Fusion of Intra- and Extracellular Forms of Vaccinia Virus with the Cell Membrane

Transcription

1 JOURNAL OF VIROLOGY, OCt. 199, p X/9/ $2./ Copyright 199, American Society for Microbiology Vol. 64, No. 1 Fusion of Intra- and Extracellular Forms of Vaccinia Virus with the Cell Membrane ROBERT W. DOMS,1,2* ROBERT BLUMENTHAL,3 AND BERNARD MOSS' Laboratory of Viral Diseases, National Institute of Allergy and Infectious Diseases,' and Laboratory of Pathology2 and Section on Membrane Structure and Function,3 National Cancer Institute, Bethesda, Maryland 2892 Received 14 May 199/Accepted 3 June 199 The membrane fusion activities of the isolated single-enveloped intracellular form of vaccinia virus (INV) and the double-enveloped extracellular (EEV) form were studied by using a lipid-mixing assay based on the dilution of a fluorescent probe. Fluorescently labeled INV and EEV from both the IHD-J and WR strains of vaccinia virus fused with HeLa cells at neutral ph, suggesting that fusion occurs with the plasma membrane during virus entry. EEV fused more efficiently and with faster kinetics than INV: approximately 5% of bound EEV particles fused over the course of 1 h, compared with only 25% of the INV particles. Fusion of INV and EEV was strongly temperature dependent, being decreased by 5% at 34 C and by 9% at 28 C. A monoclonal antibody to a 14-kilodalton envelope protein of INV that has been implicated in the fusion reaction (J. F. Rodriguez, E. Paez, and M. Esteban, J. Virol. 61:395-44, 1987) completely suppressed the initial rate of fusion of INV but had no effect on the fusion activity of EEV, suggesting that vaccinia virus encodes two or more membrane fusion proteins. Finally, cells infected with the WR strain of vaccinia virus formed syncytia when briefly incubated at ph 6.4 or below, indicating that an acid-activated viral fusion protein is expressed on the cell surface. However, WR INV and EEV did not display increased fusion activity at acid ph, suggesting that the acid-dependent fusion factor is not incorporated into virions or that its activity there is masked. A critical step in the infectious entry pathway of enveloped viruses is fusion of the viral membrane with that of the cell (15, 24, 45, 51). Fusion is invariably mediated by viral membrane proteins and may occur directly with the plasma membrane or after the virus particle is endocytosed and delivered to endosomes. In endosomes, the low-ph environment triggers a conformational change in the viral protein which leads to fusion and release of the nucleocapsid into the host cell cytoplasm (15, 24, 45, 51). The entry pathway of poxviruses, including vaccinia virus, the prototype of this family, is not well characterized. Studies of poxviruses are complicated by the fact that two structurally distinct, infectious forms exist (1, 5, 26). Virus particles assemble in the cytoplasm of the host cell and obtain a lipid envelope. This intracellular virus (INV) is released upon lysis of the infected cell and is the form that is best characterized (26). While the bulk of the virus remains cell associated, a fraction buds into a cellular organelle, identified as the Golgi apparatus, and is released into the medium (1, 12, 17). As a result of this pathway, a second lipid envelope and a unique set of glycoproteins not found in INV surround the virion (3). The amount of double-membraned extracellular enveloped virus (EEV) varies between different vaccinia virus strains and host cell types (31). For the WR strain, -55% of the total virus particles are secreted as the EEV form, whereas for the IHD-J strain, up to 4% of the particles acquire the second envelope. Until recently, studies on the entry of vaccinia virus into cells have concentrated on INV and have relied largely on electron microscopy. Whereas some studies suggest that endocytosis is the more common route of entry or may even be a prerequisite for infection (9), others have shown that fusion can occur directly at the cell surface (2, 7). More recently, Janeczko and co-workers (19) have provided * Corresponding author. strong biochemical evidence that INV fuses in a ph-independent manner, thus implicating the plasma membrane as the site of fusion. In addition, a 14-kilodalton (kda) INV envelope protein that is likely involved in fusion has been identified (4). Considerably less is known about the entry of EEV into cells. It has been suggested, however, that EEV enters cells more quickly and efficiently than its INV analog and may play an important role in widespread dissemination of the virus in vivo (5, 32, 33). However, membrane fusion has not been directly monitored for either viral form. In this study, we have used a well-characterized fluorescence assay to directly measure the kinetics and extent of INV and EEV fusion with target cells (4, 13, 22). We found that both INV and EEV from the IHD-J and WR strains of vaccinia virus fused in a ph-independent manner. In addition, a monoclonal antibody to a 14-kDa viral envelope protein reported to be involved in the fusion reaction (4) was found to suppress INV, but not EEV, fusion, implying the existence of more than one viral membrane fusion protein. MATERIALS AND METHODS Materials. Octadecylrhodamine B chloride (R18) was obtained from Molecular Probes (Eugene, Ore.). Monoclonal antibody MAbC3 (39) was a generous gift of M. Esteban, State University of New York Health Science Center, Brooklyn. Virus preparation. Confluent BSC-1 cells were infected with the indicated vaccinia virus strain at 5 PFU per cell. The cells were scraped into the medium at 48 h after infection and sedimented at 1,5 x g for 5 min. The supernatant was collected, and the virus was pelleted in an SW28 rotor at 24, rpm for 3 min. The cell pellet was suspended in 1 mm Tris (ph 9.), and the cells were lysed with a Dounce homogenizer. After a low-speed centrifugation, the supernatant was collected and the virus was pelleted as described above. The virus pellets were resus- Downloaded from on August 21, 218 by guest 4884

