plasma membrane-associated filament systems in

J. Cell Sci. 64, 351-364 (1983) 351Printed in Great Britain © The Company of Biologists Limited 1983

PLASMA MEMBRANE-ASSOCIATED FILAMENTSYSTEMS IN CULTURED CELLS VISUALIZED BYDRY-CLEAVING

DICK A. M. MESLAND* AND HERMINA SPIELEDivision of Cell Biology, The Netherlands Cancer Institute, Plesmanlaan 121,1066 CX Amsterdam, The Netherlands

SUMMARY

Substrate-attached critical-point-dried cells cleave along the level of the substrate-adherent mem-brane if removed by means of adhesive tape. The remaining membrane fragments on grids can bevisualized three-dimensionally by means of stereo transmission electron microscopy. Attachment ofcells may be achieved by active spreading of the cell, or artificially by poly-L-lysine adherence of pre-fixed cells. In 11 different cell types a filamentous network appears to remain associated with thecytoplasmic face of the membrane. In one hepatoma cell type virtually no filamentous network couldbe detected. Two general network morphologies are described: the hepatocytic network and thelymphoid network. Since no correspondence could be found between cytoplasmic structure and thestructure of the membrane-associated network, and since cells generally cleave along the level of thisnetwork, excluding cell organelles, we conclude that it comprises a distinct structural system,analogous to the membrane skeleton of the red cell membrane.

INTRODUCTION

To understand the various properties of the plasma membrane it has been suggestedthat current concepts of membrane structure, envisaged as a matrix of phospholipidsin which proteins are embedded (Singer & Nicolson, 1972), have to incorporate asystem that somehow regulates its overall shape and its dynamic behaviour(Kirkpatrick, 1979), such as the spectrin/actin system shown to exist in erythrocytes(recent reviews: Gratzer, 1981; Branton, Cohen & Tyler, 1981). This particularsystem is organized as a two-dimensional network on the cytoplasmic face of themembrane (Lang, Nermut & Williams, 1981) and linked to it by connections with itspreponderant integral protein. It controls the red cell's elasticity, its shape and thedistribution of transmembrane glycoproteins (Nicolson & Painter, 1973). At present,however, there is only limited evidence for the existence in other cells of membrane-associated filament networks analogous to those of the red cell. Despite studies thatinfer such a structure (Ben Ze'ev, Duerr, Solomon & Penman, 1979; Mescher, Jose& Balk, 1981; Ishikawa, Tsukita & Tsukita, 1982; Luna et al. 1981), directdemonstration can be obtained only by the development of ultrastructural techniques,which are able to reveal the cytoplasmic face of the plasma membrane (Boyles &Bainton, 1979, 1981; Lang etal. 1981).

Recently, we introduced dry-cleaving as a new technique to visualize this side of the

•Present address: Division TB/Biorack, ESTEC, Noordwijk, The Netherlands.

352 D. A. M. Mesland and H. Spiele

membrane in primary-cultured hepatocytes (Mesland, Spiele & Roos, 1981a). Itmakes use of the operational circumstance that critical-point-dried cells tend to breakeasily along their substrate-adherent membrane. Thus, hepatocytes, spread onFormvar-coated grids, fixed in situ and critical-point-dried, can be torn from thesubstrate by means of adhesive tape, which causes the adherent membranes of thecells to detach and be left on the grid. Combined with stereo microscopy theprocedure produced structural evidence for the existence of a distinct membrane-associated network in these cells. The network appears to be dissociable from themembrane by treatment of cells with cytochalasin B (Mesland, Los & Spiele, 19816)and structural features'suggested that it might function in the formation of coatedvesicles (Mesland et al. 1981a).

We have now extended the scope of the technique to suspended cells as well.Attachment of pre-fixed cells to poly-L-lysine-coated surfaces (Sanders, Alexander &Braylan, 1975), and subsequent dry-cleaving, allows the visualization of the cytoplas-mic face of the plasma membrane in many different cells. The results produce the firstdemonstration of a distinct membrane-associated filamentous web in a variety ofnormal and tumour cells.

