Dynamics of membrane skeleton in fused red cell membranes

Transcription

1 Dynamics of membrane skeleton in fused red cell membranes JIN CHENG-ZHI* Medical Research Council Cell Biophysics Unit, King's College, Druiy Lane, London WC2B 5RL, UK * Permanent address: Fudan University, Shanghai, People's Republic of China Summary The dynamic nature of the red cell membrane cytoskeleton was examined by the method of fluorescence redistribution after fusion (FRAF). Ghosts, labelled at the interior membrane surface, and intact cells, labelled at the outer surface with a different fluorophore, were fused and the passage of the fluorescent label into the unlabelled part of the membrane was followed in the fluorescence microscope. To achieve specific labelling of a cytoskeletal component only, a fluorescent analogue of phalloidin was used to label the actin. It was shown that there was no dissociation of the phalloidin during the time of the experiment, and no exchange of phalloidin between labelled and unlabelled actin in unsealed mixtures of ghosts. In the fused membrane pairs at 37 C the phalloidin-linked fluorescence diffused in the plane of the membrane and became distributed through the membrane envelope in 1-2 h. It was shown by gel electrophoretic analysis that no detectable loss of or damage to membrane proteins resulted from the fusion process. It is concluded that, at least in membrane pairs treated with fusogen, exchange of actin or actin-containing elements occurs within the cytoskeletal network. Key words: membrane, cytoskeleton, dynamics. Introduction The unusual mechanical properties of the red cell membrane result from the presence of a protein network that densely covers the cytoplasmic surface of the bilayer. The structural proteins that are needed to form a continuous lattice are spectrin, actin and 4.1. The spectrin is in the form of elongated tetramers, with an extended length of about 200 nm; these are attached at their extremities to short filaments (12-20 subunits) of F-actin, and the association is promoted by protein 4.1, of which there is one molecule for every spectrin tetramer end. In the electron microscope the spectrin can be seen to make up a rather regular hexagonal lattice, with the actin and 4.1 at the junctions (Byers & Branton, 1985; Shen et al. 1986). The membrane cytoskeleton interacts with the membrane by more than one path: spectrin is bound to ankyrin, which is itself attached to the major transmembrane protein, band 3; the 4.1 is attached to the intracellular domains of the transmembrane sialoglycoproteins, the glycophorins, and there is evidence that both spectrin and 4.1 may interact directly with the phospholipid bilayer (for review see Bennett, 1985). Journal of Cell Science 90, (1988) Printed in Great Britain The Company of Biologists Limited 1988 The membrane cytoskeleton is a highly durable and stable structure. At the same time the individual ternary interactions at the junction points are of only moderate strength (ternary association constant approx M~ 2 in vitro; Ohanian et al. 1984), although the short actin filaments are strongly stabilized by the spectrin and 4.1 and will not dissociate to form subunits under physiological solvent conditions, nor will they rearrange to form longer filaments (Pinder & Gratzer, 1983). Nevertheless, some evidence exists that the membrane cytoskeleton has dynamic properties. Koppel et al. (1980) have argued that the translational diffusion of transmembrane proteins in the membrane is limited by the local dissociation of the cytoskeletal network, through the interstices of which they penetrate. This diffusion is promoted by physiological concentrations of the polyanions, ATP and 2,3- diphosphoglycerate, and these ions are also able to dissociate the isolated cytoskeletons in vitro (Sheetz & Casaly, 1980). To obtain a direct measure of the extent of diffusional freedom of cytoskeletal elements we have employed the method (Fowler & Branton, 1977) of 93

