|Home | About | Journals | Submit | Contact Us | Français|
Experiments in the 1960s showed that Sendai virus, a paramyxovirus, fused its membrane with the host plasma membrane. After membrane fusion, the virus spontaneously “uncoated” with diffusion of the viral membrane proteins into the host plasma membrane and a merging of the host and viral membranes. This led to deposit of the viral ribonucleoprotein (RNP) and interior proteins in the cell cytoplasm. Later work showed that the common procedure then used to grow Sendai virus produced damaged, pleomorphic virions. Virions, which were grown under conditions that were not damaging, made a connecting structure between virus and cell at the region where the fusion occurred. The virus did not release its membrane proteins into the host membrane. The viral RNP was seen in the connecting structure in some cases. Uncoating of intact Sendai virus proceeds differently from uncoating described by the current standard model developed long ago with damaged virus. A model of intact paramyxovirus uncoating is presented and compared to what is known about the uncoating of other viruses.
Enveloped virus entry at the plasma membrane includes binding of the virion to one or more receptors, changes in the virion components, membrane fusion, and membrane uncoating. The term “membrane uncoating” is being used to describe the separation of internal virion components from the viral membrane so the internal components can enter the cell. The term “uncoating” is sometimes used to mean the release of the viral genome from the capsid or other structures that have also entered the cell, but in this review, the term “membrane uncoating” will be used to represent only the separation of the virion internal contents and the viral envelope.
Much of the original model of membrane fusion and uncoating was generally accepted as a result of a 1968 paper by Morgan and Howe (41). That paper provided strong evidence that Sendai virus (a paramyxovirus) entered a cell by fusion of the viral membrane with the cell plasma membrane. After membrane fusion, the virion rapidly lost its structure as the viral membrane merged with the host membrane and its components became part of the host membrane. The viral ribonucleoprotein (RNP) and internal proteins were released into the cytoplasm. This model of membrane uncoating is still generally accepted. For instance, in a 2007 virology text (24), this model was presented and illustrated with a figure from the Morgan and Howe paper. (The same figure is shown here as Fig. 2B.)
Later, it was shown that Sendai viruses, which had been grown in fertilized chicken eggs, had different properties depending whether they had been harvested after growth for roughly 1 day (“early harvest”) or for several days (“late harvest”). The early-harvest viruses appear to be intact, but the late-harvest viruses have a different morphology and appear to be damaged (20, 26).
This review summarizes data showing that intact early-harvest Sendai viruses uncoat quite differently from the way damaged late-harvest Sendai viruses uncoat. A model of intact paramyxovirus membrane uncoating is presented. The membrane uncoating of some other enveloped viruses that enter at the plasma membrane is compared to that described by this model.
Paramyxoviruses were used for many of the early membrane fusion experiments. The paramyxoviruses being discussed have two glycoproteins that appear as spikes on the virus. One glycoprotein is the HN protein, which binds receptors that contain N-acetylneuraminic acid and which has neuraminidase activity. The second glycoprotein is the F (fusion) protein, which is required for membrane fusion to occur. The M protein, which is thought to play a major role in virus assembly, is apposed to the inner side of the viral membrane. The M protein is small (Mr, ~38,500 to 41,500), basic, and somewhat hydrophobic but does not have a hydrophobic region long enough to go across a lipid bilayer. It is probably important in membrane uncoating. Paramyxoviruses have single-stranded RNA of negative polarity, which associates with the nucleocapsid (N) protein to form a helical ribonucleoprotein (RNP). The L and P proteins, which together form RNA-dependent RNA transcriptase, are associated with the RNP (34).
Until the end of the 1970s, studies of viral membrane fusion and uncoating often used paramyxoviruses that were grown for several days on the chorioallantoic membranes of chicken eggs with 9- to 11-day-old embryos. The chorioallantoic membrane is a highly vascular, fused layer of chorion and allantois and can be dissected from the egg. The chorioallantoic membrane cells produce infectious viruses with the F0 protein cleaved into F1 and F2 proteins. This cleavage is necessary before the F protein is active and the virus can fuse its membrane.
Many of the paramyxoviruses cause hemagglutination by binding red cells. Hemagglutination may be followed by hemolysis, which can be easily measured by recording the optical density at 575 nanometers (OD575) to measure hemoglobin release. As discussed below, hemolysis requires membrane fusion, but membrane fusion does not necessarily result in hemolysis.
