Experimental Evolution To Isolate Vaccinia Virus Adaptive G9 Mutants That Overcome Membrane Fusion Inhibition via the Vaccinia Virus A56/K2 Protein Complex

It remains unclear how the multiprotein entry fusion complex of vaccinia virus mediates membrane fusion. Moreover, vaccinia virus contains fusion suppressor proteins to prevent the aberrant activation of this multiprotein complex. Here, we used experimental evolution to identify adaptive mutant viruses that overcome membrane fusion inhibition mediated by the A56/K2 protein complex. We show that the H44Y mutation of the G9 protein is sufficient to overcome A56/K2-mediated membrane fusion inhibition. Treatment of virus-infected cells at different pHs indicated that the H44Y mutation lowers the threshold of fusion inhibition by A56/K2. Our study provides evidence that A56/K2 inhibits the viral fusion complex via the latter’s G9 subcomponent. Although the G9^H44Y mutant protein still binds to A56/K2 at neutral pH, it is less dependent on low pH for fusion activation, implying that it may adopt a subtle conformational change that mimics a structural intermediate induced by low pH.

ABSTRACT

For cell entry, vaccinia virus requires fusion with the host membrane via a viral fusion complex of 11 proteins, but the mechanism remains unclear. It was shown previously that the viral proteins A56 and K2 are expressed on infected cells to prevent superinfection by extracellular vaccinia virus through binding to two components of the viral fusion complex (G9 and A16), thereby inhibiting membrane fusion. To investigate how the A56/K2 complex inhibits membrane fusion, we performed experimental evolutionary analyses by repeatedly passaging vaccinia virus in HeLa cells overexpressing the A56 and K2 proteins to isolate adaptive mutant viruses. Genome sequencing of adaptive mutants revealed that they had accumulated a unique G9R open reading frame (ORF) mutation, resulting in a single His44Tyr amino acid change. We engineered a recombinant vaccinia virus to express the G9^H44Y mutant protein, and it readily infected HeLa-A56/K2 cells. Moreover, similar to the ΔA56 virus, the G9^H44Y mutant virus on HeLa cells had a cell fusion phenotype, indicating that G9^H44Y-mediated membrane fusion was less prone to inhibition by A56/K2. Coimmunoprecipitation experiments demonstrated that the G9^H44Y protein bound to A56/K2 at neutral pH, suggesting that the H44Y mutation did not eliminate the binding of G9 to A56/K2. Interestingly, upon acid treatment to inactivate A56/K2-mediated fusion inhibition, the G9^H44Y mutant virus induced robust cell-cell fusion at pH 6, unlike the pH 4.7 required for control and revertant vaccinia viruses. Thus, A56/K2 fusion suppression mainly targets the G9 protein. Moreover, the G9^H44Y mutant protein escapes A56/K2-mediated membrane fusion inhibition most likely because it mimics an acid-induced intermediate conformation more prone to membrane fusion.

IMPORTANCE It remains unclear how the multiprotein entry fusion complex of vaccinia virus mediates membrane fusion. Moreover, vaccinia virus contains fusion suppressor proteins to prevent the aberrant activation of this multiprotein complex. Here, we used experimental evolution to identify adaptive mutant viruses that overcome membrane fusion inhibition mediated by the A56/K2 protein complex. We show that the H44Y mutation of the G9 protein is sufficient to overcome A56/K2-mediated membrane fusion inhibition. Treatment of virus-infected cells at different pHs indicated that the H44Y mutation lowers the threshold of fusion inhibition by A56/K2. Our study provides evidence that A56/K2 inhibits the viral fusion complex via the latter’s G9 subcomponent. Although the G9^H44Y mutant protein still binds to A56/K2 at neutral pH, it is less dependent on low pH for fusion activation, implying that it may adopt a subtle conformational change that mimics a structural intermediate induced by low pH.

INTRODUCTION

Vaccinia virus is a complex enveloped virus with a large double-stranded DNA genome of approximately 190 kb. Vaccinia virus produces two infectious forms of viral particles in infected cells, termed mature virus (MV) and extracellular virus (EV) (1). MV has a single membrane that is derived from the endoplasmic reticulum (2) and consists of more than 20 membrane proteins, including four attachment proteins (A27 [3], H3 [4], D8 [5], and A26 [6]) and an entry fusion complex (EFC) of 11 components (A16 [7], A21 [8], A28 [9], G3 [10], G9 [11], H2 [12], J5 [13], L5 [14], O3 [15], L1 [16], and F9 [17]) that mediates fusion between host and viral membranes during virus entry (18–20). EV has an additional Golgi-derived membrane that, upon rupturing at the cell surface or becoming degraded in acidic endosomes, exposes the internal EFC to mediate membrane fusion with host cells (21, 22). Therefore, both MV and EV require identical viral EFCs to mediate membrane fusion with host cells (20, 23), evidenced by the strong conservation of EFC components throughout the poxvirus family (24).

Although the viral EFC is required for membrane fusion of MV and EV with host cells, previous studies have identified two sets of viral regulatory proteins that can curtail membrane fusion, i.e., A26 and the A56/K2 protein complex. These regulatory proteins employ an acid-sensitive mechanism to inhibit the EFC-mediated fusion activity of MV and EV (25–27). The MV-specific A26 protein binds to individual A16 and G9 proteins of the EFC and suppresses EFC fusion activity at neutral pH (27, 28). Only upon MV being internalized into endosomes can the acidic environment of these organelles trigger conformational changes in the A26 protein to induce EFC fusion activation and subsequent viral membrane fusion with endosomes, releasing viral cores into the cytoplasm (29). Unlike A26, the A56/K2 protein complex is expressed on cell surfaces in the early stage of virus infection to inhibit EV-mediated fusion to infected cells in a process termed “superinfection interference” (25, 30–35). Overexpression of A56 and K2 on the surfaces of HeLa cells was found to inhibit not only vaccinia EV cell entry but also MV entry (36). Coimmunoprecipitation (co-IP) experiments on the A56/K2 complex have revealed an association with A16/G9 subunits of the viral EFC (37, 38), and consistently, deletion of the A56R or K2L open reading frames (ORFs) from the viral genome resulted in cell-cell fusion of infected cells (32, 33, 39).

Both A26-mediated and A56/K2-mediated fusion inhibition are sensitive to acid conditions, which induce the dissociation of these viral fusion suppressor proteins from the EFC and prompt “acid-induced membrane fusion.” The crystal structure of the A26 protein was recently resolved (29), showing that when residues His48 and His53 of the N-terminal α2-helical domain become protonated at low pH, they form His-cation pairs with nearby basic amino acids that induce acid-dependent repulsion to create the conformational changes that result in fusion activation. Further in vitro mutagenesis and mutant virus characterization clarified the molecular mechanism by which MV undergoes acid-induced membrane fusion (29). In contrast, it had been unclear how the A56/K2 protein complex mediates membrane fusion inhibition and if acid conditions trigger similar conformational changes of A56/K2 to abrogate the inhibition of EV membrane fusion.

