We often think of the plasma membrane as a physically separable part of the cell. It is chemically distinct: a thin oily film of lipids and hydrophobic proteins that is immiscible with the water-based cytoplasm it encloses. In questions of cell physiology and metabolism the outer membrane is conveniently regarded as an isolated permeability barrier controlling the entry and egress of ions and small molecules. However, when we come to consider the mechanical properties of cells and their movements, any attempt to segregate the plasma membrane from the rest of the cell is misleading. It is true that the forces necessary for motile phenomena such as locomotion and mitosis are generated mainly in the cytoplasm, by a system of protein filaments known collectively as the cytoskeleton. But the cytoskeleton does not move within the flexible enclosure of the plasma membrane like a cat in a bag. There are extensive and substantial links between the two that give the cell surface appreciable strength and make it an important structural element in cell movements.

Perhaps the most conspicuous form taken by the membrane-associated cytoskeleton is that of a cell ‘cortex’. Early studies of free-living amoebae, later extended to echinoderm eggs and macrophages, revealed a peripheral layer of gelated cytoplasm closely associated with the inner face of the plasma membrane. This appeared to be a major determinant of cell form and was at one time thought to be responsible for many kinds of cell movements including those of vertebrate cells. More recent work has shown that contacts between the cytoskeleton and the membrane in vertebrate cells are mainly confined to discrete regions, such as intercellular junctions and microvilli, each of which has a distinctive structure and complement of proteins. At first sight this seems at variance with the idea of a uniform, and universal, cortex.

Our plan in this short essay is to describe the layer of peripheral cytoplasm associated with the plasma membrane of animal cells and to compare the uniform thick cortex found in amoebae and echinoderm eggs with the much thinner layer present in the cells of vertebrate tissues. Since the function of the cortex is to give rigidity and form to the cell surface, we will also consider the cellular mechanics of the cortex and speculate on its role in cell movements. For reasons of space and fluency we will give only a short selection of references. However, a number of recent reviews are cited that will provide greater access to the very extensive literature relating to the cell cortex.

STRUCTURE OF THE MEMBRANE-ASSOCIATED CORTEX

Amoebae and echinoderm eggs

The earliest indication of a peripheral layer of cytoplasm associated with the outer cell membrane came from studies of large freshwater amoebae of the Amoeba proteus type. These organisms have been studied by microscopists for more than 200 years, and the differentiation of their cytoplasm into regions of distinct organelle content and physical consistency was one of the earliest features to be noted (see Allen & Kamiya, 1964; Jeon, 1973; Chambers & Chambers, 1961). Already, by 1926, Mast was able to summarize more than a century of observation and debate in his classical account of movement in A. proteus (Mast, 1926). Incorporating concepts of sol and gel states of cytoplasm, he named the central, relatively fluid region the plasmasol and the tube of stiffer cytoplasm that encloses it the plasmagel (two regions now also known as endoplasm and ectoplasm, respectively). Outside the plasmagel and between it and the plasma membrane, Mast described a clear layer of variable thickness known as the hyaline layer or hyaline cortex; this was continuous with a much larger cap of similar appearance at the tip of a pseudopod (the hyaline cap) (Fig. 1).

Fig. 1.

Schematic diagram of A. proteus crawling over a solid substratum showing the regional differentiation of the cytoplasm discussed in the text. The migration is accompanied by streaming movements in the inner plasmasol region, as indicated.

Fig. 1.

Schematic diagram of A. proteus crawling over a solid substratum showing the regional differentiation of the cytoplasm discussed in the text. The migration is accompanied by streaming movements in the inner plasmasol region, as indicated.

The hyaline layer of large amoebae is therefore the doyen of cell membrane cortices. By the early 1970s this layer had been shown to contain many thin filaments and evidence had been obtained that these were probably actin (Pollard & Ito, 1970; Comly, 1973). However, as far as we are aware there have been no further investigations of its precise nature and, moreover, there is reason to believe that it is not essential for cell movements. As noted by Mast, and corroborated by many subsequent observers, the material of the hyaline layer and cap is relatively fluid; the cell surface on one side of this layer slips easily over the plasmagel on the other side, with little if any permanent attachment between the two. When the outer membrane is physically removed from the cell, the denuded cytoplasm remains able to carry out streaming movements similar to those seen in intact pseudopodia (Allen, Cooledge & Hall, 1960; Taylor & Condeelis, 1979). In a sense, therefore, A.proteus does move as a ‘cat in a bag’, although the ‘bag’ in this case includes both the plasma membrane and a tightly associated actin-rich cytoskeleton.

In other types of amoeba the structure of the cortex is better understood. In an early study, isolated membrane fragments from the relatively minute protozoa Acanthamoeba were found to be associated with large numbers of actin filaments (Pollard & Korn, 1973). In the electron microscope these appeared as a series of filaments attached end-on to the inner surface of the plasma membrane, like a fur lining. A similar layer appears to exist beneath the membrane of the amoeboid form of the slime mould Dictyostelium discoideum and has been the subject of detailed biochemical studies of the actin-membrane association (Bennett & Condeelis, 1984). Isolated membrane fragments, freed of their extrinsic protein by treatment with 0·1 M-NaOH, will bind to purified actin filaments in a controlled and specific fashion, and several membrane proteins involved in this binding have been identified (Luna, Goodloe-Holland & Ingalls, 1984; Goodloe-Holland & Luna, 1984; Stratford & Brown, 1984). Interestingly, actin filaments are able to bind not only end-on but also laterally, and are evidently able to form a feltwork of actin filaments beneath the cell surface. A lateral association of this kind could occur through the binding of actin filaments to a large number of membrane proteins each with relatively low affinity.

