Nipah Virion Entry Kinetics, Composition, and Conformational Changes Determined by Enzymatic Virus-Like Particles and New Flow Virometry Tools

ABSTRACT

Virus-cell membrane fusion is essential for enveloped virus infections. However, mechanistic viral membrane fusion studies have predominantly focused on cell-cell fusion models, largely due to the low availability of technologies capable of characterizing actual virus-cell membrane fusion. Although cell-cell fusion assays are valuable, they do not fully recapitulate all the variables of virus-cell membrane fusion. Drastic differences between viral and cellular membrane lipid and protein compositions and curvatures exist. For biosafety level 4 (BSL4) pathogens such as the deadly Nipah virus (NiV), virus-cell fusion mechanistic studies are notably cumbersome. To circumvent these limitations, we used enzymatic Nipah virus-like-particles (NiVLPs) and developed new flow virometric tools. NiV's attachment (G) and fusion (F) envelope glycoproteins mediate viral binding to the ephrinB2/ephrinB3 cell receptors and virus-cell membrane fusion, respectively. The NiV matrix protein (M) can autonomously induce NiV assembly and budding. Using a β-lactamase (βLa) reporter/NiV-M chimeric protein, we produced NiVLPs expressing NiV-G and wild-type or mutant NiV-F on their surfaces. By preloading target cells with the βLa fluorescent substrate CCF2-AM, we obtained viral entry kinetic curves that correlated with the NiV-F fusogenic phenotypes, validating NiVLPs as suitable viral entry kinetic tools and suggesting overall relatively slower viral entry than cell-cell fusion kinetics. Additionally, the proportions of F and G on individual NiVLPs and the extent of receptor-induced conformational changes in NiV-G were measured via flow virometry, allowing the proper interpretation of the viral entry kinetic phenotypes. The significance of these findings in the viral entry field extends beyond NiV to other paramyxoviruses and enveloped viruses.

IMPORTANCE Virus-cell membrane fusion is essential for enveloped virus infections. However, mechanistic viral membrane fusion studies have predominantly focused on cell-cell fusion models, largely due to the low availability of technologies capable of characterizing actual virus-cell membrane fusion. Although cell-cell fusion assays are valuable, they do not fully recapitulate all the variables of virus-cell membrane fusion. For example, drastic differences between viral and cellular membrane lipid and protein compositions and curvatures exist. For biosafety level 4 (BSL4) pathogens such as the deadly Nipah virus (NiV), virus-cell fusion mechanistic studies are especially cumbersome. To circumvent these limitations, we used enzymatic Nipah virus-like-particles (NiVLPs) and developed new flow virometric tools. Our new tools allowed us the high-throughput measurement of viral entry kinetics, glycoprotein proportions on individual viral particles, and receptor-induced conformational changes in viral glycoproteins on viral surfaces. The significance of these findings extends beyond NiV to other paramyxoviruses and enveloped viruses.

INTRODUCTION

Nipah virus (NiV) and Hendra virus (HeV) belong to the Henipavirus genus within the Paramyxoviridae family. The Paramyxoviridae family comprises important human-pathogenic and veterinary enveloped viruses that also include measles, mumps, Newcastle disease, respiratory syncytial, canine distemper, metapneumo-, and human parainfluenza viruses (1–3). NiV and HeV cause respiratory disease and severe acute encephalitis with mortality rates of 40 to 100% in humans. NiV is a priority pathogen in the USA-NIH/NIAID agenda and a potential bioterrorism agent (4, 5). Henipaviruses preferentially infect endothelial and neuronal cells, which express high levels of the host cell receptors ephrinB2 (6, 7) and/or ephrinB3 (8). Ephrin cell receptors are necessary for viral entry and for the cell-cell fusion (syncytium) that results from henipavirus infections. The NiV natural host is the fruit bat primarily from the genus Pteropus, commonly known as the flying fox. Humans have contracted NiV from pigs, directly from contact with bat saliva via contaminated fruit or palm sap, or by direct human-to-human contact (9–11).

Viral entry into host cells is an essential part of viral infections, and for enveloped viruses, viral entry requires virus-cell membrane fusion (3). However, research on the mechanism of receptor-induced membrane fusion has predominantly focused on cell-cell membrane fusion models, in part due to the relative lack of technologies capable of characterizing virus-cell membrane fusion for most enveloped viruses, including the paramyxoviruses (3, 12–14). Although these assays yield valuable information, they do not fully recapitulate all the variables of fusion of actual virions with their target cells. As examples, the viral and cellular membrane lipid and protein compositions and curvatures are drastically different (15, 16). Furthermore, because NiV is a biosafety level 4 (BSL4) pathogen, replication-incompetent forms of NiV that still provide the ability to study virus-cell membrane fusion and viral entry kinetics are highly desirable (17). In the present study, we sought to circumvent these limitations and the surrogate cell-cell fusion systems to enable the study of the mechanism(s) of virus-cell membrane fusion. Thus, we used enzymatic Nipah virus-like-particles (NiVLPs) as a model of paramyxovirus entry and developed new flow virometric tools to measure viral entry kinetics, glycoprotein proportions on individual viral particles, and receptor-induced conformational changes on viral glycoproteins on viral surfaces.