MATERIALS AND METHODS

Cell culturesRat hepatocytes were isolated and cultured in plastic Petriperm dishes or on grids as described

previously (Roos, Vande Pavert &Middelkoop, 1981; Meslande/n/. 1981a). Normal rat BRL cells,rat Reuber/H35 hepatoma cells, rat HTC hepatoma cells (all generously given by Dr R. van Wijk,University of Utrecht) and mouse M143066 hepatoma cells were cultured in Falcon flasks contain-ing 25 ml of standard medium (a 1: 1 mixture of Dulbecco's modification of Eagle's medium andHAM's F-12 medium (Flow) supplemented with 10% new-born-calf serum (Gibco), 20mM-HEPES, 4mM-glutamine and antibiotics), in a 5 % CO2/air mixture at 37 °C. Of these, M140366cells grow in suspension, the other cells attach and spread. For whole-mount electron microscopy(WMTEM) cells (4 X 106/3ml) were grown to or near to confluency on 200 mesh nickel gridssandwiched between carbon-coated Formvar and glass coverslips (8 mm X 26 mm), u.v.-sterilized,in 35 mm Falcon cell culture dishes.

After routine trypsination cells could be kept in suspension in 50 ml Falcon conical tubes contain-ing about 1 -5 X 10' cells in 10 ml standard medium bubbled with a 95 % O2/5 % CO2 mixture for1 h at 37 °C. When measured by Trypan Blue exclusion, about 95 % of viable cells were generallyobtained. A fraction of rat liver endothelial cells was isolated with the hepatocytes (Roose/ al. 1981)and kept in suspension as described above.

Mouse MB6A lymphosarcoma cells, mouse mammary carcinoma cells TA3/St and TA3/Hawere grown in ascites as described by Roos, Dingemans, Van de Pavert & Van de Bergh-Weerman(1977, 1978). Mouse 3T3 cells were cultured on glass coverslips or on grids as described by Tem-mink & Spiele (1980), except that nickel grids were used instead of gold grids.

Human small thymocytes (a gift from Dr F. A. Vyth-Dreese of this Institute) were isolated bycentrifugal elutriation as described by Figdor el al. (1982).

Dictyostelium discoideum cells were a generous gift from Dr R. van Driel (University of Amster-dam). They were cultured as described by Van Driel (1981) and spread on poly-L-lysine on carbonover Formvar-coated grids for 1 h.

Attachment of suspended cellsCells in suspension were pelleted and resuspended in 0-1 % glutaraldehyde (GA) in HMK buffer

(0-1 M-KCI, 3 mM-MgClz, 10mM-HEPES, pH6-9) at room temperature. Routinely, fixation lasts

Plasma membrane-associated filaments 353

for IS min only, but cells could be kept in fixative at 4°C for weeks without detectable structuralchanges. After fixation cells were rinsed in HMK and allowed to settle on poly-L-lysine-coatedcarbon over Formvar films (1 mg/ml poly-L-lysine for 1 h at room temperature, followed by severalrinses with HMK), supported by nickel grids or by glass coverslips, for about 1 h at 4CC (Sanderset al. 1975). Near to complete coverage of the surface by the cells should be obtained, since thisgreatly enhances the success of cleaving.

Cell preparation

Spread cells or prefixed, suspended cells attached to poly-L-lysine were fixed in 0-l % GA for15 min, washed, treated with 1 % tannic acid (Malinckrodt Inc., St Louis, Mo.) for 30 min, andpost-fixed with 0-5 % OsO4 for 30 min, all in HMK buffer. The cells were then washed with distilledwater and stained with 1 % uranyl acetate in water for another 30 min, followed by stepwise dehydra-tion in ethanol; all treatments were carried out at room temperature. Cells were critical-point-driedfrom liquid CO2 • From this step on they were either dry-cleaved, prepared for scanning electronmicroscopy (SEM), or embedded in Epon, sectioned and post-stained with lead citrate for standardelectron microscopy (EM).

Dry-cleavingCleavage of critical-point-dried cells on grids was performed by putting the grid, cell-side down,

onto the adhesive side of a clean piece of Scotch acetate tape (no. 810) fixed on an object glass slideby means of Avery double-sided self-adhesive foto-split paper, pressing it very gently with forceps(while watching with a binocular microscope) and subsequently removing it.

To dry-cleave cells on carbon-coated glass coverslips, pieces of approximately 3 mm X 15 mmwere cut from a double layer of Scotch acetate tape (no. 810) and clinical Leukosilk tape stuck tothe former's non-adhesive side. This produces relatively stiff, yet flexible, pieces that were put ontothe cell layer and removed again, producing a zone of cleaved cells between whole, uncleaved cells.Cells were subsequently embedded in Epon and sectioned perpendicular to the substrate forexamination of the dimensions of material left after cleaving.