2 VOL. 64, 199 pended in 1 mm Tris (ph 9.), and the EEV and INV were purified either on buoyant density CsCl gradients or by sucrose velocity gradient sedimentation. Buoyant density centrifugation was performed as described by Payne (3). For sucrose velocity gradient sedimentation, resuspended virus was placed over a 25 to 4% continuous sucrose gradient in 1 mm Tris (ph 9.). Centrifugation was at 15, rpm for 3 min in the SW28 rotor. Virus bands were collected, and the virus concentrated by pelleting. Fusion assay. Purified vaccinia virus (1,g of viral protein) was labeled with 2.,ul of 1 mglml R18 for 2 min at room temperature. Free R18 was removed by pelleting the virus twice in phosphate-buffered saline with.2% bovine serum albumin. Labeled virus was suspended in fusion medium, consisting of RPMI medium without phenol red and with 1 mm N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) and 1 mm 2-(N-morpholino)ethanesulfonic acid (ph 7.4), by vortexing and brief sonication. Virus was bound to HeLa cells in fusion medium at 4 C for 3 min. The virus-cell suspension was washed twice by low-speed centrifugation to remove unbound virus. The fluorescence intensity of R18-labeled virus ranged from 5 to 2% that of labeled virus incubated with Triton X-1. Fusion activity was measured as previously described (4, 8). Briefly, 1,ul of the R18-labeled vaccinia virus-hela cell suspension was added directly to 2 ml of fusion medium prewarmed to 37 C in a disposable plastic fluorescence cuvette. Cells were kept in suspension by the use of a magnetic stir bar. Fluorescence was measured by using a model 8 (SLM Aminco) or an MPF-44B (Perkin-Elmer) spectrofluorometer with a 1-s time resolution. The excitation wavelength was set to 56 nm, and the emission wavelength was set at 59 nm. A 57-nm filter was used to reduce light scatter. Experimental manipulations included varying the temperature or ph of the medium as well as the inclusion of the lysosomotropic agent ammonium chloride, in which case cells were also incubated in the presence of the compound during the virus-binding period. For trypsinization experiments, R18-labeled virus was incubated with an equal volume of trypsin (.25 mg/ml) for 3 min at 4 C. A 1-fold molar excess of soybean trypsin inhibitor was then added, and the virus was washed by centrifugation. Virus was also incubated with MAbC3 for 6 min before cell binding. To monitor fusion with erythrocytes, fresh chicken erythrocytes were washed and labeled with R18 as previously described (25). Labeled erythrocytes were then bound to the surface of BSC-1 cells that had been infected with the appropriate virus 12 h earlier. Binding was for 1 min at room temperature, after which the cells were washed with ice-cold fusion medium and scraped into a suspension. Samples of the erythrocyte-bsc-1 cell suspension were diluted into prewarmed fusion medium, and the extent of fusion was monitored as described above. Cell-cell fusion. Polykaryon formation was monitored as previously described (1). Briefly, confluent BSC-1 cells were infected with vaccinia virus. At different times after infection, cells were fixed in methanol and stained with Giemsa solution. The extent of polykaryon formation was determined at 12 h postinfection by calculating the fraction of cell nuclei present in multinucleated cells (three or more nuclei). To determine the ph dependence of polykaryon formation, virus-infected cells were immersed in prewarmed (37 C) fusion medium [phosphate-buffered saline with 1 mm 2-(N-morpholino)ethanesulfonic acid and 1 mm HEPES] at the desired ph for 2 min. The cells were then returned to normal growth medium, and polykaryon forma- E._. VACCINIA VIRUS FUSION ACTIVITY 4885 Bottom Fraction Number FIG. 1. Characterization of INV and EEV. RK13 cells were infected with the IHD-J strain of vaccinia virus and labeled with [35S]methionine at 4 h postinfection. After continuous labeling for 24 h, virus secreted into the medium or isolated after cell lysis was applied to CsCl gradients as described in Materials and Methods. Gradients were fractionated from the bottom, and the amount of virus was determined by scintillation counting. Virus particles were also treated with Triton X-1 (TX1) before centrifugation. The material present at the top of the Triton X-1 gradients represents detergent-extracted viral envelope proteins. tion was assayed 2 h later. Both the time and the temperature at low ph were varied as indicated in the text. Plaque assays. Confluent BSC-1 cells were infected with serial dilutions of purified INV or EEV from either the IHD-J or WR strain. Infections were performed in bicarbonate-negative RPMI medium with 2 mm HEPES (ph 7.6). In addition, ammonium chloride or chloroquine was added as indicated in the text. After 3 min, the cells were washed and placed in fresh RPMI medium with the appropriate lysosomotropic agent for an additional 8 h. The cells were then washed and placed in normal growth medium for an additional 28 h, after which plaques were visualized and counted after staining with crystal violet. RESULTS Characterization of INV and EEV. A variable proportion of vaccinia virus is secreted from infected cells after acquiring a second lipid envelope (31). The remaining virus is cell associated and can be easily isolated after mechanical disruption of the infected cells. Because of the presence of a second envelope, EEV can be separated from INV on buoyant density gradients. We used techniques described by Payne (3) to isolate both INV and EEV for fusion studies. Initially, we used the IHD-J strain of virus and RK13 cells, a system that has been shown to yield large quantities of EEV (31). Approximately 9% of the virus recovered from the medium of infected cells sedimented with a buoyant density characteristic of the double-enveloped virus (Fig. 1). The remaining virus cosedimented with INV. In contrast, approximately 75% of virus released by hypotonic lysis and mechanical disruption of infected cells sedimented at the INV position, with the remaining 25% representing EEV that had presumably not yet been secreted or had re-bound to the cell surface after release. When gradient-purified INV and EEV were extracted with the nonionic detergent Triton X-1, both forms cosedimented at a position distinct and more dense than INV (Fig. 1). This change indicated that the difference in INV and EEV 2 TOP Downloaded from on August 21, 218 by guest