Probably the most conspicuous and certainly the best-studied example of a membrane cortex is found in sea-urchin eggs. These cells are spherical, usually about 100μm in diameter, and have a distinct gelated layer of cytoplasm about 3·4μm thick associated with their inner surface (Hiramoto, 1957). The presence of this cortical layer, and the fact that it is firmly attached to the plasma membrane, were established in a series of delicate micromanipulation studies. For example, Hiramoto, by introducing fine glass needles into sea-urchin eggs, showed that the inner contents of the cytoplasm have the consistency of an easily stirred fluid while the peripheral layer 3-5 pm from the surface is more gel-like (Hiramoto, 1957). Attempts to push through this outer layer from the inside resulted in the deformation of the surface (Fig. 2). The existence of a cortex in these and other types of cell can also be demonstrated by allowing portions of the cell surface to stick to an adhesive coverslip and then washing the remainder of the cell away with a stream of fluid. The exposed inner surface of the cortex can then be visualized directly and its composition examined using antibodies and other specific reagents (see, e.g., Spudich & Spudich, 1979).

Fig. 2.

Experiment demonstrating the presence of a gelated cortex in sea-urchin eggs (Hiramoto, 1957). The glass needle inserted into the egg encounters mechanical resistance only when it comes within 3–5 μm of the egg plasma membrane.

Fig. 2.

Experiment demonstrating the presence of a gelated cortex in sea-urchin eggs (Hiramoto, 1957). The glass needle inserted into the egg encounters mechanical resistance only when it comes within 3–5 μm of the egg plasma membrane.

From studies of this kind it appears that echinoderm eggs, in common with the eggs of frogs and fruit-flies, have cortices composed of a thick actin-rich network. Before fertilization, the echinoderm cortex is relatively poorly ordered and contains less extensively polymerized actin. Shortly after fertilization, the surface of the egg becomes covered with numerous fine microvilli, each of which contains a bundle of actin filaments similar to those described below for the intestinal epithelial cells (Schroeder & Stricker, 1983).

Vertebrate cells

Any attempt to relate the echinoderm egg cortex to the membrane-associated cytoskeleton of vertebrate cells encounters problems of scale. A layer 3—5 μm thick would occupy most, if not all, of the contents of a fibroblast or an epithelial cell (Fig. 3).

Fig. 3.

Comparative sizes of some of the cells discussed in the text. Note that the fibroblasts appear to be relatively large since they are depicted as though flattened onto a culture substratum.

Fig. 3.

Comparative sizes of some of the cells discussed in the text. Note that the fibroblasts appear to be relatively large since they are depicted as though flattened onto a culture substratum.

Nevertheless, parallels have been drawn. In 1939, for example, Lewis compared the migration of leucocytes with that of giant freshwater amoebae and interpreted them in terms of a contractile layer of peripheral cytoplasm. Unfortunately, as we have seen, the layer of cytoplasm that generates movements in A. proteus is not firmly attached to the plasma membrane and the membrane cortex has no direct role in cell locomotion. But with regard to the membrane-associated cytoskeleton of leucocytes, Lewis’s ideas have been shown to be essentially correct. There is indeed a distinct layer of specialized cytoplasm beneath the plasma membrane of macrophages, neutrophils and lymphocytes (see reviews by Stossel, Hartwig & Yin, 1981 ; Oliver & Berlin, 1982; Loor, 1981; Hartwig, Niederman & Lind, 1985). In electron micrographs this appears as an amorphous inner coat of fine filaments from which other cytoplasmic organelles are excluded (Fig. 4). Evidence of several kinds shows that these filaments are composed largely of actin.

Fig. 4.A.

Neutrophils stained with rhodamine/phalloidin to reveal actin filaments. Note the intensity of the fluorescence in the region beneath the plasma membrane. Bar, 20 μM. B. Transmission electron micrograph of a neutrophil showing the actin feltwork beneath the residual plasma membrane. Bar, 1 μm. (Courtesy of Peter Sheterline.)

Fig. 4.A.

Neutrophils stained with rhodamine/phalloidin to reveal actin filaments. Note the intensity of the fluorescence in the region beneath the plasma membrane. Bar, 20 μM. B. Transmission electron micrograph of a neutrophil showing the actin feltwork beneath the residual plasma membrane. Bar, 1 μm. (Courtesy of Peter Sheterline.)

To learn more about the biochemistry of the cortex we would like ideally to obtain it in large quantities entirely free of other cell components. While this is difficult, it is feasible to obtain membrane fragments in reasonable amounts that consist of both plasma membrane and associated cytoskeleton. As we saw with regard to the Dictyostelium and Acanthamoeba cortices, preparations of this kind enable the links between actin filaments and integral membrane proteins to be analysed. A further step in purification used with a number of vertebrate cells is to isolate from a membrane preparation a residual cytoskeleton by detergent extraction. This results in a membrane shell, which has been extensively characterized in the case of the red blood cell, as described below, but is clearly more complex in the case of tissue cells (Apgar, Herrmann, Robinson & Mescher, 1985; Moss, 1983).

The primary source of information on the nature of the cortex itself, as distinct from its linkage to the plasma membrane, comes from studies of actin-rich gels. If sea-urchin eggs, macrophages or indeed, any of a wide range of other animal cell types are homogenized at low temperatures in buffers lacking Ca2+, watery extracts are obtained that turn into a semi-solid gel upon warming. Subsequent manipulation of the Ca2+ concentration then causes regional liquefaction and contraction, and even a form of directed streaming (Taylor & Condeelis, 1979).

Biochemical analysis of the gels permits identification of the proteins that ‘conspire’ with actin to produce the changes in physical state. These include a class of long flexible protein molecules, such as filamin, that are able to link adjacent actin filaments together whatever their relative orientation; even in small proportions, such proteins cause a solution of actin filaments to turn into a gel. Also present is a class of actin-fragmenting proteins, such as gelsolin, which can sever actin filaments in response to small increases in the concentration of Ca2+, and a number of other calcium-sensitive actin-binding proteins. The collective action of these proteins on an actin filament gel is to generate a precipitous fall in viscosity analogous to a gel-to-sol conversion. Finally, the gelating extracts from echinoderm eggs and macrophages contain myosin and associated control proteins, which enable it to produce a Ca2+-stimulated contraction of the actin filaments in a gel and, under the right conditions, a form of streaming (Weeds, 1982; Geiger, 1983).

