclonal antiserum 835. Thus, the normalized binding ratio of MAb 92 or MAb 66 to antiserum 835 were calculated and compared to that of wild-type NiV-F, normalized to 1 (Fig. 3B). There was no significant difference in binding of MAb 92 to all mutants of interest compared to wild-type F (Fig. 3B). Only the T101A mutant displayed slightly higher binding to MAb 66 (P ≤ 0.05) (Fig. 3B), and mutant N99A appeared to have slightly elevated levels of MAb 66 antibody binding, but this was not statistically significant (Fig. 3B). Both of these mutants lack an N-glycan, and thus their slight enhancement of MAb 66 binding could potentially be due to MAb 66 binding an epitope relatively more exposed by removal of an N-glycan. Overall, the lack of significant changes in conformational MAb binding to most of the fusion mutants of interest suggests that these mutants have roughly a wild-type NiV-F prefusion conformation. For the hypofusogenic mutants, our conformational antibody binding studies support the notion that these mutants are in a wild-type-like prefusion conformation, as opposed to an already triggered postfusion conformation, consistent with the results obtained in syncytial assays conducted at 32°C (Fig. 2D).

Several NiV-F HR3 residues modulate the early F-triggering and late fusion pore expansion steps.

Based on our initial structural analysis prior to this study, our hypothesis was that the HR3 region modulates early F-triggering steps. After determining that the F processing capabilities and conformational states of the HR3 hyperfusogenic and hypofusogenic HR3 mutants could not explain their fusion phenotypes, we performed a functional flow cytometric F-triggering assay, as previously described (31). This assay allowed us to test the ability of the HR3 mutants to undergo the conformational change from the prefusion to the prehairpin intermediate (PHI), upon receptor binding to G, a step we define as F-triggering. In this transient PHI conformational state, the HR1 domain is exposed (Fig. 4A), allowing the HR2-Cy5 peptide that mimics the natural HR2 domain to bind the HR1 domain on the surface of cells expressing triggered NiV-F. Levels of bound HR2-Cy5 peptide are then detected by flow cytometry (Fig. 4B). The F-triggering MFI values (Fig. 4C) were normalized to those obtained for wild-type F, and an F-triggering index was calculated by dividing the normalized F-triggering signal MFI for each mutant (MFI for wild-type F set to 100%) by the normalized cell surface expression of F on the same cells (wild-type F set to 100%) (Fig. 4D); thus, wild-type NiV-F has an F-triggering ratio of 1. This F-triggering ratio normalized to the cell surface expression levels of F is justified, as HR2-Cy5 binding signal depends on the levels of F available (expressed) on the cell surface (31).

FIG 4.

F-triggering and fusion pore formation assays point to NiV-F HR3 effects on F-triggering and fusion pore expansion. (A) Schematic of F-triggering assay. HEK 293T cells were transfected with either wild-type or mutant NiV-F, and wild-type NiV-G. F-triggering assays were performed with an HR2-Cy5 peptide construct. (B) Representative flow cytometry histograms showing binding of HR2-Cy5 to 293T cells expressing either NiV-F alone (negative control [blue]), NiV-G alone (negative control, gray), wild-type NiV-F plus NiV-G (red), and NiV-F Y97A mutant plus NiV-G (pink). (C) Examples of MFI values of HR2-Cy5 binding to 293T cells expressing either NiV-G alone (negative control), NiV-F plus NiV-G, and NiV-F Y97A mutant plus NiV-G. (D) Bound HR2-Cy5 signals were normalized to wild-type NiV-F triggering levels, set to 100. The F-triggering index was calculated by dividing percent normalized F-triggering by percent normalized F CSE. (E) Representative RLU values of DSP luciferase activity of HEK 293T cells expressing either pcDNA backbone only plus DSP_1-7 (negative control), NiV-F plus NiV-G plus DSP_1-7 (wild-type levels), and NiV-F Y97A mutant plus NiV-G plus DSP_1-7 (hyperfusogenic mutant), 8 h after target cell overlay. (F) Luciferase activity of DSP fusion pore formation levels normalized to those of wild-type NiV-F for hyperfusogenic and hypofusogenic mutants. Data are averages from 3 or more independent experiments. Error bars designate standard errors of the means. (G) DSP luciferase activity kinetics of WT F and the K98A mutant over 12 h. Data are averages from 2 experiments; error bars designate standard deviations. (H) Fusion pore formation from panel F plotted against log F-triggering indices from panel D. Green lines depict WT F levels. Statistical significance is indicated by asterisks (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001).

Overall, all hyperfusogenic HR3 mutants had significantly increased levels of F-triggering compared to that of the wild-type NiV-F (P ≤ 0.05 to P ≤ 0.01) (Fig. 4D). While the T101A and I90A mutants were triggered at the highest levels (∼4-fold and ∼2-fold, respectively) compared to wild-type NiV-F, the I96A, Y97A and N99A mutants showed milder increases, between ∼1.2- and ∼1.7-fold above wild-type NiV-F levels. These results indicate that the HR3 domain is important for early F-triggering steps. Very interestingly, the completely cell-cell fusion-dead L104A mutant displayed only a 60% reduction in F-triggering (P < 0.01), while the hypofusogenic K98A mutant yielded wild-type F-triggering levels (Fig. 4D). These data suggested that these hypofusogenic mutants may downmodulate a post-F-triggering step(s). Therefore, we proceeded to test our mutants in our fusion pore formation dual split-protein cell-cell fusion assay.