Surface-etchingMouse 3T3 cells grown on glass coverslips were fixed in 0-1 % GA in HMK for 4 min, washed

in HMK (once for 5 min), in 501T1M-NH4CI in HMK (twice for 15 min) and in HMK again (twicefor 5 min). They were then extracted with 1 % Triton X-100 in HMK for 30min, washed in HMK(3 times for 15 min) and postfixed in 0-1 % GA overnight, all at room temperature. Cells weredehydrated and critical-point-dried as described above. For transmission electron microscopical(TEM) observation of the etched surface, replicas were made by a modification of the methodintroduced by Heuser & Salpeter (1979). Dried cells were sputtered with gold/palladium in aPolaron SEM coating unit at l-4kV and 30 mA for 40 s, which deposits a uniform metal layer ofapprox. 30—50 A, followed by rotary deposition of carbon at an angle of 80 ° for 5 s in a Balzers f reeze-etch apparatus. Replicas were floated off on HF, washed in distilled water, and cell remnants weredigested in 15 % sodium hypochlorite for 1 h, washed several times in distilled water and picked upon Formvar-coated copper grids. Micrographs were reverse-printed.

Scanning electron microscopyCritical-point-dried cells were prepared for SEM as described previously (Meslandef al. 1981a).

Stereo transmission electron microscopy

Whole mount dry-cleaved preparations and replicas were observed in a Philips EM300 electronmicroscope at lOOkV acceleration voltage. Stereo micrographs were taken with a tilt angle of 6°.

Morphometric measurements

The total length of the microtubules and the total membrane surface per micrograph weremeasured with a Leitz semi-automatic image-analysis system.


RESULTS AND DISCUSSION

Fig. 1 shows the dry-cleaving procedure applied to human small thymocytes. Theadherence of pre-fixed cells to poly-L-lysine is shown by SEM in Fig. 1A and in cross-section of Epon-embedded critical-point-dried material in Fig. 1B. This figuredemonstrates the small flat attachment zone between plasma membrane and sub-strate, as well as the normal overall morphology of this previously dried cell. Cross-sections made perpendicular to the adherent membrane and associated structures, leftafter dry-cleaving, show a layer about 50 nm thick on the carbon film (Fig. lc). Fig.ID gives a low-magnification view of the adherent membranes left on grids asvisualized directly in TEM after dry-cleaving of thymocytes. This image, comparedwith the uncleaved cells in Fig. 1A, illustrates the efficiency of the procedure: virtuallyall cells are stripped of their adherent membranes, with consequent exposure of theircytoplasmic face. Except for cells densely covered with microvilli, this result wasobtained with all suspensions of cell types studied so far. In addition, since cellsadhere to the poly-L-lysine in a random fashion, as was shown by SEM of cells withpolarized morphology (Fig. 1E), membrane fragments of random areas of the surfacesof the cell are visualized by the technique. Thus, given a sufficient number of frag-ments, a reliable picture of the whole plasma membrane can be obtained.

Table 1 lists 12 cell types in which we have observed the plasma membrane'scytoplasmic face by dry-cleaving; some of them are in both a spread and a suspendedstate. Except for one case (Fig. 4, see below), all have a membrane-associated filamen-tous network, the morphology of which is rather complex and may differ from one celltype to another, with respect to filamentous density and type, and the occurrence ofother morphological features. For example, the presence of microvilli affects the localdensity of filaments, and the occurrence of coated pits and vesicles produces differentpatterns in the network (Mesland et al. 1981a). Detailed analysis of the compositionof the network must await combined dry-cleaving and immunolabelling techniques.However, from all the structural observations made, two main types of networkmorphology can be distinguished. The first, which we term the hepatocytic network,consists of a rather dense system of anastomosing filaments, as described before(Mesland et al. 1981a). Different classes of filaments, varying in diameter from about2 nm to 7 nm and often having a knobby appearance are observed in this type of