Reconstituted gels resemble in their ultrastructure and physical properties the cortical layer of cytoplasm of large cells, and several of the gel-forming proteins have a predominantly peripheral location in the cytoplasm. They are probably not unique to the cortex, since similar sets of actin-binding proteins can be isolated from vertebrate tissue cells that lack a conspicuous membrane cortex, and even from the extruded, membrane-free cytoplasm of giant amoebae. It is clear, however, that without actin-binding proteins no cortex could be formed.

Motile cells such as neutrophils and macrophages are similar to amoeboid protozoa in the sense that they persue a ‘free-living’ existence, albeit within the confines of the animal body. The other cells of the multicellular tissues of the body, such as endothelial cells, lung alveolar cells and lens epithelial cells, have a more anisotropic environment, being in contact in some regions with neighbouring cells and in other regions with the extracellular matrix. Possibly for this reason, the layer of actin and other proteins associated with the plasma membrane is more variable in thickness and in composition. Connections with the plasma membrane still exist, but they are most prominent in regions of specialized function (Geiger, 1983). We will now consider briefly the nature of the actin-membrane association in these regions.

Microvilli are found in large numbers on the luminal surface of epithelial cells in the intestine and in other tissues. Each intestinal microvillus has a core of actin filaments arranged in a parallel bundle ; and a number of proteins that crosslink the actin filaments together and link them to the membrane have been identified. Lateral association with the plasma membrane seems to be due to a series of regularly spaced, 7nm×20nm bridges composed of a 110000 Mr actin-activated ATPase, which is tightly associated with calmodulin. The ends of the actin filaments appear, in the electron microscope, to be embedded in a plaque of dense material at the tip of the microvillus. Evidently the ends of the filaments are still accessible, since incubation of isolated brush-border preparations in actin solutions results in the growth of filaments at their distal, membrane-associated ends (Mooseker, Pollard & Wharton, 1982).

It may be noted in passing that growth of actin filaments at their membrane-associated ends seems to occur elsewhere. In the almost explosive extension of the actin-based acrosomal process of Thyone sperm, the kinetics of elongation are consistent with the diffusion-limited addition of actin monomers to the distal, membrane-associated end (Tilney & Inoue, 1982). In the far slower morphogenesis of the bundle of actin in the Limulus sperm acrosomal process, there is also ultrastructural evidence that growth occurs at the membrane-associated ends (Tilney, Bonder & DeRosier, 1981). More generally, in essentially every situation in which the polarity of actin filaments attached to membranes has been examined by myosin decoration, they have been found to be attached by the end that, in isolated actin preparations, is the preferred end for growth.

One further similarity between the peripheral cytoplasm of echinoderm eggs and that of vertebrate tissue cells lies in the specialized structure formed in the course of cell division. In most, if not all, animal cells, a circumferential band of actin filaments and associated proteins known as the contractile ring forms around the equator of the cell in the final stages of mitosis. This produces a mechanical force by its constriction, which, in turn, causes a furrow to form in the cell surface and eventually pinches the cell into two. In echinoderm eggs the contractile ring is clearly part of the submembrane cortex: indeed, the more centrally located cytoplasm can be disrupted or even removed entirely without arresting the progress of the cleavage furrow. The contractile rings of vertebrate tissue cells are smaller, but otherwise closely similar to those of echinoderm eggs. Evidently, they are also closely associated with the plasma membrane, since surface receptors directly or indirectly associated with the actin filaments align with the cleavage furrow during cytokinesis (Rogalsky & Singer, 1985).

Actin filaments are also attached to the plasma membrane at sites at which the cell as a whole is anchored to external features such as another cell, the extracellular matrix or a tissue culture substratum. These sites include tight junctions and zonulae adherens between epithelial cells, neuronal synapses and the advancing margins of fibroblasts crawling in tissue culture (Fig. 5). The latter cells also form attachment points, or adhesion plaques, to the culture substratum On more proximal regions of their lower surface. In the adhesion plaque, a bundle of parallel actin filaments lying close to the lower surface of the cell, known as a stress fibre, appears to terminate at a specialized region of the plasma membrane. Here the actin filaments are mechanically linked to the membrane and through it to components of the extracellular matrix, such as fibronectin. Several of the proteins involved in his linkage have been identified, such as vinculin, talin and integral membrane proteins involved in binding to fibronectin (see, e.g., Chen et al. 1985 ; Damsky et al. 1985 ; Mangeat & Burridge, 1984). At present, however, we do not have a detailed understanding of their arrangement within the adhesion plaque.

Fig. 5.

Transmission electron micrograph of a vertical section through the leading edge of a well-spread fibroblast. Note the thick dorsal and ventral cortical layers of aligned microfilaments. ×32000.

Fig. 5.

Transmission electron micrograph of a vertical section through the leading edge of a well-spread fibroblast. Note the thick dorsal and ventral cortical layers of aligned microfilaments. ×32000.

The extracellular matrix also affects the membrane-associated cytoskeleton in lens epithelial cells. In this case actin filaments lie in bundles approximately 0·2 μm thick parallel to the basal membrane creating the characteristically smooth and quiescent lower surface. Experiments with cultured epithelial cells have shown that the organization of this actin-rich layer depends on extracellular matrix components. The attachment of these actin bundles to the plasma membrane is less well characterized than in the case of the adhesion plaques or microvilli; however, an integral membrane heparan sulphate proteoglycan as well as specific receptors for laminin and collagen may be involved (Hay, 1985).

Away from these regions of specialized function the cortex of vertebrate tissue cells, if it exists at all, is certainly less conspicuous and harder to detect. Transmission electron microscopy sometimes shows a thin region of amorphous material from which other cytoplasmic organelles are excluded (Ben Ze’ev, Dueer, Solomon & Penman, 1979). Techniques such as fast-freeze, deep-etch electron microscopy sometimes reveal fine filaments extending between the inner surface of the plasma membrane and the lateral surfaces of actin filaments (Tilney, 1983; Hirokawa, 1982). It is possible that these crosslinks are easily lost in the course of preparation for electron microscopy and that they are more widely distributed than is generally recognized.

