Analysis of VSV pseudotype virus infection mediated by rubella virus envelope proteins

Rubella virus (RV) generally causes a systemic infection in humans. Viral cell tropism is a key determinant of viral pathogenesis, but the tropism of RV is currently poorly understood. We analyzed various human cell lines and determined that RV only establishes an infection efficiently in particular non-immune cell lines. To establish an infection the host cells must be susceptible and permissible. To assess the susceptibility of individual cell lines, we generated a pseudotype vesicular stomatitis virus bearing RV envelope proteins (VSV-RV/CE2E1). VSV-RV/CE2E1 entered cells in an RV envelope protein-dependent manner, and thus the infection was neutralized completely by an RV-specific antibody. The infection was Ca2+-dependent and inhibited by endosomal acidification inhibitors, further confirming the dependency on RV envelope proteins for the VSV-RV/CE2E1 infection. Human non-immune cell lines were mostly susceptible to VSV-RV/CE2E1, while immune cell lines were much less susceptible than non-immune cell lines. However, susceptibility of immune cells to VSV-RV/CE2E1 was increased upon stimulation of these cells. Our data therefore suggest that immune cells are generally less susceptible to RV infection than non-immune cells, but the susceptibility of immune cells is enhanced upon stimulation.

when the other five immune cell lines (U937, THP-1, Raji, M8166, and Jurkat) were tested (Fig. 1B). Therefore, RV infection is established efficiently in only some non-immune cell lines, particularly trophoblast-derived cell lines.

Generation of a pseudotype VSV bearing RV envelope proteins. Data from the experiments with
RV indicated that many cell lines were nearly non-susceptible or non-permissible, or both, to RV infection. To assess the susceptibilities of individual cell lines to infection with RV, pseudotype VSVs bearing RV envelope proteins were generated. The VSV pseudotype virus genomes lack the G gene, which is replaced with a reporter protein gene, the green fluorescent protein (GFP) or firefly luciferase (FLuc) gene 8,14,18,19 . The pseudotype viruses were thus expected to show similar behaviors to RV during the process of virus entry, and the subsequent processes of gene expression are dependent on the VSV replication machinery. In the initial experiment, only RV envelope E2 and E1 proteins were provided in trans to generate the GFP gene-and FLuc gene-encoding pseudotype viruses, VSV GFP -RV/E2E1 and VSV FLuc -RV/E2E1, respectively, like other VSV pseudotype viruses 13,[19][20][21][22][23][24][25][26][27][28][29][30] . The infectivity titers for VSV GFP -RV/E2E1 and VSV FLuc -RV/E2E1 were 10-fold higher than those of the counterpart control viruses, VSV GFP -∆G and VSV FLuc -∆G, respectively, which lack envelope glycoproteins, in Vero cells ( Fig. 2A, B). Although this suggests that RV envelope proteins contribute to the infectivity of the pseudotype viruses, they seem to have little practical application because of their low infective titers. Co-expression of the Capsid (C) protein resulted in production of the pseudotype viruses, VSV GFP -RV/CE2E1 and VSV FLuc -RV/CE2E1 and these pseudotype viruses showed higher infectivity titers than VSV GFP -RV/E2E1 and VSV FLuc -RV/E2E1, respectively ( Fig. 2A, B). The infectivity titers for VSV GFP -RV/CE2E1 and VSV FLuc -RV/CE2E1 were 50-200-fold higher than those of VSV GFP -∆G and VSV FLuc -∆G, respectively. An experiment indicated that the RV C protein promotes fusion activity in RV envelope (E1 and E2) proteins by supporting the maturation or stabilizing either E2 and E1 or their interactions during intracellular transport to the cell surface 31 . We have confirmed the enhance effect by the C protein on fusion by RV envelope proteins. The surface expression level of the E1 protein with the C protein was similar to that without the C protein (Fig. 2C). The total amounts of the E1 protein in