A principal role of the matrix protein (M) of paramyxoviruses is to orchestrate viral assembly and budding (18–20). M lies underneath the viral lipid envelope and is highly abundant (21). Viral assembly generally occurs when M links ribonucleoprotein (RNP) cores with the cytoplasmic tails of viral glycoproteins, mediating the interaction(s) between RNP cores and the envelope of the nascent viral particle (22). In fact, NiV-M is essential for NiV assembly and budding, and multiple NiV-M mutants have been found to abrogate or severely impair these processes (23–27). We and other groups have used recombinant gene expression via plasmid transfection to create NiVLPs expressing the F, G, and M viral proteins (24, 27). Furthermore, we fused the codon-optimized bacterial beta-lactamase (βLa) gene to NiV-M (βLaM) to allow detection of viral entry, specifically inhibited by anti-NiV-F or anti-NiV-G neutralizing antibodies (27). One aim of the present study was to use these enzymatic βLaM-NiVLPs to obtain viral entry kinetic curves for wild-type (WT), hyper-, and hypofusogenic virions.

βLa naturally cleaves antibiotics such as penicillins and cephalosporins (28). An ester derivative of the βLa green-fluorescence substrate CCF2 (CCF2-AM) is membrane permeable (28). CCF2-AM loaded into mammalian cells is hydrolyzed into the membrane-impermeable CCF2. Upon reaction with βLa, CCF2 is further hydrolyzed, causing a disruption in resonant energy transfer and a change from green to blue fluorescence when excited by light (28, 29). For the βLa-M protein, fusion of βLa to the NiV-M N terminus does not interfere with VLP production, as the NiV-M C terminus modulates budding and viral assembly (27). Additionally, a Y105W mutation in the βLa catalytic site increased assay sensitivity nearly 2-fold (27). We previously showed that the βLa-M NiVLPs are functional. Conventional negative-staining transmission electron microscopy of purified NiVLPs revealed their glycoprotein spikes, and the NiVLPs were functional in viral binding and virus-cell entry assays (27, 30). Upon βLa-M NiVLP entry, fusion virus-cell membrane releases βLa-M into the cytosol, allowing cleavage of CCF2-AM and a shift in fluorescence activity (27). Here, we used the βLa-M NiVLPs to obtain viral entry kinetic curves, providing a new and simple way to study paramyxovirus entry.

During entry into cells, NiV's attachment (G) and fusion (F) envelope glycoproteins mediate viral binding to cell receptors and virus-cell membrane fusion, respectively. NiV-G binds ephrinB2 or ephrinB3 cell receptors (6–8, 31), and we recently reported that at least three receptor-induced conformational changes in NiV-G are necessary upon receptor binding to trigger F to execute membrane fusion (32). We also reported the kinetics of two conformational steps in NiV-F upon cell receptor binding to NiV-G, which triggers F from a prefusion to a prehairpin intermediate and ultimately to the six-helix bundle formation that executes membrane fusion (33, 34). Despite the essential role of receptor-induced conformational changes in the glycoproteins of enveloped viruses during virus-cell and cell-cell membrane fusion, technologies that detect glycoprotein conformational changes on actual enveloped virions are cumbersome, when possible (3, 35–39). Here, we show that an adaptation of flow cytometry, which we term “flow virometry,” enabled us to both estimate the relative amounts of F and G on the surfaces or individual NiVLPs and detect receptor-induced conformational changes in NiV-G. This is important, as the composition of envelope glycoproteins on viral surfaces is a factor in the relative capability of a viral particle to enter a host cell.

In the current study, we employ enzymatic NiVLPs and new flow virometric tools to detect and evaluate viral entry kinetics, viral glycoprotein composition, and receptor-induced conformational changes in NiV-G. Our ability to produce NiVLPs that contained either none, one, or both glycoproteins allows us to, as a proof of principle, measure viral entry kinetics, the relative amounts of F and G on individual viral particles, and receptor-induced glycoprotein conformational changes on these viral particles.

MATERIALS AND METHODS

Cell culture.

Human embryonic kidney 293T cells were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). CHOpgsA745 cells were obtained through the American Type Culture Collection and were previously modified to stably express ephrinB2 (CHOB2) (6). CHOpgsA745 cells were grown in minimal essential medium (MEM) containing 10% FBS, while CHOB2 cells were grown in the same medium supplemented with 1 mg G418/ml.

Expression plasmids and vectors.

PCDNA3.1 plasmids expressing the codon-optimized NiV M, F, or G genes were previously constructed and reported (7, 27, 40, 41). The PCDNA3.1 plasmids expressed codon-optimized NiV F, G, and βLaM genes each containing a C-terminal AU1, hemagglutinin (HA), or FLAG tag, respectively (27). To further improve expression of these genes, we subcloned them into the high-protein-expression vector pCAGGS (42). Each gene was isolated through a restriction enzyme digestion between the KpnI and XhoI cloning sites, purified, and then ligated into pCAGGS.

NiVLP production and purification.

NiVLPs were manufactured by polyethylenimine (PEI; Sigma Inc.) transfection of 293T cells. Briefly, 30 μg total DNA plasmids was transfected into 70% confluent 293T cells per 15-cm plate. Several proportions of βLaM:F:G expression plasmids were tested, as indicated in Fig. 1C, as the expression of individual codon-optimized genes varies widely compared to the expression of each gene's wild-type counterparts. For example, when the plasmids expressing βLaM, F, or G were used in the proportions 5:24:1 (in μg; sample labeled 24:1), this resulted in the highest levels of entry over background for the wild-type MFG NiV virions. Thus, this plasmid ratio was utilized to produce the virions used in all subsequent experiments. In the negative controls, the corresponding amount of filler DNA plasmid pCAGGS replaced the missing F or G viral protein DNA. Next, 120 μl of PEI (at 1 mg/ml water) was added to the solution to make a 4:1 (μl/μg) PEI/DNA ratio. The solution was briefly vortexed, allowed to sit at room temperature for 10 min, and then added to the 15-cm plate. Cells were incubated for 36 h before collection of the cell supernatants, which were first cleared by centrifugation at 300 × g and then layered onto 6 ml of 20% sucrose in NTE buffer (100 mM NaCl, 10 mM Tris-HCl [pH 7.5], 1 mM EDTA) in an ultracentrifuge tube. Samples were ultracentrifuged at 100,000 × g for 90 min. After spinning, the supernatant was discarded and the virus was resuspended in 500 μl of 5% sucrose in NTE buffer.