Often, Sendai virus (SeV) was the paramyxovirus used. It is a mouse parainfluenza virus type 1 and is also is referred to as hemagglutinating virus of Japan (HVJ). It takes between 14 and 20 h at 36°C for one round of replication on chorioallantoic membrane cells in embryonated chicken eggs. Sendai viruses grown for 24 h are called “early-harvest” viruses. If the viruses are not harvested after 1 day in the eggs, the progeny viruses infect additional chorioallantoic cells, and the number of viruses produced increases considerably. Therefore, viruses grown in eggs for 2 to 3 days were commonly used in early times. The viruses grown 48 h or longer are termed “late-harvest” viruses (20). Early-harvest virus preparations cause a low level of hemolysis, and some do not cause hemolysis at all. As the time of viral growth in eggs increases, the amount of hemolysis produced by the virus increases. In a late-harvest virus preparation, there are always viruses that were released just before harvest, and so there are always some viruses that have the attributes of early-harvest virus. In early-harvest preparations, there are usually some viruses that are damaged and behave like late-harvest viruses. Thus, a virus preparation is usually not purely one or the other kind of virus, but it is possible to make preparations that contain predominantly early-harvest (intact) or late-harvest (damaged) viruses.
There are considerable morphological differences between early-harvest and late-harvest Sendai viruses (HVJ) (26). These differences are shown in Fig. Fig.1.1. Negatively stained early-harvest viruses (Fig. (Fig.1B)1B) have a relatively homogeneous size and are oval shaped. Thin sections of early-harvest virus (Fig. (Fig.1C)1C) show that the RNP is regularly folded parallel to the long axis of the virions. If the virus is sectioned at right angles to the long axis (arrow), the strands of RNP are seen in cross-section and are distributed in an orderly fashion at about equal distances from each other and mainly parallel to the virus axis. A few damaged viruses are also seen in this section. The late-harvest viruses (Fig. (Fig.1A)1A) are pleomorphic, with many spherical viruses of various sizes. Many late-harvest viruses have RNP strands that are randomly distributed in the virus rather than regularly arranged in relation to the membrane. The viruses within the late-harvest preparation that have an early-harvest or intermediate virus morphology are indicated by the arrows. The internal volumes of the larger late-harvest virus are multiples of the internal volume of early-harvest virus. Therefore, it is not surprising that late-harvest viruses have lost considerable internal organization.
Freeze-fracture electron microscopy has been used for studying membrane structures (4, 45, 50). The sample is rapidly frozen, fractured, and then shadowed. Intrinsic membrane proteins are seen in fractured membranes as intramembranous particles (IMPs) in the smooth lipid part of the membrane. The exoplasmic face (E face or EF) represents the hydrophobic side of the outer monolayer. The protoplasmic face (P face or PF) represents the hydrophobic side of the inner monolayer. Late-harvest viruses have IMPs with a mode diameter of 150 Å in the E face, whereas early-harvest viruses rarely show IMPs (26).
If early-harvest viruses are purified and concentrated and then incubated for 2 days at 37°C in allantoic fluid, they develop IMPs and an appearance consistent with that of late-harvest viruses. The same is true if purified early-harvest viruses are left for 2 days at 4°C with balanced salt solution, pH 7.6. LLC-MK2 cells begin to release F0 HVJ (Sendai virus) at 10 h, and the viral number increases until 3 days of incubation. HVJ grown for 3 days in LLC-MK2 cells have the electron micrograph appearance of late-harvest virus, and freeze-fracture electron microscopy shows IMPs in the E face. Thus, the viruses seem to “age” either during a long period with cells or during a long time in different media without cells. At present, it is not known which membrane protein(s) is altered during aging to result in late-harvest virus. The authors of a study of early-harvest and late-harvest viruses suggested that the M protein could be involved (26). In the virion, the M protein is associated with the nucleocapsid and with the glycoprotein spikes and is important in viral assembly (34, 44), and so involvement of the M protein in changes causing late-harvest virus properties seems a good possibility. Noncovalent intermolecular interactions in Sendai virus and Newcastle disease virus were studied with reversible cross-linkers that had a 1.1-nm separation between their functional groups. Nucleoprotein and M protein heterodimers resulting from cross-linking were most prevalent in early-harvest viruses (37). Not all viruses are as fragile as Sendai virus. However, since Sendai viruses were used for many of the early studies of viral entry, their fragility is an important reason for ideas about membrane uncoating.
If early-harvest viruses are treated with sonication or with repeated freeze-thaw cycles, the amount of hemolysis they produce increases (20). Measurements of infectivity and hemolysis show that the infectivity goes down after each freeze-thaw cycle as the amount of hemolysis a virus causes increases. Late-harvest viruses produce considerable hemolysis. The sensitivity to chemical or physical manipulations, such as freezing and thawing, seems to exist to some degree for many kinds of viruses. This sensitivity to laboratory proceedings is sometimes also reflected in viral properties after different purification procedures. Experiments with vesicular stomatitis virus and preliminary experiments with simian virus 5 (SV5; parainfluenza virus 5) and influenza virus indicated that the viruses appropriately swelled or shrank with the change in the osmolarity of the external media (3). Below are discussed the data that show that damaged Sendai virus uncoats in a process that is different from that used by intact virus.