The dual split-protein cell-cell fusion assay was previously developed to qualitatively measure cell-cell fusion for several viruses (31, 36, 37). We recently adapted this method for NiV surface glycoproteins expressed in 293T cells (31). We transfected effector cells with NiV-F (wild-type or HR3 mutant), wild-type NiV-G, and the split-protein DSP_1-7. Ten to 12 h posttransfection, cells were overlaid in the presence of the luciferase substrate EnduRen, with target cells transfected with ephrin B2 (to maximize receptor concentration) and the complementary split-protein DSP_8-11 in a 1:1 ratio. Once a fusion pore forms that is sufficiently sized to allow enough cellular content mixing to yield a fully functional Renilla luciferase (DSP_1-7 plus DSP_8-11 full protein), an EnduRen signal is detected. Luciferase activity was measured 8 h after target cell overlay using a Tecan plate reader, as this time was determined to yield consistent signals near the top of the linear range (31, 36, 37). Luciferase activity for HR3 NiV-F mutants was measured as relative light units (RLU) (Fig. 4E) and normalized to that of wild-type NiV-F (set at 100%) (Fig. 4F). This assay therefore measures fusion pore formation, a fusion step between F-triggering and extensive fusion pore expansion (movement of nuclei between fused cells).

The hyperfusogenic I90A, I96A, Y97A, N99A, and T101A mutants all yielded significantly increased levels of normalized DSP fusion compared to that of the wild-type NiV-F (Fig. 4F). The I90A and T101A mutants yielded an ∼2-fold increase in fusion pore formation (P ≤ 0.001 and P ≤ 0.01, respectively). Other hyperfusogenic mutants (I96A, Y97A, and N99A mutants) yielded ∼1.5- to 1.7-fold increases compared to the wild-type NiV-F, fairly consistent with their F-triggering levels (Fig. 4F and H). Therefore, for the hyperfusogenic mutants, their enhanced fusion pore formation phenotypes likely stem primarily from their earlier enhanced F-triggering step. In accordance with our observation, when we plotted fusion pore formation against F-triggering, there was a significant level of positive correlation (P = 0.0014 and R² = 0.8374 [Fig. 4H]).

Interestingly however, the hypofusogenic K98A mutant (with ∼50% syncytium formation levels [Fig. 2]) yielded a slight increase in its level of DSP fusion pore formation compared to wild-type NiV-F (∼120%, P ≤ 0.05) (Fig. 4F). Moreover, when the cell-cell fusion kinetics of mutant K98A were plotted with those of wild-type F over a span of 12 h, we observed that at early time points for the DSP fusion assay (0 to 4 h), the two proteins had similar levels of fusion pore formation, and as time progressed, the K98A mutant had slightly elevated levels of fusion pore formation (Fig. 4G). This suggests that this mutant is hypofusogenic due to a step(s) subsequent to both F-triggering and fusion pore formation. Of further interest, the cell-cell fusion-dead L104A mutant, which exhibited ∼0% syncytium formation (Fig. 2), yielded 25% fusion pore formation (P ≤ 0.001). These findings indicate that at least partially, the cause for the hypofusogenic phenotype of both the K98A and L104A mutants is a step(s) subsequent to F-triggering and fusion pore formation, such as extensive fusion pore expansion.

Hypofusogenic HR3 mutants can undergo viral entry.

Both virus-to-cell fusion and cell-to-cell fusion processes utilize the same glycoprotein machinery. However, there are differences between the two membrane fusion processes, such as in membrane curvature, lipid and protein compositions, etc. Therefore, we asked whether the fusion phenotype observed in syncytial assays would reflect levels of viral entry (Fig. 5). As NiV is a biosafety level 4 (BSL4) pathogen, we utilized our established BSL2 pseudotyped viral entry assay to assess viral entry yielded by our select HR3 fusion mutants of interest. NiV-G and NiV-F (wild type or HR3 mutants) were pseudotyped onto vesicular stomatitis virus (VSV) that lacks its native G protein and contains a Renilla luciferase reporter gene, as previously described (32, 38). For our assays, VSV pseudotyped by wild-type NiV-G alone served as our negative control, as these particles allow cell receptor binding but not membrane fusion or viral entry.

FIG 5.

HR3 NiV-F VSV-rLuc pseudotyped viral entry. (A and B) Viral entry of hyperfusogenic HR3 mutants (A) and hypofusogenic HR3 mutants (B) into Vero cells. Vero cells were infected with NiV/VSV-rLuc pseudotyped virions containing wild-type NiV-G and either wild-type NiV-F or HR3 mutant NiV-F. Virions were produced in HEK 293T cells, and their genome copy numbers were measured by RT-qPCR. Equal genome copy numbers were used at the time of infection. Relative light units (RLU) were quantified and plotted against viral serial dilutions. Data shown are averages from three independent experiments. Error bars represent standard errors of the means. Statistical significance is indicated by asterisks (*, P ≤ 0.05; **, P ≤ 0.01). (C) Equivalent amount of NiV/VSV-rLuc pseudotyped virions were separated by 10% SDS-PAGE and immunoblotted to detect wild-type NiV-G and either wild-type NiV-F or HR3 mutant NiV-F.