cells were also similar between cells co-expressed with or without the C protein (Fig. 2D). Nevertheless, the level of cell-to-cell fusion was increased by ~two-fold by co-expressing the C protein (Fig. 2E). Although the detailed mechanism was unclear, the data demonstrated that the RV envelope protein expressed on the cell surface showed a better fusion activity than that expressed without the C protein. Thus, in the following experiment, the C protein was provided together with RV envelope E2 and E1 proteins. However, it should be noted that co-expression of the C protein with the E1 and E2 proteins may produce empty non-infectious RV-like-particles (RVLP). Indeed, an electron microscopic assay revealed both bullet-shaped (~61 ± 7 nm × 173 ± 18 nm; n = 5) and spherical particles (~73 ± 12 nm in diameter; n = 8), which were corresponding to VSV and RV virions, respectively 9, 32 . This suggests that pseudotype VSV and RVLP were contained in the VSV GFP -RV/CE2E1 stocks (Fig. 2F). No particles showed a combined shape of the bullet and spherical forms. Although the VSV pseudotype virus genome was expected to be incorporated into the bullet shaped particles, it was confirmed by the following experiment. Stock solutions of VSV FLuc -G, VSV FLuc -RV/CE2E1 and RVLP were independently prepared and concentrated and fractionated by sucrose-gradient ultracentrifugation. In this experiment, RVLP, which contains a subgenomic replicon RNA in which the structural genes are replaced with DNA sequences encoding the puromycin N-acetyl-transferase protein, the foot-and-mouth disease virus 2 A self-cleavage domain, and the Renilla luciferase 33 , was used as a marker for RVLP fractionation pattern. VSV FLuc -G, VSV FLuc -RV/CE2E1, and RVLP in individual fractions were detected by infecting Vero cells with the fractions and measuring the relative light unit (RLU) from the cells. The peak infectivity of VSV FLuc -RV/CE2E1 was at ninth fraction similarly to that of VSV FLuc -G, with no peak or shoulder at seventh fraction where RVLP indicated the peak infectivity ( Fig. 2G-I). The profiles demonstrated that the pseudotype VSV particles, but not RVLP, incorporated the genome.
VSV-RV/CE2E1 undergoes a similar entry process to the authentic RV. Infection of VSV FLuc -RV/ CE2E1, but not VSV FLuc -G, which has the VSV-G protein, was inhibited by anti-RV serum in a dose-dependent manner (Fig. 3A, B), confirming the functional contribution of RV proteins in VSV FLuc -RV/CE2E1 infections. RV enters cells through receptor-binding-mediated endocytosis and subsequent low pH-triggered viral-host membrane fusion [34][35][36] . To analyze the entry process of VSV FLuc -RV/CE2E1, the effects of endosomal acidification inhibitors on VSV FLuc -RV/CE2E1-infections were examined. Vero cells were pretreated with various concentrations of bafilomycin A1 and chloroquine, and infected with VSV FLuc -RV/CE2E1 and three control pseudotype VSVs bearing envelope proteins of VSV, murine leukemia virus (MLV), and measles virus (MV) (VSV FLuc -G, VSV FLuc -MLV/Env, and VSV FLuc -MV/FH, respectively). Similar to RV, entry of the authentic VSV occurs in endosomes under low pH conditions 37 , while MLV and MV enter cells in the plasma membrane under neutral pH [38][39][40] . As expected, the inhibitors blocked VSV FLuc -G infections in a dose-dependent manner, but did not block VSV FLuc -MLV/Env and VSV FLuc -MV/FH infections (Fig. 3C, D). The effect of the inhibitors on VSV FLuc -RV/ CE2E1 was similar to that on VSV FLuc -G. Another feature of RV entry is its Ca 2+ dependency 35 . Hence, the Ca 2+ requirements of VSV GFP -RV/CE2E1 infection were assessed. The number of VSV GFP -RV/CE2E1-infected cells declined severely when the level of CaCl 2 in the culture media was lowered (Fig. 3E). VSV GFP -MV/FH infectivity, which was used as a control, was reduced by only ~20% at a maximum. All these data suggest that VSV GFP -RV/ CE2E1 undergoes a similar entry process to the authentic RV.