3 4886 DOMS ET AL. TABLE 1. Virus strain Effects of lysosomotropic agents and MAbC3 on plaque formationa Form Plaque formation (%) in the presence of: NH4C1 Chloroquine MAbC3 WR INV WR EEV IHD-J INV ND IHD-J EEV ND Influenza Japan 5 1 >95 a Confluent BSC-1 cells were infected with purified INV or EEV of the indicated virus strain in the presence of either ammonium chloride (2 mm) or chloroquine (.1 mm) as described in Materials and Methods. After 36 h, the cells were stained and the number of plaques was counted. The number of plaques formed in the presence of ammonium chloride or chloroquine was then expressed as a percentage of the number of plaques formed in the absence of lysosomotropic agents. The results shown are averaged from three experiments. As a control, cells were also infected with the Japan strain of influenza virus in the presence or absence of the agents. Cells infected with influenza can bind erythrocytes as a result of expression of the influenza hemagglutinin on the plasma membrane, thus providing a convenient infectivity assay (1). After 8 h, the cells were washed with saline, after which a 1% solution of human erythrocytes in saline was added for 15 min. The cells were then washed, and the fraction of BSC-1 cells binding erythrocytes was determined. Greater than 95% of the cells bound erythrocytes when infections were carried out in the absence of lysosomotropic agents. The effect of MAbC3 on WR infectivity was determined as described by Rodriguez et al. (39). Briefly, 2 PFU of virus was incubated with a 1-2 dilution of MAbC3 for 6 min, after which the virus-antiserum sample was adsorbed to cells as usual. ND, Not determined. buoyant density was due to the presence of an additional lipid envelope in the EEV form. EEV sedimented more slowly than INV on sucrose velocity gradients, providing another method of separating the two virus forms. Since CsCl often leads to virus aggregation, sucrose velocity gradient centrifugation was routinely used to obtain virus for fluorescent labeling. Effects of lysosomotropic agents on virus entry. Agents that elevate endosomal and lysosomal ph (lysosomotropic agents) prevent penetration of viruses which rely on acidactivated fusion proteins but have no effect on viruses which fuse at neutral ph (24). Janeczko et al. have shown that lysosomotropic agents have no effect on the infectivity of INV, suggesting that this form of the virus enters cells by a ph-independent mechanism (19). We have repeated and extended these studies by using both INV and EEV. Confluent BSC-1 cells were infected with serial dilutions of either the INV or EEV form of the WR or IHD-J strain of vaccinia virus in the presence of ammonium chloride or chloroquine for 3 min. The cells were then washed and placed in normal growth medium containing the lysosomotropic agent. After 8 h, the cells were washed and returned to growth medium without the lysosomotropic agent. Plaques were counted 36 h postinfection. Ammonium chloride and chloroquine had little effect on the infectivity of either INV (in agreement with Janeczko et al. [19]) or EEV (Table 1). In addition, ammonium chloride was shown to have no effect on the fusion activity of these viruses when measured by the R18 dequenching assay (described below and shown in Fig. 3). These results suggest that both INV and EEV can fuse at neutral ph. Fusion of INV and EEV with HeLa cells. To directly monitor the fusion activities of INV and EEV, we used the well-characterized R18 dequenching assay initially described by Hoekstra et al. (13). The lipophilic R18 probe partitions into viral or cellular membranes at concentrations of about 4% and is highly self-quenched. R18-labeled virus is bound to cells at 4 C and then allowed to fuse by rapidly warming c (I) tn LL- Time (min) O 3 B a 2 1 io J. VIROL. Time (min) FIG. 2. Fusion of R18-labeled INV and EEV with HeLa cells. Purified INV and EEV were labeled with R18, washed, and bound to HeLa cells at 4 C for 3 min. The cells were washed to remove unbound virus, and 5-,ul samples were added to 2 ml of 37C fusion medium. Fluorescence dequenching was monitored for 6 min (A) and for 5 min (B). In both panels A and B, EEV fusion is shown as curve A and INV fusion is shown as curve B. Fusion of EEV, incubated at 56 C for 2 min before binding to cells, is shown in curve C in panel A. Fusion of INV, incubated at 56 C for 2 min before binding to cells, is shown in Fig. 3A. to 37 C. After fusion, the probe is rapidly diluted in the cell membrane and dequenching, with a concomitant increase in fluorescence, occurs. The extent of fusion can be determined by adding Triton X-1 at the end of the experiment, thus maximally diluting the probe and providing an endpoint. This assay has been used to characterize in great detail the fusion activities of several viruses, including vesicular stomatitis virus (4, 36), influenza virus (13, 22), and Sendai virus (14, 22). Sucrose gradient-purified INV and EEV were labeled with R18. Both the WR and IHD-J strains were used. Since the results obtained with these two strains were virtually identical, only the results obtained with WR are shown. After labeling, virus was bound to HeLa cells for 3 min at 4 C. Greater than 95% of the virus bound. Unbound virus was removed by washing, after which samples of the R18-labeled vaccinia virus-hela cell suspension were added to a 2-fold excess of prewarmed (37 C) medium directly in the spectrofluorometer. Figure 2A shows the results obtained with WR INV and EEV over a 6 min period at ph 7.4. Fusion began almost immediately and proceeded at a nearly constant rate during the course of the experiment. The results of several experiments showed that on average approximately 1% of the EEV particles fused per min compared to.5% of the INV particles. Fusion was then monitored for a 1-h period to Downloaded from on August 21, 218 by guest