To accurately compare entry levels between pseudotyped VSVs incorporating wild-type or HR3 mutant NiV-F proteins, we determined the genome copy numbers of our viral preparations and infected Vero cells with equal genome copy numbers over several orders of magnitude. Additionally, to determine the levels of incorporation of NiV-F (wild type or HR3 mutants) onto VSV pseudotyped virions, we performed Western blot analysis on equilibrated concentrations of NiV/VSV-rLuc pseudotyped virions (Fig. 5C). Interestingly, most hyperfusogenic HR3 mutants (I90A, Y97A, N99A, and T101A mutants) displayed decreased NiV F and G incorporation levels, with the exception of the I96A mutant, which had incorporation levels relatively close to that of wild-type NiV/VSV virions. In contrast, virions containing the hypofusogenic HR3 K98A and L104A mutants had incorporation levels similar to those of the wild-type NiV/VSV virions. The hyperfusogenic mutants with reduced incorporation (I90A, Y97A, N99A, and T101A mutants) also displayed decreased viral entry levels compared to that of wild-type NiV-F (P ≤ 0.05 and P ≤ 0.01) (Fig. 5A). Similarly, the cell-cell fusion-dead L104A virions had significantly diminished viral entry levels compared to those of the wild-type virions (P ≤ 0.05) (Fig. 5B). In contrast, the hyperfusogenic I96A and the hypofusogenic K98A NiV-F mutants yielded wild-type levels of viral entry. These data suggested that for most of the hyperfusogenic HR3 mutants, incorporation of the glycoproteins into the virions played a major role in determining the overall levels of viral entry. Therefore, some mutant entry phenotypes may be masked by poor levels of incorporation of F (and in some cases G) on the virions due to excessive syncytium formation in the virion-producing cells. We tried to improve the levels of incorporation of the hyperfusogenic F mutants onto VSV virions by producing them at 32°C (to limit cell-cell fusion), with no success (data not shown) (28).

Interestingly, the K98A mutant had wild-type levels of F-triggering, slightly increased levels of fusion pore formation, but reduced syncytium formation (∼50%). However, the K98A also yielded wild-type-levels of F and G incorporation and viral entry (Fig. 5B and C). These data strongly suggest that fusion pore formation is sufficient to allow viral genome entry into cells and thus luciferase activity. In support of this conclusion, the L104A mutant yielded low levels of F-triggering and fusion pore formation and no visible syncytium formation. However, the L104A NiV-F/VSV-rLuc pseudotyped virions were still able to infect Vero cells at low levels compared to the absolute negative-control G-only virions (Fig. 5B). As aforementioned, virus-cell and cell-cell membrane fusion processes likely utilize similar early membrane fusion mechanisms. However, our data are consistent with viral genome entry requiring only fusion pore formation and minimal fusion pore expansion, with syncytium formation requiring much more extensive large fusion pore expansion to allow movement of nuclei among cells. Our viral entry data are thus consistent with the hypofusogenic K98A and L104A HR3 mutants affecting the extensive large fusion pore expansion step needed for full syncytium formation. Therefore, we propose that the HR3 domain modulates both early F-triggering and late fusion pore expansion steps in the membrane fusion cascade.

Fusogenicity of HR3 mutants does not correlate with altered avidities of NiV-G and NiV-F interactions.

Some henipaviruses and morbilliviruses bind proteinaceous receptors and follow some variation of a dissociation model of interactions, in which the attachment glycoprotein initially interacts with the fusion glycoprotein prior to receptor binding and then dissociates after receptor-binding at some point during the membrane fusion process (14–16, 18, 39, 40). We previously reported a trend that favored an inverse relationship between the avidity of NiV-F and NiV-G interactions and membrane fusogenicity of several groups of henipaviral F and G mutants (10, 27–29, 31, 41). To determine if the HR3 NiV-F mutants that affected cell-cell fusion levels exhibited the expected trend of interaction with NiV-G, we performed established coimmunoprecipitation experiments, as previously described (28, 30, 31). HEK 293T cells were transfected with both NiV-G and NiV-F (wild type or mutant) at a 1:1 plasmid/plasmid ratio. Wild-type or mutant NiV-F proteins were then affinity purified using rabbit anti-AU1 to target the AU1 tag located at the C-terminal end of the F protein. The NiV-G protein that immunoprecipitated with NiV-F was detected by Western blot analysis using an anti-hemagglutinin (anti-HA) fluorescent antibody (Fig. 6A). As previously performed, to determine relative F/G binding avidities, we determined the levels of coimmunoprecipitated G relative to the overall G expression levels from lysates using a semiquantitative fluorescence-based densitometric analysis, a relatively more quantitative method than traditional chemiluminescence Western blot detection methods (28, 31). We calculated the binding of NiV-G to NiV-F as G_IP/(G_lys × F_IP) and normalized these values to the binding avidity of wild-type NiV-F, normalized to 100% (Fig. 6B).

FIG 6.