2 observing redistribution of fluorescence-labelled proteins in fused pairs of labelled and unlabelled membranes (FRAF). The results indicate, at least in the system generated by fusion with the fusogen, polyethylene glycol, a much higher degree of diffusion of cytoskcletal proteins in the plane of the membrane than had been supposed. Materials and methods Human red cells were prepared from blood, less than 1 week old. The cells were washed three times at 4 C with phosphate-buffered physiological saline (0-15 M-sodium chloride, 5 mm-potassium phosphate, pi I 7-4). Any white cells present were carefully removed each time. The cells were suspended in saline at a haematocrit of 3 % and the cells in I ml of this suspension were lysed by addition of 30 vol. of hypotonic buffer, containing 30mM-sodium chloride, 1 mm-potassium phosphate, pi I 7-4. After 5 min on ice, the membranes were pelleted by centrifugation at ^ for 10 min. The pellet was resuspended in 20 ml of the isotonic buffer and again recovered by centrifugation. The ghosts were kept on ice. For resealing, 10 ml of the isotonic buffer, containing in addition I mm-magnesium chloride and 1 mm-calcium chloride, was added to the pellet. After incubation at 37 C for 1 h, the resealed ghosts were recovered by centrifugation and resuspended in the same buffer. Fusion between resealed ghosts and intact cells was carried out on a microscope slide. Coverslips were coated with a monolayer of polylysine by immersion for 30 min in polylysine at 3^(gml~' in isotonic buffer. They were rinsed and one drop of the ghost suspension or of a mixture of ghosts and intact cells was applied to the centre. The coverslips were kept at room temperature for 30 min. After rinsing gently with isotonic buffer, several drops of polyethylene glycol (nominal M r 6000), 35% (w/w) in isotonic buffer, were added to the surface and the coverslip was kept at 37 C for 5 min. It was then rinsed by immersion in the resealing buffer, left in the buffer for 15 min and transferred to a dish containing a hypotonic medium (30 nim-sodium chloride, 1 mm-potassium phosphate, 02mM-magnesium chloride, 0-2 mm-calcium chloride, pi I 7-4). After 15 min at 37 C, the coverslip was placed, coated side down, supported at both edges on coverslips, cemented to a microscope slide. The upper face was blotted, and hypotonic buffer was introduced with a pipette into the gap between the coverslip and the slide. This assembly was placed on the microscope stage for examination by phase-contrast and fluorescence microscopy. For fluorescent labelling of intact cells a suspension at a haematocrit of 3 % in isotonic buffer was added to 8 vol. of 0-lingml~' filtered fluorescein isothiocyanate (FITC) or 30^gml~' rhodamine isothiocyanate (RITC) in the same buffer. After agitation at 0 C for 8h, the cells were washed twice with 25 ml buffer, containing 0'3 mginl"' glycineamide hydrochloride. The pellet after this treatment was washed three times more with isotonic buffer, diluted with 0-75 ml buffer and kept on ice. Unsealed ghosts, prepared as described from 1 ml of the 3 % haematocrit red cell suspension, were resuspended in 1 ml of isotonic buffer at 0 C, containing 50,l/gml~ tetramethylrhodamine phalloidin (Sigma). After agitation for 2h on ice the ghosts were pelleted, 1 ml isotonic buffer was added and the suspension was again agitated for 30 min on ice. The labelled ghosts were washed three times as before and kept on ice. For gel electrophoresis in the presence of sodium dodecyl sulphate, an 8% polyacrylamide gel was used with a stacking gel and a discontinuous buffer system (Laemmli, 1970). Results The fusion process, as observed in the light microscope, occurred over quite widely differing intervals of time. When fusogen-treated cells were transferred to a hypotonic medium, some of the pairs of cells that were already in contact showed an essentially instantaneous coalescence and a single compartment was formed, as shown by the disappearance of the dividing membrane, viewed by phase-contrast microscopy. Other pairs (or groups) of cells fused much more slowly, and the process was sometimes not complete until several hours had elapsed. Ghosts were prepared from cells that had been labelled at the outer surface with RITC. This reagent, or its analogue, FITC, under the conditions used, labels essentially only the extracellular domain of band 3 (Fowler & Branton, 1977). RITC was preferred to FITC because of its superior resistance to photoblcaching. Following fusion, or at all events the point at which the pair or cluster of original cells could be seen to acquire a common smooth envelope, fluorescence was clearly apparent at the original interface, and this remained perceptible for periods of minutes to hours. This indicates that after the fusion event, which is accompanied by the disappearance of the lipid bilayer from the interface, the major transmembrane protein, band 3, or at least a significant part of it, remains at the interface. In the absence of fluorescence a thin dividing line can also be discerned between the original compartments, consisting presumably of the membrane cytoskeleton. This is confirmed by the fluorescence of the interface after labelling with TRITC-phalloidin (see below). The fluorescent band 3 must be supposed to be associated with this cytoskeletal complex. An attempt was next made to achieve unique labelling of the internal (in particular the cytoskcletal) proteins by blocking the external binding sites on intact red cells with methyl-, phenyl- or rhodamine isothiocyanate, and then reacting them with fluorescein isothiocyanate after lysis and before resealing. This procedure was only partially successful, because complete blocking could not be achieved, and a background of fluorescein fluorescence from the outer membrane surface proved inescapable. We therefore sought a more specific means of labelling cytoskeletal proteins. For the subsequent observation of diffusion effectively irreversible labelling is essential. TRITC-phalloidin 94 Jin Cheng-zhi