4 VOL. 64, 199 VACCINIA VIRUS FUSION ACTIVITY 4887 c U) cn LL- -11 C: Time (min) LL. Time (min) C U) LL Time (min) FIG. 3. Fusion of INV and EEV at low ph and in the presence of ammonium chloride. Virus (INV in panel A and EEV in panel B) was labeled and bound to HeLa cells as described in Materials and Methods. Samples of the virus-cell suspension were added to prewarmed buffer at either ph 7 (curve A) or ph 5 (curve B). Samples of the virus-cell suspension were also incubated in the presence of 2 mm ammonium chloride for 3 min before the initiation of fusion at 37 C and also in the presence of ammonium chloride (curve C). Fusion of INV, incubated at 56 C for 2 min before binding to cells, is shown in curve D in panel A. The 56 C control for EEV is shown in Fig. 2A (curve C). determine the ultimate extent of fusion. The rates at which INV and EEV fused decreased with time (Fig. 2B). Ultimately, approximately 5% of the EEV fused, compared with 25% of the INV. To rule out the possibility of probe exchange arising in the absence of fusion, we incubated virus at 56 C for 2 min before cell binding. This treatment has been shown to abolish viral infectivity (9). We found that heat-inactivated virus still bound to cells with high efficiency (>9%) but that fusion did not occur (Fig. 2A for EEV; Fig. 3A for INV). Thus, the dequenching observed with native virus was not due to probe exchange but rather due to fusion itself. ph dependence of fusion. The fusion observed in Fig. 2 could be the result of fusion occurring with the plasma membrane at neutral ph or could be the result of virus fusing at acid ph within endosomes. To differentiate between these possibilities, we examined fusion under a variety of ph conditions as well as in the presence of ammonium chloride. We found that fusion of both INV and EEV occurred with nearly the same rates and extent within a ph range of 5. to 7.4 (Fig. 3). No more than a 2% rate difference was ever observed between high and low ph. Fusion was also unaffected when cells were incubated with ammonium chloride (Fig. 3). By contrast, similar concentrations of ammonium Time (min) FIG. 4. Effects of temperature on fusion kinetics. Samples of the virus-cell suspension were added to media at 37 C (curve A), 34 C (curve B), 31 C (curve C), and 25 C (curve D). The kinetics of INV fusion is shown in panel A; that of EEV is shown in panel B. Fusion at 28 C was only slightly faster than that observed at 25 C and is not shown for the sake of clarity. chloride inhibit the fusion of acid-activated viruses such as influenza virus (27, 46) and vesicular stomatitis virus (4) as measured by the R18 assay. Thus, we concluded that both INV and EEV fuse at neutral ph. Fusion activity was neither enhanced by nor dependent on acid ph. These results are most consistent with INV and EEV fusing directly with the plasma membrane at neutral ph. Temperature dependence of fusion. The fusion activity of enveloped viruses is usually temperature dependent (15, 24, 45, 51). We determined the temperature dependence of INV and EEV fusion by adding R18-labeled virus bound to HeLa cells to media at different temperatures. The fusion rates of INV (Fig. 4A) and EEV (Fig. 4B) showed a similar temperature dependence. The fusion rates were decreased by approximately 5% when the experiment was performed at 34 rather than 37 C (Fig. 5). The ultimate extent of fusion was also affected by temperature, with only 6% of total virus fusing at 34 C relative to 37 C over a 3-min period (Fig. 5). Longer incubations at reduced temperatures did not result in significantly increased fusion. Antibody to a 14-kDa envelope protein suppresses INV fusion. Recently, Rodriguez et al. (4) have identified a vaccinia virus protein that is likely involved in INV fusion. The 14-kDa protein forms disulfide-linked trimers in the envelope of INV (4). A monoclonal antibody to this protein, MAbC3, inhibits virus-induced cell fusion as well as penetration (but not binding) of virus into cells (38-4). We have tested the ability of this antibody, provided by M. Esteban, to inhibit fusion of both INV and EEV from the Downloaded from on August 21, 218 by guest

5 4888 DOMS ET AL. J. VIROL. CV) 1 8 EEV c a) > 6 Cu U Temperature (OC) FIG. 5. Effects of temperature on the extent of fusion. The experiments shown in Fig. 4 were repeated, with incubations lasting for 3 min. The extent of fusion at 37C after 3 min was defined as 1% for both INV and EEV. When fusion at 31 C was allowed to proceed for 2 h, only 1% additional fusion was observed. WR strain. Purified virions were incubated with MAbC3 for 6 min before cell binding. Binding was not inhibited by the antibody (data not shown), in agreement with earlier results (39, 4). The antibody was, however, strongly neutralizing for INV, in agreement with Rodriguez et al. (39; Table 1). In contrast, MAbC3 had no effect on the infectivity of EEV, suggesting that the 14-kDa protein is not displayed on the surface of EEV or that the epitope is not accessible (Table 1). We next tested the ability of MAbC3 to inhibit INV fusion. MAbC3 completely inhibited the initial rate of INV fusion, although a small degree of dequenching was observed at later times (Fig. 6). Higher concentrations of antibody had no additional effect. These data also show that fluorescence dequenching is the result of fusion, and not of binding followed by nonspecific probe exchange. Polykaryon formation. Expression of a viral fusion protein on the cell surface during the course of infection may, under the appropriate conditions, lead to syncytium formation. Thus, cells infected with viruses that fuse at neutral ph, such as Sendai virus and human immunodeficiency virus, 4 c 3 U- 2 1 Time (min) FIG. 6. Effects of MAbC3 on INV membrane fusion. R18-labeled WR INV was incubated with MAbC3 for 6 min. Cells were then added, and binding was allowed to proceed at 4 C for an additional 3 min. Fusion was then determined as for Fig. 2B. Incubation of EEV with MAbC3 had no effect on its fusion activity (not shown). Curve A shows fusion of WR INV in the absence of antibody; curve B shows fusion of WR INV that was first incubated with MAbC3. form multinucleated giant cells at neutral ph (15, 45, 51). Cells infected with acid-activatable viruses such as influenza virus and Semliki Forest virus can be induced to form syncytia by brief low-ph treatment. Thus, syncytium formation offers a simple and rapid means to initially assay for the production and cell surface expression of viral fusion factors. Earlier studies have shown that cells infected with the IHD-W strain of vaccinia virus spontaneously form syncytia, whereas cells infected with the IHD-J strain do not (16). To explore vaccinia virus-mediated syncytium formation in greater detail, cells infected with the WR, IHD-J, or IHD-W strain of vaccinia virus were assayed for fusion activity at both neutral and acid ph at different times after infection. We found that approximately 