The Sendai virus-induced fusion of HeLa cells has been studied by freeze-fracture electron microscopy. Freeze-fracture observations confirm previous scanning electron-microscope studies (1977) and show that at 4 °C virus particles bind to the cell surface and that cell agglutination results from the crosslinking by virus particles of microvilli on adjacent cells. Incubation at 37 °C initiates a change in viral envelope structure and fusion of ‘altered’ virus particles with the cell plasma membrane. Fusion of a virus particle with two crosslinked cells is probably the membrane fusion event which initiates cell-cell fusion; fusion is completed as a result of virally induced cell swelling. Lateral diffusion of viral envelope components following viruscell fusion and, in some instances, an aggregation of plasma membrane intramembrane particles occurs in swollen cells. These observations show that the mechanisms of viral envelope-cell and probably cell-cell fusion are the same as have been reported for erythrocytes. Although endocytosis of intact virus particles does occur, the specialized cell-mediated mechanism for fusion of the viral envelope with the cell plasma membrane suggests that this, and not viropexis, is the mechanism of Sendai virus infection.
Numerous morphological studies have been directed towards elucidating the mechanism of virus-induced cell fusion (Okada, 1962; Schneeberger & Harris, 1966; Hosaka & Koshi, 1968; Howe & Morgan, 1969; Bächi & Howe, 1972; Toister & Loyter, 1973; Knutton, Jackson & Ford, 1977) but thin-section studies, although clearly illustrating fusion of virus particles with the cell plasma membrane (Howe & Morgan, 1969) and cell-cell fusion (Schneeberger & Harris, 1966), have provided little insight into the actual mechanism of virus-cell fusion or answered the question whether fused viral envelopes are directly or indirectly involved in the membrane fusion event leading to cell-cell fusion. As a technique for studying membrane fusion freeze fracture has several advantages over thin sectioning and freeze-fracture studies have provided a much greater insight into the structural changes occurring during virus-cell fusion (Knutton, 1976, 1977). These studies showed that prior to fusion with the cell membrane the viral envelope undergoes a cell-mediated change in structure resulting in the production of smooth particle-free linear invaginations; it is two of these linear segments of the viral envelope which first fuse and become incorporated into the cell membrane. Subsequently, probably as a result of virally induced cell swelling (Knutton, following paper, p. 85), the entire viral envelope becomes incorporated into the cell membrane. Furthermore, direct morphological evidence was presented in favour of a cell-virus-cell bridge mechanism of cell fusion (Apostolov & Almeida, 1972), rather than the alternative cell-cell bridge mechanism (Hosaka & Koshi, 1968; Poste, 1972); more recently further examples showing the fusion between a viral envelope and two adjacent cells have provided additional evidence that this and not direct membrane fusion, is the initial membrane fusion event in cell fusion (Knutton, 1978).
These studies, however, were carried out using the fusion of erythrocytes as a model system. Consequently, it is important to know if fusion with the cell plasma membrane is the mechanism of viral entry into other eukaryotic cells and if the mechanism of virally induced cell fusion is the same as that described for erythrocytes. In this paper I present freeze-fracture observations on the fusion of HeLa cells by Sendai virus.
HeLa cells were grown on plastic Petri dishes in minimal essential medium (Eagle, 1959) containing 10% newborn-calf serum. Exponentially growing cultures were removed from culture dishes with trypsin, washed and resuspended at a concentration of 6 × 106 cells/ml in Hanks’ balanced salt solution (Hanks’ BSS) (Hanks & Wallace, 1949). Sendai virus was grown in 10 or 11-day-old chick embryos, harvested after 72 h and clarified by centrifugation. Concentrated virus was stored under liquid nitrogen until used.
For fusion experiments the virus was u.v.-inactivated and diluted with Hanks’ BSS to a concentration of 2000 haemagglutinating units/ml (HAU/ml). One millilitre of virus was added to i ml of washed cells at 4 °C. After 30 min the agglutinated cells were transferred to a waterbath at 37 °C and samples taken after various time intervals up to 60 min. Fusion was stopped by either the addition of an equal volume of 6 % glutaraldehyde in Hanks’ BSS or by chilling to 4 °C.
For freeze fracture, cells fixed for 30 min with 3 % glutaraldehyde were washed and infiltrated with 25 % glycerol in Hanks’ BSS. Samples were rapidly frozen in freshly melted Freon 22 and fractured at— 110 °C in a Denton freeze-fracture device. Platinum carbon replicas were cleaned in 5 % sodium hypochlorite, washed in distilled water and examined in a Siemens tot electron microscope. In freeze-fracture illustrations the encircled arrowhead indicates the direction of shadow.