3 Fig. 1. Ghosts labelled with TRITC-phalloidin (B) and the same after incubation with OSmgml ' non-fluorescent phalloidin (A). The exposure time was the same. The reduction in background fluorescence is due to additional washing steps in the displacement procedure. Fig. 2. Micrographs of typical fused ghost red cell combinations. The intact cells were labelled at the outer surface with fluorescein isothiocyanate (green fluorescence) and the ghosts at the inner surface with tetramethylrhodamine phalloidin (red fluorescence). These pictures were taken 90min after addition of the fusogen. Micrographs were taken by phasecontrast (A), and with filters to isolate the green (B) and the red (C) fluorescence. Note that the green fluorescence (on the transmembrane proteins) has diffused throughout the overall membrane envelope, and that the red fluorescence has also extended into the entire membrane area. Its intensity becomes progressively more uniform with time.

4

5 was eventually selected, because this is highly specific for F-actin, and labels nothing else in cells (Barak et al. 1980), and its binding to F-actin is such that the dissociation rate is immeasurably small; indeed the binding has been described as irreversible (Wieland, 1977; Grandmont-Leblanc & Gruda, 1977). TRITC phalloidin gave rise to a very stable fluorescence at the inner membrane surface. To show that it did not dissociate from the actin under the conditions and on the time scale of the diffusion experiments two types of control were performed. In the first, labelling was achieved by incubation with SO/igmP 1 TRITC phalloidin. After extensive washing to remove unbound material, underivatized phalloidin was added to a concentration of 0-5 m g m P 1. No disappearance of fluorescence from the membrane could be observed even after 10h (Fig. 1). Second, unsealed ghosts, labelled with TRITC-phalloidin and pelleted, were mixed with unlabelled unsealed ghosts. They were gently agitated together for 2h. On examination in the fluorescence microscope a mixed population of highly fluorescent and completely non-fluorescent ghosts was observed, showing that any equilibrium proportion of free phalloidin, through which exchange might occur, is of insignificant concentration. When the TRITC-phalloidin-labelled ghosts were fused with fluorescein-labelled intact cells, fused cell pairs in which one hemisphere showed red fluorescence and the other green could be seen. These were observed for long periods, and, as Fig. 2 shows, the red fluorescence advances into and eventually fills the green-fluorescent region. This movement is in the plane of the membrane, without any diffusion into the cytoplasmic space. Sequential records of the time course in a single cell pair could not be obtained because of the bleaching of fluorescence that accompanied a single exposure. However, cell pairs with differing degrees of penetration of the diffusing fluorescent label could be observed at different times. We can see no other interpretation for these observations than that the labelled actin is capable of diffusional movement of some sort within the cytoskeletal complex. To demonstrate that this is not an artefact resulting from proteolytic damage to the cytoskeletal proteins, due perhaps to the presence of calcium ions in the fusion medium, ghosts were suspended in the same medium at 37CC for 15min. The membranes were recovered by pelleting in the centrifuge and the proteins were examined by high-resolution polyacrylamide gel electrophoresis in the presence of SDS (Fig. 3). No qualitative or quantitative alteration in the protein pattern, compared with the pristine ghosts, could be detected. Sp Ac A i a Fig. 3. Gel electrophoresis, in the presence of SDS, of red cell membrane proteins from untreated ghosts (A) and those exposed to the calcium-containing medium in which fusion occurred under the conditions used in the diffusion experiments (B). Note absence of satellite bands that would result from proteolysis. Spectrin (Sp), bands 3, 4.1 and actin (Ac) are indicated. Discussion There appear to have been no earlier studies of the extent to which the red cell membrane cytoskeleton is a dynamic structure, within which exchange of proteins is possible. It seems generally to have been supposed that the complex is highly stable and not prone to local dissociation at physiological ionic strength. The remarkable ability of the cell to become distorted under shear or by perturbation with membrane-active agents (see, e.g., Mohandas & Shohet, 1978) must obviously reflect some form of rapidly reversible structural change in the membrane cytoskeleton. Local dissociation of spectrin tetramers or of spectrin from the junction points, for example, is clearly a possibility, although Elgsaeter and his associates (see, e.g., Elgsaeter et al. 1986) proposed a model of membrane flexibility, based on elastic properties of the spectrin molecules themselves. There is, however, no doubt that local sealing of an interrupted cytoskeleton does occur, as in fusion or elimination of excess membrane in immature cells (Zweige/ al. 1981). Conversely, local withdrawal of the cytoskeleton precedes endocytosis of membrane material (Hardy et al. 1979), and the externally stimulated clustering of intramembranc proteins is mirrored by a spatial rearrangement of the distribution of spectrin (Ji & Nicolson, 1974). Membrane skeleton dynamics 95