A thinner and more specialized type of membrane-associated cortex is found in circulating erythrocytes and may be an example of a structure common to all cells. The skin of the red blood cell consists of a bimolecular leaflet of lipid and a number of glycoproteins. Attached to its inner surface is a very thin layer of protein, perhaps 10 nm in thickness (Fig. 6). The make-up of this thin two-dimensional layer is fairly well understood (Bennett, 1985). A transmembrane protein, ‘band 3’, is bound to a cytoplasmic protein, ankyrin, which in turn is bound to spectrin. Spectrin itself is a long flexible protein, which forms a complex with another protein termed ‘band 4.1’, and short oligomeric lengths of actin filament stabilized by tropomyosin, thereby building an extended two-dimensional network (Byers & Branton, 1985). A notable feature of this arrangement is that actin is not attached directly to the membrane but through a series of intermediate molecules. Indeed this is also the case in microvilli and probably in stress fibres, and may be a general feature of the association of actin with membranes in vertebrate cells.

Fig. 6.

Schematic diagram of the two-dimensional network of spectrin and other proteins on the inner face of the plasma membrane of the mammalian red blood cell (see Byers & Branton, 1985).

Fig. 6.

Schematic diagram of the two-dimensional network of spectrin and other proteins on the inner face of the plasma membrane of the mammalian red blood cell (see Byers & Branton, 1985).

Spectrin-like and ankyrin-like proteins have been detected in most types of vertebrate cells (Bennett, 1985). They are closely similar to their erythrocyte counterparts and presumably can form a cortical network similar to that in the red blood cell. However, the situations are not precisely the same. At least in fibroblastic cells in culture, the spectrin family of proteins are not uniformly distributed around the entire cell cortex. They are excluded from certain regions of specialized function, such as adhesion plaques, and are detectable in the cytoplasm. While neurones appear to possess a relatively uniform peripheral layer of ankyrin and spectrin in association with actin, the particular isoforms of spectrin and ankyrin found in the axon differ from those in the perikarya and dendrites (Lazarides & Nelson, 1985). The significance of these regional differences and, indeed, the function of spectrin and ankyrin in non-erythroid cells is not known.

To summarize the points raised thus far: a membrane-associated cortex is a conspicuous feature of certain very large cells, such as giant freshwater amoebae and echinoderm eggs. In these cells the cortex consists of an actin-rich gel visible in the light microscope and mechanically distinct from the rest of the cytoplasm. In contrast, in vertebrate cells a uniform, mechanically strong cortex is only rarely found. A thin layer of this kind is evident in macrophages and neutrophils. But in most tissue cells the association between the plasma membrane and the cytoskeleton is variegated: conspicuous in certain specialized regions, such as in the core of surface extensions (microspikes and microvilli) and intercellular adhesions but weaker and transitory elsewhere. Even in regions of vertebrate tissue cells that appear to lack a well-defined cortex there may be a thin layer of spectrin and actin similar to the network underlying the membrane of mammalian red blood cells.

MECHANICAL PROPERTIES OF THE CELL CORTEX

To paraphrase a comment made by Wilhelm His a century ago, cell biology “cannot proceed independently of all reference to the general laws of matter -to the laws of physics and mechanics” (His, 1887). Certainly, until we know how cells produce and respond to physical forces we will not fully understand how they move and maintain their shape. But in turning to the mechanical properties of the cell surface we enter a grey area between whole cells and molecules in which we lack not only experimental data but also adequate concepts. Experimental difficulties exist because cells are small, and generally irregular and changing in shape. The interpretation of any parameters that one succeeds in measuring is often complicated by the variable contribution made by the cytoskeleton tightly associated with the plasma membrane and by internal contents, such as the nucleus and major organelles. Most fundamentally, it is extremely difficult to relate mechanical parameters such as moduli of elasticity and bending to molecular composition.

The mammalian red blood cell, in many ways ideal for such analysis, illustrates the state of the art. Many attempts have been made to give a mechanical explanation for its ability to adopt discoid, cup-shaped or echinocytic shapes under different conditions (see, e.g., Stokke, Mikkelson & Elgsaeter, 1986). But even in this stripped-down shell of a cell we cannot predict emergent shape or mechanical properties from the molecular details of the membrane.

But we must start somewhere. To simplify both the measurement of material properties and their interpretation in terms of cellular structures most experiments in ‘cell mechanics’ have used cells that are: (1) relatively large, (2) smooth-contoured, or (3) have a well-defined cortex (notably sea-urchin eggs and mammalian erythrocytes). They have also involved a number of ingenious devices by which small forces of known magnitude can be applied to the surfaces of cells and the resulting deflection measured. Three principal methods of applying force have been used. Those in which cells are: (1) squashed beneath a flat plate, (2) sucked in a small pipette, or (3) depressed by a small probe (Fig. 7).

Fig. 7.

Experimental approaches to the mechanical properties of the cell surface. The deformation produced in these experiments is a measure of the rigidity of the cell surface. A. The cell is squashed beneath a plate of known mass, such as a fragment of glass coverslip. B. A portion of the cell surface is sucked into the tip of a micropipette using a negative pressure of known value. C. The cell is indented by a flexible probe the bending of which indicates the force exerted.

Fig. 7.

Experimental approaches to the mechanical properties of the cell surface. The deformation produced in these experiments is a measure of the rigidity of the cell surface. A. The cell is squashed beneath a plate of known mass, such as a fragment of glass coverslip. B. A portion of the cell surface is sucked into the tip of a micropipette using a negative pressure of known value. C. The cell is indented by a flexible probe the bending of which indicates the force exerted.

The earliest estimates of the resistance to deformation of the cell surface, some of which go back to the nineteenth century, were based on the compression of seaurchin eggs by a constant weight. As reviewed by Hiramoto (1981), a cluster of eggs are sandwiched between the lower floor of a microscope chamber, on which they rest, and a glass plate is placed on their upper surface. The decrease in vertical diameter as the eggs are squashed then indicates the ease with which their surface can be deformed (the interpretation of such results is discussed below). A closely similar experimental method was recently used to follow changes in surface properties following fertilization (Schroeder, 1981). An important modification of this technique, enabling the weight applied to the cells to be varied was devised by Cole in 1932. He used a thin gold beam, 6 pm thick, 180 pm wide and 3 mm long, which could be lowered onto a cell. The force applied was then calculated from the bending of the beam.