Human non-immune cells are generally susceptible to VSV-RV/CE2E1, whereas immune cells are much less susceptible than non-immune cells. We analyzed VSV-RV/CE2E1, VSV-G, and VSV-∆G for their infectivity levels in various human cell lines and Vero cells. To compare the different pseudotype viruses, the infectivity titers were standardized by the genome copy numbers in the virus stocks. The genome copy numbers in VSV GFP -RV/CE2E1, VSV GFP -G, and VSV GFP -∆G stocks used in the experiments  [8][9][10][11] , respectively, were mixed and cultured together. The cells were transfected with pcDNA3.1-E2E1, pcDNA3.1-CE2E1 or the empty vector and incubated for 32 hours. (D) The cell lysates were subjected to immunoblotting with anti-RV E1 and anti-GAPDH antibodies. The signal intensity of the E1 protein in the cells were 5.02 × 10 10 , 1.15 × 10 11 , and 4.29 × 10 10 , respectively, per milliliter. Also the genome copy numbers in VSV FLuc -RV/CE2E1, VSV FLuc -G, and VSV FLuc -∆G stocks used in the experiments were 9.35 × 10 9 , 2.81 × 10 10 , and 1.21 × 10 10 , respectively, per milliliter. VSV-G infectivity titers were generally much higher than those of VSV-RV/ CE2E1 (Fig. 4). Not surprisingly, these data suggest that RV envelope proteins were less efficiently incorporated and/or less functional in VSV-based pseudotype virions than the authentic VSV G glycoprotein. Nonetheless, the VSV GFP -RV/CE2E1 infectivity titers were significantly higher than VSV GFP -∆G in many cell lines (Fig. 4). VSV GFP -G infectivity titers in 293T, Huh7, NJG, JAR, and JEG3 cells were similar to that in Vero cells (Fig. 4A), as were the infectivity titers of VSV GFP -RV/CE2E1 in 293T and NJG cells in Vero cells (Fig. 4A). In contrast, the infectivity titers of VSV GFP -RV/CE2E1 in Huh7, JAR, and JEG3 cells were ~10-times greater than they were in Vero cells (Fig. 4A). Vero cells are commonly used for propagating RV and it is well accepted that this cell line is susceptible to RV. These data therefore suggest that these non-immune cell lines (293T, Huh7, NJG, JAR, and JEG3) are similarly or more highly susceptible to RV than Vero cells. VSV GFP -G infectivity titers were reduced by 10 to 1,000-fold in HeLa, FLC-4, FaDu, Detroit562, HSQ89, and A549 cells, when compared with the titer in Vero cells (Fig. 4A, B). Therefore, these cell lines may be less susceptible or less permissible to VSV infection. Even when considering these observations, the infectivity titers of VSV GFP -RV/CE2E1 in these cells were significantly higher than those of VSV GFP -∆G (Fig. 4B), suggesting that these cell lines are also susceptible to RV. Similar experiments were performed using VSV FLuc -RV/CE2E1, of which the infection was highly sensitively quantified by the luciferase activity. In these experiments a VSV G protein-specific antibody, which neutralized VSV FLuc -G infection efficiently, was used to eliminate the possible effect by the residual VSV FLuc -G used for the production of VSV FLuc -RV/CE2E1. Data with VSV FLuc -RV/CE2E1 and the VSV-G neutralizing antibody confirmed the negligible or small effect by the residual VSV FLuc -G and the high susceptibility of non-immune cells to VSV FLuc -RV/ CE2E1 (Fig. 4C).
In contrast, VSV GFP -RV/CE2E1 infectivity titers remained as low as those of VSV GFP -∆G in immune cell lines, and the infectivity titer, if any, of VSV GFP -RV/CE2E1 could not be detected in most immune cell lines (data not shown). To assess the susceptibility of the immune cell lines to VSV-RV/CE2E1 more sensitively, they were infected with VSV FLuc -RV/CE2E1, and any luciferase activity by the residual VSV FLuc -G was eliminated by the VSV G protein-specific antibody (Fig. 4D). As was expected, the infectivity titers of VSV FLuc -RV/CE2E1 in immune cells were undetectable in many immune cell lines (Fig. 4D). However, significant, but very low, levels of infectivity titers were detected in THP-1, Jurkat, and MT2 cells (Fig. 4D).

Stimulation of immune cells by phorbol 12-myristate 13-acetate (PMA) increases the susceptibility of immune cells to VSV-RV/CE2E1.