6 In this sense the results described here are perhaps not as unexpected as might at first appear. Cells incubated with the fusogen under conditions leading to fusion when the membranes come into contact do not lose their stability, as occcurs when the cytoskeletal complex is damaged (see, e.g., Shohet, 1979; Coakley el al. 1978). This suggests that exposure to fusogen does not markedly weaken the cytoskeletal organization. Moreover, as judged by gel electrophoresis, there is no proteolytic or other damage to any membrane proteins. The method that we have used to follow diffusion (FRAF) necessitates a fusion step: we have preferred this to the alternative approach of fluorescence recovery after photobleaching of a fluorescence-labelled membrane (FRAP), because it avoids the uncertainties, such as possible photochemical or thermal damage, inherent in the latter (Bretscher, 1980). At the same time, we can certainly not exclude the possibility that our system has been perturbed by the fusogen, or by the process of fusion itself. What can be stated with reasonable assurance is that there is no gross structural or chemical perturbation, and the system at least approximates to the native state. It should also be noted that the process of fusion is not always easily identified or defined. The hypotonic conditions used here are those that promote fusion under the influence of Sendai virus (Knutton & Bachi, 1980). Different criteria for fusion may not give mutually concordant results (Diizgiines et al. 1987). The precise nature of the interaction between the apposed membranes does not concern us here, except to the extent that the diffusion of the fluorescence into the non-fluorescent partner proceeds in cell pairs when a single continuous membrane envelope can be seen. The results indicate clearly that the diffusion of the actin does not occur through a pool of free protein in solution, but proceeds in two dimensions in the plane of the membrane. One must thus envisage some form of exchange process within the cytoskeletal complex. Whether this involves the transport of single actin subunits or intact junctions cannot be stated. However, the evidence is that the actin protofilaments are very stable in the complex (Pinder & Gratzer, 1983) and will be more so when associated with phalloidin. By contrast, the spectrin self-association through the sites at the centre of the tetramer is relatively weak (see, e.g., Shahbakhti & Gratzer, 1986), and thus one may more readily envisage an exchange of junction units within a relatively fluid lattice. I am grateful to the Royal Society and the Sino-British Fellowship Trust for support during part of the time in which this work was carried out. I thank G. A. Dunn and W. B. Gratzer for advice and discussion. References BARAK, L. S., YOCUM, R. R., NOTHNAGEL, E. & WEBB, W. W. (1980). Fluorescence staining of the actin cytoskeleton in living cells with 7-nitrobenz-2-oxa-l,3- diazole-phalloidin. Pivc. natn. Acad. Sci. U.S.A. 77, BENNETT, V. (1985). The membrane skeleton of human erythrocytes and its implications for more complex cells. A. Rev. Biochem. 