The 1968 study by Morgan and Howe (41) included excellent electron micrographs that showed fusion of Sendai virus membranes with the membranes of chorioallantoic cells and of L cells. The continuity of the viral and host membranes was unequivocally clear, so that fusion between Sendai virus membrane and host cell plasma membrane was generally accepted as the mechanism of Sendai virus entry. Their virus was grown in chorioallantoic cavities of 11-day-old chicken embryos for an unstated amount of time. The authors comment on “pleomorphism” of the viruses, and the viruses have the appearance of late-harvest viruses.
Figure Figure22 A shows a virus that has been incubated with cells at 37°C for 6 min and has fused its membrane with the host plasma membrane. When Sendai virus binds many cell receptors, the cell membrane becomes indented to surround the bound part of the virus. At the edge of the region of binding, the host membrane returns to its normal curvature and in doing so produces a rim around the indentation where the membrane is very curved. Many of the micrographs in the Morgan and Howe study (41) show viruses that had fused their membranes with host membranes in the vicinity of the very curved membrane edge at the border of a region of binding.
Figure Figure2B2B shows a fused virus and cell with the virus structure widening as the viral membrane merges with the host membrane. Much of the viral contents have mixed with the cytoplasm below the region of fusion. In some micrographs, the contents of the virus and of cell are almost completely mixed. This supported the idea that after fusion of viral and host membranes, there is extensive mixing of the viral and host components.
Even in late-harvest preparations, there are a few early-harvest viruses. Morgan and Howe presented a micrograph wherein the viruses are small and oval and appear to be early-harvest viruses (Fig. (Fig.2C).2C). The ribonucleoprotein (RNP) is very organized and aligned with the long axis of the virus. The virus on the right is bound but not fused with the host and appears to be cut at right angles to the virus axis. The virus on the left in this micrograph has fused with the cell membrane. A connecting structure is present, and the RNP has moved a short distance down the connecting structure into the cell, yet the rest of the virion is intact. This was not commented upon, but at the time, the idea of early-harvest and late-harvest viruses had not yet been proposed. This micrograph raises the question of whether for intact viruses the RNP moves through a connecting structure (connector) without dismantling the viral coat.
Colloid osmotic swelling of red cells produces hemolysis. Red cells have hemoglobin and do not have the membrane structures, such as microvilli, which allow most cells to expand considerably. Late-harvest Sendai viruses produce a change in the plasma membrane permeability that leads to the osmotic swelling. The increase in permeability is thought to be due to the presence of the virus glycoproteins (probably the F protein) in the red cell membrane. Zn2+ and Ca2+ inhibit the changes in permeability. The damage to membranes by hemolytic viruses has been compared to the damage performed by toxins, such as Staphylococcus aureus alpha-toxin, melittin, complement, and other cytotoxic agents (2). Virus-induced hemolysis requires both membrane fusion and subsequent changes in membrane permeability.
Changes in membrane permeability also occur in host cells at the beginning of infection with late-harvest viruses, but the host cells do not lyse. The potential across the membrane of FL cells (a line from human amnion) was determined before and after infection with HVJ (Sendai virus) (42). After incubation with what appears to be late-harvest virus at 37°C, the FL cell membrane potential dropped. The membrane potential was repaired after membrane fusion was completed. The increase of membrane permeability of mouse-grown Lettrée cells seen with late-harvest Sendai virus does not occur after infection with early-harvest virus (54).
Knutton studied fusion of nonhemolytic Sendai viruses (grown 24 h or early harvest) with erythrocytes (30). Scanning microscopy showed that by 2 to 3 min at 37°C, nonhemolytic viruses had fused with the red cell. Many of the nonhemolytic fused viruses do not become completely incorporated into the red cells even after 20 to 30 min of incubation. Hemolytic viruses (grown 72 h or late harvest), however, merge into red cells and cause a change in red cell permeability, which leads to cell swelling and visible red cell-red cell fusion.