Fig. 1. Dry-cleaving of suspended cells illustrated with human small thymocytes. A.Scanning electron micrograph of dried cells. X5000. B. Dried cells embedded in Epon andsectioned perpendicular to the poly-L-lysine-coated substrate (arrow). Note the flat attach-ment zone of plasma membrane and substrate, and the normal EM image of the previouslydried cells. X9700. c. A similar preparation as in B, made after removal of the cells witha piece of Scotch adhesive tape (dry-cleaving, see Materials and Methods). It shows a thinlayer of membrane and associated structures left on the substrate. X50000. D. Low-magnification view of substrate-adherent membranes of cells dry-cleaved on grids,visuali2ed directly in TEM. All cells are stripped of their membranes, showing theircytoplasmic faces (compare with A and B). X9700. E. Uropod-containing (arrows) mouselymphosarcoma cells that demonstrate the random positions taken by cells attached to apoly-L-lysine-coated surface. X2000.


J&

0m

Fig. 1


network (Figs 2, 3, 6). The second type is the lymphoid network, which actuallyconsists of two types that may even occur simultaneously in one cell: one constitutesa system of clearly distinct, smooth filaments associated with 'spongy' conglomera-tions spread on the membrane (Figs 5, 7); the other lacks the spongy conglomerationsand is generally denser and more reminiscent of the hepatocytic network (Fig. 8). Thecircumstances that influence the occurrence of spongy conglomerations as well asvariations in filament density in human thymocytes are currently being studied in thislaboratory. Three classes of filaments, with diameters of 2-3 nm (similar to thosedescribed by Schliwa & Blerkum, 1981), 4nm and 6-7 nm, as well as an occasionalmicrotubule, can be readily distinguished in the lymphoid network (Fig. 8). Alsoribosomes are observed associated with the latter two classes of filaments. Cells havebeen classified according to their network morphologies in Table 1.

Cells that tend to flatten extremely, such as mouse 3T3 cells and rat BRL cells, donot cleave as uniformly as thicker cells. This may obscure ultrastructural detail dueto superimposition, but can also produce striking images of rough endoplasmicreticulum with its ribosomes intimately associated with a system of filaments (Fig. 6;see Wolosewick & Porter, 1979). These images therefore produce direct structuralconfirmation of the association of ribosomes with the cytoskeletal framework (Cer-vera, Dreyfuss & Penman, 1981). To circumvent superimposition, spread 3T3 cellshave additionally been subjected to surface-etching by extraction with Triton X-100after brief fixation with 0-1 % GA (Mesland et al. 1981a). This procedure dissolvesmembranes without other disturbance of ultrastructure, thus exposing themembrane-associated filamentous network at the cell surface, which can be observeddirectly by SEM or by means of replicas in the transmission microscope. As shownby the replica in Fig. 9, a complex filamentous web is revealed with different organiza-tions in the nuclear, perinuclear and marginal areas of the cell. Most prominent arethe small bundles of parallel filaments in the perinuclear region, that form a patternof their own within the filamentous network proper.

In hepatoma cells, the density of the hepatocytic network appears diminished (Figs3, 4); in M143066 mouse hepatoma cells, this is so, even to the extent of the virtualabsence of membrane-associated filaments, if compared with normal hepatocytes(compare Fig. 4, with fig. 1 of Mesland et al. 19816). By morphometric determina-tion of the number of coated pits and vesicles per (Urn2 of coated membrane surfaceit could be shown that this aberration is correlated with strongly diminished vesicula-tion of the cell's coated membrane (Mesland & Spiele, unpublished data). Also, TA3mammary carcinoma cells have rather open filamentous networks in many cases.Another interesting feature observed in the hepatocytic cells (BRL, H35, HTC)concerns the much more frequent occurrence of microtubules in the filamentous webof suspended cells than in that of spread cells. For example, suspended Reuber/H35cells show 0-67 pLm of microtubule per ^m2 membrane in contrast to only 0-10 /im inspread cells (more than 150/zm2 membrane measured). This observation demon-strates that microtubules. do associate closely with the membrane over considerabledistances (Figs 2, 3) and that the state of the cell influences the composition of themembrane-associated filamentous network (see Berlin, Caron & Oliver, 1979).


Microtubules have been found associated with lymphoid-type networks as well (Fig.8).