FIG 1.

βLaM allows measurement of viral entry kinetics. (A) Viral entry kinetics of WT NiVLPs (MFG) relative to negative controls (MF and MG). NiVLPs were produced by transfecting NiV βLaM, F, and G expression plasmids into 293T cells. After NiVLP purification, viral entry into target CHO-B2 cells preloaded with the green fluorescent CCF2-AM dye is indicated by a conversion of CCF2-AM to a blue fluorescent dye. The increase in the ratio of blue to green (B/G) fluorescence over more than 3 h is shown. All data points were normalized to the highest B/G ratio, and results from one representative experiment (of three) are shown. NiVLPs containing both F and G (MFG), F only (MF), or G only (MG) were tested. (B) Viral entry kinetics of WT NiVLPs (MFG) with the average background of MF and MG VLPs subtracted. (C) Western blot analysis of viral NiVLP preparations produced by using 5 μg M plasmid and 25 μg of a combination of F and G plasmids at an F/G ratio of 0.8:1, 2.4:1, 8:1, 24:1, or 72:1. Thus, the total amount of DNA expression plasmids per 15-cm dish was 30 μg.

Viral entry kinetics assay.

CHO-B2 cells were collected, washed with phosphate-buffered saline (PBS), and counted. Loading buffer was prepared using a CCF2-AM loading kit (Invitrogen), consisting of 3 μl CCF2-AM substrate in dimethyl sulfoxide (DMSO; 1 mM), 10 μl solution B (100 mg/ml Pluronic-F127 surfactant in DMSO and 0.1% acetic acid), 154 μl solution C (24% [wt/wt] polyethylene glycol [PEG] 400, 18% TR-40 by volume in water), 820 μl DMEM plus 25 mM HEPES (sterile filtered), and 12.5 μl solution D (probenecid in NaOH). CHO-B2 cells were added to the loading buffer (50,000 cells/sample) and incubated at room temperature for 90 min. Loaded cells were washed three times and resuspended in 60 μl of cold fusion buffer (FB; consisting of Hanks' balanced salt solution [HBSS], 25 mM HEPES, 2 mM glutamate, 1 mM CaCl₂, and 2.5 mM probenecid). A black, clear-bottom 96-well plate was placed on ice, and 109 μl FB was added to each well, followed by 11 μl of virus and 60 μl of loaded cells resuspended in FB. Virus was bound at 4°C for 1 h. Viral entry was measured by a Tecan M1000 fluorometer preheated to 37°C. Fluorescence was recorded every 3 min for >3 h. A ratio of blue-to-green (B/G) fluorescence was determined for each well at each time point. The background control B/G ratio of MG and MF viral samples was then subtracted from each sample ratio as indicated, and results were plotted against time.

Detection of protein viral incorporation by Western blot analysis.

VLP samples were also processed for glycoprotein and matrix expression through Western blot analysis. Each lane of a polyacrylamide gel was loaded with 10 μl of virus samples. Glycoproteins NiV-F and NiV-G were blotted in a 1:2,000 dilution of mouse anti-AU1 and rabbit anti-HA primary antibodies, respectively, while NiV-βlaM was blotted 1:500 with mouse anti-β-lactamase in a duplicate gel. Fluorescent secondary antibodies Alexa Fluor 647 goat anti-mouse IgG and Alexa Fluor 488 goat anti-rabbit IgG (Life Technologies, NY) were used at a 1:2,000 dilution (32, 40, 43, 44).

Flow virometry of NiV-F and NiV-G on viral surfaces.

VLP samples were stained with primary mouse anti-HA and rabbit anti-806 antibodies at a 1:300 dilution for 1 h at 4°C. The samples were then washed with FACs buffer (phosphate-buffered saline with 1% fetal bovine serum) and ultracentrifuged twice at 100,000 × g for 1 h at 4°C. Fluorescent secondary antibodies (Li-Cor Biosciences) were used at a dilution of 1:100 and allowed to bind at 4°C for 30 min, and samples were washed and ultracentrifuged once more at the same speed. NiV-F and -G were detected using a Guava easyCyte 8HT flow cytometer, in which the forward and side scatter (FSC and SSC, respectively) settings were slightly modified to detect small particles. The relatively small viral particles were differentiated from suspension buffer debris by gating in the forward versus side scatter plot (see Fig. 3A and B).

FIG 3.

Flow virometry of NiVLPs. (A and B) Graphs of forward (x axis) versus side (y axis) scatters for the buffer sample containing only 5% sucrose-NTE buffer without any VLPs (from a mock transfection) (A) or containing VLPs (B). These plots assist in gating out background particles in the buffer from our analysis, making it possible to distinguish and incorporate the VLPs into our gate. (C to F) Representative dot plots for the incorporation of NiV-F and -G on NiVLPs' surfaces. x and y axes show the levels of NiV-F and NiV-G incorporation on viral surfaces, respectively. (C) NiVLPs made with βLaM only (M only), showing the percentages of cells in each quadrant; (D) NiVLPs made with βLaM and NiV-G; (E) NiVLPs made with βLaM and NiV-F; (F) NiVLPs made with βLaM, NiV-G, and NiV-F. Representative dot plots from at least three independent experiments are shown.