Fusogenicities and avidities of G/F interactions for the HR3 fusion mutants do not correlate. (A) Western blot analysis of NiV-F coimmunoprecipitation with NiV-G. NiV-F was pulled down with protein A beads coupled with rabbit anti-AU1. Pulldown of NiV-G was detected with anti-HA PE. (B) Densitometric analysis was performed to quantitate levels of NiV-G pulldown by calculating (G_IP)/(G_lys × F_IP). Black line depicts WT F levels. Statistical significance is indicated by asterisks (*, P ≤ 0.05; **, P ≤ 0.01). (C) Avidity of F-G interactions from panel B plotted against log fusion indices from Fig. 2B.

In this study, a couple of hyperfusogenic mutants, the I96A and T101A mutants, yielded lower avidities of interactions with NiV-G compared to wild-type NiV-F (Fig. 6A and B). However, most of the HR3 mutants did not follow the previously published trend of G/F interactions. For example, the hypofusogenic L104A mutant yielded near wild-type-levels of interactions with NiV-G (Fig. 6B). Further, the hyperfusogenic I90A mutant exhibited significantly higher levels of NiV-G interactions than the wild-type NiV-F (P ≤ 0.05). Conversely, the hypofusogenic K98A mutant yielded significantly decreased avidity of G/F interactions (Fig. 6A and B). These data support the notion that the interactions between NiV-G and NiV-F may modulate the fusion process via a mechanism more complex than a simple dissociation model, at least as modulated by the NiV-F HR3 domain. Congruently, when fusion capabilities were plotted against G/F avidities, we did not observe any mathematical correlation between fusogenicity and G/F avidities (Fig. 6C). Thus, the differences observed in the avidities of interaction between NiV-G and the HR3 mutants support the notion that the F/G interactions are complex and may modulate membrane fusion by more than a single interactive step in the membrane fusion process.

DISCUSSION

Paramyxoviral entry requires both the receptor binding protein (RBP), designated G, H, or HN, and the fusion F glycoproteins. While the RBP binds the cell host receptor(s) and triggers F, ultimately F drives the membrane merging mechanism, which is fairly conserved and required during both viral entry and the pathognomonic syncytium formation. The roles of several F domains in modulating fusion are increasingly clear, including the fusion peptide, HR1, HR2, transmembrane domain, cytoplasmic tail, Ig-like domain, some residues within the F₂ domain, linker/loop regions, and the F head-stem interface regions (27, 29, 42–62). The potential role(s) of the paramyxoviral F extracellular third helical (HR3) domain, however, remains largely unknown. Recently, the HR3 (also known as HRC) domain of the related respiratory syncytial virus (RSV) in the related Pneumoviridae family was described to generally affect membrane fusion. However, the precise step(s) of the membrane fusion cascade affected for RSV remains unknown (63). Further, RSV has been reported to cause infection without its cognate attachment protein G (64). RSV-F (from the A2 and cp-52 strains) has also been reported to bind heparin or heparan-like molecules on the cell surface (65), and this is sufficient to trigger membrane fusion. In contrast, for the paramyxoviruses, including the henipaviruses, F-independent fusion has not been observed under physiological conditions, and F has not been reported to bind any host cell receptor. Therefore, the mechanisms of F-triggering between pneumoviruses and paramyxoviruses are likely significantly different.

Comparison of the crystal structure of NiV-F in its prefusion conformation with a NiV-F postfusion conformation model based on the paramyxovirus hPIV3 (22) (since there is no available postfusion structure for NiV-F) revealed conformational differences in the two structures (Fig. 1). Utilizing several new mechanistic assays, we performed a comprehensive mutational analysis of the 20-residue portion of the HR3 domain closest to the fusion peptide (N84 to L104) and demonstrated roles of the HR3 in both early F-triggering steps and late large fusion pore expansion steps. We concluded that the I90A, I96A, Y97A, N99A and T101A mutants are hyperfusogenic due to effects in early F-triggering steps (Fig. 2, 4, and 7A). Importantly, the observed changes in F-triggering were not due to significant differences in the levels of cell surface expression, proteolytic cleavage, or gross conformational configurations (Fig. 2 and 3). Further, for most HR3 mutants, F-triggering levels seemed responsible for the subsequent levels of fusion pore formation, and the two phenotypes correlated well (Fig. 4H). It is conceivable that since the HR3 region modulates F-triggering, some HR3 mutations may result in premature F-triggering. However, when we expressed the NiV-F HR3 mutants in the absence of G, we detected wild-type F prefusion conformations (Fig. 3B), and no syncytia or F-triggering was detected at 37°C, indicating the requirement of G for F-triggering. It is possible that multiple HR3 mutations may abrogate the need of G for F-triggering, which would require future studies.

FIG 7.

Model of NiV-F HR3 involvement in F-triggering and fusion pore expansion. (A) Table summarizing the steps in the fusion cascade affected by HR3 mutants. (B) Depiction of fusion cascade steps. (C) Prefusion NiV-F in a hexamer-of-trimer assembly (PDB accession code 5EVM) (21). (D) Insets of trimer-trimer interfaces in which portions of the HR3 can be observed.