5% of the IHD-W-infected cells and 5% of WR-infected cells (Fig. 7A) formed polykaryons 12 h after infection, in agreement with previous studies (16, 2). By contrast, cells infected with IHD-J did not fuse at any time after infection. When the ph dependence of polykaryon formation at 12 h postinfection was examined, we found that WR-infected cells rapidly and efficiently fused to form polykaryons after brief (2-min) treatment with low-ph buffers. Fusion was detected at ph 6.4 (Fig. 7B) and was essentially complete (>9% of all cells) at ph 5.8 and below (Fig. 7C). Treatments at ph 5.8 for as little as 3 s were sufficient to trigger complete polykaryon formation. Approximately 5% fusion occurred when the ph treatment was for 1 s (Fig. 7), whereas fusion was not observed when cells were treated with ph 5.8 buffer for only 5 s. Polykaryon formation occurred with equal efficiency when acid treatments were performed at 37 C or room temperature (Fig. 7D) but did not occur at 4 C. These results are in agreement with those of Kohno et al. (2), who also found that WR-infected cells can fuse at acid ph. Cells infected with the IHD-J or IHD-W strain of vaccinia virus did not display enhanced fusion activity at acid ph. In an attempt to more precisely define the conditions under which cell-cell fusion occurs and to compare this with the fusion activity of intact virus, we used R18-labeled chicken erythrocytes as fusion targets. Erythrocytes have long served as model membranes for viral fusion studies (3, 15). They are easily labeled with R18 and can be bound to the surface of cells expressing the hemagglutinin proteins of paramyxoviruses or orthomyxoviruses (25, 41). Once bound, the erythrocytes fuse with the plasma membrane of the host cell at neutral ph (for paramyxoviruses) or acid ph (for orthomyxoviruses). Fusion can be quantitated by the increase in fluorescence that ensues after dilution of the R18 probe into the plasma membrane (13). Since vaccinia virus expresses a hemagglutinin on the surface of infected cells (16), we determined whether erythrocytes would serve as suitable target membranes. R18-labeled erythrocytes were bound to vaccinia virus-infected (either IHD-J or WR) BSC-1 cells at 4 C. Almost all infected cells bound several erythrocytes at 12 h postinfection. Samples of the R18- labeled erythrocyte-bsc-1 cell suspension were then added to 37 C medium to allow fusion to commence. However, fluorescence dequenching characteristic of fusion was not observed, even after acid treatment (data not shown). We also failed to detect fusion between R18-labeled virus and erythrocytes, nor did we detect viral hemolytic activity. Thus, erythrocytes did not serve as targets for vaccinia virus fusion factors even though they could be bound to the surface of infected cells. The formation of polykaryons (Fig. 7) could arise from the expression of a viral fusion factor(s) on the cell surface or by Downloaded from on August 21, 218 by guest

6 Bar. ISI.r "I 6: : t:~wl~ w i.= -; v *..;A.1 FIG. 7. Polykaryon formation. Confluent BSC-1 cells were infected with the WR strain of vaccinia virus as described in Materials and Methods. After 12 h postinfection, the cells were treated with buffers at different phs for 2 min or for shorter times. The cells were then returned to normal growth medium at 37 C, ph 7.4, for 2 h, after which they were fixed and stained, and the extent of polykaryon formation was determined. (A) ph 7.4; (B) ph 6.4 for 2 min at 37 C; (C) ph 5.8 for 2 min at 37 C; (D) ph 5.8 for 2 min at room temperature. No fusion was observed when uninfected cells were acid treated. release of intact virions followed by rebinding to the cell surface with subsequent fusion. Acid-activatable viruses such as influenza virus can be bound to the surface of cells and, when treated with low ph for a brief period of time, trigger massive polykaryon formation from without (5). Dales and Kajioka (9) reported that polykaryon formation from without is not seen with poxviruses. On the other hand, Ichihashi and Oie (18) found that trypsin-treated IHD-J vaccinia virus (the INV form) induced polykaryon formation 3 h after infection. The effects of ph on vaccinia virusinduced fusion from without have not yet been ascertained. Vaccinia virus (INV or EEV) was bound to the surface of BSC-1 cells for 1 h at 4 C. The cells were washed and then immersed in prewarmed (37 C) fusion buffer at different phs. Polykaryon formation was not observed with the WR, IHD-J, or IHD-W strain. The inability of the WR strain to induce fusion from without at acid ph, in contrast to its ability to trigger massive polykaryon formation from within under similar conditions, could be due to several reasons. The acid-dependent fusion factor(s) present on the plasma membrane of WR-infected cells might have been lost, masked, or inactivated during virus purification, the component may not be incorporated into virions, or perhaps the conditions used were not appropriate. Finally, trypsintreated IHD-J virus failed to initiate polykaryon formation under any conditions, though extensive cell clumping was observed. DISCUSSION VACCINIA VIRUS FUSION ACTIVITY W.4. Ir $.. 4, v, It 41!&6., a I 41..1, 11V W.12;,I 41,.4. w, D.. Previous studies on the entry of the INV form of vaccinia virus into cells have suggested that it may fuse directly with the plasma membrane or may enter by endocytosis (2, 7, 9). These routes of entry are not mutually exclusive, at least for viruses that fuse in a ph-independent manner. In the case of Sendai virus, fusion occurs directly with the plasma membrane, though with relatively slow kinetics (14). Thus, it would not be surprising if some fraction of cell-bound virus particles are endocytosed before they undergo fusion. Conceivably, these internalized viruses may fuse from within endosomes, albeit in a ph-independent manner. By contrast, viruses that fuse at acid ph can undergo fusion only after endocytosis and delivery to acidic intracellular organelles (15, 24, 45, 51). We have been able to extend previous studies on the entry of vaccinia virus into cells by directly monitoring the mem- ra- I..'. Y.w.1-.: 1- if 11.. "." - k Downloaded from on August 21, 218 by guest VOL. 64, 199 f.._ -AC,.. P.' W-1, 4r I 4; t.- 'c 'i k' e ' *.. 4 t`.'4 j AY.I X,.?-U\ jiu.44 " ; I. A. - I? C C.