Knutton and Bächi used nonhemolytic viruses and hemolytic viruses to investigate the movement of Sendai viral antigens in red cell membranes after membrane fusion (31). All their batches of “nonhemolytic” virus showed some hemolysis. Unless stated otherwise, the virus they used had a hemolytic figure of 5% or less. They used monolayers of red cells. They detected the Sendai glycoproteins with fluorescein isothiocyanate (FITC) anti-Sendai virus antibody (or with FITC anti-Sendai virus antibody Fab fragments, which gave similar results). When hemolytic or nonhemolytic viruses were added to red cells at 4°C, the viruses did not fuse because of the temperature, and the immunofluorescence of viral antigens was granular. After 10 min at 37°C, the immunofluorescence of the hemolytic viruses became diffused in red cells, and after 1 to 2 h, all red cells showed uniform fluorescence. When nonhemolytic viruses were added to red cells, the immunofluorescence maintained the grainy pattern initially present at 4°C even after 1 h at 37°C. Therefore, the fused nonhemolytic viruses did not release their membrane glycoproteins from the fused virion membranes into the cell membranes.
Sendai viruses, strain ESW5, were incubated in the chorioallantoic sac of eggs for about 68 h, so the virus used was late harvest. The liposome composition was roughly similar to the lipids in BHK-21-F cell plasma membranes (28), and gangliosides, which contain neuraminic acid, were included to act as receptors (12). At 37°C, Sendai virus can make a ganglioside-containing liposome envelop it as in phagocytosis (although without vacuole closure) (16) and can fuse with liposomes (14). After incubation of viruses and liposomes for 2 h at 37°C, the viral RNP was deposited inside the outer layer of a multilamellar vesicle (MLV), and the viral membrane glycoproteins spread over the surface of the liposome in patches of spikes. This was consistent with results of the Morgan and Howe study of Sendai virus entry into cells. Viruses that have been incubated at 37°C with liposomes for only 1 min show fusion only at the very curved leading edge of the region where the liposome binds the virus (17). This would be the rim of the envelopment if the region were viewed in three dimensions (3D). Curvature facilitates the close approach of surfaces (13, 52). In a curved bilayer, the heads of the phospholipids of the outer layer of the bilayer are pulled away from each other, facilitating membrane fusion. Further, the membrane tension of the unbound membrane is on one side of the curvature and virus binding is on the other side, so the membrane between the two regions is probably mechanically stressed. With continued time of incubation after membrane fusion, the remaining binding between the fused virus and liposome decreases until the only remaining contact between virus and liposome is the fusion region. Since viral binding alters the properties of a region of the host membrane, it is possible that a major role of the viral F protein is to make changes in the properties of a region of the viral membrane.
Not all viruses merged into liposomes after fusion of the viral and liposomal membranes. Some fused viruses were in the process of what Morgan and Howe (41) described as uncoating. However, there were some cases in which the viruses had fused but the viral membrane glycoproteins had not diffused into the liposomes and the virion had not merged into the liposomes (17). This indicates that membrane uncoating does not necessarily occur immediately after membrane fusion and suggests the possibility of two populations of virus. Preliminary experiments with early-harvest viruses suggested that after membrane fusion, these viruses do not merge into liposomes (15), but further experiments are needed.
Figure Figure33 shows a model of entry of intact (early-harvest) paramyxovirus. The drawing gives a roughly proportional representation of the sizes of the components. Many components are not drawn, because their distributions during uncoating are unknown. This includes the M protein, which in the mature virus is associated with the inside of the membrane, and the transcriptase, which is associated with the RNP. It also includes some host proteins included in the virion, such as actin (9) and a cellular protein kinase (21).
Figure Figure3A3A represents an early-harvest virus bound to a cell. The virus is oval rather than round, and the RNP is distributed parallel to the membrane. In a two-dimensional (2D) longitudinal view, some of the RNP appears to be in the middle of the virus. However, the viral section at right angles to the viral long axis, as shown in Fig. Fig.1C,1C, indicates that the RNP that in Fig. Fig.3A3A appears to be in the middle is really running parallel and close to the membrane in the front and back of the virion. So in a 3D drawing, the RNP would be seen to be equally close to the membrane throughout the virus.
Figure Figure3B3B indicates that in the period just around the time of fusion, a connector between the virus and cell is made and there is preparation for moving the RNP. The spikes are drawn differently because the viral glycoproteins undergo a conformational change during the fusion process (36, 56), and some of the early electron microscopy studies reported changes or loss of defined spikes during fusion (27, 29).
The early-harvest membrane glycoproteins do not move into the cell membrane as the late-harvest glycoproteins do. In the early-harvest virus, the interactions between membrane proteins presumably mean the proteins are in complexes too big to diffuse in the membrane. The lipid mobility in the membrane of the early-harvest virus would not be expected to be so constrained after fusion, since lipids are much smaller than proteins and they would not have tight associations. However, the movement of labeled lipids from early-harvest virus membrane into the host membrane has not been tested.
