Studies of membrane fusion.: III. fusion of erythrocytes with polyethylene glycol

High-molecular-weight polymers of polyethylene glycol (PEG) were first shown to be potent fusogens for plant protoplasts (Kao & Kichayluk, 1974; Wallin, Glimelius & Erikkson, 1974) but more recently have been used to fuse a variety of cell types (Ahkong et al. 1975 b; Pontecorvo, Riddle & Hales, 1977). Because PEG is relatively non-toxic, yields a high incidence of cell fusion and can be used to fuse a wide variety of different cells including interspecific and interkingdom cell types (Ahkong et al. 1975b; Jones et al. 1976), it is rapidly becoming a very widely used cell fusogen. Most of the studies, however, have emphasized techniques for optimizing the production of cell hybrids; few studies have been concerned with the actual mechanism of PEG-induced cell fusion (Maggio, Ahkong & Lucy, 1976; Maul, Steplewski, Weibel & Koprowski, 1976). Thin-section observations (Maul et al. 1976) have shown that cell-cell fusion initially occurs at small localized areas of the cell surface but, because of the limited information present in images of cross-sectioned membranes, these studies have provided little insight into the actual membrane modifications taking place which allow membranes to fuse. The freeze-fracture technique, on the other hand, allows one to examine extensive en face views of membranes and also has the property of allowing integral membrane proteins and phospholipids to be distinguished morphologically and, therefore, offers the opportunity of defining how these different membrane components are involved in membrane fusion. In this paper I present freeze-fracture observations on the fusion of human erythrocytes and describe PEG-induced changes in membrane structure which show that membrane fusion occurs between small localized areas of naked lipid bilayer. Subsequent cell swelling results in the production of spherical fused cells.

PEG-induced fusion of erythrocytes was carried out essentially according to the procedure of Davidson & Gerald (1977). Human or chick erythrocytes were washed 3 times in Hanks’ balanced salt solution (HBSS) (Hanks & Wallace, 1949) and a 2% suspension (∼1·5 × 10⁸ cells/ml) used for fusion experiments. 1 ml of an erythrocyte suspension was centrifuged (1000 g, 5 min) and the cell pellet resuspended in 1 ml of a 50 % solution of PEG (6000 molecular weight) in HBSS at 4 °C. After 1 min the aggregated cells were diluted with 9 ml of HBSS and the cells centrifuged (1000 g, 5 min). The cell pellet was resuspended in 5 ml of fresh HBSS and fusion initiated by transferring the cell suspension to a waterbath at 37 °C. Samples of agglutinated cells (cells in 50 % PEG), following removal of most of the PEG at 4 °C and cells incubated for up to 60 min at 37 °C, were processed for thin section, scanning and freeze-fracture electron microscopy. Cell fusion was assessed by light microscopy.

For thin sections cells were fixed with 3 % glutaraldehyde in 0·1 M sodium cacodylate pH 7·4 for 1 h, postfixed with 1 % buffered osmium tetroxide for 2 h, blockstained with 2 % aqueous uranyl acetate, dehydrated through a graded series of ethanol and propylene oxide solutions and embedded in Epon. Sections were cut with glass knives on a Tesla ultramicrotome, stained with uranyl acetate and lead citrate and examined in a Siemens 101 electron microscope.

For scanning electron microscopy cells were allowed to settle on to gelatin-coated glass coverslips (Vial & Porter, 1975) and immediately fixed with 3 % glutaraldehyde in HBSS for 30 min. In some cases cells were postfixed for 1 h with 1 % buffered osmium tetroxide prior to being dehydrated through a graded series of acetone solutions. The coverslips were then transferred to liquid carbon dioxide and critical-point dried. Finally the coverslips were mounted on aluminium stubs, coated with a thin layer of gold, and the specimens examined in a Cambridge Type II stereoscan.

For freeze-fracture electron microscopy cells were fixed with 3 % glutaraldehyde in HBSS for 30 min, washed, and infiltrated with 25 % glycerol also in HBSS. Specimens were rapidly frozen in freshly melted Freon 22 and platinum carbon replicas made in a Denton freezefracture machine. Replicas were cleaned with bleach and washed in distilled water before being picked up on uncoated grids and examined. In freeze-fracture micrographs the encircled arrows indicate the direction of platinum shadowing.