The above data demonstrated that immune cells are much less susceptible to RV than non-immune cell lines. However, previous studies demonstrated that lymphocytes become susceptible to RV after stimulation by mitogen [41][42][43] . U937 and THP-1 cells, derived from histiocytic lymphoma and monocytic lymphoma, respectively, were stimulated with PMA, and then infected with VSV FLuc -RV/ CE2E1 or VSV FLuc -G. Unstimulated control cells were also infected with VSV FLuc -RV/CE2E1 or VSV FLuc -G. The positive effect on VSV FLuc -RV/CE2E1 entry by PMA stimulation was evident in THP-1 cells. The infectivity of VSV FLuc -G was unchanged by PMA stimulation, but the infectivity of VSV FLuc -RV/CE2E1 was increased by ~30-fold (Fig. 5). Similar effect was also observed in U937 cells. The infectivity of VSV FLuc -G and VSV FLuc -RV/ CE2E1 were increased by ~10-fold and ~200-fold, respectively (Fig. 5). Therefore, the increase in VSV FLuc -RV/ CE2E1 infection was possibly in part due to the positive effect by PMA for VSV replication. However, further increase in VSV FLuc -RV/CE2E1 suggested that the VSV FLuc -RV/CE2E1 entry by RV envelope proteins was promoted by PMA stimulation. These data suggested that immune cells become susceptible to RV infection after stimulation, although the levels are still much lower than those in non-immune cells.

Discussion
The efficiency with which RV is able to infect different cell lines differs dramatically; however, it is not known which steps restrict RV infection in cells that do not support infection by this virus. Pseudotype viruses are useful tools for analyzing the susceptibility of specific viruses separately from their intracellular replication processes. A pseudotype virus with RV envelope proteins has been reported previously 31 . This is a lentiviral pseudotype virus based on the simian immunodeficiency virus (SIV). However, the SIV vector failed to incorporate the intact forms of RV envelope proteins 31 . Therefore, the SIV pseudotype virus bearing modified RV envelope proteins with the cytoplasmic tail of the VSV G protein was used in the study 31 . In the present study, a new was normalized by those of GAPDH. The full-length images are shown in Supplementary Fig. 2. (E) The cells were incubated with low pH (pH 5.1) media for 15 min and then were cultured with the standard culture media for 8 h. The Renilla luciferase activity derived from fusion cells were measured and normalized by expression levels of the E1 protein determined in (D). (F) Electron microscope image of particles in the VSV-RV/CE2E1 stock solution. Purified virions in the VSV GFP -RV/CE2E1 stock solution were fixed with 2% paraformaldehyde, and then were negatively stained with 2% phosphotungstic acid solution. The arrows indicate spherical particles. Bar    pseudotype virus bearing the intact forms of the RV envelope proteins was generated using a VSV pseudotype system. Although the E1 protein alone causes membrane fusion and supports virus entry, the E2 and C proteins play roles in RV entry. After E1 protein binding to a receptor, RV enters cells by endocytosis 44 . Exposure to the low pH environment in early endosomes induces conformational changes in the E1 and E2 proteins and membrane fusion 34,36 concomitantly with a solubility change in the C protein 36 . The C protein also has a supportive role in membrane fusion 31 . The mechanism of fusion enhancement by the C protein is still unclear, but the C protein likely supports the maturation or stabilizes either E2 and E1 or their interactions during intracellular transport to the cell surface 31 . As expected, co-expression of the C protein promoted the production of infectious VSV pseudotype virus with RV envelope proteins (VSV-RV/CE2E1). Infection by the novel VSV-RV/CE2E1 pseudotype virus was indeed mediated by RV envelope proteins, and the virus underwent a similar entry process to that of the authentic RV as previously reported 35,44 .
Using the VSV pseudotype system, we have shown that human non-immune cell lines are generally susceptible, while most immune cell lines are much less susceptible to RV than non-immune cells. Therefore, inefficient infection of immune cell lines with the authentic RV is explainable in part by the poor susceptibility of these cell lines to RV. However, previous studies have demonstrated that immune cells can become infected with RV 41-43, 45, 46 . RV replicates in peripheral blood mononuclear cells (PBMCs), of which macrophages are the main target by RV 42 . Consistent with these observations, RV is isolatable from the PBMCs of patients naturally infected with RV 46 . Unstimulated lymphocytes poorly support RV infection 42 , but are able to support RV infection upon stimulation by mitogen [41][42][43] . Additionally, a previous study showed that Raji and Cess human B-cell lines and the U937 monocyte line all supported RV infection 45 . Our data do not necessarily contest these previous observations, because a basal level of RV infection was observed in all the immune cell lines, although the efficiency was very low. In addition, our data further demonstrated that the infection of immune cell lines was increased to a certain level by stimulation of the cells. The key point about our data is that the entry efficiency of RV was generally high in non-immune cell lines but not in immune cell lines. These observations are important in terms of understanding more about the pathology of RV.