In order to understand how the A56/K2 protein complex inhibits the viral EFC, we employed an experimental-evolution strategy involving serial passaging of vaccinia virus in cells overexpressing A56/K2 to identify adaptive mutant viruses that overcome A56/K2-mediated fusion inhibition. Subsequent viral genome sequencing of these adaptive mutant viruses revealed the mutation and consequent mechanism allowing these mutant viruses to evade A56/K2-mediated inhibition.

RESULTS

Expression of A56/K2 on HeLa cell surfaces inhibits WRΔA26 entry.

We performed experimental evolution to select for and identify adaptive vaccinia mutant viruses that could overcome the fusion inhibition mediated by the A56/K2 complex. Previously, Wagenaar et al. showed that stable expression of A56 and K2 in uninfected cells is sufficient to prevent virus entry and cell fusion (36). Therefore, we used lentiviral vectors to introduce the mammalian codon-optimized A56 and K2 ORFs into HeLa cells. We established a stable cell line, named HeLa-A56/K2, expressing high levels of the A56 and K2 proteins on cell surfaces, as detected by fluorescence-activated cell sorting (FACS) (Fig. 1A) and by immunofluorescence staining using anti-A56 and anti-K2 antibodies (Fig. 1B). Next, we chose to infect cells with WRΔA26 virus, and not the wild-type (WT) Western Reserve (WR) virus, for two reasons. First, both A26 and A56/K2 bind to the G9/A16 subunits of the EFC, raising the possibility that A26 on wild-type WR MV particles may interfere with the binding of MV to the A56/K2 protein complex on cell surfaces during experimental passaging. Second, purified EV particles specifically lack A26 protein (40), so by passaging WRΔA26 MV particles on HeLa-A56/K2 cells, we could closely approximate superinfection interference of EV entry. We infected HeLa and HeLa-A56/K2 cells with MV of WRΔA26-Venus-A4-mCherry at a multiplicity of infection (MOI) of 0.1 PFU per cell and monitored the expression of the viral early Venus marker and late A4-mCherry genes by FACS at 2 h postinfection (hpi) and 8 hpi, respectively (Fig. 1C and D). The mean fluorescence intensity in HeLa cells was set as 100% to normalize the relative fluorescence intensity in HeLa-A56/K2 cells. The data revealed that viral early gene expression in HeLa-A56/K2 cells was reduced to 22% of that in HeLa cells (Fig. 1C), and viral late gene expression in HeLa-A56/K2 cells was 11% of that in HeLa cells (Fig. 1D). Furthermore, plaques of WRΔA26 were smaller (<0.35 mm in diameter) and fewer (by 8- to 10-fold) on HeLa-A56/K2 cells than on HeLa cells (Fig. 1E). Thus, the expression of the A56 and K2 proteins on HeLa-A56/K2 cells significantly inhibited WRΔA26 MV infection, as reported previously for A56/K2 inhibition of wild-type WR MV (36).

FIG 1.

Generation of HeLa-A56/K2 cells coexpressing the vaccinia virus A56 and K2 proteins to inhibit WRΔA26 infection. (A) FACS analyses of surface expression of A56 (top) and K2 (bottom) in HeLa-A56/K2 cells. Control HeLa and HeLa-A56/K2 cells were incubated with anti-A56 or anti-K2 mAb followed by fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse secondary antibody, washed, and analyzed by FACS. The y axis shows the percentage of the cell population stained with either anti-A56 (top) or anti-K2 (bottom) mAb. The x axis represents the fluorescence intensity. (B) Immunofluorescence staining of cell surface A56 and K2 proteins using anti-A56 or anti-K2 mAbs in HeLa and HeLa-A56/K2 cells under nonpermeable conditions. Blue represents propidium iodide staining in cell nuclei. (C and D) HeLa and HeLa-A56/K2 cells were infected with MV of WRΔA26-Venus-A4-mCherry at an MOI of 0.1 PFU per cell, and the expression of the viral early Venus marker (C) and late A4-mCherry (D) genes was monitored by FACS at 2 hpi and 8 hpi, respectively. The mean fluorescence intensity in HeLa cells was set as 100% to normalize the relative fluorescence intensity in HeLa-A56/K2 cells. Experiments were repeated three times, and statistical analyses were performed using two-tailed unpaired Student’s t test. ***, P < 0.001. (E) Plaque formation of WRΔA26 on HeLa and HeLa-A56/K2 cells. HeLa and HeLa-A56/K2 cells were infected with ∼800 PFU of WRΔA26 and incubated for 3 days under an agar overlay prior to cell staining with 1% crystal violet.

Adaptive mutants that have overcome A56/K2 inhibition gradually accumulate upon experimental passaging of WRΔA26 on HeLa-A56/K2 cells for 20 generations.

We began our passaging experiments by infecting HeLa-A56/K2 cells with WRΔA26 at an MOI of 0.02 PFU per cell and then harvested cells at 3 days postinfection (dpi) (Fig. 2A). Lysates of the harvested cells were used as the P1 virus pool to initiate the next round of infection on HeLa-A56/K2 cells, with the resulting lysates being harvested as the P2 virus pool. This process was repeated 10 times to obtain a total of 10 virus pools, denoted P1 to P10, of WRΔA26^HeLa-A56/K2. We continued with another 10 passages at an MOI of 0.002 PFU per cell to obtain viral pools P11 to P20. In general, the proportion of a given viral pool used as the inoculum to generate the next pool was relatively small, at ∼0.1 to 0.01%. As a control experiment, WRΔA26 was similarly passaged in parallel on HeLa cells to obtain 20 passaged virus pools, denoted P1 to P20, of WRΔA26^HeLa. We anticipated that beneficial viral mutations to overcome A56/K2-mediated fusion suppression would be specifically selected for and accumulate in the virus pools of WRΔA26^HeLa-A56/K2 and that spontaneous nonadaptive mutations would be present in virus pools of WRΔA26^HeLa following long-term virus passaging in control HeLa cells.

FIG 2.