Suspending cells in a 50% solution of PEG results in instantaneous cell agglutination and the formation of large non-specific cell aggregates (Fig. 1A). Cells shrink and become highly distorted as a result of the formation of large planar areas of close contact between cells (Figs. 1–3). At regions of close apposition adjacent cell membranes are separated by less than 5 nm and in many cases they come into sufficiently close contact that one sees an apparent fusion of the outer dense leaflets of the adjacent ‘unit membrane’ structures (Fig. 2). However, at this stage, the 2 membranes are not fused in any strict sense of the term and removal of the PEG by washing results in separation of the agglutinated cells. By freeze-fracture electron microscopy it can be seen that PEG treatment induces dramatic structural changes in the plane of the membrane (Fig. 3). Fractures through agglutinated cell aggregates reveal the large planar areas of close cell contact (Fig. 3A). PEG induces an aggregation of the P fracture face intramembrane particles into small patches (Fig. 3A) or large aggregates (Fig. 3B) separated by islands of smooth membrane. Such particle aggregates, however, only occur at sites of cell-cell contact. Regions of cell membrane not in contact with an adjacent cell retain an essentially random distribution of membrane particles (Figs. 1B, 3A, asterisks). The tight interaction between adjacent cell membranes at sites of contact is such that the fracture plane frequently jumps from one membrane to the other (Fig. 3 A). At higher magnifications (Fig. 3B) it can be seen that there exists a complementarity between smooth and particulate regions of adjacent cell membranes. This is readily apparent because an impression of the aggregated P-face intramembrane particles can be seen on the E fracture face of the adjacent membrane. The aggregates of P-face particles can be seen to correspond to the aggregates of E-face pits; similarly there is a complementarity between smooth regions of fracture face. The E-face intramembrane particles although unaggregated are also localized in regions of fracture face associated with the aggregates of E face pits (Fig. 3B).

Following the removal of most of the PEG, cells remain agglutinated but are less tightly bound. Rather than being in close contact over large areas of cell surface, oblique fractures through agglutinated cells show that cells only remain in close contact at small (∼ 0·1 μm diameter) plaques of smooth particle-denuded membrane (Fig. 4C, D, arrows). Intercellular space is clearly visible in between the plaques of smooth membrane. That adjacent cells remain tightly bound, however, is indicated by the fact that the membrane becomes significantly distorted at these focal points of contact (Fig. 4D, arrows). In regions of fracture face no longer in close contact with an adjacent cell the intramembrane particles appear to have returned to an essentially random distribution (Fig. 4A, B).

Incubation of an agglutinated cell suspension from which most of the PEG has been removed for 30–60 min at 37 °C results in extensive cell-cell fusion and the formation of large spherical polyerythrocytes (Fig. 7). Other unfused erythrocytes also swell (Fig. 7) and haemolysis of most cells eventually takes place. Intermediate stages during the formation of spherical fused cells are revealed by taking samples after brief (0–10 min) periods of incubation. The earliest fusion event appears to involve fusion between small areas of adjacent cell membrane resulting in cells connected by small (∼ 0·1–0·2 μm diameter) cytoplasmic connexions (Fig. 5 A, B). The appearance of the intracellular matrix at this stage (Fig. 5 A) indicates that the cells are still unswollen and are not haemolysed.

Light microscopy indicates that spherical fused cells arise from cells connected by small cytoplasmic bridges by a process of cell swelling during which the cytoplasmic bridges become expanded. Expansion of the initial cytoplasmic connexions during swelling results in fusing cells having a dumbbell shape and many such cells are seen during early stages of fusion. Fig. 6 shows one polyerythrocyte resulting from fusion of at least four individual erythrocytes which has not yet expanded to a spherical shape. The dumbbell shape of the fusing cells is readily apparent. The freeze-fracture appearances of fused cells are similar to controls; there is an essentially random distribution of intramembrane particles (Fig. 6), although in many instances, PEG treatment and the subsequent incubation at 37 °C has the effect of inducing segregation of smooth membrane (Fig. 6, asterisk) and the formation of droplets identified as lipid by their morphology. Such lipid droplets frequently bleb off from either the interior or exterior cell surface (Fig. 6).

It has been proposed that cell-cell fusion occurs by a mechanism which involves three distinct stages (Knutton, 1978). Cell fusogens induce (1) close contact between adjacent cell membranes; (2) membrane fusion at small localized sites of cell contact; and (3) expansion of sites of membrane fusion to form spherical fused cells by a process of cell swelling. Although the molecular mechanisms whereby different fusogens elicit this sequence of events clearly differ, the morphological observations presented here and illustrated in diagrammatic form (Fig. 8) are consistent with such a mechanism.