Taken together, these results reveal the existence of plasma membrane-associatedfilament systems in a variety of cells (Ben Ze'ev et al. 1979). Whether.these systemscan be considered to be functional entities, membrane skeletons analogous to the red

Figs 2-5. Filamentous webs associated with the cytoplasmic face of the plasma membranein suspended cells of different types. X 35 000. Corresponding cell images are shown in theinsets. X1100.

Fig. 2. Rat liver endothelial cell exposing its typical fenestrae (/) surrounded by a back-folded membrane and abundant coated pits (large arrows) and vesicles (double arrows).Note the microtubules (small arrows) running along the plane of the membrane.

Fig. 3. Reuber/H35 rat hepatoma cell. Compared with normal hepatocytes (Meslandet al. 1981a,b) this cell has a filamentous web of reduced density, as well as reducedoccurrence of coated pits and vesicles. Many microtubules occur embedded in the web(small arrows).

Fig. 4. A. M143066 mouse hepatoma cell, in which a filamentous web is virtually absent(compare with Mesland et al. 19816). Some bound cytoplasmic remnants (large arrow)'hover' above the membrane. Microvilli are visible underneath and at the border of themembrane. The limited occurrence of coated pits and vesicles stands in marked contrastto the many areas of coated membrane in this cell (small arrows), B. Cross-section of aMl43066 cell showing rather dense cytoplasm that can hardly be related to the dry-cleavedimage. X35 000.

Fig. 5. A. Typical lymphoid web of a mouse lymphosarcoma cell. A dense pattern ofspongy conglomerations cover the membrane. As judged by the lack of microvilli andmembrane protrusions, the membrane fragment is considered to be from a rather smoothregion of the cell, opposite the uropod (see inset, and Fig. 1E). B. Corresponding cross-section image. Note the more open cytoplasm of this cell compared to that in Fig. 4B andthe consequent impossibility of relating this to the corresponding dry-cleaved image.X35 000.

Fig. 6. Stereo micrograph of a dry-cleaved, spread BRL cell (see inset, X440) showinga membrane-associated filamentous web in the background, together with endoplasmicreticulum (er), polyribosomes (medium arrows) and associated filaments. Note the smallbundle of filaments with polyribosomes connected at irregular intervals (small arrows).Seemingly, all ribosomes appear associated with filaments that occur at the surface of theendoplasmatic reticulum and can be readily observed. Vesicular bodies (large arrows;possibly lysosomes) appear to be associated with filaments as well. X50000.

Figs 7-8. Stereo micrographs of dry-cleaved human small thymocytes. The correspond-ing SEM image is shown in the inset. X1100.

Fig. 7. Demonstration of a typical web in which filaments associate with spongy con-glomerations (small arrows) spread on the membrane. Morphological transitions of thistype can be rather abrupt in uropod-containing cells (see also, de Petris, 1981). Notefilament-associated ribosomes (arrowhead). X50000.

Fig. 8. Higher magnification of a somewhat larger thymocyte that clearly shows threefilament classes: 2-3 nm filaments (small arrows), 4nm filaments (arrows) and 6-7 nmfilaments (double arrows). Also indicated are an embedded piece of microtubule (openarrow) and a small coated pit (cp). X79 000.

Fig. 9. Au/Pd-sputter-coated surface replica of spread, and surface-etched, mouse 3T3cells. Etching was performed by Triton X-100 extraction of pre-fixed cells (see Materialsand Methods) and reveals a complex filamentous web at the cell surface. The structure ofthe web differs in the nuclear region (n), peri-nuclear region (p—p) and marginal region(m) of the cell. Note the system of filament bundles in the peri-nuclear region that formsa pattern of its own within the filamentous web. The arrow indicates the continuitybetween filaments and a microvillus. X3300.


Figs 2 and 3


Figs 4 and 5. For legend see p. 358.


Fig. 9. For legend see p. 358.