Experimental passaging of WRΔA26 in HeLa and HeLa-A56/K2 cells. (A) Schematic representation of our experimental-evolution selection process. For P1 to P10 passaging, an MOI of 0.02 PFU per cell was used, and for P11 to P20 passaging, an MOI of 0.002 PFU per cell was used. (B) Plaque morphology of WRΔA26^HeLa-A56/K2 from the P1, P5, P10, and P20 virus pools on HeLa and HeLa-A56/K2 cells at 3 dpi after cell staining with 1% crystal violet. (C) Quantification of WRΔA26^HeLa-A56/K2 titers in the P1, P5, P10, and P20 virus pools shown in panel B. (D) Plaque morphology of WRΔA26^HeLa from the P1, P5, P10, and P20 virus pools on HeLa and HeLa-A56/K2 cells at 3 dpi after cell staining with 1% crystal violet. (E) Quantification of WRΔA26^HeLa titers in the P1, P5, P10, and P20 virus pools shown in panel D.

Next, we compared plaque sizes (Fig. 2B and D) and titers (Fig. 2C and E) of the P1, P5, P10, and P20 pools of WRΔA26^HeLa-A56/K2 and WRΔA26^HeLa by plating on HeLa and HeLa-A56/K2 cells. We found that the P1 pool of WRΔA26^HeLa-A56/K2 grew well on HeLa but not on HeLa-A56/K2 cells, exhibiting small plaques and low virus titers in the latter cells (Fig. 2B and C). However, we observed gradual virus adaptation through subsequent passaging rounds on HeLa-A56/K2 cells so that by the 20th passaging event, the P20 pool of WRΔA26^HeLa-A56/K2 already exhibited large plaques (Fig. 2B) and a high virus titer (Fig. 2C) on both HeLa-A56/K2 and HeLa cells. Accordingly, we concluded that repeated passaging of WRΔA26 on HeLa-A56/K2 cells had specifically selected for variants in the P20 pool of WRΔA26^HeLa-A56/K2 that had overcome the fusion inhibition imposed by the cell surface A56/K2 protein complex. In our control passaging experiment, the P1, P5, P10, and P20 pools of WRΔA26^HeLa all formed large plaques (>0.35 mm in diameter) on HeLa cells and small plaques (<0.35 mm in diameter) on HeLa-A56/K2 cells (Fig. 2D). Moreover, the titers of each of these pools of WRΔA26^HeLa on HeLa cells were always ∼8-fold higher than those on HeLa-A56/K2 cells (Fig. 2E), demonstrating that even after 20 passages on control HeLa cells, WRΔA26^HeLa remained sensitive to A56/K2-mediated fusion inhibition.

Whole-genome sequencing of virus pools reveals a specific G9 H44Y mutation in WRΔA26.

We purified viral genomes from P5, P7, P10, and P20 of WRΔA26^HeLa-A56/K2 as well as from the P10 and P20 pools of WRΔA26^HeLa. Whole-genome sequencing analyses were performed with a sequencing depth close to an average of 600 reads per base. Genome sequences were aligned with parental WRΔA26 sequences to identify genetic variations that arose during passaging on HeLa-A56/K2 and control HeLa cells. To avoid background fluctuation interfering with data interpretation, we filtered the data set with a cutoff of 10% variation and identified one mutation of the D8L ORF (C525G) that presented a 13.5% mutation frequency in the P10 and a 93.9% mutation frequency in the P20 pools of WRΔA26^HeLa (Fig. 3A). The same mutation was observed in the P20 pool of WRΔA26^HeLa-A56/K2, with a mutation frequency of 61.9%. Another unique mutation (C130T) was identified in the G9R ORF solely of WRΔA26^HeLa-A56/K2 virus pools (P7, P10, and P20), which resulted in the H44Y amino acid change of the G9 protein (Fig. 3A). Although H44 is conserved among G9 proteins of the Orthopoxvirus genus, the respective residues are Q in the parapoxvirus genus, F in yokapox virus, and R in pteropox virus (Fig. 3B). The G9^H44Y mutation frequency increased from 49.2% in the P7 pool to 87% in the P10 pool and then up to 99.8% in the P20 pool of WRΔA26^HeLa-AK (Fig. 3C). Together, these data demonstrate that the G9^H44Y genetic mutation is specifically associated with the phenotypic adaptation of WRΔA26^HeLa-A56/K2 virus pools, allowing them to overcome inhibition by the viral A56 and K2 suppressors. Since the C525G mutation of the D8L ORF occurred in pools of both WRΔA26^HeLa and WRΔA26^HeLa-A56/K2, we assume that it has little relevance for the adaptation allowing WRΔA26^HeLa-A56/K2 to overcome A56-K2 inhibition.

FIG 3.

Whole-genome sequencing identifies genome variations present in different virus pools. (A) Mutations of viral genes present in various virus pools derived from populations of WRΔA26^HeLa and WRΔA26^HeLa-A56/K2. The data set was filtered with a cutoff of 10% variation, and mutations resulting in amino acid changes are shown. N/D, not done. (B) Schematic representation of the vaccinia virus G9 protein (amino acids 1 to 340) showing the single His44Tyr mutation close to the N-terminal region. TM, transmembrane region. Black bars in the box represent conserved cysteine residues. Shown is a multiple-sequence alignment of the N-terminal 50 amino acids of the vaccinia virus G9 protein (GenBank accession number YP_232969) with its orthologues in cowpox virus (accession number NP_619884), camelpox virus (accession number NP_570475), taterapox virus (accession number YP_717397), variola virus (accession number NP_042116), monkeypox virus (accession number NP_536506), raccoonpox virus (accession number YP_009143395), skunkpox virus (accession number YP_009282780), volepox virus (accession number YP_009281834), yokapox virus (accession number YP_004821420), pteropox virus (accession number YP_009268779), pseudocowpox virus (accession number YP_003457351), bovine papular stomatitis virus (accession number NP_957955), parapoxvirus (accession number YP_009112785), Orf virus (accession number NP_957823), squirrelpox virus (accession number YP_008658478), and molluscum contagiosum virus (accession number NP_044019). The color scheme reflects amino acid conservation with a minimum threshold of 50%. (C) Percent 130C>T mutation frequency (representing His44Tyr) in the G9R ORF that accumulated across the 20 passages of our experimental-evolution analysis.

Recombinant WRΔA26-G9^H44Y virus overcomes MV and EV entry inhibition on HeLa-A56/K2 cells.