For thin sections cells were fixed for 1 h with 3 % glutaraldehyde in 0·1 M cacodylate buffer pH 7-4, postfixed 2 h with i % buffered osmium tetroxide and block stained overnight in 2 % aqueous uranyl acetate. Specimens were dehydrated through graded ethanol and propylene oxide solutions and embedded in Epon. Sections were cut with a diamond knife on a Tesla ultramicrotome and stained with uranyl acetate and lead prior to being examined.
At 4 °C, Sendai virus particles bind to the HeLa cell surface and at high cell and virus concentrations crosslinking of microvilli on adjacent cells by virus particles results in cell agglutination (Fig. 1). Because of the highly villated surface morphology of spherical suspension HeLa cells, fractures through agglutinated cells are complex. Fig. 1 shows the region of contact between two agglutinated cells both in oblique view (Fig. 1 A) and in cross fracture (Fig. 1 B). At the region of contact one sees, in addition to fractured virus particles, many round and elongated membrane profiles which represent cross- and obliquely-fractured microvilli. At low magnification (Fig. 1 A) it is difficult to distinguish between virus particles and microvilli but virus particles are readily distinguished at higher magnifications (Fig. 1 A, inset, and 1 B) by the presence of large ∼ 14-nm diameter intramembrane particles on E fracture faces (Fig. 1B, arrows) and large pits on P faces (Fig. 1 B, arrowheads). Viral envelope spikes can also be seen around the perimeter of many fractured virus particles (Fig. 1B, arrowheads). In contrast, HeLa cell plasma membrane intramembrane particles are only ∼8–9 nm in diameter. Following the interaction with virus at 4 °C both P and E HeLa cell plasma membrane fracture faces retain a random distribution of membrane particles typical of control cells.
Warming an agglutinated cell suspension to 37 °C results, after about 2 min, in a change in the morphology of most virus particles. Virus particles generate linear invaginations of the viral envelope which are characterized in freeze-fracture replicas by smooth linear ridges on E faces (Fig. 2 A,B) and complementary grooves on P faces. Changes also involve loss of the ∼ 14-nm E face intramembrane particles and the appearance of smaller ∼8-9-nm diameter particles on both P and E faces. Such ‘altered’ virus particles fuse with the plasma membrane. Fig. 2C, D show virus particles at early and late stages during fusion with the plasma membrane while Fig. 2E, F show face views of virus particles which have fused and become incorporated into the HeLa cell plasma membrane. The localized arrays of smooth E face linear ridges and P face grooves reveal the sites of viral envelope-cell fusion. The structural changes of the viral envelope and mechanism of virus-cell fusion are identical to those observed during the Sendai virus-induced fusion of human erythrocytes (Knutton, 1977) and will not be described in further detail here.
Studies with erythrocytes have strongly suggested that cell-cell fusion is initiated by fusion of a virus particle with 2 crosslinked cells since numerous examples of cell-viral envelope cell fusion have now been observed (Knutton, 1977, 1978). Consequently, examples of cell-virus-cell fusion have been searched for during the fusion of HeLa cells. Virus particles with an ‘altered’ morphology can be seen crosslinking adjacent cells (Fig. 2B) but, to date, no convincing examples of virus bridging fusion have been detected. This may not be too surprising, however, since the fusion of erythrocytes is more synchronous than for HeLa cells and, even then, it was necessary to examine many replicas of cells at early stages of fusion in order to detect a few examples of virus bridge formation. Because of their highly convoluted surface morphology the problem is even more difficult with HeLa (or other eukaryotic) cells. It should be pointed out that no examples of membrane fusion not involving a viral envelope have been observed either.
Light- and electron-microscope observations of fusing cells suggest that the formation of large spherical fused cells from individual HeLa cells takes place when cells joined by small cytoplasmic bridges, as a result of membrane fusion, swell. At intermediate stages during swelling dumbbell-shaped cells are common and Fig. 3 shows 3 partially swollen fused HeLa cells. An unfolding of surface microvilli takes place during cell swelling (Knutton et al. 1976) but at this intermediate stage many microvilli are still present. During the swelling of fused erythrocytes membrane vesicles are frequently seen inside the cell (Knutton, 1977). Although many membrane fracture faces are seen inside partially swollen fused HeLa cells at the region of expansion (Fig. 3, arrows) it is difficult to tell if any of these are plasma membrane derived. Thin-section images, however, show large vesicular profiles of villated surface membrane inside some partially swollen fused cells (Fig. 3, inset). Although it is not possible to prove from a single thin section image that such membrane profiles do represent internalized membrane, it is hard to believe that, at this stage of cell swelling, they are still connected to the external cell surface.