Detection of ephrinB2-induced conformational changes in viral surface NiV-G via antibody binding using flow virometry.

Wild-type NiVLPs (MFG samples) were bound to ephrinB2 at various concentrations (10 nM, 1 nM, 0.1 nM, or 0 nM ephrinB2) for 1 h at 4°C. At 25 to 37°C, eprhinB2 binding induces spatiotemporally sequential and distinct conformational changes in NiV-G, detected with Mab213 and Mab45, respectively (32, 34). Therefore, after incubation of pre-ephrinB2-bound wild-type NiVLPs for 15 min at 37°C with either antibody at 1:300 dilution, a fluorescent secondary antibody (as described above) was used at a dilution of 1:100 as described above to detect Mab213 or Mab45 in separate tests. Washes were performed, and antibody binding was detected by flow virometry as described in “Flow virometry of NiV-F and NiV-G on viral surfaces.”

RESULTS

Enzymatic NiVLPs can be used to quantify viral entry kinetics.

The strategy of fusing βLa to a viral gene has been used to detect viral entry into mammalian cells for enveloped viruses of several families (27, 29, 45). One of the aims of the present study was to further use βLa-NiVLPs to measure viral entry kinetics. For NiV and influenza virus, βLa VLPs have been produced by fusing βLa to the respective matrix protein (βLaM) and transfecting into 293T cells mammalian expression plasmids encoding the respective βLaM and envelope glycoproteins (27, 45). βLaM VLP-induced virus-cell membrane fusion allows βLaM conversion of the CCF2-AM substrate to a blue fluorescent product (27, 45). In the present study, we similarly produced βLaM NiVLPs in 293T cells by expressing codon-optimized genes for the NiV-F, NiV-G, and βLa-M with the Y105W mutation. These optimizations have been shown to increase the sensitivity of the entry assay (27, 40). We then further used the βLaM NiVLPs to measure viral entry kinetics in a high-throughput assay.

We previously reported that expression of βLa in target 293T cells allowed the quantification of real-time cell-cell fusion kinetics (40, 43). Here, to synchronize viral entry, we first allowed binding at 4°C of the NiVLPs to target CHO-B2 cells, which stably express the NiV and HeV ephrinB2 receptor (6, 7). Then, NiVLPs were allowed to induce virus-cell membrane fusion and content mixing by incubating the virus-cell mixture at 37°C. Viral entry was detected at 37°C using a fluorimeter 96-well plate reader by measuring green and blue fluorescence intensities every 3 min for >3 h. VLPs containing only βLaM and F (MF) or βLaM and G (MG) were used as negative controls (Fig. 1A). The blue-to-green (B/G) fluorescence ratios were normalized to the highest B/G ratio for the MFG sample at 3 h. The B/G ratios increased over time at a greater rate for the MFG samples than for the negative-control MG or MF samples, indicating that this assay is sensitive enough to measure viral entry kinetics over time. The two negative-control samples yielded similar B/G ratios throughout the experiment. Therefore, we then subtracted the average of the B/G ratios for the MG and MF negative-control samples from the B/G ratio of the MFG samples at every time point, allowing us to quantify the kinetics of viral entry of the wild-type NiVLPs (Fig. 1B). Roughly, a positive signal began between 15 and 30 min from the start of the 37°C incubation period and increased steadily until the beginning of a plateau at approximately 90 min, reaching a maximum at approximately 3 h. It is noteworthy that the 15- to 30-min delay in the start of the positive signal was longer than the delay (∼10 min) observed in equivalent βLa cell-cell fusion assays (33, 40, 43). After 3 h, no increase in the blue/green fluorescence ratio was observed. It is also noteworthy that we produced and tested NiVLPs by transfecting various ratios of M/F/G expression plasmids. The virions produced using these ratios incorporated various amounts of M, F, and G, generally correlating with the amounts of plasmids utilized (Fig. 1C). Importantly, the highest viral entry kinetic signal/noise ratios were obtained for virions produced by transfecting the 5:24:1 ratio of expression plasmids (Fig. 1B), and this plasmid ratio was further utilized to produce the virions for all subsequent studies.

Sensitive NiVLPs can distinguish differences in the entry kinetics of hyperfusogenic versus hypofusogenic NiV-F phenotypes.

We then asked whether the enzymatic NiVLP entry assay is capable of differentiating the kinetics of viral entry among wild-type, hyperfusogenic, and hypofusogenic virions. We reported that removal of two N-glycans from the NiV-F ectodomain (mutant F3F5) resulted in an NiV-F mutant protein with a >5-fold hyperfusogenic phenotype in cell-cell fusion assays and in viral entry assays, using NiV-vesicular stomatitis virus (NiV/VSV) pseudotyped viruses (40). Similarly, we reported that single-point mutations in the cytoplasmic tail KKR motif of NiV-F resulted in >3-fold hyperfusogenic (mutant K1A) or hypofusogenic (mutants K2A and R3A) phenotypes (43). Using expression plasmids for these mutant NiV-F proteins, we prepared NiVLPs containing wild-type NiV-βLaM, NiV-G, and the hyperfusogenic (K1A or F3F5) or hypofusogenic (K2A or R3A) mutant NiV-F proteins.