A different apparatus was used by Mitchison & Swann (1954). Called a ‘cell elastimeter’, this caused a bulge to be sucked from the surface of an echinoderm egg with a small pipette. From the size of the bulge, and the negative pressure needed to produce it, a parameter termed the surface ‘stiffness’ could be estimated. Microaspiration experiments of this kind have been used extensively. They have been used to follow the post-fertilization changes in sea-urchin eggs mentioned above (Wolpert, 1966). More recently, they have been used to measure the increase in the surface tension of erythrocytes following exposure to lectins (Evans & Leung, 1984; Smith & Hochmuth, 1982).

By far the most delicate and precise instrument used to date to investigate cell surface mechanics is that devised by Elson and colleagues (see Pasternak & Elson, 1985). The cell is depressed locally by a small glass probe mounted on a flexible steel wire or (more recently) on a glass beam. The wire is moved through a cyclic waveform by a piezoelectric motor and optical sensors monitor the deflections of the tip motor and of the probe. Since the tip of the probe is only 2 pm in diameter, and displacements of under 04 pm can be detected, the apparatus is able to measure the properties of very small cells, such as lymphocytes. It is even able to compare values in different regions of the same cell (Petersen, McConnaughey & Elson, 1982).

The raw data obtained in the experiments described above consist in every case of a distance, or ‘strain’ (depression of the surface or size of a bulge in a pipette or change in cell diameter), produced by a force, or ‘stress’, of known magnitude. For many purposes this empirical relationship between stress and strain is sufficient, e.g. when the properties of different kinds of cells, or of the same cell under different conditions, are to be compared. Calculation of intrinsic elastic moduli from such data is more complicated. It depends on the detailed geometry of the apparatus and of the cell; also on the time taken in the measurement, since the cell surface, in common with many biological materials, shows viscoelastic behaviour. Analysis, therefore, requires a number of simplifying assumptions.

The first assumption, without which analysis would indeed be difficult, is that the mechanical properties of the cell are dominated by its outer ‘skin’, that is, by the plasma membrane and its tightly associated cortex. As we have seen, this is probably true for echinoderm eggs and mammalian erythrocytes. Resistance to deformation in such a cell then arises from the flexural rigidity of the cortex and from its resistance to stretching (see, e.g., Taber, 1983). In a similar way, a car tyre resists deformation because it is made of thick rubber, which is difficult to bend, and because inflation with air has generated tension in the surface, which resists further areal expansion. Mitchison and Swann, in the experiments already mentioned, made an attempt to distinguish between the contributions of these two components, reaching the conclusion that flexural rigidity made a major contribution to the observed stiffness. However, this conclusion has been criticized on theoretical grounds (Wolpert, 1966; Yoneda, 1973), and other investigators have also assumed that the resistance to bending forces is negligible.

The latter view, at least, has the virtue that it makes analysis simpler. If resistance to bending is negligible, then the surface becomes a ‘thin shell’ in mechanical terms: one in which the only important forces are those in the plane of the membrane (Calladme, 1983). Two kinds of surface tension are then relevant: interfacial tension, arising from the phase boundary between the hydrophobic membrane components and their water-based surroundings, and the elasticity of the plasma membrane and its associated cortex.

The difference between these two types of surface tension is illustrated by two physical models of a flexible shell or skin : that of a soap bubble and that of a rubber balloon. In the soap bubble, tension in the skin and, hence, the inward pressure it exerts on the enclosed volume of air are direct consequences of the interfacial surface tension between soap solution and air. So long as an adequate reserve supply of soap solution exists, then the interfacial surface tension will be the same at all regions of the bubble surface independently of its size. The surface tension of a rubber balloon, in contrast, is a measure of the Hookean elasticity of the rubber of which the skin is made. As the balloon is inflated, the rubber stretches and the tension it generates increases.

Returning to the surface of the cell, we know that this must possess an areaindependent interfacial surface tension similar to that seen in a soap bubble. (One might, for example, draw parallels between the incorporation of new membrane into the surface by exocytosis to the recruitment of fresh soap solution into the skin of the soap bubble.) Yet the contribution of this interfacial tension to the overall tension of the cell is probably minor. In the original study by Cole (1932), the response of the cell to compression appeared to be dominated by a large elastic component. Even though this viewpoint has been questioned (Yoneda, 1973), most subsequent investigators have, similarly, agreed that elasticity (or more accurately viscoelasticity) is the most important characteristic of the cell surface (Evans & Hochmuth, 1978).

It will be apparent from the foregoing that we are at present a long way from making secure and objective measurements of the moduli of elasticity and bending of the cell surface. Similarly, it is not yet possible to compare, rigorously, measurements made on different types of cells with different types of apparatus. However, measurements that have been made reveal such large variations in mechanical properties that we have been tempted to make an order-of-magnitude comparison. We have therefore listed in Table 1 approximate values of ‘cell stiffness’ based on the forces found necessary to deform the surface of the cell by 1 μm, regardless of the methods used, and ignoring questions of the origins of the resistance to deformation.

Table 1.

Mechanical ‘stiffness’ of the animal cell surface

Mechanical ‘stiffness’ of the animal cell surface
Mechanical ‘stiffness’ of the animal cell surface

Conspicuous in this table is the very small, essentially unmeasurable, value for a bimolecular leaflet of phospholipid compared to that at the surface of real cells. The lipid component of the plasma membrane seems to be of negligible strength and, by itself, unlikely to provide the motive force for cell movements in amoebae or fibroblasts (although such a role has been proposed; Bretscher, 1984). It should be noted, however, that plasma membranes also contain a variable complement of integral proteins, and where these reach high concentrations, or even form crystalline arrays as in gap junctions, the plasma membrane may contribute to cell surface stiffness.

A second salient feature in Table 1 is the large difference between the values for the red blood cell and a fibroblast. Evidently the spectrin-rich network associated with the plasma membrane is much weaker than a fully developed actin-rich cortex. Indeed, it may be weak enough to be influenced by the lipid bilayer, and there is evidence that the shape of the mammalian erythrocyte is affected by the composition of the phospholipid bilayer (Kuypers et al. 1984). Finally, it may be noted in Table 1 that, as expected, surface stiffness increases with the thickness of the cortical cytoskeleton.