It was previously reported that residue V94 of measles virus F (homologous to NiV-F Y97), which differs between wild-type and Edm strains, is important in yielding differential cytopathic effects (CPE), and a V94A substitution yielded differential syncytium formation (56). Structural analysis of prefusion NiV-F shows that Y97 is located at an interface between the HR3 and fusion peptide regions, suggesting a mechanism in which this amino acid is important for fusion peptide release, a step we define as F-triggering (Fig. 1B and Fig. 4D). In another recent study, adding cysteine constraints between residue Y97 (in the HR3 region of HeV-F) and residue G131 (in the HR1 region), or between residue N100 (in the HR3) and residue A119 (in the HR1), resulted in decreased cell-cell fusion levels induced by HeV-F, supporting a mechanism in which the HR3 region is important for initial F-triggering steps (66). Further, we previously reported that N-glycan removal at NiV-F residue N99 (mutant N99Q in the HR3) highly increased cell-cell fusion levels (severalfold) over NiV F levels (27). In this study, we further determined that this hyperfusion phenotype is due to enhanced levels of F-triggering when we analyzed the similar N99A and T101A mutants, which also lack the N-glycan (Fig. 2 and Fig. 4D).

Interestingly, the hypofusogenic HR3 mutants yielded either wild-type (K98A) or decreased (L104A) levels of F-triggering. Further, fusion pore formation data indicated that the hypofusogenic phenotype of these mutants was at least partially due to modulation of a step after fusion pore formation, such as extensive large fusion pore expansion (Fig. 4F and G and Fig. 7A). Importantly, the fusion phenotypes were not a result of differences in F protein cell surface expression levels, proteolytic cleavage, or conformational configurations (Fig. 2 and 3). Further, the level of F processing for the L104A mutant is consistent with previous findings by Moll et al. that mutations upstream of the cleavage site did not affect F cleavage (67). The phenotypes of the K98A mutant indicate that the HR3 domain can have effects on extensive fusion pore expansion (Fig. 4D, F, and G). The F-triggering assay measures the proportion of F proteins that have reached the prehairpin intermediate (PHI) conformation (a step defined as F-triggering). F-triggering is measured by binding of the soluble HR2-Cy5 peptide to the exposed trimeric HR1 region in F in the PHI conformation (Fig. 4A). It is noteworthy that after the PHI folds into the 6-helix bundle (6HB) conformation, the HR2-Cy5 peptide no longer binds F (68). The levels of PHI (HR2-Cy5 binding) could be affected by the kinetics of F-triggering and the refolding to the postfusion state. Since the K98A mutant yielded wild-type levels of F-triggering and slightly elevated levels of fusion pore formation, but a lower level of syncytium formation (Fig. 4D, F, and G), we posit that the K98A mutant affects a later large fusion pore expansion step noted by lower association of nuclei during syncytium formation. We posit that the alternative scenario, that the phenotypes of the K98A mutant are due to rapid and thus nonproductive folding, is unlikely. In fact, when we measured syncytium formation at 32°C, the K98A mutant still exhibited ∼40% fusion compared to that of wild-type F, similar to the levels of K98A mutant syncytium formation at 37°C (Fig. 2D). Further, the K98A mutant was found in a prefusion conformation (Fig. 3B) and still able to infect Vero cells when pseudotyped onto VSV virions with NiV-G (Fig. 5B), which requires only small fusion pore expansion. The L104A mutant illustrates that a single point mutation in the HR3 domain can affect both F-triggering and extensive fusion pore expansion. When we performed syncytial fusion assays at 32°C, the L104A mutant was still not able to form sufficient syncytia (Fig. 2D). Interestingly, it was previously reported that Edm strains and wild-type strains of MeV differed at F amino acid position 101, homologous to the L104A substitution in NiV F (56). However, in the case of MeV F, amino acid substitutions at position 101 did not significantly alter fusogenicity, as opposed to our observations for NiV F, suggesting differences in the precise mechanism of fusion modulation (56).

Although we primarily focused on HR3 mutants that expressed at approximately wild-type levels at the cell surface, there were a few mutants that yielded reduced cell surface expression and protein production: the G85A, I86A, L87A, P89A, and L94A mutants (Fig. 2 and 3). It was previously reported that the L87A and P89A mutants within a conserved block in the HeV F₂ portion of F also had deficient cell surface expression and trimerization capabilities, respectively (55). This indicates that certain portions of the HR3 region are needed for proper protein folding and production. We speculate that such residues may be particularly helpful in folding F into its proper metastable prefusion conformation.

We performed coimmunoprecipitation experiments to measure relative binding of HR3 NiV-F mutants to NiV-G (Fig. 6). Based on prior data for other NiV-G and NiV-F mutants, for a couple of the HR3 hyperfusogenic mutants we observed the expected trend of loss in the avidities of F/G interactions, in agreement with a variation of the dissociation model (an inverse correlation between fusion capabilities and F/G interaction avidities) (10, 11, 27–29, 31, 41). However, the hyperfusogenic I90A mutant and the hypofusogenic K98A mutant did not follow this trend. Overall, we did not find a correlation between fusion activity and the avidity of G/F interactions for mutants in the HR3 domain, although there were observed differences in interactions with NiV-G. These data support the idea that the henipaviral F/G interactions are more complex than the relatively simple dissociation model. Instead, F and G may undergo several stages of relative association and dissociation during the fusion cascade. We postulate that perhaps the HR3 region (which is solvent exposed) may be a face of interaction of NiV-F with NiV-G that may influence such complex changes. However, these postulates require further experimentation.