7 489 DOMS ET AL. brane fusion process with the use of a sensitive and quantitative fluorescence assay. We found that the INV form of vaccinia virus fused at neutral ph and that acid ph neither enhanced nor diminished its fusion activity. These findings, coupled with the inability of lysosomotropic agents to inhibit viral fusion activity or infectivity (Table 1; 19), convincingly show that the INV form of vaccinia virus fuses in a phindependent manner. Thus, most virus particles probably fuse directly with the plasma membrane, though some may fuse during the brief time period between endocytosis and delivery to lysosomes and subsequent degradation. The fusion activity of EEV paralleled that of INV with one important difference: EEV consistently fused at twice the rate and efficiency of INV. These findings support previous work that has shown that EEV penetrates cells more quickly than its INV analog (35) and suggest that this is a direct result of increased fusion activity. We also found, as have others, that preparations of EEV typically had a higher PFU-to-particle ratio than similar preparations of INV. The increased fusion activity of EEV may account for some or all of this increase in infectivity. The kinetics and temperature dependence of vaccinia virus fusion are similar to those previously described for Sendai virus, which also fuses at neutral ph (13, 14). Our results agree with those of Payne and Norrby, who found that penetration of INV at 22 C was greatly reduced (35). However, in contrast to our results, Payne and Norrby found that EEV was capable of penetrating cells at this reduced temperature. However, the assay used by Payne and Norrby did not directly measure virus penetration but relied instead on the acquisition of antibody resistance by cell-associated virus. Our results, using the R18 assay, directly measured membrane fusion. We note that although influenza virus also fails to penetrate cells efficiently at reduced temperatures, it does undergo an irreversible binding step (44). A similar process could explain the acquisition of serum resistance in the absence of fusion activity seen at reduced temperatures with vaccinia virus. Likewise, Payne and Norrby found that up to 9% of cell-associated EEV became serum resistant after 1 h at 37 C, compared with 5% of INV (35). These values suggest approximately twice the extent of fusion that we have found. Several factors may conspire to inflate the apparent rate of penetration observed with the serum resistance assay. In addition to the possibility of an irreversible binding step, endocytosis of virus particles would presumably make them serum resistant even in the absence of fusion. These possibilities, coupled with differences in cell type and virus preparations, may account for this discrepancy. At present, relatively little is known about the viral proteins involved in binding and fusion. For the INV form, a 32-kDa protein has recently been shown to play an important role in virus binding (23). Interestingly, variants that lack the 32-kDa protein are still infectious, albeit with reduced titers. Thus, vaccinia virus may employ several binding mechanisms, which is perhaps not surprising given the structural complexity of the virus. Indeed, up to 3 viral envelope proteins have been identified (28), and a 54-kDa surface tubule protein has also been implicated in the binding reaction (47). As for membrane fusion, strong evidence has been presented which shows that a homotrimeric 14-kDa envelope protein plays an important role (4). In agreement with the results of Rodriguez et al. (4), we found that a monoclonal antibody to the 14-kDa protein did not inhibit virus binding but did strongly suppress fusion activity. Fusion was completely inhibited for the first 1 min, after J. VIROL. which a low rate of dequenching was observed. The failure of the antibody to completely inhibit fusion may have been due to limiting amounts of antibody, dissociation of the antibody either at the cell surface or after delivery of the virus to endosomes, or by nonspecific R18 probe transfer occurring only after long incubations. Further studies will be needed to clarify this point. Also, erythrocytes did not serve as fusion targets when bound to the surface of vaccinia virus-infected cells. Whether this reflects a requirement for a particular target membrane composition or the presence of a different receptor is not known. In contrast to INV, even less is known about the proteins involved in the binding and fusion activity of EEV. EEV acquires a second lipid envelope before its secretion from cells. Presumably, this outer envelope masks those proteins that are present only on the innermost envelope, which is acquired in the cytoplasm and covers the INV form. Indeed, we found that the antibody to the 14-kDa INV fusion protein had no effect on the fusion activity of EEV. It is not yet known whether the 32-kDa INV-binding protein is present on the surface of EEV as well. Thus, binding and fusion of EEV is likely mediated by additional proteins, with the obvious candidates being the set of glycoproteins present in the second lipid envelope (3). Of these, only the vaccinia virus hemagglutinin has been well characterized (34, 43). The membrane fusion activities of the best-characterized enveloped viruses have been shown to reside in single viral membrane proteins (11, 21, 29, 37, 