Infection of CHO-B2 cells with the wild-type or mutant NiVLPs yielded viral entry kinetics remarkably similar to those previously reported in cell-cell fusion kinetics assays for the same mutants (40, 43) (Fig. 2). For example, compared to the wild-type NiVLPs, the hyperfusogenic mutant F3F5 VLPs yielded a higher maximum signal and a steeper kinetic rate of viral entry at the time interval between 30 and 90 min (P < 0.01). We obtained the maximal fusion rates (those between times 30 min and 90 min) and maximum plateau signals (at 180 min) for each viral entry curve and compared mutant to wild-type F VLP values by performing Student's t tests and applying Bonferroni's correction factors. In contrast, the K1A mutant VLPs yielded maximum signal and kinetic rate of viral entry similar to those of the wild-type NiVLPs (40, 43) (Fig. 2) (P > 0.05). Compared to the wild-type NiVLPs, the hypofusogenic mutant K2A and R3A VLPs yielded significantly lower maximum signals and slower kinetics of viral entry (43) (Fig. 2) (P < 0.05). These results indicate that the βLaM VLPs can be used to measure the relative levels and kinetics of viral entry for hyperfusogenic and/or hypofusogenic NiVLPs.

FIG 2.

Viral entry kinetic curves of WT and mutant NiVLPs can detect differences in rates of viral entry. NiVLPs containing NiV βLaM, G, and WT or mutant NiV-F expression plasmids were produced and bound to target CHO-B2 cells loaded with CCF2-AM dye as described for Fig. 1. The change in the B/G ratio was plotted over the course of 3 h. The average background was subtracted, and results were normalized to the maximal B/G ratio of WT NiVLPs (MFG). Results from one representative experiment (of three) are shown.

As the extent of cell-cell fusion depends on the relative levels of glycoproteins on the cell surface (46), it is likely that virus-cell membrane fusion is also dependent on the relative levels of incorporation of the viral glycoproteins on individual virions that infect a cell. Therefore, we next pursued a technique capable of detecting the levels of the NiV-F and -G glycoproteins on individual viral particles. Western blot analysis, for example, is capable of measuring the overall levels of F and G in a viral preparation (30, 40, 43) but is unable to discern the proportions of NiVLPs that contain both F and G, F only, G only, or neither F nor G. Flow virometry, a variation of flow cytometry, allowed us to achieve our goal.

Flow virometry can quantify the relative levels of NiV-F and -G on individual NiVLP surfaces.

To determine whether the differences in viral entry kinetics observed for the various mutant virions were due to differences in the levels of glycoprotein incorporation into virions, we measured the relative amounts of wild-type NiV-G and wild-type or mutant NiV-F on the virion surfaces. Using the advantages of flow cytometry, we adapted this technique by slightly increasing the sensitivity of the FSC and SSC (which measure particle size and granularity, respectively), relative to the FSC and SSC voltages used in regular flow cytometry. Thus, we developed flow virometry as a new approach to determine the incorporation levels of NiV F and G on the surface of individual VLPs.

Briefly, we fluorescently labeled NiVLPs using glycoprotein-specific antibodies as detailed in Materials and Methods. We then used the high-throughput capabilities of flow cytometry to measure the 488-nm and 647-nm fluorescence intensities of 5,000 individual NiVLPs per sample in less than 1 min. Despite having performed several washes and ultracentrifugations of our fluorescently labeled VLPs, it is entirely possible that small cell debris or other small particles in our filtered viral suspension NTE buffer might have been present in our samples and thus could be mistaken for VLPs. Therefore, we first verified that NiVLPs were distinguishable from background noise resulting from small particles in the NTE buffer in which the VLPs were resuspended. The small particles in the buffer (Fig. 3A, from mock-transfected cells, particles outside the elliptical gate) were clearly distinct from VLPs (Fig. 3B, particles inside the elliptical gate). Clearly, the sample with NiVLPs (Fig. 3B) contained a much greater proportion of high FSC and high SSC particles (primarily VLPs) than the sample with NTE buffer alone (mock transfected) in the same gate (Fig. 3A).

NiVLPs gated as in Fig. 3B were then plotted for their F signals (x axis) against their G signals (y axis). Similar plots were generated for M only, MG, MF, and MFG viral samples (Fig. 3C to F, respectively). As expected, the MFG and MF virions had higher F incorporation signals than the F-negative control “M only” or “MG” virions. Similarly, the MFG and MG virions had higher G signals than the G-negative control “M only” or “MF” virions. It is worth noting that the proportion of G-positive VLPs appeared to be greater for the MFG VLPs than for the MG VLPs, suggesting that the presence of F positively affects the incorporation of G into virions. We also noticed that in the MFG VLP samples, generally the VLPs with higher G signals corresponded with VLPs with higher F signals. These data agree with a prior report that suggests that NiV-F can positively affect the levels of NiV-G in virions, as observed by Western blotting (24).

By gating into the bottom left quadrant the majority of virions that contained no F or G using a four-quadrant gate, as shown in Fig. 3C to F, the percentages of virions positive for F and G (F⁺G⁺), F only (F⁺), or G only (G⁺) or those negative for F and G (F⁻G⁻) were calculated. We know that for the negative-control M-only viral sample (Fig. 3C), the percentage of F⁻G⁻ (bald particles) should be 100%, and the percentages of G⁺, F⁺G⁺, and F⁺, corresponding to virions in quadrants 1, 2, and 4, respectively, should be zero. Therefore, the percentages in quadrants 1, 2, and 4 for the M-only sample, 1%, 12%, and 6%, respectively, were the background subtracted from the percentages in the respective quadrants in all other viral samples. The final background-corrected percentages of F⁺G⁺, G⁺, and F⁺ VLPs were plotted in pie charts and are shown in Fig. 4A to G for the different VLP samples. Although this calibration step undoubtedly increased the accuracy of our results, relatively small background stains remained (3% G⁺ in the MF sample, and 6% F⁺ in the MG sample [Fig. 4B and C]). These background-positive stains suggest that the level of inaccuracy of our estimated measurements is approximately equal to or less than 6%.