The connector is roughly 250 to 300 Å long. The viral glycoprotein spikes are 120 Å long, so the connector is over twice as long as the spikes on a free virus. The length could be due to rearrangement of viral components but seems more likely to reflect participation of both viral and host components. There are many possibilities for these host components, for instance, cytoskeletal components. A connector is also sometimes seen at early times after fusion with late-harvest virus. It seems possible that small host internal components, e.g., ATP, could move into the virus though the connector and alter the proteins inside the virion.
Figure Figure3C3C shows that even as the RNP starts to move out of the virus, the early-harvest virus membrane continues to keep its general shape and the glycoproteins do not diffuse into the host cell membrane. In the free early-harvest virus, the RNP probably has most or all of its length associated with the membrane through the M protein. The conformational changes of F and HN proteins after receptor binding may be involved in the release of the nucleocapsid from its association with the membrane and M protein. Activation of the viral internal contents may also help release the RNP.
Figure Figure3D3D is based mostly on extrapolation of data or on the data for other viruses. The size of the RNP illustrates why diffusion is unlikely to bring the RNP to necessary host cell components. When the original ideas of uncoating developed in the 1960s, it was not recognized that the late-harvest virus had been damaged, and there was no idea of cell trafficking. Since then, it has been found that diffusion of large macromolecules in the cytoplasm is extremely impaired (51). Thus, in the 1960s, it was assumed that release of the RNP and associated internal proteins into the cytoplasm beneath wherever on the cell the virus happened to bind would be followed by diffusion of the RNP to the host components needed for replication. As shown in Fig. Fig.3,3, the passage of the early-harvest RNP through the connector would allow the presentation of the viral genome to be organized. How the RNP starts to have its end moved into the connector is not known. Once the end of the RNP enters the cell, it probably associates with cellular structures, and they may participate in drawing it into the cell. In Fig. 3C and D, the gray and beige structures represent unspecified cell cytoplasmic components that are involved in the RNP trafficking and replication. The positioning of the viral contents for trafficking is probably part of uncoating.
Late-harvest HVJ (Sendai virus) membrane glycoproteins that had diffused into the membranes of Ehrlich ascites tumor (EAT) cells were identified by ferritin-conjugated anti-F and anti-HN antibodies but were not seen as spikes. The viral glycoproteins found in the EAT cell membrane after fusion with late-harvest HVJ were later taken up by coated vesicles in cells (27). There are no data on what happens to the intact early-harvest virus membrane after Sendai viral contents have entered the cell. The empty viral membrane may be degraded, or there could be release of the empty membrane and its fragments into the environment, as shown for Sindbis virus (see below).
The question of whether the same general model of membrane uncoating applies to other viruses, with details and mechanisms that vary from virus to virus, then arises. Published micrographs showing entry of enveloped viruses usually came from experiments designed to investigate membrane fusion. Rarely did anyone do the sort of experiment needed to focus on membrane uncoating.
Methods in light and electron microscopy now available (10, 48) hold promise for more understanding of membrane uncoating. When cryomicroscopy is used, the viral structure is not disturbed by the fixatives and stains which are required for transmission microscopy. For viruses with high levels of symmetry, it is possible to do single-particle imaging by averaging data from many similar particles (8). For pleomorphic viruses with variable shapes and sizes and for complexes with viruses or viral components, electron tomography may be applicable. For electron tomography, the sample is rotated over a range of tilts, and the 2D images so obtained are appropriately weighted and combined computationally so that they make a 3D image (40). Electron tomography is used with samples that have undergone ultrarapid freezing or after the sample has been fixed, stained, and sectioned. There is a limit to the thickness of the sample (0.5 to 1 μm) that can be used for electron tomography.
Below are discussed two viruses that have structures that are very different from that of paramyxoviruses. One common theme is that when a virus binds to its receptor, major changes in the virion usually result, and these changes seem to be a necessary part of membrane fusion and uncoating.
The RNA is a positive strand and is surrounded by 240 units of capsid protein, which form an icosahedron. A membrane surrounds the capsid. The envelope glycoproteins, E1 and E2, form heterodimers that link together into an icosahedral lattice on the outside of the membrane, with E2 penetrating the membrane to link with the capsid proteins. Thus, the membrane is between two protein layers and is not easily available unless the E1 and E2 proteins undergo considerable changes during entry (6). As is the case with the paramyxoviruses (22, 23), a thiol-disulfide exchange reaction is necessary for Sindbis virus infection (1). The exchange reaction used by Sindbis virus requires a neutral to alkaline pH.
Despite the considerable differences between the structures of alphaviruses and of paramyxovirus, there are similarities in the sorts of damage that that allow the viruses to cause hemolysis and to cause changes in infected host cell permeability.