Possibly the most interesting recent findings in cell mechanics concern changes in the surface stiffness of a cell with physiological activities. In the sophisticated measurements made by Elson and associates referred to above, the properties of lymphocytes were found to change in response to various kinds of treatment. Exposure of the cell to lectins or antibodies caused aggregation of receptors on the cell surface and produced a large increase in surface tension and decrease in receptor mobility (Edelman, 1976). Both responses occurred even if crosslinking was confined to one small region of the cell surface, so that it was not considered due simply to the formation of a rigid ‘crust’ on the cell surface. Here, as in studies of the fertilization of sea-urchin eggs, the membrane cortex seems capable of a profound reorganization, almost a change of state.

Besides their relevance to lymphocyte capping, phenomena of this kind are of great interest from the point of view of the response of cells to solid surfaces (which can act, in a sense, rather like crosslinking agents). A change in the cortex in response to contact with a surface is likely to be of major importance in cell locomotion and possibly also in the substrate dependence of cell growth shown by most freshly explanted animal tissue cells.

The surface stiffness of sea-urchin eggs shows interesting temporal and regional changes following fertilization. There is initially a global increase in tension (which may be related to the tendency of even small vertebrate cells to round-up before division) followed by focal relaxation at the two opposite poles. In the circumferential band midway between the poles, at the site of the forming contractile ring, surface stiffness or tension remains high throughout the second phase of general relaxation.

White & Borisy (1983) have discussed these changes of surface tension in a theoretical paper, using them as the basis of a model for cytokinesis. In their hypothesis, the surface tension is produced by linear contractile elements in the plane of the surface -a metaphor for the membrane-associated actin-rich cortex. As changes in surface tension occur (mediated, in the White—Borisy model, by influences spreading from the two mitotic spindles) the linear contractile elements move from regions of lower to regions of higher surface tension, changing as they do so from a random orientation to alignment with the equator of the cell. A suitable physical analogy might be that of an elastic net stocking drawn tightly around the cell, which is loosened at either pole by cutting some of the threads. The tension of the remaining network will automatically draw the remaining strands together around the middle of the cell while at the same time changing their orientation until they are circumferentially aligned (Fig. 8).

Fig. 8.

Mechanical analogy for the changes in surface tension accompanying cytokinesis. The ‘cell’ on the left is enclosed in a network of elastic threads. Cutting some of these threads in the polar regions causes the remainder of the network to migrate to the equatorial region (right). Note that the threads will also tend to adopt an equatorial orientation.

Fig. 8.

Mechanical analogy for the changes in surface tension accompanying cytokinesis. The ‘cell’ on the left is enclosed in a network of elastic threads. Cutting some of these threads in the polar regions causes the remainder of the network to migrate to the equatorial region (right). Note that the threads will also tend to adopt an equatorial orientation.

Although a sea-urchin egg is much larger than a vertebrate cell, and designed to survive in a very different environment, we suspect that the changes described above may be of widespread significance. Certainly, cytokinesis follows a similar course in most kinds of animal cells, so that even if we cannot at present measure such smallscale changes, it is not unreasonable to think that they exist. Moreover, the concept of regional changes in surface tension produced by the internal cytoskeleton, and especially microtubules, seems to fit several kinds of vertebrate cell movement.

Take fibroblast migration as an example. At the leading edge of the cell a highly motile flexible surface produces lamellipodia and microspikes while at the sides of the cell the surface is quiescent and smoothly rounded. As noted by Vasiliev (1982), these regional differences correspond to the distribution of microtubules, which in general tend to lie parallel to the long axis of the cell ; and regional differences are lost when the microtubules break down in response to colchicine treatment. Furthermore, microspikes and lamellipodia, as well as particles on the cell surface and submembrane assemblies of actin filaments known as arcs (Heath, 1983), all flow backwards from the leading edge of the cell to the smoother more quiescent regions. Evidently, cycles of actin-containing structures occur in such cells, probably carrying with them other membrane components.

It seems reasonable to us to see this rearward flow of surface material as arising from regional variations in surface tension in the membrane and associated cytoskeleton. The movements on the surface of a fibroblast would then conform to a pattern similar to that described above for a dividing echinoderm egg with certain modifications (for example, it must involve compensatory flow of membrane and cytoskeletal components through the cytoplasm to the advancing ‘low tension’ end of the cell). A very similar picture can be painted for the surface movements that occur on the growth cone at the advancing tip of a growing nerve axon (Bray & Chapman, 1985). In this case the surface at the actively moving (putatively low-tension) region of the growth cone moves back to the axonal cylinder, where it appears to be stabilized by interactions with the longitudinally aligned microtubules and neurofilaments.

We thank Elliot Elson and Tom Pollard for their helpful comments on this article.