FIG 4.

Proportions of NiVLPs containing surface NiV-G and/or NiV-F. (A to G) Percentages of NiVLPs containing both F and G (F⁺G⁺), F only (F⁺), G only (G⁺), or neither F nor G (bald particles, F⁻G⁻). As M-only particles (Fig. 3C) do not contain any F or G, we can safely use the percentages of VLPs in Fig. 3C quadrants 1, 2, and 4 as background for the percentages of VLPs in each of these quadrants, respectively, for all NiVLP samples that contain F and/or G. The final background-subtracted percentages of NiVLPs that contain both NiV F and G (F⁺G⁺, blue), F only (F⁺, red), G only (G⁺, green), or neither F nor G (F⁻G⁻, magenta) are shown in pie charts. (H) Western blot analysis of all NiVLP samples, showing the incorporation levels of NiV-F and NiV-G into the NiVLP samples analyzed. Results from one representative experiment (of three) are shown.

Interestingly, the wild-type MFG samples had the highest percentage of F⁺G⁺ VLPs (∼50%), while the F mutant viral samples had lower proportions of F⁺G⁺ VLPs (14 to 34%). It has been reported that for NiV, not only M but also F plays an important role in viral assembly and budding. The same report suggested that F may assist in the recruitment of G into VLPs (24). Our data suggest that the mutant NiV-F proteins tested are less efficient than the wild-type NiV-F in executing VLP assembly and budding and/or recruiting G into VLPs.

To complement our studies, a Western blot analysis of the various VLP samples was performed (Fig. 4H). Overall, for most samples the relative amounts of F and G in the viral samples were in agreement with those calculated by flow virometry. For example, the presence of F appeared to enhance the incorporation of G into virions, as observed when comparing MG virions with virions that also incorporated F, such as MFG or K1A, K2A, and R3A VLPs (Fig. 4H). One exception was mutant F3F5, which did not enhance the overall incorporation of G into VLPs. Also, the incorporation of the F3F5 mutant protein itself was lower than those of the wild-type or K1A, K2A, or R3A mutant F proteins (Fig. 4H). This may contribute to the relatively lower levels of G incorporation into the F3F5 virions. In addition, the level of G-only virions in the F3F5 viral sample (8%, Fig. 4D) was greater than those in other viral samples. These data suggest that higher levels of G and F3F5 in this viral sample are found in separate viral particles compared to the wild-type virions, in agreement with NiV-F having a role in recruiting G into virions.

It is important to note that the relative proportions of F⁺G⁺ VLPs in the viral samples as detected by flow virometry do not need to match the relative amounts of F and G observed by Western blotting, as mutant virions may have relatively different distributions or densities of F and G on their surfaces. Although Western blot analysis is unable to provide information on the relative distributions or densities of F and G on viral surfaces, flow virometry is. For example, while the amounts of F and G Western blot signals in the K2A viral sample appeared to be similar to those observed in the other viral samples (Fig. 4H), the percentage of F⁺G⁺ viral particles in that sample was the lowest (Fig. 4F). An explanation for these data may be that the K2A sample has a lower proportion of virions, with more-densely packed F and G. In contrast, for the F3F5 sample, the F and G proteins may be less densely packed in the VLPs that contain them, thus yielding lower F and G Western blot signals than the other mutant or wild-type virions tested (Fig. 4H). In fact, when we quantified the levels of F and G glycoproteins in these viral samples, we first observed that the levels of F and G per particle increased with particle size for the MFG samples (Fig. 5A to E). In addition, the levels of F and G were lower in the F3F5 particles, and at least for the larger viral particles, the levels of F were greater in the K2A samples than the respective levels in the wild-type viral samples (Fig. 5). These data indicate that a combination of Western blot analysis and flow virometry is useful for fully characterizing glycoprotein densities in viral particles.

FIG 5.

Quantification of NiV-F and NiV-G incorporation onto VLPs of various sizes. (A) A sample of MFG VLPs is separated into three gates based primarily on their forward scatter (small versus medium versus large VLPs). Dot plots of F incorporation versus G incorporation for the small (B), medium (C), or large (D) VLPs from panel A are shown. (E) Mean fluorescence intensity (MFI) of NiV-F or NiV-G viral surface levels obtained as shown in panels A to D. Averages and standard deviations from four experiments are shown.

Importantly, in these two examples, despite the relatively lower levels of F and G in the F3F5 sample and higher levels of F and G in the K2A samples, we found higher and lower levels of entry of these VLPs, respectively, than for the wild-type VLPs. This corroborates that the increased or decreased levels of viral entry, respectively, for the F3F5 and K2A VLPs are likely due to the intrinsic membrane fusion-inducing capacities of these F mutant proteins and not simply a result of the levels of incorporation of F and G at the surface of these virions. These results also indicate that overall the hyperfusogenic and hypofusogenic phenotypes of the F3F5 and K2A samples in cell-cell fusion assays are reproducible in viral entry VLP assays.

Detection of specific receptor-induced conformational changes in NiV-G on NiVLP surfaces.

We have reported that incubation of F/G-containing membranes with receptor-containing membranes at 4°C can be used to synchronize cell-cell or virus-cell membrane fusion (33, 34). In addition, we recently reported that receptor binding induces at least three spatiotemporal conformational changes in NiV-G on cell surfaces, necessary to induce membrane fusion, detected using conformational monoclonal antibodies and flow cytometric analyses (32, 34). Furthermore, we reported that receptor binding to NiV-G induces conformational changes in F on viral surfaces, detectable by Raman spectroscopy (30). However, Raman spectroscopy was unable to detect receptor-induced conformational changes in NiV-G on NiV virions.