Alphaviruses do not normally bind red cells at a neutral pH, but at pH 5.8, they do bind and cause hemagglutination (49). “Fresh virus” is virus grown in BHK-21 cells, centrifuged at 10,000 × g for 30 min to remove debris, and then used directly with no further purification. At pH 5.8, fresh virus did not cause hemolysis. If, however, the fresh Semliki Forest viruses were frozen and thawed 10 times, they produced hemolysis even though the amount of hemagglutination did not change. However, Sindbis virus was less fragile, since this treatment resulted in only a small amount of hemolysis, while the hemagglutination remained the same. Some of the Sindbis viruses and the Semliki Forest viruses were purified either by polyethylene glycol precipitation or by vacuum dialysis and banding on a sucrose gradient. The viruses that had dialysis and sucrose gradient purification had a higher hemolytic capacity per ng than did the viruses purified with only precipitation. So both freeze-thawing and purification procedures can alter viruses enough to change their capacity to cause hemolysis.
Such cells have nonspecific pores that can passage ions and molecules of weight up to 900 Da (35). This change in permeability does not occur in the presence of Zn2+, Ca2+, or La3+. Inhibiting permeability changes with Zn2+ did not inhibit cell-cell fusion. Koschinski at al. (32) showed that at low pH, permeability appears in target membranes during viral entry, in the plasma membrane of infected cells during viral multiplication, and in isolated virus particles. La3+ prevents the permeability from appearing in all these situations. When Semliki Forest viruses were added to cells at low pH, La3+ did not inhibit the formation of viral plaques or viral yield. It was proposed that at low pH, the E1 molecule damages the membrane and might act like the membrane-damaging protein toxins. An increase in the conductance of a planar lipid membrane was noted upon the addition of freeze-thawed or sonicated Sendai virus or upon the addition of Semliki virus at pH 5.2 (55). It appears that when alphaviruses or paramyxoviruses are damaged and the viral membrane proteins have changed conformation, the proteins can cause a change in permeability. It is interesting to speculate about whether such altered proteins can cause pathology.
In a paper that addressed mainly the effects of pH changes on virus structure, a final section showed transmission electron micrographs of Sindbis viral entry at neutral pH (43). Sindbis virus strain SVHR produces virus with a low particle/PFU ratio, usually between 1 and 10 (18). For the studies on Sindbis virus entry at pH 7.2 (43), including the micrographs shown in Fig. Fig.4,4, the SVHR strain of Sindbis virus was used, and the particle/PFU ratio was 1 (D. T. Brown, personal communication). Sindbis viruses were adsorbed to BHK cells at 0°C, and the temperature was then raised to 37°C, all at pH 7.2. The sample was fixed very shortly after transfer from 0°C to 37°C, and the fixed virus-cell complexes were labeled with virus-specific antiserum and 60-Å gold bead-conjugated goat anti-rabbit antibodies. This allowed not only intact viruses but also virus membranes that had been emptied of their contents to be recognized. Figure Figure4A4A shows a virus that is attached to a cell. After attachment, a connector between the virus and cell (which the authors referred to as a pore) is formed (Fig. (Fig.4B).4B). The dense material is still within the virus. Figure Figure4C4C also shows the continuity of the virus and host membranes well (bottom arrow). The electron density (top arrow) is reduced in the virus, which could indicate changes in the viral core or could be secondary to microscopic procedure. In Fig. Fig.4D,4D, the electron-dense region is changing shape, and a part of it approaches the connector. In Fig. Fig.4E,4E, an electron-dense region is no longer visible, but there appears to be stranded material (arrow) that goes from the virus into the cell. In Fig. Fig.4F,4F, there is a structure that has no electron-dense material and is released from the cell. It is identified as viral by antibodies conjugated with gold beads. This release of an empty viral membrane has not been noted before, but the use of labeled antibodies meant such a structure would be revealed. If the envelope membrane disengages from the cell instead of merging with the host membrane, then not only would the cell not have viral proteins on its surface until the virus replicates but the released membrane pieces could serve as immunologic decoys.
The connector between the virus and cell had an average width of around 60 Å when the virus still contained the electron density and an average width of around 126 Å when the electron density was gone. The length of the connection is about 94 Å and did not change with the disappearance of the electron-dense structure from the virion. Of the attached viruses, 26% appeared empty after being added to cells at 37°C, whereas only 2 to 3% appear empty when they were left with cells at 4°C. The micrographs of Sindbis viral entry in many ways fit the model presented above for intact paramyxovirus membrane uncoating. A connector has been formed, and there are indications that the RNA moves through this connector into the cell. The viral membranes remain on the outside of the cell. In addition, this study adds the idea that emptied viral membranes are released from the cell after the passage of their internal contents into the cell.