Allen
,
R. D.
,
Cooledge
,
J. W.
&
Hall
,
P. J.
(
1960
).
Streaming in cytoplasm dissociated from the giant amoeba, Chaos chaos
.
Nature, Land
.
187
,
896
899
.
Allen
,
R. D.
&
Kamiya
,
N.
(
1964
).
Primitive Motile Sy stems in Cell Biology
.
New York, London
:
Academic Press
.
Apgar
,
J. R.
,
Herrmann
,
S. H.
,
Robinson
,
J. M.
&
Mescher
,
M. F.
(
1985
).
Triton X-100 extraction of P185 humour cells: evidence for a plasma membrane skeleton structure
.
J. Cell Biol
.
100
,
1369
1378
.
Bennett
,
H.
&
Condeelis
,
J.
(
1984
).
Decoration with myosin subframent-1 disrupts contacts between microfilaments and the cell membrane in isolated Dictyostelium cortices
.
J. Cell Biol
.
99
,
1434
1440
.
Bennett
,
V.
(
1985
).
The membrane skeleton of human erythrocytes and its implication for more complex cells. A
.
Rev. Biochem
.
54
,
273
304
.
Ben Ze’ev
,
A.
,
Dueer
,
N.
,
Solomon
,
F.
&
Penman
,
S.
(
1979
).
The outer boundary of the cytoskeleton: Lamina derived from plasma membrane probes
.
Cell
17
,
859
865
.
Bray
,
D.
&
Chapman
,
K.
(
1985
).
Analysis of microspike movements on the neuronal growth cone
.
J. Neurosci
.
5
,
3204
3213
.
Bretscher
,
M. S.
(
1984
).
Endocyosis: relation to capping and cell locomotion
.
Science
22A
,
681
686
.
Byers
,
T. J.
&
Branton
,
D.
(
1985
).
Visualization of the protein associations in the erythrocyte membrane skeleton
.
Proc. natn. Acad. Sci. U.S A
.
82
,
6153
6157
.
Calladine
,
C. R.
(
1983
).
Theory of Shell Structures
.
Cambridge University Press
.
Chambers
,
R.
&
Chambers
,
E. L.
(
1961
).
Explorations in to the Nature of the Living Cell
.
Cambridge, Mass
. :
Harvard University Press
.
Chen
,
W. T.
,
Hasegawa
,
E.
,
Hasegawa
,
T.
,
Weinstock
,
C.
&
Yamada
,
K. M.
(
1985
).
Development of cell surface linkage complexes in cultured fibroblasts
.
J. Cell Biol
.
100
,
1103
1114
.
Cole
,
K. S.
(
1932
).
Surface forces of Arbacia eggs
.
J. cell. comp. Physiol
.
1
,
1
19
.
Comly
,
L. T.
(
1973
).
Microfilaments in Chaos carolinensis’. Membrane association, distribution and heavy meromyosin binding in the glycerinated cell
.
Cell Biol
.
58
,
230
237
.
Damsky
,
C. H.
,
Knudsen
,
K. A.
,
Bradley
,
D.
,
Buck
,
C. A.
&
Horwitz
,
A. F.
(
1985
).
Distribution of cell substratum attachment (CSAT) antigen on myogenic and fibroblastic cells in culture
,
J. Cell Biol
.
100
,
1528
1539
.
Edelman
,
G. H.
(
1976
).
Surface modulation in cell recognition and cell growth
.
Science
192
,
218
226
.
Evans
,
E. A.
(
1973
).
New membrane concept applied to the analysis of fluid shear- and micropipette-deformed red blood cells
.
Biophys. J
.
13
,
941
954
.
Evans
,
E. A.
&
Hochmuth
,
R. M.
(
1978
).
Mechanochemical properties of membranes
.
Int. Rev. Cytol
.
10
,
1
64
.
Evans
,
E. A.
&
Leung
,
A.
(
1984
).
Adhesivity and rigidity of erythrocyte membrane in relation to wheat germ agglutinin binding
.
J. Cell Biol
.
98
,
1201
1208
.
Geiger
,
B.
(
1983
).
Membrane-cytoskeleton interaction
.
Biochim. biophys. Acta
737
,
305
341
.
Goodloe-Holland
,
C. M.
&
Luna
,
E. J.
(
1984
).
A membrane cytoskeleton from Dictyostelium discoideum. III. Plasma membrane fragments bind predominantly to the sides of actin filaments
.
J. Cell Biol
.
99
,
71
78
.
Hartwig
,
J. H.
,
Niederman
,
R.
&
Lind
,
S. E.
(
1985
).
Cortical actin structures and their relationship to mammalian cell movements
.
In Subcellular Biochemistry
(ed.
D. B.
Roodyn
), vol.
11
, pp.
1
49
.
New York
:
Plenum Press
.
Hay
,
E. D.
(
1985
).
Matrix-cytoskeletal interactions in the developing eye
.
J. cell. Biochem
.
27
,
143
156
.
Heath
,
J. P.
(
1983
).
Behaviour and structure of the leading lamella in moving fibroblasts. I. Occurrence and centripetal movement of arc-shaped microfilament bundles beneath the dorsal cell surface
.
J. Cell Sci
.
60
,
331
354
.
Hiramoto
,
Y.
(
1957
).
The thickness of the cortex and the refractive index of the protoplasm in sea urchin eggs
.
Embryologia
3
,
361
374
.
Hiramoto
,
Y.
(
1981
).
Mechanical properties of dividing eggs
.
In Mitosis/Cytokinesis
(ed.
A. M.
Zimmerman
&
A.
Forer
), pp.
398
418
.
New York
:
Academic Press
.
Hirokawa
,
N.
(
1982
).
Cross-linker system between neurofilaments, microtubules and membranous organelles in frog axons revealed by the quick-freeze, deep-etching method
.
J. Cell Biol
.
94
,
129
142
.
His
,
W.
(
1887
-88).
On the principles of animal morphology
.
Proc. R. Soc. Edinb
.
15
,
287
298
.
Jeon
,
K. W.
, editor
(
1973
).
The Biology of Amoeba
.
New York, London
:
Academic Press
.
Kuypers
,
F. A.
,
Roelofsen
,
B.
,
Berendsen
,
W.
,
Op Den Kamp
,
J. A. F.
&
Van Deenen
,
L. L. M.
(
1984
).
J. Cell Biol
.
99
,
2260
2267
.
Lazarides
,
E.
&
Nelson
,
W. J.
(
1985
).
Expression and assembly of the erythroid membraneskeleton proteins ankyrin (globin) and spectrin in the morphogenesis of chicken neurons
.
J. cell. Biochem
.
27
,
423
441
.
Lewis
,
W. H.
(
1939
).
The role of a superficial plasmagel layer in changes of form, locomotion and division of cells in tissue cultures