Based on our present findings, we delineate a model of the fusion cascade in which the NiV-F HR3 domain primarily acts in early F-triggering or fusion peptide release steps and this F-triggering capacity translates downstream to the later steps of the membrane fusion cascade, after NiV-G receptor binding (Fig. 7A and B). By collaborative efforts we recently reported that NiV-F trimers can form higher-order multimeric structures: hexamers of trimers (21). In that study, based on trimer-trimer interface mutants affecting cell-cell fusion and viral entry phenotypes, we proposed that F hexamer-of-trimer oligomerization may play a role in F-triggering. Interestingly, the HR3 region is located at shared interfaces between NiV-F trimers in the hexamer-of-trimer structure (Fig. 7C and D) (21). Therefore, our data are consistent with a mechanism in which the HR3 region may use the hexamer-of-trimer oligomeric arrangement to modulate F-triggering as well as downstream steps, such as large extensive fusion pore expansion during the pathological syncytium formation for the henipaviruses. For example, it is possible that the K98A and L104A mutants may abrogate F-triggering and/or large fusion pore expansion capacities via the F hexamer-of-trimer arrangement.

As proof of concept, we performed NiV/VSV-rLuc pseudotype viral entry assays with the hyper- and hypofusogenic HR3 domain mutants. Although there is a limit to utilizing the VSV pseudotyped system for paramyxoviral entry, the NiV F glycoproteins pseudotyped onto the VSV virion are what drive the virus-cell fusion process. Most of the hyperfusogenic mutants resulted in decreased infectivity, largely due to a lack of incorporation into virions (Fig. 5). However, the hypofusogenic mutants were incorporated into virions and were able to infect cells at wild-type levels (K98A) or to reduced levels (L104A) despite displaying low to completely abrogated levels of syncytium formation, respectively. This supports the model that these mutants have fusion pore formation and perhaps limited fusion pore expansion but lack expansive large fusion pore expansion capability, since only a small fusion pore is likely required for viral entry, but much larger fusion pore expansion is likely required for full cell-cell fusion during the pathological syncytium formation.

Future experiments are needed to determine whether the HR3 region is involved in hexamer-of-trimer formation and whether such an oligomeric state may be required or preferable for optimal F-triggering and optimal fusion pore expansion (Fig. 7C and D). Additionally, the effects of multiple mutations within the HR3 region of the same F protein remain to be studied. Importantly, the HR3 domain has a high level of sequence homology with the equivalent HR3 domains of other henipaviruses (95%, 76%, and 61% identities with Hendra, Kumasi, and Cedar henipaviruses, respectively [Fig. 8A]), as well as a significant level of sequence homology with the HR3 sequences of other paramyxoviruses (∼42% to 52% identities) (Fig. 8B). Moreover, the HR3 domain is structurally conserved across paramyxoviruses (Fig. 8C).

Thus, if our hypothesis is correct, we posit that the potential role(s) of membrane fusion modulation uncovered in this study for the NiV F HR3 may be conserved beyond NiV in other henipaviruses and perhaps in other paramyxoviruses. Remarkably, our results indicate that the HR3 domain plays important roles in both early F-triggering and large extensive fusion pore expansion steps in the membrane fusion cascade during syncytium formation. Furthermore, due to the roles of the NiV-F HR3 in these early and later steps of the membrane fusion process, and its relatively solvent-exposed structure (69), the HR3 domain may constitute a great new candidate for antiviral and vaccine development strategies.

MATERIALS AND METHODS

Cells and their maintenance.

Human embryonic kidney HEK293T cells (293T) and Vero cells were grown at 37°C in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 5% CO₂. 293T cells were used for transfections and Vero cells were used for NiV/VSV-rLuc pseudotyped viral entry assays.

Plasmid construction and utilization.

Mammalian codon-optimized NiV-F and NiV-G from the Malaysian strain of NiV were previously constructed in pcDNA3.1+ vectors (70). NiV-G has an HA tag at its C terminus (extracellular) and NiV-F has an AU1 tag at its C-terminal end (intracellular). HR3 scanning alanine point mutants spanning amino acids 84 to 104 were generated using a QuikChange mutagenesis kit using NiV-F AU1 in pcDNA3.1+ plasmid as a template. Mutants were created in pcDNA3.1+ vectors and then cloned for improved expression into a mammalian pCAGGS vector by restriction enzyme digestion using KpnI and XhoI ligation (Invitrogen and New England BioLabs [NEB]). Full gene sequences were confirmed.

Transfections.

Transfections were performed utilizing a variety of transfection reagents: Turbofect (Thermo Scientific), Lipofectamine 2000 (Invitrogen), and polyethyleneimine (PEI) depending on the cell type used. Transfection reagents were used under manufacturers’ recommendations.

Cell surface expression of the glycoproteins.