To verify that the H44Y mutation of the G9 protein could overcome A56/K2-mediated fusion inhibition, we generated a recombinant WRΔA26-G9^H44Y virus that expresses the G9^H44Y mutant protein as well as a green fluorescent protein (GFP) to facilitate the isolation of virus clones (Fig. 4A). We also generated a revertant vaccinia virus, WRΔA26-G9-Rev, expressing a wild-type G9R protein and a GFP marker as a control (Fig. 4A). Immunoblot analyses of infected cells demonstrated that both WRΔA26-G9^H44Y and WRΔA26-G9-Rev expressed a G9 protein of the same size as that in the parental WRΔA26-WT virus (Fig. 4B). To ensure that WRΔA26-G9^H44Y had not reverted to wild-type G9R during the isolation process, we purified the viral genomes from virus stocks, amplified the G9R ORF by PCR, and sequenced the resulting DNA fragments to reconfirm clonal accuracy (data not shown). Next, we infected HeLa and HeLa-A56/K2 cells with MV of WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev at a low MOI of 0.02 and harvested cells at 24 hpi to determine MV titers (Fig. 4C). All three viruses grew equally well on HeLa cells, but the titers of WRΔA26-G9^H44Y on HeLa-A56/K2 cells at 24 hpi were significantly higher than those of WRΔA26-WT and WRΔA26-G9-Rev, suggesting that the G9^H44Y mutation had overcome fusion inhibition by A56/K2 proteins (Fig. 4C). To confirm that the increased viral yield was due to enhanced MV entry, we infected HeLa and HeLa-A56/K2 cells with ∼200 PFU of each virus at 37°C for 60 min, washed off any unbound virus, added an agar overlay, and then counted the numbers of plaques on both cell types at 3 dpi. We divided the number of plaques on HeLa-A56-K2 cells by the respective number on HeLa cells to normalize the efficiency of virus entry into HeLa-A56/K2 cells. As shown in Fig. 4D, WRΔA26-WT and WRΔA26-G9-Rev formed fewer plaques on HeLa-A56/K2 cells than on HeLa cells, whereas we observed equivalent numbers of WRΔA26-G9^H44Y plaques on both cell types, supporting that the G9^H44Y mutation had enhanced virus entry into HeLa-A56/K2 cells.

FIG 4.

Generation of recombinant WRΔA26-G9^H44Y virus. (A) Schematic representations of the WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev recombinant viruses. Luciferase and guanine phosphoribosyl transferase (GPT) selection markers replaced the A26L locus in all three viruses. The WRΔA26-G9^H44Y and WRΔA26-G9-Rev recombinant viruses expressed a p11k-driven G9^H44Y mutant and the wild-type G9 protein, respectively, as well as a GFP marker driven by the p7.5K promoter. (B) Immunoblotting of G9 proteins in virus-infected cells collected at 24 hpi. GFP is a marker that is present in the WRΔA26-G9^H44Y and WRΔA26-G9-Rev viruses, and D8 is a viral envelope protein that serves as a positive control. (C) Growth of the WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev viruses in HeLa and HeLa-A56/K2 cells. HeLa and HeLa-A56/K2 cells were infected with each virus at an MOI of 0.02 PFU per cell, and the cells were harvested at 0 and 24 hpi for virus titer determination. The y axis represents virus growth, as determined by dividing virus titers at 24 hpi with the respective values at 0 hpi. (D) Relative efficiency of virus entry into HeLa and HeLa-A56/K2 cells. HeLa and HeLa-A56/K2 cells were infected with ∼200 PFU of each virus for 60 min, washed with PBS to remove unbound viruses, and overlaid with agar, and the numbers of PFU, indicating successful entry events, were determined. The values on the y axis were calculated using the formula (PFU on HeLa-A56/K2 cells/PFU obtained on HeLa cells) × 100%, representing the relative efficiency of entry of each virus into HeLa-A56/K2 cells normalized to that for HeLa cells. (E) Plaque morphologies of WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev on HeLa and HeLa-A56/K2 cells after staining with 1% crystal violet. (F) Plaque size quantification for the WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev viruses on HeLa and HeLa-A56/K2 cells. Single plaques on individual cells were randomly selected, and the diameters of 30 plaques were determined using ImageJ. The y axis represents the normalized plaque size, calculated by dividing the plaque diameter on HeLa-A56/K2 cells with the respective value on HeLa cells. Experiments in panels C, D, and F were repeated three times, and data were analyzed using two-tailed unpaired Student’s t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

The viral EFC is also important for facilitating EV-cell fusion during cell-to-cell spreading (41), so we anticipated that WRΔA26-G9^H44Y should also enhance EV spreading on HeLa-A56/K2 cells. Therefore, we measured the sizes of plaques formed by WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev on HeLa and HeLa-A56/K2 cells at 3 dpi. As shown in Fig. 4E, although all three viruses formed large plaques on HeLa cells, only WRΔA26-G9^H44Y formed mid- to large-sized plaques on HeLa-A56/K2 cells, whereas WRΔA26-WT and WRΔA26-G9-Rev formed tiny plaques on HeLa-A56/K2 cells (Fig. 4E; quantification in Fig. 4F). Taken together, these results show that the H44Y mutation of the G9 protein is sufficient to improve vaccinia MV entry as well as EV spreading in otherwise resistant HeLa-A56/K2 cells.

The vaccinia virus A56 and K2 proteins on infected cell surfaces function to prevent EV progeny back-fusion (35–37), but A56R or K2L knockout results in a cell fusion phenotype upon EV spreading in viral plaques (32). We wanted to investigate if the G9^H44Y mutation overcomes A56/K2-mediated inhibition of fusion by EV. GFP- and red fluorescent protein (RFP)-expressing HeLa and HeLa-A56/K2 cells were mixed at a 1:1 ratio and subsequently infected with WRΔA26-WT, WRΔA26-G9^H44Y, or WRΔA26-G9-Rev, and we monitored the development of plaque morphology with a cell fusion phenotype at 1 to 2 dpi. As shown in Fig. 5A, WRΔA26-WT and WRΔA26-G9-Rev produced plaques lacking a cell fusion phenotype on both HeLa and HeLa-A56/K2 cells. In contrast, WRΔA26-G9^H44Y developed plaques with a robust cell fusion phenotype on HeLa cells. It also produced plaques with a cell fusion phenotype on HeLa-A56/K2 cells, although the extent of cell fusion was less obvious. Imaging quantification of cell fusion within 14 plaques of each virus confirmed that the viral EFC harboring the G9^H44Y mutation was less inhibited by the A56/K2 protein complex (Fig. 5B).

FIG 5.