Here, taking advantage of flow virometry, we asked whether we could detect receptor-induced conformational changes in NiV-G on viral surfaces of NiVLPs. We used the same conformational antibodies that we had previously used to detect two distinct receptor-induced conformational changes in NiV-G on cell surfaces, Mab213 and Mab45 (32, 34). Upon receptor ephrinB2 binding to NiV-G, the binding of Mab213 to NiV-G decreases and the binding of Mab45 to NiV-G increases, in this spatiotemporal order (32). A soluble form of the ephrin-B2 receptor is sufficient to induce receptor-induced conformational changes in NiV-G on cell surfaces (32, 34). Therefore, we measured the ability of soluble ephrinB2 to induce conformational changes in NiVLPs. Congruent with the cell surface conformational changes previously reported, increasing amounts of soluble ephrinB2 receptor binding to NiVLPs MFG decreased the relative binding levels of Mab213 to NiV-G but increased the relative binding levels of Mab45 to NiV-G (Fig. 6). To our knowledge, this is the first time that flow virometry is utilized to both quantify the relative proportions of glycoproteins in a viral sample and detect receptor-induced conformational changes in individual virions.

FIG 6.

Quantification of ephrinB2-induced conformational changes in NiVLP surface NiV-G. Plot of the log of ephrinB2 molarity (x axis) versus the percentage of normalized binding of conformational antibodies Mab213 or Mab45 (y axis). Binding of Mab213 or Mab45 was measured in separate experiments at increasing concentrations of soluble ephrinB2 prebound at 4°C for 1 h and then allowed to exert a conformational change in NiV-G on NiVLP surfaces at 37°C for 15 min. Upon receptor ephrinB2 binding to NiV-G, the binding of Mab45 to NiV-G was enhanced, while the binding of Mab213 to NiV-G was decreased, consistent with results obtained on cell surfaces (32, 34, 44). Results from one representative experiment (of three) are shown.

DISCUSSION

The enveloped virus glycoproteins are responsible for viral entry (virus-cell membrane fusion) and for syncytium formation during viral infections (cell-cell membrane fusion), among other functions. Virus-cell membrane fusion is essential in the life cycle of enveloped viruses (3, 12). However, many of the steps that viral glycoproteins take during viral entry and syncytium formation remain unclear (2, 3, 12). The receptor-induced conformational changes that enveloped glycoproteins undergo during membrane fusion have remained of great interest for decades. Understanding those glycoprotein conformational changes is leading to the development of antiviral drugs that block viral entry. For example, enfuvirtide is an FDA-approved peptide drug currently used against HIV-1 to block a conformational change of the fusion protein gp41, required for membrane fusion (47, 48). Therefore, understanding the receptor-induced conformational changes that enveloped virus glycoproteins undergo during membrane fusion will likely expand the repertoire of potential antiviral therapeutic targets and agents. Identifying the specific conformational changes needed for membrane fusion in a rapid fashion offers many potential applications in the field of virology (3). To our knowledge, this is the first study that employs the tools to simultaneously detect viral entry kinetics, viral surface relative compositions, and receptor-induced glycoprotein conformational changes while the glycoproteins are embedded on enveloped virions.

For the paramyxoviruses, what we currently know about receptor-induced conformational changes in their enveloped glycoproteins has been studied primarily using soluble glycoprotein forms (3, 49) or glycoproteins expressed on cell surfaces (3, 12, 34, 50). Although these studies provide valuable information, they do not fully replicate virus-cell membrane fusion. For example, the viral and cellular membrane lipid and protein compositions and curvatures are drastically different, with virions having more-positive curvatures than cells (15, 16). In addition, the use of relatively high temperatures to induce conformational changes, for example, increases the uncertainty of the significance of the results observed. In contrast, the glycoproteins in the virion envelope have a truly native conformation: full-length, embedded in the membrane, and in the proper context with other envelope proteins. Although there are viruses for which assays have been developed to study conformational changes in virions (for examples, see references 35, 36, 37, 38, and 39), by far, such methods tend to be complex and time-consuming. In the present study, we developed tools to study viral glycoprotein conformational changes in virions in situ that are selective, accurate, and rapid enough to be appropriate for studies conducted in real time with equipment available in most laboratories.

Obtaining X-ray crystal structural information from full-length viral envelope glycoproteins is problematic because their hydrophobic transmembrane domains are embedded in a lipid membrane. Thus, structural studies have focused on ectodomain glycoprotein forms. The study of receptor-induced viral glycoprotein conformational changes have also focused on viral glycoprotein soluble ectodomain forms (for examples, see references 49, 51, and 52) or on full-length wild-type glycoproteins expressed on cell surfaces (for examples, see references 32 and 34). Despite soluble ectodomains normally binding their respective cell receptors, how accurately their structures and receptor-induced structural changes compare with their membrane-bound full-length wild-type counterparts is uncertain. Furthermore, soluble glycoprotein ectodomains tend to adopt postfusion conformations, sometimes limiting our ability to characterize essential receptor-induced conformational changes (49, 53). Analysis of receptor-induced conformational changes of wild-type full-length glycoproteins embedded in cellular or viral membranes is preferred. However, even then, differences between the roles of the receptor-induced conformational changes in virus-cell and in cell-cell membrane fusions exist. Therefore, the study of receptor-induced conformational changes on actual viral surfaces is highly desirable but currently uncommon for enveloped viruses due to technical limitations (35–39) and extremely uncommon for the paramyxoviruses, which contain two envelope glycoproteins incorporated at fairly unknown proportions into virions.