48). An interesting possibility raised by this study is that vaccinia virus encodes several membrane fusion proteins. First, an antibody to the 14-kDa INV fusion protein failed to inhibit fusion of EEV, suggesting that EEV must contain a different viral fusion factor. Second, whereas WR vaccinia virus fused in a ph-independent manner, WR-infected cells formed polykaryons after brief acid treatment (Fig. 7; 2). The efficiency and kinetics of this process were similar to those displayed by several well-characterized acid-activated viral fusion proteins, including influenza virus hemagglutinin and vesicular stomatitis virus G protein (4, 49). This finding suggests that an additional, acid-activated viral fusion protein is expressed, though it may not be incorporated into virions. Alternatively, its activity may have been lost during virus isolation, or additional viral or cellular proteins may modulate its activity. Indeed, Seki et al. (42) have shown that vaccinia virus hemagglutinin may actually serve to inhibit viral fusion. Thus, vaccinia virus may produce several fusion proteins with different properties whose activities, in turn, can be regulated by additional viral gene products. It is of interest to note that herpes simplex virus encodes two proteins, gb and gd, which can independently catalyze acid-dependent polykaryon formation (6). Their activities also seem to be regulated by additional viral inhibitory factors (6). The fusion of EEV with the plasma membrane of host cells presents an interesting topological problem. If EEV is truly double enveloped, then fusion of the outer membrane will result in the introduction of a single-enveloped form of the virus into the cytoplasm of the host cell. In effect, fusion of EEV would be expected to release INV into the cytoplasm. How then is the innermost membrane removed and the virus uncoated? Obviously, fusion with a second cellular membrane would result in the delivery of the viral nucleocapsid to a site that is topologically equivalent to the extracellular space. Perhaps the viral capsid penetrates its sole remaining membrane by a non-fusion-dependent mechanism, which Downloaded from on August 21, 218 by guest

8 VOL. 64, 199 would be unique among the enveloped viruses studied thus far İn conclusion, vaccinia virus, like members of several other virus families, fuses in a ph-independent manner, thus implicating the plasma membrane as the major site of viral entry. The increased efficiency of EEV fusion relative to INV may help explain the more infectious nature of the secreted form. Finally, the assay used here will allow even more detailed studies of vaccinia virus fusion activity in the future and will undoubtedly help in the characterization of additional viral fusion and regulatory proteins. ACKNOWLEDGMENTS We thank Debi Sarkar, Michael Clague, Christian Schoch, and Anu Puri for helpful conversations and assistance with the R18 assay and computer graphics. We also thank Stuart Issacs and Christian Schoch for critically reading the manuscript, Norman Cooper for providing the HeLa cells, and Mariano Esteban for his generous gift of MAbC3. LITERATURE CITED 1. Appleyard, G., A. J. Hapel, and E. A. Boulter An antigenic difference between intracellular and extracellular rabbitpox virus. J. Gen. Virol. 13: Armstrong, J. A., D. H. Metz, and M. R. Young The mode of entry of vaccinia virus into L cells. J. Gen. Virol. 21: Blumenthal, R Membrane fusion. Curr. Top. Membr. Transp. 29: Blumenthal, R., A. Bali-Puri, A. Walter, D. Covell, and. Eidelman ph-dependent fusion of vesicular stomatitis virus with Vero cells. Measurement by dequenching of octadecyl rhodamine fluorescence. J. Biol. Chem. 262: Boulter, E. A., and G. Appleyard Differences between extracellular and intracellular forms of poxvirus and their implications. Prog. Med. Virol. 16: Butcher, M., K. Raviprakash, and H. P. Ghosh Acid ph-induced fusion of cells by herpes simplex virus glycoproteins gb and gd. J. Biol. Chem. 265: Chang, A., and D. H. Metz Further investigations on the mode of entry of vaccinia virus into cells. J. Gen. Virol. 32: Clague, M. J., C. Schoch, L. Zech, and R. Blumenthal Gating kinetics of ph-activated membrane fusion of vesicular stomatitis virus with cells: stopped-flow measurements by dequenching of octadecylrhodamine fluorescence. Biochemistry 29: Dales, S., and R. Kajioka The cycle of multiplication of vaccinia virus in Earle's strain L cells. Virology 24: Doms, R. W., M.-J. Gething, J. Henneberry, J. White, and A. Helenius Analysis of a variant influenza hemagglutinin that induces fusion at elevated ph. J. Virol. 57: Florkiewicz, R. Z., and J. K. Rose A cell line expressing vesicular stomatitis virus glycoprotein fuses at low ph. Science 225: Hiller, G., and K. Weber Golgi-derived membranes that contain an acylated viral polypeptide are used for vaccinia virus envelopment. J. Virol. 