WRΔA26-G9^H44Y mutant virus forms plaques with a cell fusion phenotype on HeLa and HeLa-A56/K2 cells. (A) Images of plaques formed by the WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev viruses on GFP- and RFP-expressing HeLa and HeLa-A56/K2 cells. HeLa or HeLa-A56/K2 cells expressing GFP and those expressing RFP were mixed at a 1:1 ratio and infected with each virus. Plaques were photographed at 2 dpi, as described in Materials and Methods. (B) Quantification of cell fusion in plaques in panel A from 14 images for each virus. The percentage of cell fusion was quantified using the image area of GFP⁺ RFP⁺ double-fluorescent cells divided by that of single-fluorescent cells. Although WRΔA26-G9^H44Y and WRΔA26-G9-Rev harbor a GFP marker, its expression level was relatively low relative to the fluorescence intensities in HeLa-GFP and HeLa-A56/K2-GFP cells, so viral GFP was calibrated as the background value. Data from a total of 14 plaques were subjected to statistical analyses using two-tailed unpaired Student’s t test. ***, P < 0.001.

H44Y mutation of the G9 protein does not alter its ability to bind to the A56/K2 protein complex or other EFC components.

Since it had previously been proposed that A56/K2 directly binds the A16/G9 subunits of the EFC to regulate the fusion activity of the EFC (36–38), we used co-IP to investigate if the G9^H44Y mutation alters the ability of G9 to bind to A16 or the A56/K2 protein complex. First, we cotransfected plasmids expressing A16 with wild-type G9 or the G9^H44Y mutant into 293T cells, and our co-IP showed that both wild-type G9 and mutant G9^H44Y pulled down A16 protein equally well (Fig. 6A). Next, we assessed if the G9^H44Y mutation affects interactions between G9/A16 and the A56/K2 protein complex. Wild-type G9 or G9^H44Y mutant plasmids were cotransfected with three plasmid types expressing the A16, A56, and K2 proteins into cells that had been infected with VTF7-3 to induce high-level gene expression (Fig. 6B). Co-IP of A56 brought down the K2, A16, and wild-type G9 proteins. The replacement of the wild-type G9 plasmid with the G9^H44Y mutant plasmid did not alter these co-IP results (Fig. 6B). We then performed reciprocal co-IP using Myc antibody to pull down the wild-type G9 or G9^H44Y mutant protein. Both the wild-type and mutant G9 proteins brought down A16, A56, and K2 equally well (Fig. 6C). Finally, we investigated if the G9^H44Y mutation affects EFC formation, but our co-IP results revealed little difference between the wild-type G9 and G9^H44Y mutant proteins in terms of binding to EFC subunits (Fig. 6D).

FIG 6.

Coimmunoprecipitation of the G9^H44Y protein with A16, the A56/K2 protein complex, and multiple components of the viral EFC in virus-infected cells. (A) Co-IP of G9^H44Y with A16 in transiently transfected cells. 293T cells were transfected with plasmids expressing Myc-G9, Myc-G9^H44Y, and/or hemagglutinin (HA)-A16 and harvested 24 h later for co-IP with anti-Myc agarose beads as described in Materials and Methods. The blot was probed with anti-G9 and anti-A16 antibodies. (B and C) Co-IP of the G9^H44Y/A16 complex with the A56/K2 protein complex. 293T cells were infected with VTF7-3 and transfected with plasmids expressing Myc-G9, Myc-G9^H44Y, HA-A16, A56-GFP, and/or K2-Flag, and lysates were harvested 24 h later for co-IP with anti-GFP (B)- or anti-Myc (C)-conjugated agarose beads. Blots were probed with anti-GFP, anti-Flag, anti-G9, or anti-A16 antibodies. (D) Co-IP of the G9^H44Y protein with EFC components in virus-infected cells. 293T cells were either mock infected or infected with individual viruses at an MOI of 5 PFU per cell, harvested at 24 hpi for co-IP with anti-G9-conjugated agarose beads, and analyzed by immunoblotting with anti-EFC antibodies, as shown at the right side of each panel.

The WRΔA26-G9^H44Y mutant virus has altered pH sensitivity to acid-induced membrane fusion in HeLa-A56/K2 cells.

Although our co-IP assays revealed that both the wild-type G9 and mutant G9^H44Y proteins bound to A56/K2 equally well at neutral pH, we could not exclude the possibility that they exhibit differential responses during the acid-dependent fusion activation process. To explore if the WRΔA26-G9^H44Y mutant virus exhibits any change in acid sensitivity, we infected HeLa-A56/K2 cells with MV of WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev for 1 h so that the incoming viruses could bind to A56/K2 on cell surfaces. We then treated these MV-bound cells briefly with buffers with a range of pHs (pH 7.4, 6.5, 6.0, 5.5, or 4.7), washed the cells again with growth medium, and then monitored cell-cell fusion 1 to 2 h later (Fig. 7A). Using this strategy, we found that the WRΔA26-G9^H44Y virus triggered a low level of cell-cell fusion at neutral pH, suggesting that cell surface A56/K2 proteins efficiently bound WRΔA26-G9^H44Y and inhibited the subsequent membrane fusion usually triggered by vaccinia viruses. When infected HeLa-A56/K2 cells were treated with low-pH buffers, both WRΔA26-WT and WRΔA26-Rev triggered cell-cell fusion only at the lowest pH of 4.7. In contrast, cells infected with the WRΔA26-G9^H44Y mutant virus triggered robust cell fusion at pH 6.0, 5.5, and 4.7, as quantified in Fig. 7B. This higher pH threshold of membrane fusion activation by the WRΔA26-G9^H44Y mutant suggests a reduced inhibitory effect of A56/K2 in terms of suppressing EFC-G9^H44Y fusion activity. We speculate that the G9^H44Y mutant protein may mimic an “intermediate structure” of the wild-type G9 protein at the low pH that facilitates viral EFC dissociation from the A56/K2 protein complex. Thus, we conclude that the H44Y mutation of the G9 protein promotes the easy dissociation of the viral EFC from A56/K2.

FIG 7.

Acid titration of virus-mediated fusion activation in HeLa-A56/K2 cells. (A) HeLa-A56/K2 cells expressing GFP and those expressing RFP were seeded at a 1:1 ratio. The next day, these cells were pretreated with 40 μg/ml cordycepin for 60 min and subsequently infected with MV of WRΔA26, WRΔA26-G9^H44Y, or WRΔA26-G9-Rev at an MOI of 100 PFU per cell. Cordycepin was present in culture media throughout the experiments to inhibit viral early gene expression. After infection at 37°C for 60 min, cells were treated with a buffer of pH 7.4, 6.5, 6.0, 5.5, or 4.7 at 37°C for 3 min, washed with growth medium, further incubated at 37°C for 1 h, and then photographed using a Zeiss Axiovert fluorescence microscope. (B) Quantification of percentage of cell fusion from the images in panel A. Images from three independent experiments were quantified using the image area of GFP⁺ RFP⁺ double-fluorescent cells divided by that of single-fluorescent cells. Data from three independent experiments were subjected to statistical analysis using two-tailed unpaired Student’s t test. *, P < 0.05; ***, P < 0.001. Cyan asterisks represent the level of significant differences between WRΔA26-G9^H44Y and WRΔA26-G9-Rev. Blue asterisks represent the level of significant differences between WRΔA26-G9^H44Y and WRΔA26-WT.