.
Arch. exp. Zellforsch
.
23
,
1
7
.
Loor
,
F.
(
1981
).
Cell surface-cell cortex transmembranous interactions with special reference to lymphocyte functions
.
Cell Surf. Rev
.
7
,
253
335
.
Luna
,
E. J.
,
Goodloe-Holland
,
C. M.
&
Ingalls
,
H. M.
(
1984
).
A membrane cytoskeleton irom Dictyostelium discoideium. II. Integral proteins mediate the binding of plasma membranes to F-actin affinity beads
.
J. Cell Biol
.
99
,
58
70
.
Mangeat
,
P.
&
Burridge
,
K.
(
1984
).
Actin-membrane interaction in fibroblasts: What proteins are involved in this association?
J. Cell Biol
.
99
,
955
1035
.
Mast
,
S. O.
(
1926
).
Structure, movement, locomotion and stimulation of amoeba
.
J. Morph. Physiol
.
41
,
347
425
.
Mitchison
,
J. M.
&
Swann
,
M. M.
(
1954
).
The mechanical properties of the cell surface. I. The cell elastimeter
.
J. exp. Biol
.
31
,
443
460
.
Mooseker
,
M. S.
,
Pollard
,
T. D.
&
Wharton
,
K. A.
(
1982
).
Nucleated polymerisation of actin from the membrane-associated ends of microvillar filaments in the intestinal brush border
.
J. Cell Biol
.
95
,
223
233
.
Moss
,
D. J.
(
1983
).
Cytoskeleton-associated glycoproteins from chicken sympathetic neurons and chicken embryo brain
.
Eur.J. Biochem
.
135
,
291
297
.
Oliver
,
J. M.
&
Berlin
,
R. D.
(
1982
).
Mechanisms that regulate the structural and functional architecture of cell surfaces
.
Int. Rev. Cytol
.
74
,
55
94
.
Pasternak
,
C.
&
Elson
,
E. L.
(
1985
).
Lymphocyte mechanical response triggered by crosslinking surface receptors
.
J. Cell Biol
.
100
,
860
872
.
Peterson
,
N. O.
,
Mcconnaughey
,
W. B.
&
Elson
,
E. L.
(
1982
).
Dependence of locally measured cellular deformability on position on the cell, temperature and cytochalasin B
.
Proc, natn. Acad. Sci. U.SA
.
79
,
5327
5331
.
Pollard
,
T. D.
&
Ito
,
S.
(
1970
).
Cytoplasmic filaments of Amoeba proteus. I. The role of filaments in consistency changes and movement
.
J. Cell Biol
.
46
,
267
289
.
Pollard
,
T. D.
&
Korn
,
E. D.
(
1973
).
Electron microscopic identification of actin associated with isolated amoeba plasma membranes
.
J. biol. Chem
.
248
,
448
450
.
Rogalski
,
A. A.
&
Singer
,
S. J.
(
1985
).
An integral glycoprotein associated with the membrane attachment sites of actin microfilaments
.
J. Cell Biol
.
101
,
785
801
.
Schroeder
,
T. E.
(
1981
).
The origin of cleavage forces in dividing eggs. A mechanism in two steps
.
Expl Cell Res
.
134
,
231
240
.
Schroeder
,
T. E.
&
Stricker
,
S. A.
(
1983
).
Morphological changes during maturation of starfish oocytes: surface ultrastructure and cortical actin
.
Devi Biol
.
98
,
373
384
.
Smith
,
L.
&
Hochmuth
,
R. M.
(
1982
).
Effect of wheat germ agglutinin on the viscoelastic properties of erythrocyte membrane
.
J. Cell Biol
.
94
,
7
11
.
Spudich
,
A.
&
Spudich
,
J. A.
(
1979
).
Actin in Triton-treated cortical preparations of unfertilised and fertilised sea urchin eggs
.
J. Cell Biol
.
82
,
212
226
.
Stratford
,
C. A.
&
Brown
,
S. S.
(
1984
).
Isolation of an actin-binding protein from membranes of Dictyostelium discoideum
.
J. Cell Biol
.
100
,
727
735
.
Stokke
,
B. T.
,
Mikkelsen
,
A.
&
Elgsaeter
,
A.
(
1986
).
The human erythrocyte membrane skeleton may be an ionic gel
.
Eur. biophys. J
. (in press).
Stossel
,
T. P.
,
Hartwig
,
J. H.
&
Yin
,
H. L.
(
1981
).
Actin gelation and the structure of and movement of cortical cytoplasm
.
In Cytoskeletal Elements and Plasma Membrane Organization. Cell Surface Rev
., vol.
7
, pp.
138
168
.
Amsterdam
:
North-Holland
.
Taber
,
L. A.
(
1983
).
Compression of fluid-filled spherical shells by rigid indenters
.
J. appl. Meeh
.
50
,
717
722
.
Taylor
,
D. L.
&
Condeelis
,
J. S.
(
1979
).
Cytoplasmic structure and contractility
.
Int. Rev. Cytol
.
56
,
57
144
.
Tilney
,
L. G.
(
1983
).
Interactions between actin filaments and membranes give spatial organization to cells
.
Modern Cell Biol
.
2
,
163
199
.
Tilney
,
L. G.
,
Bonder
,
E. M.
&
Derosier
,
D. J.
(
1981
).
Actin filaments elongate from their membrane associated ends
.
J. Cell Biol
.
90
,
485
494
.
Tilney
,
L. G.
&
InouÉ
,
S.
(
1982
).
The acrosomal reaction of Thyone sperm. II. The kinetics and possible mechanism of acrosomal process elongation
.
J. Cell Biol
.
93
,
820
827
.
Vasiliev
,
J. M.
(
1982
).
Spreading and locomotion of tissue cells: factors controlling the distribution of pseudopodia
.
Phil. Trans. R. Soc. Lond. B
299
,
159
167
.
Weeds
,
A.
(
1982
).
Actin-binding proteins: regulators of cell architecture and motility
.
Nature, Lond
.
296
,
811
816
.
White
,
J. G.
&
Borisy
,
G. G.
(
1983
).
On the mechanism of cytokinesis in animal cells
.
J. theor. Biol
.
101
,
289
316
.
Wolpert
,
L.
(
1966
).
The mechanical properties of the membrane of the sea urchin egg during cleavage
.
Expl Cell Res
.
41
,
385
396
.
Yoneda
,
M.
(
1973
).
Tension at the surface of sea urchin eggs on the basis of ‘liquid drop’ concept
.
Adv. Biophys
.
4
,
153
190
.