We recently showed that confocal micro-Raman spectroscopy can be used to simultaneously detect distinct glycoproteins and analyze viral glycoprotein conformational changes in NiV-F in NiVLPs or NiV envelope glycoproteins pseudotyped onto vesicular stomatitis virus (VSV) particles in situ (30). Unfortunately, the Raman spectroscopy studies were able to detect only NiV-F but not NiV-G receptor-induced conformational changes. In addition, this technique is not readily available in the average virology laboratory. The new flow virometry studies here were able to detect conformational changes in NiV-G and likely have broader utility, as they require only a flow cytometer. In addition, the combination of viral glycoprotein composition and receptor-induced viral glycoprotein conformational changes in a single assay is possible.

Interestingly, comparison of NiVLP entry kinetics with cell-cell fusion kinetics suggested that overall viral entry may occur at a slower kinetic rate than cell-cell fusion. While the cell-cell fusion blue-to-green fluorescence shifts over background appeared approximately 10 min after the start of the cell-cell fusion reaction at 37°C (40, 43), viral entry blue-to-green fluorescence shifts over background appeared 15 to 30 min after the start of the virus-cell fusion reaction (Fig. 1B and 2). There are two potential explanations that we can envision (not mutually exclusive) for this time difference. (i) We speculate that the relatively more positive curvature of virion membranes compared to that of cell membranes would require greater levels of energy to undergo the negative curvature of “membrane dimples” postulated as necessary during the hemifusion and fusion pore formation steps of the membrane fusion cascade (3, 54). (ii) Viral entry may require an additional step, such as partial M dissociation, after membrane fusion for the βLaM protein to be in direct contact with its CCF2-AM substrate (3). How NiVLP entry kinetics compares to live NiV entry kinetics remains to be determined.

It is noteworthy that the F3F5 mutant showed significantly higher levels of viral entry than the wild-type MFG virions despite the F3F5 virions having the lowest levels of incorporation of F and G (Fig. 4 and 5). On the other hand, while the K1A mutant virions had levels of F and G incorporation similar to those of the wild-type MFG virions, the K1A virions yielded levels of viral entry relatively similar to or only slightly higher than those of the wild-type MFG virions (Fig. 2). This is remarkably consistent with the levels of cell-cell fusion previously reported for these mutants (40, 43). Because the F3F5 and K1A mutants both yielded approximately 4- to 6-fold-higher levels of cell-cell fusion than their wild-type counterparts, the K1A mutant must be hyperfusogenic for reasons other than its kinetics of membrane fusion, such as fusion pore expansion, as previously discussed (33).

It is also noteworthy that while the R3A mutant yielded cell-cell fusion kinetic rates above background, the K2A viral entry kinetic curve was basically indistinguishable from background levels, in spite of mutant K2A being incorporated into virions (Fig. 4). In the cell-cell fusion kinetics assays, the K2A mutant was also observed to be more hypofusogenic than the R3A mutant, but the K2A cell-cell fusion kinetics was still above background levels (43). All these data put together indicate either that the sensitivity of the viral entry kinetics assay may be lower than the cell-cell fusion assay or that the NiV-F protein behaves differently on the surface of cells and on virions.

A combination of flow virometry, Western blot analysis, and Raman spectroscopy promises to be highly informative in the characterization of enveloped virions, their relative glycoprotein compositions, and their receptor-induced conformational changes. Each technique offers its own advantages and disadvantages, and their results complement one another. For example, the combination of Western blot and flow virometry analyses suggested that the K2A mutant virions may have a lower proportion of virions with more-densely packed F and G than the wild-type NiVLPs (Fig. 4 and 5). Interestingly, we previously reported that the K2A mutant F protein had a looser association with lipid rafts than the wild-type NiV-F protein (43). As lipid rafts have been shown to be involved in assembly and budding for other enveloped viruses, it is possible that the relatively different levels of association of mutant K2A with lipid rafts may affect the distribution of F and G on the viral surface.

Importantly, our studies show that flow virometry can monitor receptor-induced conformational changes in viral glycoproteins. This may potentially provide a simpler and more rapid method for monitoring how hypo- or hyperfusogenic enveloped virus mutants function during viral entry. Important to the characterization of wild-type NiVLPs, we showed that approximately 50% of NiV virions had both G and F glycoproteins on their surface. Despite this relatively high percentage of F⁺G⁺ virions, it is possible that most of these virions do not contain the right proportion of F and G on their surfaces or the correct arrangement of F and G on the viral surfaces for the optimal likelihood of successful viral entry. Interestingly, the presence and levels of NiV-F were directly correlated with the presence and levels of NiV-G (Fig. 3F). This suggests that the incorporation of NiV-G is highly dependent on the incorporation of NiV-F on a per-virion basis. Once again, how the distribution of F and G on NiVLPs compares to that on live NiV remains to be determined.

In summary, in the present study we showed the feasibility of performing viral entry kinetic studies using VLPs. We showed that such technologies are sensitive enough to dissect differences between hyperfusogenic and hypofusogenic mutant VLPs. We described how flow virometry can be used to assess the glycoprotein composition of VLPs, aiding the interpretation of viral entry kinetics results. Furthermore, we showed that flow virometry can be used to assess receptor-induced conformational changes in the attachment glycoprotein. Therefore, the technologies described here provide fast, practical, and sensitive methods to study viral entry, composition, and receptor-induced conformational changes. Our results promise to expedite the characterization of VLPs to be used in studies of vaccine development (for an example, see reference 55), viral entry, and receptor-induced glycoprotein conformational changes that occur during membrane fusion on actual viral particles. These advances apply not only to NiV but also to other important paramyxoviruses and other enveloped viruses, promising to significantly advance the viral entry field.