DISCUSSION

Vaccinia virus exhibits a broad range of host infectivity, using multiple membrane proteins with high redundancy to bind to cell surface carbohydrates. However, the multisubunit nature of the vaccinia virus EFC complicates efforts to understand how the virus executes membrane fusion during viral entry. Currently, most of the known virus fusion proteins are grouped into three classes (42). Class I fusion proteins contain an N-terminal fusion peptide and a triple-stranded helical structure for trimer formation. Class II fusion proteins contain an internal fusion loop for membrane insertion (43). Class III fusion proteins are often considered hybrid molecules, exhibiting features common to class I and class II fusion proteins (44). X-ray crystallization approaches are useful for identifying novel structural features of viral fusion proteins, but thus far, only the L1 (45) and F9 (46) proteins of the vaccinia virus EFC have been crystallized, and their structures do not share significant homology with the known viral fusion proteins mentioned above.

In this study, we employed an experimental evolutionary approach that involved passaging viruses under conditions that selected for viral mutations, which facilitated escape from A56/K2-mediated membrane fusion inhibition. Although the A56/K2 proteins are known to inhibit membrane fusion mediated by the viral EFC, inhibition is not absolute. Accordingly, we titrated a relatively low level of input virus, in the range of 0.1 to 0.01%, during experimental evolution in order to obtain differential inhibition of MV entry on HeLa-A56/K2 cells relative to that on HeLa cells. Such a low MOI input forced the virus to replicate for multiple rounds in each passage before harvesting.

From our serial passaging experiments, we identified the H44Y mutation of the G9 protein by whole-genome sequencing in several virus pools. Our WRΔA26-G9^H44Y recombinant virus also confirmed that the H44Y adaptive mutation is sufficient to overcome A56/K2-mediated fusion suppression of EV and MV. Although the G9^H44Y mutant protein facilitates escape from A56/K2-mediated membrane fusion, our co-IP experiments did not reveal any biochemical differences contributed by the H44Y mutation since both the wild-type G9 and G9^H44Y mutant proteins bound to A16, A56/K2, and other EFC components. Notably, vaccinia virus expresses three major forms of A56, i.e., proteins of 85 kDa, 62 kDa, and 58 kDa (47–49), and a monoclonal antibody (mAb) recognizing the 85-kDa form was shown to inhibit vaccinia virus infection and syncytium formation (36, 39). Wagenaar et al. previously showed that the vaccinia virus G9/A16 protein subcomplex bound to A56/K2 proteins in detergent-treated extracts; however, which form of the A56 protein was brought down by the G9 protein in co-IP experiments was not clear (38). Our co-IP experiments revealed that both wild-type and mutant G9 proteins preferentially and consistently interacted with the two underglycosylated forms of the A56 protein (58 kDa and 62 kDa) instead of the fully glycosylated form (Fig. 6B). How the glycosylation of the A56 protein influences its interaction with the G9/A16 proteins will be interesting to study in the future.

Since WRΔA26-G9^H44Y could trigger cell fusion at pH 6, we hypothesize that the mutant G9^H44Y protein adopts an intermediate conformation more prone to acid-induced membrane fusion. Residue H44Y is located within the N-terminal region of the G9 protein, which was previously reported to play a role in fusion regulation since the insertion of a small tag at the N-terminal region of the G9 protein increased spontaneous fusion at neutral pH (11). Although H44 is conserved among G9 proteins of the Orthopoxvirus genus, the respective residues are Q in the parapoxvirus genus, F in yokapox virus, and R in pteropox virus (Fig. 3B). Why the H44Y mutation was selected out of our experimental-evolution assay to overcome A56/K2-mediated inhibition is currently unclear. It would be very informative to obtain the crystal structure of the G9 protein in order to fully explain how the H44Y mutation modulates EFC fusion activity. Finally, the G9 protein is known to interact with the A16 protein in the EFC, but we did not select out any mutation of the A16 ORF during the 20 passages of our experimental-evolution assay. Why this might be is not clear. The acid sensitivity of A56/K2 inhibition of membrane fusion by the EFC suggests that a biochemical regulatory mechanism is at play, which may involve acid-dependent conformational changes. This possibility requires further investigation. The advantage of our experimental-evolution approach is that we could identify functionally important mutations, but it has the limitations of being time-consuming and necessitating narrow-spectrum selective pressure. In the future, we need to identify other functionally important motifs or domains in subunits of the vaccinia virus EFC in order to further dissect the membrane fusion mechanism mediated by the poxviral EFC.

MATERIALS AND METHODS

Cell culture, reagents, and viruses.

BSC40 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) plus 10% calf serum (CS). CV-1 cells, HeLa and other HeLa-derived cell clones, as well as 293T cells were cultured in DMEM plus 10% fetal bovine serum (FBS; Invitrogen). The pLKO AS3w.neo, pLKO AS3w.puro, pLKO AS3.1 EGFP3′, pLKO AS3.1w tRFP-C, pCMV-ΔR8.91, and pMD.G lentiviral vectors were obtained from the National RNAi Core (Academia Sinica, Taiwan). The pMyc-G9R, pHA-A16L, pA56R-GFP, and pK2L-Flag plasmids were described previously (28). We generated the pMyc-G9R His44Tyr mutant plasmid using in vitro mutagenesis (QuikChange Lightning site-directed mutagenesis kit; Agilent Technologies Inc.). The wild-type Western Reserve (WR) strain of vaccinia virus, the A26L deletion virus (WRΔA26), WRΔA26-A4mCherry, and VTF7-3 (ATCC VR-2153) were propagated in BSC40 cells, and virus titers were determined by plaque assays as described previously (27, 28, 50). The generation of HeLa cells expressing GFP or RFP was performed as described previously (27, 29). Anti-H3 (4), anti-D8 (5), anti-G3 (10), anti-A4 (51), anti-L1 (28), anti-G9 (28), anti-L5 (28), and anti-A16 (28) antibodies were all described previously. Anti-A56 monoclonal antibody (mAb) was provided by Yasuo Ichihashi (39, 52). Anti-K2 mAb was provided by Richard W. Moyer (53). Anti-Flag mAb was purchased from Sigma Inc. Anti-GFP antibody was purchased from Bioscience Inc. Anti-A28, anti-H2, anti-F9, and anti-O3 are rabbit antibodies raised against the following four synthetic peptides, respectively: A28(92–112) (N′-QAEVGPNNTRSIRKFNTMQQC-C′), H2(81–102) (N′-KLESDRGRLLAAGKDDIFEFKC-C′), F9(80–107) (KDRRAIAEEIGIDLDDVPSAVSKLEKNC-C′), and O3(14–35) (N′-CSWLSYSYLRPYISTKELNKSR-C′).