Even among non-immune cell lines RV infection efficiency differs dramatically, although our data using a pseudotype virus system shows that non-immune cells are generally susceptible to RV. These data suggest that the permissibility to RV infection differs considerably among non-immune cell lines. One possible factor modulating the RV infectivity is the host innate immune system. RV is highly sensitive to interferon (IFN) 3,47,48 , and the infectivity of individual cell lines with RV is affected by their capacity for IFN production and response to IFN. Vero cells are defective in IFN production 49,50 , and JEG3 and JAR trophoblast cell lines, which support RV infection most efficiently, are refractory to IFN 51,52 . Therefore, the activity of innate immunity in individual cells at least partially determines the cell or tissue tropism of RV infection. CRS is a major concern in RV infections. The high infectivity of RV in trophoblast cell lines is an interesting observation that could open a door towards understanding the establishment of RV infection in the fetus after it crosses the placenta. In the fetus, persistent infection of the endothelial cells of the fetal vessels with RV is probably a cause of CRS, because it may induce vascular abnormalities, resulting in dysfunctional or abnormal development of multiple fetal organs or tissue 53 . Perelygina et al. have shown that human fetal endothelial cells are persistently infected with RV 54,55 . Further studies on the cell and tissue tropism of RV and the molecular mechanisms involved in such tropism are essential if we are to understand the pathophysiology of rubella and CRS. The new RV pseudotype system established in the present study will make a positive contribution to these additional studies. Plasmid constructs. The expression plasmid encoding the precursor protein for the structural proteins (C, E2 and E1) of the RV Hiroshima strain (pcDNA3.1-SP/C 1-300 ) has been described previously 33 . The expression plasmid encoding only E2 and E1 proteins of the RV Hiroshima strain (pcDNA3.1-E2E1) has been described previously 33 . The expression plasmids encoding the envelope proteins of VSV (pC-VSV-G), the Edmonston vaccine strain of MV (pCA7PS-Ed-H and pCXN-Ed-F), and MLV (pFBASALF) have also been described previously 13, 58-60 . pcDNA3.1-SP/C 1-300 is hereafter referred to as pcDNA3.1-CE2E1.

CCID 50 infectivity titration of RV in various cell lines.
A stock solution of rHS was serially diluted 4-fold. Next, monolayers of non-immune (Vero, SH-SY5Y, SK-N-MC, 293T, HUEhT-1, Caco-2, HeLa, FLC-4, Huh7, FaDu, Detorit562, A549, HSQ89, NJG, JAR and JEG3) cells in 96-well plates were cultured with the diluted rHS samples at 35 °C. After a 4 day-incubation period, the cells were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. Then, the RV-infected cells were detected by an indirect immunofluorescent assay using the mouse monoclonal antibody specific for the RV C protein and an Alexa Fluor 594-conjugated goat anti-mouse secondary antibody (Thermo Fisher Scientific). The CCID 50 was calculated using the Spearman-Karber formulation 61,62 . Immune (Raji, THP-1, U937, M8166, Jurkat, MT2) cells in 96-well plates (2.0 × 10 4 cells/well) were also cultured with the 4-fold diluted rHS samples for 4 days at 35 °C, and then fixed with 4% paraformaldehyde. The fixed cells were transferred to V-bottomed 96-well plates and centrifuged to remove the 4% paraformaldehyde fixing solution. The cells were then permeabilized with 0.5% Triton X-100. After removing the permeabilizing solution, indirect immunofluorescent staining was performed using the mouse monoclonal antibody specific for the RV C protein and an Alexa Fluor 594-conjugated goat anti-mouse secondary antibody (Thermo Fisher Scientific). The cells were transferred to flat-bottomed 96-well plates to observe the RV-infected cells using a fluorescence microscope. To discriminate the fluorescent signals from the RV-infected cells from non-specific fluorescent signals, cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (Lonza).

Pseudotype virus generation.