239T cells grown in 6-well plates were transfected with wild-type NiV-F or HR3 mutants using Turbofect and 2 μg of DNA for 16 to 24 h. Cells were subsequently collected in phosphate-buffered saline (PBS) alone or PBS containing EDTA at 4°C for 15 to 30 min. Cells were then transferred to a 96-well plate and primarily stained with 1:1,000 rabbit polyclonal antibody 834 or 835 (specific to NiV-F) diluted in PBS plus 1% bovine serum albumin (BSA) for an hour as previously performed (70). Cells were washed 3 times with PBS and then incubated with secondary fluorescent goat anti-rabbit Alexa Fluor 647 or 488 antibody at a 1:500 dilution. Next, cells were washed 2 or 3 times with PBS buffer and fixed in 0.5% paraformaldehyde (PFA). Negative controls for CSE experiments were cells transfected with the pcDNA3.1+ vector. Fixed cells were then analyzed in a second-generation Millipore Guava easyCyte flow cytometer.

To ensure that the presence of NiV-G did not alter NiV-F cell surface expression, cells lacking the receptor (PK13) were transfected with wild-type NiV-G and wild-type NiV-F or HR3 mutants of interest at a 1:1 ratio using Lipofectamine and 2 μg of DNA for 24 h. There was no significant difference observed in CSE between wild-type NiV-F and HR3 mutants in the presence or absence of NiV-G (data not shown).

Cell-cell syncytial assay.

293T cells were grown in 6-well plates to form an evenly distributed monolayer. Cells were then transfected ∼24 h after seeding with NiV-G and NiV-F wild-type or HR3 mutant analogs in a 1:1 ratio or 1:3 ratio. Transfections were done using Turbofect, with 2 μg of DNA per well. Cells were washed carefully with PBS, fixed in 1% PFA at 37°C for 30 min, and left in tissue culture water for 16 h posttransfection. To measure cell-cell syncytium formation at 32°C, cells were transfected as described above at 37°C for 5 h and subsequently moved to 32°C for 11 h. Cells were washed carefully with PBS and fixed in 1% PFA at 25°C for 16 h posttransfection. Cells were then analyzed under a Zeiss microscope at a magnification of ×200. Nuclei inside syncytia (each nucleus resulting from an individual fused cell joining a syncytium) in 5 fields per well were counted. The negative control for syncytial experiments was cells transfected with pcDNA3.1+ vector or NiV-G only. Only syncytia that had 4 or more nuclei were counted to exclude mitotic events. Averages of the 5 fields were calculated and normalized to the average yielded by wild-type NiV-F as previously performed (28).

WB analyses.

293T cells were grown in 6-well plates and subsequently transfected with NiV-F or HR3 mutant plasmids using Turbofect and 2 μg of total DNA. Cells were collected 24 to 36 h posttransfection and then lysed with 1× radioimmunoprecipitation (RIPA) buffer. Lysates were centrifuged at a relative centrifugal force (rcf) of 21,000 at 4°C. Subsequently, supernatants were incubated with 1× SDS buffer plus 2% β-mercaptoethanol (BME) at room temperature for 10 min and loaded onto 10% acrylamide–Tris-HCl gels. Proteins were separated at 100 V for 2 h and then transferred onto polyvinylidene difluoride (PVDF) membranes at 0.5 A for 1.5 h. Membranes were incubated in Li-COR blocking buffer overnight. Primary antibody incubation was performed by staining membranes with 1:1,000 mouse anti-AU1 (Biolegend) to detect NiV-F for 1 h. The membrane was then washed 3 times with PBS plus blocking buffer for ∼10 min per wash. Subsequently, the membrane was incubated with secondary goat anti-mouse Alexa Fluor 488 and goat anti-rabbit Alexa Fluor 647 (Thermo Scientific Fisher). The membrane was imaged using a Bio-Rad imager. Densitometric analysis was done by calculating the density of dark pixels. Percent cleavage was determined by calculating the percent processed F (F₁ signal) from total expression of F (precursor [F₀ signal] plus processed F₁ signal). Negative controls for WB analysis were cell lysates obtained from cells mock transfected with pcDNA3.1+ vector alone.

Binding of NiV-F conformational antibodies by flow cytometry.

NiV-F or HR3 mutants of interest were transfected into 293T cells for 24 h using 2 μg of DNA. After the transfection period, cells were collected and split to be stained with polyclonal anti-F 835 antibody or MAb 92 or MAb 66 at a 1:1,000 dilution in PBS plus 1% BSA. To detect binding of these primary antibodies, we probed with secondary goat anti-rabbit 488 antibody at a 1:500 dilution. These experiments were performed basically as previously described (29).

F-triggering assay for hypofusogenic mutants.

F-triggering assays were performed as previously described, with further modifications for optimization using an HR2-Cy5 peptide (31, 68). 293T cells were grown in 6-well plates. Twenty-four hours later, at around 80 to 90% confluency, cells were transfected with PEI with a total 2 μg/well at a 1:3 G/F ratio of DNA expression plasmids. Negative controls for F-triggering assays were 293T cells transfected with NiV-G only. Wild-type NiV-F or hypofusogenic HR3 mutants were transfected with NiV-G. Ten to 14 h posttransfection, cells were examined under the microscope to assess fusion levels prior to collection. Transfected cells were washed with PBS and resuspended in PBS plus 1% BSA. Transfected cells or effector cells were incubated and also functioned as target cells. Cells were incubated in the presence of 2 μM HR2-Cy5 peptide for 1 h at 4°C. Subsequently, cells were placed at 37°C for 30 min to allow for F-triggering. The 30-min incubation period was selected based on previous F-triggering kinetics assays, in which most F-triggering occurred within a 15-min time frame (68). Therefore, a 30-min incubation period is sufficient for most F-triggering to occur. Cells were finally washed 3 times with PBS and resuspended in 300 μl of 1× PBS. Cells were then immediately analyzed using a second-generation Millipore Guava easyCyte flow cytometer following previously described gating strategies (68).