The authors of this paper (43) suggested that classical membrane fusion did not occur but that entry was similar to that seen for poliovirus (19), which has no membrane. The E1 and E2 protein icosahedron cover the alphavirus membrane so it unavailable for contact with the host cell membrane, but changes of the E1 and E2 proteins after binding of the virion may solve that problem. At present, there are questions about the exact identity of the alphavirus receptors.
Another controversial point is whether alphaviruses can infect by entering at the plasma membrane or if they can enter only by endocytosis and fusion out of the endocytic vacuole at low pH. The latter is generally the favorite model (33). After low-pH treatment, alphavirus membranes fuse with cell plasma membranes or with liposomes (which lack receptors) (5, 43, 53). That alphaviruses can enter the cell surface at neutral pH is supported by the fact that the micrographs show entry at the cell surface at pH 7.2. Further, 26% of the virions are emptied of their contents during incubation at 37°C. For a number of other viruses, there have been reports by some groups that entry can be from the cell surface at neutral pH and from other groups that entry of the same virus is from the endosome at low pH. It is possible that for some viruses, entry can be by both mechanisms.
Herpes simplex virus 1 (HSV-1) is a pleomorphic virus composed of many proteins (11). The membrane has about 12 glycoproteins, of which 5 are involved in viral entry. The virus has its choice of several virus receptors (47). Inside the membrane are the tegument and a capsid with a diameter of 125 nm that contains a 152-kbp double-stranded DNA. The tegument contains more than 20 proteins, which have a variety of roles (25). Some of the tegument proteins attach to some of the membrane glycoproteins, and some attach to the nucleocapsid. The tegument occupies about two-thirds of the virion interior but it does not totally surround the nucleocapsid. The nucleocapsid has an asymmetric position in the virion. On one pole of the virion, the nucleocapsid is close to the viral membrane, and on the other pole, the tegument separates the nucleocapsid from the membrane by about 30 to 35 nm. The spikes tend to be sparse over the virion pole, where the nucleocapsid is close to the membrane and to be more densely packed over the pole that has the tegument directly under it (11). Cryoelectron tomography studies of HSV-1 membrane fusion had led to the suggestion that the virion is functionally bipolar as well as structurally asymmetric (39). The complexity of the virion probably makes it necessary to alter many protein associations before uncoating is complete. The virus enters some cells by fusion of viral membrane with the host plasma membrane; in the present paper, only entry of the virus by viral and host plasma membrane fusion is discussed. The DNA viruses, with the exception of poxviruses, replicate their nucleic acid in the nucleus, so uncoating of HSV-1 releases a nucleocapsid, which must reach the nucleus. The capsid travels a good part of the distance to the nucleus by associating with microtubules and dynein (46). How the nucleocapsid reaches the microtubule is unknown and probably involves an interface that is a continuum with uncoating.
A problem in investigating virus membrane uncoating is obtaining virus preparations that have a low percentage of damaged or defective viruses. A low particle/PFU ratio is one measure of a good virus preparation. While care with the virus preparation can markedly lower the particle/PFU ratios, at present it is not possible to obtain exceedingly low ratios with many viruses. Damaged and defective viruses may well have difficulties in viral entry.
Döhner et al. (7) demonstrated that the quality of HSV-1 affects its success in entry and in reaching the nucleus. Their paper is directed at clarifying whether the small capsid protein VP26 is essential for efficient dynein-mediated capsid transport on microtubules. They made HSV1-ΔVP26 (HSV-1 that lacked VP26). Since they thought the infectivity of the mutant virus might be altered, they studied the protein/PFU, particle/PFU, and genome/PFU ratios of virions. They noted that the literature reported HSV-1 with particle/PFU ratios ranging from 5 to more than 1,000. Their best genome/PFU ratios were 15 to 25. If they removed the nonencapsidated DNA before calculating the ratios, the ratios were 7 to 10. They discuss many of the factors that affect the genome/PFU ratio, although they did not include sonication. Their preparation was based on the methods in reference 46 except that they used Nycodenz instead of sucrose gradients. In the preparation described in reference 46, after the initial pelleting of the virus, the virus was resuspended by water bath sonication.
Döhner et al. determined the cellular positions of the viral capsid and of gD, the membrane glycoprotein that binds receptors, by using fluorescent microscopy with capsid and gD antibodies and fluorescently labeled secondary antibodies. After 30 min of incubation at 37°C, the capsid and gD protein were randomly distributed over the cell. With the wild-type (wt) virus (KOS), most capsids did not colocalize with the gD protein, but with HSV1-ΔVP26, only about half of the capsids were separated from the gD protein. Thus, the wt either fused or uncoated more efficiently than did HSV1-ΔVP26. The authors suggested the particles with continued association of the capsid and gD were virions either still associated with the plasma membrane or inside endosomes.