The importance of cell swelling in cell-cell fusion is indicated by carrying out fusion in the presence of concentrations of saccharides which inhibit cell swelling (Maeda et al. 1977). In this case no cell fusion is observed (Fig. 4B) in contrast to cells fused in the absence of sugars where cell swelling does occur (Fig. 4A). In swollen cells, identified in freeze-fracture replicas by a sparsity or lack of microvilli, lateral diffusion of viral envelope components occurs following incorporation of the viral envelope into the cell membrane; segments of smooth linear viral membrane diffuse away from the original site of fusion (Fig. 5) and eventually disappear altogether. A further membrane modification of some swollen cells is the aggregation of P face intramembrane particles (Fig. 6).
During virus-induced cell fusion endocytosis of intact virus particles has been observed in a few instances; virus particles are seen in the cell inside membrane-bounded compartments (Fig. 7).
The morphological observations presented here show that the mechanism of fusion of Sendai virus particles with the HeLa cell plasma membrane is identical to that described in detail for the fusion of virus particles with the erythrocyte membrane (Knutton, 1977, 1978). Fusion involves a cell-mediated, temperature-dependent reorganization of viral envelope structure to produce smooth, probably lipid bilayer (Branton, 1971), segments of viral envelope which interact and fuse with the cell plasma membrane. No modification of the HeLa cell membrane at sites of fusion was detected. An aggregation of intramembrane particles, which has been considered to be important in cell-cell fusion (Bächi & Howe, 1972; Bächi, Aguet & Howe, 1973) clearly does not occur at the time of virus-cell fusion (i.e. 2-3 min after warming to 37 °C) but much later when cells are fused and swollen. Although virus particles can enter membrane-bounded compartments inside the cell by endocytosis, the specialized viral mechanism for fusion of the viral envelope with the cell plasma membrane providing direct entry of the nucleocapsid into the cell suggests that this, and not viropexis (Dales, 1973) is the mode of Sendai virus infection of permissive cells.
Studies of Sendai virus-induced fusion of erythrocytes (Knutton, 1978) have led to the conclusion that cell-cell fusion involves 3 distinct stages: (1) the membranes of adjacent cells are brought into close contact (i.e. cell agglutination), (2) membrane fusion occurs at one (or a small number) of sites of close cell contact to produce cells connected by small cytoplasmic connexions. In the case of erythrocytes this membrane fusion event appears to involve the simultaneous fusion between a virus particle and two crosslinked cells, and (3) expansion of cells connected by small cytoplasmic connexions to form spherical fused cells occurs by a process of permeability-induced cell swelling. Although the observations presented here confirm the existence of stage 1 (cell agglutination) and stage 3 (cell swelling), the initial membrane fusion event (stage 2) has proved elusive, probably for the reasons discussed earlier. Nevertheless, membrane fusion (stage 2) must occur. What these studies have, so far, been unable to determine is whether membrane fusion occurs directly between cells or whether membrane fusion is mediated by a bridging virus particle. The observation of ‘altered’ virus particles bridging 2 cells, virus particles which are capable of fusing with both cells, suggests that cell-viral envelope-cell bridge formation can occur and is probably the initial membrane fusion event in HeLa cell fusion as well as in erythrocyte fusion. Further studies, however, are clearly required to provide direct evidence to prove this point.
Osmotic cell swelling appears to be the driving force which then expands cells connected by small cytoplasmic bridges to form fused cells (Knutton, 1978). Loss of all cell surface microvilli (Knutton et al. 1976, 1977) and the observed internalization of plasma membrane-derived vesicles inside swollen fused cells gives some indication of the drastic effects swelling can have on cell structure. Membrane modifications occurring as a result of cell swelling now appear to be responsible for such phenomena as the aggregation of intramembrane particles and the rapid lateral mobility of viral antigens (Bächi et al. This aspect of the fusion process will be discussed more fully in the accompanying paper.
A comparison of the morphological observations reported here with those previously reported in greater detail for the fusion of human erythrocytes (Knutton, 1977, 1978) suggests that the mechanism of fusion is probably the same for both cell types. Since virus-cell fusion takes place within 2–3 min of transferring an agglutinated cell suspension to 37 °C, the difference in time taken to observe a high incidence of fused erythrocytes (2–3 min) and HeLa cells (15–30 min) suggests that swelling occurs very rapidly for erythrocytes but much more slowly with HeLa cells. This difference presumably reflects the different metabolic capacity of these 2 cell types to overcome the effect of a membrane permeability change induced as a result of virus-cell fusion.
The author is grateful to Mrs Diane Jackson for expert technical assistance and the Cancer Research Campaign for financial support.