54, BRETSCHER, M. (1980). Lateral diffusion in eukaryotic cell membranes. Trends Biochem. Sci. 5, VI VII. BYERS, T. J. & BRANTON, D. (1985). Visualization of the protein association in the erythrocyte membrane skeleton. Proc. natn. Acad. Sci. U.S.A. 82, COAKLEY, W. T., BATER, A. J. & DEELEY, J. O. T. (1978). Vesicle production on heated and stressed erythrocytes. Biochim. biophys. Act a 512, DOZGONES, N., ALLEN, T. M., FEDOR, J. & PAPAHADJOPOULOS, D. (1987). Lipid mixing during membrane aggregation and fusion: Why fusion assays disagree. Biochemistry 26, ELGSAETER, A., STOKKE, B. T., MIKKELSEN, A. & BRANTON, D. (1986). The molecular basis of erythrocyte shape. Science 234, FOWLER, V. & BRANTON, D. (1977). Lateral mobility of human erythrocyte integral membrane proteins. Xatiire, bond. 268, GRANDMONT-LEBLANC, A. & GRUDA, J. (1977). Affinity chromatography of heavy meromyosin and heavy meromyosin subfragment-1 on F-actin stabilized by phalloidin. Can.J. Biochem. 55, HARDY, B., BENSCH, K. G. & SCHRIER, S. L. (1979). Spectrin rearrangement in erythrocyte ghost endocytosis. J. CellBiol. 82, Jl, T. H. & NlCOLSON, G. L. (1974). Lectin binding and perturbation of the outer surface of the cell membrane induces a transmembrane organizational alteration at the inner surface. Proc. natn. Acad. Sci. U.S.A. 71, KNUTTON, S. & BACHI, T. (1980). The role of cell swelling and haemolysis in Sendai virus-induced cell fusion and in the diffusion of incorporated viral antigens.,7- Cell Biol. 42, KOPPEL, D. E., SHEETZ, M. P. & SCHINDLER, M. (1980). Matrix control of protein diffusion in biological membranes. Proc. natn. Acad. Sci. LS.A. 78, LAEMMLI, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, Land. 227," MOHANDAS, N. & SHOHET, S. B. (1978). Control of red cell deformability and shape. Curr. Top. Hemat. 1, OHANIAN, V., WOLFE, L. C, JOHN, K. M., PINDER, J. C, Lux, S. E. & GRATZER, W. B. (1984). Analysis of the ternary interaction of the red cell membrane skeletal proteins, spectrin, actin and 4.1. Biochemistly 23, Jin Cheng-zhi

7 PINIDER, J. C. & GRATZER, W. B. (1983). Structural and dynamic states of actin in the erythrocyte. J. Cell Biol. 96, SHAHBAKHI, T. F. & GRATZER, W. B. (1986). Analysis of the self-association of human red cell spectrin. Biochemistry 25, SHEETZ, M. P. & CASALY, J. (1980). 2,3- Diphosphoglycerate and ATP dissociate membrane skeletons. J. biol. Chem. 255, SHEN, B. W., JOSEPHS, R. & STECK, T. L. (1986). Ultrastructure of the intact skeleton of the human erythrocyte membrane. J. Cell Biol. 102, SHOHET, S. B. (1979). Reconstitution of spectrin-deficicnt spherocytic mouse erythrocyte membranes. J. din. Invest. 64, WIELAND, T. (1977). Modification of actins by phallotoxins. Nattoivissenschaften 64, ZWEIG, S. B., TOKUYASU, K. T. & SINGER, S. J. (1981). Membrane-associated changes during erythropoiesis. On the mechanism of maturation of reticulocytes to erythrocytes. J. sitpramolec. Struct. 17, (Received 21 October Accepted 25 January I9SS) Membrane skeleton dynamics 97

8