To generate VSV GFP -RV/CE2E1, 293T cells, cultured in 6-well collagen-coated plates (8.0 × 10 5 cells/well), were transfected with the pcDNA3.1-CE2E1 expression plasmid using branched polyethylenimine (Sigma-Aldrich). At 32 h posttransfection, the cells were infected with VSV GFP -G at a multiplicity of infection (MOI) of 3.0 13,16,22,30,63 . The VSV GFP -G genome lacks the G gene, which is replaced with the GFP gene 13 . On the viral envelope the VSV GFP -G particles contain the G protein, which is provided in trans using the pC-VSV-G expression plasmid 13 ). The cells were washed four times with DMEM and incubated with DMEM containing 10% FBS. After 24 h, the culture supernatants containing VSV GFP -RV/ CE2E1 were harvested and centrifuged at 10,000 × g for 5 min at 4 °C to remove the cell debris. VSV FLuc -RV/ CE2E1 was generated similarly to VSV GFP -RV/CE2E1 using VSV FLuc -G, which encodes the FLuc gene, instead of the GFP gene 14,28 , at a MOI of 0.3. VSV GFP -RV/E2E1 and VSV FLuc -RV/E2E1 were also generated similarly to VSV GFP -RV/CE2E1 and VSV FLuc -RV/CE2E1, respectively, using pcDNA3.1-E2E1 instead of pcDNA3.1-CE2E1. VSV GFP -G, a gift from Dr. M. A. Whitt (University of Tennessee, TN), and VSV FLuc -G were propagated as reported previously 13,14 . VSV GFP -∆G and VSV FLuc -∆G, which lack envelope proteins, were generated using pcDNA3.1 + (Thermo Fisher Scientific) instead of pcDNA3.1-CE2E1 13,14 . VSV FLuc -MLV/Env was generated similarly to VSV FLuc -RV/CE2E1 using pFBASALF 58 instead of pcDNA3.1-CE2E1. VSV GFP -MV/FH and VSV FLuc -MV/ FH were also generated similarly to VSV GFP -RV/CE2E1 and VSV FLuc -RV/CE2E1, respectively, using pCXN-Ed-F and pCA7PS-Ed-H 30, 59, 60 instead of pcDNA3.1-CE2E1. To generate VSV GFP -MV/FH and VSV FLuc -MV/FH, a fusion-blocking peptide (Z-D-Phe-Phe-Gly) (The Peptide Institute, Osaka, Japan) was used to inhibit syncytium formation during the production of these pseudotype viruses, as reported previously 30,59 . VSVΔG/GFP-*G 63  Electron microscopy image of viral particles. The stock solution of VSV GFP -RV/CE2E1 was centrifuged at 1,500 × g for 10 min at 4 °C twice to remove cell debris. The viruses were collected by centrifugation at 120,000 × g for 1 h at 4 °C onto 20% (wt/vol) sucrose cushions 9 . The pellets were resuspended in phosphate-buffered saline (PBS). The ultracentrifugation step was repeated. After fixation with 2% paraformaldehyde in PBS, the viruses were placed on a carbon-coated grid for 45 s, rinsed with distilled water, stained with 2% phosphotungstic acid solution and observed with an electron microscope (TEM-1400, JEOL, Tokyo) operating at 80 kV as previously reported 64 . Flow cytometry for detection of the E1 protein. 293T   Immunoblotting for detection of the E1 protein. 293T cells were transfected with pcDNA3.1-CE2E1 or pcDNA3.1-E2E1 expression plasmids. At 36 h posttransfection, the cells were analyzed by immunoblotting using anti-RV E1 or anti-GAPDH antibody as previously reported 33 . The signal intensity of the E1 protein was normalized by that of GAPDH.