Döhner et al. looked at the distribution of the capsid and of gD, after 3 h of incubation at 37°C, from a wt HSV-1 (KOS) with a genome/PFU ratio of 20, from a wt HSV-1 (F) with a genome/PFU ratio of 132, and from HSV1-ΔVP26 with a genome/PFU ratio of 48. The wt (KOS) virus with a genome/PFU ratio of 20 had a larger proportion of capsids that reached the nucleus, and only a little of the gD protein remained in the cell, whereas the wt (F) virus with a genome/PFU ratio of 132 had fewer capsids at the nucleus, and the amount of gD protein was very noticeable throughout the cell and at the periphery. The HSV1-ΔVP26 preparation with a genome/PFU ratio of 48 had less nuclear targeting than did the wt HSV-1 with a genome/PFU ratio of 20 and had a little more gD labeling remaining in the cell than did the wt HSV-1. A HSV1-ΔVP26 preparation with a ratio of 69 had more gD in the periphery than did the preparation with a ratio of 48. Movement of capsids from HSV1-ΔVP26 to the nucleus was found to be dependent on microtubules and dynein function. Döhner et al. concluded that since HSV-1 capsids could reach the nucleus in the absence of VP26, other HSV-1 proteins must be able to recruit dynein. They point out that if they had compared a preparation of HSV1-ΔVP26 with a high genome/PFU ratio to a wt HSV-1 with a low genome/PFU ratio, they could have reached the erroneous conclusion that VP26 is necessary for nuclear targeting. They have also succeeded in demonstrating that HSV-1 with a high genome/PFU ratio has defects in the entry process, although the nature of the defect is not clear.
Even the good preparations of wt HSV-1 have a considerable number of noninfective particles, which must be considered in an electron microscope study. Electron microscopy by Morgan and Howe of infection with late-harvest Sendai virus showed two different patterns of uncoating, representing damaged and intact virus (41). At the time, they did not expect two virus populations and so did not recognize the second pattern as being other than a variation of the general pattern. The presence of two different populations was noted in studies of late-harvest Sendai virus with liposomes (17). The data from immunofluorescent microscopy and electron microscopy should substantiate and complement each other. With many viruses, it will be impossible to get preparations with virus as intact as those of early-harvest Sendai virus and of Sindbis virus, so studies of membrane uncoating will have to be designed to account for the presence of both intact virus and damaged virus.
The goal of this review is to arouse enthusiasm for investigation of uncoating, which is more complex than previously thought. Further work is needed to confirm the model described in this review. Studies of various enveloped viruses that enter at the plasma membrane should show whether uncoating requires a series of steps that is specific for each virus.
The above data indicate that, at least for paramyxoviruses, membrane uncoating does not conform to the old model, which is very commonly assumed. Binding of virus to receptors leads to membrane fusion and formation of a connecting structure. The viral contents appear to move into the cell in what appears to be an organized process that probably involves virus and host participation. The altered membrane glycoproteins and probably the viral internal components participate in uncoating. Viral binding can also activate host pathways (38). A detailed description of the viral and probably host components that make the connecting structure is needed. How the nucleoprotein or core moves through the connecting structure into the cell should be investigated.
Damage to the virus is a major factor in altering the process of membrane uncoating. For many viruses, damage can result from conditions of growth, methods of purification, details of storage, or procedures such as freeze-thawing or sonication. The virus strain also can be important.
How difficult it will be to demonstrate the uncoating of the infective virus will depend upon the virus preparation quality. This can be estimated by using the particle/PFU ratio or the genome/PFU ratio. For some viruses, there are special tests for viral integrity, such as the amount of hemolysis the virus causes. If the particle/PFU ratio is not too high, uncoating of infective particles can be distinguished from the uncoating of noninfective particles by looking at enough samples to distinguish the different patterns of uncoating For viruses with a very high particle/PFU ratio, this might not be experimentally practical.
It is important to screen each virus to find the timing of the most interesting steps in viral entry. For membrane fusion of Sendai virus, Sindbis virus, and herpes simplex virus, many of the most interesting micrographs showed events that occurred in the first few minutes of incubation at 37°C.
How uncoating coordinates with cell trafficking of the viral contents should be investigated.
Further investigations are likely to reveal steps in membrane uncoating that have not been surmised, because uncoating has not been much considered. The details and mechanisms of membrane uncoating will probably vary from virus to virus.
Published ahead of print on 28 July 2010.