F-triggering assay for hyperfusogenic mutants.

F-triggering assays resulted in better negative-control-to-test ratios when performed before substantial fusion occurred. Therefore, we performed F-triggering assays with hyperfusogenic mutants using a slightly different approach. Cells were transfected as aforementioned and maintained at 37°C for 6 to 8 h before being moved to 32 to 33°C to allow for efficient transfection but to prevent complete fusion in 293T cells. Transfecting cells at 32 to 33°C has previously been reported to not reduce transfection efficiency (38). Then cells were tested for F-triggering as described above for the hypofusogenic mutants.

Fusion pore formation (dual split-protein fusion) assay.

293T cells were seeded on a ViewPlate 96-well plate and transfected as previously described (31, 36, 37). Effector cells, grown in the ViewPlate 96-well plate, were transfected with G, F, and dual split protein (DSP_1-7) plasmids at a 2:6:4 ratio using 0.2 μg of DNA/well. Target 293T cells were transfected with ephrin B2 plasmid and DSP_8-11. Ten to 12 h after posttransfection, effector cells and target cells were incubated with live-cell EnduRen substrate and were incubated together in a 1:1 ratio. Eight hours after overlay, luciferase activity in fused cells containing the active dual split green fluorescent protein (GFP) and Renilla luciferase protein was measured as previously described (31, 36, 37). Negative controls for this assay included 293T cells transfected with pcDNA and DSP_1-7.

Generation and quantification of VSV pseudotyped virions with wild-type NiV-G and HR3 NiV-F mutants.

293T cells were seeded onto 10-cm plates and transfected at 90 to 95% confluency. Cells were transfected using PEI with G and F DNA expression plasmids at a G/F ratio of 1:3, using 8 μg/plate. Eight to 10 h posttransfection, cells were infected with a previously constructed recombinant VSV whose VSV-G gene sequence had been replaced with a Renilla luciferase reporter gene (VSV-ΔG-rLuc) (10, 27, 29, 31). NiV/VSV-rLuc pseudotyped virions were collected at ∼48 h posttransfection and purified via centrifugation over a 20% sucrose cushion. Virions were resuspended in NTE buffer (150 mM NaCl, 40 mM Tris-HCl, 1 mM EDTA [pH 8.0]) plus 5% sucrose and stored at –80°C. Viral RNA was extracted using the E.Z.N.A. viral RNA kit (Omega Bio-tek). Subsequently, VSV genome copy numbers were reverse transcribed and quantified by reverse transcription-quantitative PCR (RT-qPCR) using UltraPlex 1-Step (Quantabio) with a TaqMan Ind-1 spec probe.

To quantify NiV/VSV-rLuc pseudotyped virion infections, Vero cells were seeded in a 96-well plate and infected at ∼30% confluency. NiV/VSV-rLuc virion genome copy numbers were equilibrated and subsequently serially diluted to infect Vero cells. Infections were done for 2 h in infection buffer (1% FBS in PBS) and then fresh medium was added. Twenty-four hours postinfection, Vero cells were lysed, and Renilla luciferase activity was measured using a Renilla luciferase assay system kit (Promega) and a Tecan Spark plate reader. Relative light units were then plotted against serial dilutions as previously performed (10, 27, 29, 31).

Coimmunoprecipitation assay.

293T cells were transfected with NiV-G and NiV-F or HR3 region mutants at a 1:1 ratio. Cells were collected 24 h posttransfection and lysed with 1× RIPA buffer. NiV F protein was pulled down with rabbit anti-AU1 antibody coupled to protein G beads (Miltenyi). Anti-HA phycoerythrin (PE) (Miltenyi) was used to detect coimmunoprecipitated NiV-G with an HA tag at the C terminus, and mouse anti-AU1 and secondary goat anti-mouse 647 were used to detect immunoprecipitated NiV-F with an AU1 tag at the C terminus. All antibodies were used at a 1:1,000 dilution. NiV-G association was detected via Western blot analysis, as previously described (31).

Statistical analyses.

Statistical analyses were performed using GraphPad Prism, version 8. For comparisons between HR3 mutant NiV-F analogs or “test values” to a known value of the normalized wild-type NiV-F (always set as 100% or 1.0, depending on the analysis), we performed one-sample, two-tailed t tests. We performed unpaired, two-sample, two-tailed t tests for comparisons between HR3 NiV-F mutants and wild-type NiV-F F-processing levels and between NiV/VSV pseudotyped viral entry assays. Nonsignificant differences (P > 0.05) are unmarked; in figures, “*” indicates a significant difference (P ≤ 0.05); “**” indicates a very significant difference (P ≤ 0.01), and “***” and “****” indicate very high significant differences (P ≤ 0.001 and P ≤ 0.0001, respectively).