A DSP-based fusion assay. A DSP-based-fusion assay was performed as described previously 56,57 . A fusion protein of Renilla luciferase and GFP was split into two fragments designated dual split proteins, DSP 1-7 and DSP 8-11 56 . Although each fragment lacks the activities as Renilla luciferase and GFP, they reassemble and become functional when they are expressed within an identical cell simultaneously. 293CD4/DSP 1-7 and 293FT/DSP [8][9][10][11] cells constitutively expressing DSP 1-7 and DSP [8][9][10][11] , respectively, were mixed and cultured together, and transfected with pcDNA3.1-CE2E1, pcDNA3.1-E2E1 or an empty pcDNA3.1 vector. At 32 h posttransfection, the cells were incubated with a low pH (pH5.1) culture media for 15 min to induce a cell-to-cell fusion by RV envelope proteins followed by the incubation with a standard culture media for 8 hours at 37 °C. Then the Renilla luciferase activity derived from the reassembled DSPs by a cell fusion were measured using the Renilla Luciferase Assay System (Promega, Madison, WI) and GloMax 20/20 Luminometer (Promega). The expression levels of Renilla luciferase were normalized by the total expression levels of E1 proteins detected by an immunoblotting as described above.
Sucrose gradient fractionation analysis of pseudotype viruses. The pseudotype virus particles from VSV FLuc -RV/CE2E1 and VSV FLuc -G in 10 ml stock solutions were precipitated using PEG-it Virus Precipitation Solution (System Biosciences, Mountain view, CA), and the precipitated virus particles were then resuspended in 0.5 ml of DMEM. RVLP, which have been reported previously 33 were generated and also precipitated and resuspended in 0.5 ml of DMEM, as were VSV FLuc -RV/CE2E1 and VSV FLuc -G. The resuspended VSV FLuc -RV/CE2E1, VSV FLuc -G and RVLP were layered onto 12 ml of a 20 to 40% (wt/vol) sucrose density gradient and fractionated by ultracentrifugation in an SW41 rotor (Beckman Coulter, Tokyo, Japan) at 145,000 × g for 2 h. Each 1 ml of the fractionated samples in the gradient was collected from the top to the bottom, and a small part of each fraction was diluted with an equal volume of DPBS. Next, the diluted samples were inoculated into the Vero cell culture media in the 96-well plates. For VSV FLuc -RV/CE2E1 and VSV FLuc -G, the RLU from the cells was detected after 24 h of incubation using the Bright-Glo Assay System (Promega) and POWER SCAN HT (BioTek, Winooski, VT), after which RLU was used as an indicator of infection with the pseudotype viruses. For RVLP, the RLU from the cells was detected after a 72 h incubation period using the Renilla Luciferase Assay System and GloMax 20/20 Luminometer, after which RLU was again used as an indicator of infection with RVLP.
Neutralization assays for VSV FLuc -RV/CE2E1 and VSV FLuc -G. Goat antiserum against RV and an unimmunized normal goat serum were 4-fold serially diluted in DMEM. VSV FLuc -RV/CE2E1 and VSV FLuc -G, which express FLuc, were mixed with the diluted sera and incubated for 1 h at 4 °C. Then, the Vero cell monolayers in 96-well plates were infected with VSV FLuc -RV/CE2E1 and VSV FLuc -G pretreated with the sera. At 24 h p.i., the RLU from the cells was measured using the Bright-Glo Assay System and POWER SCAN HT. Analysis of calcium dependency in pseudotype virus infections. Calcium dependency in the pseudotype virus infections was assessed using a method similar to that reported previously 35 . Vero cells in 96-well plates were incubated with VSV GFP -RV/CE2E1 and VSV GFP -MV/FH, which express GFP, for 2 h at 4 °C, and then washed with pre-chilled DPBS, twice. The cells were then cultured in DMEM free from CaCl 2 (Thermo Fisher Scientific) or containing various concentrations of CaCl 2 for 1 h at 37 °C. The cells were then cultured in DMEM containing 2 mM CaCl 2 and 10% FBS at 37 °C. At 24 h p.i. the number of GFP-expressing cells, which correspond to VSV GFP -RV/CE2E1-or VSV GFP -MV/FH-infected cells, was counted using a fluorescence microscope. Genome copy number measurements for VSV GFP -RV/CE2E1, VSV GFP -G, and VSV GFP -∆G. The genome copy number for VSV GFP -RV/CE2E1, VSV GFP -G, and VSV GFP -∆G was analyzed by reverse transcription-quantitative PCR (RT-qPCR) using a set of primers specific for the VSV N gene, as reported previously 19 .

Infectivity titration of VSV
Statistical analysis. Two-tailed t-tests were used to determine significant differences among pseudotype virus infectivity titers in various cell lines. Data Availability. The datasets analyzed during the current study are available from the corresponding author on reasonable request.