Construction of a fluorescent recombinant vaccinia virus expressing early Venus and late A4-mCherry protein, WRΔA26-Venus-A4-mCherry.

Viral genomic DNA containing an early Venus expression cassette flanked by J4L and J5R sequences was obtained from John H. Connor (54, 55) and used as the template to obtain a PCR fragment, which was subsequently cloned into a TOPO plasmid, resulting in pTopo-J4R-Venus-J5L, in which Venus is expressed from a viral early promoter of the C11R gene (54). The plasmid was sequenced to ensure accuracy and subsequently transfected into CV-1 cells that had been infected with WRΔA26-A4mCherry (50). The lysates were harvested at 24 h postinfection (hpi), and virus titration was performed on BSC40 cells to isolate the WRΔA26-Venus-A4-mCherry virus through multiple rounds of plaque purification by fluorescence microscopy until 100% purity was reached.

Construction of a stable cell line expressing the A56/K2 protein complex.

The vaccinia virus A56R and K2L genes were synthesized using codons optimized for mammalian expression (GenScript Inc.). The vaccinia virus A56R and K2L ORFs were inserted into the vectors pLKO AS3w.neo and pLKO AS3w.puro and subsequently packaged into lentiviral vectors (provided by the RNAi Core, Academia Sinica, Taiwan). HeLa cells were infected with lentiviral vectors expressing A56 and cultured in medium containing G418 (1 mg/ml) for 2 weeks. These A56-expressing HeLa cells were subsequently infected with lentivirus expressing K2 and selected in medium containing puromycin (1 μg/ml) for 1 week. The resulting cell clones were assessed for A56 and K2 protein expression using anti-A56 and anti-K2 mAbs, respectively, by fluorescence-activated cell sorting (FACS) and immunofluorescence by confocal microscopy, as described previously (50). HeLa-A56/K2 cells expressing GFP or RFP were subsequently generated using FACS as described previously (27, 29). In brief, HeLa-A56/K2 cells were infected with lentiviral vectors expressing GFP or RFP (RNAi Core, Academia Sinica, Taiwan) and selected for GFP-positive (GFP⁺) or RFP-positive (RFP⁺) cells by FACS analyses.

Experimental passaging of WRΔA26 viruses on HeLa and HeLa-A56/K2 cells for adaptive mutant virus selection.

HeLa-A56/K2 cells (10⁶ cells/100 mm) were infected with WRΔA26 at a multiplicity of infection (MOI) of 0.02 PFU per cell and cultured for 2 to 3 days until infection had completed. The cells were harvested, denoted the P1 pool of WRΔA26^HeLa-A56/K2, and then used to infect fresh HeLa-A56-K2 cells again at an MOI of 0.02 PFU per cell to obtain the P2 pool. This process was repeated 10 times to obtain a total of 10 virus pools, denoted P1 to P10, of WRΔA26^HeLa-A56/K2. We then continued with 10 more passages at an MOI of 0.002 PFU per cell to obtain the P11 to P20 pools. As a control, HeLa cells were infected and passaged with WRΔA26 as described above to obtain 20 virus pools, denoted P1 to P20, of WRΔA26^HeLa. To observe the plaque phenotypes of these passaged viruses, the P1 to P20 virus pools of WRΔA26^HeLa-A56/K2 and WRΔA26^HeLa were simultaneously titrated on HeLa and HeLa-A56/K2 cells to determine plaque numbers and sizes at 3 days postinfection (dpi) after 1% crystal violet staining.

Viral genome sequencing and data analyses.

Vaccinia viral genomic DNA isolation, sequencing, and data analyses were performed as previously described, with minor modifications (29). All viral genomic DNA was purified and quantified using a Qubit 3.0 fluorometer and a Qubit dsDNA BR assay kit (Life Technologies, Carlsbad, CA, USA). Whole-genome sequencing libraries were prepared using a Kapa Hyper preparation kit (Roche, Kapa Biosystems) protocol for Illumina platforms. Genomic DNA (1 μg) was sheared to an average length of 350 bp using an M220 focused ultrasonicator (Covaris). For each library, 100 ng fragmented DNA was end repaired and 3′ adenylated using a proprietary master mix before being ligated to the barcoded adapter from an Illumina index kit (Illumina Inc., San Diego, CA, USA). The adapter-ligated DNA was enriched using a Kapa Hyper library preparation kit with nine cycles of PCR and then purified using Kapa Pure beads (Roche, Kapa Biosystems). The quality and size distribution of the libraries were verified on an Agilent 2100 bioanalyzer using an Agilent DNA high-sensitivity kit. Concentrations of the libraries were determined using a Kapa library quantification kit (Roche, Kapa Biosystems). Libraries were sequenced using NextSeq paired-end (PE) 150-cycle version 2.5 reagents on an Illumina NextSeq 500 system (Illumina Inc., USA) at the Genomics Core Facility of the Institute of Molecular Biology, Academia Sinica (Taiwan). To obtain 150-bp paired-end reads, base calling was conducted using bcl2fastq (version 2.17.1.14) conversion software, and BCL files were converted to FASTQ format.

The resulting sequence data were uploaded to CLC Genomics Workbench 11.0.1 (Qiagen, Aarhus, Denmark) for raw sequence trimming, sequence mapping, and variant detection. Raw sequencing reads were trimmed by removing adapter sequences and low-quality sequences (Phred quality score of <Q30) and sequencing fragments shorter than 30 nucleotides. Sequencing reads were mapped to the human genome (GRCh38; from ftp.ensembl.org/pub/release-82/fasta/homo_sapiens/dna/) with stringent parameters (mismatches cost of 2, insertion cost of 3, deletion cost of 3, minimum fraction length of 0.9, and minimum fraction similarity of 0.9). All host genome sequences that met the above-mentioned parameters were removed, before mapping the remaining paired-end reads to the vaccinia virus WR genome (GenBank accession number NC_006998) (56). We removed duplicate reads, performed a local realignment, and then used the Basic Variant Detection tool in CLC Genomics Workbench 11.0.1 to call single nucleotide polymorphisms (SNPs) and insertions/deletions (indels) with the following customized parameters to identify mutation positions: (i) a minimum frequency of 1% and a minimum coverage of 10 reads and (ii) a minimum quality of SNPs/indels of >Q25 and a neighborhood quality (upstream/downstream 5 bases) of >Q20.