cell membrane skeleton, or merely part of the cytoplasmic microtrabecular network(Wolosewick & Porter, 1979), artificially separated by cleaving, remains to beanswered. However, several observations seem to favour the former possibility. First,we could not find a correspondence in density between cytoplasmic structure in cross-sectioned material and the filamentous networks in dry-cleaved material. Such corres-pondence might be expected, since Wolosewick (1980) showed that stereomicrographs of the cytoplasm in thick sections have microtrabecular morphology. Onthe contrary, Ml43066 cells have a dense cytoplasm and a highly reduced network,while MB6A lymphosarcoma cells have more open cytoplasm and dense, differentnetworks (compare Fig. 4 with Fig. 5). Another explanation for the virtual absenceof a filamentous network might be, however, that under the conditions applied theassociation of plasma membrane and filament system in M143066 cells is weak anddissociates during cleaving (possibly comparable to the cytochalasin B effect in nor-mal hepatocytes; Mesland et al. 19816). This possibility obviously does not detractfrom the significance of the altered structure observed. Second, with the notableexception of very flattened cells (Fig. 6), there is almost complete exclusion of organ-elles in dry-cleaved preparations. Thus there is a preferential tendency to break alongthe level of the filament system, indicating an anisotropy between the latter and thecytoplasm. Organelle-excluding cell cortices have been observed in a number of celltypes (Mooseker & Tilney, 1975; Goldman, Yerna & Schloss, 1976; Begg, Rodewald& Rebhun, 1978) and also the findings of Wolosewick & Porter (1979), using high-voltage EM of whole-mount preparations, hint at the existence of a subplasmalemmalsystem distinct from the cytoplasm proper.

In conclusion, we think we have visualized membrane-associated filament systems.And although their occurrence in cells other than erythrocytes may not be unexpected(Kirkpatrick, 1979; Ben Ze'ev et al. 1979; Mescher et al. 1981; Luna et al. 1981),the demonstration of their detailed structure by dry-cleaving may contribute sig-nificant new information about the plasma membrane and its mechanisms of control,both in normal cells and in malfunctioning tumour cells.

We acknowledge the use of SEM facilities at the Laboratory for Electron Microscopy andMolecular Cytology of the University of Amsterdam. We are grateful to Dr E. Roos for his helpful-ness and critical support of this study, and Drs C. A. Feltkamp and O. P. Middelkoop for valuablediscussions. The technical assistance of I. V. Van de Pavert and N. Ong is acknowledged.

REFERENCES

BEGG, D. A., RODEWALD, R. & REBHUN, L. I. (1978). The visualization of actin filament polarityin thin sections. Evidence for the uniform polarity of membrane-associated filaments..7. CellBiol.79, 846-852.

BEN ZE'EV, A., DUERR, N., SOLOMON, F. & PENMAN, S. (1979). The outer boundary of thecytoskeleton: Lamina derived from plasma membrane proteins. Cell 17, 859-865.

BERLIN, R. D., CARON, J. M. & OLIVER, J. M. (1979). In Microtubules (ed. K. Roberts & J. S.Hyams), pp. 443-485. London, New York, Toronto, Sydney, San Francisco: Academic Press.

BOYLES, J. & BAINTON, D. F. (1979). Changing patterns of plasma membrane-associated filamentsduring the initial phases of polymorphonuclear leukocyte adherence. J. Cell Biol. 82, 347-368.


BOYLES, J. & BAINTON, D. F. (1981). Changes in plasma-membrane-associated filaments duringendocytosis and exocytosis in polymorphonuclear leukocytes. Cell 24, 905-914.

BRANTON, D., COHEN, C. M. & TYLER, J. (1981). Interactions of cytoskeletal proteins on thehuman erythrocyte membrane. Cell 24, 24—32.

CERVERA, M., DREYFUSS, G. & PENMAN, S. (1981). Messenger RNA is translated when associatedwith the cytoskeletal framework in normal and VSV-infected HeLa cells. Cell 23, 113-120.

DE PETRIS, S. (1981). Distribution of Con A receptors in thymocytes and spleen cells during cellmovement. J. Ultrastruct. Res. 74, 267-283.

FIGDOR, C. G., VYTH-DREESE, F. A., BONT, W. S., SPITS, H. & D E VRIES, J. E. (1982). Isolation

of human thymocytes differing in maturation state and function by centrifugal elutriation.Thymus 4, 243-256.

GOLDMAN, R. D., YERNA, M-J. & SCHLOSS, J. A. (1976). Localization and organization ofmicrofilaments and related proteins in normal and virus-transformed cells. J. supramolec. Struct.5, 155-183.