PEG does induce cell agglutination and the formation of very close contacts between adjacent cell membranes. Normally, the plasma membranes of adjacent cells visualized by transmission electron microscopy do not approach each other closer than ∼ 20 nm. This is both because of mutual electrostatic repulsion between 2 closely apposed membranes and because of the exclusion volume of the plasma membrane glycoproteins and glycocalyx macromolecules (Maroudas, 1975). PEG may facilitate close cell contact (< 5 nm) by reducing the exclusion volume of membrane glycoproteins. In addition to facilitating close membrane contact PEG induces an aggregation of intramembrane particles. Once close apposition (i.e. < 5 nm) between charged membranes is achieved, an aggregation of intramembrane particles could occur by direct electrostatic displacement (Gingell & Ginsberg, 1978). In the case of erythrocytes it is known that the intramembrane particles are associated with the major charge-bearing groups of the cell surface (Pinto da Silva & Nicolson, 1974). Although other mechanisms of intramembrane particle aggregation are possible (Gingell & Ginsberg, 1978), e.g. spectrin aggregation (Elgsaeter, Shotton & Branton, 1976), that contact-mediated electrostatic displacement is the mechanism in this case is supported by the observation that intramembrane particle aggregation occurs only at sites of close cell-cell contact and not over the remainder of the cell surface. Furthermore, removal of PEG reverses the process of particle aggregation. An aggregation of intramembrane particles was suggested in the study of PEG-induced fusion of mouse L cells (Maul et al. 1976) but these authors failed to examine cells in the presence of high concentrations of PEG. Although a direct correlation between thin-section and freeze-fracture images cannot be made, it seems likely that the regions of membrane showing very close contact (i.e. where there is an apparent fusion of the outer dense leaflets of adjacent ‘unit membrane’ structures) represent complementary regions of intramembrane particle-denuded lipid bilayer since, on removal of most of the PEG, cells only remain in close contact at small plaques of smooth membrane and thin-section images show that fusion of the ‘unit membrane’ structures also occurs at these plaques.

In agreement with previous studies with mouse L cells (Maul et al. 1976), PEG-induced fusion of erythrocytes is also seen to occur initially at small localized areas. Since membranes remain in close contact only at small plaques of smooth intramembrane particle-denuded membrane following removal of PEG and since the cytoplasmic continuities initially formed are the same order of magnitude it seems reasonable to conclude that membrane fusion has occurred at the plaques of smooth membrane. There is now a considerable body of evidence to suggest that membrane fusion occurs via lipid-lipid interactions between protein-depleted regions of adjacent apposed membranes (Satir, Schooley & Satir, 1973; Papahadjopoulos, Poste & Shaeffer, 1973; Orci, Perrelet & Friend, 1977; Lawson et al. 1977) and a mechanism of fusion based on the interaction and fusion of naked lipid bilayers has been proposed (Ahkong, Fisher, Tampion & Lucy, 1975 a). The observations presented here suggest that PEG-induced membrane fusion also occurs by such a mechanism.

The formation of a spherical fused cell, following membrane fusion at one (or more) small localized areas of cell contact, must involve membrane redistribution and expansion of the cytoplasmic continuities. Both light and electron microscopy suggest that this occurs by a process of cell swelling. It is clear that erythrocytes fused at 37 °C for 30 min are all swollen and, in most cases, haemolysed; at earlier stages, fusing cells with partially expanded cytoplasmic continuities are common. Biochemical studies have shown that during PEG-induced cell fusion there is a change in membrane permeability (Knutton, Micklem & Pasternak, unpublished observations). Loss of cation asymmetry, as is known to take place during Sendai virus-induced cell fusion (Poste & Pasternak, 1978), would then result in entry of water and cell swelling. The origin of the membrane permeability change is still unknown in the case of PEG treatment but the freeze-fracture observations show that PEG does perturb membrane structure quite drastically, sometimes in an irreversible manner, and frequently results in the segregation of lipids and the blebbing off of lipid droplets from the cells. Such membrane perturbations could lead to changes in membrane permeability. On the other hand, lipid segregation may be the consequence of changes in membrane permeability since it has been shown that conditions which cause a contraction of the actin-spectrin meshwork compress the lipid bilayer of the membrane causing it to bleb off particle-free vesicles (Elgsaeter et al. 1976). Entry of external Ca²⁺ for example, could be one possible cause of lipid blebbing. Nevertheless, membrane permeability changes, whatever their origin, which result in cell swelling do appear to provide the driving force which expands cells connected by small cytoplasmic connexions to form spherical fused cells.

I am grateful to Mrs Diane Jackson for excellent technical assistance and the Cancer Research Campaign for financial assistance.