We also used the paired-end reads, after removing host genome sequences, to generate mutant and revertant viral genomes by de novo genome assembly but excluding the terminal repeat sequences of vaccinia virus. Using the alignment program MAFFT version 7 (57, 58), we aligned our parental WRΔA26 strain, P5 pool, P7 pool, P10 pool, and P20 pool viral genome sequences with the reference WR strain (GenBank accession number NC_006998) to identify differences among viruses.

Generation of recombinant WRΔA26-G9^H44Y and WRΔA26-G9-Rev vaccinia viruses.

A G8R-GFP-G9R cassette containing the p11k-driven G9R ORF, the p7.5K-driven GFP gene, and 500 bp of the G8R sequence at the 3′ end was cloned into the pBluescript plasmid using the NEBuilder HiFi DNA assembly cloning kit to obtain the pBluescript-G8R-GFP-G9R plasmid. We then generated a G9R His44Tyr mutant plasmid using in vitro mutagenesis (QuikChange Lightning site-directed mutagenesis kit; Agilent Technologies Inc.). To generate recombinant WRΔA26-G9^H44Y and WRΔA26-G9-Rev viruses, we infected CV-1 cells with WRΔA26 at an MOI of 5 PFU per cell. The CV-1 cells were subsequently transfected with the pBluescript-G8R-GFP-G9R plasmid and the pBluescript-G8R-GFP-G9RH44Y plasmid and cultured for 3 days before cell harvesting for the next round of infection to enrich for GFP-positive cells using FACS. These GFP-positive cells were used to further purify GFP-positive recombinant virus plaques after five rounds of plaque purification.

Growth analyses of the WRΔA26-WT, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev viruses in HeLa and HeLa-A56/K2 cells.

HeLa cells (2 × 10⁵ cells/well) and HeLa-A56/K2 cells (2 × 10⁵ cells/well) in 6-well plates were infected with each virus at an MOI of 0.02 PFU per cell, washed, and incubated with growth medium at 37°C, and the cells were then harvested at 0 and 24 hpi for virus titer determination. We determined virus growth by dividing virus titers at 24 hpi with the respective values at 0 hpi, as previously described (29).

Vaccinia MV-triggered cell fusion from without at neutral and acidic pH.

Cell-cell fusion induced by vaccinia MV infections was performed as described previously (27, 29). In brief, HeLa-A56/K2 cells expressing GFP and those expressing RFP were mixed at a 1:1 ratio and seeded into 96-well plates. The next day, these cells were pretreated with 40 μg/ml cordycepin or 3′-deoxyadenosine (Sigma) for 60 min and subsequently infected with MV of either WRΔA26, WRΔA26-G9^H44Y, or WRΔA26-G9-Rev at an MOI of 100 PFU per cell. Cordycepin was present in culture media throughout the experiments to inhibit viral early gene expression. After infection at 37°C for 60 min, cells were treated with phosphate-buffered saline (PBS) of various pHs (pH 7.4, 6.5, 6, 5.5, or 4.7) at 37°C for 3 min, washed with growth medium, further incubated at 37°C for 1 h, and then photographed using a Zeiss Axiovert fluorescence microscope. The percentage of cell fusion was quantified using MetaXpress software as the image area of GFP⁺ RFP⁺ double-fluorescent cells divided by that of single-fluorescent cells, as described previously (29). Experiments were repeated three times, and statistical analyses were performed using Student’s t tests in Prism (version 5) software (GraphPad). Statistical significance is represented as indicated in the figure legends.

Vaccinia EV-triggered cell fusion from within.

HeLa or HeLa-A56/K2 cells expressing GFP and those expressing RFP were mixed at a 1:1 ratio and seeded into 60-mm plates. The next day, cells were infected with only 20 to 50 plaques of either WRΔA26, WRΔA26-G9^H44Y, or WRΔA26-G9-Rev. The resulting plaque formation was monitored from 1 to 2 dpi, and individual plaques were photographed at 2 dpi using a Zeiss LSM710 fluorescence microscope. Although WRΔA26-G9^H44Y and WRΔA26-G9-Rev harbor a GFP marker, its expression level was relatively low relative to the fluorescence intensities in HeLa-GFP and HeLa-A56/K2-GFP cells, so viral GFP was calibrated as the background value. The percentage of cell fusion was quantified using MetaXpress software as the image area of GFP⁺ RFP⁺ double-fluorescent cells divided by that of single-fluorescent cells, as described previously (29). Experiments were repeated twice, and statistical analysis was performed with a total of 14 plaques using Student’s t tests in Prism (version 5) software (GraphPad). Statistical significance is represented as indicated in the figure legends. Alternatively, we performed EV-mediated fusion assays using Giemsa solution (Merck Inc.), monitored syncytium formation directly under a light microscope, and found that the resulting data were consistent with the outcomes of fluorescence microscopy.

Coimmunoprecipitation and immunoblot analyses.

Cell lysates used for coimmunoprecipitation (co-IP) were prepared from cells as previously described (28), under three different conditions: (i) 293T cells (8 × 10⁶ cells/100 mm) were transfected with 3 μg pMyc-G9R and pHA-A16L using a calcium phosphate transfection protocol and harvested 24 h later; (ii) 293T cells (8 × 10⁶ cells/100 mm) were infected with VTF7-3 at an MOI of 5 PFU per cell before being transfected with 3 μg of pMyc-G9R, pHA-A16L, pA56R-GFP, and pK2L-Flag using a calcium phosphate transfection protocol and then harvested 24 h later; and (iii) 293T cells (8 × 10⁶ cells/100 mm) were infected with the WRΔA26, WRΔA26-G9^H44Y, and WRΔA26-G9-Rev viruses at an MOI of 5 PFU per cell and harvested at 24 hpi. Cells were then lysed in a lysis buffer consisting of 20 mM Tris-HCl (pH 8.0), 200 mM NaCl, 0.5% NP-40, 1 mM EDTA, and a protease inhibitor cocktail containing 2 μg/ml aprotinin, 1 μg/ml leupeptin, 0.7 μg/ml pepstatin, and 1 mM phenylmethylsulfonyl fluoride (PMSF). Lysates were cleared by low-speed centrifugation; incubated at 4°C overnight with anti-Myc, anti-GFP (1:100), or anti-G9 (1:50) antibody conjugated to agarose beads (Sigma) or protein A beads (GE Healthcare); washed four times with lysis buffer; and centrifuged. The immunoprecipitates were then resuspended in SDS-containing sample buffer, separated by SDS-PAGE, and analyzed by immunoblot analyses as described previously (28).