55: Hoekstra, D., T. D. Boer, K. Klappe, and J. Wilschut Fluorescence method for measuring the kinetics of fusion between biological membranes. Biochemistry 23: Hoekstra, D., K. Klappe, T. D. Boer, and J. Wilschut Characterization of the fusogenic properties of Sendai virus: kinetics of fusion with erythrocyte membranes. Biochemistry 24: Hoekstra, D., and J. W. Kok Entry mechanisms of enveloped animal viruses. Implications for fusion of intracellular membranes. Biosci. Rep. 9: Ichihashi, Y., and S. Dales Biogenesis of poxviruses: interrelationship between hemagglutinin production and polykaryocytosis. Virology 46: VACCINIA VIRUS FUSION ACTIVITY Ichihashi, Y., S. Matsumoto, and S. Dales Biogenesis of poxviruses: role of A-type inclusions and host cell membranes in virus dissemination. Virology 46: Ichihashi, Y., and M. Oie Adsorption and penetration of the trypsinized vaccinia virion. Virology 11: Janeczko, R. A., J. F. Rodriguez, and M. Esteban Studies on the mechanism of entry of vaccinia virus in animal cells. Arch. Virol. 92: Kohno, K., J. Sambrook, and M.-J. Gething Effect of lysosomotropic agents on the entry of vaccinia virus into CV-1 cells. J. Cell. Biochem. Suppl. 12: Konder-Koch, C., B. Burke, and H. Garoff Expression of Semliki Forest virus proteins from cloned complementary DNA. I. The fusion activity of the spike glycoprotein. J. Cell Biol. 97: Loyter, A., V. Citovsky, and R. Blumenthal The use of fluorescence dequenching measurements to follow viral membrane fusion events. Methods Biochem. Anal. 33: Maa, J.-S., J. F. Rodriguez, and M. Esteban Structural and functional characterization of a cell surface binding protein of vaccinia virus. J. Biol. Chem. 265: Marsh, M., and A. Helenius Virus entry into animal cells. Adv. Virus Res. 36: Morris, S. J., D. P. Sarkar, J. M. White, and R. Blumenthal Kinetics of ph-dependent fusion between 3T3 fibroblasts expressing influenza hemagglutinin and red blood cells. J. Biol. Chem. 264: Moss, B Replication of poxviruses, p In B. N. Fields, D. M. Knipe, R. M. Chanock, M. S. Hirsch, J. L. Melnick, T. P. Monath, and B. Roizman (ed.), Virology, 2nd ed. Raven Press, New York. 27. Nussbaum, O., and A. Loyter Quantitative determination of virus-membrane fusion events. Fusion of influenza virions with plasma membranes and membranes of endocytic vesicles in living cultured cells. FEBS Lett. 221: Oie, M., and Y. Ichihashi Characterization of vaccinia polypeptides. Virology 167: Paterson, R. G., S. W. Hiebert, and R. A. Lamb Expression at the cell surface of biologically active fusion and hemagglutinin/neuraminidase proteins of the paramyxovirus simian virus 5 from cloned cdna. Proc. Natl. Acad. Sci. USA 82: Payne, L. G Polypeptide composition of extracellular enveloped vaccinia virus. J. Virol. 27: Payne, L. G Identification of the vaccinia hemagglutinin polypeptide from a cell system yielding large amounts of extracellular enveloped virus. J. Virol. 31: Payne, L. G Significance of extracellular enveloped virus in the in vitro and in vivo dissemination of vaccinia. J. Gen. Virol. 5: Payne, L. G., and K. Kristensson Extracellular release of enveloped vaccinia virus from mouse nasal epithelial cells in vivo. J. Gen. Virol. 66: Payne, L. G., and E. Norrby Presence of haemagglutinin in the envelope of extracellular vaccinia virus particles. J. Gen. Virol. 32: Payne, L. G., and E. Norrby Adsorption and penetration of enveloped and naked vaccinia virus particles. J. Virol. 27: Puri, A., J. Winick, R. J. Lowy, D. Covell,. Eidelman, A. Walter, and R. Blumenthal Activation of vesicular stomatitis virus fusion with cells by pretreatment at low ph. J. Biol. Chem. 263: Riedel, H., C. Konder-Koch, and H. Garoff Cell surface expression of fusogenic vesicular stomatitis virus glycoprotein from cloned cdna. EMBO J. 3: Rodriguez, J. F., and M. Esteban Mapping and nucleotide sequence of the vaccinia gene that encodes a 14-kilodalton fusion protein. J. Virol. 61: Rodriguez, J. F., R. Janeczko, and M. Esteban Isolation and characterization of neutralizing monoclonal antibodies to vaccinia virus. J. Virol. 56: Rodriguez, J. F., E. Paez, and M. Esteban A 14,-Mr Downloaded from on August 21, 218 by guest

9 4892 DOMS ET AL. envelope protein of vaccinia virus is involved in cell fusion and forms covalently linked trimers. J. Virol. 61: Sarkar, D. P., S. J. Morris,. Eidelman, J. Zimmerberg, and R. Blumenthal Initial stages of influenza hemagglutinininduced cell fusion monitored simultaneously by two fluorescent events: cytoplasmic continuity and lipid mixing. J. Cell Biol. 19: Seki, M., M. Oie, Y. Ichihashi, and H. Shida Hemadsorption and fusion inhibition activities of hemagglutinin analyzed by vaccinia virus mutants. Virology 175: Shida, H Nucleotide sequence of the vaccinia virus hemagglutinin gene. Virology 15: Stegmann, T., F. P. Booy, and J. Wilschut Effects of low ph on influenza virus. Activation and inactivation of the membrane fusion capacity of the hemagglutinin. J. Biol. Chem. 262: Stegmann, T., R. W. Doms, and A. Helenius Proteinmediated membrane fusion. Annu. Rev. Biophys. Biophys. J. VIROL. Chem. 18: Stegmann, T., H. W. M. Morselt, J. Scholma, and J. Wilschut Fusion of influenza virus in an intracellular acidic compartment measured by fluorescence dequenching. Biochim. Biophys. Acta 94: Stern, W., and S. Dales Biogenesis of vaccinia: relationship of the envelope to virus assembly. Virology 75: White, J., A. Helenius, and M.-J. Gething Haemagglutinin of influenza virus expressed from a cloned gene promotes membrane fusion. Nature (London) 3: White, J., J. Kartenbeck, and A. Helenius Membrane fusion activity of influenza virus. EMBO J. 1: White, J., K. Matlin, and A. Helenius Cell fusion by Semliki Forest, influenza and vesicular stomatitis viruses. J. Cell Biol. 89: White, J. M Viral and cellular membrane fusion proteins. Annu. Rev. Physiol. 52: Downloaded from on August 21, 218 by guest