GRATZER, W. B. (1981). The red cell membrane and its cytoskeleton. Biochem.J. 198, 1-8.HEUSER, J. E. & SALPETER, S. R. (1979). Organization of acetylcholine receptors in quick-frozen,

deep-etched, and rotary-replicated Torpeda postsynaptic membrane. J . Cell Biol. 82, 150-173.ISHIKAWA, H., TSUKITA, S. & TSUKITA, S. (1982). Association of the cytoskeleton with cell

membranes, a concept of plasmalemmal undercoat. In Biological Functions of Microtubules andRelated Structures (ed. Sakai, H. Mohri & G. G. Borisky), pp. 377-390. Tokyo: AcademicPress.

KIRKPATRICK, F. H. (1979). New models of cellular control; membrane cytoskeletons, membranecurvature potential, and possible interactions. Biosystems 11, 93-109.

LANG, R. D. A., NERMUT, M. V. & WILLIAMS, L. D. (1981). Ultrastructure of sheep erythrocyteplasma membranes and cytoskeletons bound to solid supports. .7. Cell Set. 49, 383-399.

LUNA, E. J., FOWLER, V. M., SWANSON, J., BRANTON, D. & TAYLOR, D. L. (1981). A membrane

cytoskeleton from Dictyostelium discoideum. I. Identification and partical characterization of anactin-binding activity. J . Cell Biol. 88, 396-409.

MESCHER, M. F., JOSE, M. J. L. & BALK, S. P. (1981). Actin-containing matrix associated withthe plasma membrane of murine tumour and lymphoid cells. Nature, Land. 289, 139-144.

MESLAND, D. A. M., LOS, G. & SPIELE, H. (19816). Cytochalasin B disrupts the association offilamentous web and plasma membrane in hepatocytes. Expl Cell Res. 135, 431-435.

MESLAND, D. A. M., SPIELE, H. & Roos, E. (1981a). Membrane-associated cytoskeleton andcoated vesicles in cultured hepatocytes visualized by dry-cleaving. Expl Cell Res. 132, 169-184.

MOOSEKER, M. S. & TILNEY, L. G. (1975). Organization of an actin filament-membrane complex.Filament polarity and membrane attachment in the microvilli of intestinal epithelial cells. J. CellBiol. 67, 725-743.

NICOLSON, G. L. & PAINTER, R. G. (1973). Anionic sites of human erythrocyte membranes. II.Antispectrin-induced transmembrane association of the binding sites for positively chargedcolloidal particles. .7. Cell Biol. 59, 395-406.

Roos, E., DINGEMANS, K. P., VAN DE PAVERT, I. V. & VAN DE BERGH WEERMAN, M. A. (1977).

Invasion of lymphosarcoma cells into the perfused mouse liver. J. natn. CancerInst. 58, 399-407.Roos, E., DINGEMANS, K. P., VAN DE PAVERT, I. V. & VAN DE BERG WEERMAN, M. A. (1978).

Mammary carcinoma cells in mouse liver: infiltration of liver tissue and interaction with Kupffercells. Br.J. Cancer 38, 88-99.

Roos, E., VAN DE PAVERT, I. V. & MIDDELKOOP, O. P. (1981). Infiltration of tumour cells intocultures of isolated hepatocytes. J. Cell Sci. 47, 385-397.

SANDERS, S. K., ALEXANDER, H. L. & BRAYLAN, R. C. (1975). A high-yield technique forpreparing cells fixed in suspension for scanning electron microscopy.^. Cell Biol. 67, 476-480.

SCHLIWA, M. & VAN BLERKUM, J. (1981). Structural interaction of cytoskeletal components. J .Cell Biol. 90,222-235.

SINGER, S. J. & NICOLSON, G. L. (1972). The fluid mosaic model of the structure of cell mem-brane. Science, N.Y. 175, 720-731.

TEMMINK, J. H. M. & SPIELE, H. (1980). Different cytoskeletal domains in murine fibroblasts. jf.Cell Sci. 41, 19-32.

VAN DRIEL, R. (1981). Binding of the chemoattractant folic acid by Dictyostelium discoideum cells.Eur.J. Biochem. 115, 391-395.


WOLOSEWICK, J. J. (1980). The application of polyethylene glycol (PEG) to electron microscopy.J. CellBiol. 86,675-681.

WOLOSEWICK, J. J. & PORTER, K. R. (1979). Microtrabecular lattice of the cytoplasmic groundsubstance. Artifact or reality. J . Cell Biol. 82, 114-139.

{Received 25 April 1983-Accepted 16 June 1983)

plasma membrane-associated filament systems in

Documents